Anthony Robert Booth

Intuitions
Intuitions
EDITED BY
Anthony Robert Booth

and
Darrell P. Rowbottom
1
1
Great Clarendon Street, Oxford, ox2 6dp,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© the several contributors 2014
The moral rights of the authors have been asserted
First Edition published in 2014
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2013957956
ISBN 978–0–19–960919–2
Printed and bound by
CPI Group (UK) Ltd, Croydon, cr0 4yy
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
In memory of E. J. Lowe, a gentle master.
Contents
Contributors ix
Introduction 1
Anthony Robert Booth
art I The Ontological and Epistemological

P
Standing of Intuitions
1. The Rational Roles of Intuition 9
Elijah Chudnoff
2. Intuitions: Their Nature and Probative Value 36
Ernest Sosa
3. Empirical Evidence for Rationalism? 50
Joel Pust
4. Moderate Intuitionism: A Metasemantic Account 68
Michael Johnson and Jennifer Nado
5. Intuition, ‘Intuition’, Concepts and the A Priori 91
C. S. I. Jenkins
Part II Intuitions in Disciplines or Sub-Disciplines

6. Intuitions in Science: Thought Experiments as Argument Pumps 119
Darrell P. Rowbottom
7. Novice Thought Experiments 135
Roy Sorensen
8. Moral Intuitionism, Experiments, and Skeptical Arguments 148
Mark van Roojen
9. Linguistic Intuitions in Context: A Defense of Non-Skeptical
Pure Invariantism 165
John Turri
Part III Challenges

10. The Challenge of Sticking with Intuitions through Thick and Thin 187
Jonathan M. Weinberg and Joshua Alexander
viii Contents
11. Sceptical Intuitions 213

Duncan Pritchard
12. Who Needs Intuitions? Two Experimentalist Critiques 232
Jonathan Jenkins Ichikawa
13. Grasp of Essences versus Intuitions: An Unequal Contest 256
E. J. Lowe
14. X-Phi without Intuitions? 269
Herman Cappelen
Index 287
Contributors
Joshua Alexander, Siena College

Anthony Robert Booth, University of Sussex
Herman Cappelen, University of St Andrews
Elijah Chudnoff, University of Miami
Jonathan Jenkins Ichikawa, University of British Columbia
C. S. I. Jenkins, University of British Columbia
Michael Johnson, Lingnan University
E. J. Lowe, University of Durham
Jennifer Nado, Lingnan University
Duncan Pritchard, University of Edinburgh
Joel Pust, University of Delaware
Darrell P. Rowbottom, Lingnan University
Roy Sorensen, University of Washington, St Louis
Ernest Sosa, Rutgers University
John Turri, University of Waterloo
Mark van Roojen, University of Nebraska, Lincoln
Jonathan M. Weinberg, University of Arizona
Introduction
Anthony Robert Booth†
Nineteen ninety-eight saw the publication of a volume of essays entitled Rethinking

Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry, edited by
Mike DePaul and William Ramsey. In an introductory essay, Gary Gutting wrote that
the volume was motivated by the recognition that philosophy, perhaps more so than
any other discipline, seems to be driven by our intuitions. The traditional philosophi-
cal practice of conceptual analysis, for instance, seems to go as follows. We wonder
“What is x?” We try to define x in terms of necessary and sufficient conditions. We then
endeavour to determine whether our definition of x is correct. We consider whether
it is prone to counter-examples, cases where our definition intuitively does not apply,
i.e. we test it against our intuitions. If the definition is prone to these counter-examples,
we conclude that it cannot be the right definition of x, or propose an error-theory to
explain why the intuition is mistaken; if it is not so prone, we tentatively conclude that
it is a good definition. We await further counter-examples.
According to Gutting, what was pressing at the time was that “a very sizable group
of ” philosophers held a more qualified view about how intuitions ought to be used in
conceptual analysis. These philosophers thought that although philosophical theories
should sometimes be revised when they clash with our intuitions, on other occasions
the intuitions should be rejected instead, particularly when they are also at odds with
our background beliefs. The process, they purported, is directed at our attaining reflec-
tive equilibrium. It was thus now necessary, according to Gutting, for philosophers to
articulate and defend this new practice. Furthermore, these philosophers owed us an
account of the epistemological pedigree these intuitions were meant to have, as this
was being put under pressure from “an unexpected quarter,” a nascent movement in
cognitive psychology. Psychologists working in this area were moving towards think-
ing of the practice of conceptual analysis as bogus, since “the data seem to show that,
except for some mathematical and geometrical concepts, it is not possible to use sim-
ple sets of conditions to capture the intuitive judgments people make regarding what
With thanks to Darrell Rowbottom for useful editorial suggestions on earlier drafts of this Introduction.
†
2 Anthony Robert Booth
falls under a given concept” (Gutting 1998, p. 25). Gutting saw the arrival of this move-
ment as throwing Western Analytical Philosophy into a “crisis.”
It is probably fair to say that Rethinking Intuition did not resolve that crisis. But it did
contain some classic and influential papers. Among them was George Bealer’s trench-
ant defense of the autonomy and authority of philosophy qua a priori discipline and
popularization of the idea that we should treat intuitive judgements as akin to percep-
tual judgements.
There have been two crucial developments, at least, over the past fifteen years. First
is the rise (or “resurgence”) of analytic metaphysics and of a school of philosophy that
rejects both the preoccupation with ordinary language and the idea that the default
methodology of philosophy is conceptual analysis. Adherents to this school typically
reject the idea that philosophy is all about giving definitions or that it is about defin-
ing concepts—philosophy ought to be directed at the world, not merely at our con-
cepts and language. A monumental defense of a particular take on this modern line
of thought is Timothy Williamson’s The Philosophy of Philosophy. For him, and many
others, “intuitions” are not to be thought of on the analogy of perception but rather as
judgements, or propensities to make judgements, about counterfactuals. And as such,
they are un-mysterious—congruous with the scientific world view. The prevalence
of a mode of thinking like Williamson’s has plausibly made it the case that, de facto,
Western Analytic Philosophy has not seen itself as undergoing a crisis as deep as that
which Gutting had envisaged. Nonetheless, second, the nascent challenge from the
cognitive psychologists has matured and evolved into a fully-fledged sub-discipline
of philosophy, with its own journal, and, so I’ve heard, its own musical anthem! It is
now known as Experimental Philosophy, commonly as X-Phi, and while its theoretical
object has remained much the same as it was at its inception, its methods have become
increasingly sophisticated.
So the broad questions that motivate this volume are: what is the epistemologi-
cal and ontological standing of intuitions in light of the important work done (by
Williamson, Sosa, and Bealer, among others) since Rethinking Intuition? Is that
standing variable across disciplines, and can investigating how intuitions are used
in disciplines beside philosophy help us configure the right way of thinking about
them? Has a more sophisticated X-Phi deepened or brought into sharper focus its
challenge to Western Analytic Philosophy? Does contemporary analytic meta-
physics constitute a new challenge to the intuitions-based practice of conceptual
analysis?
We have thus structured the volume as follows. The chapters in Part I address the
first question (concerning the epistemological and metaphysical standing of intui-
tions); the chapters in Part II, the second question (concerning how intuitions are used
in disciplines besides philosophy); and the chapters in Part III, the third and fourth
questions (concerning X-Phi and contemporary analytic metaphysics). We have
secured a mix of older and younger contributors to the volume. We have done this in
Introduction 3
order to reflect newer trends as well as the developed perspectives of those who have
been involved in the relevant debates for longer.
Indeed, the first chapter of Part I, “The Rational Roles of Intuition” is authored by
Elijah Chudnoff, a newcomer to the profession. His chapter concerns whether intui-
tion has a broader role to play than that which I canvassed earlier; that is, whether
intuitions can do more than bring “theoretical inputs” to a game of conceptual analy-
sis or reflective equilibrium. According to Chudnoff, intuitions can also have a role in
guiding action and, in particular, in guiding inference (conceived of as a theoretical
action). He makes the claim plausible by arguing that, contra Williamson, intuition is
akin to perception and, moreover (intriguingly), unlike belief or assertion, has both
mind-to-world and world-to-mind direction of fit. Chudnoff ’s chapter is followed
by “Intuitions: Their Nature and Probative Value” by Ernest Sosa, a well-established
behemoth of the intuitions literature. In this chapter, Sosa argues that intuitions
should not be understood in accordance with the perceptual model. However, draw-
ing on earlier work, he suggests that this is not to embrace the idea that intuitions are
merely products of linguistic or conceptual competence; instead, he argues that they
are deliverances of our understanding propositions. Yet, Sosa now argues, the fact that
intuitions are based on understanding relevant propositions cannot, alone, secure
their appropriate epistemic standing. What we need, he continues, is a competence
account of the rationality of intuition. In the chapter that follows, “Empirical Evidence
for Rationalism?,” Joel Pust also discusses the epistemic standing of intuitions. In par-
ticular, he discusses what he calls “Moderate Rationalism,” the view that having an
intuition that p gives one prima facie epistemic justification for believing that p. The
central question Pust takes up concerns how Moderate Rationalism ought to be justi-
fied; that is, can it (and should it) be empirically justified, or justified a priori? Pust
offers a careful exposition of the arguments in favour of the view that it should be jus-
tified empirically, and argues that none succeed. He also presents a case for thinking
that Moderate Rationalism should be justified a priori, “from the armchair”—taking
his cue from Anscombe’s and Armstrong’s defense of the idea that beliefs about causal
relations can be directly justified, given our ability to directly observe irreducible
causal relations between events (e.g., that the bread cut the knife). Michael Johnson’s
and Jennifer Nado’s chapter “Moderate Intuitionism: A Metasemantic Account”
goes back to the idea, criticized by Sosa, that the epistemic probity of intuitions is
best defended if we think of intuitions as the products of linguistic competence.
Here, Johnson and Nado offer a novel account of the reliability of intuition along the
model of a disposition-based meta-semantic theory. In the last chapter of Part I, C. S.
I. Jenkins does some very helpful, careful work to catalogue the disparate ways the
word ‘intuition’ has been put to use by analytic philosophers. Her catalogue is struc-
tured along four “bundles of symptoms,” viz.: commonsensicality, a prioricity, imme-
diacy, and metaphilosophical. She suggests that we should distinguish conceptions of
intuition associated with the commonsensicality bundle from those associated with
the a prioricity bundle, but that both of these conceptions are associated with the
metaphilosophical and immediacy bundles. Finally, drawing on some of her earlier

work, and having disambiguated between the various notions of intuition, she argues
that many of the arguments against one notion of intuition—strongly associated with
the a prioricity bundle—miss their targets.
Part II of this volume is about whether intuitions have a role to play that is unique
to philosophy, or to particular areas of philosophy. One fertile area of investigation
on this topic seems to be the use of thought experiments in disciplines outside of
philosophy, in particular the hard sciences. Are thought experiments put to differ-
ent methodological use in the hard sciences than in philosophy? In the first chapter
of Part II “Intuitions in Science: Thought Experiments as Argument Pumps,” Darrell
Rowbottom explores questions concerning the methodological status of thought
experiments in science and the ontological status of the intuitions they give rise to.
Against James Robert Brown, Rowbottom argues in favour of the view he calls scien-
tific intuition empiricism (contrasted with scientific intuition rationalism) according
to which the intuitions that thought experiments in science give rise to are ultimately
derived from experience. Rowbottom also discusses John Norton’s view that thought
experiments in science are arguments, but argues that they are best construed as argu-
ment pumps instead. On Rowbottom’s view, then, there is a structural parallel between
thought experiments in science and actual physical experiments. If Rowbottom is
right, then only a very naturalistic conception of philosophy’s methodology will be
confluent with the use of intuition in science. Taking a different approach to this issue,
in the next chapter of Part II, “Novice Thought Experiments,” Roy Sorensen questions
whether data concerning lay-intuitions on thought experiments in philosophy should
carry the same theoretical weight that novice experiments in physics carry for the pro-
fessional physicist—that is to say, none at all. As he puts it: “No physicist (in his role as
a physicist) surveys the conflict between undergraduate physics experiments and the
experiments of their professors.”
And, according to Sorensen, the same is true as regards thought experiments in
physics. So unless we think that philosophy is methodologically special, then there is
surely expertise in philosophy, as there is in any other discipline. Further, given that
physicists also use thought experiments, we probably ought to be careful about think-
ing that philosophy is methodologically special. Intuitions figure not only in philos-
ophy. And, according to Sorensen, the epistemic role and value of intuitions is tied
up with the way in which they reflect learning, both in philosophy as elsewhere. If
Sorensen is right, it seems that the challenge from X-Phi somewhat misfires, directed,
as it is, at highlighting the mismatch between expert and ordinary intuition. I think it
is interesting to note that this dialectical response to X-Phi, a kind of companion in
guilt strategy, may put pressure on the autonomy of philosophy, as traditionally con-
ceived. One might wonder, then, whether one of the lasting legacies of X-Phi will be a
re-affirmation of a particular kind of naturalistic conception of philosophy.
An area of philosophy where the role of intuitions seems particularly prominent is
ethics. Ethical Intuitionism as a position in moral epistemology still enjoys widespread
Introduction 5
support. But, as Mark van Roojen points out in the next chapter of Part II, “Moral
Intuitionism, Experiments and Skeptical Arguments,” it is still important that intui-
tionists in ethics respond to the argument against relying on intuitions brought for-
ward by experimental philosophers. Van Roojen argues that the ethical intuitionist
must take on board some of these challenges and accept merely a “moderate” intui-
tionism. The moderate intuitionist, for Pust, accepts a foundationalist epistemology
that requires intuitions to function as anti-sceptical regress-stoppers and figure in
theory-testing roles, but nonetheless be inferential judgements. If this moderate view
is tenable, the ethical intuitionist can accept the experimentalist claims vis-à-vis the
unreliability of intuitive judgements, without eschewing wholesale intuitions’ making
a positive contribution to moral epistemology. If this is the role that intuitions play in
ethics, one might wonder: is this the role intuitive judgements play in all of philosophy?
In the last chapter of Part II, “Linguistic Intuitions in Context: A Defense of
Non-Skeptical Pure Invariantism,” John Turri investigates the way that philosophers
have used linguistic intuitions in epistemology to adjudicate the debates between con-
textualists and invariantists. Turri argues that the pure invariantist—the holder of the
view that the truth conditions for knowledge ascriptions are not context sensitive and
that only theoretical considerations legitimately affect the strength of one’s epistemic
situation—can explain the intuitions made salient by contextualist cases to her advan-
tage with the theoretical aid of indirect speech acts, in particular indirect requests and
denials. Turri gleans some overall methodological lessons from this regarding when
truth does, and does not, matter when we are questioning the appropriateness of lin-
guistic behaviours.
Part III opens with Weinberg and Alexander’s “The Challenge of Sticking to
Intuitions through Thick and Thin.” The authors of this chapter offer us a state-of-
the-art account of what X-Phi takes its central target to be and how the debate on intui-
tions has moved on and evolved in light of the experimentalists’ critique. In particular,
they discuss how a certain notion of intuition—what they call a “thick” notion—has
been deployed to meet the challenge from X-Phi. Weinberg and Alexander argue,
unsurprisingly, that such a move brings with it its own set of problems and, in some
modalities, fails to adequately immunize intuition-driven philosophical methodology
from the challenge. According to at least some of the more prolific exponents of X-Phi,
then, philosophy has not survived their challenge.
The last few chapters of this collection comprise less X-Phi partisan responses to
the experimentalists’ challenge, and alternative views of philosophical methodology.
One way to resist the X-Phi challenge is to appeal to the possibility of there being a
divergence between the philosophical intuitions of experts and non-experts. If there
is such a divergence, and it is the experts’ intuitions that carry the evidential weight in
philosophy, then data about the ‘folk’ is irrelevant to a project aimed at casting doubt
on the legitimacy of current philosophical methodology. This is, as we saw, and in
rough strokes, the line that Sorensen defends, and it is also the line taken up by Duncan
Pritchard in this volume’s “Sceptical Intuitions.” In this chapter, Pritchard illustrates
how intuitional judgements are sometimes significantly informed by expertise via a

case-study regarding the debate about radical scepticism.
A different line of defense against the challenge from X-Phi is to argue that the
role of intuition in philosophy has been somewhat exaggerated. This is the broad
line that the authors of the next three chapters, in very different ways, take up.
In his chapter “Who Needs Intuitions?,” Jonathan Jenkins Ichikawa usefully sur-
veys the different ways in which the experimentalists’ critique can be motivated
and examines whether it is predicated on the assumption that intuitions occupy
a central role in philosophical inquiry. The next chapter is E. J. Lowe’s “Grasp of
Essences versus Intuitions: An Unequal Contest.” As I mentioned in opening, one
of the interesting developments in philosophy since the publication of Rethinking
Intuition has been the rise of what is being called “Analytic Metaphysics.” This could
be viewed as something of a historical surprise, given that it has come alongside
the development of X-Phi and its concomitant naturalistic worldview. So then it is
not so surprising to find that many current practitioners of analytic metaphysics
eschew an intuition-driven philosophical methodology. But what might an alterna-
tive methodology, suitable for the ends of metaphysical inquiry, be? In his chapter,
Lowe argues that we can legitimately engage in (realist) metaphysics because we are
rational, and so able to grasp the essences of at least some mind-independent enti-
ties. To grasp an entity’s essence for Lowe, however, is not to properly understand a
relevant concept (or concepts) related to the entity—this is where Lowe’s favoured
methodology parts company with intuition-driven philosophy. Intuitions for Lowe
are no more than psychological states, incapable of constituting reliable evidence
as regards the nature of reality. The kind of judgement underpinning philosophical
inquiry (as well as theoretical science, Lowe argues) must be thought of as having
the power to reveal the mind-independent truths about essence. In his recent book
Philosophy without Intuitions (Oxford University Press 2012) Herman Cappelen has
also argued against the idea that philosophical inquiry does and should accord a cen-
tral role to intuitions. In the last chapter of this volume “X-Phi without Intuitions?,”
Cappelen addresses the rejoinder from the X-Phi camp, that a close analogue of the
X-Phi critique can be levied against any alternative non-intuition-centered concep-
tion of philosophical practice. Cappelen argues that the rejoinder is unsuccessful,
or, at least, underdeveloped—that is, the experimentalists’ job, if it is a legitimate
job, is far from over.
On behalf of my co-editor—Darrell Rowbottom—as well, I would like to thank the
following for their help in making this volume possible: Peter Momtchiloff, Mike De
Paul, an anonymous reader for Oxford University Press, and several anonymous refer-
ees whose reports we independently solicited.
PA RT I
The Ontological and
Epistemological Standing
of Intuitions
1
The Rational Roles of Intuition
Elijah Chudnoff †
1.1. Introduction
We have attitudes—beliefs, desires, hopes, fears, intentions—and we perform
actions—mental ones such as counting sheep before falling asleep and bodily ones
such as making the bed after waking up. Some of these attitudes and actions are more
reasonable than others. We have experiences, such as perceptions, bodily sensations,
recollections, imaginings, and, I would add, intuitions. Some of these play roles in
making some of our attitudes and actions more reasonable than others. By the rational
roles of a type of experience I mean the roles experiences of that type play in making
some of our attitudes and actions more reasonable than others.
In this chapter I will explore the rational roles intuitions play. Two have been dis-
cussed widely recently:
Justifier: Intuitions justify beliefs.
Evidence: Intuitions are evidence for beliefs.
I don’t assume these are the same rational role. I discuss both of them briefly below. My
main aim in this chapter, however, is to defend the view that intuitions play an addi-
tional rational role. To a first approximation:
Guidance: Intuitions guide actions.1
Here is the plan.
In section 1.2, I set out some assumptions I will make about the nature of intuition.
In section 1.3, I discuss the justifier and evidence roles. In sections 1.4 to 1.6, I make
a case for thinking that intuitions play the guidance role. The argument proceeds by
† I would like to thank John Biro, Robert D’Amico, Kirk Ludwig, Greg Ray, Darrell Rowbottom, Michael
Slote, Gene Witmer, and two anonymous reviewers at OUP for their helpful feedback on earlier drafts of
this paper.
1 In this chapter I focus on mental actions, though I think intuitions play a role in guiding some bodily
actions as well.
10 Elijah Chudnoff
reflection on the transmission of justification through inference. According to infer-

ential internalists, in order to gain justification for believing the conclusion of an
argument by inferring it from the premises in that argument, one must “see” that the
premises support the conclusion. In section 1.4, I motivate this view and endorse the
idea that one’s “seeing” such a support relation consists of one’s having an intuition.
Recently, Paul Boghossian (2008 [2003]) has pressed a regress argument against infer-
ential internalism inspired by Lewis Carroll’s “What the Tortoise Said to Achilles”
(1995 [1905]). In section 1.5, I review Boghossian’s argument, isolating what I take to
be its main premise. In section 1.6, I develop a response to Boghossian’s argument that
requires intuitions to play the guidance role. In section 1.7, I elaborate on the view that
intuitions play the guidance role in response to two objections. And in section 1.8,
I take up the question of what it is in virtue of which intuitions play a guidance role,
when they do so.
1.2. What Are Intuitions?

Consider three theses about perception:
(1) Perceptual experiences are sui generis experiences; they should not be identified
with doxastic attitudes or dispositions—such as beliefs, or inclinations to believe.2
(2) Perceptual experiences possess presentational phenomenology; whenever you
have a perceptual experience representing that p, there is some q (maybe = p)
such that—in the same experience—it perceptually seems to you that q, and you
seem to be sensorily aware of the chunk of reality that makes q true.3
(3) Perceptual experiences fit into your stream of consciousness like experiential
atoms; they are not constituted by your other experiences, such as your imagin-
ings and conscious thoughts.4
In my view intuition is similar to perception with respect to the first two points, and dis-
similar with respect to the third. That is, I endorse the following theses about intuition:
(4) Intuition experiences are sui generis experiences; they should not be identified
with doxastic attitudes or dispositions—such as beliefs, or inclinations to believe.5
2 Cf. Jackson (2009 [1977]), Evans (1981), Peacocke (1983), Searle (1983), Foster (2001), Huemer (2001).
Armstrong (1968) is a well-known defense of the opposing view; see also Glüer (2009).
3 Cf. McDowell (1994), Robinson (1994), Sturgeon (2000), Foster (2001), O’Shaughnessy (2002), Crane
(2005), and Johnston (2006). They all agree that perception possesses presentational phenomenology,
though not all adopt the same gloss on what this amounts to. I explore the nature of presentational phenom-
enology further in Chudnoff (2012).
4 Contrast the views of some earlier writers according to which perceptual experiences—as opposed
to mere sensations—are supplemented by imagination. For discussion see Strawson’s “Imagination and
Perception,” in Strawson (2007).
5 Cf. Bealer (1998, 2000, 2002) and Huemer (2001, 2008). For arguments in favor of the opposing view
see: Williamson (2004, 2005, 2007) and Earlenbaugh and Molyneux (2009).
The Rational Roles of Intuition 11
(5) Intuition experiences possess presentational phenomenology; whenever you

have an intuition experience representing that p, there is some q (maybe = p)
such that—in the same experience—it intuitively seems to you that q, and you
seem to be intuitively aware of the chunk of reality that makes q true.6
(6) Intuition experiences fit into your stream of consciousness like experiential
molecules; they are constituted by your other experiences, such as your imagin-
ings and conscious thoughts.7
I have argued for theses (4) through (6) at length elsewhere (Chudnoff 2011a, 2011b,
2012, 2013a, and 2013b). Here I will briefly indicate some motivation for accepting them.
Consider the following two claims:
If a < 1, then 2 – 2a > 0 (A)
√ 7 + √10 > √ 3 + √17 (B)
Both (A) and (B) are truths we can come to know. But there is a difference. For most of
us, (B) is only knowable by calculation or testimony. (A), on the other hand, is some-
thing that it is possible to just “see,” i.e. intuit to be true, perhaps after a moment or two
of reflection.
Contrast the experience you have when you intuit (A) with the experience you have
when you consciously judge (B), say because you calculate it or receive testimony that
it is true. A natural way to characterize what distinguishes the intuitive way of becom-
ing convinced that (A) is true is this. In this case, you are not compelled by authority or
argument to believe that if a < 1, then 2 – 2a > 0; nor do you just find yourself mysteri-
ously tempted to believe this proposition. Rather, the proposition is made to seem true
to you by your apparent insight into the bit of mathematical reality that makes it true,
namely the dependence of 2 – 2a on a. This is why I say that intuitions have presen-
tational phenomenology. If intuitions have presentational phenomenology, however,
then they shouldn’t be identified with doxastic attitudes or dispositions. One might
have a doxastic attitude or disposition in light of having an intuition experience with
presentational phenomenology, but the doxastic attitude or disposition itself is some-
thing else, since it is possible to have such an attitude or disposition, even a conscious
one, without having any presentational phenomenology.
6 This view—though not my way of putting it—was more common among earlier writers on intuition. In
Chudnoff (2011b), I give reasons for attributing it to René Descartes, Edmund Husserl, Bertrand Russell, and
Kurt Gödel. I would add certain moral intuitionists such as John Balguy and Richard Price to the list of his-
torical proponents; see their works excerpted in D. D. Raphael (1969). Among more recent writers, Panayot
Butchvarov and Laurence BonJour seem to me to defend similar views; see Butchvarov (1970) and BonJour
(2005).
7 Cf. Husserl (1975, 2001), Parsons (1980, 2007), and Tieszen (1989, 2005). I believe Husserl was the first to
defend this view. It was common ground among those in the phenomenological tradition; see, for example,
Reinach (1911), Gurwitsch (1964), and Lévinas (1995).
12 Elijah Chudnoff
Reflection on example intuitions seems to me to provide some motivation for

accepting (4) and (5). What about (6)? Here the considerations are a bit more involved.
At least three observations are relevant.
First, in having an intuition a proposition appears to you to be true.
Second, the same proposition can appear to you to be true in different ways, and these
different ways are significant enough so that we should count them as part of the identity
of the intuitions with which they are associated. For example, maybe you think that one
intuition might justify believing a proposition more than another and that this is due to
the fact that it makes the proposition appear to be true in a clearer manner than the other.
Here is an example from Descartes: if you try to intuit that a chiliagon has more sides
than a 999-sided figure by imagining it, your intuition will be less clear than if you relied
solely on your intellectual grasp of the difference between 1,000 sides and 999 sides.
Third, the differences in ways propositions appear true in intuitions are correlated
with differences in associated reflections. In the example from Descartes the differ-
ences in the intuitions are associated with differences in accompanying thoughts and
imaginings. A straightforward approach to individuating intuitions so that their iden-
tities include ways propositions appear to be true in them and not just the propositions
that do appear to be true in them is to take such thoughts and imaginings to be parts of
the intuitions. This suggests that (6) is true, i.e. that intuitions are constituted by other
experiences such as thoughts and imaginings.
1.3. Intuitions as Justifiers and as Evidence

Take your intuition that if a < 1, then 2 – 2a > 0. You have no reason to reject this claim
or distrust your intuition. So, plausibly, your intuition makes it the case that you have
justification for believing that if a < 1, then 2 – 2a > 0. Suppose, further, that you take
your intuition at face value. You thereby form a justified belief that if a < 1, then 2 – 2a >
0. Your belief is justified because it is based on your intuition.
The foregoing suggests that at least some intuitions are justifiers. But it also suggests
that we should distinguish between two ways in which such intuitions are justifiers.
Following standard terminology, some intuitions are propositional justifiers and some
intuitions are doxastic justifiers. If an intuition is a propositional justifier, then it makes
it the case that you have justification for a belief. If an intuition is a doxastic justifier,
then it plus the fact that you base a belief on it make it the case that your belief is justified.
If intuitions are justifiers, then it is natural to ask: in virtue of what does an intuition
play the justifier role, when it does so? There are a number of options one might pur-
sue. Two initial ideas are:
Reliabilism: if an intuition plays the justifier role, it does so in virtue of being a reli-
able indicator of the truth of its content (cf. Bealer 1998a, 1998b, 1999; Goldman
2007; Peacocke 2003; and Sosa 2007, 2009).
Phenomenalism: if an intuition plays the justifier role, it does so in virtue of having a

certain phenomenology with respect to its content (cf. Huemer 2005).
I call these initial ideas, since they require and have received further elaboration. My
own view is that Phenomenalism is the preferable starting point, and that it requires
one simple elaboration: the relevant phenomenology is presentational phenomenol-
ogy, as characterized in section 1.2. Whether this is the correct view will not make a dif-
ference to the rest of my discussion, so I will not take up its defense here (see Chudnoff
2011a for the defense).
So far I have framed my discussion in terms of justification. Much recent work on
intuition, however, is framed in terms of evidence. The question I want to consider
now is: how might intuitions being evidence relate to them being justifiers?
On one way of thinking about evidence, there isn’t much to say in response to this
question since “evidence” is just a terminological variant of “justifier.” That is:
Your evidence consists of whatever is a propositional justifier for you, i.e. whatever
makes it the case that you have justification for believing something.
If this is how we think of evidence, then, insofar as we agree that intuitions play the jus-
tifier role, we should understand the idea that intuitions are evidence just as we under-
stand the idea that intuitions are justifiers.
But there is another way of thinking about evidence, on which the relation-
ship between intuitions as justifiers and as evidence is less clear. We might put it
like this:
Your evidence consists of considerations that epistemically count in favor of or
against your having certain beliefs.
This formulation leaves open two issues. One issue is about the ontology of evi-
dence: are the considerations that constitute evidence facts or propositions? Another
issue is about the conditions on possessing evidence: must the considerations that con-
stitute your evidence be known or believed, or believed with justification, or prop-
ositionally justified for you, etc? These are important questions. But the issues I am
concerned with do not hinge on answers to them.
If we think of evidence as epistemically favorable considerations, then insofar as we
use “intuition” to pick out a kind of experience, we should agree that intuitions are not
evidence, since experiences are not considerations.
Suppose we adopt this second way of thinking about evidence and we use “intuition”
to pick out a kind of experience. Given that your intuitions are not themselves your
evidence, how do the intuitions that you have stand with respect to your evidence? Say
you intuit that p. What follows about your evidence? Here are some possibilities:
—Your evidence now includes the consideration that p.
—Your evidence now includes the consideration that you have had the intuition
that p.
14 Elijah Chudnoff
Suppose that this is all that follows about your evidence. Then it appears that there is
a problem. What evidence do you have for believing that p? The consideration that p
seems question-begging.8 The consideration that you have had the intuition that p is
about your own psychology, and, one might worry, even if it lends some support to
believing p, the support it lends is very slight (cf. Goldman 2007; Earlenbaugh and
Molyneux 2009a and b; Williamson 2007; Ichikawa 2013; Cath 2012).
I don’t think this is much of a problem. Suppose you don’t have very good evidence
for believing that p—the considerations available to you are either question-begging or
psychological. Still you might be justified in believing that p to a very high degree. The
reason why is that even if your intuition is not itself evidence, and its occurrence does
not ensure that you have good evidence for believing that p, still, it is a justifier, and it
might justify you in believing that p to a very high degree. The moral is that epistemic
rationality cannot be understood wholly in terms of evidence, if evidence is under-
stood in the second way we have distinguished, as consisting of epistemically favorable
considerations. More precisely, the following claim fails to hold: if your intuition expe-
rience representing that p justifies you in believing that p, then the justification you
thereby have for believing that p consists of having evidence for believing that p. If it
strikes you as incongruous to say that you might have a high degree of justification for
believing that p though only slight evidence for believing that p, then that just militates
in favor of understanding evidence along the lines of the first way, distinguished above,
as consisting of justifiers. Then the claim—that if your intuition experience represent-
ing that p justifies you in believing that p, then the justification you thereby have for
believing that p consists of having evidence for believing that p—will hold trivially,
since the evidence you have will just be the justifier, i.e. the intuition experience.
1.4. Inferential Internalism

The aim of the next four sections is to make a case for thinking that intuitions guide
action and that this is a distinct rational role from the justifier and evidence roles. The
argument will focus on a puzzle about inference.
Consider the following argument:
(1) Every even number is divisible by 2.
(2) The number of pigs in the pen is even.
(3) So, the number of pigs in the pen is divisible by 2.
Say you know (1) from school and (2) from counting. You “see” that (1) and (2) support
(3). So you infer (3) from (1) and (2) and thereby come to know that the number of pigs
in the pen is divisible by 2.
8 Cf. Glüer (2009) on this issue, as it comes up in thinking about the relationship between perceptual
experiences and reasons for belief.
In what does your “seeing” that (1) and (2) support (3) consist? Plausibly, it consists
of your having an intuition experience that represents that (1) and (2) support (3). This
is a historically popular idea—at least among rationalists.9 Here are two considerations
in favor of it. First, the subject matter of the claim that (1) and (2) support (3) is similar
to the subject matter of typical claims that intuition justifies—e.g. the claim that if a <
1, then 2 – 2a > 0. Both are claims about non-empirical matters. It could be that there
are two or more distinct sources of justification for claims about non-empirical mat-
ters, but this view is prima facie unattractive and should be avoided if possible. Second,
experiences of the sort that make the claim that (1) and (2) support (3) evident are
similar to typical intuition experiences. Specifically, they possess the characteristics of
intuition listed in section 1.2: they are sui generis, presentational, and constituted by
thoughts and imaginings.
The puzzle about inference concerns the sort of transition that occurs from your
intuition that (1) and (2) support (3) to your inferring (3) from (1) and (2). The puzzle
is that there are both reasons to think that the transition from intuition is required for
knowing by inference and reasons to think that the transition from intuition is not
required for knowing by inference. The solution I will propose is that the reasons for
thinking that the transition from intuition is not required for knowing by inference
depend on an assumption about intuition that should be rejected. The assumption is
that intuition has solely mind-to-world direction of fit, like a belief or an assertion.
I will suggest that intuition sometimes has both that and world-to-mind direction of
fit, like a desire or a command. In these cases intuition is what Ruth Millikan (1995) has
called a pushmi-pullyu representation: it both describes a state of affairs and directs an
action.
Inferential internalists think that the transition from intuition or some analogous
mental state is required for knowing by inference. The balance of this section is dedi-
cated to explaining in more detail what this view is and what motivation there is for
adopting it.
Here are a few recent formulations of inferential internalism:
[a]‌The inferential internalist is committed to the view that for S to be justified in believing P on
the basis of E, S must not only be justified in believing E but must be justified in believing that E
makes probable P (where E’s entailing P can be viewed as the upper limit of E’s making probable
P) (Fumerton 2006, p. 101).
[b] (Simple Inferential Internalism): A deductive inference performed by S is

warrant-transferring just in case (a) S is justified in believing its premises, (b) S’s justification for
believing its premises is suitably independent of his justification for believing the conclusion,
and (c) S is able to know by reflection alone that his premises provide him with a good reason for
believing the conclusion (Boghossian 2008 [2003], p. 268).
9 For example, see Descartes’ Rules in Descartes (1985), Ewing’s “Reason and Intuition,” in Ewing (1968
[1941]), Pollock (1974), and BonJour (1998).
16 Elijah Chudnoff
[c]‌In order for one to have positive epistemic status ɸ in virtue of believing P on the basis of R,
one must believe that R evidentially supports P, and one must have positive epistemic status ɸ in
relation to that later belief as well (Leite 2008, p. 422).
There are important differences among these formulations of inferential internalism.

I’ll mention four. First, [a]‌and [c] are more general than [b]: whereas [b] is restricted to
deductive inference, [a] and [c] range over at least all sorts of inference, and [c] maybe
even further over cases of epistemic basing that do not involve inference. Second,
whereas [a] and [c] aim to give necessary conditions on the acquisition of inferential
justification, [b] aims to give necessary and sufficient conditions on the acquisition of
inferential justification. Third, [a] and [b] do not require you to have a belief about the
relation between premise and conclusion in your inference, but [c] does. Formulation
[a], for example, requires that you have justification for a belief about the relation
between premise and conclusion, but it is possible to have justification for believing
something, even if you do not believe it. Formulation [b], likewise, requires that you be
able to know by reflection alone about the relation between premise and conclusion,
but it does not require that you actually do know, or even have any belief about the mat-
ter. Formulation [c], on the other hand, requires that you have such a belief. Fourth,
[a] uses one epistemic notion—justification; [b] uses three—justification, being in a
position to know, and warrant transmission; and [c] uses a schematic letter covering a
range of positive epistemic notions.
These differences aside, there is an obvious family resemblance holding among [a]‌,
[b], and [c]. From them I distill the following view:
(II-) S knows that p by inferring p from q1 . . . qN only if S intuits that q1 . . . qN support p.
This view is more general than [b]‌since it is about all inferences, but potentially less
general than [c] since it is about inferential justification only, and not epistemic basing
more generally. It is similar to [a] and [c] in that it aims to give a necessary condition,
not a sufficient condition, but it is less committal than them, as it is silent on all condi-
tions on inference save the one that distinguishes inferential internalism from other
views. It commits to the view that the mental state representing the support relation
is an intuition. Given this commitment, (II-) does not require that S must believe that
q1 . . . qN support p, since it is possible to intuit something and not believe it. Finally, it
uses the notion of knowledge, since analogous principles that use epistemic notions
weaker than knowledge imply (II-).
(II-) does not seem to me to be a proper formulation of inferential internalism, at
least insofar as it is committed to a principle about knowledge.10 For that, I think we
need to add something:
10 It might be sufficient if reformulated as a principle about propositional justification. But, as indicated

above, if one is an inferential internalist about propositional justification, then one should be an inferential
internalist about doxastic justification and states, such as knowledge, requiring doxastic justification.
S knows that p by inferring p from q1 . . . qN only if S infers p from q1 . . . qN in part
(II)
because S intuits that q1 . . . qN support p.
Though none of [a]‌, [b], or [c] suggests the additional condition—what we might call
the becausal condition11—there are reasons to include it.12 I will mention three.
First, there is a strategic reason: (II) is stronger than (II-), so if it can be defended, so
can (II-). Adding the becausal condition doesn’t hurt strategically.
Second, there is a dialectical reason: even though Boghossian does not formulate
the becausal condition in [b]‌, in arguing against inferential internalism he takes it to
be committed to more than just (II-), and (II) is a plausible articulation of just what
more.13 Of course, an inferential internalist might then just reply to Boghossian by
distinguishing (II-) from (II) and claiming to endorse the former, not the latter.14 But
this is unsatisfying—and the reason why is the third, and most important, reason for
adopting formulation (II).
The third reason is that the most compelling motivation for inferential internalism
motivates (II) as much as it motivates (II-). The most compelling motivation for infer-
ential internalism derives from reflection on certain examples. Consider the following
two arguments.
Argument A
( A1) Connie and Cyndi are a cone and a cylinder with the same base and height.
(A2) Therefore, Cyndi encloses a greater volume than Connie.
Argument B
( B1) Connie and Cyndi are a cone and a cylinder with the same base and height.
(B2) Therefore, Cyndi encloses three times the volume of Connie.
Imagine Smith. Smith doesn’t have any particular mathematical expertise. But sup-
pose he has justification for believing (A1)—someone tells him it is so, or he measures
it himself, or whatever. From (A1) he infers (A2). Plausibly, he now also has justifica-
tion for believing (A2). Suppose, on the other hand, he has justification for believing
(B1)—we’re just relabeling (A1). From (B1) he infers (B2). Is it plausible in this case to
say that he has justification for believing (B2)? I think not. Why?
On the face of it, it is because he can intuit that (A2) follows from (A1), but he can-
not intuit that (B2) follows from (B1). It might take him a moment to intuit that (A2)
11 How exactly to understand the becausal condition is an issue I will discuss below.

12 Brewer seems to endorse such a becausal condition in Brewer (1995).
13 There might be a principle stronger than (II-) but weaker than (II) that best fits the conception of infer-
ential internalism Boghossian has in mind when he is arguing against the view. See n.17. Since I am defend-
ing the stronger principle, (II), whether this is so does not matter for my purposes.
14 I am partly inclined to think that Leite’s response to Boghossian’s Carrollian argument consists in doing
precisely this: endorsing (II-), rejecting (II). See Leite (2008, pp. 429–32). But I am not confident that this is
a correct interpretation. Leite says that one’s appreciation of the relation between premise and conclusion
must “play a role” without “doing something” in one’s inference (see p. 432).
18 Elijah Chudnoff
follows from (A1), but it is certainly within his capabilities. It is difficult to imagine him
intuiting in a similar way that (B2) follows from (B1), however. To do this, he would
have to intuit the exact ratio of the volume of a cone to the volume of a cylinder with
the same base and height. And that is beyond his limited capabilities.
Now suppose that while Smith does intuit that (A1) supports (A2), this intuition
plays no role in accounting for why he makes the inference he does. Suppose he just
ignores his intuition and makes the inference anyway. Does he, in this re-imagined
case, gain justification for believing (A2)? No. The reason why not is that even though
Smith intuits that the premise supports the conclusion, he does not infer the conclu-
sion from the premise in light of this intuition, but independently of it. This observa-
tion suggests that Smith must not only intuit that (A1) supports (A2) in order for his
inference to give him justification for believing (A2), but, also, must make his inference
in part because he has this intuition. To summarize: reflection on examples motivates
(II) as much as it motivates (II-).15
1.5. Boghossian’s Carrollian Argument

Consider an inference that accords with modus ponens (MPP):
(1) If today is the 20th, then Martha Argerich is playing today in Carnegie Hall.
(2) Today is the 20th.
(3) Martha Argerich is playing today in Carnegie Hall.16
Suppose (II) is true. So: in order to know (3) by inferring it from (1) and (2), one must
intuit that (1) and (2) support (3), and one must infer (3) from (1) and (2) at least in part
because of this intuition. This raises questions of two sorts.
First, there are questions about one’s intuition that (1) and (2) support (3). What is its
precise content? For example, is it about MPP inferences in general, or about this particu-
lar MPP inference? I will set these questions aside for now. I return to them in section 1.7.
Second, there are questions about the becausal condition. What exactly is it for
one’s inference to be made in part because of one’s intuition? Boghossian poses a sim-
ilar question: “We can ask how my knowledge of the validity of the inference from
(1) and (2) to (3) is supposed to bear on my warrant to infer (3)?”17 His main reason
15 The point I am making here parallels a more familiar point about justified belief. In order to have a justi-
fied belief that p it does not suffice to have a belief that p and justification for believing that p: one must base
one’s belief that p on one’s justification for believing that p. For further discussion, see Feldman and Conee
(1985).
16 The example is from Boghossian (2008 [2003], p. 267).
17 Boghossian (2008 [2003], p. 274). It is because Boghossian takes this to be a question that inferential
internalists must face that I believe he thinks inferential internalism is committed to more than (II-). But it
is because there is a difference between an intuition bearing on one’s warrant for inferring and an intuition
bearing on one’s inferring that I believe (II) might be too strong to capture his conception of inferential inter-
nalism. As pointed out above, this doesn’t matter for my purposes. See n.18 for a reason to think a principle
stronger than (II-) but weaker than (II) is still too weak.
for rejecting inferential internalism is that he does not think that this question has a
satisfying answer:
But it is very hard to see, once again, how my putatively justified judgment that my premises
entail my conclusion could bear on my entitlement to draw the conclusion in anything other
than inferential form, thus:
(iv) This particular inference from (1) and (2) to (3) is valid.

(v) If an inference is valid, then anyone who is justified in believing its premises and knows
of its validity is justified in inferring its conclusion.
Therefore,
(vi) Anyone who is justified in believing the premises of this inference is justified in believ-
ing its conclusion.
(vii) I am justified in believing the premises (1) and (2).
Therefore,
(viii) I am justified in inferring (3).
Even if we conceded, then, that we have rational insight into the validity of specific inferences,
we do not escape the threat of circularity that afflicts the internalist account. Once again, an abil-
ity to infer justifiably according to MPP is presupposed (Boghossian 2008 [2003], pp. 274–5).
In Boghossian’s argument (iv) is the content of my intuition that (1) and (2) support

(3). How does the content of this intuition bear on my inference from (1) and (2) to
(3)? Boghossian claims that it can only do so by figuring in another inference, namely
the inference from (iv) and supplementary premises to (viii), the conclusion that I am
justified in inferring (3). If all this is so, then the inferential internalist is in trouble. One
problem, which Boghossian points out, is that the inference from (iv) and supplemen-
tary premises to (viii) invokes MPP, thus launching us on the sort of regress Carroll
(1995 [1905]) illustrates in his dialogue between the Tortoise and Achilles. Another
problem, which Boghossian does not point out, is that there is an additional ques-
tion about how possession of the information in (viii) itself bears on my inference.
Possessing the information that I am justified in inferring (3) is one thing; inferring
(3) is another. By the conclusion of Boghossian’s argument, I still haven’t inferred (3),
only that I am justified in inferring (3).18
18 Boghossian describes himself as exploring how (iv) might bear on my entitlement to infer (3), not on
my inferring (3). Perhaps, then, this additional problem is off his radar. But it shouldn’t be. Suppose the
inferential internalist has a good story about how (iv) bears on my entitlement to infer (3). There is still the
question: how does my entitlement to infer (3) bear on my inferring (3)? Just as one might possess evidence
for a belief, but believe independently of it, say on the basis of wishful thinking, so one might possess entitle-
ment for an inference, but infer independently of it, and so without transmitting justification from premises
to conclusion. Again, see Feldman and Conee (1985) for discussion of justified belief.
20 Elijah Chudnoff
The main premise in Boghossian’s case against inferential internalism is this:

Main Premise: In order to make an inference from some premises to a conclusion
because of one’s intuition that those premises support that conclusion, one must take
the claim that those premises support that conclusion as a premise in an inference.
Might the inferential internalist simply deny the Main Premise, and thereby deflect
Boghossian’s criticism? While I do think that the inferential internalist ought to deny
the Main Premise, I do not think that doing so in itself constitutes an adequate response
to Boghossian’s criticism. The Main Premise is plausible. I think it is unavoidable given
a certain assumption about the nature of the intuition one is, according to inferential
internalists, supposed to have of the relation between premises and conclusion in an
inference. In the next section, I will explore this assumption and how it is bound up
with the Main Premise.
1.6. Intuition in Action

What must intuition be like so that Boghossian’s Main Premise is false? This is the
question that I want to address in this section.
There are two assumptions that I will make. First, inferring is a mental action.19 And
second, the becausal relation between inference and intuition is not merely causal; it is
a rational transition.20 What is a rational transition? I do not have a definition to give.
Suppose you believe that p because it perceptually seems to you that p. This transition
from perception to belief is not merely causal since it can make the belief rational.
Suppose you Φ because you intend to Φ. This transition from intention to action is
not merely causal since, provided the intention is rational, it can make the action
rational. My second assumption is that inferring a conclusion from some premises
because you intuit that the premises support the conclusion is also, like these two tran-
sitions, not merely causal since it can make the inference have the property of being
justification-transmitting. That is, inferences made in light of intuitions that their
premises support their conclusions succeed in transmitting justification you have for
believing their premises to their conclusions.
Let us say that a rational transition from a mental state is direct just in case it does
not consist in taking the content of that mental state as a premise in an inference. With
19 Compare: “In making inferences, a being is ipso facto an agent” Burge (1998). Both Peacocke (2008)
and Gibbons (2009) agree. Strawson (2003) disagrees, and though Mele (2009) does not discuss inference
in particular, he develops a position toward mental action in general that is similar to Strawson’s. Sir Peter
Strawson and Alfred Mele do agree with Tyler Burge, Christopher Peacocke, John Gibbons, and me on this
much: when we make an inference we are doing something for which we are immediately responsible. We
are responsible and this distinguishes inferences from sneezes and hiccups. And this responsibility is imme-
diate in the sense that we are responsible and not just because we are responsible for some upstream cause of
our inference. The assumption that inferences are mental events for which we are immediately responsible is
likely strong enough for my purposes here. I cannot explore the issue in any further detail, however.
20 In this I am in agreement with Brewer (1995). It is worth emphasizing that being not merely causal is
compatible with being causal.
the above assumptions and this stipulation in place, our question can be rephrased this
way: what must intuition be like so that it is possible for there to occur a direct rational
transition from it to a mental action, specifically an inference?
Boghossian considers two paradigms: belief and perception. But neither seems to
provide us with a good model.21 Take belief first: one way to make a rational transi-
tion from a belief is to take it as a premise in an inference. The inference might be
theoretical leading to another belief. Or it might be practical leading to an action—and
perhaps even a mental action. But in neither case is the rational transition direct, for it
consists in taking the belief as a premise in an inference.
Take perception then. One way to make a rational transition from perception is to
take it at face value—i.e. to form the belief that p just because it perceptually seems to
you that p. This leads to a belief, however, not an action, and so not a mental action.
Perhaps there is another way to think of this sort of transition. Suppose taking a percep-
tion at face value is making a judgment, and that making a judgment is a mental action.
Inference, then, might stand to intuition as taking at face value stands to perception: an
inference is the mental act that occurs when you take your intuition that some prem-
ises support some conclusion at face value. While I think that something like this is
correct, the analogy with perception does not help us to see how it can be. When you
take a perception at face value you form a belief that shares some of its content. So, if
inferring were just taking an intuition at face value, then it would result in a belief that
shares some of the intuition’s content, i.e., presumably, a belief that some premises sup-
port some conclusion. But this is not what results from an inference. What results is a
change in the epistemic dependencies among your beliefs: after inferring, you believe
the conclusion inferred, and your belief in it is epistemically dependent on your beliefs
in the premises from which it is inferred.
Let us consider one other rational role perception might play. While walking you
might take into account what you perceive in negotiating obstacles, but without, let us
suppose, first forming beliefs about your environment and then taking these beliefs
as premises in practical inferences about how to move. Suppose you step to the side
because you perceive an obstacle. Is this a direct rational transition from perception to
action? Perhaps it is, but, again, it does not provide us with a good model for intuition.
The reason why not is that it is a transition that occurs in the context of a background
activity: you step to the side because you perceive an obstacle while walking. This is not
a case in which a perception alone—without help from other mental states, or a back-
ground activity—gets you walking in the first place.22
21 This claim seems to me to hold only assuming, as I am in the present discussion, orthodox conceptions
of belief and perception on which both only have a mind-to-world direction of fit, about which more later.
22 One might defend the view that intuition works like perception does in the context of a background
activity by arguing that when we make inferences because of what we intuit we do so in the context of a back-
ground activity of thinking, or reasoning, or working our way toward an inference, or something else. This
view seems implausible to me. Sometimes we just make an inference, and this isn’t part of any larger endeavor.
22 Elijah Chudnoff
Reflections like these motivate Boghossian’s Main Premise. There is, moreover, reason
to think that if belief and perception were the only available paradigms on which to model
appreciation then Boghossian’s Main Premise would be compelling. Let us see why.
Perception and belief have mind-to-world—as opposed to world-to-mind—direc-
tion of fit. The difference is illustrated by a famous example from Anscombe (1957,
p. 56). A man is shopping around town getting the items on a list that his wife gave
him. A detective is following him making a list of all the items that he purchases. Let us
suppose that both man and detective have done their jobs well, so that their lists read
the same. The man’s list has items-to-list direction of fit: the items on the list are given
and the list directs the man to purchase those items. It has a directive function. The
detective’s list has list-to-items direction of fit: the items purchased are given and the
list describes which items have been purchased. It has a descriptive function. Similarly,
some mental states, such as beliefs and perceptions, have mind-to-world direction of
fit. The world is given and they function to describe it. Other mental states, such as
desires and intentions, have world-to-mind direction of fit. Their contents are given
and they function to direct their subjects to satisfy those contents.
I have picked out the two different directions of fit by their association with two dif-
ferent functional roles. One might wonder whether (i) a state has its direction of fit in
virtue of its functional role, (ii) a state has its functional role in virtue of its direction
of fit, (iii) a state’s functional role is identical to, or includes as a part, its direction of
fit, or (iv) a state has its direction of fit and its functional role in virtue of other facts
about it, which facts ensure that the direction of fit and functional role line up in the
way I have indicated. My approach here will be to remain neutral on this issue. For my
purposes what matters is that directions of fit and functional roles line up as I have
indicated: world-to-mind states direct and mind-to-world states describe. What ulti-
mately explains this is an issue I will leave unresolved.23
Above we ran through some considerations that suggested, roughly, that one can-
not directly rationally respond to a belief or a perception with an action. You can take
a belief into account by taking it as a premise in an inference, which inference might
result in action. You can take a perception into account by endorsing it with a belief, or
maybe by relying on it to guide an antecedent activity. But you cannot, it seems, take
such states into account by just acting on them. Why? A natural idea is that it is precisely
because of their direction of fit. Consider, then, the following general principle:
(Inertia) It is impossible to make a direct rational transition from a mental state
with solely mind-to-world direction of fit to an action.24
23 For further discussion see: Humberstone (1992), Velleman (1992), Smith (1994), Millikan (1995), Platts
(1997), Sobel and Copp (2001), Jacobson-Horowitz (2006), and Tenenbaum (2006).
24 If taking a perception at face value is a mental action, then (Inertia) requires qualification for which
perhaps something close to the following would do: (Inertia*) Aside from taking a perception at face value, it
is impossible to make a direct rational transition from a mental state with solely mind-to-world direction of
fit to an action. I will set this complication aside. For first, it isn’t clear that taking an experience at face value
is a mental action. And second, even if it is, this doesn’t affect my discussion, since, as pointed out above, the
transition from appreciation to inference cannot be modeled on taking a perception at face value.
The qualifications “direct” and “rational” are essential. On one natural view of causation, it
is metaphysically possible for anything to cause anything. So it is metaphysically possible
for a belief or a perception to cause an action. But this is compatible with (Inertia) because
(Inertia) is about rational transition not mere causation. Surely beliefs and perceptions
can play some role in rationally guiding action. But again this is compatible with (Inertia)
because (Inertia) is about direct rational transition not rational transition in general.
Many, and likely most, philosophers will find (Inertia) or a nearby principle attrac-
tive.25 The so-called Humean theory of motivation entails it. Here is Michael Smith’s
(1994, p. 92) formulation of that theory’s central tenet:
R at t constitutes a motivating reason of agent A to Φ iff there is some Ψ such
(P1)
that R at t consists of an appropriately related desire of A to Ψ and a belief that
were she to Φ she would Ψ.
Given the plausible assumption that a motivating reason is just a mental state to which
one can directly rationally respond with an action, the only if direction of (P1) entails
(Inertia).26 The Humean theory of motivation is stronger than (Inertia): that is, the
Humean theory entails (Inertia), but (Inertia) does not entail the Humean theory.
So anti-Humeans can accept (Inertia). And, in fact, many do. Many anti-Humeans
defend their view precisely by trying to show how acceptance of (Inertia) is compatible
with rejection of the Humean theory of motivation. Some argue that some beliefs have
world-to-mind direction of fit.27 Others argue that some beliefs, or perceptions, are
inseparable from desires.28 Finally, others argue that a motivating state with a world-
to-mind direction of fit can arise out of, or consist in, the presence of other states that
just have mind-to-world direction of fit.29 What all this suggests is that (Inertia) and
nearby principles are philosophically well-entrenched.
Now we are in a position to demonstrate Boghossian’s Main Premise.
(1) Inference is a mental action; and the becausal relation between inference and
intuition is rational, not merely causal. [Assumptions]
(2) It is impossible to make a direct rational transition from a mental state with
solely mind-to-world direction of fit to an action. [Inertia]
(3) A rational transition from a mental state is direct just in case it does not con-
sist in taking the content of that mental state as a premise in an inference.
[Definition]
(4) Intuition has solely mind-to-world direction of fit. [Premise]
25 I will generally suppress the qualification “or a nearby principle,” taking (Intertia) to stand for itself and
nearby principles.
26 One might argue that a motivating reason need not be a mental state. See Dancy (2003). Even if this
proves correct, it would require only superficial modifications to my discussion here.
27 For discussion, both pro and con, see: Altham (1986), McNaughton (1991), Smith (1994), Little (1997),
Jacobson-Horowitz (2006), and Tenenbaum (2006).
28 For discussion see Nagel (1970), McDowell (1978, 1979), and Dancy (1993, 2003).
29 See Dancy (1993, 2003).
24 Elijah Chudnoff
(5) Boghossian’s Main Premise: In order to make an inference from some premises
to a conclusion because of one’s intuition that those premises support that con-
clusion, one must take the claim that those premises support that conclusion as
a premise in an inference. [From (1), (2), (3), and (4)]
The demonstration is valid. The only question is: are all of its premises true? (1) sets out
plausible background assumptions about inference and the becausal relation between
inference and intuition that I will not call into question. (3) is just a definition. So the
only candidates for rejection are (2) and (4). As pointed out above, (2) is philosophi-
cally well-entrenched, and, it seems to me, for good reason: it is very plausible.
I believe we should give up (4). Intuition does not have solely mind-to-world direc-
tion of fit. One argument in favor of this view is a modus tollens argument that appeals
to inferential internalism. The idea is that if (4) is true, then so is Boghossian’s Main
Premise, and if that is true, then inferential internalism is false, but inferential inter-
nalism is true, so we should reject (4). I find the considerations in favor of inferen-
tial internalism persuasive, so I find this argument persuasive. One might worry that
it is dialectically problematic since it might appear illegitimate to assume inferential
internalism. This worry seems misplaced to me, however. Boghossian recognizes the
force of the considerations in favor of inferential internalism and argues that this force
is overridden by a stronger Carrollian argument against inferential internalism. The
strength of that argument, however, depends on the assumption that intuition has
solely mind-to-world direction of fit. Once we recognize that this assumption is not
mandatory, we have the option of rejecting it. One way to remove this option is to give
positive considerations in favor of thinking that intuition does have solely mind-to-
world direction of fit, so that this is no longer an assumption, but an independently
supported premise. In the absence of such considerations, however, the modus tollens
argument is dialectically legitimate. Still, one might want a reason to reject (4) that is
independent of commitment to inferential internalism.
Another argument in favor of thinking intuition has world-to-mind direction of fit
appeals to the same sorts of considerations one might appeal to in arguing that desire
has world-to-mind direction of fit. Since desire is the paradigm example of a men-
tal state with world-to-mind direction of fit, it is rare to find arguments in favor of
thinking that it does have world-to-mind direction of fit. But if one aimed to give such
an argument, here is how it might go. Reflection on the roles desires play in our lives
suggests that we can directly rationally respond to them with actions, so by (Inertia),
they do not have solely mind-to-world direction of fit. Similarly, one might argue as
follows. Reflection on the roles intuitions play in our lives suggests that we can directly
rationally respond to them with actions—e.g. inferences—so by (Inertia), they do not
have solely mind-to-world direction of fit. One might challenge (Inertia). One might
challenge the claim about what reflection on the roles intuitions play suggests. One
might challenge the reliability of this reflection; that is, perhaps it does suggest the role
I have described, but it is mistaken. None of these options seems plausible to me. I have
already reviewed (Inertia). Reflection on the roles instances of a kind of mental state
play in our lives might be limited and fallible in what it tells us about that mental state,
but it does seem like a reliable source of information about very basic features of those
roles. Finally, reflection does suggest that there doesn’t need to be any intermediate
inference between intuition that the premises in an argument support its conclusion
and inferring that conclusion from those premises.
The foregoing supports the following:
World-to-Mind: in some cases, to intuit that some premises support some conclu-
sion is, at least in part, to be in a mental state that has world-to-mind direction of fit.
If the World-to-Mind thesis is correct, and intuition isn’t like perception or belief—on
orthodox conceptions of these states—then what is it like? To compare intuitions to
desires seems silly.30 A comparison to intentions seems more plausible, but still forced.
A better comparison is to states such as felt commands, demands, and obligations that
can be thought of as mental imperatives.31 This fits with the metaphors expressed in
phrases such as “being moved by the force of reason” or “being compelled by reason.”
Most likely, however, any comparison to other states will be more or less misleading,
since intuitions are sui generis mental states.
In specifying the content with respect to which an intuition has world-to-mind
direction of fit I will use the general notion of direction. To intuit that some prem-
ises support some conclusion, then, is, at least in part, to be directed to do something,
namely to believe that conclusion on the basis of those premises. It is important to read
this claim properly: the direction is to [believe that conclusion on the basis of those
premises], not just to [believe that conclusion]. One of the things we can do is form
beliefs. Another of the things we can do is base some beliefs on others, that is, modify
the epistemic dependencies that hold among our beliefs.32 What I am suggesting, then,
is that in some cases intuitions direct one to do this.
Here, then, is a way to think about the transition from intuition to inference. Take
the modus ponens argument from (1) if today is the 20th, then Martha Argerich is play-
ing today in Carnegie Hall, and (2) today is the 20th, to (3) Martha Argerich is play-
ing today in Carnegie Hall. Suppose you infer (3) from (1) and (2) because you intuit
that (1) and (2) support (3). Your intuition is a mental state that directs you to believe
(3) on the basis of your beliefs (1) and (2). Your inference is the mental action you
30 Some philosophers use “desire” to pick out the general category of mental states with world-to-mind
direction of fit. I am not following this technical usage here.
31 Commands, demands, and obligations are not mental. The mental imperative is the impression a com-
mand, demand, or obligation makes when felt. Maurice Mandelbaum appeals to mental imperatives in
exploring the phenomenology of moral experience; see Mandelbaum (1955). Another area where philoso-
phers have found it useful to appeal to mental imperatives is in work on the nature of pain. Some writers
defend an imperatival theory of pain; for discussion, see Klein (2007, 2010) and Hall (2008).
32 Of course, in some cases in order to do this we also have to form a belief. This is what happens when we
reason our way to a new belief.
26 Elijah Chudnoff
perform when you do what you are directed to do. That is, your inference is the men-
tal action of believing (3) on the basis of (1) and (2). Earlier I considered the idea that
inference stands to intuition as taking at face value stands to perception. This analogy
gave the wrong results. Now we have an improved analogy. Inference stands to intui-
tion as action stands to direction. This analogy gives the correct results, since follow-
ing a direction to believe one thing on the basis of other things does precisely result in
believing one thing on the basis of other things.
The World-to-Mind thesis defuses Boghossian’s Carrollian argument. It doesn’t
defuse the argument by simply removing one way of demonstrating its Main Premise,
for perhaps there are other ways of demonstrating the Main Premise. Rather, the
World-to-Mind thesis defuses Boghossian’s Carrollian argument by providing a con-
ception of intuition that allows us to see why that argument’s Main Premise is false. If
intuition sometimes has a world-to-mind direction of fit, then it is the sort of state to
which one can directly rationally respond with a mental action. The comparison of
intuition with felt commands, demands, and obligations helps to make this clear. Your
intuition directs you to believe some conclusion on the basis of some premises, and
your inference is the mental action in which you do what you are directed to do.
1.7. Objections and Elaborations

The defense of the World-to-Mind thesis developed in the previous section is incom-
plete. In this section I introduce two elaborations in response to two objections.
The first objection derives from Gilbert Harman’s discussion of the difference
between truths about good reasoning and truths about logical consequence (Harman
1986). Suppose S intuits that P supports Q. So S is in a mental state that directs him to
base a belief that Q on his belief that P. This suggests that the only way for S to rationally
respond to his intuition is to believe Q on the basis of P. But—it seems—another thing
S might do is revise his belief in P. That is, his intuition that P supports Q might just as
well lead him to give up P as believe Q on the basis of P.
It will help to set out the worry in the form of an inconsistent triad:
1. If S intuits that P supports Q, then S is in a mental state that directs him to believe
Q on the basis of P. [World-to-Mind]
2. If S is in a mental state that directs him to believe Q on the basis of P, then the
only way for S to rationally respond to this mental state is to believe Q on the
basis of P. [Premise]
3. But it is possible for S to rationally respond to his intuition that P supports Q by
revising his belief that P. [Harman’s Observation]
There are two natural replies to this worry.
The first is to reject (2). Directions in general can be resisted. They throw their
weight in favor of an action, but they do not compel it. This is clear from the possibility
of receiving inconsistent directions—i.e. directions that cannot jointly be satisfied. It

is possible to be directed to Φ, take this direction seriously, be rational, and not Φ. The
reason why is that all things considered you might have most reason to do something
other than Φ. For example, you might receive a more compelling direction to not-Φ.
So (2) is false, and (1) and (3) are compatible with each other.
Another reply is to argue that expressing Harman’s Observation with (3) is making
precisely the confusion that Harman aimed to expose. Distinguish (3) from (3*):
(3*) It is possible for S to rationally respond to his intuition that Q is a logical conse-
quence of P by revising his belief that P.
(3*) expresses Harman’s Observation, but (3) does not. The difference is that (3*) is
about the logical consequence relation and (3) is about what I have been calling the
support relation. So far I have not emphasized this distinction since it did not bear on
the discussion. But now it does. There are many different relations in the vicinity. What
I have in mind in talking about the support relation is the relationship one’s belief that
P has to believing that Q when one has a reason to base a belief that Q on one’s belief
that P. So when S intuits that P supports Q, what S intuits is that S has a reason to base a
belief that Q on S’s belief that P. As Harman points out, this is different from S intuiting
that Q is a consequence of P, since Q might be a consequence of P without S having a
reason to base a belief that Q on S’s belief that P.
Note that (3) might still be true. But if it is true, this is because reasons are defeasible.
And that is the possibility captured by the first reply I made to the worry. What does
not seem possible is for S to rationally respond to an undefeated intuition that P sup-
ports Q—i.e. an undefeated intuition that S has a reason to base a belief that Q on S’s
belief that P—by revising his belief that P. For if it is rational for S to revise his belief
that P, then, all things considered, S no longer has a reason to base a belief that Q on S’s
belief that P.
Now I turn to the second objection. I have set out considerations in favor of thinking
that intuition has world-to-mind direction of fit. But it is obvious that intuition, like
perception, has mind-to-world direction of fit. How can the considerations be squared
with this obvious fact?
My reply has already been foreshadowed in how I formulated the World-to-Mind
thesis: to intuit that some premises support some conclusion is, at least in part, to
be in a mental state that has world-to-mind direction of fit. This leaves open the
possibility that intuition also has mind-to-world direction of fit. And this is the
reply to the present objection that I propose. Intuitions have both mind-to-world
and world-to-mind direction of fit. They are what Millikan calls pushmi-pullyu
representations.
There are reasons to think intuition is not an isolated case. Millikan (1995) men-
tions animal signals, performative utterances, inner representations that animals
use to navigate, intentions, and mental representations of social norms as exam-
ple pushmi-pullyu representations. Some philosophers defend the view that moral
28 Elijah Chudnoff
judgments are pushmi-pullyu representations.33 And Timothy Bayne (2010) has

recently argued that experiences of agency, such as the feeling of performing some
action, are pushmi-pullyu representations. Any given case is bound to raise its own
controversies. But the category is defensible, and if there are both considerations in
favor of assigning a representation a mind-to-world and a world-to-mind direction of
fit, then these considerations jointly constitute a reason to assign it both. This seems to
me to be the case with at least some intuitions.
1.8. The Ground of Intuitive Guidance

The eighteenth-century moral sense theorist, Francis Hutcheson, skeptically asked his
rationalist contemporaries “What is this conformity of actions to reason?” John Balguy
answered (in Raphael 1991 [1969], p. 455):
. . . [the] question amounts plainly to this: what does a reasonable creature propose in acting rea-
sonably? Or what is it that induces his will to take council of his understanding? As if this were
not the very essence of a rational action!
The question I will pursue in this section—what is it in virtue of which an intuition can
play an action-guiding rational role?—is modeled on Hutcheson’s. The answer I will
propose is modeled on Balguy’s.
The view that I will defend is this: intuitions play action-guiding rational roles in
virtue of playing justifying rational roles. Even though action-guiding roles are dis-
tinct from justifying roles, it can be, and I am inclined to think that it is, true that what
explains why intuitions can play action-guiding roles is that they play justifying roles.
Consider Smith. In virtue of what does his intuition that (A1)—that Connie and Cyndi
are a cone and a cylinder with the same base and height—supports believing that
(A2)—that Cyndi encloses a greater volume than Connie—guide his inference? The
answer, in my view, is that it is because his intuition justifies him in believing that (A1)
supports (A2).
The main reason for this view derives from the observation about rationality that
Balguy seems to me to have in mind. In expanding on Balguy’s claim, it will be helpful
to have a simpler claim to serve as a model. Consider the following:
In a state with the appropriate laws, parking next to a fire hydrant constitutes park-
ing illegally.
The above seems straightforward enough, but there are three observations worth
emphasizing. First, “In a state with the appropriate laws,” is essential. Without the laws,
parking next to a fire hydrant might not be illegal. Second, nothing needs to happen
in addition to parking next to a fire hydrant in order for you to park illegally: given
33 This case has received the most discussion. See, for example, Altham (1986), McNaughton (1991), Smith
(1994), Little (1997), Jacobson-Horowitz (2006), and Tenenbaum (2006).
the laws, parking next to a fire hydrant counts as parking illegally. This is the point of
invoking the notion of constitution. Third, if your car is parked next to the fire hydrant
then it is parked illegally, but it is still also parked next to the fire hydrant. It is not as if
the one property takes the place of the other.
As I understand Balguy, he is making a similar point about rationality, grasping rea-
sons, and guidance of the will. Here is how I would put it:
In a rational person, a mental state—e.g. an intuition—that justifies believing that
one has a reason to Φ constitutes a mental state that guides one to Φ.
Here are three observations analogous to those made above, though in reverse order.
I will make them in relation to Smith and his inference of (A2) from (A1). First, if
Smith’s intuition that (A1) supports (A2) justifies him in believing that (A1) supports
(A2), then it guides his inferring (A2) from (A1), but it still also justifies him in believ-
ing that (A1) supports (A2). Just as with the car, it is not as if the one property takes the
place of the other. So one intuition can play both the justifying and the action-guiding
role. Second, nothing needs to happen in addition to Smith’s intuition justifying him in
believing that (A1) supports (A2) for it to guide his inferring (A2) from (A1). In partic-
ular, Smith does not need to go through any extra bit of reasoning, so there is no threat
of a Carrollian regress reemerging at this point. Third, “In a rational person,” plays a
role similar to that played by “In a state with the appropriate laws.” Part of what it is for
a state to have the appropriate laws is for it to be a state in which parking next to a fire
hydrant constitutes parking illegally. Similarly, one way to capture a popular idea about
rationality is to say that part of what it is for a person to be rational is to be a person for
whom mental states that play justifying roles with respect to reasons constitute mental
states that play action-guiding roles.34 One might worry about pressing the analogy too
far. The state has laws and it is in virtue of these that parking next to a fire hydrant con-
stitutes parking illegally. Maybe there are laws of rationality and it is in virtue of these
that mental states that play justifying roles with respect to propositions about reasons
constitute mental states that play action-guiding roles. But perhaps rationality cannot
be articulated by a set of laws. All that matters for my purposes here is that whatever
the nature of rationality turns out to be, that nature will make it so that insofar as a
person is rational, mental states that play justifying roles with respect to propositions
about reasons constitute mental states that play action-guiding roles for that person.
This view of the ground of intuitive guidance provides a basis for responding to an
objection to the view that intuitions are pushmi-pullyu representations. The objection
34 Compare: “a rational person who judges there to be compelling reason to do A normally forms the
intention to do A, and this judgment is sufficient explanation of that intention and of the agent’s acting on it
(since this action is part of what such an intention involves). There is no need to invoke an additional form
of motivation beyond the judgment and the reasons it recognizes, some further force to, as it were, get the
limbs in motion” (Scanlon 1998, pp. 33–4). And: “For me to be a theoretically rational person is not merely
for me to be capable of performing logical and inductive operations, but for me to be appropriately con-
vinced by them: my conviction in the premises must carry through, so to speak to a conviction in the conclu-
sion” (Korsgaard 1986 [1997], p. 14).
30 Elijah Chudnoff
can be put like this. Suppose S’s intuition that P supports Q both directs S to believe Q
on the basis of P, and presents it as true that P supports Q. If there are these two parts
of the intuition—the directive part and the descriptive part—then there must be some
story about how an inference that is made in response to the intuition relates to both
parts. If it is just a response to the directive part, then it is just as if the inference is made
in ignorance of the information that P supports Q. If it is just a response to the descrip-
tive part, then it is a violation of (Inertia). If it is a response to some inferential integra-
tion of the two parts, then it will lead to a Carrollian regress. In brief, introducing the
second direction of fit seems to have put us back in square one.
The reply to this objection is that there is a fourth option. The first two options seem
clearly mistaken to me. In my view when S infers Q from P because he intuits that P
supports Q his inference is a response to the intuition as a whole, not to one or another
part of it. This raises the question, however: how are the two parts fused together into
a whole to which S might respond? The third option described above suggests it is by
inferential integration—i.e. as two parts of an inference. But this is also clearly mis-
taken since it leads to Carrollian regress. What is needed is an account of how the two
parts are fused together into a whole that does not appeal to their inferential integra-
tion. The account of intuitive guidance suggests such an account: the descriptive part
constitutes the directive part.
Another objection to the view that intuitions play action-guiding roles in virtue of
playing justifying roles is that it renders intuitions dispensable as guides to action. I’ve
been arguing that intuitions play action-guiding roles that cannot be assimilated to,
even if they are explained by, their justifying roles. But, one might wonder now, what
was the point? Suppose intuitions could play action-guiding roles in virtue of justify-
ing beliefs about reasons for action. Then they needn’t, since all guidance might derive
from the beliefs so justified. All intuitions need to do is justify beliefs, and then these
beliefs guide our actions.
Two initial points. First, note it simply does not follow from the claim that intui-
tions guide action in virtue of justifying beliefs about reasons for action that beliefs
about reasons for action can themselves guide action. Humeans could very well be
right about the motivational inefficacy of belief. While I think there is something to
this idea, I will not take it on as a commitment. Second, one might concede that intui-
tions are dispensable as guides to action, but still think they do in fact guide action, and
so think it is worth shedding light on the matter. It is not as if every worthy object of
investigation must be indispensible. That said, let’s see if intuitions really are dispensa-
ble as guides to action.
It seems to me that between the two—beliefs and intuitions—beliefs are more dis-
pensable than intuitions as guides to action. Contrast the following two claims. The
first is my gloss on Balguy: in a rational person, a mental state—e.g. an intuition—
that justifies believing that one has a reason to Φ constitutes a mental state that guides
one to Φ. The second is a similar-looking alternative: in a rational person, a mental
state—e.g. a belief—that represents that one has a reason Φ to constitutes a mental
state that guides one to Φ. I think we should be less confident in this alternative claim.
It might be that consistency requires acting in accordance with what you believe your
reasons for action are. But suppose these beliefs are unjustified. Then insofar as you are
rational, you should give them up, not act in accordance with them. So it seems that
if beliefs play action-guiding roles, this depends on their being justified by intuitions.
On the other hand, if intuitions about reasons for action play action-guiding roles,
their ability to do so does not depend on their leading to beliefs. You can act in light
of the justification you have for believing that you have a reason to act without form-
ing the belief that you have that reason to act. In fact, this is likely the norm. Beliefs
are mental states with a certain degree of permanency. Most of our actions have little
significance beyond the moment of their occurrence. It would be silly to form standing
beliefs about what reasons you have for all the actions you perform in life.35 So it seems
that if intuitions play action-guiding roles their doing so does not depend on their
leading to beliefs. There is reason to think, then, that beliefs are more dispensable than
intuitions as guides to action.
References
Altham, J. (1986). “The Legacy of Emotivism,” in G. Macdonald and C. Wright (eds), Fact,
Science, and Morality. Oxford: Blackwell, pp. 275–88.
Anscombe, G. E. M. (1957). Intention. Cambridge, MA: Harvard University Press.
Armstrong, D. M. (1968). A Materialist Theory of the Mind. London: Routledge.
Audi, R. (1994). “Dispositional Beliefs and Dispositions to Believe,” Noûs 28, pp. 419–34.
——. (1997). Moral Knowledge and Ethical Character. New York: Oxford University Press.
——. (2005). The Good in the Right: A Theory of Intuition and Intrinsic Value. Princeton,
NJ: Princeton University Press.
Bayne, T. (2010). “Agentive Experiences as Pushmi-Pullyu Representations,” in J. Aguilar, A.
Buckareff and K. Frankish (eds), New Waves in the Philosophy of Action. Basingstoke: Palgrave
Macmillan, pp. 219–36.
Bealer, G. (1998a). “A Theory of Concepts and Concept Possession,” Philosophical Issues 9, pp.
261–301.
——. (1998b). “Intuition and the Autonomy of Philosophy,” in M. R. DePaul and W. Ramsey
(eds), Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry.
Lanham, MD: Rowman & Littlefield, pp. 201–39.
——. (1999). “A Theory of the A Priori,” Pacific Philosophical Quarterly 81, pp. 1–30.
Boghossian, P. (2003). “Blind Reasoning,” Proceedings of the Aristotelian Society, Supplementary
Volume 77, pp. 225–48; reprinted (2008) in Content and Justification: Philosophical Papers.
New York: Oxford University Press, pp. 267–89.
BonJour, L. (1998). In Defense of Pure Reason: A Rationalist Account of A Priori Justification.
Cambridge and New York: Cambridge University Press.
35 In terms of Audi’s distinction between dispositional beliefs and dispositions to believe one might say
that in many cases the relevant intuitions generate dispositions to believe, though not dispositional beliefs.
See Audi (1994).
32 Elijah Chudnoff
BonJour, L. (2005). “In Defense of the A Priori,” in M. Steup and E. Sosa (eds), Contemporary
Debates in Epistemology. Oxford: Blackwell, pp. 98–104.
Brewer, B. (1995). “Mental Causation: Compulsion by Reason,” Proceedings of the Aristotelian
Society, Supplementary Volume 69, pp. 237–53.
Burge, T. (1998). “Reason and the First Person,” in B. S. C. Wright and C. MacDonald (eds),
Knowing Our Own Minds. Oxford: Oxford University Press, pp. 243–70.
Butchvarov, P. (1970). The Concept of Knowledge. Evanston: Northwestern University Press.
Carroll, L. (1995 [1905]). “What the Tortoise Said to Achilles,” Mind 104, pp. 278–80.
Cath, Y. (2012). “Evidence and Intuition,” Episteme 9, pp. 311–28.
Chudnoff, E. (2011a). “The Nature of Intuitive Justification,” Philosophical Studies 153, pp.
313–33.
——. (2011b). “What Intuitions Are Like,” Philosophy and Phenomenological Research 82, pp.
625–54.
——. (2012). “Presentational Phenomenology,” in S. Miguens and G. Preyer (eds), Consciousness
and Subjectivity. Piscataway, NJ: Transaction Books, pp. 51–73.
——. (2013a). “Awareness of Abstract Objects,” Noûs 47, pp. 706–26.
——. (2013b). “Intuitive Knowledge,” Philosophical Studies 162, pp. 359–78.
Cleve, J. van (1984). “Reliability, Justification, and the Problem of Induction,” Midwest Studies In
Philosophy 9, pp. 555–67.
Crane, T, (2005). “The Problem of Perception,” in E. N. Zalta (ed.), Stanford Encyclopedia of
Philosophy, available at: <http://plato.stanford.edu/entries/perception-problem/> [last
accessed 11 February, 2014].
Crisp, R. (2008). Reasons and the Good. New York: Oxford University Press.
Cullity, G., and Gaut, B. (eds) (1998), Ethics and Practical Reason. Oxford: Oxford
University Press.
Dancy, J. (1993). Moral Reasons. Oxford: Blackwell.
——. (2003). Practical Reality. New York: Oxford University Press.
Darwall, S. (1997). “Reasons, Motives, and the Demands of Morality,” in S. Darwall, A.
Gibbard, and P. Railton (eds), Moral Discourse and Practice: Some Philosophical Approaches.
New York: Oxford University Press, pp. 305–12.
——. (2002). “Ethical Intuitionism and the Motivation Problem,” in P. Stratton-Lake (ed.),
Ethical Intuitionism: Re-evaluations. Oxford: Oxford University Press.
Darwall, S., Gibbard, A., and Railton P. (eds) (1997). Moral Discourse and Practice: Some
Philosophical Approaches. Oxford: Oxford University Press.
DePaul, M., and Ramsey, W. (1998). Rethinking Intuition: The Psychology of Intuition and its Role
in Philosophical Inquiry. Lanham, MD: Rowman & Littlefield.
Descartes, R. (1985). The Philosophical Writings of Descartes: Volume 1, trans. J. Cottingham, R.
Stoothoff, and D. Murdoch. Cambridge: Cambridge University Press.
Dreier, J. (1997). “Humean Doubts about the Practical Justification of Morality,” in G. Cullity
and B. Gaut (eds), Ethics and Practical Reason. Oxford: Oxford University Press, pp.
81–100.
——. (2006). Contemporary Debates in Moral Theory. Oxford: Blackwell.
Earlenbaugh, J., and Molyneux, B. (2009a). “Intuitions are Inclinations to Believe,” Philosophical
Studies 145, pp. 89–109.
——. (2009b). “If Intuitions Must Be Evidential then Philosophy is in Big Trouble,” Studia
Philosophica Estonica 2, pp. 35–53.
Evans, G. (1982). The Varieties of Reference. Oxford: Oxford University Press.

Ewing, A. C. (1968 [1941]). “Reason and Intuition,” in Non-Linguistic Philosophy.
London: Routledge. Reprinted from Proceedings of the British Academy 27.
Feldman, R., and Conee E. (1985). “Evidentialism,” Philosophical Studies 48, pp. 15–34.
Foster, J. A. (2000). The Nature of Perception. New York: Oxford University Press.
Fumerton, R. (2006). Epistemology. Oxford: Blackwell.
Gibbons, J. (2009). “Reason in Action,” in L. O’Brien and M. Soteriou (eds), Mental Actions.
Oxford: Oxford Scholarship online DOI: 10.1093/acprof:oso/9780199225989.001.0001.
Glüer, K. (2009). “In Defence of a Doxastic Account of Experience,” Mind and Language 24, pp.
297–327.
Gödel, K. (2001). Collected Works: Volume III: Unpublished Essays and Lectures.
New York: Oxford University Press.
Goldman, A. I. (2007). “Philosophical Intuitions: Their Target, Their Source, and Their
Epistemic Status,” Grazer Philosophische Studien 74, pp. 1–26.
Hall, R. J. (2008). “If It Itches, Scratch!,” Australasian Journal of Philosophy 86, pp. 525–35.
Harman, G. (1986). Change in View. Cambridge, MA: MIT Press.
Hooker, B. (2002). “Intuitions and Moral Theorizing,” in P. Stratton-Lake, Ethical
Intuitionism: Re-evaluations. Oxford: Oxford University Press.
Huemer, M. (2001). Skepticism and the Veil of Perception. Lanham, MD: Rowman & Littlefield.
——. (2005). Ethical Intuitionism. New York: Palgrave Macmillan.
Humberstone, I. L. (1992). “Direction of Fit,” Mind 101, pp. 59–83.
Husserl, E. (1975). Experience and Judgment, 1st ed. Evanston, IL: Northwestern University Press.
Ichikawa, J. J. (2001). Logical Investigations, Volume 2, rev. ed. London: Routledge.
——. (2013). “Who Needs Intuitions? Two Experimentalist Critiques,” in A. Booth and D.
Rowbottom (eds), Intuitions. Oxford: Oxford University Press.
Jacobson-Horowitz, H. (2006). “Motivational Cognitivism and the Argument from Direction of
Fit,” Philosophical Studies 127, pp. 561–80.
Jackson, F. (1977 [2009]). Perception: A Representative Theory. Cambridge and
New York: Cambridge University Press Archive.
Johnston, M. (2006). “Better than Mere Knowledge? The Function of Sensory Awareness,” in T.
S. Gendler and J. Hawthorne (eds), Perceptual Experience. Oxford: Oxford University Press,
pp. 260–90.
Klein, C. (2007). “An Imperative Theory of Pain,” Journal of Philosophy 104, pp. 517–32.
——. (2010). “Response to Tumulty on Pain and Imperatives,” Journal of Philosophy 107,
pp. 554–7.
Korsgaard, C. (1986 [1997]). “Skepticism about Practical Reason,” reprinted in S. Darwall, A.
Gibbard, and P. A. Railton, Moral Discourse and Practice. New York: Oxford University Press.
Leite, A. (2008). “Believing One’s Reasons are Good,” Synthese 161, pp. 419–41.
Lévinas, E. (1995). The Theory of Intuition in Husserl’s Phenomenology. Evanston,
IL: Northwestern University Press.
Liao, S. M. (2008). “A Defense of Intuitions,” Philosophical Studies 140, pp. 247–62.
Little, M. O. (1995). “Seeing and Caring: The Role of Affect in Feminist Moral Epistemology,”
Hypatia 10, pp. 117–37.
——. (1997). “Virtue as Knowledge: Objections from the Philosophy of Mind,” Noûs 31,
pp. 59–79.
34 Elijah Chudnoff
Mandelbaum, M. (1955). Phenomenology of Moral Experience. Baltimore, MD: Johns Hopkins

University Press.
McDowell, J. (1978). “Are Moral Requirements Hypothetical Imperatives?,” Proceedings of the
Aristotelian Society, Supplementary Volume 52, pp. 13–29.
——. (1979). “Virtue and Reason,” The Monist 62, pp. 331–50.
——. (1982). “Criteria, Defeasibility, and Knowledge,” Proceedings of the British Academy 68, pp.
455–79.
McNaughton, D. (1991). Moral Vision: An Introduction to Ethics. Oxford: Blackwell.
Mele, A. (2009). “Mental Action: A Case Study,” in L. O’Brien and M. Soteriou
(eds), Mental Actions. Oxford: Oxford Scholarship online DOI: 10.1093/acprof:
oso/9780199225989.001.0001.
Millikan, R. (1995). “Pushmi-Pullyu Representations,” Philosophical Perspectives 9, pp. 185–200.
Nagel, T. (1970). The Possibility of Altruism. Oxford: Oxford University Press.
O’Brien, L., and Soteriou, M. (2009). Mental Actions. Oxford: Oxford Scholarship online
[DOI: 10.1093/acprof:oso/9780199225989.001.0001].
O’Shaughnessy, B. (ed.) (2003). Consciousness and the World. New York and Oxford: Oxford
University Press.
Parsons, C. (1980). “Mathematical Intuition,” Proceedings of the Aristotelian Society 80, pp.
145–68.
——. (2007). Mathematical Thought and Its Objects. New York: Cambridge University Press.
Peacocke, C. (1983). Sense and Content: Experience, Thought, and Their Relations. Oxford: Oxford
University Press.
——. (2003). The Realm of Reason. New York: Oxford University Press.
——. (2008). Truly Understood. Oxford: Oxford University Press.
Platts, M. (1997). Ways of Meaning: An Introduction to a Philosophy of Language, 2nd ed.
Cambridge, MA: MIT Press.
Pollock, J. (1974). Knowledge and Justification. Princeton, NJ: Princeton University Press.
Pust, J. (2000). Intuitions as Evidence, 1st ed. London: Routledge.
Railton, P. (1997). “On the Hypothetical and Non-Hypothetical in Reasoning about Belief
and Action,” in G. Cullity and B. Gaut (eds), Ethics and Practical Reason. Oxford: Oxford
University Press .
Raphael, D. D. (ed.) (1991 [1969]). British Moralists 1650–1800, 2 vols. Oxford: Oxford
University Press.
Ross, D. (2003). The Right and the Good, 2nd ed. New York: Oxford University Press.
Scanlon, T. M. (1998). What We Owe to Each Other. Cambridge, MA: Harvard
University Press.
Searle, J. R. (1983). Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge
University Press.
Sidgwick, H. (1981). Methods of Ethics, 7th ed. Cambridge, MA: Hackett Publishing Company.
Smith, M. (1994). The Moral Problem. Oxford: Blackwell.
Smith, Q. (2009). Epistemology: New Essays. New York: Oxford University Press.
Sosa, E. (2007). “Intuitions: Their Nature and Epistemic Efficacy,” Grazer Philosophische Studien
74, 1, pp. 51–67.
——. (2009). A Virtue Epistemology: Apt Belief and Reflective Knowledge, Volume II.
Stratton-Lake, P. (2002). Ethical Intuitionism: Re-evaluations. Oxford: Oxford

University Press.
Strawson, G. (ed.) (2003). “Mental Ballistics: The Involuntariness of Spontaneity,” in Real
Materialism and Other Essays. New York: Oxford Scholarship online [DOI:10.1093/acprof:
oso/9780199267422.003.0010].
——. (ed.) (2008). Real Materialism and Other Essays. New York: Oxford Scholarship online
[DOI:10.1093/acprof:oso/9780199267422.003.0010].
Strawson, P. F. (2008 [1974]). Freedom and Resentment and Other Essays, 2nd ed.
London: Routledge.
Sturgeon, S. (2000). Matters of Mind. London: Routledge.
Svavarsdottir, S. (1999). “Moral Cognitivism and Motivation,” Philosophical Review 108, pp.
161–219.
Tenenbaum, S. (2006). “Direction of Fit and Motivational Cognitivism,” in R. Shafer-Landau
(ed.), Oxford Studies in Metaethics, vol. 3. Oxford: Oxford University Press.
Tieszen, R. (1989). Mathematical Intuition: Phenomenology and Mathematical Knowledge.
New York: Springer.
——. (2005). Phenomenology, Logic, and the Philosophy of Mathematics. Cambridge: Cambridge
University Press.
Velleman, J. D. (1992). “The Guise of the Good,” Noûs 26, pp. 3–26.
Villanueva, E. (ed.) (1998). “Preface,” in Concepts. Atascadero, CA: Ridgeview Publishing.
Wallace, R. J. (2006). “Moral Motivation,” in J. Dreier (ed.), Contemporary Debates in Moral
Theory. Oxford: Blackwell, pp. 182–96.
Williamson, T. (2004). “Philosphical ‘Intuitions’ and Scepticism about Judgement,” Dialectica
58, pp. 109–53.
——. (2007). The Philosophy of Philosophy. Malden, MA and Oxford: Wiley-Blackwell.
Wright, C., Smith, B. C., and Macdonald, C. (eds) (2000). Knowing Our Own Minds.
2
Intuitions: Their Nature and
Probative Value
Ernest Sosa
2.1. Introduction
What follows will first take up the nature of intuitions in general. Then we ask which of
these are rational. If only some are, finally, what distinguishes them?
Intuition is traditionally modeled on perception, particularly on vision. We begin by
exploring this comparison. Visual knowledge and visually justified or competent belief
depend on visual experience, or so they do on the account assumed here. Does any-
thing play for intuition the epistemic role played by sensory experience for perception?
2.2. What Is an Intuition? Seemings and

Experiences
You see a foreign word clearly displayed arm’s-length away in bright light. Say it’s the
Spanish word desplegar (to display). The array that gives content to your experience
has a certain definite character; it is a specific arrangement of curved and straight lines.
Yet in your first encounter with that complex shape you may well not be attracted to
assent to the proposition concerning a certain phenomenally distinctive shape, that
the item encountered displays that shape. Of course you can point to the shape and
indicate it. But you lack even so much as a phenomenal discriminating concept of that
shape under which you could detect its instances and recognize them. A fluent enough
Spanish speaker, by contrast, does have such a phenomenal concept and is able to see
that the array of shapes before him does display that complex shape. The experience
must therefore be distinguished from the seeming. In Spanish speakers the experi-
ence prompts the seeming, and even the corresponding assent, but in most others it
does not. Nor does this distinction require a sense-datum view. What follows in fact
assumes a view of sensory experience as having propositional content, as a sort of
propositional attitude, even when this attitude is not linguistically expressible.
Intuitions: Their Nature and Probative Value 37
Experiencing sensorily as if p is anyhow distinct from its seeming to one as if p (i.e.,

from one’s being intellectually attracted to assent accordingly). The two states (state
types) are constitutively distinct. How similar are they epistemically?
In our example, someone fluent in a certain language can share a visual experi-
ence with someone ignorant of that language when they both view a word of it in
good light. They both would see a certain array of straight and curved lines. This
experience is as passive as is an experience of pain. One’s visual experience of that
bit of syntax cannot be changed at will, nor be much affected by direct choice. That’s
how it is for the visual sensory experience of the complex arrangement of lines. Also
passive is the seeming prompted in a fluent speaker. Neither state is modifiable at
will, by direct choice. Once he considers the question, the fluent speaker is bound to
suffer the prompted attraction or seeming, as it seems to him that the relevant word
of his language is there displayed before him. He has no choice as to how that seems
to him.
Nevertheless, there is the following difference. The seeming is rationally based; the
experience is not. The seeming manifests the subject’s rational agency; the experience
does not, any more than does a pain. Someone with a headache is not in pain moti-
vated by a reason, but he is attracted to assent by a reason. Is the pain a reason why,
but not a reason for which? I doubt it. When I am aware that my headache is starting,
I have a reason for so believing, namely, that it is indeed starting. I am responding
to my experience with an appropriate belief, and there is a reason within my con-
sciousness why I so believe, and for which I so believe.1 That stands in contrast to the
neurological reasons why I so believe without believing for those reasons. It is surely
relevant that I can make my reason explicit in the first case, without special empiri-
cal inquiry, but not in the second case, that of the neurological causes. I can avow my
reason for thinking I’m in pain by saying that I so believe because I am in pain. Not so
in the second case.2
Factive reasons are distinct from stative reasons. Pains are perhaps only factive rea-
sons, not stative. If so, then pains are not themselves mental states on the rational basis
of which one might form beliefs: they are hence not stative reasons. But they can still be
factive reasons: i.e., facts (or truth-makers for such facts) that can be taken in through
beliefs that do then constitute stative reasons. Suppose you say that your pain (or, bet-
ter, the fact that you are in such pain) is a reason for your belief that you are in pain.
On our present view, this would be like saying that a certain speedometer reading (or,
better, the fact that the speedometer reads as it does) is my reason for believing my
1 Thus if I am asked why I think I have a headache (and I bother to answer such a silly question) I can cor-
rectly say “because I do,” which is not like answering by describing some neurological condition. Rather, it
explains a rational basis that I have for so thinking, a reason for which I think as I do, and not just a reason
why I so think, that provides a merely causal, neurological explanation. The remainder of this chapter will
occasionally rely crucially on such a notion of rational basing or rational motivation.
2 Pains have only one foot in the space of reasons: they constitute reasons but are not themselves based on
reasons.
38 Ernest Sosa
speed to be such and such. Of course the speedometer reading is not a stative reason,
since for one thing it is not a mental state. It is, rather, a factive reason. Pains too might
be reasons only in this factive way, even though they are mental states. Similarly, sen-
sory experiences might be reasons only in this factive way, even if they too are mental
states. Once we view sensory experiences as mental states with propositional content,
however, it does become tempting to consider them stative reasons, and not just factive
reasons. Let’s explore this issue a bit further.
We are comparing four entities in respect of how they can function as reasons for a
subject: first, a red surface that he sees; second, a pain that he undergoes; third, a sen-
sory experience that he hosts, as of a red surface; fourth, an intellectual seeming: e.g.,
the subject’s inclination or attraction to think he sees a red surface. Each of these can
of course function as a factive reason (or as the truth-maker for a factive reason).
Moreover, the intellectual seeming can also function as a stative reason, given its role
in the subject’s cognitive dynamics, its bearing as he deliberates on the question as to
what he then sees. Even more clearly, the red surface cannot function as a stative reason
for that subject, since it is no mental state of his, or of anyone’s. What of the other two
entities? I mean the pain, and the visual experience.
With regard to these two it is not so clear what to say. The pain is of course a men-
tal state of that subject. But it has no propositional content.3 Arguably, then, it can-
not function as a motivating reason, as a reason for which the subject believes he is
in pain. Such a reason must be appropriately related to an argument that the subject
could rehearse. What is this appropriate relation? What is required is that the states in
question have a propositional content that constitutes a premise, or at least a step, in
such an argument, whose conclusion is the proposition for which the stative reason is
a reason: i.e., the proposition that one might properly believe with the stative reason as
one’s motivating reason. On this conception pains obviously cannot function as stative
reasons, for lack of propositional content.
That however gives no reason to deny that a sensory experience (as opposed to a
pain) might function as a stative reason. As we are conceiving of them, sensory expe-
riences do have propositional content. Thus the sensory experience that our subject
undergoes, as of a red surface, has the propositional content that he then sees a red sur-
face, or faces a red surface, or some such. Unlike the pain, then, the sensory experience
can bear the required relation to an argument whose conclusion is the proposition for
which (or for believing which) the stative reason is a reason.
On the other hand, consider experiences generally, whether with or without propo-
sitional content. When an experience helps to rationally justify a corresponding belief
(a believing by the subject, at a given time) it must function as a rational basis. But this
it can do only if the subject responds systematically to such experience with such belief,
3 Or so I assume here for the sake of argument, although it is uncertain that this assumption is correct.
If pains do all of them have propositional content, then such states as feeling drowsy, or nauseous, etc., may
serve us better.
or at least with a corresponding seeming (an attraction to believe). If the subject does
respond systematically that way, however, then it becomes plausible that at some level
he can tell when an experience of that sort is present and that at some level he has a
belief—or a commitment, which need not be linguistically formulated, or even formu-
lable, by that subject—that an experience of that sort is positively relevant to the truth
of the belief that he forms, or the attraction that he experiences, based on an instance
of that experience. This way of thinking may suggest that experiences, even those with
propositional content, need not be viewed as stative reasons. They can remain mere
factive reasons, along with red surfaces. They enter our rational dynamics only when
registered by a subject, who forms a corresponding belief, perhaps one that remains
implicit.
That suggestion is in doubt as soon as we focus on such “registering” of an experi-
ence, the registering required for the triggering of the relevant modus ponens reason-
ing. I mean the reasoning that proceeds via the implicit conditional belief that if one
is experiencing thus then such and such is the case. The problem is that such implicit
reasoning will justify its conclusion only if its premises are justified. Since one of the
premises enters the subject’s cognitive dynamics only by being thus “registered,” this
registering will need to be justified. But we cannot explain how it is justified by appeal
to some further modus ponens reasoning, involving a prior “registering” of the experi-
ence in question, on pain of a vicious regress. It seems inevitable that we must recog-
nize the status of some experiences as stative reasons, and not just as factive reasons.
This includes some experiences with no propositional content, such as pains.
That all being so, propositional sensory experiences plausibly function not only as
factive reasons, but also as stative reasons. For our purposes, in any case, it does not
much matter whether they do so or not. Either way, propositional sensory experiences
are not themselves motivated by reasons for which they are hosted. Intellectual seem-
ings, by contrast, are often thus motivated. For example, its seeming to me that I see
(or face) a red surface is rationally motivated by my visual experience as of seeing (or
facing) a red surface.
Sensory experiences are not themselves based on motivating reasons, unlike intel-
lectual seemings. Experiences are distinct from seemings not only constitutively but
also functionally. We return to this later, but first we take a closer look at seemings.4
2.3. Deliberating and Pondering

What is it to ponder a factual question, or to deliberate on a practical choice?
You ponder or deliberate by weighing reasons in an attempt to reach a decision. In
yes/no cases the overall balance may favor one side enough that you affirm or choose
4 Apart from its interest in epistemology, this distinction between seemings and experiences provides a
promising framework for understanding duck-rabbit and other aspect switches, where something remains
the same (the experience) while something shifts (the seeming).
40 Ernest Sosa
accordingly. And if the process has been adequate, that decision will be justified. What
is it to weigh reasons, to assign a weight on the scales of judgment?
Our talk of assigning weights is of course metaphorical. We might equally speak of
attraction or repulsion.5 We assign greater weight to one consideration than to another
if and only if it attracts us more. Even if a resultant attraction favors one side, moreover,
this may not decide the matter. A resultant attraction will yield a decision only when its
intensity rises above a certain threshold.
Suppose the weighing—the pondering or deliberation—is conscious. In that case,
conscious attractions will take their place alongside conscious desires and beliefs. If
a friend you trust tells you something, this attracts you to believe what he says, but if
a second friend tells you the opposite, this attracts you to believe the opposite. If one
friend is more trusted, his testimony will have greater weight; it will attract more. But
the resultant attraction may or may not suffice to determine belief. Even if it overcomes
the opposing attraction, it may remain only an inclination, without determining out-
right belief.6
I take that to be a familiar situation, of a sort we all face daily. There are many other
instances. Even when you believe in the sincerity of an interlocutor, for example, shifty
eyes and fidgety fingers may insinuate doubt. Here emphatic assertion may betoken
sincerity, while shiftiness and fidgeting suggest insincerity. Take another case: Even
when you affirm the result of a calculation, doubt may remain if you have erred in the
past. Here careful calculation may inspire confidence while the many mistakes you
have made in the past give you pause.
Something can make the affirmative seem right on a certain question even while
something else does the same for the negative (or at least for suspending). Such exam-
ples are legion. Suppose you measure Müller-Lyer lines. Even while this makes it seem
to you that they are congruent, the visual attraction to think them incongruent still
remains, at least while the lines are in view. Opposing attractions might thus coexist
concurrently in a single mind. What follows will refer to such attractions mainly as
“seemings.”
2.4. Sources and Foundations

In visual perception there is an attraction to assent whose source is a correspond-
ing visual experience. By ‘source’ here I mean a rational source or basis. This means a
motivating psychological state. Such reasons or rational bases are those for which one
believes, or desires, or judges. These are reasons why, but they are distinctive. Thus, if
5 “Assigning weights” suggests more active agency than “being attracted.” Later we shall touch on this
issue of how active, or even voluntary, is the thinker’s conduct as he deliberates or ponders.
6 A fuller discussion would distinguish between rebutters and undercutters. A rebutter of a seeming that
p would be an opposing seeming, that not-p. An undercutter normatively undercuts the attraction to assent
to p without constituting an attraction to assent to not-p.
I now have a yen for ice cream prompted by seeing the words ‘ice cream’, this visual
experience is a reason why I have my conscious desire, but it is not a reason for which
I have that desire. Seeming to see those words may help explain my desire causally,
without rationally motivating it. Similarly, if some bad news puts me in a bad mood,
and I am annoyed at you for arriving one minute late, my mood may be a cause but not
a rational motivator of my annoyance. The only reason I have for being annoyed at you
may be your lateness, even if that is a terrible reason.
We can now explain a way in which sensory experience, the phenomenal given, has
been thought to serve as a foundation for human knowledge. Experience is a rational
basis for perceptual belief, but it does not itself admit justification. This is how it func-
tions as a regress-stopper, as a rational foundation. Experience is a state that gives jus-
tification without needing it in turn. Experiences are supposed to be rationally passive
so as to lie beyond rational justification or unjustification. They just happen to us, inde-
pendently of our rational agency, which is why they are not rationally assessable.
That, then, is how experience is supposed to serve as a foundation for perceptual
knowledge. Does anything play that same role for intuitive knowledge? It might be
thought that intuitive seemings do so. When it is part of a paradoxical cluster such as
the sorites, an intuitive attraction can remain strong even once seen to be false. And
the same goes for perceptual illusions such as the Müller-Lyer. The lines can still seem
incongruent even once we know them to be congruent.
Intuitive seemings do have a perceptual analogue, but it is not sensory experiences.
The analogue of intuitive seemings is rather perceptual seemings. Such seemings,
whether perceptual or intuitive, differ crucially from experiences. Unlike experiences,
seemings are evaluable as to whether they are rationally, epistemically appropriate.
Experiences are like pains. They are just there, independently of our rational agency.
When we are in pain, there are reasons why that is so, but they are not motivating rea-
sons: they are not reasons for which we are in pain. We do not enter or remain in a state
of pain for a reason that motivates our doing so. Sensory experiences are like that. We
have no motivating reason for our visual experience of light rather than darkness when
we open our eyes in the sun. There are of course reasons aplenty why we have that expe-
rience, but they are not rationally motivating. They are reasons why, but not reasons for
which we experience as we do.
Seemings, however, unlike experiences, can have a rational basis. My reason for
being attracted to think that p, for example, may be the testimony of a trusted friend,
while my reason for being attracted to think the opposite is the contrary testimony of
another friend.
That being so, intuition, unlike perception, seems to involve no regress-stopper
of the sort defined. Intuitive seemings are no better than perceptual seemings as
regress-stoppers of that sort. Neither kind of seeming is a state beyond epistemic justifi-
cation. Neither is exempt from epistemic assessment in respect of its rational propriety.
Consequently, intuitive seemings, intuitive attractions to assent, seem unable to
acquire their proper epistemic standing in the way of perceptual seemings: i.e., by
42 Ernest Sosa
being properly based rationally on a regress-stopper—namely, the sensory experi-

ence—beyond justification and unjustification.
That is how it seems, but not perhaps how it is. Is there not after all a plausible
regress-stopper for intuition? Why not the understanding or grasp of the proposition
intuited? Although intuitions are not based on corresponding experiences, they are
not baseless. Why does it seem to me that nothing is self-diverse? Just because I under-
stand the question; that is all it takes. I am attracted to think that nothing is self-diverse
simply through understanding that content well enough. Such seemings are based on
understanding of their propositional contents, in a way that is also subject to rational
assessment. It quite properly seems to me that nothing is self-diverse, based just on my
understanding. All one needs for determining whether nothing is self-diverse is just to
understand the question well enough. That suffices for answering it, and for doing so
properly. One answers as one does simply because one understands the question well
enough. That is all it takes to properly motivate one’s answer. Pains, by contrast, have
no more rational basis than do the visual sensory experiences that people undergo
when they view a long foreign word whose meaning they do not know.
Just how deep is the analogy between experience and understanding, however, as
sources of foundational justification? Before we take this up, we need first to take a
step back.
2.5. Latitudinarianism
Let’s consider seemings, attractions to assent, in general, whether intuitive or not.
All seemings, it may be thought, are created equal, epistemically on a par in the
following way.
General latitudinarianism
If it seems to S that p, then this seeming is ipso facto epistemically justified for S. In
other words, if S is attracted to assent to p, then S is ipso facto epistemically justified
in being so attracted.
This is clearly wrong for rationally based seemings in general, where the attraction
is rationally motivated, with the subject attracted for an ulterior reason. Surely the
reason can be a bad reason, in which case the seeming would seem unjustified, epis-
temically inappropriate. It might seem to me that I will have a good day, based on read-
ing my horoscope. Since my seeming here is less than rational, this refutes general
latitudinarianism.
Are foundational seemings special? Are all foundational seemings epistemically on a
par? What, first, are such seemings?
A seeming is foundational if, and only if, it is based sufficiently on a regress-stopping
reason: i.e., on a state that can provide justification to attitudes that it rationally
founds, without requiring or even allowing justification of its own, since it is too
passive a state to be rationally assessable.
Here are some examples of such foundational seemings:

a. It seems to S that he has a headache, and his rational basis, the reason why it
seems to him that way, the reason for which he is attracted to assent to that con-
tent, is that he does then have a headache.
b. It seems to S that he faces a fire, and the reasons why he is so attracted are that it
looks that way, feels that way, and sounds that way.
c. It seems to S that 3 + 2 = 5, and his basis, his reason for this is just that he ade-
quately understands that very content.
d. It seems to S that squares have four sides, and his basis, his reason for this is just
that he adequately understands that very content.
With this view of foundational seemings in the background, here now is the thesis
under consideration:
Latitudinarianism of foundational seemings
All foundational seemings are ipso facto epistemically justified.
This may seem obviously unacceptable, since the subject’s mind can be full of counter-
acting information, with the upshot that it should not seem to her that p, even though it
does foundationally seem that way to her.
To this objection there is a plausible reply. The possibility of such counteracting
information shows only that the subject’s resultant seeming might properly contra-
dict her foundational seeming. And that resultant seeming might coexist compatibly
with an initial, foundational seeming that is appropriately in the affirmative. So, an
epistemically justified resultant seeming that not-p is compatible with an epistemically
justified initial (prima facie) seeming that p.
So far, foundational latitudinarianism is defensible. Next let us look more closely at
intuitions, in particular at intuitive seemings. Properly restricted, here is the view:
Latitudinarianism (of intuitions)
All intuitive seemings are ipso facto initially, prima facie, rationally intuitive, and
thus epistemically justified.
What seems distinctive of intuitive knowledge is that simply understanding the spe-
cific propositional content properly grounds its appearance of truth. The sheer under-
standing serves on its own as a proper basis for that seeming, helping thus to make it
rational. Here’s another way to put the point:
A seeming is rationally intuitive if and only if it acquires some epistemic standing—
some rational, probative value—on a basis that includes nothing other than the sub-
ject’s understanding of the relevant propositional content.
Would any such understanding-based seeming amount to a non-inferentially justi-
fied, rational intuition? Would every seeming acquire such epistemic standing and
probative value, simply through the subject’s understanding of its very content? This is
what we next take up.
44 Ernest Sosa
2.6. Irrational, Epistemically Ill-Formed Intuitions

Intellectual seemings are themselves epistemically evaluable. A consideration might
attract too much, or too little; intuitive attractions are no exception. The bare enter-
taining of a proposition with perfectly fine understanding can attract too much. For
example, the attraction might be just a bias or superstition. Here the causal source is a
kind of social pressure that bypasses cognitive mechanisms of proper inference, testi-
mony, or perception. This pressure might lead a child to give a proposition the wrong
intuitive weight, and this might then linger into adulthood through normal cognitive
inertia, even if that proposition should exert no attraction whatsoever absent support
by ulterior reasons. Enculturating social pressure might thus anchor an inappropriate
attraction, an improper bias.
We have distinguished between (a) the operation of improper enculturation, and
(b) the rational operation of normal cognitive mechanisms that continue to exert their
influence via memory. An attraction can also be grounded or reason-based not directly
but through memory. Although originally acquired through conscious inference,
let’s say, it now remains stored in memory, its conscious basis long forgotten. None of
brainwashing, hypnosis, or enculturation works that way; none is reason-based at all,
much less consciously so.7
These potentially flawed sources can all give rise to bad attraction that is still intui-
tive, since rationally based just on understanding. Current attraction through the
earlier operation of such mechanisms has no rational source beyond the sheer under-
standing of the proposition, with its determinate content. (Please note well: the sort of
case imagined is one where at no time does the subject have any rational basis for being
attracted as he is. I am supposing that the flawed sources, such as enculturation, need
not operate through the provision of reasons to the subject enculturated.) In such a
case mere understanding is nearly always a poor reason, even when the proposition is
perfectly well understood. This is especially clear when it is not even true that p, so that
there is no fact there to exert the rational attraction.8 Here it is the sheer grasp of the
content that attracts one unaided; one has no rational basis for being attracted to assent
beyond one’s understanding of it. Only the enculturating mechanisms cause the attrac-
tion, and they need not accomplish this through rational basing on the part of the sub-
ject. As far as cognitive processing goes, at no point can the subject adduce reasons for
7 I argue in terms of mechanisms of belief-formation that are reason-based, since I take even perception
to be one such. But my argument could be framed about as well in terms of mechanisms that operate through
grounds or bases other than reasons (if there are any), since the mechanisms that I wish to contrast with
intuition do not operate through grounds or bases in any case.
The point here is delicate. Even with these non-reason-based mechanisms, there may be a kind of basing
on reasons, since belief may perhaps always be based in part on understanding. So, when I speak of beliefs
that are not reason-based, please understand this as ‘not reason-based except possibly for understanding’.
8 One would more naturally say: because of the fact that p rather than because of what seems to be the
fact that p. My latter more cautious formulation is meant to respect the difference between ignorance and
being attracted to believe, except at most by saying that the relevant content just seems
obviously true given just his understanding of it.
However, the point must be put delicately. One is not attracted simply by adequate
understanding in a sense implying that whenever one understood a proposition thus
adequately, one would be equally attracted to assent. This is obviously unacceptable,
if only because one generally understands a proposition no more adequately than its
negation. One is attracted rather to assent by adequate understanding of the content in
question, where the very content is no less essential to the attraction than is the under-
standing of it. The negation of that content would not attract, despite one’s understand-
ing it no less well.
2.7. What Distinguishes Intuitions with Probative

Standing?
Assuming that intuitions can derive from unfortunate enculturation, then, what distin-
guishes those that are rational? It cannot be the quality of the perfectly good understand-
ing itself that makes the difference. Why not? Again, because our understanding of a
proposition will normally be no better than our understanding of its negation. Someone
might be drawn to believe the negation, based just on his understanding of it, but only
because of subliminal enculturation. That attraction, however, that seeming, does not
acquire proper standing that way. Take someone brainwashed by a cult so that it seems
to him that he has no reason whatsoever to avoid hurting anyone. Nothing beyond his
sheer understanding of that proposition rationally yields the corresponding seeming
and even the corresponding belief. He has no motivating reason for that seeming; he has
none now, nor did he ever have any. But this fact does not render his attitude rational.9
Again, not all seemings motivated rationally by nothing beyond one’s understanding of
their contents are thereby rendered rational.10
irrationality. A belief that q might turn out false even if derived from flawless rational performance. Suppose
one judges that p based essentially on a supporting belief that q. Since it is false that q, one would speak falsely
if one cited as one’s reason just the fact that q. I see no better way to protect one’s rationality, and one’s rational
basis, against this untoward outcome, than to say that, even if it is false that q, so that one’s reason cannot be
the fact that q, one’s reason might still perfectly well be what one takes to be the fact that q. Alternatively, one
might lay out one’s rationale by displaying one’s argumentative basis. And this can remain one’s “rationale”
even if an essential premise turns out to be false.
The same plausibly applies not only to one’s rational bases for beliefs but also to one’s rational bases for
seemings. (Again, one might alternatively just replace the “reason” formulations with “rationale” formula-
tions. This would reduce or eliminate the implication of factivity.)
9 I use the word ‘rational’ in a broad sense, which does not require rational attitudes to be based on ulte-
rior reasons.
10 Of course, it is rationally more appropriate to have a belief in line with a resultant seeming provided the
seeming is strong or intense enough; indeed, such a seeming may just amount to the corresponding belief,
provided it is a resultant seeming. But the epistemic acceptability of the seeming, and perhaps thereby of the
corresponding belief, surely goes beyond the mere fit of the belief with the seeming.
46 Ernest Sosa
What feature must a seeming have, beyond deriving from such sheer understand-
ing, if it is to be rationally intuitive?
Simplicity is not the key, since no proposition will ever be that much simpler than
its negation. (And besides it is often the more complicated negation that is intuitively
justified.)
Might simple truth be the required feature? Is it the combination of understanding
and truth that makes the difference? No, this will not stand comparison with paradoxes
such as the sorites. Some proposition in the paradoxical cluster, at least one, is both false
and rationally intuitive. When hard thought reveals a solution, one that requires reject-
ing some proposition in the cluster, this does not entail our total lack of justification for
intuitive attraction to that proposition. Rather, it follows only that reflection has now
defeated our rationally intuitive attraction. The present outcome is hence analogous to
the outcome of one’s measuring the lines in a Müller-Lyer case. Anyone ignorant of the
illusion would surely be justified to consider the lines incongruent, based on his per-
ceptual experience. Just so, anyone ignorant of the sorites paradox would be justified
in his intuitive attraction to each member of the cluster taken singly, before they are
put together to yield the clash. Suppose one of these seemings is eventually shown to
be false. Even that seeming has epistemic standing for that subject, certainly, before it is
shown to be false (and perhaps even afterwards). It is properly, rationally involved in the
subject’s deliberation or pondering, at least before the resolution of the paradox. Even
afterwards it is far from clear that its attractive force is removed rather than overcome.
We have considered some proposed explanations of what makes an intuition
rational, but none has survived scrutiny. The source of the rationality of an intuition
cannot be the sheer understanding of the proposition intuited. No degree of under-
standing will suffice on its own, even combined with the proposition’s simplicity, com-
bined with its truth, or combined with its simplicity and its truth.
2.8. A Competence-Based Account

Here is a further option.
A proposition is intuited when the subject is attracted to assent just through under-
standing it. Such intuitive attraction, such an intuition, is rational when it is owed to
a competence manifest in the subject’s attraction to assent, one that reliably enough
discerns the true from the false. Such a competent ability would be manifest not
only in the attraction to which it gives rise, but also in the attraction’s being correct,
by having a true content.
What is to be said for or against this option?
It is instructive to compare intuition more closely with perception. Visual percep-
tual knowledge normally involves a visual experience that gives rise to a visual seem-
ing. Normally when a visual seeming corresponds to a visual experience, it is thereby
justified, and yields in turn a corresponding belief, itself thereby justified. However,
such perception is not a good model for successful intuition.
Intuitions are best understood as intellectual seemings or attractions. Unlike visual

experiences, seemings are not sources of our attraction to assent. They are rather the
attractions themselves. An attraction is intuitive when it has no rational basis other
than the subject’s conscious understanding of the proposition whose truth is intuited,
with its specific content. The conscious understanding is not the intuition, however; it
is not what is distinctively characteristic of intuitive justification. After all, conscious
belief always involves conscious understanding, even when the belief or the seeming is
not intuitive, but introspective, perceptual, or inferential.
What then is distinctively involved in intuitive justification? What distinguishes it
from perceptual, sensory justification? Here again is a point of difference:
What distinguishes intuitive justification is that the entertaining itself (with ade-
quate understanding) of that specific content exerts its attraction while rationally
unaided.
Intuitions are reason-based in a way that does not go beyond conscious grasp of the
specific propositional content. Intuitive justification involves nothing quite like the
visual sensory experience that in perceptual knowledge sustains one’s visual seem-
ing. Intuitive justification involves no pre-seeming experience that can prompt and
thereby justify a corresponding seeming, nothing other than the sheer understanding
of that content.
2.9. Intuition and Perception

Despite all our qualms and reservations, intuitions and their role in a priori knowledge
still seem importantly similar to perceptions and their role in empirical knowledge.
How does the competence approach accommodate this? Even in the case of percep-
tion, I have argued, seemings are operative, intellectual seemings, attractions to assent.
The difference is that perceptual seemings have a basis other than sheer understand-
ing; they have a further basis in the experiences that figure separately in the psychology
of the subject. Only when combined with experience does conscious understanding
here yield the attraction, while conscious understanding by itself exerts no attraction.
Intuitions by contrast are attractions to assent that do derive from unaided conscious
understanding of their propositional contents. A rational intuition, moreover, derives
standing and probative value by being competently formed even while rationally based
thus on nothing beyond the sheer understanding of its specific content.
Sensory experience figures as a rational basis in perceptual knowledge and justi-
fication. Foundational justification is said to require a regress-stopping state that
provides justification without requiring it in turn, a state undergone passively. Such
a state cannot be entered or exited through voluntary choice. It even lies beyond the
subject’s rational agency, so that it is not entered for a motivating reason. Sensory states
are of course like that. They provide justification without requiring it in turn, and are
48 Ernest Sosa
undergone quite passively. You enter or exit them neither by choice nor based on rea-
sons; they are beyond your rational agency.
Even when a sensory state provides justification, we would like to know just how
this mode of justification does its work. Does such a state always justify its correspond-
ing seeming or belief? The “speckled hen” problem shows that this cannot be right,
or so I have argued elsewhere (Sosa 2002). How then does one here acquire justifica-
tion? An alternative account requires competence instead. Take those who justifiably
believe that they face a collection of three items, based immediately on a correspond-
ing visual experience. These are distinguished by their ability to subitize (reliably) that
many items. Your belief that you see eleven items in your visual field has no compara-
ble justification, absent any comparable ability to subitize beyond the usual seven or so
items. Without that ability there can be no such justification, not even when the items
seen in your visual field do number eleven. Suppose what accounts for your belief is
just that you have been hypnotized, or temporarily wired, so that you will respond to
eleven items in your experience with a corresponding seeming. In that case, your belief
is not just accidentally related to the fact believed. One might hence attribute a kind of
“knowledge” to you, if only metaphorically. But your state falls short. It is certainly not
knowledge, and seems less than epistemically justified.
Will the exercise of competence suffice, even in the absence of any regress-stopping
state beyond understanding of the content involved? There is no apparent reason to
require that our basic non-inferential mechanisms of proper belief fixation must
always operate through such an ulterior mental basis, beyond sheer understanding.
This is not only implausible abstractly. It is implausible also when we consider basic
arithmetic, geometric, and logical beliefs in the human mind. No such ulterior rational
basis seems discernible for such beliefs.
As for contingent beliefs, consider also our competences to discern cardinality, or
shape, or color. Some such competences are reason-based: the subject acquires his
belief or seeming based on some guiding ulterior mental state. Even our perceptual
competences, however, are not plausibly required to operate (inevitably, necessarily)
through the guidance of ulterior states. Thus, consider the phenomenon of blindsight.
No person-level experience guides blindsight beliefs about the orientation of cer-
tain lines. Yet the subject’s seemings and beliefs about the faced lines are still reliably
formed. In any case, suppose in the actual world visual experiences do after all serve
to guide blindsight beliefs. Even so, in nearby worlds such beliefs can derive reliably
from purely subpersonal mechanisms. If so, how are they to be assessed epistemically?
Suppose they can indeed be as reliably competent as are simple arithmetic and geo-
metric beliefs, and can thereby constitute knowledge. If so, then perceptual knowledge
of one’s environment requires no foundational regress-stoppers beyond understand-
ing of the content involved. What really matters is rather that the belief should mani-
fest a relevant competence. This is so whether or not it is guided by regress-stopping
states other than understanding. Even if there are such states, they must still guide the
subject through competence. On the other hand, if there are no such basis states, as with
blindsight, we still need a guiding competence if the subject is to attain justification

and knowledge, even if this competence operates not through rational basing, but only
subpersonally.11
There is of course a way in which experiences are epistemically evaluable, as accurate
or veridical, as adroit or competent, and even as apt, as veridical because competent.
Only with seemings, however, do we ascend to the space of the conceptual, to rational
deployment of concepts.
2.10. Just How Can Understanding Function as a

Source of Epistemic Standing for Intuitions?
Fundamental, intuitive rational beliefs are based at least on understanding of the prop-
ositions believed, or so it has been argued above. It is not, however, just the under-
standing of a proposition, whatever its content, that gives a proper basis for believing
it. Otherwise, it would also constitute a basis for believing its negation, which must
be equally well understood. Not even the highest pitch of clarity and distinctness will
suffice. After all, whenever you attain that pitch for any given proposition, you will
attain the same pitch for its negation. What suffices is rather the being understood
(shared by a proposition and its negation pretty much equally) along with the specific
content of that very proposition. Can we go beyond this to some general feature that,
when combined with the being understood, will properly yield acceptance? As we have
seen, neither simplicity nor truth is such a feature, either singly or in combination.
Nevertheless, something distinguishes simple truths of arithmetic or geometry, for
example, making them suitable objects of immediate acceptance upon understanding,
and giving them their attraction to normal human minds universally (upon under-
standing). Whatever it is, whether innate or socially instilled, its yield is uniform and
general enough to suggest dispositions at work (whether wholly individually seated, or
partially socially seated). Given how epistemically benign they are, finally, such dispo-
sitions seem not inappropriately considered “competences.”
References
Sosa, E. (2002). “Privileged Access,” in Q. Smith and A. Jokic (eds), Consciousness: New
Philosophical Essays. New York: Oxford University Press, pp. 273–95; also in BonJour, L., and
Sosa, E. (eds) (2003). Epistemic Justification. Oxford: Blackwell.
11 According to the account defended here, the seemings of the competently blindsighted might well be
allowed to count as intuitions and indeed as rational intuitions. I myself find this surprising but perhaps
acceptable. Anyone who disagrees can delimit the intuitions of interest, and the rational intuitions, so as to
bring them into better alignment with the tradition. Thus, they might be restricted by the character of their
content, including its modal status, or they might be delimited by their distinctive reliance on sense organs;
even the blindsighted depend on open eyes. It may be interesting to argue about such restrictions, but for
present purposes they seem unnecessary.
3
Empirical Evidence for
Rationalism?
Joel Pust†
3.1. Introduction
Rationalism of a moderate variety has recently enjoyed the renewed interest of epis-
temologists. As I shall use the term, “moderate rationalism” designates the view that
a person’s having a rational intuition that p prima facie justifies them in believing that
p. For present purposes, I shall simply assume that a rational intuition is a proposi-
tional attitude, distinct from belief or judgment in which a proposition seems true,
or is presented as true, in a distinctive manner to the subject of the attitude.1 On views
according to which perceptual experience, introspective experience, or memorial
experience also have propositional content which is presented to a subject, further
work is required to distinguish the exact manner of the presentation of a proposition
which is distinctive of rational intuition.
My focus in this essay is on the suggestion that adherents of moderate rationalism
ought to provide an empirical defense of their doctrine (hereinafter “MR”), either
because empirical evidence is necessary to justify the acceptance of MR or because
empirical evidence would be dialectically sufficient to compel their empiricist oppo-
nents to accept MR. More specifically, I shall be concerned with three arguments.
According to the first argument, the causal-requirement argument, empirical evi-
dence is necessary in order to justify the claim that any actual token belief is based on
rational intuition and MR requires such a claim for its justification. According to the
second argument, the reliability argument, empirical evidence is necessary in order
to justify the claim that a putative source of evidence is reliable and MR requires such
† For helpful comments on previous drafts of this chapter, I thank my colleagues at the University of
Delaware, an audience at the University of Connecticut, an audience at the University of Missouri, and two
anonymous referees for Oxford University Press.
1
A detailed discussion of the nature of rational intuitions can be found in Pust (2012).
Empirical Evidence for Rationalism? 51
a claim for its justification. According to the third argument, the empirical-case argu-
ment, certain sorts of empirical evidence would be dialectically sufficient to resolve the
traditional dispute between empiricists and rationalists in the rationalists’ favor.
3.2. The Causal-Requirement Argument

MR can be understood in a variety of ways. The following four variations will prove
useful in what follows:
[MR1] If S has a rational intuition that p, then S is prima facie justified in believing
that p.
[MR2] Some person at some time is prima facie justified in believing that p in vir-
tue of having a rational intuition that p.
[MR3] Some person at some time has a prima facie justified belief that p in virtue
of having a rational intuition that p.
[MR4] Some person at some time prima facie justifiably believes that p in virtue of
having a rational intuition that p.
MR1 is a necessary proposition and does not entail the existence of any contingent
entity, state or event. MR2 to MR4 are contingent propositions as they existentially
quantify over contingent beings and their contingent properties. As I intend them to
be read, MR3 and MR4 (but not MR2), imply that someone has both a rational intui-
tion that p and a belief that p. MR3 holds that someone’s belief is justified and MR4
holds that someone’s believing is justified.
The so-called “basing requirement” is often introduced in terms of a distinction
between propositional justification and doxastic justification. The former obtains when
a proposition is justified for a given person—when, that is, they have a justification for
believing it, whether or not they actually do so believe. The latter requires, in addition to
propositional justification, that the person believe the proposition. It is often also held
to require that the person base their belief in the proposition on some ground which
provides them with propositional justification. So, for example, a person who has good
reason, and hence propositional justification, to believe that a given politician is of ques-
tionable character, but who believes the proposition entirely out of prejudice, may be
said to fail to base their belief on their justification and so to lack doxastic justification.2
In what follows, I will focus on having a justified belief that p and justifiably believ-
ing that p (the epistemic properties at issue in MR3 and MR4), both of which are often
thought to require that S’s belief that p be based on a ground g in virtue of which S is
propositionally justified in believing that p (i.e. that g is a reason for which the belief
2 A somewhat similar distinction is sometimes drawn between justified belief and justifiable belief, with
the former notion being akin to doxastic justification in requiring basing on a suitable ground and the latter
being akin to believing while possessing propositional justification but failing to base one’s belief on a pos-
sessed suitable ground.
52 Joel Pust
is held). It is also widely thought that S’s belief that p is based on ground g only if S’s
belief that p is caused, causally sustained by, or suitably counterfactually dependent on
a ground g (Audi 1983; Swain 1981; but see Foley 1987; Ginet 1983). I shall avoid inde-
pendent discussion of the basing requirement and directly address the suggestion—
“the causal requirement”—that a necessary condition of a token belief (or person’s
believing) being justified is that it be suitably causally related to some ground, g, which
suffices to propositionally justify for the person the content of their belief.3
With these ideas in place, we can address the core motivation of the
causal-requirement argument, a motivation which is entirely general and has nothing
particularly to do with rationalism. If, as the causal requirement maintains, justified
belief requires a causal connection between the belief in question and some suitable
ground, then it appears that justification for believing the proposition that some token
belief, b, is (prima facie) justified requires reason to think that b is suitably causally
related to a ground, g. That is, justified acceptance of the claim that some token belief
is justified requires, given the causal requirement, reason to think that the belief bears
certain causal relations to other mental states. Indeed, justification for the claim that
some particular belief or believing is justified by some particular ground requires rea-
son to think that those particular items are causally related. This raises the question of
whether and by what means one might have such justification.
The suggestion cannot be, on pain of a regress, that one must have a justified belief
in the existence of a suitable causal connection between b and g in order for one’s
first-order belief b to be justified. Rather, the concern has to do with propositional
justification of a claim about our having justified beliefs or justifiably believing various
propositions. It is, in other words, a concern for the epistemologist, one concerned to
justify the claim that various actual beliefs are indeed justified.
On a suitably broad conception of one’s armchair epistemic resources, one’s intro-
spective powers constitute such a resource. However, introspection might initially
appear ill-suited to discern the causal relations which hold between one’s own mental
states. Indeed, as Robert Audi, a proponent of the causal condition on justified belief
(Audi 1983), notes “it is simply not clear how one can have introspective access to the
relevant causal relations, and hence to what” proponents of the causal requirement
“take to be a crucial condition of one’s justification” (Audi 1989, p. 310). The general
worry is put even more forcefully by Richard Fumerton:
[I]‌f we understand the basing relation in causal terms, we should beware of philosophers specu-
lating about which beliefs are or are not justified. One needs empirical evidence to support a
causal hypothesis, and it strikes me that philosophers are rarely in possession of the empirical
evidence they would need in order to support a psychological claim about what is or is not caus-
ing a given belief. (2002, pp. 206–7; see also Fumerton 1995, p. 92; 2007, pp. 40–1)
3 The proponents of the causal requirement agree that the causal relation between the ground and the
belief must be of the right sort and that not just any causal connection will do. Such a further necessary con-
dition raises complex issues which I will not address here.
In Fumerton’s hands, skepticism about our armchair access to evidence sufficient to

justify causal claims regarding our beliefs is used to support the suggestion that phi-
losophers, qua philosophers, ought to “concern themselves only with the question of
whether there is justification for us to believe this or that proposition” (2002, p. 207),
i.e. with propositional justification for particular propositions, rather than with the
doxastic justification of particular beliefs.
Now that we have before us the alleged general problem for the armchair episte-
mologist, we may turn to the form it takes in connection with MR. In his excellent
study of a priori justification (2003), Albert Casullo maintains that an epistemo-
logical theory such as MR can provide an “account of how our beliefs are in fact
justified” only if “the processes to which the theory appeals . . . actually play some
role in acquiring or sustaining the beliefs in question.” He maintains, moreover,
that “empirical evidence is necessary to show that the cognitive processes in ques-
tion [such as having rational intuitions] can, in some more robust sense [than
logical possibility], play a role in producing or sustaining the beliefs in question”
(2003, p. 171). More specifically, he maintains that the resources of the armchair do
not suffice to justify MR because “introspection cannot,” given the causal require-
ment on basing, “provide assurance that what one takes to be the justification for
believing a certain proposition is true is the basis of one’s conviction that it is true”
(2003, p. 176).
Given that one’s own beliefs are the ones to which one has the best armchair access,
we may distill two arguments from these suggestions. The first is directed against the
claim that armchair resources could suffice to justify the view that any of one’s own
beliefs is justified by a rational intuition and the second is directed against the view that
armchair resources could suffice to justify the view that one justifiably believes some
proposition in virtue of having a rational intuition.
The Causal-Requirement Argument—Version 1
[1]‌ One is justified in believing: S’s belief that p is justified only if S’s belief that p is
causally related to a proper ground g.
[2]‌ If one is not justified in believing that one’s belief that p satisfies what one is justi-
fied in believing to be a necessary condition of being justified, then one is not
justified in believing that one’s belief that p is justified.
[3]‌ Hence, if one is not justified in believing that one’s belief that p is causally related
to a proper ground g, then one is not justified in believing that one’s belief that p
is justified. [From [1] and [2]]
[4]‌ Introspection and rational intuition cannot justify one in believing that one’s
belief that p is causally related to a proper ground g.
[5]‌ Hence, introspection and rational intuition cannot justify one in believing that
one’s belief that p is justified. [From [3] and [4]]
[6]‌ Hence, non-introspective empirical evidence is required for one to be justified
in believing MR3. [From [5] and MR3]
54 Joel Pust
The Causal-Requirement Argument—Version 2

[1′] One is justified in believing: S justifiably believes p only if S’s belief that p is caus-
ally related to a proper ground g.
[2′] If one is not justified in believing that one’s believing that p satisfies what one is
justified in believing to be a necessary condition of one’s justifiably believing p,
then one is not justified in believing that one justifiably believes p.
[3′] Hence, if one is not justified in believing that one’s believing that p is causally
related to a proper ground g, then one is not justified in believing that one justi-
fiably believes p. [From [1′] and [2′]]
[4′] Introspection and rational intuition cannot justify one in believing that one’s
believing that p is causally related to a proper ground g.
[5′] Hence, introspection and rational intuition cannot justify one in believing that
one justifiably believes p. [From [3′] and [4′]]
[6′] Hence, non-introspective empirical evidence is required for one to be justified
in believing MR4. [From [5′] and MR4]
While these arguments purport to preclude an armchair justification for MR3 and
MR4, they do not raise any objection to MR2 or to our having armchair justifica-
tion for MR2. (As well they should not, given that only rational intuition would
appear to propositionally justify the premises of the arguments for their propo-
nents.) Rather, they contest the claim that one could have propositional justifica-
tion for believing that MR3 or MR4 are true in the absence of empirical evidence of
a non-introspective variety. In the remainder of this section, I shall argue that these
arguments fail.
According to premise [1]‌, S’s having a belief that p which is propositionally justified
is not sufficient for S’s belief that p to be justified. According to premise [1′], S’s hav-
ing a belief that p which is propositionally justified is not sufficient for S to justifiably
believe p. I shall suggest that the causal requirement is not at all plausible as a necessary
condition on justified belief, though it is more plausible as a requirement on justifi-
ably believing. Moreover, I shall suggest that both premises have less plausibility with
respect to direct rather than to inferential justification and with respect to occurrent
rather than standing belief.
Consider a case in which a person has the doxastic attitude mandated by her evi-
dence, grounds, or reasons but in which her attitude is not causally related to the pos-
sessed grounds. Suppose, for example, a scientist has very strong evidence for the
truth of some theory but is caused to believe the theory by a desire that it be true.
According to one version of the causal requirement at issue, her belief would fail to be
justified. However, what exactly about her belief is defective? It is, we are supposing,
supported by her total evidence. This fact, it seems, is sufficient to undermine the sug-
gestion that she ought to have some other doxastic attitude (such as disbelief or with-
holding) toward the theory. As her actual attitude is the epistemically fitting one for
the epistemic situation in which she finds herself, it seems her actual belief is beyond
epistemic criticism.
What might be more plausibly said is that the believer rather than her belief is the
proper subject of criticism. She, it might be said, exhibits her epistemic viciousness in
believing something simply because she wishes it to be true. So, she doesn’t justifiably
believe the theory even if her belief is justified (Ginet 1983). Whatever the plausibility of
such a claim, it appears to support only the causal requirement on justifiably believing
and not that on justified belief. After all, reasons must be given for thinking that a per-
son’s failings must necessarily infect her beliefs, even those beliefs which are a product
of her failings (Foley 1987, p. 203).
Moreover, whatever plausibility attaches to the causal requirement on justified belief,
is, I think, a result of an excessive focus on cases of indirect justification and of standing
(but non-occurrent) belief. Perhaps my non-occurrent beliefs are justified (when they
are) partly in virtue of some causal relation they bear to various grounds I have or once
had. Perhaps my inferentially justified beliefs are justified partly in virtue of some such
relation. However, matters appear quite different if we turn our attention to the most
plausible cases of direct justification of occurrent belief. Suppose, for example, that one
now is attentively aware of a red patch in one’s visual field and now has an occurrent
belief that one now is presented with such a red patch. Surely such a belief is justified
no matter what more might be true about its causal basis. Similarly, suppose that one
experiences intense pain as if in one’s thigh and has also the occurrent belief that one
is now in pain. Surely such an occurrent belief is justified regardless of whatever causal
relations it bears to its object. Finally, suppose that it now strikes one that the taller than
relation is transitive and one also occurrently believes that proposition. Surely such
a belief is justified no matter what caused it. Such cases, I think, clearly show that, no
matter how things stand with inferential or non-occurrent beliefs, there is no causal
condition on direct justification of occurrent conscious belief. Moreover, the rational-
ist proponent of MR3 claims for rational intuition merely the power to directly prima
facie justify such occurrent belief. So, premise [1]‌, which alleges all justified belief to be
subject to the causal requirement, is false.4
What, however, of premise [1′]? Do the cases outlined above undermine its war-
rant? To my mind, the matter is not clear. Insofar as the property at issue in [1′] is the
property of a person, it is at least somewhat plausible that her possessing that property
has something to do with the manner in which she arrives at belief. If, for example, it
is possible to have a rational intuition that p while having one’s occurrent belief that p
caused by wishful thinking, then, as noted earlier, one’s being so moved to belief seems
4 Here it is worth noting that, in spite of his claim that philosophers qua philosophers should defend
only theses regarding propositional justification, Fumerton’s own acquaintance-based account of direct or
immediate justification seems so strong as to render it quite unclear what additional epistemic good would
be provided by a causal relation. As he notes, “when everything that is constitutive of a thought’s being true
is immediately before consciousness, there is nothing more that one could want or need to justify a belief ”
(1995, p. 75).
56 Joel Pust
criticizable. What is less clear is whether that fact implies that one does not justifiably
believe. In this respect, then, the second version of the causal-requirement argument
is the stronger one.5 Nonetheless, we need not settle the status of [1′] as both versions
of the causal-requirement argument have a further contestable premise and are, in the
end, invalid.
Let us turn to consider premises [4]‌and [4′]. Those premises hold that introspec-
tion and rational intuition cannot provide us with justification for claims that our own
mental states stand in particular causal relations. Roughly put, they assert that intro-
spection might justify the belief that I have a token belief b and the belief that I have
some putative internal ground g, but not the belief that the former is caused by the lat-
ter. Fumerton, again, claims that “however one understands causation, it seems to me
patently absurd to suppose that one can discover through introspection what is caus-
ing one to believe what one believes” (2007, p. 41). Fumerton’s claim, however, seems
to admit of a reading on which what is absurd is that one can ever have introspective
grounds for such a causal claim and one on which what is absurd is that one can always
do so. The stronger claim is indeed implausible. The sort of case which Fumerton pro-
vides as illustration of the impotence of introspection is one involving the causes of
one’s own philosophical views. This is a case of inferential justification and so here,
as above, it is important that we not allow the plausibility of various claims regarding
inferential justification to influence our views of the status of non-inferential justifica-
tion. What premises [4] and [4′] require is the falsehood of the weaker claim. Whether
or not it is ultimately defensible, the weaker claim is not absurd.
Given the assumption that rational intuition does not provide us with direct justifi-
cation for believing claims about token causal relations,6 the central question in evalu-
ating both [4]‌and [4′] is whether or not we can have introspective justification for
believing some ground, g, caused some belief, b. Whether this is in fact so depends on
difficult questions regarding the nature of causal relations, the content of introspective
states, and the epistemology of such states. However, contrary to the suggestions of
Casullo and Fumerton, on some combinations of such views, it may turn out that we
can introspectively justify the required causal claims.
Some philosophers have held that beliefs about causal relations can be directly justi-
fied because one can directly observe an irreducible causal relation between two token
events (Anscombe 1971; Armstrong 1968).7 Such philosophers claim one can simply
directly observe that the knife cut the bread, the tree was felled by the axe, or that one’s
body is being pressed upon by some heavy object. Moreover, it has been argued that the
obtaining of a causal relation between two mental events can be directly introspected.
5 I thank two anonymous referees for correction here.

6 Some minimal conceptions of a priori justification, on which such justification is merely justification by
a non-empirical process, would allow that one could have a priori justification for causal claims.
7 Others have held, implausibly I think, that causal relations are directly observable because causal
relations between events are reducible to directly observable non-causal relations between the events
(Ducasse 1926).
Evan Fales (1990), for example, appeals to a case in which one directly introspects pres-
sure as if on one’s body and argues that in such a case one can directly introspectively
discern the presence of a causal relation between two events. Others have appealed to
the alleged direct introspectability of the causal connection between an act of willing
and some subsequent mental event.
That observation or introspection of such relations is direct in the relevant sense will
likely be contested by those who reject direct realism. They will hold that no property
or relation can be directly apprehended in perception or introspection unless it is not
possible for such a property or relation to be absent while a phenomenologically indis-
tinguishable mental state is present. More precisely, it may be maintained that a token
property or relation is directly observed or apprehended only if it is given in experi-
ence in the sense that it is not logically possible for two qualitatively indistinguishable
experiences to differ in respect of whether they instantiate the property or relation.
Given such a conception of directness, it is difficult to maintain that causal relations
are directly known by perception. It is less clear whether they can be known directly by
introspection. Given some occasion on which one experiences a color sensation and
forms the belief that one is now undergoing such an experience, might one’s belief be
caused by some other event? Perhaps it may be, and if so it appears that we cannot gain
the relevant sort of direct knowledge of causal relations.
However, more liberal accounts of non-inferential perceptual justification may lend
support to the notion that one can have immediate prima facie perceptual justification
for causal claims. Such liberal accounts (Pollock 1974; Huemer 2001; Pryor 2000) typi-
cally involve the following two commitments: (a) perceptual or experiential states have
propositional content (for the subject), propositional content not restricted to propo-
sitions about the properties instantiated in the experience; and (b) a subject who hosts
such a perceptual state is defeasibly but immediately justified in believing its proposi-
tional content. One may, for example, be directly prima facie justified in believing that
one has a hand simply in virtue of having an experience that one has a hand.
Exactly which propositional contents can and do serve as propositional objects for
such presentational or representational perceptual states remains debatable. If one can
have a perceptual experience that one has a hand, can one have an experience that one
is human, middle-aged or was born in New Orleans? While I am inclined to an abste-
mious account of the contents of perception, those who are not might well hold that
causal relations can be perceptually experienced.8 If so, then even if causation cannot
be directly observed or introspected in the stronger factive sense, it may still be that
one can have non-inferential perceptual justification for the claim that the window was
broken by the rock or the house swept away by the river.
Indeed, such a doctrine may plausibly be extended to the introspective realm and
one might be said to have direct introspective justification for the claim that some
8 Indeed, Siegel (2009) argues on phenomenological and explanatory grounds that we can visually expe-
rience that things are causally related. Note that “visually experience that p” is here used non-factively.
58 Joel Pust
token belief was caused by some experience, volition, or process of inference. One
might, for example, decide to imagine being in a room in one’s childhood home and
immediately thereafter be presented with faint images of said room. One might, on
this view, have a (fallible) introspective experience of a causal relation, i.e. an intro-
spective experience with a singular causal claim as its content. Similarly, it might seem
to one that one occurrently believes that not-not-p implies p because one has an occur-
rent intuition with that content. One would then have, on the envisaged view, direct
prima facie introspective justification for a singular causal claim regarding one’s men-
tal states.
The possibilities for non-inferential justification of singular causal claims are, I have
maintained, broader than some have imagined, and if any are ultimately defensible,
premises [4]‌and [4′] of the causal basing argument are false. In that case, the rational-
ist could directly justify by introspection the causal claims (implausibly) required by
[1] and (perhaps plausibly) required by [1′]. In fact, however, I am quite sympathetic to
the view that one cannot gain direct justification for causal claims and that their justifi-
cation is always inferential. Still, even if that is so, the conclusions of the two arguments
under consideration, [6] and [6′], don’t follow from the preceding premises. From the
inadequacy of rational intuition and introspection to directly justify causal claims it
does not follow that more than armchair resources are required to provide one with
justification for believing that one’s belief that p is suitably causally related to a rational
intuition that p. More precisely, that one cannot directly justify a causal claim by intro-
spection and rational intuition does not imply that one cannot do so inferentially by the
use of the aforementioned faculties conjoined with memory of purely mental events.9
Such an indirect justification would still, in virtue of relying on no claims requiring
perceptual justification, be broadly non-empirical.
In fact, we may be able, entirely from the armchair, to gain further introspective evi-
dence for causal claims regarding rational intuitions which we cannot gain for sensa-
tions or experiences. To see this, it is important to note that some mental events can be
produced in us as an immediate result of other mental events under our direct power.
One can, for example, produce in oneself a rational intuition simply by considering a
suitable proposition. However, one cannot, as Locke noted, produce sensory or expe-
riential states in oneself on mere consideration. I cannot, simply by considering the
proposition that I am seeing a chair, produce a perceptual state with the requisite con-
tent. So the rationalist has a distinct advantage over the empiricist in justifying the
causal claims which the causal-requirement argument alleges she must justify. The
empiricist’s ability to justify the analogous claims is hostage to fortune in that, in virtue
of the contents of perception, it almost always requires the cooperation of the occur-
rence of suitable sensations outside her direct control.
9 On the extent to which reliance on memory compromises a priority see BonJour (1998, pp. 124–9). Even
if reliance on memory is involved in any complex inferential justification and such reliance compromises the
a priority of the conclusion, such a reliance is not an appeal to any experiential source.
Though I have argued that each version of the causal basing requirement argument
fails to support its conclusion, it is worth considering to what extent moderate ration-
alism would be imperiled if either or both of the arguments were to succeed. The moti-
vation for considering MR3 and M4, it may be recalled, was Casullo’s suggestion that
any interesting version of rationalism must address “the epistemic status of our actual
beliefs” and “how our beliefs are in fact justified.” A perfectly natural reading of these
phrases takes them only to refer to whether there is justification for the beliefs which
we in fact hold. I have suggested that this is equivalent to MR3. If so, then MR4 is, while
interesting, not at issue in considering the epistemic status of our beliefs. Even if my
account of what MR3 requires is rejected, the question of whether we have proposi-
tional justification for what we believe seems the more foundational question in episte-
mology and in philosophy more generally.10
Indeed, we may go further and wonder just how much it would matter to the pro-
ponent of MR if people rarely believed that which they were propositionally justi-
fied in believing in virtue of the content of their rational intuitions. In one important
sense, the answer would be, “not much.” The rationalist’s primary claim is MR2—that
we are justified, in virtue of having rational intuitions, in believing various proposi-
tions. Whether or not we believe those propositions is, of course, independent of the
question of the justificatory power of our rational intuitions. However, that empiri-
cal evidence is required for us to have justification to believe even the weakest and
most central rationalist theses is the contention of the argument which will occupy our
attention in the next section.
3.3. The Reliability Argument

The second alleged need for empirical investigation in order to justify the acceptance of
moderate rationalism is based on the claim that truth conduciveness is required for jus-
tification. If so, “if one is to offer evidence that a particular process is a source of a priori
justification, one must offer evidence in support of the claim that beliefs based on that
process are likely to be true” (Casullo 2003, p. 171). More generally, those who maintain
that some putative source of justification is indeed such a source must provide reasons
for thinking that said source will produce more true beliefs than false ones. Casullo
claims, moreover, that “the claim that a process is truth conducive, or more minimally,
that it is not error conducive is a contingent general claim that can only be supported
by empirical investigation” (p. 171).11 Hence, in order to justifiably accept that rational
10 In support of this contention, notice that proponents of the causal requirement never seek to show that
anyone satisfies the causal requirement with respect to belief in the premises of their arguments. Instead,
they are content merely to show, just as Fumerton suggests they should be, that we are justified in believing
the premises, i.e. that we have propositional justification for the premises and hence for the conclusions.
11 Casullo sometimes seems to suggest that the rationalist must offer empirical evidence that defeating
evidence against the reliability of rational intuition is not available. While I agree that if a particular a pri-
ori justification is empirically defeasible, one must have justification to believe that one lacks empirically
60 Joel Pust
intuition is a source of justification, one must have empirical evidence that it is truth
conducive. Let us regiment these remarks in the following argument:
The Reliability Argument
[1]‌ A process, ϕ, is a source of prima facie justification only if ϕ is reliable.
[2]‌ Any substitution instance of “process ϕ is reliable” expresses a contingent gen-
eral proposition.
[3]‌ One can be justified in believing a contingent general proposition only on the
basis of empirical investigation.
[4]‌ Therefore, one can be justified in believing that ϕ is a source of prima facie justi-
fication only on the basis of empirical investigation. [From [1]–[3]]
[5]‌ Therefore, one can be justified in believing rational intuition is a source of prima
facie justification only on the basis of empirical investigation. [From [4]]
[6]‌ Therefore, one can be justified in believing moderate rationalism only on the basis
of empirical investigation. [From [5] and the definition of moderate rationalism]
An initial and obvious objection to premise [1]‌is the standard demon world objec-
tion to process reliabilism (Cohen 1984). If reliability requires that a process produce
mostly truths in the circumstances in which it is deployed, then reliabilism seems falsi-
fied by the possibility of persons who experience exactly as we do and believe as we do
but who are massively deceived. Such a case appears, on the assumption that we are
justified in believing as we do on the basis of our experiences, to show that reliability of
that sort is not a necessary condition on justification generally.
Perhaps the argument would be strengthened by beginning with the claim that jus-
tification to believe that a putative source of evidence is justifying requires reason to
think that the source is truth-conducive. So long as those persons massively deceived
would have good reason to think reliable a process which was in fact unreliable, such a
constraint would be secure from demon world counterexamples. This would amount
to the following:
[1′] One is justified in believing that a process ϕ is a source of prima facie justifica-
tion only if one is justified in believing that ϕ is reliable.
Even if this alternative repairs the initial step of the argument, premise [2]‌is false.
There are necessarily reliable (even if not infallible) processes. Reliability, as propo-
nents of the generality problem for reliabilism frequently point out, is a property
attaching to types of processes. Even if it is contingent that there are any actual process
tokens of some type, it is necessary that certain types are reliable. The process-types of
believing necessary truths or of proving theorems from true axioms or forming beliefs
defeating evidence in order to be ultima facie justified, it does not follow that one must have such justifica-
tion in order to be prima facie justified by a rational intuition in believing a given proposition. Moreover, to
allow that particular rational intuitions are empirically defeasible is not necessarily to allow that the justifica-
tory power of rational intuition generally is empirically defeasible.
about one’s conscious mental states by careful introspection are all plausibly necessar-
ily reliable process types. However, [2] cannot be abandoned in favor of
[2′] “The process of believing on the basis of undefeated rational intuition is reli-
able” is a contingent general claim,
because [2′] is intuitively implausible.
We may set these worries about the initial steps of the argument aside, as it can be
shown that if MR is true, then whether MR is a necessary truth or contingent truth,
one has entirely non-empirical reasons for thinking that S’s having a rational intuition
that p prima facie justifies S in believing that p. To see this, notice that rational intuition
is directly self-certifying.12 It is self-certifying in that, intuitively, belief that p is prima
facie justified for one when one hosts a rational intuition that p. So, belief in MR is jus-
tified if MR is true. Hence, it follows that [5]‌is false and, moreover, that the conclusion
of the argument is mistaken.
Another, broadly inductive, argument for the same conclusion is available. Consider
the individual rational intuitions in virtue of which one is, given MR, justified in believ-
ing various propositions. Intuitively, each of them suffices to prima facie justify belief
in its content. Hence, one has, given consideration of a sufficient number and variety
of propositions endorsed by rational intuition, inductive justification for believing that
rational intuitions provide prima facie justification for their contents, i.e. for MR.
These justifications are entirely non-empirical, depending not even on introspective
awareness of some contingent mental state, but only on the content of rational intuition(s).
What intuition testifies is that a state of the sort in question suffices to justify belief in its
content. While one has to host such a state in order to properly grasp the proposition
that it justifies, that one ever hosts such a state need not play any justificatory (rather than
enabling) role. Recall here the traditional rationalist’s distinction between one’s grasp of
a proposition depending on suitable experiences and one’s justification for believing the
grasped proposition depending on suitable experiences. Those features which play a role
in enabling us to grasp a proposition need not be part of what justifies us in believing it.
It may be objected that these self-certifying arguments presuppose the truth of MR.
Indeed, they do. However, this is entirely appropriate as the reliability argument was
supposed to show that belief in MR is not justified without empirical evidence even if
MR is true. That conclusion is false. Furthermore, while relying on intuition for evi-
dence of its justificatory power is epistemically circular in the sense that it involves
relying on a putative source of evidence to justify its own status as evidence, it does
not involve any kind of bootstrapping or track-record justification of the sort alleged
by some to be epistemically objectionable (Alston 1993; Fumerton 1995; Vogel 2000;
Cohen 2002). The justifications just outlined do not, unlike bootstrapping arguments,
12 Indeed, it is unique in being directly self-certifying because it is the only source which directly justifies
propositions regarding epistemic properties such as justification and knowledge.
62 Joel Pust
involve justificatory appeal to one’s having had any intuitions at all. This is especially
clear in the case of the general justification. It is, however, true as well of the broadly
inductive justification as one is not inferring, as in a track-record or bootstrapping
argument, from past instances of an actual intuition justifying some proposition to
the claim that intuition has justificatory power. Instead, one is appealing directly to the
fact that intuitions with the contents in question propositionally justify those who host
them in believing corresponding propositions.
Of course, those who think that bootstrapping or track-record arguments are epis-
temically unproblematic (Bergmann 2004; Van Cleve 2003) would have no objection
to a standard track-record argument for the reliability of rational intuition. Supposing
the truth of MR, a track-record argument for the reliability of one’s rational intuitions
can be given along the standard model. Consider again the set of rational intuitions
one has. Given MR, one is prima facie justified in believing each one of them is true. As
one may also know introspectively that each one of those propositions is the content of
one of one’s own rational intuitions, one may then be justified in believing that rational
intuition is reliable by the following non-demonstrative argument:
P1 and I have the intuition that P1
P2 and I have the intuition that P2.
Etc.
Therefore, rational intuition is reliable.

If any form of track-record argument for the reliability of some belief source is accepta-
ble, then a track-record argument for the reliability of rational intuition seems the most
plausible candidate. Moreover, the track-record argument just presented is still broadly
(though not entirely) non-empirical, requiring only introspection and rational intui-
tion for the justification of its premises. Hence, even if the reliability of rational intuition
is contingent, if track-record arguments are ever acceptable, premise [3]‌is false.
Finally, we ought to note here that the reliability argument is entirely general and so,
if it were sound, the empiricist would have to answer to its demands as well and pro-
vide empirical evidence that putative sources of empirical evidence are reliable. That
such an empirical defense would presuppose what it sets out to show seems quite clear.
Indeed, it would have to be a track-record argument of some kind because experi-
ence and introspection deliver no direct verdicts regarding justification. So, if such a
track-record argument for the claim that perception justifies beliefs is acceptable, then
so also is the rationalist’s justification of the reliability of rational intuition by appeal to
rational intuition. Indeed, it should be regarded as more acceptable in virtue of the fact
that arguments other than the track-record sort can be provided for rational intuition.
I conclude that moderate rationalism can, as its traditional proponents have main-
tained, be justified by direct appeal to the justificatory power of rational intuition(s).
If its justification requires reason to think it is reliable, that conclusion can be deduced
from the fact that it is justificatory or directly justified by rational intuition itself. These
defenses are epistemically circular, but as such circularity is in any case unavoidable,
that fact alone cannot impugn rationalism. Indeed, the rationalist is alone in having
recourse to a directly self-certifying faculty. Furthermore, if epistemically circular
track-record arguments are ever acceptable, there is a straightforward non-empirical
track-record argument for the reliability of rational intuition in one’s own case to
which no empiricist proponent of the reliability argument can consistently object.
3.4. The Empirical-Case Argument

Perhaps the most interesting way in which empirical evidence has been alleged rel-
evant to the defense of moderate rationalism is the recent suggestion that empiri-
cal evidence of the reliability of rational intuition would compel a consistent radical
empiricist to concede defeat and accept the existence of a priori justification of the sort
to which the rationalist appeals. Here is Casullo’s formulation of the idea:
[R]‌ationalists can successfully resolve their dispute with empiricists by offering empirical
support for the general claim that rational insight is truth-conducive. If their supporting case
involves only general principles that empiricists regard as justified and the evidence meets the
standards that empiricists regard as adequate for justification, then reasonable empiricists must
concede that there are cogent grounds for endorsing rationalism. (Casullo 2000, pp. 45–6)
Central to understanding and evaluating this suggestion is Casullo’s further claim that
empiricists need not be skeptics with respect to a suitable range of the propositions
which rationalists allege are justified a priori. Empiricists accept, for example, that we
have justified belief in various logical, mathematical, and other propositions. Hence,
the argument continues, if there is some way of arguing that the contents of rational
intuitions are reliable which is acceptable by the empiricist’s own lights, the empiricist
should concede that rational intuition provides at least propositional justification for
its contents, i.e. that MR2 is true.13
Consider the following doctrines:
Anti-Rationalism [AR]—S’s having a rational intuition that p does not justify S
(even prima facie) in believing that p.
Non-Skeptical Anti-Rationalism [NSAR]—AR but we are justified in believing
many of the propositions which are the contents of rational intuitions.
The central suggestion of what I shall call “the empirical-case argument” is that con-
sistent proponents of NSAR must accept that MR is true if provided an argument for
13 Similar suggestions appear in Kornblith (2000, p. 83) who claims that an empirical case for the exist-
ence of reliable rational intuitions would suffice for his accepting that there are a priori justified beliefs and
in Rey (1998, p. 33) who suggests that we could have empirical evidence that we had a priori knowledge of a
certain sort.
64 Joel Pust
the general claim that rational intuitions are reliable which depends only on proposi-
tions (and inferential principles) which they accept.
We should distinguish between two kinds of non-skeptical empiricists—internal-
ist empiricists and externalist empiricists. According to internalist empiricists, only
experience provides justification for belief and it does so because it provides a reason
for belief which is internally accessible in some suitable sense. According to externalist
empiricists, only experiential processes provide justification for belief and they do so
in virtue of some factor, such as actual world reliability, which is not necessarily inter-
nally accessible in the relevant sense.
Now, I take it that the envisaged strategy is one in which we have a kind of empiri-
cal calibration of rational intuition. Sticking to the crude inductive track-record argu-
ment, we have an argument with a suitably large number of premises of the form
“p and S has the rational intuition that p” and a conclusion of the form “S’s rational
intuitions are reliable.” Such an argument must be such that the empiricist has rea-
son to regard each conjunct of each premise as empirically justified. Supposing that
the empiricist is justified in believing a suitable number and variety of such premises
then she is justified in accepting the conclusion. According to the empirical-case argu-
ment, the empiricist might thereby arrive, by an argument which is acceptable to her
qua empiricist, at the conclusion that her rational intuitions are reliable. Would this
amount to showing, by her lights, that some version of MR is true?
Contrary to the empirical-case argument, it would not. While such an empiricist
might then accept that she is prima facie justified in accepting the contents of her
rational intuitions, she would not thereby be accepting MR. Rather, she would be
accepting the following doctrine:
[MR*] If S has a rational intuition that p and S has independent empirical reason
to regard her rational intuitions as reliable, then S is prima facie justified in
believing that p.
Crucially, the ultimate justification S has for believing p, in virtue of satisfying the ante-
cedent of MR*, is still an empirical one. The fact that S has a rational intuition that p
is not sufficient for S to be prima facie justified in believing p. MR, by contrast, holds
that having a rational intuition that p is sufficient for prima facie justification. The
empiricist can be brought to accept MR* without giving up her empiricism and with-
out giving up AR, properly understood. While proponents of MR and MR* agree that
a person is justified in believing when she has a rational intuition, they do not agree
about the fundamental epistemic principle at issue.
Consider, for comparison, the debate between reductionist and anti-reductionist
views of the epistemology of testimony. On a reductionist view, one’s justification for
accepting the contents of putative testimony is grounded in independent reason for
thinking that testimony, whether of persons generally or of the particular person in
question, is reliable. On the anti-reductionist view, testimonial justification is basic or
sui generis and requires no independent grounding. Instead, while certain experiences
might be required in order to entertain a testimonial content, one’s justification for

accepting the content derives solely from its status as an assertion or piece of testi-
mony. It would be quite odd to think that if a reductionist provided a reductive justifi-
cation for a general presumption in favor of testimony, they would thereby have given
up their reductionist position. More generally, the existence of a sound track-record
argument for the acceptance of derivative or non-basic sources of evidence does not
render the sources so calibrated basic sources of evidence. Hence, it is incorrect to
think that any empiricist would give up their empiricism if they accepted, on empirical
grounds, that there was reason to trust rational intuition.
Moreover, it is actually quite unclear how far such an empirical justification of the
reliability of rational intuition could be appropriately extended. Casullo insists that
empiricists are not skeptics as they wish to allow we are justified in believing math-
ematical and logical propositions (or at least those required for empirical science).
Still, many historical and present empiricists are skeptics with respect to many of the
propositions which the rationalists allege are justified a priori—moral propositions,
modal propositions, etc. Indeed, some are skeptical regarding those portions of math-
ematics or logic which are not required by empirical science.14 Hence, even if an induc-
tive case can be made, on empirical grounds, that rational intuition is a reliable source
regarding some of logic and mathematics, this is not necessarily a vindication of the
general thesis of the rationalist which is a claim about rational intuition generally and
is not restricted to certain domains. To show that rational intuition is even a derivative
general source, the envisaged inductive justification of MR* would need to extend not
just beyond the particular propositions which the empiricists regard as justified in a
given domain, it would have to extend to domains wherein many empiricists would
allege we lack entirely empirical reasons for belief. Such an argument would be difficult
to mount.
Assessing the empirical-case argument is more difficult if we consider instead exter-
nalist empiricists. Let us take as our paradigm externalist a process reliabilist accord-
ing to whom S is (ex ante) justified in believing p if and only if there is a reliable process
available by means of which S could arrive at belief that p and other available pro-
cesses would not, if used, result in S not believing p (Goldman 1979). As an externalist,
the reliabilist likely thinks of the division between various sources of belief as of little
fundamental importance. If reliance on rational intuitions is in fact reliable, reliabi-
lism would imply that such beliefs are justified (modulo various worries about defeat
(Casullo 2003, pp. 128–46)). Our question, however, is not what reliabilism implies
but what a consistent reliabilist would be rationally required to accept in light of the
empirical-case argument.
Setting aside the concerns just noted regarding the generality of the reliability claim
which would be justified by such an inference, a reliabilist who was initially neutral on
14 Here I am in disagreement with Casullo who claims that the “fundamental disagreement between apri-
orists and radical empiricists is not over the scope of human knowledge” but over the source (2003, p. 161).
66 Joel Pust
the question of the reliability of rational intuition, and who accepted the reliability of
the relevant inferences and empirical belief-forming processes would be persuaded by
the calibration just outlined. Moreover, as we saw in our discussion of the reliability
argument in section 3.3, she would also have available a variety of non-empirical argu-
ments which, if MR is true, could lead her to justified belief in it and so she would also
be ex ante justified in believing MR, if MR is true.
What we are now considering, however, is an externalist empiricist, i.e. a reliabil-
ist who denies (for whatever reason) that rational intuition is reliable but accepts that
empirical sources and inductive inference are reliable. It is difficult to determine what
a rationally consistent externalist empiricist would be compelled to do by the provi-
sion of the argument in question. The crucial issue is whether or not the conclusion
would be defeated by her antecedent belief in empiricism.15 If we suppose that she has
a standing belief in the unreliability of rational intuition, a variety of beliefs which she
takes to be produced by a reliable empirical process and a variety of beliefs, produced
by a reliable introspective process which she regards as reliable, that rational intui-
tion has agreed with the reliable empirical processes, then there is plainly some ten-
sion here. Does rationality require that she give up AR, her belief in the reliability of
those empirical judgments which appear to agree with rational intuition, or her view
that the calibration argument is a cogent argument? I don’t think the matter is clear
and so the claim of dialectical efficacy is uncertain with respect to the anti-rationalist
non-skeptical externalist.
Suppose, however, that reasons can be found to conclude that such an externalist
should be moved to accept that beliefs produced by rational intuition are ex post justi-
fied and that we are ex ante justified in believing many propositions by the availability
of the process of relying on rational intuition. Even if this were so, however, the impor-
tance of this result would be diminished by the fact that a genuine externalist is not,
I think, much concerned with the traditional debate between empiricists and rational-
ists as that is a debate framed in internalist terms and shot through with internalist pre-
conceptions. Even if the envisaged externalist accepts that there are non-experiential
reliable sources of beliefs, she was never a real party to the original debate.
3.5. Conclusion
I conclude that the justification of all of the forms of MR outlined above does not
require any non-introspective empirical evidence, that acceptance of the core version
of MR can be justified entirely a priori, and that any traditional empiricist who remains
unconvinced by the rationalist’s traditional arguments should remain unconvinced by
such empirical arguments for the reliability of rational intuition as can be produced.
Moderate rationalism can and should be justified entirely from the armchair.
15 The question of whether or not unjustified beliefs can serve as defeaters is a difficult one (Bergmann
2006, pp. 165–8). It seems to me that they cannot serve as defeaters of propositional justification.
References
Alston, W. (1993). The Reliability of Sense Perception. Ithaca, NY: Cornell University Press.
Anscombe, G. E. M. (1971). Causality and Determinism. Cambridge: Cambridge University Press.
Armstrong, D. (1968). A Materialist Theory of the Mind. London: Routledge & Kegan Paul.
Audi, R. (1983). “The Causal Structure of Indirect Justification,” Journal of Philosophy 80, pp.
398–415.
——. (1989). “Causalist Internalism,” American Philosophical Quarterly 26, pp. 309–20.
Bergmann, M. (2004). “Epistemic Circularity: Malignant and Benign,” Philosophy and
Phenomenological Research 69, pp. 709–27.
——. (2006). Justification Without Awareness. New York: Oxford University Press.
Casullo, A. (2000). “The Coherence of Empiricism,” Pacific Philosophical Quarterly 81, pp. 31–48.
——. (2003). A Priori Justification. New York: Oxford University Press.
Cohen, S. (1984). “Justification and Truth,” Philosophical Studies 46, pp. 279–95.
——. (2002). “Basic Knowledge and the Problem of Easy Knowledge,” Philosophy and
Ducasse, C. J. (1926). “The Nature and Observability of the Causal Relation,” Journal of
Fales, E. (1990). Causation and Universals. New York: Routledge.
Foley, R. (1987). The Theory of Epistemic Rationality. Cambridge, MA: Harvard University Press.
Fumerton, R. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield.
——. (2002). “Theories of Justification,” in P. K. Moser (ed.), The Oxford Handbook of
Epistemology. Oxford: Oxford University Press.
——. (2007). “What and About What is Internalism?,” in S. Goldberg (ed.), Internalism and
Externalism in Semantics and Epistemology. New York: Oxford University Press.
Ginet, C. (1983). “Justification of Belief: A Primer,” in C. Ginet and S. Shoemaker (eds),
Knowledge and Mind: Philosophical Essays. New York: Oxford University Press.
Goldman, A. (1979). “What is Justified Belief?,” in G. Pappas (ed.), Justification and Knowledge.
Dordrecht: Reidel, pp. 1–23.
Huemer, M. (2001). Skepticism and the Veil of Perception. Lanham, MD: Rowman & Littlefield.
Kornblith, H. (2000). “The Impurity of Reason,” Pacific Philosophical Quarterly 81, pp. 67–89.
Pollock, J. (1974). Knowledge and Justification. Princeton, NJ: Princeton University Press.
Pryor, J. (2000). “The Skeptic and the Dogmatist,” Noûs 34, pp. 517–49.
Pust, J. (2012). “Intuition,” in Edward N. Zalta (ed.), The Stanford Encyclopedia of Philosophy,
available at: <http://plato.stanford.edu/archives/win2012/entries/intuition/> [last accessed
October 7, 2013].
Rey, G. (1998). “A Naturalistic A Priori,” Philosophical Studies 92, pp. 25–43.
Siegel, S. (2009). “The Visual Experience of Causation,” Philosophical Quarterly 59, pp.
519–40.
Swain, M. (1981). Reasons and Knowledge. Ithaca, New York: Cornell University Press.
Van Cleve, J. (2003). “Is Knowledge Easy—or Impossible: Externalism as the Only Alternative to
Skepticism,” in S. Luper (ed.), The Skeptics. Aldershot: Ashgate.
Vogel, J. (2000). “Reliabilism Leveled,” Journal of Philosophy 97, pp. 602–23.
4
Moderate Intuitionism:
A Metasemantic Account
Michael Johnson and Jennifer Nado
4.1. Introduction
Intuitions have for many years been considered indispensable to philosophical meth-
odology. Recently, however, a growing body of empirical work has indicated that intui-
tions may be subject to various sorts of undesirable variation. These findings strongly
suggest that philosophers have substantially overestimated the epistemological worth
of intuition; it has even been suggested that intuition must be excluded from philo-
sophical practice. Nonetheless, given that there’s been no dramatic revolution in philo-
sophical method, most philosophers seem to be hesitant to relegate intuition to the
dustbin entirely. There is a strong temptation to say that intuitions must have at least
some evidential weight—though they obviously occasionally go astray. Given their
ubiquity in reasoning (philosophical and otherwise), it’s difficult to accept the idea that
our intuitions could be so unreliable that they’d have to be wholly abandoned as an
evidential source. Therefore, in this chapter, we’ll explore the potential for a ‘moderate’
account of intuition.
Moderate intuitionism (as we’ll call it) recognizes that intuitions are generally reli-
able, but also frequently in error on certain classes of cases. On the methodological
side, it allows that some revision of our philosophical practices might be in order,
while stopping short of a complete rejection of an arguably central philosophical
tool. Of course, moderate intuitionism can’t be defended solely by an appeal to its
pleasant consequences; what’s needed is an account that explains why a moderate
stance is appropriate. That is to say, there should be a theory that provides an expla-
nation of the genuine yet somewhat fragile connection to the truth that moderate
intuitionism aims to ascribe to intuitions. Our goal in this chapter is to outline such
a theory.
A plausible place to start looking for the needed theory is in language. Our ten-
dency to assume that intuition must be generally reliable seems linked to the fact
Moderate Intuitionism: A Metasemantic Account 69
that it is impossible for our intuitions about the meanings of words to be substan-
tially misleading. It’s not at all plausible to suppose, for example, that ‘dog’ might
(in the epistemic sense) turn out to refer to cats. This has led some philosophers
to suspect that there exists some kind of deep connection between intuitions and
meanings.
Unfortunately, however, the most popular existing metasemantic accounts don’t
seem to give us any kind of explanation for this apparent link—a fact that is particu-
larly odd, given that philosophers tend to defend their preferred metasemantic account
on the basis of intuitions. Thus, while intuitions regarding the reference of names in
certain circumstances are standardly employed to defend Kripkean causal-historical
accounts, such accounts don’t themselves provide us with any particular reason to
suppose that we would have intuitions in accord with them. Nothing internal to the
account explains why we should have causal-historical intuitions rather than, say,
descriptivist ones.
We propose to develop an outline for a metasemantic account which ties facts about
meanings to dispositions to apply words when in possession of complete information.
As we’ll show, an account of this type both fits our metasemantic intuitions, and pre-
dicts a link between intuition and meaning that could underwrite the former’s reli-
ability. At the same time, the account also predicts that intuitions will fail under certain
conditions. This fits pretty well with how the empirical evidence on intuition seems to
be turning out. We take this general fit between our account and the evidence to pro-
vide support both for the metasemantic account itself (or some other account along
the same lines), as well as for the moderate approach to intuition.
4.2. Moderate Intuitionism: Motivations

The nature of intuition itself is notoriously difficult to pin down. Rather than attempt-
ing to defend any particular analysis of intuition, we will just stipulate that by ‘intuition’
we mean to refer at the very least to spontaneous, not-obviously-inferential judgments
such as those that frequently occur in response to thought experiments.1 Our primary
goal is to defend a moderate stance on the epistemological status of the sorts of things
philosophers are inclined to call ‘intuitions’; as far as we can tell, our argument is com-
patible with a fairly broad range of particular, considered views on the nature of intui-
tion. As we’ll discuss later, it is even compatible with the claim that there is no unified
mental state-type underlying uses of the term ‘intuition’.
The empirical arguments that have been recently marshaled against intuitions are
well known, so we’ll only briefly rehearse them here. In short, intuitions vary—both
interpersonally as well as intrapersonally. Evidence suggests that intuitions vary across
cultures (Weinberg et al. 2001; Machery et al. 2004), genders (Buckwalter and Stich
1 We characterize intuitions as judgments here. Others may prefer inclinations to judge, or seemings that
can lead to judgments. Nothing in what follows hangs on this issue.
70 Michael Johnson and Jennifer Nado
2014), and socioeconomic groups (Haidt et al. 1993; Weinberg et al. 2001); and that
they are sensitive to emotional states (Wheatley and Haidt 2005; Schnall et al. 2008),
order of presentation (Swain et al. 2008), and more. Since these factors are plausibly
irrelevant to the phenomena the intuitions are meant to describe—that is, e.g., the facts
about knowledge don’t vary as a function of the cultural background of the attributor—
this variation suggests that intuition is not wholly tracking the truth. Thus, variation
data has been taken to show that intuitions are unreliable, and that they are therefore
unsuited for use in philosophical theorizing.
Of course, this last step is open to serious question. If the findings are robust, they
show (barring relativism) that at least some intuitions are in error. But, it may be
argued, this merely demonstrates that intuition is fallible—a fact that few philoso-
phers would deny. As Ernest Sosa has noted, perception is also susceptible to various
interfering factors that cause it to occasionally go awry, but perception is nonetheless a
quintessentially respectable evidential source. “The upshot is that we have to be careful
in how we use intuition, not that intuition is useless” (Sosa 2007, p. 105). Intuition isn’t
perfect, but that’s no reason to think that it should never be employed in philosophical
argumentation.
In fact, there are reasons to suspect that intuition can’t be totally rejected as an evi-
dential source without invoking radical skepticism, both with regard to philosophi-
cal inquiry as well as generally. As George Bealer (1992) and Joel Pust (2000) have
noted, it seems that even arguments against intuition invoke intuitions—for example,
intuitions that beliefs formed by unreliable processes are unjustified. Even worse for
the anti-intuitionist, Timothy Williamson (2007) has argued persuasively that there
is simply no clear distinction between ‘philosophical’ intuitions and everyday cases of
concept application. If this is right, however, arguments against the reliability of intui-
tion threaten to generalize to arguments against the reliability of concept application—
a remarkably skeptical position.
More generally, it’s patent that intuitions have at least some tie to the truth. The pos-
sibility that all or even most of our intuitions are mistaken does not warrant serious
consideration. No one is worried about the possibility that murdering innocents for
pleasure is in fact the summum bonum; nor is there any reason to believe that the term
‘consciousness’ in fact correctly applies to all objects within 100 meters of the Eiffel
Tower. In a less silly vein, there appears to be little variation on intuitions regarding
certain central cases of philosophical categories like knowledge. Almost all epistemo-
logical theories agree in dismissing random guessing as a source of knowledge—and
empirical evidence from Weinberg et al. (2001) suggests that there is cross-cultural
agreement on such cases as well.
For several reasons, then, the most attractive stance on intuition seems to be that it
is significantly less epistemologically respectable than the most optimistic among us
might have hoped, but still broadly reliable and capable of providing evidence for or
against certain philosophical hypotheses. As noted in the Introduction, we will call
this position ‘moderate intuitionism’.
4.3. Moderate Intuitionism: In Search of

Theoretical Explanation
Moderate intuitionism is attractive, but it needs explanatory support. Why should
there be a link between intuition and truth at all? In the case of perception, we have a
fairly thorough theoretical understanding of why beliefs based, e.g., on vision should
be more or less reliable. We can tell a story about the causal mechanisms by which
looking at dogs produces the belief that there are dogs. The story begins with light
reflected from dogs hitting cones on the retina, and continues with edge detection,
the segmentation of the scene into objects and their parts, and so on. The details of
the story also explain how vision sometimes produces beliefs about dogs in situations
where there are none. Given the fact that illusory dogs arise only rarely, the story pre-
dicts that visually-based dog beliefs will be generally reliable. Of course, the story is not
complete—there are gaps in our understanding, e.g., at the level of consciousness. But,
importantly, there’s an account which explains more or less why—and how—the pres-
ence of dogs tends to lead to true dog beliefs.
Our understanding of the means by which vision produces true beliefs is thus rea-
sonably good. In other cases, as with memory, we do not yet have as many details, but
we can conceive of the general form an account would take. With intuition, however,
the situation is otherwise. It’s not at all obvious how to explain the link between, for
example, knowledge and our intuitions about knowledge. Indeed, this ignorance has
itself occasionally been fodder for the anti-intuitionist argument, as when Jonathan
Weinberg notes that the “hopefulness” of intuition as an evidential source is under-
mined by the fact that “We just do not seem to know much about the underlying psy-
chology of the propositional seemings that we term ‘intuitions’ ” (Weinberg 2007,
p. 336). Tellingly, the most persuasive defenses of intuition—such as those of Sosa,
Bealer, Pust, and Williamson mentioned above—are in a sense ‘negative’. They argue
against anti-intuitionism, rather than in favor of intuitionism; they simply advert to
the practical impossibility of abandoning intuition, or to the fact that such a move
is unwarranted given the evidence at hand. Positive defenses of intuition, based on
explanations of how intuitions actually produce true beliefs, are occasionally offered—
however, in many cases, the explanations are ultimately unsatisfying.
When defenders of intuition do attempt to provide positive characterizations of the
nature of the intuitive process, they frequently employ somewhat metaphorical expla-
nations in terms of ‘perceiving’ or ‘grasping’ the truth of a proposition. For instance,
Laurence BonJour claims that, when one has an intuition, one is “able to see or grasp or
apprehend in a seemingly direct and unmediated way that the claim in question cannot
fail to be true” (BonJour 1998, p. 101). Another popular term is ‘understanding’—Sosa’s
view is that “we manifest a competence that enables us to get it right on a certain subject
matter, by basing our beliefs on the sheer understanding of their contents” (Sosa 2007,
p. 102). Standardly, an intimate link with ‘reason’ is supposed to be involved—terms
such as ‘rational insight’ (BonJour) and ‘rational seeming’ (Bealer) have been offered to
describe the intuitive process. In all the above cases, however, the phraseology employed
does not give us much in the way of explanation of the particular mechanisms by which
such capacities supposedly operate. How does one come to grasp or understand the truth?
What process underlies rational seeming, such that those seemings tend to be true?
The following is a common and potentially promising idea for unpacking the above
explanations: intuitions tend to be true because the reliability of intuition is guaran-
teed by the existence of some sort of constitutive relationship between intuitions on the
one hand and meanings or concepts on the other. Alvin Goldman, for instance, has
argued that “it’s part of the nature of concepts . . . that possessing a concept tends to give
rise to beliefs and intuitions that accord with the contents of the concept” (Goldman
2007, p. 15). Similarly, Bealer writes that intuitions possess a “strong modal tie to truth,
[which] is simply a consequence of determinately possessing the concepts involved” in
the intuition (Bealer 1996, p. 2). And Frank Jackson claims that “the business of consult-
ing intuitions about possible cases is simply part of the overall business of elucidating
concepts” (Jackson 1998, p. 33). The idea, then, is that intuitions are intimately tied up
with concepts in a way that guarantees that intuitions will reflect truths involving those
concepts. The existence of a constitutive link of this sort, if adequately explained, could
provide the theoretical support for moderate intuitionism that we’re seeking.
The basic idea of a constitutive tie between intuitions and concepts can, however,
be expanded on in several ways. On Goldman’s view, for instance, “possessing a con-
cept makes one disposed to have pro-intuitions toward correct applications and
con-intuitions toward incorrect applications—correct, that is, relative to the contents
of the concept as it exists in the subject’s head” (Goldman 2007, p. 15). In other words,
intuition reveals psychological facts: your intuition that x is a case of knowledge is evi-
dence that x falls under your knowledge concept, where ‘concept’ is used in a narrow
psychological sense to indicate a particular mental representation which “is fixed by
what’s in its owner’s head” (Goldman 2007, p. 13).
Although the truth of Goldman’s account would guarantee that intuitions are evi-
dence regarding psychological facts, it would not automatically guarantee that intui-
tions are themselves generally true or generally lead to true beliefs.2 This would be
a separate, further claim; one which Goldman does not make. It is compatible with
Goldman’s account to claim that an intuition of the form ‘the Gettier case is not a case
of knowledge’ provides evidence for a certain psychological fact, while failing to pro-
vide evidence that the Gettier case is not a case of knowledge. Whether or not this failure
arises depends on our account of the truth conditions of intuitions, and whether those
truth conditions are determined by something other than the psychological concept
that produces the intuition. So at the very least, Goldman would need to make supple-
mentary claims in order to defend a move from ‘intuitions are evidence for facts about
2 Whether or not it makes sense to call an intuition ‘true’ will depend on one’s particular view of intui-
tions. If intuitions are beliefs, then they can be true or false. If they are inclinations to believe, or intellectual
seemings, or something similar, it may make more sense to merely say that they produce true beliefs without
being themselves truth-evaluable.
one’s personal concept of knowledge’ to ‘intuitions are evidence for facts about knowl-
edge’. Because moderate intuitionism is a claim that intuitions are generally reliable
tout court, rather than just reliable indicators of the nature of personal psychological
concepts, we’ll leave Goldman’s account to the side.3
Bealer’s version of the constitutivity approach, on the other hand, is
non-psychological. Bealer has suggested that intuitions possess a ‘strong modal tie’ to
the truth—necessarily, intuitions are true most of the time, or under normal circum-
stances. This strong modal tie is itself cashed out in terms of ‘determinate concept pos-
session’, which is the sort of concept possession one has when one possesses a concept
without ‘misunderstanding or incomplete understanding’ (Bealer 2000, p. 11); how-
ever, Bealer clearly means to use ‘concept’ to express something graspable by multiple
persons, rather than something individual an.d purely psychological. Unfortunately,
Bealer provides no explicit argument for the claim that humans in fact ever attain such
determinate concept possession; instead, he tends to rest with the claim that there is no
intrinsic barrier to achievement of determinate concept possession.
It’s clear, however, that on Bealer’s account, determinate concept possession is made
possible at least in part by the fact that philosophical terms, unlike natural kind terms,
are ‘semantically stable’—that is, knowledge of their conditions of application does not
require any contingent knowledge about the speaker’s external environment. Conversely,
“an expression is semantically unstable iff the external environment makes some contri-
bution to its meaning” (Bealer 1996, p. 23). Thus, the constitutive tie between intuitions
and concepts seems to involve a claim about the meanings of terms. If semantic stability
is a feature that enables determinate concept possession, and thereby a strong modal tie
to the truth, this suggests that our intuitions about philosophical terms and their applica-
tions are reliable in virtue of certain metasemantic facts. Though we need not be commit-
ted to Bealer’s particular account of the constitutive link between intuition and truth, the
idea that said link involves metasemantic facts is promising. Let’s pursue it.
4.4. Metasemantics and Constitutivity

As mentioned above, one way that intuitions might have a constitutive tie to the truth
is via their relation to metasemantics. Metasemantics is the theory of why expressions
mean what they do (i.e. in virtue of what do they mean what they do?), rather than some-
thing else, or nothing at all. Now suppose, presumably contrary to fact, that the correct
metasemantic view is one according to which the facts about meaning are fully deter-
mined by intuitions. If the connection were particularly direct, in a sense to be explained
presently, then there would be a neat story to tell explaining the reliability of intuition.
3 Goldman offers his account as a way to defend the idea that intuitions can provide evidence for philo-
sophical claims. As part of the account, he proposes that understanding concepts in the personal psycho-
logical sense is the primary goal of philosophical inquiry. We disagree with him on this point, but will not
argue it here.
The crudest version of such a metasemantic view might look something like this:
Crude Metasemantic Constitutivity (CMC): Term T applies to object O in the
mouth of speaker S in virtue of the fact that S intuits that T applies to O.
If CMC were true, semantic application intuitions—that is, intuitions that a term T
applies to object O—would be a sort of self-fulfilling prophecy. Having the intuitions
makes the intuitions true, because the intuitions themselves ground the meaning facts
that the intuitions are about.
This model could accommodate more than just intuitions about the application of
terms, under a reasonable assumption.
Assumption: For each T in S’s vocabulary, then if there is a property P that T
expresses, P is such that:
(i) An object4 O is P iff T applies to O.
(ii) S intuits that O is P iff S intuits that T applies to O.5
Let’s take an example. If I intuit that case C is an instance of knowledge, then by (ii)
I intuit that ‘knowledge’ applies to C. So, by CMC, ‘knowledge’ does apply to C. By (i) if
‘knowledge’ applies to C, then C is knowledge, and my (non-semantic) intuition that
C is knowledge is true. Thus, though CMC is a metasemantic claim, its truth would
underwrite reliability for intuitions that are not explicitly linguistic.
What this suggests is that a close tie between intuitions and meanings, through a
metasemantic theory that connects them, as given in CMC, would be a promising
and powerful tool for connecting intuitions with the truth of their contents. There are,
however, at least three reasons to think that nothing quite this crude is going on. First,
as we’ve pointed out, intuitions vary under circumstances that don’t plausibly alter the
truth of the facts at issue (e.g. the order in which cases are presented to subjects to elicit
their intuitions). CMC, however, doesn’t admit any separation between intuition and
truth; according to CMC as stated, there’s no possibility of error in intuition (granting
the Assumption). At the very least, a less crude account would need to require that
speakers were attentive, or appropriately trained, or something similar before their
intuitions ‘count’ as determining meaning.
Second, and more seriously, there’s reason to think that the crude model given
in CMC is altogether too internalist. Our hesitation to commit to a fully internalist
conception of meaning is not merely based on intuition, but on general theoretical
grounds. The worry is well illustrated by a passage from Jerry Fodor:
. . . words can’t have their meanings just because their users undertake to pursue some or other
linguistic policies; or, indeed, because of any purely mental phenomenon, anything that hap-
pens purely ‘in your head.’ Your undertaking to call John ‘John’ doesn’t, all by itself, make ‘John’
4 Or event, or what have you.

5 An example: For T = ‘water’, there is a property P = being water, such that when you intuit something is
water, you intuit that ‘water’ applies to it, and when ‘water’ does in fact apply to it, it is water.
a name of John. How could it? For ‘John’ to be John’s name, there must be some sort of real
relation between the name and its bearer; and intentions don’t, per se, establish real relations.
This is because, of course, intentions are (merely) intentional; you can intend that there be a cer-
tain relation between ‘John’ and John and yet there may be no such relation. A fortiori, you can
intend that there be a semantical relation . . . and yet there may be no such relation. (1990, p. 98)
If, like us, you accept Fodor’s claim, then application intuitions can’t be self-ful-
filling in the way described by CMC, because something purely mental (viz., my
intuition that T applies to O) cannot in itself bring about a real relation (T’s actually
applying to O).
Finally, though somewhat paradoxically, there are well-known intuitive reasons
for thinking that intuitions just don’t bear that close a tie to meanings. That is, several
thought experiments and related actual cases (Kripke’s ‘Peano’ and ‘Gödel/Schmidt’
cases, Putnam’s Twin Earth) elicit externalist intuitions, and have been used to sup-
port externalist metasemantic theories wherein the purely mental has little say in the
application-conditions of our terms. CMC is, of course, inconsistent with this externalist
approach. So much, then, for CMC—we’ll have to look for a less direct link between intu-
ition and meaning. Perhaps one can be found in the externalist theories just mentioned.
4.5. Externalist Metasemantics and Intuition

Consider the following prompt:
You’re of course familiar with the name ‘Gödel’. Likely, you believe that Gödel
proved that arithmetic is incomplete. Suppose what I tell you now is true. There were
two men, X and Y. X discovered a proof that arithmetic is incomplete—indeed, the
same proof that frequently has the name ‘Gödel’ attached to it in articles and text-
books today. X, however, had the name ‘Schmidt’ but not ‘Gödel’ written on his birth
certificate. Another man, our Y, killed X in secret and stole his proof before X had a
chance to publish it. Y then published the proof under the name he (Y) had on his
birth certificate, which happened to be ‘Gödel’. This is why the proof now frequently
has the name ‘Gödel’ attached to it in articles and textbooks today.
Again, supposing all this to be true, what of your original belief? Did Gödel prove
that arithmetic is incomplete?
A fair number of influential philosophers are inclined to answer the question nega-
tively,6 and historically this has been taken as evidence that the intuition is a negative
one: Gödel did not prove the incompleteness theorem, if the situation is as stated in the
prompt. Further, intuitions on this case, on Hilary Putnam’s Twin Earth case (Putnam
1975), and so on, have historically been taken to provide strong evidence for certain
metasemantic theories and against others.
Beginning with Saul Kripke (1972), who formulated it.

6
Let’s give a quick example of how such arguments go. Most individuals who have
heard of Gödel likely only have heard that he proved the incompleteness of arithmetic.7
On descriptivist accounts, a name refers to whatever satisfies the mentally represented
description that the speaker associates with that name; thus, on a descriptivist account,
‘the individual who proved the incompleteness of arithmetic’ is one of the few candi-
date descriptions that can determine the meaning of ‘Gödel’. However, this predicts
that in the scenario described in the prompt, Gödel did in fact prove that arithmetic is
incomplete—for it is a tautology that the individual who proved the incompleteness of
arithmetic proved the incompleteness of arithmetic. The prediction of descriptivism is
thus at odds with the intuition, and this is interpreted as evidence that descriptivism is
false.8
Note that the intuitions on such cases aren’t taken merely as crudely suggestive, but
as fine-grained tools for getting at the metasemantic reality. For example, Gareth Evans
(1973) introduces several examples that are intended to show that Saul Kripke’s sug-
gested model can’t be right, even though it comports with our intuitions most of the
time. One such case involves the reference of ‘Madagascar’. On Kripke’s account, ref-
erence is determined by an initial baptism, and maintained by causal links between
speakers through which reference is transmitted. However, the landmass that was
initially baptized ‘Madagascar’ is in fact a portion of the African mainland, and thus
Kripke’s account inaccurately predicts that ‘Madagascar’ refers to that region. The
attention paid to such counterexamples suggests that getting the intuitions right ‘most
of the time’ isn’t enough; you have to get all the intuitions right, or at least all the ones
that seem fairly systematic. But why should any of this be so? Why think that metase-
mantic intuitions of this sort are reliable guides to metasemantic facts?
Normally in science, we use theories to generate predictions (with the aid of cer-
tain background assumptions, etc.), and we design experiments to test those predic-
tions; the results of those experiments can then potentially confirm or disconfirm the
theory. But consider the application of this idea to externalist metasemantic theories.
Kripke’s theory of reference—the view that name N refers to the object O that stands
at the end of a certain historical chain beginning with a baptism event and succes-
sive N-inheritance from speaker to speaker—does not at all predict that anyone will
have intuitions that accord with it. There’s nothing contradictory about Kripke look-
ing at Evans’ ‘Madagascar’ case and saying “No, ‘Madagascar’ refers to the portion of
7 As a reviewer points out, many will only have heard that he is a logician: of course, this is even more
problematic for a descriptivist.
8 Even defenders of descriptivism have accepted the evidential force of the intuition. For example, vari-
ous philosophers have attempted to outline a version of descriptivism that makes predictions in accordance
with these and other thought experiments in the literature (e.g. Katz 1994). Such a version might say that the
description associated with ‘Gödel’ that determines the latter’s referent is ‘the individual upon whose birth
certificate appears the name “Gödel’.” In this case, Gödel would not have proven the incompleteness of arith-
metic in the scenario described, since the individual upon whose birth certificate appears the name ‘Gödel’
in the scenario did not prove the incompleteness of arithmetic in the scenario. It’s worth noting that such a
view is clearly designed to comport with the intuitions.
the mainland baptized as such. That’s what my theory says.9 What’s intuition got to do
with it?” So the fact that our metasemantic intuitions do or don’t accord with Kripkean
theory seems irrelevant to whether or not we have confirmation of it. The argument
from intuition to Kripkeanism seems to be missing a crucial step—one which links the
theory with the intuitions.
It is, in fact, not obvious how such a link might be provided in the absence of some
kind of constitutivity hypothesis. The intuition in question involves a claim about
the reference relation; there’s been a temptation, therefore, to claim that the intui-
tion reflects the subject’s theory—we might even say her concept—of reference. Such
a theory of reference might be either explicit, or tacit. The explicit option, of course, is
unhelpful; if one’s explicitly-held metasemantic theory is causing the intuition about
which individual ‘Gödel’ refers to in the prompt, then the ‘intuition’ is just a statement
of the consequence of one’s philosophical theory. The other option involves the idea
that individuals possess something like a tacit ‘folk metasemantics’ that allows them,
when given a non-semantic description of a situation, to produce a judgment about
what the semantic facts are. But the problem remains: why think that such judgments
provide evidence about the metasemantic facts themselves?
Consider a parallel. People have natural intuitions about how non-animate objects
will move when subjected to various forces. The tacit principles associated with these
intuitions are known collectively as ‘folk physics’, and they appear to roughly resemble
the principles of the (false, discredited) medieval impetus theory. Obviously, no physi-
cist is of the mind that physics should strive to capture either the content of folk physics
or the intuitions it generates. Physicists are concerned with scientific theories about
how actual objects subject to actual forces behave, not psychological theories regard-
ing how people think they will behave. Folk physics is irrelevant.
Much the same could arguably be said for semantics. Attributing meaning or
reference to natural language expressions serves a scientific, predictive, explana-
tory purpose. The assigned contents have to meet certain constraints of which most
individuals are not aware: they need to be objective enough to be communicable;
context-independent enough to be compositional; fine-grained enough to capture cer-
tain logical relations, etc. So, if tacit folk metasemantics drives judgments on the Gödel
case, this seems of little moment—why does folk metasemantics have anything more
to do with metasemantics than folk physics has to do with physics? Metasemantics is
an empirical theory about why expressions mean what they do, rather than something
else or nothing at all; it is not a psychological theory about why people think things
mean what they do.
In general, there’s no obvious reason to think that ‘folk theories’ are a good guide to
true theory. So we still have no reason to suppose that, for externalist metasemantic
9 Well, this would be somewhat unlikely, given that Kripke claims not to be offering a theory. For our pur-
poses, we’ll adopt the convenient fiction that Kripke’s remarks on the metasemantics of names and natural
kind terms constitute a theory that he endorses.
accounts, our intuitions about reference are anything less than independent of the
facts about reference. As the case of folk physics shows intuitions can, quite generally,
‘float free’ from the facts. Perhaps evolutionary considerations suggest that the intui-
tions must line up to some extent with the facts, but again, as the case of folk physics
shows, this requirement need not be very substantial.
If the folk metasemantics approach is to provide the link between intuition and
theory, then one needs to claim that metasemantics is in some kind of epistemically
privileged position. This is, in fact, prima facie tempting—for, while physics involves
external phenomena, metasemantics is in a sense about us. The use of language is a
human activity, and so it might seem that humans would have special insight into its
workings. One might try to make a parallel with reliance on grammatical intuitions in
linguistics;10 in linguistics, after all, it’s common to assume that there is an intimate tie
between grammatical intuition and grammatical truth, one which simply arises from
the nature of language (in other words, a constitutive tie). So the analogy with gram-
mar might lead us to propose some sort of constitutive link—not directly between
intuitions and meanings, but between folk metasemantics (which generates intui-
tions) and meaning.
However, with regard to the grammar analogy, a couple of points are in order. First,
a close tie between represented grammar rules and grammar facts need not preclude
a divide between grammar intuitions and grammar facts: multiple center embeddings
(e.g. “A theory that a philosopher that a grant agency awarded money endorses is true”)
seem unacceptable, but are in accord with English grammar (on standard accounts).11
More importantly, though, it’s relatively clear why grammaticality should be deter-
mined by internal rules and representations—because its function is to explain, among
other things, the learnability of languages. The fact that an infinite set of expressions
is learnable from a finite, highly impoverished set of stimuli is explained by a severe,
innate restriction on the class of potential grammars. If the principles restricting pos-
sible grammars were not internal to the mind, they couldn’t do the explanatory work
required of them. These principles (with their learned parameters set) determine what
is grammatical (but of course not what is parsable). Thus grammaticality is internal (if
only partly accessible through intuition).
On the other hand, it’s far less plausible to suppose that what semantics is supposed
to explain is a purely internal affair. Representation is a relation between us and the
world. And as Fodor urged in the last section, such relations can’t hold “just because
their users undertake to pursue some or other linguistic policies; or, indeed, because of
any purely mental phenomenon, anything that happens purely ‘in your head’ ” (1990,
10 Cf. Stich 1996, pp. 41–2.

11 The reason things work out like this in our linguistic theories is that grammaticality is a feature that’s
supposed to explain the learnability of languages, and the evidence suggests that our language parsers (the
generators of our intuitions of acceptability) only employ fallible heuristics for detecting grammaticality.
(It’s easier and faster to parse when you allow yourself to get certain systematic classes of cases wrong.)
p. 98). Merely representing a folk semantics does not bring it to pass that meanings are
determined by that folk semantics.
If this is right, then there’s reason to doubt the hypothesis of a constitutive link
between folk metasemantics and metasemantics itself. Worse, though, even the exist-
ence of such a link wouldn’t really support standard causal metasemantic accounts—
because, as we’ll discuss presently, such accounts don’t even capture the metasemantic
intuitions that were supposed to motivate their acceptance.
4.6. From Causal to Dispositional Metasemantics

Supposing we did represent a tacit metasemantic theory of a broadly Kripkean sort,
one would expect to find that our intuitions about reference track the predictions of
Kripkean theory rather closely. But, in fact, they depart from such predictions in very
significant ways. For example, it seems intuitive that descriptive terms or languages
are possible. Suppose that English and Cantonese differ in the following respect: in
English, name N refers to object O iff N is acquired through an anaphoric chain that
terminates in an ostensive or descriptive baptism of O with N. In Cantonese, N refers
to O iff O is the unique object that satisfies a description D that speakers associate with
N. It strikes us that this is possible. Now, suppose that our intuitions are right and it is
in fact possible. Then, the fact that name-referent pairs <N, O> in English stand in an
anaphoric-baptism relation would be an interesting generalization about English, but
not an explanation of why N bears the reference relation to O. Pairs <N, O> in other
languages might stand in the selfsame relation, but nevertheless not be referentially
related.
There are in fact plausible real-world examples of descriptive terms: for instance,
the empty expression ‘phlogiston’. Phlogiston was a substance that the alchemist and
physician J. J. Becher implicated in the combustion and rusting of materials. Oxygen is
crucial for both of these processes, and was arguably the substance with which Becher
was confronted when introducing the term. Yet ‘phlogiston’ is universally taken not
to apply to oxygen, but rather to have an empty extension—and even, according to
Kripke, a necessarily empty extension. We seem to take the descriptions associated by
Becher with ‘phlogiston’ to be constitutive of its nature: since oxygen is not released in
combustion or in oxidation, phlogiston is not oxygen. It’s apparently more important
to us whether the substance Becher postulated had the properties he postulated for it
than whether there is a single substance involved in the processes he identified. Our
intuitions here run counter to the predictions of causal accounts.
Another well-known anti-Kripkean intuition: the original baptismal event can be
irrelevant to the current meaning of an expression (even in cases where there’s no
intentional reference shift, as when I call my dog ‘Aristotle’), as can the descriptions
under which objects or kinds are baptized. In Evans’ ‘Madagascar’ case, although a
portion of the mainland of Africa is the baptized location that stands at the end of
our ‘Madagascar’-chain, and although at no point did anyone involved in the chain
intentionally shift the reference to Madagascar, still it is the island and not a portion of
the mainland that is properly so-called now. Since where we’ll be taken if we ask to go
to ‘Madagascar’ is more important to us than what was originally called ‘Madagascar’,
‘Madagascar’ means the former, and not the latter.
Similarly, ‘jade’ was introduced presumably with about the same understanding with
which ‘water’ and ‘gold’ were introduced: as a putative substance kind. Jadeite and neph-
rite have similar, though not identical, appearance, hardness, beauty, and economic
value. When it was discovered in the nineteenth century that the two types of jade were
separate minerals, it was open to us to reject ‘jade’ as a classifier or to use it to apply to just
one of the minerals. However, it was more important to us to have a term that applied to
minerals of a certain appearance and value than it was to cut solely at the joints of nature.
Compare fool’s gold, which we had no interest in classing with the element Au.
Can we capture this group of intuitions, as well as the externalist intuitions men-
tioned previously, with a single metasemantic account—while retaining the idea of
a constitutive link? One possibility: an account that runs not through a single tacit,
broadly Kripkean metasemantic theory, but instead through a variety of less unified
represented rules. Let’s look at an example. Keith Donnellan (1993) argues that the
“force”12 of our intuitions in Twin Earth cases is strong evidence that we are follow-
ing a hard-to-consciously-access, internal-to-the-mind “semantic rule” that leads us
to infer from the premise that something shares the underlying nature of the para-
digm cases of water to the conclusion that it is water (pp. 157–8). Donnellan then urges
something like the following constitutive connection between semantic rules and
metasemantic facts:
Donnellan’s Metasemantic Constitutivity (DMC): (For any T, R, O, S) if term T bears
relation R to object O, then T applies to O in the mouth of speaker S in virtue of the
fact that S follows an internal semantic rule to infer from the claim that T bears R to
O to the claim that T applies to O.
Donnellan concludes that “there may be a sense in which what is ‘in our heads’ deter-
mines the extension of a term such as ‘water’ ” (Donnellan 1993, p. 158). The idea here
is that semantic rules (for example, the rule to infer from the claim that the liquid in
lakes and rivers around here was baptized ‘water’ to the claim that ‘water’ applies to the
aforementioned liquid) ground the metasemantic facts. Thus our intuitions in Twin
Earth cases are reliable, because the semantic rule we follow that nothing lacking the
underlying nature of the paradigm cases of water is water makes it true that no such
thing is water. The rule at once generates the intuition and grounds the fact that the
12 Curiously, Stephen Stich (1996, p. 47) argues from the fact that for many prompts, we don’t have firm ref-
erence intuitions, to the conclusion that our internally represented metasemantic theories underdetermine
many cases. The analogous conclusion here would be that although the force of Twin Earth intuitions might
establish that we represent and follow a semantic rule for ‘water’, the lack of force in many other, similar cases
shows that not every expression is governed by a semantic rule (or that some are governed by indeterminate
rules).
intuition is about. This position solves the problem of how an ‘externalist’ metaseman-
tic theory can be connected with intuition: the ‘externalist’ theory itself is grounded by
an internal representation of it.
But notice, too, that DMC could at least in principle account for the anti-Kripkean
intuitions just discussed. Descriptive terms and languages are possible according to
DMC, because it’s possible for expressions to be governed by descriptive semantic
rules. So if ‘phlogiston’ has a descriptive semantic rule governing it, we could explain
why there isn’t any phlogiston; and if ‘jade’ is governed by a semantic rule, like ‘ “jade”
refers to the economically interesting kind encompassing such-and-so paradigm
cases’, we could explain jade’s current disjunctive status; and similarly if ‘Madagascar’ is
governed by a semantic rule like ‘ “Madagascar” refers to the place you get taken when
you ask to go to “Madagascar” ’. So we have in DMC a constitutive account with the
flexibility to account for any metasemantic intuition one happens to have, by postulat-
ing a represented rule that underlies that intuition.
However, we want to argue against this particular version of constitutivity for two
reasons. First, as we’ll argue in section 4.7, it’s implausible that any of the terms just
discussed are or were actually governed by the semantic rules just hypothesized. But
more fundamentally, as we have already emphasized more than once, purely inter-
nal things like semantic rules don’t plausibly engender, by themselves, representation
relations. What’s needed is something that at once has the externalist virtues of the
anaphoric-baptismal account and the internalist, constitutivity-granting virtues of the
semantic rules account.
Here’s our attempt at embodying these virtues, through a somewhat less direct
approach to constitutivity. The proposed account,13 which we’ll call a ‘dispositional’
theory of reference, is as follows:
Dispositional Constitutivity (DC): A linguistic expression E means some object,
property, kind, relation, etc., X, in the mouth of speaker S, in virtue of the fact that S
would be disposed to apply E to X if S had all the relevant information.
‘Relevant information’ consists of the facts F that would, were S to be apprised of
F, influence S’s dispositions to apply E.14 The inclusion of the ‘relevant information’
clause creates a gap between one’s current disposition to apply a term and that which
13 We would like to emphasize that this is not an attempt at a naturalization of intentionality: the account
contains clearly intentional terms like ‘apply’ and ‘relevant information’ (defined in terms of ‘apprising’ S).
14 This may even include semantic facts; for instance, one’s disposition to apply the term ‘dog’ may be
influenced by semantic facts about what the term ‘dog’ means in the mouths of experts. However, when
determining that an expression E means X in the mouth of speaker S, one fact that should be excluded from
the relevant information is the very fact being determined by that information, namely, ‘expression E means
X in the mouth of speaker S’. We think this exclusion is motivated, since the fact in question is not ‘indepen-
dently grounded’. When a set of facts grounds another fact, the latter fact cannot be in the grounding set;
things do not ground themselves. Thus, the fact that expression E means X in the mouth of speaker S cannot
be part of the set of facts, knowledge of which determines the dispositions that ground the meaning of E in
the mouth of S.
determines the meaning of the term. Thus, the fact that you are disposed to apply ‘cow’
while viewing a horse on a dark night is not determinative of the meaning of ‘cow’;
for, if you were apprised of certain facts, you would no longer be so disposed. Looking
ahead, ‘relevant information’ often involves the sort of information provided in the
course of a thought experiment—e.g., the information that the watery stuff on Twin
Earth is not chemically identical to the watery stuff on Earth.
The basic motivation behind the dispositional proposal is this. Fodor argued that
nothing purely mental could by itself establish semantic relations. But we need not
consider the purely mental by itself: “linguistic policies don’t make semantic relations;
but maybe they make causal relations, and maybe causal relations make semantic rela-
tions” (1990, p. 99). Maybe we do represent tacit metasemantic theories, or semantic
rules, or whatever. Those things, in themselves, can’t determine what means what. But
they might dispose us to behave in certain ways: accepting the semantic rule ‘ “water”
applies to the liquid in lakes and rivers around here’ might dispose us to withhold
‘water’ applications from substances we know to not be the liquid in lakes and riv-
ers around here. Such a disposition isn’t purely mental: it’s a real relation between our
application behavior and the extra-mental events of the extra-mental world. And
according to the account on offer, this disposition (subject to the relevant information
clause) establishes a semantic relation between ‘water’ and H2O.
What we’ve just given is only the barest possible sketch of a metasemantic account.
But we’re not particularly interested in defending an account of reference; instead,
we’re interested in outlining a method for elucidating a constitutive account of the evi-
dential status of intuition, without the need for an internalist metasemantics. What
we will try to show here is that something roughly like the account on offer is true,
and preferable to either a straight-up anaphoric-baptismal account, or to a semantic
rule account like DMC. This is a good thing, because if it’s true, we’ll have an explana-
tion that can underwrite moderate intuitionism. The principal idea, to be elaborated
upon, is that our judgments (intuitions) regarding the application of terms given cer-
tain prompts are reliable but fallible indicators of our dispositions to apply terms in the
circumstances described by those prompts. The latter dispositions are constitutively
linked to the metasemantic facts15—to which the terms correctly apply—and thus our
intuitions are reliable but fallible indicators of that to which our terms correctly apply.
4.7. Some Relevant Details

It’s important that we establish that the dispositional account just offered, or some-
thing roughly similar, is true, or roughly true. If it isn’t, then it can’t explain intuition’s
15 Only when, of course, the prompts specify enough relevant information—that is, when our disposi-
tions when given the specified information are not substantially different from what our dispositions would
be when given all relevant information. This will of course be a matter of degree, as will the strength of the
constitutive link.
connection with the truth. And our goal here is to explain intuition’s (fallible) connec-
tion with the truth. That being our goal, however, we can’t spend too long on a defense
of our theory, or we’ll never get to what we want to do with it. So we propose to just
review how it handles the problem cases for the anaphoric-baptismal account, and in
so doing illustrate how we intend it to work.
According to the dispositional theory, the reason why descriptive names, descrip-
tive kind terms, or other descriptive expressions E are possible is that it is possible
that a speaker S be disposed to apply E to objects only when S takes it that they satisfy
a certain description D. S may apply E to objects that don’t satisfy D, under the mis-
taken assumption that they do; however, were S given information that distinguishes
the objects that don’t satisfy D from the ones that do (that is, were S not to be mistaken
about them), she would no longer be disposed to apply E to them. Thus, E applies, in S’s
mouth, to all and only objects satisfying D. On this account, it is entirely possible that
the relevant description and its integral role in the meaning of E was never represented
as a semantic rule by S. S might well accept a non-descriptivist tacit semantic rule like
‘ “phlogiston” applies to the most natural substance in the vicinity of paradigm exam-
ples X, Y, and Z’. And maybe oxygen was just such a substance. But for us, none of that
is relevant: if S is truly disposed to reject the application of ‘phlogiston’ to anything
upon learning that its release from an object does not cause oxidation (say), and no
other relevant information could change her mind, then ‘phlogiston’ is empty.
Consider an alternate set of dispositions. It might be that though S now rejects the
application of ‘phlogiston’ to anything, she would apply ‘phlogiston’ to oxygen, were
she to learn more of the history of phlogiston theory, or of the behavior of oxygen, or
whatever. In such a case, the individual would be now lacking relevant information, in
the sense of lacking information that would influence her dispositions to apply ‘phlo-
giston’: thus her current lack of a disposition to apply ‘phlogiston’ to oxygen would be
only a fallible indicator of her disposition to apply ‘phlogiston’ to oxygen, if she had all
relevant information.
As indicated by the examples earlier, our intuitions often seem to track, not causal
factors, but simply distinctions that we find important to make—and what makes such
distinctions important can vary from case to case. ‘Jade’ provides a paradigm example.
‘Jade’ is problematic for causal accounts; it’s presumably introduced as a natural kind
designator, but it isn’t a natural kind. The baptism is defective, yet ‘jade’ is non-empty—
it means either jadeite or nephrite. Donnellan’s semantic rule account doesn’t clearly
provide a good account for jade, either—for it’s implausible that jade is disjunctive
because of some explicitly represented ‘escape clause’ in its semantic rule that called
for disjunctive contents when unified ones weren’t to be had.16 Instead, the explana-
tion simply seems to be that the economic and cultural importance of jade trumps our
joint-carving interests in this particular case.
Pace Bealer (2002), who makes just such a claim.

16
Again contra Donnellan, it’s highly implausible that this fact about our interests
was ever embodied in a tacit semantic rule. But the dispositional account gets the
case right; ‘jade’ refers to jadeite or nephrite simply because, even knowing these to
be distinct minerals, and even under circumstances where we can distinguish them,
we are still disposed to call both ‘jade’. Our dispositions reflect certain highly contin-
gent social and economic facts. Similarly, although tea is closer to chemically pure
water than the stuff in the Hudson River, the latter but not the former is water. Again,
this is just how we choose to call things, given our fairly contingent and idiosyncratic
interests, and we are not moved by what was baptized what, or what is more ‘kindy’,
or whatever.
Frequently, how others are disposed to apply terms is relevant to how we are dis-
posed to apply them—we defer to experts (more generally, to others). The details of
this deference are quite messy. Geoff Nunberg once17 gave the following example: ‘carp’
means something different to each community—as you go from pond to pond, dif-
ferent fish are called ‘carp’ and no one is motivated by considerations of what other
speakers a town over call ‘carp’ to change their practice. The case of the color term
‘puce’ is somewhat different. Many of us would be likely to reconsider our application
of that term upon encountering disagreement from another English speaker. However,
Americans are apparently not as motivated by what French speakers have to say on the
matter, even though the term was originally borrowed from the French—in the United
States, ‘puce’ is a purplish brown, while in France it is a dark reddish brown. ‘Liberty’
seems different still: a Frenchman’s use of ‘liberté’ may well influence how I decide
to apply ‘liberty’. In the limiting case, I defer to no one: no information about other
speakers’ usage can sway me. These are the terms of my private language. Wittgenstein
thought this to be impossible, but we don’t: in such a case, my expressions are true of
that to which I’m disposed to apply them (when I’m not in error about what the things
are that I’m applying the expressions to). Our account derives semantic values deter-
mined by “deference to experts” as a special case. Each of us decides which experts are
relevant to the application of our terms, and our terms apply to whatever we would
apply them to, when given the relevant facts about expert applications.
As we’ve noted, the standard causal-historical picture is apparently limited in scope: it
applies sometimes (to ‘water’) but not always (to ‘phlogiston’ or ‘jade’) and not neces-
sarily (‘water’ could have meant watery stuff). Why does it apply to some natural kind
terms and not to all such terms, or to other kinds of terms? Again, one possible story is
Donnellan’s DMC: that with each term we associate a semantic rule that grounds the
meaning determining facts for that term. So if the semantic rule instructs us to apply
‘water’ to the local stuff baptized ‘water’, then that’s what is water. But this view is unsatis-
fying: it’s implausible that ‘phlogiston’ is empty because of being governed by a ‘descrip-
tive’ semantic rule—why should ‘phlogiston’ have been assigned a descriptive rule,
while ‘water’ received a causal rule? But again, the dispositional account can handle this
At a talk attended by one of the authors.

17
case, for it’s much more plausible that our dispositions were influenced in various subtle
ways by contingent facts regarding the history of science, or what have you.
Our interests dispose us to go one way or another in various cases, and we’ve never
been wrong following our dispositions when we had all the relevant information. It just
couldn’t turn out that even though we know the whole story about jade, and we’re con-
tent to call both jadeite and nephrite ‘jade’, nevertheless, ‘jade’ means in our mouths,
right now, just jadeite. This strongly suggests a close connection between our dispo-
sitions to apply terms (under full information) and what those terms mean. What’s
more, the disposition story isn’t perniciously internalist, like the story about semantic
rules: a disposition to apply a term to a thing is a real relation between the term and the
thing, just as a disposition to sleep during lectures is a real relation between a student
who has it and the lectures he attends. So, we propose, the dispositional account is a
good place to look for a theory of reference that can support moderate intuitionism.
4.8. The Dispositional Theory and Moderate

Intuitionism
As we mentioned earlier, we think that a broadly dispositional metasemantics can pro-
vide the needed link between intuitions and truth. It does this via a claim that our
application intuitions are generally a guide to what terms apply to which things. But
how exactly are our intuitions a guide to what terms apply to which things? Let’s look
at a particular example. Suppose an English speaker from 1750, George, is given the
following prompt:
H2O/XYZ: You’re of course familiar with the word ‘water’. Likely, you believe that
water is the tasteless, odorless, drinkable clear liquid that flows in lakes and streams.
Suppose what I tell you now is true. All tasteless, odorless, clear, etc. liquid on planet
Earth is composed of particular amounts of specific kinds of stuff. I could, as it were,
write down a ‘recipe’ for this tasteless, odorless, etc. stuff, and I could also write reci-
pes for everything else. The ‘ingredients’ in these recipes would be the most basic
things in the universe, of which there are about six score.
Now I know that all this sounds strange, but imagine it to be true. Indeed, stranger still,
imagine that far away there is another planet much like Earth, with verdant forests,
hills, valleys, mountains, lakes and streams. And in the lakes and streams on this other
planet flows a tasteless, odorless, drinkable clear liquid. However, this stuff, this liquid,
has a very different recipe from the tasteless, colorless, clear, etc. liquid around here.
Again, supposing all I’ve told you is true, does your word ‘water’ truly apply to the
tasteless, odorless, clear, etc. liquid on this other planet?
Let’s suppose George says “No, that wouldn’t be water.” What we want to suggest is that
George’s response is good evidence that if he were given all relevant information, he would
apply ‘water’ to H2O but not XYZ, and thus ‘water’ for him now means H2O but not XYZ.
The way this works is that the prompt asks George to suppose that the watery stuff on
Earth has a certain underlying nature. We don’t know if this information is relevant or
not in advance, but if it is, then he’s been provided with relevant information. Then the
prompt stipulates that there is a sample of stuff that has a different underlying nature.
We don’t rely on George to tell apart H2O and XYZ, we just tell him we have an instance
of XYZ. So George is mentally ‘simulating’ a case where he is apprised of (maybe) all
information relevant to ‘water’ application, where he can tell apart two samples, and
is asked to further simulate how he would apply ‘water’ in these circumstances. To the
extent that George’s simulations of how he would respond in certain circumstances
mirror how he would in fact respond, George’s application intuitions in response to the
prompt are evidence for what his terms mean.
Evidence, we say, but no certain criterion. First, there’s no guarantee that the under-
lying nature of the substances involved is the only piece of information relevant to
George’s ‘water’ applications. We (and George as well) can only guess what informa-
tion might be relevant from our own intuitions, from past experience, and from what
we think we know about what’s important. It might be that future science cares only for
teleological (as opposed to compositional) kinds, and that H2O and XYZ have one and
the same purpose.18Apprised of this information, George might lean toward applying
‘water’ to XYZ as well. As already mentioned, dispositions are responsive to a myriad
of highly contingent social, economic, and cultural factors which may not be repre-
sented in the information provided by the thought experiment prompt. Second, peo-
ple are imperfect at predicting their own behavior. I might view myself as cool under
pressure, and predict that I would be disposed to risk my life to save a child from a
burning building; my real dispositions under such circumstances might be rather dif-
ferent. Similar considerations apply to assessment of one’s linguistic dispositions.
Our metasemantic account predicts that linguistic application intuitions—intui-
tions about when a term applies to an object—will be generally reliable to the extent
that a prompt eliciting them specifies enough relevant information. Yet the account
also predicts that intuitions are fallible in many cases, such as those described above.
Of course, what we ultimately want is an account on which intuitions, including
non-linguistic intuitions, are similarly reliable-yet-fallible. Fortunately, we’ve already
discussed how to make this step. All that is required is one particular assumption,
which we will repeat here:
Assumption: For each T in S’s vocabulary, if there is a property P that T expresses, P
is such that:
(i) O is P iff T applies to O.

(ii) S intuits that O is P iff S intuits that T applies to O.
18 By ‘teleological kind’ we mean a kind defined by its purpose, such as a heart. By ‘compositional kind’ we
mean a kind defined by its material composition, such as gold.
Given this assumption, one’s intuition that ‘knowledge’ applies to case C will be just as
reliable as one’s intuition that case C is a case of knowledge.
What does this mean for philosophical intuitions? It means that the intuitions of
English speakers about, e.g., consciousness will be exactly as reliable as their intui-
tions regarding their disposition to apply the term ‘consciousness’ under possession
of all relevant information (and mutatis mutandis for French speakers, etc.). And such
intuitions can err for exactly the reasons discussed above. For example, as mentioned
above, a thought experiment may not always successfully stipulate all information that
is relevant to determining an individual’s dispositions to apply a term. The course of
future neuroscience and psychology could quite plausibly affect our dispositions to
apply the term ‘consciousness’; insofar as thought experiments don’t specify such facts
about the future scientific world picture, they leave room for error.
4.9. Clarificatory Issues
Before we conclude, there are a few necessary points of clarification. First: the dispo-
sitional account is not directly constitutive, in the following sense. On our account,
intuitions are not constitutively related to meaning facts—dispositions are. But there
is still a very intimate (but non-constitutive) link between intuitions and truth, for the
simple reason that there is an intimate (but non-constitutive) link between intuitions
and the dispositions that do underwrite facts about meaning.
Our emphasis on dispositional facts as the ground for metasemantic facts means
that we are neither under pressure to recognize any sort of special conceptual compe-
tence underlying intuition, nor any sort of a priori insight into meanings. This distin-
guishes our account from certain superficially similar accounts, such as that offered
in Chalmers and Jackson (2001). David Chalmers and Frank Jackson propose that we
have a priori access to certain facts about meaning, in the form of ‘application condi-
tionals’. One such conditional, for example, might be G →~K, where G is a description
of a Gettier scenario and K is a claim that a certain mental state of that scenario’s pro-
tagonist falls under the extension of knowledge. More generally, they claim, we have
a priori access to conditionals of the form E → T, where T is a statement character-
izing the extension of a given term, and E is sufficiently detailed information about a
given possible world—thus, we know a priori that if some situation E is actual, then the
extension of term T is such and so.
However, on our account, no such a priori access is implied. The facts about disposi-
tions that ground reference are empirical facts; they just happen to be facts to which
we have particularly direct epistemic access, in the same sense that, e.g., we have fairly
direct access to features of our own personality. But our knowledge of such facts is a
posteriori, and is not in any way a special sort of ‘conceptual knowledge’ with a spe-
cial epistemic status. Further, unlike Chalmers and Jackson, we don’t recognize any
special second ‘dimension’ of meaning (A-intensions/primary intensions). Finally,
we’re not anywhere near as optimistic as Chalmers and Jackson about the prospects for
conceptual analysis via thought experiments—because, as mentioned earlier, thought

experiments are generally not guaranteed to specify all relevant information. Relevant
information can include deeply varied and idiosyncratic facts; in many cases, we sim-
ply won’t be able to construct a useful thought experiment without knowing a whole
lot more truths about the actual world.
It’s worth mentioning that our account is compatible with all sorts of views on the
nature of intuition. But more importantly, it’s compatible with several more or less
deflationary views on intuition. Our account doesn’t imply that intuition is the special
provenance of philosophy, for example—we agree with Williamson (2007) that there is
no principled distinction between ‘philosophical’ intuitions and everyday judgments
involving concept application. After all, our account provides the same explanation for
our philosophical intuitions as it does for our intuitions about how to apply the term
‘footstool’.
Our account is also compatible with the claim (defended by Nado, Forthcoming)
that intuition is fairly heterogeneous, in the sense that the actual psychological pro-
cesses underlying, e.g., moral intuitions are likely quite different from those underly-
ing, e.g., mathematical intuitions. The reliability of intuition isn’t found in some specific
psychological process that intuitions have in common—instead, it simply reflects the
fact that immediate reactions, e.g., to thought experiments, however such reactions
are produced, tend to provide some indication of the dispositions that underwrite
meaning. This is so, plausibly, because those psychological processes (whatever they
are) are likely to be partially determinative of the relevant dispositions—the complex
psychological processes governing folk psychology, for instance, are likely partially
determinative of how we would be disposed to apply the term ‘belief ’ when in pos-
session of full information. Where that link is weaker, intuition will be increasingly
fallible, as in the case of folk physics.
With the clarificatory notes in order, we’re now ready to summarize the argument as
we see it. Current empirical data on intuition provides us with good reason to adopt
a moderate stance on intuition, but it’s not immediately clear what explains the truth
of that moderate stance. A claim that there is a constitutive link between intuition and
meaning would support moderate intuitionism; this link could plausibly come from
metasemantic theory. Unfortunately, popular externalist theories don’t motivate the
needed link, and internalist theories are theoretically undesirable. Locating the con-
stitutive tie in dispositions to apply terms under full information solves the dilemma.
Even better, it fits remarkably well with the original variation data. Subjects in differ-
ent cultures are likely to have different dispositions to apply terms, due to the influ-
ence of various contingent social and cultural factors on the subjects’ interests and
beliefs about the world. Some of these differences are quite likely to evaporate under
conditions of increased information. Some may not. To the extent that the differences
do disappear, we can assign error to the original intuitions of one or the other group.
To the extent that the differences do not disappear, we must recognize a difference in
meaning.
With regard to the intrapersonal variation data, the situation is somewhat more com-
plex. Take as an example a study by Schnall et al. (2008) which showed that moral judg-
ments become harsher when subjects are exposed to disgusting surroundings. Thus,
intuitions appear to vary according to emotional state. The inclination is to assign error
to the emotionally heightened subjects, and our account can potentially explain why
this is so. The variability of intuition is in and of itself relevant information—plausibly,
when provided with the information that their own disposition to apply moral judg-
ments is being influenced by their heightened emotional state, subjects would have the
disposition to defer to their own moral judgments under more neutral conditions. This
is exactly analogous to ordinary cases of deference to experts. This is of course empirical
speculation, but what’s important is that it demonstrates the general method by which
the dispositional account might explain away certain variant intuitions as not reflective
of meaning-generating dispositions, and as therefore in error.
The empirical data suggest moderate intuitionism. The dispositional account, if true,
would provide an explanation for the link between intuition and truth that moderate
intuitionism requires. Given the plausibility of moderate intuitionism, we take this to
provide abductive support for the truth of the dispositional account. But in addition,
the dispositional account has plausibility in and of itself; and since it predicts moderate
intuitionism, we take this as reason to endorse moderate intuitionism. As we see it, the
two positions form a virtuous circle—each provides evidential support for the other.
And we think the overall picture, at least in broad outline, is likely to be pretty close
to true.
References
Bealer, G. (1992). “The Incoherence of Empiricism,” Proceedings of the Aristotelian Society,
Supplementary Volume 66, pp. 99–138.
——. (1996). “On the Possibility of Philosophical Knowledge,” Philosophical Perspectives 10,
pp. 1–34.
——. (2000). “A Theory of the A Priori,” Pacific Philosophical Quarterly 81, pp. 1–30.
——. (2002). “Modal Epistemology and the Rationalist Renaissance,” in T. Gendler and J.
Hawthorne (eds), Conceivability and Possibility. Oxford: Oxford University Press, pp.
71–126.
Buckwalter, W., and Stich, S. (2014). “Gender and Philosophical Intuition,” in J. Knobe and
S. Nichols (eds), Experimental Philosophy, vol. 2. Oxford: Oxford University Press, pp. 307–46.
Chalmers, D., and Jackson, F. (2001). “Conceptual Analysis and Reductive Explanation,”
Philosophical Review 110, pp. 315–60.
Donnellan, K. (1993). “There Is a Word for that Kind of Thing: An Investigation of Two Thought
Experiments,” Philosophical Perspectives 7, pp. 155–71.
Evans, G. (1973). “The Causal Theory of Names,” Proceedings of the Aristotelian Society,
Supplementary Volume 47, pp. 187–208.
Fodor, J. A. (ed.) (1990). “A Theory of Content II,” in A Theory of Content and Other Essays,
Cambridge, MA: MIT Press.
Goldman, A. (2007). “Philosophical Intuitions: Their Target, Their Source, and Their Epistemic
Status,” Grazer Philosophische Studien 74, pp. 1–26.
Haidt, J., Koller, S., and Dias, M. (1993). “Affect, Culture, and Morality, or Is It Wrong to Eat Your
Dog?,” Journal of Personality and Social Psychology 65, pp. 613–28.
Jackson, F. (1998). From Metaphysics to Ethics: A Defence of Conceptual Analysis.
Oxford: Clarendon Press.
Katz, J. (1994). “Names Without Bearers,” Philosophical Review 103, 1, pp. 1–39.
Kripke, S. (1972). “Naming and Necessity,” in D. Davidson and G. Harman (eds), Semantics of
Natural Language. Dordrecht: D. Reidel, pp. 253–355.
Machery, E., Mallon, R., Nichols, S., and Stich, S. (2004). “Semantics, Cross-cultural Style,”
Cognition 92: B1–B12.
Nado, J. (Forthcoming). “Why Intuition?,” Philosophy and Phenomenological Research.
Pust, J. (2000). “Against Explanationist Skepticism Regarding Philosophical Intuitions,”
Philosophical Studies 106, pp. 227–58.
Schnall, S., Haidt, J., Clore, G. L., and Jordan, A. H. (2008). “Disgust as Embodied Moral
Judgment,” Personality and Social Psychology Bulletin 34, pp. 1096–109.
Sosa, E. (2007). “Experimental Philosophy and Philosophical Intuition,” Philosophical Studies
132, pp. 99–107.
Stich, S. (1996). Deconstructing the Mind. Oxford: Oxford University Press.
Swain, S., Alexander, J., and Weinberg, J. (2008). “The Instability of Philosophical
Intuitions: Running Hot and Cold on Truetemp,” Philosophy and Phenomenological Research
76, pp. 138–55.
Weinberg, J. (2007). “How to Challenge Intuitions Empirically Without Risking Skepticism,”
Midwest Studies in Philosophy 31, pp. 318–43.
——, Nichols, S., and Stich, S. (2001). “Normativity and Epistemic Intuitions,” Philosophical
Topics 29, pp. 429–60.
Wheatley, T., and Haidt, J. (2005). “Hypnotically Induced Disgust Makes Moral Judgments More
Severe,” Psychological Science 16, pp. 780–4.
Williamson, T. (2007). The Philosophy of Philosophy. Oxford: Blackwell.
5
Intuition, ‘Intuition’, Concepts and
the A Priori
C. S. I. Jenkins†
5.1. Introduction
It’s not news that the word ‘intuition’ sometimes seems to be used in significantly dif-
ferent ways by different philosophers (see e.g. BonJour 2000, p. 102 and Russell 2007,
§4). Indeed, it’s not news that any even slightly interesting word sometimes seems to be
used in significantly different ways by different philosophers.
In this chapter, however, I shall attempt to put some structure on (at least some of)
the apparently different uses of ‘intuition’. This structure may be useful for understand-
ing how certain importantly different features of the things described using the word
‘intuition’ could be accidentally slurred over. In section 5.2, I argue that “intuition”-
hood1 is associated with four bundles of symptoms: a commonsensicality bundle, an a
prioricity bundle, an immediacy bundle, and a metaphilosophical bundle.
Then in section 5.3, I suggest that at least two different conceptions of intuition
should be distinguished: one which is primarily associated with the commonsensi-
cality symptom-bundle and one which is primarily associated with the a prioricity
symptom-bundle. I tentatively suggest that ‘intuition’ as used by philosophers is best
regarded as ambiguous, having (at least) a commonsensicality sense and an a prioricity
† Thanks to my audiences at University of Southern California, University of British Columbia, University

of Reading, University of Sheffield, and University of Cologne for many helpful questions and comments on
earlier drafts of this material. Thanks also to Michael DePaul and an anonymous referee, whose comments
enabled many substantive late-stage improvements. I am also particularly grateful to Jonathan Ichikawa for
extended discussion of these issues.
1 Throughout this chapter, I’ll use single inverted commas to indicate mention rather than use of a word,
and to mark quotations. I’ll use double inverted commas for scare quotes (as here). With the scare quotes
I am using rather than mentioning the word, but wish to indicate that such usage may be complicated or
problematic (for example, because of the potential for unobvious contextual shiftiness in the term used).
Every use of ‘intuition’ in this paper could be enclosed in scare quotes, but I’ll restrict them to passages where
I am particularly keen to draw attention to potential complications.
92 C. S. I. Jenkins
sense. I shall also suggest that ‘intuitions’ in both these senses are commonly associ-
ated with both the immediacy symptoms and the metaphilosophical symptoms, which
might be expected to lead to some confusion and conflation. Section 5.4 discusses a
much simpler view concerning the meaning of ‘intuition’ in philosophy, but offers
grounds for thinking this view is mistaken.
Finally, in sections 5.5 and 5.6, I’ll look at some of the attacks on intuition (or bet-
ter: things called ‘intuition’) as an epistemic source. Such attacks need to be carefully
targeted on particular conceptions of intuition before they can be successfully (or even
clearly) made out. I aim eventually to argue that one significant kind of philosophi-
cal “intuition,” related to a prioricity and conceptual truth, can be defended against a
range of typical epistemological challenges.
5.2. Four Bundles of Symptoms

First, let me clarify that I am interested in examining the kinds of core philosophical
uses of ‘intuition’ that seem to crop up in multiple areas of philosophy and metaphi-
losophy. I am not here considering the more specific ways in which the word ‘intuition’
is used in discussions of Kantian philosophy, in intuitionistic mathematics, or in intui-
tionistic logic. Although these uses may be interestingly related to those on which I am
focusing, this isn’t the place to explore such relationships.
The core uses of ‘intuition’ in which I am interested are (at least loosely) associ-
ated with certain bundles of symptoms. I don’t mean to suggest here that any or all of
these symptoms are or should be taken as characteristic or definitive of “intuitions” in
any sense (though some are, by some philosophers). I am merely claiming that these
symptoms are commonly associated with something’s being an “intuition” and/or
“intuitive.”
Bundle one contains two symptoms. The first symptom in bundle one is that of being
commonsensical, everyday and/or the opinion of the folk. One negative spin on that
symptom is provided by Timothy Williamson’s ‘[c]‌rude empiricists’ (he does not say
who they are), who ‘regard “intuition” as an obscurantist term for folk prejudice . . .’
(Williamson 2007, p. 2). In a similar vein, sometimes philosophers think of intuitions
as ‘preanalytic and untutored judgments’ (Kim 1994). A more positive spin involves
associating this commonsensicality with the second symptom in bundle one: lack of
(theoretical) contamination. Alvin Goldman (2005, p. 406) raises this second symp-
tom to salience when he says that intuiting involves ‘spontaneous application of [a]
concept uncontaminated by an intuitor’s prior theorizing, if any’.
I want to begin my discussion of bundle two by focusing on the a prioricity symp-
tom. A strong connection between a prioricity and intuitionhood is suggested by such
authors as Ernest Sosa (2009, p. 103: ‘One project of analytic epistemology is a priori
theorizing about the nature . . . of human knowledge . . . Any such practice gives prime
importance to intuitions . . .’), Goldman (2007, p. 19: ‘Defenders of intuition-driven
Intuition, ‘Intuition’, Concepts and the A Priori 93
methodology hold that intuitions provide evidence, or warrant . . . . The warrant in

question is commonly held to be of the a priori variety’), and George Bealer (1996,
p. 122: ‘Among our various theoretical beliefs, some are deemed to have a priori justifi-
cation. This occurs for beliefs arrived at by . . . (1) canvassing intuitions, . . .’).2
Given the association between “intuitions” and a prioricity, one might expect to find
that other symptoms associated with a prioricity are also associated with “intuition”-
hood. And, indeed, the second symptom in bundle two is the necessity (or at least felt
necessity) of the intuited subject matter. Bealer, for example, says that ‘if x intuits that
P, it seems to x that P and also that necessarily P’ (1998, p. 207). Something similar
is maintained by Laurence BonJour (2000, p. 102), though BonJour prefers the term
‘rational insight’ to ‘intuition’, because (like me) he believes the latter term is used in
different ways by philosophers and hence is potentially confusing.3
The third bundle-two symptom, also strongly associated with a prioricity, is that of
being either generated or epistemically underwritten (or both) by conceptual analy-
sis and/or conceptual competence and/or linguistic competence and/or understand-
ing. Williamson (2007, p. 23) writes that ‘[l]‌inguistic or conceptual philosophers treat
intuitions . . . as the deliverances of linguistic or conceptual competence’ (though he
does not say who these philosophers are), and David Papineau (2007, §2.4) says that
‘[t]he view that conceptual analysis plays no important role in philosophy might seem
to belied [sic] by the importance that philosophers attach to intuitions’. In a similar
vein, Bruce Russell (2007, §3) suggests that: ‘a person might have an intuition that a
proposition like “bachelors are unmarried” is true based on understanding the con-
cepts involved . . . ’, and Kirk Ludwig (2007, p. 135) states that he ‘will use “intuition” to
mean an occurrent judgment formed solely on the basis of competence in the concepts
involved in response to a question about a scenario, or simply an occurrent judgment
formed solely on the basis of competence in the concepts involved in it’.
Sosa (2009, pp. 104–5) has recently argued that focus on linguistic competence in
connection with intuition is misplaced, and that our understanding of propositions
is what matters. He thinks that ‘[t]‌o intuit is to believe an abstract proposition merely
because one understands it and it is of a certain sort . . .’ (Sosa 1998, pp. 263–4, my
emphasis). I am inclined to agree, though given that my current purpose is simply to
record the symptoms commonly associated with “intuition”-hood, the inclusion of an
association with linguistic competence in bundle two is nevertheless appropriate.
2 I should note, however, that Bealer explicitly claims that only some intuitions are the kind of ‘rational’ or
‘intellectual’ seemings that he associates with a prioricity.
3 BonJour identifies three senses of ‘intuition’: one on which it amounts to what he calls ‘rational insight’,
one on which it means anything which is not arrived at through an explicitly discursive process and is hence
(hopefully) pre-theoretic, and a third which is Kant’s sense. I shan’t discuss this interpretative hypothesis
further here, though it should become clear how BonJour’s first two senses relate to my symptom-bundles,
and hence to the two notions of intuitionhood that I’m trying to pin down in this chapter.
94 C. S. I. Jenkins
The fourth and final symptom in bundle two is that intuitions are sometimes taken
to be the upshot of some special faculty of ours.4 Williamson (2007, p. 2) claims that
‘[o]‌ne apparently distinctive feature of current methodology in the broad tradition
known as “analytic philosophy” is the appeal to intuition. Crude rationalists postulate
a special knowledge-generating faculty of rational intuition.’ (He does not say who the
‘crude rationalists’ are. But he is right that this association between intuition and a spe-
cial faculty is sometimes made, even if usually only by those who go on to say that it is
a ‘crude’ one.)
My third symptom-bundle consists of two immediacy symptoms, which I shall label
‘immediacy1’ and ‘immediacy2’. Immediacy1 is a matter of being direct in the sense of
non-inferential, or at least of no obvious inferential provenance. Jennifer Nagel (2007,
p. 793) is talking about this symptom when she says that ‘[t]‌he expression ‘epistemic
intuition’ is sometimes used very broadly, as a label for any immediate (or not explicitly
inferential) assessment of any claim of interest to epistemologists . . .’ So is Goldman,
when he says that ‘[a] phenomenological feature [rational intuitions] share is that
they come from “I know not where” ’ (2007, p. 11). Similarly, we have Russell (2007,
§4): ‘Perhaps “intuition” is being used [by experimental philosophers] in a broader
sense to mean “whatever seems obvious to a person on reflection, where that seeming
obvious is not based on inference” . . .’
It is not always made clear, when intuitions are said to be immediate in the sense of
immediate1, whether the non-inferentiality is supposed to be epistemic or psychologi-
cal. (None of the authors mentioned in the previous paragraphs clarifies their remarks
in this respect.) To be epistemically non-inferential, an intuition must not depend epis-
temically on previously justified propositions. To be psychologically non-inferential, an
intuition must not, as a matter of psychological fact, have been (explicitly or otherwise)
derived by the subject from premises to which she was previously committed. Within
the category of psychological immediacy1, one might further distinguish between
(1) not having been derived via any explicit or conscious inference and (2) not having
been derived via any inference whatsoever, even a subpersonal or non-conscious one.
To say that intuitions are immediate2 is to say that they are (or seem) obvious and/or
spontaneous and/or natural and/or compelling. The last-quoted passage from Russell
(2007) shows that he associates immediacy2 with intuitionhood. Nagel also says that an
intuition ‘has a certain immediacy, like a simple perceptual judgment’ (2007, p. 794).
Janet Levin describes modal intuitions as ‘those clear, peculiarly compelling concep-
tions of what can or cannot be— . . .’ (2007, p. 253) and for Shaun Nichols et al., an intui-
tion is ‘a spontaneous judgment about the truth or falsity of a proposition’ (2003, p. 246,
n.3). One difference between immediacy2 and immediacy1 is that only the former is
4 The relationship between symptoms three and four in this bundle is worthy of comment. If concep-
tual competence (say) is regarded a normal (i.e. unspecial) faculty, the two symptoms may be in tension.
Symptoms three and four may then be best regarded as alternatives to choose between (perhaps as possible
routes to making sense of the a prioricity symptom). Thanks to an anonymous referee here.
likely to be straightforwardly a matter of phenomenology or how an intuition presents;

the latter is plausibly determined by the correct psychological and/or epistemological
account of the subject, which may include various elements of which she is introspec-
tively unaware (and perhaps even incapable of becoming aware).5,6
Bundle four consists of three metaphilosophical symptoms of intuitionhood. First,
it is sometimes said that intuitions are starting points and/or foundations for philo-
sophical enquiry. Hilary Kornblith, for example, writes: ‘Appeals to intuition play a
foundational role in a good deal of philosophical theory construction’ (2007, p. 28).
A negative spin on the status of intuitions as starting points or foundations is offered by
Matthew Liao (2008, p. 248), who puts things this way: ‘When philosophers run out of
arguments, they often appeal to intuitions.’
It is worth pointing out that this symptom is related to the immediacy symptoms
of bundle three: for example, it can sound odd to call something foundational if it
has been inferred from, or rests epistemically upon, something else, for that makes
it sound as if the ‘something else’ has a better claim to be foundational. Nevertheless,
being metaphilosophical, this symptom is not the same symptom as any of the imme-
diacy symptoms, which don’t concern philosophy in particular.
To say that intuitions provide starting points or foundations is not necessarily to
say that they provide incontrovertible ones. Brian Weatherson (2003), for example,
thinks philosophers should aim for the best trade-off between capturing intuitions
about cases and systematic overall theories, and that this trade-off can involve deny-
ing some intuitive claims. Also worth noting is that global and local versions of this
symptom may be distinguished. The global version says that intuitions considered en
masse provide starting points for philosophical enquiry considered en masse. A local
version says that intuitions about particular things provide starting points for (some or
all) particular philosophical enquiries. Adopting a local version of the starting-point
claim allows for the possibility that some of the intuitions which are starting points
in (say) modal epistemology could be the upshot of prior philosophical theorizing in
(say) modal metaphysics.
The second symptom in bundle four is that reliance on intuitions is taken to be
characteristic, or at least distinctive, of what is sometimes called traditional Western
analytic philosophy.7 Goldman (2007, p. 1), for example, writes: ‘One thing that dis-
tinguishes philosophical methodology from the methodology of the sciences is its
5 Thanks to an anonymous referee for noting this in their comments.

6 There are many issues about the phenomenology of intuition (some of which may be related to the
immediacy symptoms) into which I shan’t delve too deeply here. Most notably, there is debate about whether
an intuition should be described as a kind of experiential state or ‘seeming’ (as Bealer 1998 believes) or not
(as Sosa 2006 argues). I am yet to be convinced that anything in this vicinity is sufficiently consistently asso-
ciated with intuitionhood (particularly in first-order philosophical practice, as opposed to metaphilosophi-
cal reflection) to count as a symptom for my current purposes.
7 By ‘traditional Western analytic philosophy’, I am attempting to refer to a tradition rooted in the work of
such ancient Greek authors as Plato and Aristotle, not only to specific sub-traditions wherein philosophers
are more specifically engaged in projects of “conceptual analysis.”
96 C. S. I. Jenkins
extensive and avowed reliance on intuition.’ Jonathan Weinberg et al. (2012, p. 256)
describe ‘[t]‌he practice of appealing to intuitive judgments concerning esoteric cases’
as having been ‘long standard in analytic philosophy’. There is, of course, some con-
nection between its being analytic philosophy that is supposedly so distinguished and
some of the bundle-two symptoms—most obviously, those involving conceptual com-
petence or analysis. (See e.g. Stich 1988, p. 578.)
The third symptom in the fourth bundle is that intuitions are or provide evidence
and/or warrant, particularly (although not necessarily exclusively) in philosophy. ‘If
something is intuitive, this tends to count in favor of a position, and if something is
counterintuitive, this tends to count against the position’ writes Liao (2008, p. 248; note
that Liao is here merely reporting a practice, rather than endorsing it). Goldman notes
that ‘[t]‌he evidential weight accorded to intuition is often very high . . .’ (see e.g. 2007,
p. 1). And Saul Kripke advocates treating intuition this way: ‘Of course, some philoso-
phers think that something’s having intuitive content is very inconclusive evidence in
favor of it. I think it is very heavy evidence in favor of anything, myself ’ (1981, p. 42).
Let me stress that the foregoing list of symptoms is not supposed to be exhaustive.
Other symptoms associated with intuitionhood (which no doubt are related in inter-
esting ways to those on my list) include: being the upshot of thought experiments (see
e.g. Ludwig 2007) and: being verdicts on (singular) cases, thought-experimental or
otherwise (see e.g. Nagel 2007, p. 793).
5.3. What’s Up?
Obviously, one needn’t assume that all the symptoms listed in section 5.2 go along
together, in the sense that something will either possess them all or lack them all. One
needn’t even assume that the symptoms in one particular bundle go along together.
But the potential for disconnects is especially clear across bundles. It is easy to see, for
example, that a prioricity and commonsensicality differ significantly in extension just
by considering theorems of advanced mathematics (a priori but not commonsensical)
and propositions like Grass is green (commonsensical but not a priori).
It is not uncommon to find a prioricity (and/or related bundle-two symptoms)
regarded as characteristic of things called ‘intuitions’ by certain authors, while other
authors characterize something called ‘intuition’ primarily by appeal to commonsen-
sicality (and/or other bundle-one symptoms). All of Williamson’s ‘crude empiricists’,
‘crude rationalists’, and ‘linguistic or conceptual philosophers’ (2007, pp. 2–3) are appar-
ently doing one or other of these things. I of course am not claiming these are the only
two ways of characterizing “intuitions,” only that they are two pervasive and classic
ways.8 (At least, they are sufficiently classic to merit Williamson’s focus on—possibly
8 Sometimes, for example, ‘intuition’ is at least used (if not explicitly defined) in ways that suggest some-
thing narrower may be going on. On one apparently reasonably common narrower use, the things labeled
‘intuitions’ are some but not all of the deliverances of a priori reasoning: maybe the most basic or obvious of
them. (See, e.g., Weatherson 2003.) Thanks to Fiona Woollard for bringing this kind of use to my attention.
somewhat caricatured—versions thereof at the outset of his discussion.) More recent,

metaphilosophically sophisticated characterization attempts are interesting, but in my
opinion these two broad approaches capture how many—perhaps even most—of the
less reflective, first-order, philosophical appeals to “intuitions” are intended. The fol-
lowing passage from John Symons (2008, p. 68) conveys a similar message:
Contemporary accounts of intuition oscillate between the folksy and the rarefied: Intuition is
sometimes understood to be a peculiarly aprioristic faculty while elsewhere it is portrayed as the
most ordinary, commonsense level of thinking; accessible to all of us. George Bealer describes
intuition as a sui generis propositional attitude which, at the same time, serves as the source of
all (non-stipulative) a priori knowledge (2002, 73). Elsewhere, we find “intuition” and “com-
monsense” being used interchangeably. Saul Kripke, for example, contrasts intuitions with ‘phi-
losopher’s notions’ and regularly identifies intuitive content as the kind of thing to which the folk
would readily agree. (1980, p. 42)9
If this much is granted—i.e. that there are two quite different, but classic, ways of char-
acterizing things called ‘intuitions’—how should we diagnose this situation? Are there,
perhaps, two quite different, but equally classic, types of competing theory of some one
thing, intuition? This seems to be Williamson’s presumption; he discusses the view of
the ‘crude empiricists’ (that “intuition” is simply folk prejudice) and that of the ‘lin-
guistic or conceptual philosophers’ (that “intuition” is the deliverance of linguistic or
conceptual competence) in the same breath, without suggesting that the two groups
might be discussing different things.
This approach strikes me as an uncharitable, however. It could be that these are two
competing hypotheses and one or other is straightforwardly incorrect. But it would be
pretty surprising if careful, intelligent philosophers were systematically mistaking the
deliverances of conceptual competence for common sense, or vice versa. More charita-
ble interpretations of the split are available.
The first is that there is some ambiguity, or some other kind of contextual shifti-
ness, in the term ‘intuition’ as used by philosophers. It could be that in the mouths of
some philosophers ‘intuition’ refers (approximately) to the deliverances of common
sense, while in the mouths of others it refers (approximately) to the deliverances of
conceptual competence. This is charitable in that neither tradition need be straight-
forwardly wrong about anything on this view (except insofar as the different users find
themselves arguing about whose account of intuition is correct: i.e. tacitly or other-
wise adopting the uncharitable interpretative hypothesis described above). Perhaps we
would have to attribute further mistakes were each group to claim some kind of special
status for intuition that can only be played by one of the two candidates (being uniquely
distinctive of or foundational for analytical philosophical enquiry, for example). But
I’m not aware of any particular evidence of that happening.
A test (an imperfect one, but a test nonetheless) for contextual shiftiness is whether
or not one can hear as acceptable a sentence which would need to involve a shift in
9
Page 41 of Kripke 1981 is also highly relevant to this claim.
98 C. S. I. Jenkins
meaning in order to come out true. For example, evidence of the ambiguity in ‘bank’
can be given by pointing out that it can be acceptable to say ‘The ducks are near the
bank but not near the bank’, where the first occurrence of ‘bank’ refers to a river bank
and the second to a financial institution. Claims like ‘This is an intuition although it
is not an intuition’ are difficult to make sense of in the absence of cues, but to my ears
there is nothing wrong with ‘This is something I rationally intuit, but I admit that it
is not pretheoretically intuitive’, and once cued with this latter sentence, the former
sounds potentially acceptable (at least to me).
A second relatively charitable interpretation, which wouldn’t require the acceptabil-
ity of this kind of sentence, is that ‘intuition’ is semantically general, and its extension
includes both the deliverances of common sense and the deliverances of conceptual
competence. It could then be suggested that philosophers who might appear be to using
‘intuition’ to refer only to the deliverances of linguistic competence are in fact simply
using the semantically general term but (consciously or otherwise) ignoring some of
the things that it covers, because they are not relevant to the discussion. Bealer (see e.g.
1996, p. 123) is one philosopher who is explicit that he thinks the extension of ‘intuition’
includes more than just the rational intuitions which are of most interest to him.
On this type of view, the concept intuition expressed by the semantically general
term ‘intuition’ might be best regarded as a kind of family resemblance concept, such
that possessing enough of the symptoms in bundles one to four qualifies something as
an intuition (where certain symptoms might be more heavily weighted than others, or
otherwise of particular significance), but it is difficult or impossible to give tidy neces-
sary and sufficient conditions on intuitionhood.
I think it may be somewhat indeterminate which of the charitable hypotheses is cor-
rect, at least as concerns the many first-order philosophical uses of ‘intuition’ where no
definition is provided. In fact, I suspect that even if one or other of the two hypotheses
could be settled upon, a healthy amount of semantic indeterminacy should probably
be attributed to the target word in any case. For example, some philosophers do not
explicitly define ‘intuition’ but appear to be tapping into a bundle-two-driven concep-
tion without it being entirely clear just which bundle-two type symptoms are taken to
be characteristic of an intuition. (§2 of Markie 2008 uses ‘intuition’ in roughly this way.
Peter Markie intentionally doesn’t give a precise definition: he is overviewing a tradi-
tion which isn’t always precise on this point.) If we were to take such uses as different in
meaning from other, bundle-one-driven, uses (as classically exemplified by e.g. Kripke
1981), as on the ambiguity/shiftiness hypothesis, it would still be reasonable to take the
word’s meaning as somewhat indeterminate with respect to which bundle-two symp-
toms are really doing the work.
Interestingly, it is consistent with the semantic generality view that ‘intuition’ is a
functional term, with some bundle(s) of symptoms characterizing the relevant func-
tional role and, perhaps, various different kinds of thing capable of serving as realizers
of that role. One might characterize the role in terms of the bundle-four symptoms
(for example), and then say that both the deliverances of common sense and the
deliverances of conceptual competence realize this role, so that both fall within the
extension of ‘intuition’.10
I shall not attempt to settle the question of which of the two charitable hypotheses
is (more nearly) correct. I suspect the first is probably closer to the truth; in particular,
it seems to me that some philosophers use ‘intuition’ in a bundle-one driven sense,
whereby commonsensicality and/or related features are important for whether or not
it counts as an “intuition,” but features related to a prioricity are irrelevant. And I think
that some philosophers use ‘intuition’ in a bundle-two-driven sense, whereby a prio-
ricity and/or related features are important but features related to commonsensicality
are irrelevant. (Let me stress again, though, that I am not suggesting these are the only
uses.) But defending a semantic theory about philosophers’ use of the word ‘intuition’
is not my goal here (interesting though the pursuit of that goal would be). For the most
part, I just want to get clear that there is this divergence (however explained) between
uses associated with bundle one and uses associated with bundle two.
I shall talk for the remainder of this chapter as if the shiftiness hypothesis were cor-
rect, and ‘intuition’ has (at least) two different extensions. But readers should feel
free to substitute my talk of (e.g.) the bundle-one-driven sense of ‘intuition’ with talk
of uses of ‘intuition’ which are best understood as (implicitly or explicitly) restricted to
bundle-one-type intuitions, despite being semantically general.
Later in this chapter, I will be defending intuitions in the bundle-two-driven sense
against some now-standard objections to epistemic reliance on “intuitions.” It may
well also be possible to defend intuitions in the bundle-one-driven sense against simi-
lar objections, but that is not part of my project. My aims are, firstly, to make clear that
the best way to make progress in this rather murky territory is to divide the challenge
to “intuitions” into distinct challenges to distinct conceptions of intuitions, and then
secondly, to begin the task of addressing those challenges as directed against the con-
ception that interests me most.
One reason why one might expect a certain amount of unclarity, confusion, and/
or conflation in this arena is that intuitions in both the bundle-one-driven sense and
the bundle-two-driven sense could naturally be—and, indeed, are—associated with
the various symptoms in bundles three and four. One can see, for example, how being
immediate2 might be associated with being commonsensical or everyday. Often, when
asked for an opinion on something one hasn’t previously considered, on which com-
mon sense delivers a verdict, the commonsense answer is (the) one that seems natu-
ral and compelling. (Of course, not everything that is obvious or natural to someone
who has undergone years of training and reflection will be commonsensical, and
vice versa. So these are not the same symptom.) Plausibly, it is also often not clear to
someone who is in fact relying on common sense where his or her belief comes from
(either psychologically or epistemically speaking). Common sense beliefs are likely
I am grateful to a member of my audience at USC for discussion of this sort of functionalist approach.
10
100 C. S. I. Jenkins
to be ones we imbibe without particularly noting their provenance or thinking about

their justification. So immediacy1 is also predictably associated with intuition in the
bundle-one-driven sense. Immediacy1 might also be uncritically associated with the
other bundle-one symptom: that of being uncontaminated by theory. If a belief is
not arrived at inferentially, it might be assumed that this means there is less scope for
such contamination to be introduced than when a belief is inferred from (theoretical
or theory-laden) premises. (I do not mean to suggest that this association should go
unchallenged, only that one can imagine it being uncritically made.)
In a similar vein, philosophical associations between immediacy1 and a prioricity
(and related bundle-two symptoms) are not hard to come by. At least some a priori
knowledge or justification must be epistemically non-inferential according to those
who follow BonJour (1998) in thinking that an unacceptably radical scepticism is the
only alternative to this position. Nor is it unusual to hear the deliverances of a priori
reasoning described as ‘compelling’ or ‘natural’ or in similar terms. One might think,
for example, of Kurt Gödel’s famous claim that the axioms of set theory ‘force them-
selves upon us as being true’ (Gödel 1947, p. 484) or of the supposed connections
between a prioricity and ‘self-evidence’ (see e.g. Audi 1998, pp. 101–3). Hence the asso-
ciation with immediacy2 is predictable too.
Likewise, the claim that philosophers get (at least some of) their starting points,
foundations, and/or warrants from a priori sources is hardly an uncommon one.
Certainly Plato and Descartes believed that a priori sources could provide both war-
rants and epistemic foundations for their theorizing. But then neither is it unusual, in
contemporary philosophy at least, to hear it said that these same bundle-four symp-
toms are exhibited by the commonsensical or the ‘platitudinous’. The work of Frank
Jackson (see e.g. 1998) and David Lewis (see e.g. 1972) demonstrates a clear commit-
ment to this association.
I have so far been ignoring the question of ontological category: the question of
whether “intuitions” are (taken to be) propositions, attitudes, acts, or processes. My
suspicion is that in order to best accommodate the way philosophers talk we should
allow that any of these things can count as “intuitions,” though (and again there
is probably some indeterminacy here) it may be best to say that different senses of
‘intuition’ are in play when these different categories are in its extension. This is not a
dimension of difference that need particularly concern us here, however, since it does
not impact upon the main distinction that I wish to draw between bundle-one-driven
and bundle-two-driven senses of ‘intuition’. I don’t mean to foreclose the question of
whether there is more than one sense of each kind, but for ease of expression I shall
continue to talk as if there is exactly one of each.
To conclude this section, I would briefly like to draw attention to the differ-
ence between something’s being called ‘an intuition’ and its being called ‘intui-
tive’. The word ‘intuitive’ seems to be much more closely related to ‘intuition’ in the
bundle-one-driven sense (and to the bundle three symptoms) than it is to ‘intuition’
in the bundle-two-driven sense. That intuitiveness is associated with one but not the
other sense of ‘intuition’ is an important (though not, as far as I can tell, widely appreci-
ated) piece of evidence that there are substantial differences between them.
5.4. A Simple View
Before I move on, I should discuss a simple view about what intuitions are which has
not so far impacted upon my discussion. I discuss it now in order to explain why I don’t
think it should have an impact: I think the view is mistaken. The view in question is that
an intuition is simply any old belief or disposition to believe. Something along these
lines is suggested by some remarks in Lewis 1983 (p. x), but a clear statement of it is
given by Peter van Inwagen (1997, p. 309):
Our “intuitions” are simply our beliefs—or perhaps, in some cases, the tendencies that make
certain beliefs attractive to us, that “move” us in the direction of accepting certain propositions
without taking us all the way to acceptance.
Sosa (1998), Williamson (2007, p. 309) and Jonathan Ichikawa (MS) are also sympa-
thetic to this view. It is, in effect, a much more extreme version of the semantic general-
ity hypothesis considered earlier.
One datum that can be offered in support of the simple view (see Ichikawa MS) is
that all kinds of things can be labelled as ‘intuitions’ or ‘intuitive’. Not only necessary or
a priori things count; contingent, a posteriori claims are often classified as intuitions
(or the kinds of propositions p such that one can have an intuition that p). John Dupré
(1996, p. 386) for example, describes: ‘. . . the natural intuition that humans are, some-
times, causally efficacious in the world around them.’ The things classified as intuitions
may, similarly, be either general (knowledge requires belief) or particular (Smith does
not know that the person who will get the job has ten coins in his pocket11). The view that
intuitions are simply beliefs or dispositions to believe accommodates this wide variety
of applications of the term ‘intuition’ in a simple, unified way.
Allowing mere dispositions to believe to count as intuitions is an important ele-
ment of the view, for sometimes it is tempting to categorize things as intuitive but fail
to believe them (Basic Law V or naïve set-theoretic comprehension, for example).
However, the corresponding problem in the other direction cannot be similarly han-
dled. If one has a belief that one is not willing to categorize as an intuition, the view is
in trouble. And it seems that philosophers do indeed talk this way. For example, here is
Weatherson (2003, p. 3):
It does not seem to be the case, in the relevant sense, that 643 × 721 = 463,603. Unless one is rather
good at mental arithmetic, there is nothing that 643 × 721 seems to be; it is out of the reach of
intuition.
Yet the proposition that 643 × 721 = 463,603 is something that Weatherson believes.
See Gettier 1963.
11
To accommodate this kind of data about how philosophers use ‘intuition’, we might
weaken the claim so as to allow that the inclusive sense of ‘intuition’ under consid-
eration here is not the only one. But then the view loses its appealing simplicity and
unificatory power.
Moreover, the supposedly motivating datum regarding the variety of uses of ‘intui-
tion’ can be more conservatively accommodated by the view I am defending, accord-
ing to which there is at least one sense of ‘intuition’ on which neither necessity nor a
prioricity is required of the things so labelled, and both general and particular propo-
sitions may count. The bundle-one-driven sense that I’ve described is such a sense.
When necessity and/or a prioricity are taken to be characteristic of “intuitions,” this
need not be interpreted as philosophical error, but can be attributed to the fact that
philosophers who say such things are talking about something different.
For the remainder of this chapter I shall be defending a particular account of intuition
in the bundle-two-driven sense against some of the supposed challenges to epistemic
reliance on intuition. These challenges, I allege, cannot be made out in the absence of
a clear conception of which thing it is that is being challenged. This is not sufficiently
appreciated in the contemporary debates about “intuitions”; Weinberg et al. (2012), for
example, launch into a paper-length critique of what they call ‘intuitions’ without dis-
cussing what they mean by the term, so it is impossible to interpret the critique with suf-
ficient clarity to begin a careful assessment of its merits. It is difficult even to say whether
it can reasonably be interpreted as addressing itself to any single target, given how radi-
cally differently the word ‘intuition’ appears to be used in the bodies of philosophical
literature they presumably take themselves to be engaging. We need to get clearer.
5.5. The First Challenge: Unclarity

The challenges to epistemic reliance on “intuitions” are many and varied, and I shall
address only a sample in the remainder of this chapter. These are the following:
(1) The nature, workings, target(s) and/or source(s) of intuitions are unclear.
(2) Why should we think intuitions are reliable, epistemically trustworthy, a source
of evidence, etc.?
(3) Intuitions exhibit cultural variation/intra-personal instability/inter-personal
clashes.
(4) There is no way to calibrate intuitions against anything else.
(5) It is not naturalistically respectable to give epistemic weight to intuitions.
Challenges in the vicinity of (1) are discussed by Goldman (2007), Cummins (1999),
and Williamson (2007), the third of whom expresses considerable disquiet about the
situation (p. 215):
“Intuition” plays a major role in contemporary analytic philosophy’s self-understanding. Yet
there is no agreed or even popular account of how intuition works, no accepted explanation of
the hoped-for correlation between our having an intuition that P and its being true that P. Since
analytic philosophy prides itself on its rigor, this blank space in its foundations looks like a meth-
odological scandal.
These sorts of complaints may be reasonable, insofar as they are targeted on authors
who want to rely on “intuition” but have not defined it and explained what it does
and how it works. (Of course, it wouldn’t be fair to demand e.g. a complete account of
exactly which propositions can be intuited, under the guise of seeking a characteriza-
tion of intuition’s ‘targets’. But some idea of the kinds of things it can deliver is not too
much to ask for.)
I have argued (in Jenkins 2008 and elsewhere) that there is an account of a priori
knowledge and justification which, I think, can help make the requisite sense of intui-
tions in the bundle-two-driven sense. I suggest for consideration a view on which con-
ceptual examination can be an a priori way of learning truths about the world because
the concepts under examination can encode information about the world. That is to
say, a view on which the structure of our concepts and the relationships between them
mirror the structure of the world.12 This would mean that facts about the world can be
read off our concepts, much as information about the world can be read off a map of
the world.
According to the view, moreover, this mirroring of the world’s structure is not an
accident. It occurs because our concepts are sensitive to our experience of the world,
which in turn is sensitive to the structure of the world. (This is what I call the ‘ground-
ing’ of concepts in experience.) Experience thus plays a crucial role in accounting for
the workings of (this kind of) intuition. It is only because of the crucial mediating
role of experience that conceptual examination counts as a source of knowledge and
justification. The knowledge and justification arrived at are thus fairly described as
empirical. This to some extent problematizes the claim that the resulting knowledge
and justification are also a priori. If ‘a priori’ means simply non-empirical, nothing can
be both empirical and a priori.
However, I argue (see e.g. Jenkins 2008, §9.5) that other common definitions of ‘a
priori’ actually do (surprisingly) allow for some knowledge and some justification
to be both a priori and empirical, and that it is both more useful and quite in keep-
ing with traditional thinking about the a priori to adopt one of these definitions. For
example, if we say that a priori knowledge is independent of empirical evidence (rather
than independent of experience altogether), we can argue that knowledge secured
through the examination of concepts which are empirically grounded is (empirical but
also) a priori, since the role played by experience in grounding our concepts is not an
12 I do not propose that this mirroring relation need be entirely straightforward; for example, I don’t
require that every predicate-like concept correspond to some worldly property. It might instead be that
some of our predicate-like concepts are compounds of more basic concepts which do correspond to worldly
properties, and that these more complex concepts encode (or contribute to the encoding in our concep-
tual schemes of) information about those properties. (See Jenkins 2008, §4.4.) For the sake of clarity, I shall
ignore such complications in what follows.
evidential role. (Evidence is evidence for propositions. Grounding for concepts need

not be construed as evidence for propositions. See Jenkins 2008, ch. 4.) We also pre-
serve other characteristic features of a prioricity: conceptual examination is sufficient
for knowledge of certain propositions without empirical testing, and this knowledge is
different in kind (not just different in degree, as Quine 1951 would have it) from ordi-
nary empirical knowledge.
This is not the appropriate place for detailed exposition of the foregoing account of
the nature of a priori knowledge and justification. But what I want to point out here is
that one can understand intuition(s) in the bundle-two-driven sense as consisting in
the act, process, and/or deliverances of the examination of empirically grounded con-
cepts. My suggested account of the a priori delivers up something to say in response to
the challenge that the nature or workings of intuitions are unclear. Plenty is said about
the source(s) by this account, too: the proximal sources of intuitions are our concepts,
and the distal source is experience. The targets are clarified somewhat, but thus far
only insofar as the account is taken to apply to all a priori knowable propositions. The
targets of intuition on this account are those subject matters that are knowable a priori,
including (in my view) mathematics, logic, and some other parts of philosophy.
As I mentioned earlier, it’s not reasonable to demand exact details as to the scope
of intuition. But one question potentially of interest to philosophers working on the
methodology of philosophy (which is where much of the current debate about the
nature of intuitions is playing out) is the following: which parts of philosophy might be
informed by intuitions which are the deliverances of the examination of empirically
grounded concepts? Can an ethical concept like right be empirically grounded? An
aesthetic one like beauty? A metaphysical one like existence? An epistemological one
like knowledge? My answer to this question has two parts.
The first part is to say that I think any or all of these philosophical concepts could in
principle be accounted for as empirically grounded. The fact that they seem in some
sense quite “distant” from experience is no bar to this being an option on the table. It
would be overly simplistic to assume that the empirical grounding of a concept like
(say) right requires that one literally sees, hears, etc. rightness. As one example of a way
to approach the empirical grounding of abstract-sounding concepts, consider that we
might think of empirical concept grounding as holistic. On a (Quine-inspired) holistic
approach, one’s entire conceptual scheme (including ethical, aesthetic and metaphysi-
cal concepts) gets to be empirically grounded in virtue of its success in making sense
of the totality of our experience. The whole scheme shares at once in holistic empiri-
cal grounding. For further development of the respects in which the idea of empirical
concept grounding should not be oversimplified, see Jenkins 2008, §5.3.
The second part of my answer is that which philosophical concepts it would be
appropriate to count as empirically grounded is a matter that different metaphiloso-
phers will approach with different methodologies and different results. My central aim
in this paper is just to put the option of a concept-grounding approach on the table
as an account of intuitions in the bundle-two-driven sense, and describe how it may
work. In section 5.6, however, I will say a little more about the kind of methodology
I might use to tackle this question of specifics. One more misconception I do want
to clear away at this stage, however, is the idea that the only way for (say) our concept
knowledge to map the world’s structure is for knowledge to be a “natural kind,” where
the “natural kinds” are thought of as including only those kinds which track the most
sparse or fundamental distinctions in the world. I believe that the world has a lot more
structure to it than its fundamental structure. It may be that knowledge is a real but
non-fundamental thing and our concept knowledge encodes information about that
real but non-fundamental thing that we can learn by conceptual examination.
In my view, the concept-grounding approach respects the a prioricity symptom
of intuition, as well as the association with conceptual competence and/or analysis.
Necessity is commonly associated with a prioricity and with being the kind of thing
that can be learned through conceptual examination, so the association with necessity
symptoms associated with intuition is also accounted for.13 There is no special faculty
associated with conceptual examination on my view, so the final symptom from bun-
dle four is being dropped. But I think that preserving the first three symptoms is easily
enough for continuity with the tradition of bundle-two-driven thinking about intui-
tion, and that belief in a special faculty is in fact a rather disreputable associate, with
which defenders of intuitions who hope to be taken seriously (particularly by those
who call themselves ‘naturalists’; see section 5.6) would do well to break ties.
This way of understanding the nature of intuition in the bundle-two-driven sense
could, I think, be of some help in moving forward the debate between Goldman
and Kornblith concerning the scope and importance of intuition. Goldman (e.g.
2007) argues that philosophical reliance on intuition is a way of finding out about our
concepts (though not about the things of which they are concepts), and that as such it
is of significance to philosophers. (That Goldman has a bundle-two-driven conception
of intuition is, indeed, evinced partly by this assumed relationship between concepts
and intuition.)
Kornblith (2007, p. 46), on the other hand, writes:
[O]‌ur concepts are not plausibly viewed as the target of philosophical understanding . . . it is the
extra-mental phenomena themselves which are the real targets of philosophical analysis: knowl-
edge, justification, the good, the right, and so on, not anyone’s concepts of these things . . . . The
standard philosophical procedure [i.e. reliance on intuition] cannot be redeemed by viewing it
as an attempt to provide an understanding of our mental representations instead of the phenom-
ena which they are representations of.
The structure of this debate suggests that we have to choose between thinking of intui-
tion as a way of finding out about the world or a way of finding about our representa-
tions of the world (our concepts). Kornblith and Goldman agree that it can’t be the
former, and disagree about its significance if it is merely the latter.
I explore the application of the account to modal knowledge in Jenkins 2010.

13
My view is that intuition in the relevant sense can be a way of finding out about the
world via the examination of concepts. So it is not the case that we face the kind of
choice that forms the background to the Goldman–Kornblith debate. Our concepts
don’t float free from the phenomena which they represent; they are sensitive to the
world because they are sensitive to experience.
One of Kornblith’s reasons for denying that intuition is a way of finding out (much)
about the world is that he thinks (many) interesting philosophical concepts, such as
our concepts of knowledge, causality, and responsibility, stand for natural kinds. And
natural kinds, he thinks, are not the sorts of things about whose deep natures one can
learn by means of conceptual examination. Instead, they are ‘susceptible to straightfor-
ward empirical inquiry’ (2007, p. 47). This latter claim is a substantive one, with which
I am not sure I agree. But I don’t intend to argue that issue here. Nor do I intend to deny
that (many) philosophically interesting concepts stand for natural kinds. Conceding
both points to Kornblith may require us to allow that he is right about the limitations
of the scope of intuition as a philosophically useful source of information. But my con-
cern is not with precisely delimiting intuition’s scope, but with understanding in gen-
eral terms how it works and what sorts of things it can achieve.
Instead, I want to examine another strand in Kornblith’s thinking, which if correct
does threaten to undermine my account of the nature of intuition qua source of jus-
tification and knowledge. The question of whether or not philosophically interesting
concepts pick out natural kinds is not actually crucial, according to Kornblith (2007,
pp. 36–7):
It really doesn’t matter, for present purposes, whether knowledge and other targets of philo-
sophical analysis are natural kinds. . . . [E]‌ven if the topics of philosophical interest typically cor-
respond to . . . socially constructed kinds, it remains true that the concepts of the folk, and the
concepts of philosophers as well, need not accurately characterize these socially constructed
categories. Just as any individual’s concept of aluminum may contain substantial errors or omis-
sions, any individual’s concept of a semiconductor, or Chippendale furniture, or of socially con-
structed categories generally, may contain substantial errors or omissions. So the gap between
concept and category does not disappear simply because we have moved from natural kind con-
cepts to socially constructed ones. And once we recognize that our concepts, whether the con-
cepts at issue are those of the folk or of theoreticians, may fail to characterize the categories they
are concepts of, the philosophical interest of our concepts thereby wanes.
Kornblith is right to note, and I want my account to accommodate, the possibility of

intuition’s leading us astray. On accounts of intuition whereby it has something to do
with the examination of concepts, i.e. accounts of the kind Kornblith here attacks, the
possibility of possessing concepts that ‘contain errors or omissions’ can be granted as
one way in which intuition can go wrong. (In fact I think there are various ways this
can happen; see Jenkins 2008, p. 238.)
However, it is simply not true that the mere possibility of error removes the epistemic
interest of concepts as a source of information about the world. The fact that there is a
‘gap’ between ‘concept and category’ is no more worrying than the fact that there is a
‘gap’ between how things look or sound to us and how they actually are. The fallibility of
ordinary sense experience is not a reason to regard it as philosophically uninteresting,
or to doubt that it is a (generally) reliable epistemic source. What is required is not the
absence of any gap whatsoever, but the existence of some sort of epistemically relevant
connection between concepts and the things of which they are concepts. I believe there
is such a connection, and that it is mediated by experience.
There is more to say about why we should think intuition actually is a reliable epis-
temic guide to the world (as opposed to why we should reject Kornblith’s reasons
for thinking it isn’t). I shall come back to this question in section 5.6. First, however,
I should note that I am not the only philosopher who explicitly ties an account of (some
kind of) intuition to a corresponding account of a priori knowledge and justification.
Bealer, for example, does the same. And indeed, Bealer has some resources to offer a
response to Kornblith which is somewhat similar to the one I just described.
Bealer (e.g. 2000) argues that ‘determinate possession’ of a concept guarantees reli-
ability (which he admits falls short of infallibility, but takes to be sufficient for justifi-
cation) in the application of that concept across a range of cases. Indeed, determinate
possession of a concept is so defined as to be possible only if one is so reliable. Hence he
can address the challenge to explain the nature and the workings of intuition by saying
that intuitions are reliable seemings which result from the determinate possession of
concepts. And because (unlike me) he does not think experience plays any epistemic
role in this process, he can straightforwardly say that this is also an account of how a
priori knowledge and justification are possible.
My primary concern about Bealer’s way of doing things is that he offers no satisfy-
ing account of how it comes about that we determinately possess any concept. Given
how special determinate concept possession is on his view (so special as to guarantee
reliability in a certain kind of judgment), failing to explain how we have it means that a
key element of our success in securing a priori justification or knowledge is left unex-
plained.14 My appeal to experience in explaining how concepts come to be grounded is
supposed to fill just this kind of explanatory lacuna.
5.6. The Remaining Challenges

Even supposing that the nature of intuition and its deliverances is clear, it is fair to ask
why we should believe that intuition so conceived is a reliable way of learning truths
about the world. Relatedly, one can reasonably question the evidential status of intui-
tions (that is, one can reasonably ask whether having the intuition that p is any kind of
evidence that p is true), and in general whether intuition so conceived will exhibit the
metaphilosophical (bundle-four) symptoms.
Goldman (2007) raises challenges in this vicinity, in order to explain why he thinks
intuition can only be a source of information about concepts (and not about the things
For more details, see Jenkins 2008, pp. 65–7.

14
of which they are concepts). In the following passage, he is challenging the view that
intuition is a source of knowledge concerning Platonic forms in particular, but an
essentially similar challenge could be raised for any claim to the effect that intuition
enables us to secure knowledge and/or justified belief about a non-conceptual realm
(p. 7):
If someone experiences an intuition that the protagonist in a selected Gettier example doesn’t
know the designated proposition, why should this intuitional experience be evidence that the
form KNOWLEDGE is such that the imaginary protagonist’s belief in this proposition doesn’t
“participate” in this form? What connection is there . . . ?
On my view, having an intuition that p is reliably correlated with, and is evidence

that, p. In answer to Goldman’s final question (or rather, to the equivalent question
directed against my view rather than a defender of intuition as a source of knowledge
of Platonic forms), I postulate empirically mediated connections between the world’s
structure and the structure of our concepts, which is what secures the reliability and
evidential status of intuitions in the bundle-two-driven sense.
As I have suggested elsewhere (Jenkins 2008, p.139) one can run an analogue of the
no-miracles argument in support of this view. The standard no-miracles argument (see
e.g. Smart 1963, p. 39 and Putnam 1975, p. 75) runs along roughly the following lines:
1. Our best scientific theories are extremely useful/successful.
2. This usefulness/success would be a miracle if the theories weren’t true.
3. We shouldn’t believe in miracles.
Therefore
4. Our best scientific theories are true.
My analogue runs as follows:
1. (At least some of) our concepts are extremely useful/successful.
2. This usefulness/success would be a miracle if the structure of those concepts15
weren’t mirroring elements of the world’s structure.
3. We shouldn’t believe in miracles.
Therefore
4. The concepts in question mirror the world’s structure.
The usefulness of concepts may be regarded as at least partly determined by the role
they play in enabling us to formulate and entertain the scientific theories mentioned
in the no-miracles argument. If not all concepts can be covered by this argument (i.e.
if they are not all useful and/or successful) then we may need to limit the trustwor-
thy kind of intuition to that involving only the examination of such concepts as are
15
Or, perhaps, that of the ultimate constituents of the concepts; see Jenkins 2008, p. 127.
covered. This is my take on the issue mentioned in the previous section concerning
how to approach the question of specifically which philosophical concepts it would be
appropriate to regard as empirically grounded.
I am not here essaying the project of determining which philosophical concepts are
empirically grounded. (To do so would be to go far beyond the scope of this chap-
ter.) I am describing the kind of methodology I would use for approaching that task.
However, I should note that the no-miracles argument, here as elsewhere, needs to
be treated with considerable care. Premise 2 and its analogues are obviously at risk of
falsehood in cases where serious rival explanations of usefulness/success are on the
table. One thoroughgoing recent incarnation of such concerns can be found in the
literature surrounding the influential argument of Sharon Street’s 2006. Street argues
that a ‘Darwinian’ explanation of our evaluative attitudes, which accounts for their
usefulness/success in evolutionary terms, is more scientifically respectable than, and
in tension with, one which posits objective moral truths which are reflected in those
attitudes.
So hard questions will need to be asked about when the usefulness/success of our
concepts is of the kind which demands an explanation in terms of the world’s real
structure in order to be non-miraculous, and when an alternative explanation in other
terms would suffice. For what it’s worth, I suspect that the success of a range of core
metaphysical and epistemological concepts such as existence and knowledge demands
an explanation in terms of real structure to which they correspond. In support of this
suspicion I would appeal to the depth and breadth of the range of explanatory work
that thinking about the world using our concepts knowledge and existence can do. (The
idea that these concepts play such explanatory roles is not unique to me; for some rel-
evant considerations about the explanatory role of thinking in terms of knowledge, for
example, one might look to Williamson 2000 (§2.4),16 and for something similar con-
cerning existence, one might look to Sider 2011 (§9.6.4).)17
Again for what it’s worth, I have no noticeable inclinations as to whether the suc-
cess of concepts like right and beauty demands the same, but then I have not spent a
comparable amount of my philosophical career to date thinking about these concepts
and the things (if any) they stand for. One further important thing to note, however, is
that even if (say) pure ethical concepts like right and good did not turn out to be empiri-
cally grounded concepts which map the structure of the world, that would not mean
we couldn’t give a concept grounding account of the intuitions that ethicists appeal
to in their philosophizing. Metaethicists, for example, might be employing intuitions
delivered by various empirically grounded concepts to philosophize about the concept
right and about related language. And perhaps it is worth emphasizing again at this
16 To be clear, though, I don’t think this consideration speaks in favour of any Williamson-inspired view
to the effect that the concept knowledge and/or knowledge itself should be regarded as in some sense prior to
other concepts or things. What it suggests to me is that knowledge comes somewhere, not that it comes first.
17 Somewhat similar caveats apply here as in n.16 above.
juncture that my aim in this paper is only to defend one kind of intuition against chal-
lenges. There may be other kinds which can withstand the challenges too, and these
other kinds of intuitions may be at work in areas like ethics whether or not right and
good are grounded concepts.
Of course, our concepts’ having a structure that mirrors that of the world is one thing;
our being appropriately sensitive to the structure of our concepts in using our intui-
tion is another. We could have just the right concepts but be completely wrong in our
intuitive beliefs, if we are insensitive to the structure that our concepts (and the world)
actually have. However, one can run something closer to the original no-miracles
argument at this level; the usefulness of the beliefs arrived at intuitively gives us some
evidence of our own reliability in detecting the structure of our concepts.18
The next challenge I want to consider comes from Robert Cummins (1998).
Cummins argues that intuition is ‘epistemically useless’ (p. 126), for philosophers at
least, since either it cannot be calibrated (checked for correctness) using another, inde-
pendent, source as a guide, or it is redundant, because we have another, independent,
source of the same information.
Cummins does not define ‘intuition’ or give any explicit clues as to what he means
by it, so it hard to be sure which (if any) of the various candidate properties discussed
above he might associate with that term. However, his list of possible sources of philo-
sophical intuition suggests some contact with both bundle one and bundle two, since
it includes both ‘ordinary beliefs’ and ‘concepts’. But intuition as I am understanding
it for the purposes of this section of this chapter can be calibrated, at least to some
extent, using good old-fashioned ordinary empirical confirmation. For example, we
have (I think) both an intuitive route and a straight inductive route to the belief that 2 +
2 = 4. Nor is there any reason to think the existence of an alternative route to this infor-
mation renders the intuitive route redundant. For one thing, why should it be intui-
tion that is redundant here rather than ordinary empirical confirmation? For another,
being able to discover an object’s shape by means of either sight or touch doesn’t in any
philosophically interesting sense render either sight or touch ‘redundant’ with respect
to this question.19
Another reasonable reply to the concern about calibration is that it invokes too much
epistemic internalism, and moreover internalism of a quite specific kind. The demand
for calibration amounts to a demand for double-checking, and an epistemic external-
ist20 can reply that one needn’t double-check in this way in order for something to be
a source of knowledge. (A reply along these lines is outlined in Goldman 2007, §3.)
A young child, for example, can secure visual knowledge without undertaking any
18 I am grateful to my audience at the University of Cologne for helping me develop this point.

19 This calibration might be taken to provide something like the kind of ‘partial certification’ described
by Weinberg et al. (2012), although, as they point out, it could be argued that intuition’s reliability across
this particular range of cases cannot be projected if there is some reason to think the sample is biased or the
results are otherwise unprojectible. (I don’t know what reason might sensibly be offered here, however.)
20 See Jenkins 2008, §2.2 for my preferred take on the characterization of externalism.
obligation to calibrate her visual apparatus against alternative sources of the same
information. Certain kinds of internalists might also argue that that double-checking
isn’t required for intuition to be a source of knowledge; certain kinds of internalist
might want to maintain, for example, that all that is required for a subject’s intuition to
be a source of knowledge that p is that the subject be aware of the intuition in question
(and perhaps that it is the basis for her belief that p). Perhaps the situation would be dif-
ferent if evidence could be provided that the faculty in question was unreliable. I shall
argue shortly, however, that the familiar kinds of putative evidence to that effect are not
such as to give rise to this kind of problem.
Cummins (1998) also suggests that if one understands intuitions as ‘generated by
concepts’ (p. 119), and if concepts are taken to be mental representations (which is my
preferred view), then concepts can ‘generate philosophical intuitions only by func-
tioning as a pointer to something else: an explicit or tacit theory . . .’ (p. 121). In which
case, he suggests, epistemic reliance on intuitions simply amounts to reliance on one’s
(explicit or tacit) theory. But no argument is offered as to why concepts (understood
as mental representations) can only generate intuitions by functioning as pointers to
something else. Concepts, as I understand them, are capable of encoding information
about the world which we can read off through conceptual examination (see Jenkins
2008, ch. 4).
The third challenge under consideration in this section states that we have evidence
that (at least some kinds of) intuitions are unreliable because intuitions differ between
people, and even a single person’s intuitions can vary over time. It is certainly true that
philosophers report ‘clashes of intuition’, though it is less clear that they are reporting
clashes of intuition in the bundle-two-driven sense that interests me in this chapter.
The notion of intuition discussed in certain influential works of ‘experimental phi-
losophy’ (such as Weinberg et al. 2001) is clearly pretty distant from the bundle-two
conception of intuitions. (This possibility is also noted by Russell 2007, §4.) Weinberg
et al. take ordinary people’s considered verdicts on a particular thought-experimental
case to be “intuitions” in the sense that interests them, without regard to whether those
verdicts are a priori or the deliverance of conceptual examination or otherwise pos-
sessed of bundle-two symptoms.
Some responses to this challenge could work regardless of how intuitions are con-
strued. For example, there is the thought (developed in Sosa 1998) that the mini-fictions
told to subjects to elicit intuitive responses may get filled out in different ways by dif-
ferent subjects, so that when they are asked whether in the envisaged scenario p is true,
they are (in effect) answering different questions (that is, questions about different
envisaged scenarios) rather than giving different answers to the same question.
Alternatively, one could argue that the responses given by different groups of sub-
jects concern slightly different concepts, both (or all) of which get expressed using the
same word ‘knows’. This possibility is suggested by Jackson (1998, p.32), and defended
by Sosa (2009, §3).
But even if intuitions in the bundle-two-driven sense sometimes are genuinely in

tension with one another, we should bear in mind that the number of clashes is going to
be very small compared with the massive (and unreported) agreement of intuitions (in
the relevant sense) among philosophers, and indeed everyone else. According to the
account I develop in my 2008, propositions like All bachelors are unmarried, All vixens
are female and 2 + 2 = 4 are among those that can be known via conceptual examina-
tion. There is overwhelming agreement in intuitions on such uncontroversial propo-
sitions as these, and there are overwhelmingly many such propositions. Focusing on
the small number of cases where we might be able to point to non-agreement in the
deliverances of intuition (in the bundle-two-driven sense) is natural and sensible,
because these cases are interesting and potentially problematic. But it shouldn’t lead to
a skewed sense of their significance. To argue that intuition is unreliable (in any epis-
temically problematic way) on the grounds that intuition sometimes goes wrong is no
better than to argue that vision is unreliable (in an epistemically problematic way) on
the grounds that we are sometimes subject to optical illusions.21
The final challenge I want to consider concerns naturalistic respectability. ‘Intuition,
after all, is a traditional hallmark of rationalism, an oft-mentioned source of a priori
warrant. . . . How can a priori warrant be reconciled with epistemological naturalism?’,
asks Goldman (2007, p. 19).
Addressing this challenge requires that we first catch our naturalist. (For a more
detailed discussion of how different kinds of naturalism relate to a prioricity, see
Jenkins Forthcoming.) What kinds of doctrines that go by the name ‘naturalism’ could
conflict with belief in intuitions (in the bundle-two-driven sense)? Most obviously,
there are those who call themselves ‘naturalists’ simply because they reject a priori
knowledge and justification. (See e.g. Devitt 2005.) They won’t be happy with intui-
tions if they are associated with the a priori.
Of course, if I’m right that intuitions can be or provide a priori knowledge and/or
justified belief, then intuitions are going to be unacceptable to naturalists in this sense.
But I offer an account of how such things are possible which renders the knowledge
and justification in question ultimately empirical. So this is no ordinary kind of a prior-
ism, and I certainly do not believe in any kind of special faculty or in any unexplained
epistemic powers or virtues. If these naturalists are ultimately motivated by empiri-
cism, they should consider that my proposal respects empiricism, whilst accommodat-
ing the apparently distinctive features of intuition in the bundle-two-driven sense.
Others mean by ‘naturalism’ the view that philosophy, or perhaps all disciplines,
should be conducted as the natural sciences are (or should be) conducted. On the
assumption that reliance on intuition in the relevant sense is not part of the method-
ology of the natural sciences, this kind of ‘naturalism’ could also cause problems for
21 It is also important to note that the a posteriority of our knowledge concerning agreement and disagree-
ment does not entail the a posteriority of intuitive knowledge itself. (See Ichikawa Forthcoming.)
a commitment to the methodological appropriateness of relying on intuition in (for

example) philosophy.
But in fact, it is plausible that scientists both do and should rely on intuition (in the
relevant sense) all the time. On the assumption, which I am happy to make, that truths
of mathematics and logic are the kinds of things that can be known intuitively, any
scientist who ever uses some mathematics or some logic (even implicitly) could be
relying on intuition.
And, like scientists, philosophers rely on other things besides intuition (in the
bundle-two-driven sense), such as the results of (other) scientists. Many philosophers
of time, for example, choose their theories based at least partly on what our best cur-
rent scientific theories say about the nature of time. Philosophers might rely on intui-
tion more often and/or more self-awarely than other sorts of enquirers, but not in such
a way as to make for a serious disconnect with other intellectual enterprises.
References
Audi, R. (1998). Epistemology: A Contemporary Introduction to the Theory of Knowledge, 2nd ed.
London: Routledge.
Bealer, G. (1996). “A Priori Knowledge and the Scope of Philosophy,” Philosophical Studies 81,
pp. 121–42.
——. (1998). “Intuition and the Autonomy of Philosophy,” in M. DePaul and W. Ramsey (eds),
Rethinking Intuition: The Psychology Of Intuition And Its Role In Philosophical Inquiry.
Oxford: Rowman & Littlefield, pp. 201–39.
Devitt, M. (2005). “There is No A Priori,” in M. Steup and E. Sosa, (eds), Contemporary Debates
in Epistemology. Oxford: Blackwell, pp. 105–15.
Dupré, J. (1996). “The Solution to the Problem of the Freedom of the Will,” Philosophical
Perspectives 10, pp. 385–402.
Gettier, E. (1963). “Is Justified True Belief Knowledge?,” Analysis 23, pp. 121–3.
Gödel, K. (1947). “What is Cantor’s Continuum Problem?,” American Philosophical Monthly
54; reprinted in P. Benacerraf and H. Putnam (eds) (1964), Philosophy of Mathematics.
Cambridge: Cambridge University Press, pp. 470–85.
Goldman, A. (2005). “Kornblith’s Naturalistic Epistemology,” Philosophy and Phenomenological
Research 71, pp. 403–10.
——. (2007). “Philosophical Intuitions: Their Target, Their Source, and Their Epistemic Status,”
Grazer Philosophische Studien 74, pp. 1–26.
Jackson, F. (1998). From Metaphysics to Ethics: A Defence of Conceptual Analysis. Oxford: Oxford
University Press.
Jenkins, C. S. I. (2008). Grounding Concepts: An Empirical Basis for Arithmetical Knowledge.
Oxford: Oxford University Press.
——. (2010). “Concepts, Experience and Modal Knowledge,” Philosophical Perspectives 24, pp.
255–79.
Jenkins, C. S. I. (2013). “Naturalistic Challenges to the A Priori,” in A. Casullo and J. Thurow

(eds), The A Priori in Philosophy. New York: Oxford University Press, pp. 274–89.
Ichikawa, J. (2013). “Experimental Philosophy and A Priority,” in A. Casullo and J. Thurow (eds),
The A Priori in Philosophy. New York: Oxford University Press, pp. 45–64.
——. (MS). “Intuitions and Begging the Question,” available at: <http://www.jonathanichikawa.
net/Home/research/wip/ibq.pdf> [last accessed October 9, 2013].
Kim, J. (1994). “Explanatory Knowledge and Metaphysical Dependence,” Philosophical Issues 5,
pp. 51–69.
Kripke, S. (1981). Naming and Necessity. Oxford: Blackwell.
Kornblith, H. (2007). “Naturalism and Intuitions,” Grazer Philosophische Studien 74, pp. 27–49.
Levin, J. (2007). “Can Modal Intuitions be Evidence for Essentialist Claims?,” Inquiry 3, pp.
253–69.
Lewis, D. (1972). “Psychophysical and Theoretical Identifications,” Australasian Journal of
——. (1983). Philosophical Papers, Volume I. Oxford: Oxford University Press.
Liao, S. M. (2008). “In Defence of Intuitions,” Philosophical Studies 140, pp. 247–62.
Ludwig, K. (2007). “The Epistemology of Thought Experiments: First Person versus Third
Person Approaches,” in Midwest Studies in Philosophy 31, pp. 128–59.
Markie, P. (2008). “Rationalism vs. Empircism,” in E. N. Zalta (ed.), The Stanford Encyclopedia
of Philosophy, available at: <http://plato.stanford.edu/entries/rationalism-empiricism/> [last
accessed October 9, 2013].
Nagel, J. (2007). “Epistemic Intuitions,” in Philosophy Compass 2, pp. 792–819.
Nichols, S., Stich, S., and Weinberg, J. (2003). “Meta-skepticism: Meditations in
Ethno-Epistemology,” in S. Luper (ed.), The Skeptics: Contemporary Essays. Aldershot: Ashgate
Publishing, pp. 227–47.
Papineau, D. (2007). “Naturalism,” in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy,
available at: <http://plato.stanford.edu/entries/naturalism/> [last accessed October 9, 2013].
Putnam, H. (1975). “What Is Mathematical Truth?,” in his Mathematics, Matter and
Method: Philosophical Papers, Volume I. Cambridge: Cambridge University Press.
Quine, W.V.O. (1956). “Two Dogmas of Empiricism,” Philosophical Review 60, pp. 20–43.
Russell, B. (2007). “A Priori Justification and Knowledge,” in E. N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy, available at: <http://plato.stanford.edu/entries/apriori/> [last
Sider, T. (2011). Writing the Book of the World. Oxford: Oxford University Press.
Smart, J. J. C. (1963). Philosophy and Scientific Realism. London: Routledge and Kegan Paul.
Sosa, E. (1998). “Minimal Intuition,” in M. DePaul and W. Ramsey (eds), Rethinking Intuition: The
Psychology of Intuition and Its Role in Philosophical Inquiry. Oxford: Rowman & Littlefield, pp.
257–69.
——. (2009). “A Defence of Intuitions,” in D. Murphy and M. Bishop (eds), Stich and His Critics.
Oxford: Blackwell, pp. 101–12.
——. (2006). “Intuitions and Truth,” in P. Greenough and M. Lynch (eds), Truth and Realism.
Oxford: Oxford University Press, pp. 208–26.
Stich, S. (1988). “Reflective Equilibrium, Analytic Epistemology and the Problem of Cognitive
Diversity,” Synthese 74, 3, pp. 391–413; reprinted in J. Kim, E. Sosa, and M. McGrath (eds),
Epistemology: An Anthology. Oxford: Blackwell, pp. 571–83.
Street, S. (2006). “A Darwinian Dilemma for Realist Theories of Value,” Philosophical Studies 127,
pp. 109–66.
Symons, J. (2008). “Intuition and Philosophical Methodology,” Axiomathes 8, pp. 67–89.
van Inwagen, P. (1997). “Materialism and the Psychological-Continuity Account of Personal
Identity,” Philosophical Perspectives 11, pp. 305–19.
Weinberg, J., Crowley, S., Gonnerman, C., Swain, S., and Vandewalker, I. (2012). “Intuition and
Calibration,” Essays in Philosophy 13, pp. 256–83.
Weatherson, B. (2003). “What Good Are Counterexamples?,” Philosophical Studies 115,
pp. 1–31.
Williamson, T. (2000). Knowledge and Its Limits. Oxford: Oxford University Press.
——. (2007). The Philosophy of Philosophy. Malden, MA and Oxford: Blackwell.
PA RT I I
Intuitions in Disciplines or
Sub-Disciplines
6
Intuitions in Science: Thought
Experiments as Argument Pumps
Darrell P. Rowbottom†
6.1. Thought Experiments and Intuitions

If intuitions play a distinctive role in the empirical sciences—if they are significant in
scientific method not only in so far as scientists occasionally do logic or mathemat-
ics—then this is presumably in thought experiments.1 Indeed a rather seductive view,
prima facie, is that intuitions (and intuition statements) are to thought experiments as
perceptions (and observation statements) are to experiments; and as such, that they
play an evidential role. (I use ‘intuition’ to mean ‘intuitive judgement’. Following Alison
Gopnik and Eric Schwitzgebel (1998, p. 77), a judgement is intuitive if it is ‘not made
on the basis of some kind of explicit reasoning process that a person can consciously
observe’.2) At the very least, this appears to be correct for many thought experiments,
such as the one due to Simon Stevin that I will present in the next section.
† I am especially grateful to Jim Brown and John Norton for detailed feedback on an initial version of this
chapter. I have also benefitted from comments from Peter Baumann, Tony Booth, Duncan Pritchard, and
Oxford University Press’s anonymous referees. Finally, I should also like to thank audiences at Dubrovnik
(39th Annual Philosophy of Science Conference), Lingnan University, and Durban (39th Annual
Conference of the Philosophical Society of Southern Africa, 2013). I cannot hope to name everyone who
has said something useful, but I recall having benefitted, in particular, from comments made by Jamin Asay,
Marco Buzzoni, Rafael de Clerq, David Papineau, Dean Peters, and Neven Sesardić. My work on this essay
was supported by the Sydney Centre for the Foundations of Science (where I held a Visiting Fellowship), the
British Academy (by way of a Postdoctoral Fellowship), and Lingnan University (by way of travel grants).
1 Naturally there may be other uses. Following Gopnik and Schwitzgebel (1998), for example, one might
say that intuitions can be used as hypotheses too. But my own view is that the real work of science goes on
in the so-called context of justification rather than the context of discovery. In short, what we do with our
hypotheses when we have them is the crucial part; see Rowbottom (2011a). This is not to deny that imagina-
tion is important, in order to generate new theories. It is worth adding that the limits of our imagination
(and hence our intuitions) may bear on the realism issue. For example, if our hypotheses are appropriately
constrained by experience, and the unobservable world is significantly different from the observable one,
then this tells against scientific realism. See Rowbottom (2012).
2 For a treatment of the many different uses of ‘intuition’, see Jenkins (2014).
120 Darrell P. Rowbottom
Thought experiments are interesting because they appear, on the surface, to be

means by which to delimit ways the world might be from the armchair. In some special
cases, what’s more, such delimitation results in only one possibility remaining; and
one is then able to derive its actuality. This might not be terribly surprising if the kinds
of possibility involved were strictly or narrowly logical—if they merely concerned
the laws of logic and the definitions of terms, e.g. recognitions that an object cannot
be light and heavy simultaneously—but they sometimes appear to be metaphysical
or nomic.
This chapter is written with a spirit of humility. I do not think that we are in a position
to determine what thought experiments in science rest on (and the nature of the intui-
tions therein); instead, I believe the best we can do is to explore the possibility space,
and continually refine the various resultant positions in response to immanent and
transcendent criticisms.3 This is a kind of gentle progress where we get a better under-
standing of the options, and their relative merits and demerits; and I think it is progress
worth making. Personally, I am interested in the prospects of the view that thought
experiments (and the intuitions therein) rest, ultimately, on experience. In particular,
I am concerned with articulating and defending the claims that: (A) the raw concep-
tual materials of natural scientific thought experiments are derived from experience;
and (B) our ‘intuitions’ about how these concepts interrelate also derive from experi-
ence (where this includes learning how to use words, e.g. by ostensive definition). Call
this scientific intuition empiricism, to be contrasted with scientific intuition rationalism
(which involves the denial of at least one of these theses). I will try to persuade you of its
viability and attractiveness, by appeal to theoretical virtues such as simplicity, but I do
not think I have any evidence that it is true. (Equally, I don’t think its opponents have
any evidence that it is false.) But naturally, those who disagree with my stance on infer-
ence may think that I succeed in doing rather more than I take myself to be capable of!
After giving an example of a thought experiment, I will look at one of the most
developed—and one of the most bold and interesting—versions of scientific intuition
rationalism, which is due to James Robert Brown. I will then endeavour to tease out a
novel problem for this. My strategy is to be as sympathetic as possible, and to buy into
many of Brown’s premises that I personally reject, rather than concentrate on the more
foundational issues that divide us. In short, I am aiming for an immanent approach to
criticism; and one of the virtues of this, I hope, is that it will help the reader to seriously
entertain Brown’s position on intuitions and thought experiments.
I will subsequently devote my attention to an alternative account of thought exper-
iments championed by John Norton, according to which they are arguments. I will
argue that this is not quite correct either, with reference to different understandings
3 This is mainly because of my stance on ampliative inferences, i.e. that they are not truth-conducive in
science itself, let alone in (relatively) speculative metaphysics and epistemology. In this kind of context, in
particular, inference to the best explanation is often appealed to; but I take this to be pragmatic, at best. (One
problem, of course, is that we cannot be sure what needs explanation and what doesn’t; we have to have a
terminus somewhere.) For more on this, see Rowbottom (2011a).
Thought Experiments as Argument Pumps 121
of the thought experiment of Stevin, but that thought experiments are best seen as
argument pumps. In this, I will suggest, they are similar to actual physical (or ‘real’)
experiments.
6.2. Stevin’s Thought Experiment

Before I launch my discussion of different accounts of thought experiments, it will help
to have an example in mind; the one I choose is not only a personal favourite, but also
has special significance in the account of Brown that I discuss in the next section. It
concerns a contemporary of Galileo’s, namely Simon Stevin, who derived the principle
of the inclined plane by considering a string of equally spaced balls, each of equal mass,
draped over a triangular prism. (If liked, one may follow Ernst Mach in instead imag-
ining a uniform chain.) See figure 6.1.
Stevin’s reasoning was simple but brilliant. First, the chain will remain stationary.
Second, the lower portion of the chain (between S and V below the prism) is sym-
metrical, and therefore pulls the remainder of the chain (between S and T and T and V)
equally clockwise and anti-clockwise. Now we need only imagine cutting the chain at
S and V, and can see that the remainder of the chain will remain stationary. ‘Hence: on
inclined planes of equal heights equal weights act in the inverse proportion of the
lengths of the planes’ (Mach 1893 [1960], p. 34). We have our principle of mechanics.
Where does an intuition enter? At the very least, this is in the initial claim that the
chain will remain stationary; whether there are any intuitions involved in the other
steps, or whether appeal to (either folk or somewhat more developed formal) physics
is all that is needed, we can leave as moot. Personally, I am sympathetic to what Mach
says here: that we expect the chain not to move on the basis of our experience of similar
scenarios, e.g. of hanging a coat on a peg. Others take a different view.
Before we continue, I should point out that some might object to the very way that
I have presented the thought experiment. I said only ‘the chain will remain stationary’;
S V
A C
Figure 6.1 Stevin’s Thought Experiment

and this follows the summary of Mach (1893 [1960], p. 34), according to which ‘the
assumption from which Stevin starts [is] that the endless chain does not move’. But
Brown (2010, p. 3) treats matters rather differently, by starting instead with only the
upper portion of the chain present on the triangular prism (i.e. between S and V on top
of the prism). He continues:
How will the chain move? . . . There are three possibilities: It will remain at rest; it will move to the
left, perhaps because there is more mass on that side; it will move to the right, perhaps because
the slope is steeper on that side. Stevin’s answer is the first: it will remain in static equilibrium.
The second diagram below [equivalent to my Figure 6.1] clearly indicates why. By adding the
links at the bottom we make a closed loop which would rotate if the force on the left were not bal-
anced by the force on the right. Thus, we would have made a perpetual motion machine, which
is presumably impossible.
An interesting question, which I will come back to, is whether Brown and I have pre-
sented two different (but highly similar) thought experiments. (In fact, Mach’s ini-
tial presentation was different again; so, as we will later see, is a reconstruction due to
Norton (1996).4) But I wish to emphasize that Brown (2010, p. 4) thinks: ‘The assump-
tion of no perpetual motion machines is central to the argument, not only from a logical
point of view, but perhaps psychologically as well.’ My presentation did not mention
perpetual motion at all; nor do I think it should have. When Brown writes of ‘three pos-
sibilities’, I see infinitely many possibilities. The chain could move for n seconds and stop
(and maybe start again m seconds later, and so on, but not perpetually). Or have a small
propensity to move and a large propensity not to move. Or explode. Or transmute into a
bird and fly away. That is, I think, for all we know independently of experience.
Presumably Brown would agree that these are possibilities, of a logical variety, if
pushed. Ultimately, our disagreement would be on why it is reasonable to disregard
them, or to rule them out. We will return to this.
6.3. Brown’s Account of Thought Experiments

Brown (1991, 2001) argues that we can grasp some laws of nature a priori, by intuition
alone. To some, this view will appear absurd. But it is less so, at least, when it is set in
proper context. First, Brown has an independent view that intuitions play a central
role in mathematical inquiry, in so far as they allow us to comprehend mathematical
entities qua abstracta; and hence, they already play a role in science, albeit indirectly, in
so far as mathematical results are required in science.5 Second, Brown thinks that laws
4 Here’s what Mach (1893 [1960], p. 33) initially says, considering the scenario in Figure 6.1: ‘The chain will
either be in equilibrium or it will not. If we assume the latter to be the case, the chain, since the conditions
of the event are not altered by its motion, must, when once actually in motion, continue to move forever,
that is, it must present a perpetual motion, which Stevinus deems absurd. Consequently only the first case is
conceivable. The chain remains in equilibrium.’ I discuss this presentation in more detail later.
5 Note also that nothing I say here questions the existence of mathematical explanations of physical phe-
nomena, as suggested by Mark Colyvan (2001) and Alan Baker (2005).
of nature should be understood to reflect connections between universals, following

Fred Dretske (1977), Michael Tooley (1977), and David Armstrong (1983). In the words
of the latter:
Suppose it to be a law that Fs are Gs. F-ness and G-ness are taken to be universals. A certain rela-
tion, a relation of non-logical or contingent necessitation, holds between F-ness and G-ness.
(Armstrong 1983, p. 85)
Brown also holds this view independently of his view on intuitions, in light of the prob-
lems with other accounts of laws, especially the regularity account championed by the
likes of Alfred Ayer (1956). In particular, the notion that laws are just universal state-
ments to which we take particular attitudes runs the risk of conflating epistemic and
metaphysical issues, according to some philosophers. The line goes that there may be
laws even if we cannot identify them. So even if we can only glean whether a given
regularity lacks exceptions—in fact, on my own view, it is hard to see how we could
even do that6—it does not follow that some regularities are not ‘special’, in so far as they
are necessary in some non-epistemic sense.
I am unconvinced that we should be Platonists concerning mathematics—uncon-
vinced, in particular, that the prospects for an empiricist line on mathematics are as
bad as Brown (2010, pp. 75–8) suggests—and also doubt that the regularity view of
laws, or a somewhat more sophisticated derivative form of anti-realism concerning
laws, is dead. But naturally I cannot treat these issues seriously in this chapter. So what
I propose to do is just to accept these premises of Brown. If anything, this stacks the
deck against empiricism concerning thought experiments.
Now Brown (2010, ch. 2) also develops a taxonomy of thought experiments, as
depicted in Figure 6.2. The overarching division is between destructive and construc-
tive types. The former refute or disconfirm existent theories/hypotheses. The latter
result in new findings. They may articulate existing theories (in mediative cases), lead
to the positing of theories (in conjectural cases), or result in (and confirm) new theo-
ries (in direct cases).7 There is also a special class of thought experiments, Platonic
ones, which are simultaneously destructive and constructive (and direct). Brown takes
these, and these alone, to grant insight into connections between universals.
Now according to Brown (2010, p. 40), Stevin’s thought experiment, discussed pre-
viously, is direct but not Platonic because it does not contain a destructive component;
specifically, it is not true that it ‘destroys or at least presents serious problems for a the-
ory’ (Brown 2010, p. 33). However, it seems to me that the distinction between direct
and Platonic is arbitrary from an epistemic perspective, in so far as it is historically
6 See, for example, Rowbottom (2010).

7 Conjectural cases are different from direct cases in so far as they do not result in any one theory in
particular. Rather, the ‘point of such a thought experiment is to establish some (thought-experimental) phe-
nomenon; we then hypothesize a theory to explain that phenomenon’ (Brown 1991, p. 40). Stevin’s thought
experiment, which Brown classifies as direct, does not seem to be like that. It appears instead that a lone
theory ‘falls out’.
TE
Destructive Constructive
Direct Conjectural Mediative
Platonic
Figure 6.2 Brown’s Taxonomy of Thought Experiments
contingent. To be more specific, any direct thought experiment will implicitly rule
out (or tell against) theories that are inconsistent with those that it tells in favour of.
Imagine, for example, that Stevin had started with the general theory that ‘For any
non-looped chain, there will be at least one triangular prism—with angle ACB at an
appropriate value, between zero and ninety—that it will fall off when draped over.’
(Perhaps he did start with this theory, as a matter of fact, but abandoned it in his pres-
entation of the thought experiment.) Then the thought experiment would have been
Platonic, it seems to me, and have involved grasping contingent relations between uni-
versals on Brown’s view. I therefore take it to be a pertinent example. (Note, nonethe-
less, that nothing I will argue in the remainder of this section requires that Stevin’s
thought experiment be classified one way or the other. My argument will still go
through.)
Having accepted many premises that I personally reject, I will now argue that empir-
icism concerning thought experiments is nonetheless a reasonable position, due to
a key lacuna in the Platonic account of Brown. This involves a mystery that lies at its
heart, concerning how we perceive/intuit relations between universals. But before
I come to explain this, there is an important caveat.
It may be tempting to think that intuitions are mysterious in a way that perceptions
are not at all, but I follow Brown (2010, pp. 81–2, p. 108) in thinking that this is a mis-
take. Even if we take all of our physics at face value—in a way that I, as a philosopher
with sceptical tendencies, am hesitant to do—it seems that our story ceases at a crucial
juncture. Electromagnetic waves of a range of frequencies enter the eye, forming an
image on the retina. Waves of particular frequencies cause our photoreceptor cones
(and/or rods, and/or ganglions) to fire, meaning that the photoreceptive pigments
in these change conformation. Ultimately, through a complex biochemical pathway,
this results in a membrane potential and an electrical signal being generated. In some
cases, e.g. in the case of rods, this signal is strongly amplified (so that even a lone pho-
ton can be detected).
But now what? I do not claim to be an expert in neuroscience, but I know that
although the story may go on a little further, i.e. that we may be able to say which parts
of the brain ‘light up’ when we’re seeing things (rather than, say, when we have our eyes
closed), it ends before it gets anywhere near explaining how we generate perceptual
beliefs. To see this, just consider the distribution of photoreceptor cones, which are pri-
marily responsible for our colour vision in daylight, in the eye. They are most densely
packed in the central region (about 2% in total) of the retina, the macula, which we use
when we perform tasks that require higher resolution vision (such as reading) than is
needed, say, to be aware of movement. Yet although this explains some aspects of our
visual experiences, e.g. that things appear sharper in the centre of our visual fields, it
fails to explain others.8 For example, why do so many people fail to notice when they are
losing peripheral vision (when they would notice if their macula degenerated)? Saying
that half of the visual cortex is devoted to processing macular signals does not answer
the question. Even if we add the dubious assumption of mind–brain identity, and dis-
regard the problem of qualia, what precisely is the brain doing? There are gaps between
photoreceptor cones. But there are no gaps in our visual field. We don’t really under-
stand what is going on. And this is after granting far more than I ever would if pushed.
In fact, it is plausible that the mystery of perception underlies several traditional
debates in philosophy of science, in so far as (often relatively undefended) assump-
tions about perception serve to influence stances on the so-called ‘aim of science’ and
the means by which to achieve it.9 For Mach (1893 [1960]), for example, the aim of
science is to save the phenomena just because there are no physical objects, above and
beyond bundles of sense impressions, to speak of. And for Bas van Fraassen (1980), the
distinction between the observable and the unobservable has special significance, in
part if not in whole, because he is a direct realist.10 So, on this view, we do not infer the
existence of observable physical things, in the way we do the existence of unobservable
physical things on the account of a scientific realist.
However, I do think that there is a key mystery in Brown’s account that lacks a cor-
relate in our best accounts of perception, and which may be avoided by appeal to
(relatively uncontroversial claims concerning) perception. Let’s start, to reiterate,
by accepting Brown’s premises; let’s imagine that there are universals, genuine intui-
tions involving abstracta in mathematics, and that laws of nature involve connections
between universals.
8 Even here, I am perhaps being too generous. In the words of Rossi and Roorda (2010): ‘Visual resolution
decreases rapidly outside of the foveal center. The anatomical and physiological basis for this reduction is
unclear.’
9 I am not a fan of the phrase—for reasons I explain in Rowbottom (Forthcoming)—but use it here
because of the role it plays for van Fraassen (1980).
10 James Ladyman (2000) concurs that van Fraassen appears to be a direct realist. At the very least, it is
clear from the following passage that he is not an indirect realist:
Such events as experiences, and such entities as sense-data, when they are not already under-
stood in the framework of observable phenomena ordinarily recognized, are theoretical
entities. They are, what is worse, the theoretical entities of an armchair psychology that can-
not even rightfully claim to be scientific. I wish merely to be agnostic about the existence of
the unobservable aspects of the world described by science—but sense-data, I am sure, do
not exist. (van Fraassen 1980, p. 72)
But now let us consider how we come to be acquainted with—or to use the language
of Bertrand Russell (1911), gain knowledge by acquaintance of—the universals relevant
to the laws of natural science. The short answer, again following Russell (1912), appears
to be that this is via their instantiation by observable things. So in short, we become
acquainted with universals such as circularity via acquaintance with circular concrete
things.11 At the very least, there are many universals used in scientific laws that we
would not have become acquainted with in the absence of specific sensory modalities.
It is uncontroversial, for example, that no human can grasp redness unless they have
seen a red thing; so if we were all born blind, and remained blind, no-one would speak
of redness at all. It is also hard to deny that experience is the sole means by which we
become acquainted with the universals that are the special concern of natural scientific
theories; those relating to charge and mass, for example. And Brown does not offer any
argument to the contrary.
So far, there is no problem for Brown’s view. But let us now ask how we become
acquainted with the contingent relations between universals that Brown discusses.
Let us concede, for the sake of argument, that some necessary relations are appar-
ent; let’s accept it is clear that all circular objects are shaped objects, for instance,
due to a connection between the relevant universals (circularity/circular-ness and
shaped-ness). (As the connection between these two kinds of property is manifest,
it seems we must accept this when we’ve granted the existence of the universals.12)
This presumably entails that we have the ability to intellectually grasp some meta-
physical modalities. But how about the contingent relations that Brown takes to be
the basis of scientific laws, i.e., that determine modalities in the nomic dimension?
How exactly does one spot the connection between being extended and being mas-
sive, for example?
Here, I contend, Brown has nothing to say. But the empiricist who buys into the
same account of laws may, by contrast, say that we come to posit (and perhaps even
confirm the existence of) such connections through experience of co-instantiation of
the relevant universals.13 There is a long tradition in the philosophy of science of think-
ing in this way, which goes back to at least Francis Bacon (1620 [1994]).
Moreover, the empiricist can explain why we only sometimes spot the contingent
connections between universals; this is because we only sometimes spot that the uni-
versals in question are co-instantiated, and/or only sometimes pay attention to the
11 A critic might urge that we only ever see approximately circular things, after Plato. However, this seems
presumptuous in so far as we may possess sense impressions, distinct from the physical things we observe, or
at least ways of perceiving circularly (on an adverbial theory of perception). In any event, for present purposes
I am content to accept that seeing things that approximately instantiate some universal is sufficient for grasp-
ing that universal.
12 I should add that I think that there is a mere conceptual necessity, not a metaphysical necessity, here. For
the record, I do not believe in universals.
13 For an anti-inductivist, which Brown (2010, p. 32) is emphatically not, the desirability of appealing to
intuition may be greater. This is because experience could only ever lead us to falsify claims about how uni-
versals were contingently linked. It could never confirm them. See Rowbottom (2010).
relevant patterns of co-instantiation. I cannot see a rival rationalist story. In summary,

then, the empiricist is able to provide an answer to two questions that the rationalist
cannot on Brown’s very own preferred account of laws.
6.4. Norton on Thought Experiments as

Arguments
If Brown’s account of thought experiments is rejected, then what are we to put in its
place? The most popular alternative, which has been championed by Norton (1991,
1996, 2002, 2004) in a series of papers, in which he covers an impressive array of his-
torical examples, is that thought experiments are simply arguments which:
(i) posit hypothetical or counterfactual states of affairs, and (ii) invoke particulars
irrelevant to the generality of the conclusion. (Norton 1991, p. 129)
What kinds of arguments are they? ‘A very broad range of argument forms should be
allowed . . . [including] inductive argument forms’ (p. 129). And it follows that some
thought experiments are considerably better than others: ‘A good thought experiment
is a good argument; a bad thought experiment is a bad argument’ (Norton 1991, p. 131).
It would therefore seem that deductive cases, where the ‘particular-free conclusion fol-
lows deductively from the premisses’ (p. 131) are exemplars of good thought experi-
ments; and gladly, Norton (1996, p. 350) thinks Stevin’s is just such a case.
So on Norton’s view, we can decide between empiricism and rationalism concerning
scientific thought experiments by examining the grounds for belief in (or sources of)
their premises.14 (Naturally it might turn out that neither epistemological view is cor-
rect across the board; the premises of different thought experiments may be grounded
in a variety, even a wide variety, of fashions.) In the case of Stevin’s thought experiment,
at least as I have presented it, experience appears to be enough. It is a matter of expe-
rience that chains don’t move in actual circumstances similar to those hypothetical
circumstances outlined by Stevin.
But are all thought experiments really arguments? Michael Bishop (1999) argues
that they are not, by asking us to consider a well-known exchange between Niels Bohr
and Albert Einstein. Without getting bogged down in unnecessary detail, the story is
as follows. Einstein presented a thought experiment that he took to refute a view of
Bohr. And although Bohr could not respond satisfactorily at the time, he was able to do
so the very next day by discussing the same hypothetical/counterfactual circumstances.
What’s more, Einstein agreed with Bohr’s response; he accepted, that is to say, that he
had previously been mistaken.
14 Indeed, Norton (1991, p. 142) states: ‘I do not rule out the possibility that thought experiments can allow
us to gain access to universals as Brown argues. However I urge that they may do so only insofar as these
universals can be accessed via argumentation.’
Now Bishop’s concern is that this would mean that Einstein changed his mind about
what was a good argument, on Norton’s view. In the words of Bishop (1999, pp. 539–40):
[On Norton’s view,] two tokens of a thought experiment are tokens of the same thought exper-
iment-type if and only if they are tokens of the same t-argument [thought-argument] type. . .
[But] the assumption that the arguments Bohr and Einstein proposed were type-identical
makes a muddle of the episode. Consider what happened. At first, Einstein proposed a
t-argument that Bohr did not accept . . . . Later, Bohr proposed a t-argument that forced Einstein
to disown his original t-argument. Here is what one must say if one believes that Einstein and
Bohr were presenting tokens of the same t-argument-type: Bohr did not accept Einstein’s
t-argument, so he presented Einstein with a token of that same argument; and then when faced
with a token of his own t-argument, Einstein proceeded to disavow that very argument.
But might Norton not respond by suggesting that Bohr presented a different (type of)
thought experiment in response to Einstein? Bishop (1999, p. 540) suggests that this is
wrong because ‘Einstein could have rightly accused Bohr of changing the subject’, and
he did not. So the natural thing to say, Bishop thinks, is that Einstein and Bohr were
discussing the same (type of) thought experiment, but developed different arguments
on the basis of it.
However, Norton (2002) has a convincing response. He points out that the thought
experiments (and thus arguments) might be construed as different yet highly similar.
In particular, he suggests that there is just one key difference: Einstein’s thought experi-
ment occurs in classical spacetime, whereas Bohr’s occurs in relativistic spacetime. As
such, Einstein could not reasonably have accused Bohr of changing the subject. Rather,
it became clear that his initial thought experiment (implicitly) presumed that his very
own view of spacetime was incorrect! The relevance of Bohr’s thought experiment,
which introduced relativistic considerations but was otherwise identical, was evident.
6.5. Thought Experiments as ‘Argument Pumps’

I nevertheless agree with the verdict of Bishop (1999) that thought experiments are
not arguments, and wish to argue for it with a different example. Specifically, I wish to
return to Stevin’s thought experiment, and consider how different philosophers pre-
sent it. In particular, I will be interested in the different way that they present the argu-
ment for the principle of the inclined plane.
Recall my initial presentation of Stevin’s thought experiment; I took it to be a fun-
damental premiss that the joined string, depicted in Figure 6.1, does not move.15 And
I probably took this from Mach (1893 [1960], p. 34), who writes of ‘the assumption from
which Stevinus starts, that the endless chain does not move’ at one point. However, this
15 This is how I remembered Mach’s presentation of Stevin’s argument when I began work on this chapter.
I also have evidence for this claim, because I have written on Stevin’s thought experiment previously in a
handbook entry. See Rowbottom (2014).
is not a premiss of the argument presented by Mach (1893 [1960], p. 33) on the page
beforehand:
The chain will either be in equilibrium or it will not. If we assume the latter to be the case, the
chain, since the conditions of the event are not altered by its motion, must, when once actually
in motion, continue to move forever, that is, it must present a perpetual motion, which Stevinus
deems absurd. Consequently only the first case is conceivable. The chain remains in equilibrium.
The symmetrical portion . . . may, therefore, without disturbing the equilibrium be removed.
Assuming that the translation is faithful, this treatment is imprecise. Clearly the second
case is conceivable; I can picture, in my mind’s eye, such a chain moving unceasingly.16
Such movement is judged impossible because of assumptions such as ‘the conditions of
the event are not altered by its motion’ and ‘perpetual motion . . . [is] absurd’. But there
is nothing in the scenario described by Stevin—as depicted in Figure 6.1—to suggest
that such assumptions are true. These are imported. One might think that ‘The chain
remains in equilibrium’ for many different reasons; indeed, one might be convinced
that the chain would not move, as I was, without even considering the notion of per-
petual motion. (I believe my conviction rests on experience of other similar scenarios;
that is, on old experience.)
Now Brown (2010, p. 3), as we saw in the initial presentation of Stevin’s thought
experiment, instead considers an open chain lying over a triangular prism, and then
considers what would happen if links were added so as to create the situation depicted
in Figure 6.1: ‘[by] adding the links at the bottom we make a closed loop which would
rotate if the force on the left were not balanced by the force on the right’. This proceeds
in the opposite direction to Mach, who begins with the scenario in Figure 6.1 and asks
us to work out what will happen when the string is cut at particular points (or more
accurately, when the symmetrical portion is somehow ‘removed’). I take this to show
that it is only central to the thought experiment to consider the difference between
two possible setups; the means by which the transition is made is irrelevant. (And in
fact, there is no need for a transition at all. It suffices to consider two different chains,
draped over two different prisms of the same type.) However, it is fascinating that such
transitions do play a part in the arguments offered by Mach and Brown. And this sug-
gests to me that thought experiments are not arguments.
I suppose one might instead conclude that Mach and Brown present slightly different
thought experiments. This seems rather odd, however, when both take themselves to be
reporting on one and the same experiment. In short, it seems to me more natural to say
that they have constructed different arguments on the basis of the same thought experi-
ment than it does to think that they have constructed different thought experiments
as well as different arguments. I contend that both grasped the key point that Stevin
sought to make, and the importance that understanding what happens in one kind of
16 One might raise the difference between prima facie and ideal conceivability, introduced by Chalmers
(2002), in an attempt to defend Mach’s view. However, it seems to me that the second case is ideally conceiv-
able. See also Sidelle (2002, §1).
hypothetical setup (with a string/chain forming a closed loop) has for understanding
what occurs in another (involving an open string/chain).17 Bishop (1999, p. 540) offers
an additional line of argument in support of this conclusion, by driving a wedge between
the notion of repeating an experiment and that of repeating an argument:
Thought experiments, like real experiments, can be repeated. And in order to repeat an experi-
ment, it is not necessary (or even possible) to duplicate the original in all its details. In fact, if one
thinks that an experiment has been botched, it would be folly to try to duplicate it, mistake and all.
Even putting the aforementioned difference of presentation to one side, moreover,

there are many other differences in the way Stevin’s thought experiment, and/or the
argument to be derived from it, are presented. Mach asks us to consider two possi-
bilities; equilibrium and non-equilibrium. Brown asks us to consider three; the chain
stays still, the chain moves clockwise, and the chain moves anti-clockwise. Mach
appears not to think the mention of perpetual motion is crucial to Stevin’s argument.18
But Brown (2010, p. 4) thinks that ‘the assumption of no perpetual motion machines
is central’. And Norton (1996, p. 350) seems to agree, when he presents the argument in
the following ‘capsule form’:
1. Assumption: The experimental arrangement is as shown in Figure [6.1].
2. Assumption: Perpetual motion is impossible.
3. From 1 and 2: The chain is in static equilibrium.
Let’s stick with Norton, for a moment, and also look at his treatment of the scenario
depicted in Figure 6.1 in slightly greater detail:
If the distribution of weight of the chain were to lead it to rotate about the prism, then it would
do so endlessly, since the distribution of the homogeneous chain about the prism would be unal-
tered by such rotation. (Norton 1996, p. 349)
Again, there are some puzzles here. Why should we assume that the distribution of the
homogenous chain about the prism would be causally unaltered by such rotation? And
must the density of the material forming the chain remain uniform under rotation?
(Imagine that the motion warmed the balls on the string to a temperature where they
started to melt.) My point in raising these possibilities is just to illustrate how much is
being imported in order to make these arguments seem plausible. Certainly it is hard
to see explicitly how the ‘argument is deductive’, as Norton (1996, p. 351) claims, unless
17 Alternatively, an advocate of the view that thought experiments are arguments might suggest that I have
reconstructed Stevin’s argument in a superior, or simply different, way. Indeed, it seems that Mach (1893
[1960]), from whom many philosophers will first have learned of Stevin’s thought experiment, does a rather
fine job of improving on Stevin’s lengthy and complicated initial presentation. At the very least, the presenta-
tions are strikingly different. See Kühne (2001, pp. 320–2).
18 At least, recall, this seems true if Mach (1893 [1960], p. 34) genuinely thought that ‘the assumption from
which Stevin starts [is] that the endless chain does not move’. As I earlier pointed out, however, Mach does
mention perpetual motion in his initial discussion of the thought experiment, just one page before. So per-
haps the correct view is that his treatment is not internally consistent, and/or that he had two distinct argu-
ments in mind.
we are expected to assume a good deal of physics.19But if we assume this physics, the
worry is that the principle of the inclined plane is also assumed (although not explic-
itly), and is not believed directly on previous experience.
So why do I think that all these different arguments derive from the same thought
experiment? The answer is that they all concern highly similar ‘hypothetical or coun-
terfactual states of affairs’. To be more specific, they all involve the positing of the same
hypothetical objects (e.g. chains) undergoing similar kinds of manipulations (e.g.
being cut or extended and joined up). Put simply, my view is that a thought experiment
is just an experiment that involves ‘hypothetical or counterfactual states of affairs’, and
which serves as an argument pump. Not any old argument will appropriately relate to
those states of affairs, in any given context; but many arguments will.
Does the ‘argument pump’ view have any special virtues? First, following something
akin to Roy Sorensen’s (1992, p. 214) principle of parity, it is satisfying that we may say
that actual experiments are ‘argument pumps’ too.20 We often disagree about what an
actual experiment shows in so far as we construct different arguments on the basis of
it, as opposed to questioning its results. For example, it is commonplace to ‘explain away’
apparent conflicts between theory and experimental results by appeal to factors that
have not been considered in generating predictions. Thus any good secondary school
pupil can explain why experiments in mechanics don’t give quite ‘the right answer’ by
appeal to friction, and so forth.
Second, it becomes explicit that the kind of theory-ladenness involved in obser-
vations in standard experiments also occurs in thought experiments. In each kind of
experiment, theoretical context affects which arguments are generated and/or thought
to be sustainable on the basis of the states of affairs. The only difference is whether the
states of affairs are actual rather than hypothetical and/or counterfactual.
Third, the ‘argument pump view’ is attractive because it is explicitly compatible
with the operational perspective on scientific thought experiments championed by
It may help to compare the more detailed reconstruction of Laymon (1991, p. 170):
19
Tx = x is a situation of the sort described by Stevin, namely, one where there’s a prism with
hypotenuse parallel to the ground, with a rope wrapped around it that has equally spaced
balls attached and that is friction free, &c.
Ex = x is in equilibrium.
Px = x is a situation where there is perpetual motion (of Aristotelian type) . . . [O]‌ur three
premises are:
(1) ∃x(Tx & x = a)
(2) Ea ∨ ~Ea
(3) ~∃x(Px)
Stevin’s argument that the rope is in equilibrium can be represented as:
(4) ∃x(Tx & x = a) & ~∃x(Px) & (Ea ∨ ~Ea) → Ea (claimed logical fact)
(5) Ea (by 1, 2, 3, 4 and some logical cousin of modus ponens)
In short, my worries concern premiss 4, which I take Norton to be arguing for in the previous quotation.
20 Sorensen (1992, p. 214) proposes ‘a parity thesis: thought experiments are arguments if and only if exper-
iments are arguments’. And my proposal is compatible with ‘thought experiments are argument pumps if
and only if experiments are argument pumps’.
Marco Buzzoni (2008), where ‘all [scientific] thought experiments may conceivably
become real experiments and all real experiments may be conceived as realized [sci-
entific] thought experiments’. I take this to be an independently appealing viewpoint,
at least in so far as ‘experiment’ takes the same referent. However, I think that thought
experiments can conceivably ‘become real experiments’ not in the sense that they
might become actual, but only in so far as they could be performed in different possible
worlds. (I do not believe in actual frictionless planes, rather than approximations to
these, for instance. And part of the distinct value of thought experiments is that they
can involve situations that are physically impossible, in addition to those difficult or
impossible to realize in practice.) In any event, it is interesting to see what Buzzoni
(2008, p. 69) says about Stevin’s experiment:
Stevin’s thought experiment is not demonstratively powerful because it can be reconstructed as
an argument; rather, it can be reconstructed as an argument because, as soon as we see the appa-
ratus built by Stevin and follow through the few steps of his experiment, we are persuaded of its
validity. If we reconstructed Stevin’s thought experiment as an incorrect argument, we would
question our reconstruction rather than the experiment.
So it is possible for the holder of the ‘argument pump’ view to agree with the gist of the idea
that ‘we analyse and appraise thought experiments by reconstructing them explicitly as
arguments and testing them against just those standards which we apply to arguments of
other forms’ (Norton 1991, p. 142). What she will deny is that there is any reconstruction of
the thought experiment, rather than reconstruction of the line of reasoning of some author
based on the thought experiment. And in fact, the holder of the ‘argument pump’ view
can venture an explanation of why it is necessary for so much reconstruction to occur.
The problem is not typically in grasping the hypothetical and/or counterfactual states
of affairs. Rather, it consists in grasping what someone else thinking about those states
of affairs takes them to show. As already implied, in the previous mention of theory-lad-
enness, this kind of thing happens with standard experiments too. For some realists, the
Casimir effect—that two parallel uncharged conducting plates, in a vacuum, attract one
another when in close proximity—shows the existence of virtual photons. For me, and
indeed some scientists, it does not; and that’s because we take virtual photons to be mere
aids to calculation, mere instruments, rather than legitimate physical posits.21
The ‘argument pump’ view is also as amenable to scientific intuition empiricism as
Norton’s alternative. Specifically, its advocate may hold that thought experiments are: ‘not
some kind of mysterious new window onto the physical world’ (Norton 1991, p. 142).
Rather, the premises that play a part in the (good) arguments that it is possible to pump from
thought experiments may rest on experience, rather than an additional faculty or source.
Might Norton rejoin that thought experiments are not (exactly) arguments, but
instead classes of highly similar arguments? My response is twofold. First, this would
constitute a genuine shift in his position, towards mine. (Indeed, I do not think this
For more on this, see Rowbottom (2012).

21
is a shift that Norton would want to make. But it is worth considering nonetheless.)
Second, this view seems undesirable for pragmatic reasons, at the bare minimum.
(And this is true even if one thinks that the matter reduces, when such a shift is made,
to a verbal dispute about ‘experiment’.) It requires, unnecessarily, that we use ‘experi-
ment’ differently in one context than in another; for physical (or ‘real’) experiments
can be repeated, we are apt to say, just by bringing about relevantly similar actual
states of affairs. If we can say the same about thought experiments, without any loss in
explanatory power, so much the better. Among other things, doing so encourages us
to look to physical experiments to understand the conditions for hypothetical and/or
counterfactual states of affairs to be part of an experiment. For example, just as to acci-
dentally knock a glass over is not to perform a physical experiment, so writing fiction is
not necessarily to perform a thought experiment. Experiments involve states of affairs
intentionally ‘brought about’—physically or mentally— in order to solve problems.22
Bibliography
Armstrong, D. (1983). What Is A Law of Nature? Cambridge: Cambridge University Press.
Ayer, A. J. (1956). “What Is A Law of Nature?,” Revue Internationale de Philosophie 10, pp. 144–65.
Bacon, F. (1620 [1994]). Novum Organum, trans. P. Urbach and J. Gibson. La Salle: Open Court.
Baker, A. (2005). “Are There Genuine Mathematical Explanations of Physical Phenomena?,”
Mind 114, pp. 223–38.
Bishop, M. A. (1999). “Why Thought Experiments Are Not Arguments,” Philosophy of Science
66, pp. 534–41.
Brown, J. R. (1991). The Laboratory of the Mind: Thought Experiments in the Natural Sciences, 1st
ed. London: Routledge.
——. (2010). The Laboratory of the Mind: Thought Experiments in the Natural Sciences, 2nd ed.
London: Routledge.
Buzzoni, M. (2008). Thought Experiment in the Natural Sciences: An Operational and
Reflexive-Transcendental Conception. Würzburg: Königshausen and Neumann.
Chalmers, D. J. (2002). “Does Conceivability Entail Possibility?,” in T. S. Gendler and J.
Hawthorne (eds), Conceivability and Possibility. Oxford: Clarendon Press, pp. 145–200.
Colyvan, M. (2001). The Indispensability of Mathematics. Oxford: Oxford University Press.
Dretske, F. (1977). “Laws of Nature,” Philosophy of Science 44, pp. 248–68.
Gopnik, A., and Schwitzgebel, E. (1988). “Whose Concepts Are They, Anyway? The Role of
Philosophical Intuition in Empirical Psychology,” in M. R. DePaul and W. Ramsey (eds),
Rethinking Intuition. Lanham, MD: Rowman & Littlefield, pp. 75–91.
22 Moreover, rather interestingly, it appears that physical experiments can be fit, roughly at least, into
the classification scheme for thought experiments proposed by Brown, which I discussed in section 6.3.
Certainly, physical experiments can be either destructive or constructive on most views of scientific method;
they can refute and/or confirm/corroborate theories. Furthermore, they can illustrate phenomena that lead
us to conjecture theories (i.e. be ‘conjectural’). They can also ‘articulate’ theories (i.e. be ‘mediative’) on the
view of Kuhn (1996, ch. 3); see Rowbottom (2011b, 2011c) for some examples. Whether they can be ‘direct’
may be somewhat more controversial. But think of it this way. They can prompt ‘direct’ deductive arguments
for one theory in particular, if appropriate premises are assumed.
Jenkins, C. S. I. (2014). “Intuition, ‘Intuition’, Concepts and the A Priori,” in A. R. Booth and D. P.
Rowbottom (eds), Intuitions. Oxford: Oxford University Press, pp. 91-115.
Kuhn, T. S. (1996). The Structure of Scientific Revolutions, 3rd ed. Chicago: University of
Chicago Press.
Kühne, U. (2001). “Die Methode des Gedankenexperiments,” unpublished PhD thesis, University
of Bremen. (Abridged version published (2005) as: Die Methode des Gedankenexperiments.
Frankfurt am Main: Suhrkamp Verlag.)
Ladyman, J. (2000). “What’s Really Wrong with Constructive Empiricism? Van Fraassen and
the Metaphysics of Modality,” British Journal for the Philosophy of Science 51, pp. 837–56.
Laymon, R. (1991). “Thought Experiments by Stevin, Mach and Goy: Thought Experiments
as Ideal Limits and as Semantic Domains,” in T. Horowitz and G. J. Massey (eds), Thought
Experiments in Science and Philosophy. Lanham, MD: Rowman & Littlefield, pp. 167–91.
Mach, E. (1893 [1960]). The Science of Mechanics: A Critical and Historical Account of Its
Development, 6th ed. La Salle: Open Court.
Norton, J. D. (1991). “Thought Experiments in Einstein’s Work,” in T. Horowitz and G. Massey
(eds), Thought Experiments in Science and Philosophy. Lanham, MD: Rowman & Littlefield,
pp. 129–48.
——. (1996). “Are Thought Experiments Just What You Thought?,” Canadian Journal of
——. (2002). “Why Thought Experiments Do Not Transcend Empiricism,” in C. Hitchcock
(ed.), Contemporary Debates in the Philosophy of Science. Oxford: Blackwell, pp. 44–66.
——. (2004). “On Thought Experiments: Is There More to the Argument?,” Philosophy of Science
71, pp. 1139–51.
Rossi, E. A., and Roorda, A. (2010). “The Relationship between Visual Resolution and Cone
Spacing in the Human Fovea,” Nature Neuroscience 13, pp. 156–7.
Rowbottom, D. P. (2010). “Evolutionary Epistemology and the Aim of Science,” Australasian
Journal of Philosophy 88, pp. 209–25.
——. (2011a). Popper’s Critical Rationalism: A Philosophical Investigation. London: Routledge.
——. (2011b). “Stances and Paradigms: A Reflection,” Synthese 178, pp. 111–19.
——. (2011c). “Kuhn Vs. Popper on Criticism and Dogmatism in Science: A Resolution at the
Group Level,” Studies in History and Philosophy of Science 42, pp. 117–24.
——. (2012). “The Instrumentalist’s New Clothes,” Philosophy of Science 78, pp. 1200–11.
——. (2014). “Educational Research as Science?,” in A. Reid, E. P. Hart, and M. Peters (eds),
A Companion to Research in Education. Dordrecht: Springer, pp. 145–53.
——.(Forthcoming). “Aimless Science,” Synthese. DOI: 10.1007/s11229-013-0319-8.
Russell, B. (1911). “Knowledge by Acquaintance and Knowledge by Description,” Proceedings of
the Aristotelian Society 11, pp. 108–28.
——. (1912). The Problems of Philosophy. London: Oxford University Press.
Sidelle, A. (2002). “On the Metaphysical Contingency of Laws of Nature,” in T. S. Gendler and J.
Hawthorne (eds), Conceivability and Possibility. Oxford: Clarendon Press, pp. 309–36.
Sorensen, R. A. (1992). Thought Experiments. Oxford: Oxford University Press.
Tooley, M. (1977). “The Nature of Laws,” Canadian Journal of Philosophy 7, pp. 667–98.
van Fraassen, B. C. (1980). The Scientific Image. Oxford: Clarendon Press.
7
Novice Thought Experiments
Roy Sorensen†
7.1. Abolition or Universal Suffrage?

Experimental philosophers pose an egalitarian dilemma: either abolish the appeal to
intuition or include everybody’s intuitions as data. Intuitions are spontaneous unsup-
ported premises, not conclusions. Consequently, intuitions from novices are as cred-
ible as intuitions from experts. Granted, philosophers are knowledgeable about the
theories they hope to test by intuition. But intuitions are deployed to test theories, not
expound them.
Experimental philosophers warn armchair philosophers that knowledge of the-
ory influences the formation and interpretation of intuitions (Knobe and Nichols
2007). Novices, by virtue of their philosophical innocence, improve the quality of
the data—either by amplifying the reliability of our stock of intuitions (Nahmias
et al. 2006) or by exposing the unreliability of intuitions (as stressed by Weinberg
et al. 2001).
Novices are more diverse than philosophers. This diversity is claimed to ameliorate
biases from gender, culture, and socioeconomic class. The experimentalist urges us to
take a step back so that more data comes into view.
7.2. An Academic Universal

Let us take a further step back, one from which scientists and their novices are co-visible
with the philosophers and their novices. This cross-disciplinary perspective reveals
the uniformity of the expert–novice relationship. Novice perceptions and novice intui-
tions are ignored throughout the humanities and the sciences.
† Predecessors of this article were presented at the Thought Experiments workshop at the University
of Toronto (May 22, 2009), St. Andrews Arché Center (December 8, 2009), and the Society for Exact
Philosophy (March 20, 2010). Special thanks to Stephen Biggs, John Doris, Joshua Knobe, Ronald Mallon,
Shaun Nichols, and Darrell Rowbottom.
136 Roy Sorensen
Experimental philosophers aim to follow the example set by scientists. This means
emulating patterns of curiosity. But it also means emulating patterns of incuriosity.
In scientific practice, the experiments of novices (and even advanced amateurs)
never overturn expert experiments. Indeed, the conflict is automatically resolved in
the opposite direction.
No physicist (in his role as a physicist) surveys the conflict between undergraduate
physics experiments and the experiments of their professors. Physicists presuppose
that there is massive conflict. They regard systematic aspects of the divergence as edu-
cational. Students learn from the mistakes exposed when they deviate from profes-
sionally conducted experiments. Novice experiments are never used to challenge or
corroborate textbook results (Kuhn 1963).
This is not because the novice experiments are only intended as practice. Scientists
have made serendipitous discoveries with experiments they intended only as practice
or merely as classroom demonstrations. In 1820 Hans Christian Ørsted used a compass
needle to illustrate the independence of electricity and magnetism. When the needle
moved in response to current flow in a wire, Ørsted realized his lecture demonstration
proved the reverse of what he intended. The intentions generating evidence are irrel-
evant to what that evidence establishes.
Science textbooks have many mistakes. When top physicists vet the textbooks, they
are appalled (Feynman 1985, pp. 288–302). Yet they never appeal to student experi-
ments to vindicate their corrections.
A psychologist might conduct a survey of novice experiments. And physicists are apt
to be interested in the psychology of physics. But the psychology of physics is not phys-
ics. And the psychology of philosophy is not philosophy.
As an epistemologist, I follow developments in the psychology of reasoning and the
psychology of perception. I read experimental philosophy in the same interdiscipli-
nary spirit. Although some of their early work is crude polling, experimental philoso-
phers have learned from professional experimenters. Consequently, their articles look
more and more like psychology articles—because that is what they are!
7.3. Ignoring Novice Thought Experiments

What goes for experiment goes for thought experiment. Physicists heed their own thought
experiments but ignore the thought experiments of novices. They do not care, as physi-
cists, whether novice answers vary with gender, socioeconomic status, culture, and so on.
Experimental philosophers fear that as students take more philosophy courses, they
come to agree with the classic thought experiments. Yet physics instructors hope for
the same trend as applied to physics thought experiments. The physics instructors are
dismayed when their graduate students continue to manifest intuitions that deviate
from Newton’s laws of motion.
What we learn from an experiment is not limited to what that experiment aims
to demonstrate. An experiment can reveal that the experimenter is left-handed or
Novice Thought Experiments 137
hypertensive or epileptic. Some of this collateral information can be scientifically rel-

evant. For instance, students tend to get the experimental result they expect to get.
Examples of this expectancy bias led Robert Rosenthal (1978) to wonder whether the
expectancy bias affects professional experiments. It does. Bias in published experi-
ments is notoriously common. D. L. Sackett (1979) presents a taxonomy of fifty-six
biases in clinical studies.
The parapsychology experiments of Joseph Banks Rhine taught methodologists
the “decline effect.” Rhine would get a strong result for clairvoyance but then see the
scale of the effect decrease or disappear. Jonathan Schooler has replicated this failure
of replication (Lehrer 2010). It has been generalized to Schooler’s own research on
verbal overshadowing, Ander Møllers’ link between sexual attraction and symmetry,
and biomedical research (Ioannidis 2005). Some of the decline effect can be attributed
to familiar causes: regression to the mean, the greater margin of error that plagues
the small samples of pilot studies, and the publication bias for positive results. But to
account for all of the effect, we are forced into the direction indicated by Rosenthal.
Experimenters shoehorn data to fit hypotheses.
Experimental philosophers have found biases in novice thought experiments
that may generalize to professors. For instance, students performing Keith Lehrer’s
Truetemp thought experiment are vulnerable to an order effect (Swain et al. 2008). In
earlier thought experiments, I tended to preface knowledge attributions I find ques-
tionable with paradigm cases of correct knowledge attributions. But given that novices
can be biased, I worried that I might be biasing colleagues. The Truetemp bias was only
investigated for novices, not experts. Nonetheless, to be on the safe side, I now rand-
omize my presentation.
The authors of the Truetemp study present their survey results as alarming. Yet
their evidence is far less comprehensive than Rosenthal’s evidence for the expectancy
bias or the decline effect. The authors of the Truetemp study do not worry that these
results throw their Truetemp experiment in doubt. Instead, they correctly regard the
expectancy bias as a nuisance that needs to be controlled. The same sense of propor-
tion should be applied to their discovery of their order effect. Order effects have been
demonstrated for master chess players and many other types of experts (Lewandowsky
et al. 2009). Since the chess masters perform well despite these shortcomings, the
impact on performance must be minor.
7.4. The Resilience of Superior Performance

Professional baseball players are notoriously superstitious (Burger and Lynn 2005).
They harbor many of the same prejudices as amateurs. They commit many of the same
fallacies. Their robustly superior performance shows that these shortcomings are
swamped by positive factors. Showing that an expert is vulnerable to the same weak-
nesses as a novice does not go far in demonstrating that the expert fails to deliver supe-
rior performance.
138 Roy Sorensen
Indeed, the persistence of framing effects, biases, and illusions invites the further
conjecture that the expert’s flaws are sometimes not worth correcting. Just as an organ-
ism has a limited budget for removing parasites and pathogens, it has a limited budget
for removing cognitive inefficiencies.
Re-training also carries some risk of backfiring. A pitcher who takes apart his game
to remove one bad habit may wind up unable to put his game back together again.
Performance is sometimes globally enhanced by combination effects that can be
undone by a single local “improvement.”
For the most part however, the resilience of superior performance is compatible
with perfectionist attention to rectifying small flaws. Individually insignificant defects
often sum to a significant problem. In Moneyball: The Art of Winning an Unfair Game,
Michael Lewis relates how conventional baseball fell into a sloppy, arrested state of
development. An ardent baseball fan, Bill James, developed more detailed measures
of performance. His original goal was to prevail in a fantasy baseball competition.
However, James’ new statistics suggested new strategies and tactics such as ‘Never bunt’
and ‘Get on base by walking’. Fantasy baseball is blind to the physical appearance of
players and so is more objective. Eventually, economic exigency led the manager of the
impoverished Oakland Athletics, Billy Beane, to apply the fantasy strategies to actual
team management. Beane’s success has spread Mr. James’ revolution to other teams.
7.5. Intuition as an Automated Response

A philosopher who innately grasps the distinction between left and right can more
easily follow the thought experiments in Ned Block’s 1974 article “Why Do Mirrors
Reverse Left/Right But Not Up/Down?” A philosopher who must reason through
questions about left–right orientation will overload his working memory.
Intuition is an unreasoned, automatic response. However, the automation normally
reflects learning. Indeed, learning is itself a matter of automating responses that ini-
tially take conscious reasoning—what K. Anders Ericsson calls “deliberate practice.”
Critical thinking is flexible but taxes our small budget of attention. Automation trans-
fers the cognitive burden to unconscious processes that trade flexibility for speed and
reliability.
The physical intuitions of a physics professor are more reliable than the physical
intuitions of those newly enrolled in his “Introduction to Physics.” When students
are asked to picture the path of a ball exiting a curved tube, they visualize the ball as
continuing its curved path—instead of the straight continuation of those who have
been taught Newton’s laws of motions (McCloskey et al. 1980). See also Gentner and
Gentner (1983), for example, on novice and expert conceptions of water flow.
Learning overwrites intuitions rather than erases them. Thus a distracted expert
may answer like a novice. Even when focused, he is apt to have both the common sense
intuitions and the intuitions acquired through study. Part of his expertise lies in his
sense of which of these spontaneous judgments should be heeded and which should be
ignored. A physicist lecturing on pendulums will release a tethered bowling ball from
directly in front of his face. As the ball returns, the physicist will fear that the ball will
hit him in the face. Yet the physicist will also have the dominating intuition that the ball
cannot touch him. He acquired this intuition from a study of the law of equal height
for pendulums. This second intuition will give him the courage to stand his ground.
He can remain in the ball’s path with the same enjoyable ambivalence he experiences
when riding in a roller coaster at an amusement park.
Galileo’s research on pendulums led to experiments in which the ball rolled on an
inclined plane rather than swung by a tether. His anticipation of Newton’s first law is
based on idealized variations of his inclined plane experiments (frictionless planes,
perfectly rigid bodies, and so on). Galileo had no direct test for the accuracy of these
answers. However, research transformed Galileo into an authority on these hypotheti-
cal issues.
7.6. Intuition and Speech Perception

Expert chess players do not consider more possibilities than novice chess players. They
consider possibilities of a higher quality. Instead of focusing on individual pieces (the
bishop being on a black square), the expert sees complex unities (such as a king-side
castle behind a fianchettoed bishop). Short-term memory is limited to about seven
chunks of information. However, people learn how to organize the information into
larger chunks:
A man just beginning to learn radio-telegraphic code hears each dit and dah as a separate chunk.
Soon he is able to organize these sounds into letters and then he can deal with the letters as
chunks. Then the letters organize themselves as words, which are still larger chunks, and he
begins to hear whole phrases. (Miller 1956)
Mastering chess is like learning how to hear telegraphic messages. It is more of a per-
ceptual ability than a reasoning ability.
The analogy with speech perception carries through to learning logic itself. The logic
student cannot improve his basic capacity to reason. However, he can learn how to
organize information about arguments into larger chunks. His logical vocabulary will
help him recognize, appraise and reconstruct arguments.
7.7. Holistic Advantages

Expertise is surprisingly local. Being an expert at chess does not make you an expert
at checkers—despite the fact that they are played on the same board. Indeed, in cases
of negative transfer, expertise in one area diminishes performance in another area (as
when a tennis player imports a two handed backhand into squash).
Chunky intuitions can be decomposed into parts for debugging. The chunks permit
positive transfer to new applications, allowing the expert to reason more effectively by
140 Roy Sorensen
analogy. Major breakthroughs occur by importing techniques from one discipline to

another. Micro-analogies play an important role in breeding up the stock of cases used
in the literature on personal identity, functionalism, and utilitarianism.
Since the chunks are intersubjective, they make the judgment portable. Unlike a
novice, experts have a social network to test and hone these judgments. They know
who to ask and how to ask. And they are more likely to get responses.
The expert also has a holistic advantage over the novice. The expert is familiar with a
constellation of comparison cases. Whereas freshmen evaluate a Gettier case in isola-
tion, the epistemologist considers the case against a background of similar cases (such
as those compiled in Robert Shope’s The Analysis of Knowledge). Comparative, contex-
tualized judgment is more reliable than absolute judgment.
7.8. Targeting Persuasive Thought Experiments

Disagreement between experts who run the same thought experiment weakens evi-
dential value. The same point holds for experiments. Acupuncture has been proven
effective in all forty-seven trials conducted in China, Taiwan, and Japan between 1966
and 1995. Yet only 56% of the studies show any therapeutic effect in the United States,
Sweden, and the United Kingdom (Lehrer 2010).
Such disagreements invite the suspicions so popular with skeptics about science.
In particular, they accuse thought experiments of circularity; since theory shapes the
intuitions that purport to test the theory, the method only yields deadlock between the
experts.
Theory influences everything: perception, calculation, and intuition. Often the
influence is surprisingly strong. In medical research, this has necessitated blind stud-
ies, double-blind studies, and even triple-blind studies (to ensure that the statistician is
not influenced). In economics, the influence is often surprisingly weak. Maurice Allais
reports:
During the 1952 Paris Colloquium, I had [Leonard Jimmie] Savage respond over lunch to a list
of some 20 questions. His answers to each were incompatible with the basic axioms of his own
theory. He was immediately troubled, and asked for time to think. A week later he told me that
his reactions during our lunch conversation had indeed run counter to his own axioms, but
that after further reflection, he had concluded that his responses, contrary to his axioms, were
explained by the fact that he had behaved irrationally. (Allais 1979, p. 533)
Allais thinks the circularity lies in Savage’s attempts to discount his own intui-
tions: Savage “defined rationality by his axioms, and he judged his answers irrational
because they infringed the axioms!” (Allais 1979, p. 533).
One might expect experimental philosophers to focus on cases in which philoso-
phers behave like Savage and explain away the data as a performance error. But, in
an effort to document the influence of the method they wish to debunk, experimen-
tal philosophers actually concentrate on cases in which philosophers capitulate to
hypothetical scenarios. For instance, they emphasize that Gettier’s cases overturned
the long consensus that knowledge is justified true belief. Life-long proponents of JTB
conceded defeat.
Experimental philosophers note that Saul Kripke presented his thought experi-
ments to audiences composed almost entirely of description theorists. The majority
accepted the thought experiments as refutations of their views. Experimental philoso-
phers typically open their articles by documenting the persuasive power of the thought
experiment they wish to undermine. Machery et al. say of Kripke’s Gödel-Schmidt
case: “it has turned out that almost all philosophers share the intuitions elicited by
Kripke’s fictional cases, including most of his opponents. Even contemporary descrip-
tivists allow that these intuitions have falsified traditional forms of descriptivism . . .”
(Machery et al. 2004, p. 48)
Or consider Harry Frankfurt’s counterexample against the assumption that a per-
son is morally responsible for what he has done only if he could have done otherwise.
Al Mele and David Robb report “Most contributors . . . have deemed the objection
successful. Compatibilists have used it in criticizing incompatibilism . . . , and some
incompatibilists have attempted to accommodate Frankfurt’s moral in refined incom-
patibilist views.” (Mele and Robb 2001, p. 242) Further examples: Lehrer’s Truetemp
was accepted by many externalists (Swain et al. 2008, p. 140) as was Laurence BonJour’s
Norman case accepted by externalists (Goldman 1986, p. 112). Newcomb’s problem led
Richard Jeffrey to abandon evidential decision theory. J. J. C. Smart admits to have the
anti-utilitarian intuition elicited by examples of “punishing the innocent” (Doris and
Plakias 2007).
7.9. The Topic Neutrality of Expertise

Learning theorists have become experts on expertise. News of the existence of such
meta-experts has been recently spread by the St. Paul of the social sciences, Malcolm
Gladwell (2008). Gladwell is a brilliant popularizer who makes the trade-offs needed
to convert dry scholarship into best sellers: simplification, concretization, and topical-
ity. After Gladwell hooks our interest, we are free to reverse the trade-offs by returning
to the primary sources.
The 10,000-hour rule says that expertise requires 10,000 hours of practice. By ‘prac-
tice’ the meta-experts do not mean mere participation. I swim an hour of laps each day
but I do not use the experience to become a better swimmer. I am not coached. I do not
study swimming manuals. I do not race. I do not even monitor my performance. I just
swim for exercise. When swimming, I lapse into reverie rather than make effortful
adjustments to my strokes and kicks.
In contrast, when I studied philosophy, I was diligent. I sought training from peo-
ple who were certified as experts—doctors in philosophy. The professors specialized.
Some of them taught me logic, the diagnosis of fallacies, and the analysis of arguments.
Others concentrated on specimens of excellent performance—generally organized
142 Roy Sorensen
chronologically (as in the history of philosophy) or as those with leading roles in clas-
sic debates (such as Bertrand Russell, Peter Strawson, and Keith Donnellan in the
debate over definite descriptions). I also learned informally from peers, seeking out
those who were better than I was, exploiting opportunities to debate. I went to confer-
ences, received referee reports, and wrote draft after draft after draft. I still do.
Most experts can only manage four or five hours per day of the intense concentra-
tion needed to improve performance. (Extra time beyond that is largely wasted and
can even backfire as “burn out.”)
Fifty hours is generally sufficient to turn a novice into an amateur with basic compe-
tence (Ericsson and Smith 1991, p. 37). Competence is achieved by automating learned
behavior. But to reach higher levels, one must resist full automation. Most people who
engage in deliberate practice never reach elite status. They become proficient amateurs
in an arrested state of development. When they stop, they get rusty and fall below their
active level. For instance, I played chess diligently until age seventeen, and then quit.
After several inactive years I reverted to being an average chess player. This fits longi-
tudinal studies of elite swimmers who quit and then revert to being average swimmers.
Those who wish to sustain their level of performance must continue at least
small-scale participation. Expertise is destroyed by simple lack of use. An expert swim-
mer who is out of the water for just one month significantly declines. Philosophers who
stop philosophizing should experience comparable declines and an eventual reversion
to novice status.
Thought experiments figure heavily in analytic philosophy. The method is well
suited for speculative issues because thought experiments are the fastest and cheap-
est form of evidence. The method is especially useful when the subject matter is con-
troversial. Your audience need only invest five minutes of attention. If the thought
experiment is persuasive, it is easily disseminated. If the thought experiment fails to be
persuasive, the cost is low to all concerned.
The 10,000-hour rule is field neutral. It works for physicists, novelists, musicians,
soldiers, hockey players, and burglars. Generally, serious learners practice 1,000 hours
a year. The corollary is that it takes about ten years to become an expert.
7.10. Novices vs. Amateurs

Only about fifty hours of deliberate practice is needed to become competent—the level
beyond novice marked by ‘amateur’. As amateurs become more proficient, they often
perform low-level tasks under the supervision of an expert.
Amateur experiments conform to a yet broader pattern of apprenticeship. Experts
never treat conflicts with novices as alarming counter-evidence. They expect some
deviation because the performance requires many rapid, accurate sub-actions which
become automatic. Conscious effort is slow and error prone. Unless the experiment is
especially simple or closely supervised, the novice will make mistakes.
Amateurs are more accurate than experts in estimating how long something will
take to learn (Hand 1999). Whereas novices are overconfident and experts have forgot-
ten most of the false paths, freshly minted amateurs have the relevant memories and
temptations. Since businesses must estimate how much time customers will need to
master new technologies, they are advised to seek counsel from freshly trained ama-
teurs, not just the experts. Since experimenters are experts, they also suffer “the curse
of expertise.” They underestimate how much preparation is needed to competently
run an experiment. Students will appear stupid, slow, and irresponsible with dead-
lines. Since laboratory experiments are public, there is opportunity to witness students
floundering. In the case of thought experiments, the chief symptom is a clearly wrong
answer. Since experimental philosophers deny that there is a clearly mistaken answer
(to a philosophical thought experiment), they are especially vulnerable to the curse of
expertise.
Laboratory assistants and graduate students have made significant discoveries. Most
of these have been feats of recognition: discovering a new species or a well-preserved
embryo in a dinosaur egg or a lost poem by Robert Frost.
But there are some early career discoveries that required ingenuity and became
enshrined as classic results. In 1958, Matthew Meselson (a graduate student) and
Franklin Stahl (a postdoctoral student) devised the first empirical demonstration that
DNA replicates in the way predicted by the Watson-Crick double helix model. The
historian of science Frederic Holmes (2001) described the Meselson-Stahl experiment
as the most beautiful experiment in biology.
Early meteorology incorporated amateurs, such as the philosopher John Locke,
to collect widespread weather data. More recently, astronomy has incorporated
high-level amateurs into its intellectual division of labor. This remarkable process has
been chronicled by Timothy Ferris in Seeing in the Dark: How Backyard Stargazers
Are Probing Deep Space and Guarding Earth from Interplanetary Peril. Thanks to tech-
nological advances in communication and observation, an international cadre of
amateurs is now allocated territories of space to monitor, especially for unpredictable
phenomena such as comets.
Ferris has documented a genuine revolution in the relationship between experts
and amateurs. Yet it would be dwarfed by the revolution advertised by revolutionary
experimental philosophers.
7.11. Comparison with the Verbal Dispute Defense

The diplomatic response to experimental philosophy is to deny that there is any disa-
greement between philosophers and novices.
The first experimental philosopher, Arne Naess, distributed questionnaires on
‘democratic’, ‘free enterprise’, and ‘true’. Several philosophers dismissed his surveys
by invoking Rudolph Carnap’s concept of explication. Instead of reporting on usage,
144 Roy Sorensen
the philosopher is re-shaping the term for theoretical purposes (as logicians did for
‘valid’).
However, explication does not make ordinary usage completely irrelevant. After
all, the explicator is trying to preserve useful patterns. For instance, Alfred Tarski
self-consciously engaged in an explication of ‘true’. He avoids complete fidelity with
the ordinary usage of ‘true’ because he thinks the liar paradox shows that the ordinary
usage is inconsistent. Tarski only analyzes a relativized counterpart of ‘true’, ‘true in
language L ’. Yet, after the usual caveats about hazards of survey research, Tarski gin-
gerly welcomes Naess’s data:
Therefore, I was by no means surprised to learn (in a discussion devoted to these problems) that
in a group of people who were questioned only 15% agreed that “true” means for them “agreeing
with reality,” while 90% agreed that a sentence such as “it is snowing” is true if, and only if, it is
snowing. Thus, a great majority of these people seemed to reject the classical conception of truth
in its “philosophical” formulation, while accepting the same conception when formulated in
plain words (waiving the question whether the use of the phrase “the same conception” is here
justified). (1944, p. 360)
Explication is intermediate between reportive definition and stipulation—although

Carnap himself is ambivalent about how intermediate (Carnap 1963, p. 934). One could
make experimental philosophy irrelevant by retreating further out on the continuum
toward brute stipulation. For instance, Stephen Biggs interprets Kripke’s schmidentity
passage in Naming and Necessity (pp. 108–9) as substituting stipulation for intuition
(as a back-up strategy). Metaphysicians bypass empirical issues of language use by
changing the topic to hypothetical languages such as schmenglish, which has schm-
analogues of core concepts.
This retreat to stipulation makes one vulnerable to J. L. Austin’s charge that the
invented meaning is uninteresting. Recall the slogan “Philosophical problems arise in
ordinary language and so must be solved in ordinary language.”
This slogan is too sweeping. Philosophers have invented lots of interesting terminol-
ogy: heterological, tonk, grue, quus, gunk, telishment, quasi-memory. Theories of lan-
guage have implications about all possible languages. So a hypothetical language can
be a counterexample to such a theory. Nevertheless, this still leaves the scope of Biggs’s
maneuver incommensurate with many of Kripke’s ambitions.
Austin recommends reading the dictionary. In effect, he does miniature corpus
studies. For instance, in “A Plea for Excuses,” Austin draws invidious comparisons
between the nuanced distinctions of a defendant and the clunky legal terminol-
ogy of the presiding judge. Even Austin, however, permits tidying up of ordinary
language.
Ernest Sosa is the most prominent exponent of the diplomatic strategy (e.g. Sosa
2007). He says that once we set aside verbal disputes with “the folk,” experimental phi-
losophy merely shows a need to exercise greater care. We must be vigilant against per-
formance errors (bias, equivocation, etc.). The raised standard needs to be realistic lest
we condemn observation along with intuition.
My objection to experimental philosophy is intended to be more confrontational.

I think the amount of conflict between philosophers and novices is of the same mag-
nitude as the conflict between physicists and novices. I think the resolution of the con-
flict is the same.
7.12. Five Prophesies

First, actuarial evidence (number of publications, percentage of presentations at phi-
losophy conventions, job advertisements, etc.) suggests that experimental philosophy,
like the new laboratory economics, is here to stay.
Second, for reasons summarized in John List’s 2003 backlash against Amos Tversky
and Daniel Kahneman’s penetration into economics, there will be a schism between
laboratory experimental philosophy and (the yet to materialize) field experimental
philosophy.
Third, as experimental philosophy becomes more entrenched, it will moderate
(partly because it will no longer need to make provocative claims to gain attention).
Experimental philosophy will be presented as complementing thought experiment
methodology.
Fourth, experimental philosophy will have a persistent, questioned status, such
as many philosophers assign to the history of philosophy. Many philosophers deny
that history of philosophy is philosophy. They appeal to other fields that sharply
distinguish history of F from F. History of physics is not physics. History of psy-
chology is not psychology. The only exception is history itself: history of history
is history. So why are philosophy organizations pouring scarce resources into
non-philosophy? Those who regard experimental philosophy as covert psychology
will lodge the same complaint: experimental philosophy is a cuckoo-bird in the
nest of philosophy.
My fifth and final prophecy is that these strange bedfellows, historians of philosophy
and experimental philosophers, will converge on Arne Naess. They will belatedly hail
this Norwegian philosopher as the founder of experimental philosophy.
References
Allais, M. (1979). “The So-called Allais Paradox and Rational Decisions Under Uncertainty,” in
M. Allais and O. Hagen (eds), Expected Utility Hypothesis and the Allais Paradox. Dordrecht
and Boston: Reidel, pp. 437–681.
Biggs, S. (2009). “What Surveys Don’t Show about Reference,” Central APA presentation.
Block, N. (1974). “Why Do Mirrors Reverse Left/Right But Not Up/Down?,” Journal of Philosophy
71, pp. 259–77.
Burger, J. M., and Lynn, A. L. (2005). “Superstitious Behavior Among American and Japanese
Professional Baseball Players,” Basic and Applied Social Psychology 27, pp. 71–6.
Carnap, R. (1963). “P. F. Strawson on Linguistic Naturalism,” in P. A. Schlipp (ed.), The Philosophy
of Rudolf Carnap. La Salle, IL: Cambridge University Press.
146 Roy Sorensen
Doris, J. M., and Plakias, A. (2007). “How to Argue about Disagreement: Evaluative Diversity
and Moral Realism,” in W. Sinnott-Armstrong (ed.), Moral Psychology, Volume 2: The
Cognitive Science of Morality. Cambridge, MA: MIT Press.
Ericsson, K. A., and J. Smith (eds) (1991). Toward a General Theory of Expertise: Prospects and
Limits. New York: Cambridge University Press.
Ferris, T. (2002). Seeing In The Dark: How Backyard Stargazers Are Probing Deep Space And
Guarding Earth From Interplanetary Peril. New York: Simon & Schuster.
Feynman, R. (1985). Surely You’re Joking, Mr. Feynman. New York: W. W. Norton.
Gentner, D., and Gentner, D. R. (1983). “Flowing Waters or Teeming Crowds: Mental Models
of Electricity,” in D. Gentner and A. Stevens (eds), Mental Models. Hillsdale, NJ: Lawrence
Erlbaum.
Goldman, A. (1986). Epistemology and Cognition. Boston, MA: Harvard University Press.
Gladwell, M. (2008). Outliers: The Story of Success. New York: Little Brown.
Hand, P. (1999). “The Curse of Expertise: The Effects of Expertise and Debiasing Methods on
Predictions of Novice Performance,” Journal of Experimental Psychology: Applied 5, pp. 205–11.
Holmes, F. L. (2001). Meselson, Stahl, and the Replication of DNA: A History of “The Most
Beautiful Experiment in Biology.” New Haven, CT: Yale University Press.
Ioannidis, J. P. (2005). “Why Most Published Research Findings Are False,” PLoS Medicine 2, 8.
DOI: 10.1371/journal.pmed.0020124. PMID 16060722.
Knobe, J., and Nichols, S. (2007). “An Experimental Philosophy Manifesto,” in J. Knobe and S.
Nichols (eds), Experimental Philosophy. New York: Oxford University Press, pp. 3–14.
Kripke, S. (1980). Naming and Necessity. Princeton, NJ: Princeton University Press.
Kuhn, T. S. (1963). “The Function of Dogma in Scientific Research,” in A. C. Crombie (ed.),
Scientific Change. New York: Basic Books, pp. 347–69.
Lehrer, J. (2010). “The Truth Wears Off,” The New Yorker (December 13).
Leslie, A., Knobe, J., and Cohen, A. (2006). “Acting Intentionally and the Side-effect
Effect: ‘Theory of Mind’ and Moral Judgment,” Psychological Science 17, pp. 421–7.
Lewandowsky, S., and Thomas, J. (2009). “Acquisition, Limitation, and Control,” in F. T. Durso
(ed.), Human Factors and Ergonomics, Volume 5. Santa Monica, CA: Human Factors Society.
List, J. (2003). “Does Market Experience Eliminate Market Anomalies?,” Quarterly Journal of
Economics 118, pp. 41–71.
Machery, E., Mallon, R., Nichols, S., and Stich, S. (2004). “Semantics, Cross-cultural Style,”
Cognition 92, B1–B12.
McCloskey, M., Caramazza, A., and Green, B. (1980). “Curvilinear Motion in the Absence of
External Forces: Naïve Beliefs About the Motion of Objects,” Science 210, pp. 1139–41.
Mele, A., and Robb, D. (2001). “Rescuing Frankfurt-Style Cases,” in L. W. Ekstrom (ed.),
Agency and Responsibility: Essays on the Metaphysics of Freedom. Boulder, CU: Westview,
pp. 241–54.
Meselson, M., and Stahl, F. W. (1958). “The Replication of DNA in Escherichia coli,” PNAS 44,
pp. 671–82.
Miller, G. A. (1956). “The Magical Number Seven, Plus or Minus Two: Some Limits on Our
Capacity for Processing Information,” The Psychological Review 63, pp. 81–97.
Naess, A. (1938) ‘Truth’ as Conceived by Those Who Are Not Professional Philosophers. Oslo: Jacob
Dybwad.
Nahmias, E., Morris, S., Nadelhoffer, T., and Turner, J. (2006). “Is Incompatibilism Intuitive?”
Philosophy and Phenomenological Research 73, pp. 28–53.
Rosenthal, R. (1978). Experimenter Effects in Behavioral Research. New York: Irvington.
Sackett, D. L. (1979). “Bias in Analytic Research,” Journal of Chronic Diseases 32, pp. 51–63.
Shope, R. K. (1983). The Analysis of Knowledge. Princeton, NJ: Princeton University Press.
132, pp. 99–107.
Swain, S., Alexander, J., and Weinberg, J. M. (2008). “The Instability of Philosophical
76, pp. 138–55.
Tarski, A. (1944). “The Semantic Conception of Truth: and the Foundations of Semantics,”
Weinberg, J., Nichols, S., and Stich, S. (2001). “Normativity and Epistemic Intuitions,”
Philosophical Topics 29, pp. 429–60.
8
Moral Intuitionism, Experiments,
and Skeptical Arguments
Mark van Roojen†
8.1. Introduction
Much argument and investigation in normative ethics seems to rely on reasoning
that deploys intuitions—somewhat immediate judgments—about claims that do not
depend for their force on antecedent commitments. Judith Jarvis Thomson (1985,
p. 1409), in the course of investigating the principles that govern permissible killing,
asked herself whether it would be permissible to push one person in front of a moving
trolley to save a larger number of people further down the track. Her judgment was
that shoving a fat man off a bridge in service of even a life-saving goal was not permis-
sible. And this judgment then constrained her theorizing about appropriate principles.
Bernard Williams (Williams and Smart 1973, p. 97), in the course of arguing against
utilitarianism in particular and consequentialism generally asks himself and his read-
ers whether George, an imagined unemployed chemist with scruples about weapons,
should take a job researching biological and chemical weapons when doing so would
prevent another more enthusiastic applicant from taking the job and designing more
lethal bombs. Williams expects his audience to conclude with him that George should
not take the job and uses that response to argue against consequentialism on the
grounds that it cannot support this verdict. In both of these well-known arguments,
the judgments about the particular hypothetical examples do not seem to be derived
† Sincere thanks to Al Casullo, John Gibbons, Elizabeth Harman, David Henderson, Walter
Sinnott-Armstrong, Michael Tooley, Brian Weatherson, and the students in my undergraduate seminar of
spring 2009 for discussion of the ideas and literature in this chapter. Many thanks also to Selim Berker,
David Chavez, Joe Mendola, and John Turri, all of whom gave me very helpful comments after reading an
earlier draft. Thanks also to Amanda Marshall, my commentator on a previous version of this chapter at the
Rocky Mountain Ethics Workshop in summer 2009, to members of the audience on that occasion, and to an
audience at Lewis and Clark College in April 2011. Furthermore, two kind anonymous reviewers for Oxford
University Press made several helpful suggestions and caught several instances of sloppiness on my part.
I thank them as well.
Moral Intuitionism, Experiments, and Skeptical Arguments 149
from principles already accepted by all those who agree with Thomson or Williams
about the examples. Rather, the responses of those who think about the examples for
the first time seem to be spontaneous in some way, and not obviously dependent on
prior philosophical commitments for their epistemic status. In fact, many people with
opposing theoretical commitments also grasp the force of the examples in question.
Philosophers often identify such spontaneous responses as “intuitions,” and in fact
self-identified moral intuitionists have made such responses one of the building blocks
in an overall foundationalist moral epistemology. Robert Audi, Roger Crisp, Mike
Huemer, Russ Shafer-Landau, and Philip Stratton-Lake have all defended versions of
epistemological intuitionism as an appropriate model for justification in normative eth-
ics.1 Though their proposals differ in important details, they have all claimed that intui-
tion can provide rather strong warrant for the normative judgments that are its ethically
relevant upshot. In particular they all seem to claim that intuitive judgments provide
justification sufficient for belief and knowledge absent defeaters. They also commit
themselves to the claim that the relevant intuitive judgments are non-inferential, what-
ever that comes to. Given this status, judgments yielded by intuition would be fit to per-
form two roles in moral epistemology, roles that are to some extent highlighted by the
examples with which we started. First, they can serve as foundational judgments in need
of no further justification on the basis of which we can go on to believe further things
about morality. This sort of foundational role requires judgements whose non-derivative
epistemic status allows them to serve as regress stoppers. Intuitionists think that the
non-inferential yet justified nature of the intuitions make them especially fit for that task.
Second, intuitive judgments can help us to decide between different equally coherent
normative theories on the basis of their fit with justified intuitive judgments. Since their
justificatory status depends on something other than the way in which they are related
to other things, they can retain that status even when they come into conflict with coher-
ent normative theories. Thus they can be used to test such theories and decide between
otherwise equally attractive competitors.
Even while this sort of intuitionism is gaining popularity among moral episte-
mologists, some experimental philosophers and philosophers using experimental
results have attacked philosophical reliance on intuitions as unreliable and hence
unwarranted. Often these sorts of criticism are directed at non-consequentialists by
consequentialists. That should be no surprise given the examples I gave as illustra-
tions. But these criticisms have also been urged by skeptics about morality. Walter
Sinnott-Armstrong has been one of the leading proponents of such general skepticism
and his arguments are perhaps the most worked out of the empirically based challenges
1 Not all of these theorists think that the primary locus of intuitive justification is in particular judgments
about hypothetical cases of the sort I use as examples. Audi, for example, focuses primarily on prima facie
principles identifying relevant moral properties. They do all seem to be classifiable (in Audi’s terminology)
as moderate foundationalists—that is, foundationalists whose claims are more moderate than those who
require certainty or indefeasibility with respect to base-level claims. I’ll use that label to refer to this sort of
view at various points in the text.
150 Mark van Roojen
to intuitionism. His strategy effectively uses varying empirical results to argue that the
putative outputs of intuition don’t have the reliability they would need to have to justify
belief in their contents.
This chapter is an attempt to figure out what an intuitionist should say in response
to these challenges. I am going to argue that at least some of the results cited by
Sinnott-Armstrong and others should shake intuitionist confidence that intuition
alone can provide justification sufficient for justified belief and knowledge. But I will
also argue that intuitively-generated judgments can play the needed regress-stopping
and theory-testing roles that intuitionists need them to play without providing justi-
fication sufficient for belief.2 Furthermore, I will argue that candidate intuitive judg-
ments need not be entirely non-inferential to play the two epistemic roles.
8.2. Intuitions as Evidence Favoring One Coherent

Theory Over Another
The dialectic rehearsed at the beginning of this chapter illustrates the role intuitions
play in deciding between conflicting but internally coherent normative theories.
Much moral philosophy attempts to provide an account of our obligations or reasons
for action, either in a comprehensive way or in some more limited domain. For just
about each such enterprise moral philosophers have come up with equally coherent
but incompatible accounts. Sometimes these accounts conflict with antecedent com-
mitments and we choose between them on that basis.3 At other times parties to the
debate propose that we consider scenarios involving choices for which the competing
coherent theories provide different advice. Parties aim to find examples for which one
option has considerable intuitive appeal apart from any fit with either of the contend-
ing theories. If judgments about enough such examples of sufficiently divergent sorts
fit better with one theory than another, this fact can be used to support one theoretical
position over another. Furthermore, some of us find certain more abstract principles
intrinsically more plausible than others, again apart from any support from one of the
contending coherent theories. And we can use these judgments to choose between
theories as well. In fact it is hard to see how much extant normative theory would exist
without recourse to both of these sorts of judgments.
Epistemological intuitionists in ethics4 aim to offer a story about either or both of
these sorts of judgments, a story which makes sense of their use for choosing between
2 What I mean by justification sufficient for belief is justification that would make belief a rational
response to the source of that justification. Or, what I tentatively take to be equivalent, I have in mind justifi-
cation which would render belief based on the relevant sources knowledge, absent falsity or defeaters.
3 The genesis of these commitments is often unclear, so that, appeal to these commitments is neutral
between a purely coherentist model of justification and a more foundationalist model. That’s why I focus on
newly formed judgments.
4 Within the ethics literature the term ‘intuitionism’ can designate either an epistemic view or a norma-
tive position of the sort that Ross defended where various prima facie principles are traded off against one
another. Here I use it in the epistemic sense.
theories. One requirement is that the epistemic status of the relevant judgments
indeed be independent of the theories they are used to decide between. Otherwise
the particular judgments in question would offer no additional reason to choose
one theory over another, beyond the fact that the first is an alternative theory to the
second. For this reason many intuitionists have stipulated that intuitions must be
“non-inferential.” The thought is that absence of inferential support secures inde-
pendence from other judgments including the theories between which we seek
grounds to choose.
8.3. Intuitions as Regress Stoppers

Consider the following argument:
Regress
(R1) If any person S is ever justified in believing a normative claim that p then S
must be able to infer p from other beliefs of S.
(R2) Any inference must have either (a) no normative premises or (b) some nor-
mative premises.
(R3) No person is ever justified in believing a normative claim that p by an infer-
ence without normative premises.
(R4) No person S is ever justified in believing a normative claim that p by an infer-
ence with a normative premise unless S is also justified in believing the nor-
mative premise itself.
(R5) No person is ever justified in believing any normative claim p by a chain of
inferences that includes p as an essential premise.
(R6) No person is ever justified in believing any normative claim p by a chain of
inferences that go on infinitely.
(R7) No person S can ever be justified in believing a normative claim that p.
This sort of argument for skepticism, based on the rejection of justificatory regresses,
is familiar to most of us. The formulation here is a simplified version of an argument
from Sinnott-Armstrong’s Moral Skepticisms.5 It usefully highlights the attractions
of normative intuitionism as providing a way to avoid regresses without lapsing into
skepticism. Most current moral intuitionists deny the first premise of this argument
and thus avoid its conclusion. They think that some normative beliefs can be justified
sufficiently even though they cannot be validly inferred from other beliefs. And what
this idea comes to for most of its proponents is one of two things. Either it amounts to
5 The formulation is simplified from Sinnott-Armstrong’s and so this is not quite a quotation, though sev-
eral of the premises are verbatim transcriptions of his. See Sinnott-Armstrong (2006, pp. 74–7) for his some-
what longer version. I think that the simplification doesn’t affect either the main point of Sinnott-Armstrong’s
argument or the appropriateness of my suggested response to it, though it does make the overall argument
easier to follow.
152 Mark van Roojen
the claim that some normative beliefs are themselves self-evident insofar as under-
standing the content of these beliefs can be sufficient for justifying a thinker in believ-
ing them even in the absence of inferential support from other beliefs, or it amounts to
the claim that intuitions are a propositional attitude distinct from belief (for instance
“seemings”) but capable of conferring justification or warrant on belief in their con-
tents. The first option has the intuitions themselves play the role of warranted or justi-
fied beliefs that are not inferred from any other thing. The second has the beliefs which
are supported by intuition serve as the beliefs that are justified or warranted without
being inferred from any other belief.6
8.4. Undermining Intuition

These uses of intuition have come under attack both from moral skeptics and from
advocates of normative theories who wish to resist particular uses of putative intuition
to favor alternative theories. One main line of attack is empirical. Various results in
experimental psychology are deployed to undermine our confidence in the reliability
of any capacity to form justified beliefs on the basis of intuition. If intuition is not suf-
ficiently reliable to justify believing its outputs it won’t be able to underwrite sufficient
confidence in those outputs to stop regresses or decide between competing theories.
Once again, one of the main proponents is Sinnott-Armstrong:
Unreliable
(U1) If our moral intuitions are formed in circumstances where they are unreli-
able, and we ought to know this, then our moral intuitions are not justified
without inferential confirmation.
(U2) If moral intuitions are subject to framing effects then they are not reliable in
many circumstances.
(U3) Moral intuitions are subject to framing effects in many circumstances.
(U4) We ought to know this.
(U5) Therefore our moral intuitions in those circumstances are not justified with-
out inferential confirmation.7
Sinnott-Armstrong lends support to steps (3) and (4) by citing various psychological
experiments involving subjects’ responses to questions about what to do in various
hypothetical scenarios. Among other things they show that people’s answers can be
influenced by how a case is described to them. One line of argument uses Kahneman
and Tversky’s famous Asian disease experiment to suggest the hypothesis that people’s
6 Audi (1996) takes the first option, as do most of the rest on my list, though Huemer (2005, 2008) takes
the second.
7 Sinnott-Armstrong (2008b, p. 52). It is worth noting that this argument is really two different argu-
ments; one that employs unreliability on its own and a second argument which supplements that argument
with knowledge of the unreliability or some related internalism-friendly access constraint.
differential willingness to avoid killing as opposed to letting die could be explained by

framing effects.8 Another (Petrinovich and O’Neill 1996) cites studies which show that
people shift their responses to cases when the description of the very same case uses
saving language rather than language of resulting deaths. Insofar as the mere descrip-
tion of a case must be irrelevant to how one ought to act in that case, these cases show
that the subjects’ responses are influenced by framing. Sinnott-Armstrong, of course,
uses this argument to support skepticism about moral knowledge. But non-skeptics
who disagree with various theories might also take comfort, inasmuch as their propo-
nents rely on putative intuitions about cases or about more-or-less plausible abstract
principles.
I’m not going to deny that in these cases the responses were affected by framing.
Nor am I going to deny the conclusion of the last argument suitably understood.
Furthermore, I’m prepared to concede that Unreliable works when deployed against
most current versions of intuitionism. But, I am going to argue, this is because both
those intuitionists and Sinnott-Armstrong share an assumption which has artificially
restricted the options for intuitionism.9 Put simply, that assumption is that every
regress-stopper must itself have non-inferential justification sufficient for justified
belief. My thesis will be that intuitionists should avail themselves of the resources
that this assumption leads them to ignore so that they can resist challenges of the sort
erected by Sinnott-Armstrong and his allies.
8.5. An Unconsidered Option
Notice that the first premise of Regress admits only justification relations that
allow inference from another of the agent’s justified beliefs. And that suggests that
regress-stoppers must themselves be beliefs. Insofar as they are beliefs it then seems
also required that they themselves come with justification sufficient for belief. After
all, how would inference from some belief you are not justified in having generate jus-
tification? Several intuitionists have noted this point and reacted by identifying intui-
tion with a distinct cognitive state, such as a seeming. And, as I understand some of
them, they think that such seemings don’t themselves require justification though they
do confer justification or warrant on beliefs with the same content. Thus, this way of
denying the first premise of Regress relies on recognition that some regress-stopping
options are being ignored.
8 Sinnott-Armstrong continues in the tradition of Tamara Horowitz (1998) who used Kahneman and
Tversky’s (1979) prospect-theoretic explanation of choices to attack arguments from cases in support of a
doing/allowing distinction. I wrote a response to that attempt (van Roojen 1999). While I still agree with
most of what I wrote then, this chapter is an attempt to get closer to the core issue than I now think I did
there. For another response see Kamm 1998.
9 It might be most fair to Sinnott-Armstrong to note that he may just be working within the assumptions
of the intuitionists his argument target, rather than having a commitment to the claim himself, which I think
should be denied.
154 Mark van Roojen
Still, so long as it is part of the view that these states confer justification sufficient for
belief on beliefs with the same content, the resulting view can still be undermined by argu-
ments like Unreliable. Let’s grant for the moment that intuitions are a distinct kind of cog-
nitive state, one requiring no justification. Since they require no justification, arguments
showing that they are unreliable in their upshot don’t undermine their intrinsic epistemic
status. Even so, it is part of the theory that they can confer positive epistemic status to
beliefs that are related to them in the appropriate way—that is, to beliefs that have the same
content as they do. As long as the view goes on to hold that the positive status in question
must be sufficient to make those beliefs justified (at least in the absence of defeaters), unre-
liability in the intuitions themselves will carry over to undermine that claim.
However, there is another way to live with the empirical results supporting
Unreliable, while allowing intuitions to play a regress-stopping role. That is to
notice that a regress-stopper need not get all of its justification non-inferentially or
non-relationally. The level of justification provided by intuitions need not be on its
own sufficient to believe their contents without more support from other sources. All
it really has to do is to put them in a favorable enough position that relations of coher-
ence with other similar judgments can corroborate them all so that they all pass the
threshold of justification sufficient for belief together.
Though a view like this is not currently popular, and though I’ve not been able to
find anyone in the ethics literature who seems to hold it, it is not entirely unfamiliar.
Bertrand Russell at one time advanced such a view for epistemology in general. After
contrasting versions of foundationalism and coherentist accounts of epistemology and
rejecting a pure coherence view, he writes:
But in a modified form the coherence theory can be accepted. In this modified form it will say
that all, or nearly all, of what passes for knowledge is in a greater or lesser degree uncertain; that
if principles of inference are among the prima-facie materials of knowledge, then one piece of
prima-facie knowledge may be inferrible from another, and thus acquire more credibility than it
had on its own account. It may thus happen that a body of propositions, each of which has only
a moderate degree of credibility on its own account, may collectively have a very high degree of
credibility. But this argument depends on the possibility of varying degrees of intrinsic credibil-
ity, and is therefore not a pure coherence theory . . . (Russell 1948, p. 157)10
10 Laurence BonJour (1985) distinguishes moderate foundationalism—the view that foundational beliefs
must be prima facie though perhaps defeasibly justified—from weak foundationalism, roughly the view
I’m advocating here, before dismissing the latter. He credits weak foundationalism to Russell, followed by
Goodman (1952), Firth (1964), and Scheffler (1967, ch. 5). As I read them, only Russell clearly defends a spe-
cifically weak foundationalist view. The others note the view in the course of arguing for views of a sort that
include both moderate and weak foundationalism. A referee points me to Chisholm’s Theory of Knowledge,
especially as explicated in Richard Foley’s paper “Chisholm’s Epistemic Principles” (1997). His inclusion of
propositions which are “probable for you,” insofar as you would be more justified to believe them than not,
does roughly capture the strength of justification I mean to indicate, since believing a proposition could be
unwarranted even when belief is a better response to one’s evidence than disbelief in that proposition. A sec-
ond referee points me to the clearest contemporary defense of such views that I now know of, an impressive
paper by James Van Cleve (2011, pp. 337–80). A more detailed version of that paper was presented at the 2009
Rutgers Epistemology Conference.
My suggestion is that a view of this sort, which allows moral intuition to generate moral
propositions with a certain degree of credibility short of that sufficient to justify belief,
will enable intuitionists to stop regresses and to resist Sinnott-Armstrong’s argument.
This way of thinking about the role of regress-stoppers highlights a good way to
think about the relevance of reliability to epistemic investigation. The really important
thing is whether taking some putative source of information into account in coming to
conclusions makes the overall belief-forming practice of which it is a part more reliable
than it would be without it. A somewhat unreliable process can be part of a more reli-
able overall process, and the overall process might be less reliable if it lacked the only
somewhat reliable sub-part. Condorcet’s jury theorem gives us a simple model: voting
in a population where each member has only a slightly higher probability than a half
of drawing the right conclusion turns out to be more reliable than the decision of any
individual. That example is overly simple for a number of reasons, but it does show that
the suggestion that repeated use of an only somewhat reliable process can generate a
more reliable process. Of course, moral enquiry is more complicated than voting for or
against a proposition on a ballot. But the complication doesn’t affect the parallel point
in that domain. An only somewhat reliable intuitive judgment generating process can
be part of an overall process of reaching reflective equilibrium about a subject matter
that is more reliable as a result of incorporating it. What really needs to be shown to
undermine the practice of relying on intuition is that the overall package is unreliable
or that the resulting procedure would be more reliable if it jettisoned reliance on intui-
tions entirely.
As already noted, regress-stopping and using intuition to decide between equally
coherent theories are necessary to enable normative theorizing to generate any war-
ranted verdicts about normative matters. If there is no non-inferential or non-relational
form of justification, and if justification must come to an end somewhere, we’ll get no
successful justification. Similarly, without a way of picking certain plausible proposi-
tions as those a theory should accommodate as well as those that are possible, there will
be no choosing between equally coherent competing theories. On the assumption that
there are truths to be discovered about moral matters, a method which renders no ver-
dicts is less reliable at finding out the truth than a method which renders verdicts which
are true more often than not. There’s still much work to be done to vindicate such a
method, for it would involve not just talking about inputs but also coherence-invoking
principles for turning the somewhat privileged but not yet therefore fully justified prop-
ositions into propositions we can rationally come to believe. We haven’t canvassed such
principles here.11 Still, we have some reason to be optimistic that such an overall view could
succeed, insofar as Condorcet’s theorem provides a simple model.
11 Van Cleve’s (2011) paper nicely canvases various formal accounts of how this might go at a very high
level of abstraction.
156 Mark van Roojen
8.6. Paradigm Moral Intuitions Are Not Beliefs Nor

Are They Sufficiently Justified to Believe
Sinnott-Armstrong is not being unfair to his real-life opponents when he targets the strong
claim that intuition is sufficient to justify belief. Each of the most prominent current moral
intuitionists accepts the claim that is the main target of Unreliable. Robert Audi, Roger
Crisp, Mike Huemer, Russ Shafer-Landau, and Philip Stratton-Lake all defend views
according to which unaided intuition yields justification sufficient for knowledge.12 All
allow that the justification provided by intuition can be defeated, but absent such defeaters
intuition justifies beliefs, and when beliefs result, those resulting beliefs count as knowl-
edge. It is that claim which Sinnott-Armstrong effectively targets.
When one reads the experimental papers underlying much of Sinnott-Armstrong’s
argument, one might be surprised to discover that the judgments elicited by research-
ers often don’t fit philosophical accounts of intuition. For example, many of the canon-
ical moral intuitionists require that intuitive judgments be non-inferential; yet often
there is just no evidence in the experiments to indicate that the responses weren’t
arrived at through inference. Subjects are asked their opinions about hypothetical
cases, but that in itself doesn’t tell us whether there was inference involved in forming
these opinions. And it isn’t completely obvious what sort of evidence would show the
judgments to be non-inferentially generated. I’ll come back to this briefly at the end of
this chapter.
More importantly for my purposes at present, there is often little evidence to indi-
cate that the responses were indicators of belief in the appropriate content13 or that
the subjects took themselves to have justification for such judgments. In Kahneman
and Tversky’s Asian disease experiment, respondents were asked to choose between
options for what to do. And while it is plausible that their overall beliefs played a role
in their choices, there’s no special reason to think that they believed they were doing
the right thing as opposed to (for example) believing their choice was most likely to
be right.14 In the Petrinovich and O’Neill 1996 study using trolley cases (upon which
12 Audi’s view is very strong: “one does not have an intuition with that proposition as its content until one
believes it” (1996, p. 110). Crisp, in defending a view he attributes to Sidgwick, concurs (2002, p. 64, p. 72 n.)
Shafer-Landau equates intuitionism with the view that there are non-inferentially justified moral beliefs
(2008, p. 83). Stratton-Lake follows Audi in thinking that unsupplemented intuition generates knowledge
(2002b, p. 18). Huemer requires only that the justification be sufficient for belief absent defeating evidence
(2005, p. 105). In fairness to Huemer, I should note that his view is closest to the sort I advocate. In a recent
paper on related matters (2008, pp. 379–80) Huemer notes that it is possible for coherence relations to
ratchet-up the level of justification provided by intuition which start out with only weak justification, and he
notes that this sort of leveraging can get going even when the original level of justification is rather weak due
to the unreliability of the intuitive judgments with which we start.
13 The requirement that intuitions are beliefs is not universal among intuitionists. But it can be striking
that some authors tell us what intuitions are supposed to be by following one or another characterization
from the intuitionist literature and then include nothing in their experiments to assure that they are testing
for something of this sort.
14 Sinnott-Armstrong himself notes that views about the subject’s moral beliefs have to be inferred from
their choices in this experiment.
Sinnott-Armstrong heavily relied), respondents were offered six options for react-
ing to a claim ranging from strongly agree to strongly disagree (with no option for
don’t know). Given that format, there’s no reason to think that they were expressing
belief in the target claim when they answered with any of the non-extreme responses.
“Somewhat agree” does not mean believe.
In fact there seems to me to be positive reason to think that the responses weren’t
beliefs and some reason also for the subjects not to take themselves to have full justi-
fication for such beliefs. Reflection on our own responses to the sorts of hypotheticals
that inform moral theorizing should, I think, reinforce the conclusion. My own reac-
tions to many trolley cases are somewhat tentative; though now that I’ve had a chance
to reflect on them and put them together with other things I think I have a greater
degree of confidence in some of my judgments. At this point, however, they are no
longer free of support from other plausible hypotheses and their justification is not
purely on the basis of considering the cases in isolation.
Perhaps further experiments could be conducted to show that subjects in fact believe
the answers they give in the relevant experiments. My own suspicion is that they don’t,
that instead the subjects express credences that fall short of belief. But my overall argu-
ment doesn’t really depend on that point. I think that even were the subjects to believe
the intuitive judgments they support in these cases, we should still deny that intui-
tion has provided them justification sufficient for belief. I think this is so even before
we are confronted with the relevant undermining experiments. Still, I think, it would
somewhat strengthen my case if test subjects were in fact expressing credences that fall
short of full beliefs in the content of the judgments they express. I could then suggest
that their own judgments roughly reflect the support those judgements actually have.
The liability to framing-effects discovered by the researchers could just be explained as
a reflection of the fact that people’s confidence in these judgments is weak enough that
it can be easily shifted, even by, in the end, irrelevant factors. And this changeability
could in fact reflect an appropriate response to an inconclusive epistemic situation.
Fans of intuition might try to use the mismatch between descriptions of the target
judgments and the actual judgments surveyed as part of a push-back strategy. They
might wish to argue that the experiments do not actually test the reliability of intui-
tion, since they do not target attitudes that fit the intuitionists’ definitions. I think this
would be a mistaken response to the mismatch. The experimenters were not being
arbitrary when they chose to study attitude-formation of the sort that they consid-
ered. Well-chosen hypothetical choice scenarios play an important role in moral
argument and they seem to function in much the way regress-stoppers must.15 So the
15 Of course, judgments about the plausibility of principles might well be another place where intuition
plays a role, and also judgments about similarity of cases. Russell in fact seemed to regard judgments about
which principles of reasoning are plausible to be a paradigm case of the contents that carry some intrinsic
credibility. And I think that we only have the materials to use intuitions about particular cases in a process
of reaching reflective equilibrium if we have other judgments at more general levels that get the same sort of
initial credence from non-relational sources.
158 Mark van Roojen
experimenters were looking at the right phenomena if they wanted to study good can-
didates for intuitive justification in the moral realm. And this remains so even when
the most prominent intuitionists themselves characterize intuition in ways that would
rule out the judgments in question.
8.7. How this Helps

If I think the push-back is mistaken, why did I think it was worth pointing out the
mismatch between the intuitionists’ characterization of the target judgments and the
judgments under empirical study? Because the mismatch suggests that intuitionists
should rethink their characterizations of intuitive judgments. I’m arguing that many of
the paradigms for such judgments don’t, even independently of Sinnott-Armstrong’s
argument, fit with a central intuitionist claim—that intuition by itself provides them
with justification sufficient for belief. An intuitionism of the sort I propose doesn’t
need them to have any stronger initial intuitive justification. It can accept all of the
premises and even the conclusion of Unreliable.
8.8. Objections from Fans of Intuition and a Reply

I speculate that the absence of this sort of view from the debate is to be explained partly
sociologically. Recent epistemic views have been widely influenced by the suggestion
that evidence is knowledge16 and, at least on a standard interpretation, this would rule
out foundational justification that falls short of underwriting belief. But there is at least
one other sort of consideration that seems to count against the view. Versions of it have
been articulated by Wilfrid Sellars, Donald Davidson, John McDowell and others, and
a related idea finds a home in Huemer’s version of intuitionism.
Using a Sellarsian metaphor, the idea is that for intuition to provide evidence it must
generate inputs that are already within the space of reasons. This is taken to mean, on
the one hand that intuitions must involve propositional attitudes, and on the other that
the contents of those propositional attitudes must be of a sort to bear the appropriate
relations to the conclusions they are used to support. McDowell attributes the view to
Davidson in his treatment of experience:
[I]‌f we conceive of experience in terms of impacts on sensibility that occur outside the space of
concepts, we must not think we can appeal to experience to justify judgements or beliefs . . . . The
space of reasons does not extend further than the space of concepts . . .
But Davidson thinks experience can be nothing but an extra-conceptual impact on sensibility.
So he concludes that experience must be outside the space of reasons. According to Davidson,
experience is causally relevant to a subject’s beliefs and judgements, but it has no bearing on
their status as justified or warranted. Davidson says that, “nothing can count as a reason for
For instance, Williamson (2000).

16
holding a belief except another belief ” (p. 310), and he means in particular that experience can-
not count as a reason for holding a belief.17
It’s no surprise to find the objection articulated in a perceptual context since percep-
tion is often used analogically to explain intuition in a moral context.
Mike Huemer’s endorsement of direct realism as part of his intuitionism is one way
to avoid the position Davidson winds up in. Huemer suggests that it is a mistake to take
perception or intuition as providing evidence from which we infer conclusions about
the world.
[T]‌he flaw consists in a basic misunderstanding of the structure of a foundationalist theory of
knowledge. Intuitionism does not hold that from ‘I have an intuition that p’ one may infer ‘p’;
nor does the principle of Phenomenal Conservatism hold that ‘It seems to me that p’ is a reason
for ‘p’. Those would be claims about inferential justification. Phenomenal Conservatism and my
version of intuitionism are versions of foundationalism: they hold that we are justified in some
beliefs without the need for supporting evidence. The role of conditions (1), (3) and (5) [about
intuition, experience and memory] in the theory of justification is that of conditions under
which certain beliefs . . . rather than that of evidence supporting those beliefs. (Huemer 2005,
pp. 120–1)
And Huemer takes the direct realist maneuver—predicated on the thought that per-
ception or intuition provide their objects directly to be used in forming further conclu-
sions—to require these objects to be propositions which we are now fully justified in
believing—provided that they have been presented by perception or intuition. That
we perceive or intuit them is a precondition for their being justified, but it is not itself a
justification. And this avoids the sort of problem that Davidson only avoids by denying
any foundation at all, since the deliverances of perception and intuition are proposi-
tional and of the right sort to serve as justifications for further thoughts.
If the basic positive idea here is right18 then there is still both something right and
something wrong with the precise way it has been developed in the quoted passage.
We can get the benefits of transparency without going all the way to requiring that the
inputs be believed or that belief be warranted. It may well be right to think that the psy-
chological claim, ‘That it seems to me that p,’ is not evidence but rather a precondition
for the proposition that p to be evidence. And further, Huemer might be right to think
that what intuition provides is a proposition—again that p. But it is wrong to think we
can only get those two benefits at the cost of thinking that we must believe p for the
proposition that p to be an input to justification. At least as far as this argument goes,
17 McDowell (1994, p 14), quoting Davidson (1986).

18 I’m ambivalent about it. On the one hand, in normal cases, the representational content of perception
or of putative intuition seems very closely connected to the nature of their content. In general a perception
with a content that represents, for example, a car on the street is better evidence about what is going on in the
street than it is about what is going on elsewhere. And that is because that is what it is about. But, on the other
hand, sometimes a judgment can support conclusions about facts with entirely unrelated content such as
when we infer some further proposition with unrelated content because it is a constituent of the best expla-
nation of the original judgment. See Harman (1977, ch. 1).
160 Mark van Roojen
we need think no such thing. What matters to the ability of our intuitive judgments to
rationalize further judgments is for their content to have implications for the contents
of these further judgments. The problem, as Huemer noted, was that a psychological
claim about how things seem to me doesn’t carry any commitment incompatible with
things being some other way. So a judgment with that psychological content can’t put
rational pressure on these further judgments, at least absent bridge premises we may
not be in a position to accept. This argument thus constrains the content of the judg-
ments contributed by intuition. It does not, however, constrain our confidence in them
nor does it constrain the level of warrant for that confidence.
My suggestion, then, is to allow judgments with the relevant content into our think-
ing to justify beliefs even when the judgments are not full beliefs. We can characterize
such judgments as accepting p with a certain credence; in ordinary English we might
call it “finding p plausible.” Huemer’s argument doesn’t show that there is anything
wrong with this way of talking. If its root idea is correct, it does show we should not
take the epistemically relevant content of that attitude to be ‘p is plausible’. Rather we
should say the epistemically relevant content is ‘p’. But we can accept that without
requiring the relevant state of mind to be belief, and without requiring that we have
justification sufficient to believe that content. The main point is that someone who
justifiably finds p plausible stands in an epistemically interesting relationship to the
proposition p and that said proposition is the right kind of thing to be an epistemic
reason to conclude yet further things, at least if it is true. It is the content of the state,
and not our degree of confidence in that content, that is epistemically relevant to fur-
ther conclusions and which makes other propositions epistemically relevant to it.19 At
the same time, an intuitionist should say, it is standing in the right sort of relation to
that proposition—considering it in the right way for it to show its plausibility—which
allows it to play an appropriate epistemic role. This claim is analogous to Huemer’s
claim about the role of believing an intuitive judgment. The main difference is that the
present view allows the state to play its epistemic role with less by way of justification.20
8.9. Further Adjustments to Intuitionism

So far I’ve argued that a more moderate intuitionism than yet proposed by even mod-
erate intuitionists in the ethics literature is well placed to live with the kinds of results
Sinnott-Armstrong has effectively deployed against these intuitionists. And I have
argued that it also fits better than less moderate versions with the actual use of putative
19 Schroeder 2008 has some interesting discussion about how to think about this sort of thing, especially at
the end of that article.
20 For what it is worth, this section of the chapter is the one I’m least sure of, due partly to my ambivalence
about the underlying worry about the space of reasons and about the transparency response to that worry.
My confidence in the main point of this chapter is greater than my confidence that we need to respond to the
space-of-reasons objection.
intuitions by ethical theorists. I want now to suggest one further trimming of intui-
tionist theoretical commitments.
Intuitionists want intuition to play an autonomous role in moral theorizing. If the
content of a putative intuition is just the upshot of claims a theorist already accepts, it
won’t be much help in choosing between different consistent theories. Nor will intui-
tions be good candidates for regress-stoppers if they’re just the upshot of even tacit
reasoning from yet further judgments, already accepted. Ethical intuitionists have thus
often built these desiderata into their canonical descriptions of intuition by requiring
that intuitive judgments be “non-inferential.” This strikes me as problematic for a cou-
ple of reasons.
Stipulating that intuitions be non-inferential will fail to exclude some judgments
without the required independence while excluding others that have it. It is going to
be very hard to know in many cases that a judgment is non-inferential if we admit
that there may be unconscious, tacit, or even just unnoted inferences. A person’s back-
ground beliefs can affect even that person’s first, seemingly immediate, reaction to a
question about a situation. So judgment may seem entirely non-inferential and yet
depend on such unconscious or unnoticed inferences. If there is reason to worry about
the effects of background beliefs, the stipulation isn’t going to help us with the prob-
lem. While we can say that only the independent judgments count as intuitions, we
won’t be in a position to say which seemingly immediate judgments are intuitions and
which are merely putative intuitions. As long as we don’t know which are which, we are
going to have to treat them the same in actual reasoning.
At the same time, the inferential nature of a judgment doesn’t in and of itself con-
stitute a troublesome limit on the autonomy of the target judgment. Worrisome
dependence exists only when the conclusions of the judgment-forming process don’t
go beyond the commitments already built into the judgments we are trying to decide
between. And as far as regresses go, you can’t find your way out of an argumentative
circle by adding in premises generated by computing the deductive consequences of
the premises you already have. But then the worry only applies when the upshot of
the inferential processes depends deductively only on the premises already available.
A judgment that is the upshot of both new information and inference therefrom won’t
be disqualified since the new information provides independent input into the overall
process.
We can see this by considering an analogy. Empirical theorizing depends upon evi-
dence, and for certain kinds of claims, such experiential evidence is apt for stopping
argumentative regresses and also for helping us choose between theories on the basis
of fit with such evidence. While it may be controversial, many philosophers think that
the nature of experience is influenced by a subject’s background beliefs and training
(e.g. see Harman 1973, ch. 11). Whether that view is in fact correct or not, it is not an
objection to it that theory-laden experiential data cannot play the role that experience
is supposed to play in theory confirmation. It can play that role so long as experience
adds something to the stock of information with which we want the theory to fit.
162 Mark van Roojen
This suggests that the needed independence should be put in a more positive way.
It isn’t so much that we want to exclude inference as that we want to be sure to include
inputs that don’t depend solely on inference from prior beliefs. Intuition should pro-
vide judgments whose contents contain more information than whatever they get by
non-ampliative inference from information the subject already has. Such judgments
provide claims that would not be reachable from existing information by purely
deductive methods.
8.10. Wrapping Up
I’ve now argued that a more moderate intuitionism than yet proposed by even moder-
ate intuitionists in the ethics literature is well placed to live with the kinds of results
Sinnott-Armstrong has effectively deployed against these intuitionists. I have also
argued that this view fits better than less moderate versions of intuitionism with
the actual use of putative intuitions by ethical theorists. Finally, I’ve suggested that
intuitionists who insist that intuitive judgments be non-inferential should drop that
requirement. That’s all we need.
Accepting these claims will be harder for some ethical intuitionists than others.21
To the extent that a proposed epistemology of intuitive judgment relies on the claim
that the contents of intuitive judgments are self-evident, these proposals will be quite
revisionary. On the other hand, just citing the self-evidence of a claim as an explana-
tion of how we come to know it is structurally little different from citing the intrinsic
plausibility of a claim. And insofar as the current proposal requires that this status be
supplemented by coherence relations to other plausible judgments, it actually provides
more of an explanation of how belief in such judgments comes to be warranted.
Furthermore, intuitionists of the moderate foundationalist stripe are already com-
mitted by their own theories to supplementing intuitive justification in this way. They
accept that intuitive justification is defeasible. And it is plausible that our actual epis-
temic situation is one in which we have encountered defeaters sufficient to render
uncritical acceptance of our intuitive judgments unjustified by moderate foundation-
alist lights. As Russ Shafer-Landau (2003, p. 265) claims, persistent moral disagreement
may be a defeater for intuitive belief so that one will need to bolster this belief by find-
ing a place for it among other supportive beliefs. Recall also that Sinnott-Armstrong’s
Unreliable contained two grounds for worry about intuition—(1) that it is unreliable in
the way demonstrated by the various experiments he cites, and (2) that we now should
know this. Moderate intuitionists should think that knowledge of enough evidence
21 It looks to me like Huemer’s intuitionism is easily able to accommodate my proposed emendations.

As I noted above, he already realizes that positive epistemic status short of warrant for belief can be boot-
strapped by coherence into warrant for belief in particular cases. And he already has an account on which
intuitive judgments themselves are not beliefs. So he would only need to drop something from his official
(2005) account of intuition. For the rest of the folks cited, accepting my main points would require greater
modification of their stated views.
undermining the reliability of intuition will count as a defeater for belief in even
self-evident propositions. So, now that we’ve looked at the evidence, moderate intui-
tionists should think that we need to find further relational support for these beliefs to
restore them to their prior justified epistemic status.
The big difference, then, between extant moderate intuitionist views and the even
more moderate view I propose here is about the status of our judgments prior to being
faced with evidence we already have. And on that issue their view gives more hos-
tages to fortune than the one I suggest they should adopt. For that sort of justification
requires more by way of reliability than the revisionary yet more moderate proposal,
and we have some reason to worry that we won’t get it.
References
Audi, R. (1996). “Intuitionism, Pluralism and the Foundation of Ethics,” in W. Sinnott-Armstrong
and M. Timmons (eds), Moral Knowledge. Oxford: Oxford University Press, pp. 101–36.
BonJour, L. (1985). The Structure of Empirical Knowledge. Cambridge, MA: Harvard
University Press.
Chisholm, R. (1989). Theory of Knowledge, 3rd ed. Engelwood Cliffs, NJ: Prentice Hall.
Crisp, R. (2002). “Sidgwick and the Boundaries of Intuitionism,” in P. Stratton-Lake (ed.), Ethical
Intuitionism. Oxford: Clarendon Press, pp. 56–75.
Davidson, D. (1986). “A Coherence Theory of Truth and Knowledge,” in E. Lepore (ed.), Truth
and Interpretation: Perspectives on the Philosophy of Donald Davidson. Oxford: Blackwell, pp.
307–19.
Firth, R. (1964). “Coherence, Certainty and Epistemic Priority,” Journal of Philosophy 61, pp.
545–57.
Foley, R. (1997). “Chisholm’s Epistemic Principles,” in L. E. Hahn (ed.), The Philosophy of
Roderick Chisholm. Chicago: Open Court, pp. 241–64.
Goodman, N. (1952). “Sense and Certainty,” Philosophical Review 61, pp. 160–7.
Harman, G. (1973). Thought. Princeton, NJ: Princeton University Press.
——. (1977). Morality. Oxford: Oxford University Press.
Horgan, T., and Timmons, M. (2006). Metaethics After Moore. Oxford: Oxford University Press.
Horowitz, T. (1998). “Philosophical Intuitions and Psychological Theory,” Ethics 108, pp.
367–85.
Huemer, M. (2005). Ethical Intuitionism. New York: Palgrave MacMillan.
——. (2008). “Revisionary Intuitionism,” Social Philosophy and Policy 25, pp. 368–92.
Kahneman, D., and Tversky, A. (1979). “Prospect Theory: An Analysis of Decision Under Risk,”
Econometrica 47, pp. 263–91.
Kamm, F. (1998). “Moral Intuitions, Cognitive Psychology, and the Harming-Versus-Not-Aiding
Distinction,” Ethics 108, pp. 463–88.
McDowell, J. (1994). Mind and World. Cambridge, MA: Harvard University Press.
Petrinovich, L., and O’Neill, P. (1996). “Influence of Wording and Framing Effects on Moral
Intuitions,” Ethology and Sociobiology 17, pp. 145–71.
Russell, B. (1948). Human Knowledge: Its Scope and Limits. New York: Simon & Shuster.
Scheffler, I. (1967). Science and Subjectivity. Indianapolis: Bobbs-Merrill.
164 Mark van Roojen
Schroeder, M. (2008). “Having Reasons,” Philosophical Studies 139, pp. 57–71.

Shafer-Landau, R. (2003). Moral Realism. Oxford: Clarendon Press.
——. (2008). “Defending Ethical Intuitionism,” in W. Sinnott-Armstrong (ed.), Moral
Psychology, Volume 2. Cambridge, MA: MIT Press, pp. 83–96.
Sinnott-Armstrong, W. (2002). “Moral Relativity and Intuitionism,” Philosophical Issues
(Realism and Relativism) 12, pp. 305–28.
——. (2006). “Moral Intuitionism Meets Empirical Psychology,” in T. Horgan and M. Timmons
(eds), Metaethics After Moore. Oxford: Oxford University Press, pp. 339–66.
——. (ed.) (2008a). Moral Psychology, Volume 2. Cambridge, MA: MIT Press.
——. (2008b). “Framing Moral Intuitions,” in W. Sinnott-Armstrong (ed.), Moral Psychology,
Volume 2. Cambridge, MA: MIT Press, pp. 47–75.
——. (2008c). “How to Apply Generalities: Reply to Tolhurst and Shafer-Landau,” in W.
Sinnott-Armstrong (ed.), Moral Psychology, Volume 2. Cambridge, MA: MIT Press, pp.
97–106.
——., and Timmons, M. (1996). Moral Knowledge? Oxford: Oxford University Press.
Stratton-Lake, P. (ed.) (2002a). Ethical Intuitionism. Oxford: Clarendon Press.
——. (2002b). “Introduction,” in P. Stratton-Lake (ed.), Ethical Intuitionism. Oxford:
Clarendon Press.
Thomson, J. J. (1985). “The Trolley Problem,” Yale Law Journal 94, pp. 1395–415.
Tversky, A., and Kahneman, D. (1981). “The Framing of Decisions and the Psychology of Choice,”
Science 211, pp. 453–8.
Van Cleve, J. (2011). “Can Coherence Generate Warrant Ex Nihilo? Probability and the Logic of
Concurring Witnesses,” Philosophy and Phenomenological Research 82, pp. 337–80.
van Roojen, M. (1999). “Reflective Moral Equilibrium and Psychological Theory,” Ethics 109, pp.
846–57.
Williams, B., and Smart, J. J. C. (1973). Utilitarianism: For and Against. Cambridge: Cambridge
University Press.
9
Linguistic Intuitions in Context:
A Defense of Non-Skeptical Pure
Invariantism
John Turri†
He either knows for certain, or he heard from the right people. I will ask him. And
he will tell me.
—Batman1
9.1. Introduction
Epistemic invariantism is the view that the truth-conditions of knowledge ascrip-
tions don’t vary across contexts. Epistemic purism is the view that purely practical
factors can’t directly affect the strength of your epistemic position. The combination
of purism and invariantism, pure invariantism, is the received view in contemporary
epistemology. It has lately been criticized by contextualists, who deny invariantism,
and by impurists, who deny purism. A central charge against pure invariantism is
that it poorly accommodates linguistic intuitions about certain cases. In this chapter,
I develop a new response to this charge. I propose that pure invariantists can explain
the relevant linguistic intuitions on the grounds that they track the propriety of indi-
rect speech acts, in particular indirect requests and denials. In the process we learn
† For helpful conversations and feedback on this paper, I thank Anthony Booth, Ian MacDonald, Rachel
McKinnon, and Darrell Rowbottom. As always, special thanks go to Angelo Turri. This research was sup-
ported by the Social Sciences and Humanities Research Council of Canada, the National Endowment for
the Humanities, the British Academy, the Association of Commonwealth Universities, and an Ontario Early
Researcher Award.
1 Quote-o-rama, “Batman: As the Crow Flies,” available at <http://www.otd.com/~paul/Quote/batman.

html> [last accessed October 21, 2013].
166 John Turri
an important methodological lesson about how to effectively marshal linguistic intui-

tions in favor of—or in opposition to—a theory.
First I will explain pure invariantism and its rivals, contextualism and impurism
(section 9.2). Then I will present a central argument against pure invariantism (section
9.3). Next I will review some previous responses to the argument (section 9.4). Lastly
I will develop a new response (sections 9.5 and 9.6) and conclude with a methodologi-
cal lesson (section 9.7).
9.2. The Received View Explained

One of the liveliest philosophical debates over the past two decades concerns
the semantics of knowledge ascriptions. Competing views here divide into two
camps: contextualist and invariantist.
Contextualists claim that the truth-conditions of knowledge ascriptions, such as ‘I
know that P’ or ‘He doesn’t know that P’, vary with features of the speaker’s context.2
The hallmark of contextualism—“the result that contextualists insist on” (DeRose
2009, p. 61)—is that a speaker in one context could truly say ‘S knows that P’, while
simultaneously a speaker in a different context could refer to the same person at the
same time with respect to the same proposition, and truly say ‘S doesn’t know that P’, It
is possible that some such disagreements are merely verbal, contextualists say, because
the speaker’s context determines how strong an epistemic position S must be in with
respect to P in order for the speaker to speak truthfully by saying ‘S knows that P’.
Consequently, S might meet the laxer requirements selected by the one speaker’s con-
text but fail to meet the stricter requirements selected by the other speaker’s context,
and this is why they could both speak truthfully.
Invariantists deny the hallmark of contextualism. The truth-conditions of knowl-
edge ascriptions, according to invariantism, are insensitive to features of the speaker’s
context. Every context selects the same requirements for being in a strong enough
epistemic position for knowledge. No matter how different our respective contexts are,
if you say ‘S knows that P’, and I say ‘S doesn’t know that P’, and we’re simultaneously
referring to the same person by ‘S’ and the same proposition by ‘P’, then our disagree-
ment is not merely verbal. At least one of us speaks falsely.
A second debate has emerged over the past decade, which cuts across the contex-
tualism/invariantism debate. It concerns the nature of knowledge itself, in particular
how it relates to purely practical matters. Suppose that Naomi and David both have
a true belief that P, that they both base their belief on equally good and compelling
evidence, that they are equally reliable on the question at hand, that they are equally
alert to counterevidence, and so on. In short, on all relevant truth-related matters,
Naomi and David are equal. In virtue of this, let’s say that they’re in an equally strong
epistemic position relative to P. Given that they both have a true belief and are in an
2 A quick note on quotations: I use single quotes to mention or name expressions; I use double quotes as
scare quotes and for direct quotation.
A Defense of Non-Skeptical Pure Invariantism 167
equally strong epistemic position, could it nevertheless turn out that only one of them
knows P? Could it turn out that, say, Naomi knows P, but David doesn’t? Could a purely
non-epistemic (i.e. a “practical”) feature of David’s situation prevent him from know-
ing P? For example, if David has more at stake than Naomi does on whether P is true,
and holding all else equal, could that prevent David from knowing P? Purists answer
‘no’. Impurists answer ‘yes’.3 The hallmark of impurism is that “what makes a true belief
into knowledge is not entirely an epistemic matter” (Stanley 2005, p. 2).
Leading contextualists endorse purism.4 By contrast, a debate rages between pur-
ists and impurists in the invariantist camp. Impure invariantists claim that although
the truth-conditions of knowledge attributions are not sensitive to the speaker’s con-
text, they nevertheless are sensitive to practical features of the context of the subject
under evaluation (i.e. the S in ‘S knows that P’). Pure invariantists deny both that the
truth-conditions of knowledge attributions vary with the speaker’s context and that
they vary with changes in the practical features of the subject’s context.5
A third debate, as ancient as the previous two are recent, cuts across all the positions
discussed so far: the debate over skepticism. Do we know most, or at least many, of the
things we ordinarily take ourselves to know? Or to put the matter in a way contextual-
ists might find more congenial, do we usually, or at least often, speak truthfully when
we say people ‘know’ things? Skeptics answer ‘no’, non-skeptics answer ‘yes’. These are
not perfectly precise views, since ‘many’, ‘most’, ‘usually’, and ‘often’ are vague terms.
Skepticism comes in degrees.
We could end up with eight different positions, depending on how we settle the
three debates.6
• Non-skeptical pure invariantism (Williamson 2005; Turri 2010a)
• Skeptical pure invariantism (Unger 1975)
• Non-skeptical impure invariantism (Hawthorne 2004; Stanley 2005)
• Skeptical impure invariantism
• Non-skeptical pure contextualism (DeRose 2009; Cohen 2005)
• Skeptical pure contextualism
3 ‘Purism’ is Jeremy Fantl and Matt McGrath’s (2009) term. It is also called “intellectualism” (Stanley
2005; DeRose 2009). Impurism is also called “practicalism” (Grimm 2011) and “pragmatic encroachment”
(Jon Kvanvig’s coinage).
4 Keith DeRose (2009, p. 188 n.4, p. 189) remarks, “One of the intuitive attractions of contextualism is
that it allows one to uphold intellectualism [i.e. purism] while delivering certain desired results about key
test cases,” and that his “allegiance to intellectualism” has motivated his sustained defense of contextualism.
See also Stewart Cohen (2005). However, Fantl and McGrath (2009, pp. 35–6) point out that there might be
hints of impurism in David Lewis’s (1996) defense of contextualism. Fantl and McGrath even express some
sympathy for impure contextualism (see also McGrath 2010, §6).
5 Timothy Williamson (2005) calls impure invariantists “sensitive invariantists” and pure invariantists
“insensitive invariantists.” DeRose (2009) calls pure invariantists “classical invariantists”; Jason Stanley
(2005) and John MacFarlane (2005) call them “strict invariantists”; Baron Reed (2010) calls them “stable
invariantists.”
6 Here I set aside consideration of assessment relativism about knowledge attributions. See MacFarlane
(2005).
168 John Turri
• Non-skeptical impure contextualism (Greco 2010)

• Skeptical impure contextualism
Some of these positions are unoccupied in the current literature and will be ignored
here. (It’s hard to even imagine any motivation for, say, skeptical impure contextual-
ism or skeptical impure invariantism.) Some important participants are sometimes
difficult to classify. For example, Jeremy Fantl and Matt McGrath are typically classed
with John Hawthorne and Jason Stanley as non-skeptical impure invariantists, but
Fantl and McGrath’s primary allegiance is to the less specific position of non-skeptical
impurism, and they express openness to non-skeptical impure contextualism (see
Fantl and McGrath 2009, p. 53; McGrath 2010, §6).
Impurism and contextualism are both “shifty” positions. They entail that either the
semantic content or truth-value of knowledge ascriptions can shift in ways that traditional
epistemology simply overlooks. By contrast, pure invariantism is a fully “stable” position.
It entails that the semantic content and truth-value of knowledge ascriptions neither do
nor can shift in the ways that impurists and contextualists claim that they can and often do.
Pure invariantism is orthodoxy. And since philosophers have traditionally been
(mostly) non-skeptical, non-skeptical pure invariantism has the distinction of being
the received view in contemporary epistemology. Impurists and contextualists are phil-
osophical rebels. Is their rebellion justified?7
9.3. The Received View Contested

Why rebel against the received view? Some argue for skepticism or express sympathy for
it (e.g. Unger 1975; Reed 2007; BonJour 2010). But there have always been skeptics and
they have always been a distinct minority. The leading contemporary challenge to ortho-
doxy is motivated by linguistic intuitions about verbal behavior. Many philosophers
argue that these intuitions provide strong evidence against pure invariantism. They
argue that either impurism or contextualism best explains the intuitive linguistic facts.8
Simple thought experiments elicit the relevant intuitions. In particular they are
elicited by descriptions of entirely realistic pairs of cases which seemingly reflect how
7 An overzealous and irreverent traditionalist might characterize the debate as “the embattled pure
hearted fending off the combined strength of the imps and cons.” A less colorful but more temperate and
dignified characterization of the debate is offered in the main text.
8 Impurists and contextualists have an in-house debate over which view better explains the data, once pure
invariantism has been eliminated. I won’t concern myself with that debate here. The important point for pre-
sent purposes is that impurists and contextualists agree that pure invariantism can’t do the job. (Assessment
relativists like MacFarlane (2005) argue that their view is even better than impurism and contextualism.)
I should note that Stanley (2005), whom I’m lumping in with the generic opponents of orthodoxy, has a more
nuanced and slightly ambivalent take on the role intuitions play in the dialectic. “These intuitions are not
intended simply to be data for an epistemological theory, as the grammaticality of various sentences may be
taken to be data for a syntactic theory. Rather, the role of my appeal to our intuitions about these particular
cases is to make vivid our commitment to the conceptual connection between knowledge and practical rea-
soning” (Stanley 2005, pp. 97–8; compare p. 12). Earl Conee (2013, p. 76) also expresses some ambivalences
about whether the data are best described as “intuitions” or “intuitive responses” based on something else.
competent speakers use ‘know’. One case, call it ‘LOW’, features a protagonist (‘Low
Pro’) who sincerely utters ‘I know P’ in an ordinary “low-stakes” setting. The key intui-
tion here is that Low Pro’s verbal behavior is natural and appropriate. The other case,
call it ‘HIGH’, features a protagonist (‘High Pro’) who sincerely utters ‘I don’t know P’
in a “high-stakes setting.” The key intuition here is that High Pro’s verbal behavior is
natural and appropriate. Importantly, the cases are set up so that Low Pro and High
Pro are in equally strong epistemic positions with respect to P: they have the same
evidence, are equally reliable, are equally alert to counterevidence; are just as confident
that P is true. And they are otherwise similarly situated, with one exception: in HIGH
the stakes are considerably higher because much more rides on whether P is true. It is
perhaps a harmless oversimplification to say that the only difference between Low Pro
and High Pro is that High Pro has more to worry about.9
Consider a concrete pair of such cases.10
LOW FLIGHT: Stewart is in the Atlanta airport, waiting to board his flight. A fellow
traveler seated nearby looks up from his laptop, stretches, turns to Stewart and says,
“I’ve been traveling all day and it’ll be a relief to get home to Detroit. A layover would
be annoying. Say, do you happen to know whether this is a direct flight to Detroit?”
With his itinerary in hand, Stewart answers, “Yes, I do—it’s direct to Detroit.”
HIGH FLIGHT: Stewart is in the Atlanta airport, waiting to board his flight.
Suddenly a man dressed in a uniform and carrying a small hardshell cooler rushes
down the concourse, stops in front of Stewart’s gate and breathlessly says to Stewart,
“I’m an organ courier transporting a kidney to a patient in Detroit. I need a direct
flight to Detroit, or the kidney will spoil. Do you know whether this is a direct
flight to Detroit?” With his itinerary in hand, Stewart answers, “Sorry, I don’t know
[whether it is]. You should check with an airline official.”
In each case, based on the itinerary Stewart believes throughout that the flight is direct
to Detroit, and his belief is true.
The cases reflect how we ordinarily speak.11 Speakers are disposed to be more gen-
erous in their ‘knowledge’ attributions in LOW cases, more sparing in their ‘knowl-
edge’ attributions in HIGH cases, and likely to deny ‘knowledge’ in HIGH cases. This is
9 Jonathan Schaffer (2006) has a very different take on which data best support contextualism, focusing
instead on the “contrast” and “question sensitivity” of knowledge ascriptions.
10 The pair is inspired by Cohen’s (1999, p. 58) widely discussed airport case, but whereas Cohen’s case
involves both elevated stakes and the explicit mention of a specific error possibility, my cases don’t involve
the mention of specific error possibilities. This will be important below when we consider Patrick Rysiew
and Jessica Brown’s defense of pure invariantism.
11 Or so it seems to theorists reflecting on the matter from the armchair. As with other aspects of actual
patterns and tendencies in ordinary usage of ‘know’, the matter is ripe for empirical investigation. As much
recent work in experimental philosophy and psychology has suggested, sometimes we’re surprised by
what we find when we carefully look and see (e.g. see Beebe and Buckwalter 2010; Beebe and Jensen 2012;
Beebe 2013; Weinberg et al. 2001; Swain et al. 2008; Feltz and Zarpentine 2010; Starmans and Friedman
2012; Schaffer and Knobe 2012; Sripada and Stanley 2012; Pinillos 2012; Myers-Schulz and Schwitzgebel 2013;
Murray et al. 2013; Turri 2012, 2013, Under review; and Turri and Friedman Forthcoming).
170 John Turri
all “utterly natural” and “integral to the ordinary use of ‘know’ ” (Williamson 2005, p. 217;
compare Stine 1976, p. 274).
But why should this undermine pure invariantism? How do we go from the propri-
ety and naturalness of such behavior to the denial of pure invariantism? According to
Timothy Williamson:
Presumably, the endorsement rests on a methodological principle of charity, by which, very
roughly, we should prefer to interpret speakers as speaking . . . truly rather than falsely (ceteris
paribus). Shifting standards seems to give us more flexibility to assign to ‘know’ a charitable
reference. (Williamson 2005, p. 220; compare Fantl and McGrath 2009, ch. 2, and Davis 2007,
p. 430)12
Keith DeRose’s most recent work (2009, ch. 2) confirms this by clarifying “the method-
ology that takes us from the data to a contextualist conclusion,” or more generally, from
the data to the denial of pure invariantism. The data are the “intuitive” facts that the
cases feature sincere, natural and appropriate speech, along with the observation that
the speech isn’t based on mistaken beliefs about the situation. DeRose calls this “the
best possible type of evidence” we could have against invariantism in epistemology,
and exactly the same type of evidence that leads us to reject, say, invariantism about
indexicals or gradable adjectives. DeRose’s key methodological claim is that there is a
“general presumption” that when competent speakers “are not basing their claims on
some false beliefs they have about underlying matters of fact, how they naturally and
appropriately describe a situation, especially by means of common words, will be a
true description.” Consequently, DeRose concludes, it is “a bad strike against” a theory
“if it rules [that such descriptions are] false, as it seems invariantism will have to rule
with respect to one or the other of ” High Pro and Low Pro (2009, pp. 50–1, p. 67).
In short, if we combine (a) the fact that sincere, natural, appropriate,
non-misinformed descriptive uses of ‘know’ in ordinary language are guided by
noticeably different standards in different contexts, and (b) the charitable meth-
odological principle that such descriptive uses are true, then we have evidence that
“militate[s]‌strongly” against pure invariantism.13 Pure invariantism is too inflexible
to charitably explain all the data. It predicts that when knowledge ascriptions in dif-
ferent contexts superficially contradict one another, at least one of them is false. And it
predicts that knowledge doesn’t come and go depending on how much is at stake. But a
charitable explanation of our linguistic behavior leads us to reject at least one of those
predictions. Our linguistic behavior is shifty. Pure invariantism can’t shift gears.
12 Note that Williamson is characterizing the line of thought, not endorsing it. He defends pure
invariantism.
13 DeRose also reports intuiting directly that the utterances in question are true (e.g. 2009, p. 49 n.2). But
he doesn’t simply rest with this. He acknowledges that the intuition that a speaker’s utterance is “appropriate”
is stronger than the intuition that it’s true (2009, p. 50). DeRose then proceeds to argue that appropriateness
is powerful evidence for the truth, as I describe in the main text. This is a good approach because it begins
with less controversial data—intuitions about the generic propriety of speech—upon which all parties to the
debate are more likely to agree.
To help organize the discussion, I’ll understand the argument against pure invar-
iantism as follows, where ‘proper’ abbreviates ‘sincere, natural, appropriate and not
based on misinformation’.
(Anti-PI)
1. Low Pro’s and High Pro’s utterances are proper. (Premise)
2. If their utterances are proper, then both of their utterances are true.
(Premise: Charity Principle)14
3. So both of their utterances are true. (From 1 and 2)
4. If both of their utterances are true, then pure invariantism is false. (Premise)
5. So pure invariantism is false. (From 3 and 4)
9.4. The Received View Defended

How should pure invariantists respond? The argument is valid, so they must reject at
least one premise. This section reviews several previous responses. The next two sec-
tions develop a new response.
Here is one strategy for rejecting line 4.15 Knowing P requires confidently enough
believing P.16 But high stakes induce people to be epistemically cautious, perhaps
excessively so. We need more evidence to maintain the same level of confidence.
Recall that High Pro and Low Pro have the same evidence. The high stakes erode High
Pro’s confidence to the point where he no longer knows P. So High Pro speaks truth-
fully when he denies that he knows. But this is because he fails to satisfy the invariant
requirements of knowledge, not because the meaning of ‘knows’ or the requirements
of knowledge are shifty.
In response, critics deny that High Pro is best understood as losing confidence in
the relevant sense. At least, the case doesn’t have to be interpreted that way. There is
another way of understanding it that suffices for the critic’s purpose. DeRose (2009,
pp. 190–3) argues that this objection to line 4 presupposes an “unstable” conception of
confidence, whereas a “stable” conception is “more natural” and “correct.” On the sta-
ble conception, high-stakes situations don’t typically diminish our confidence. Rather,
the same level of confidence is present in both LOW and HIGH. This level of confi-
dence disposes us to act one way in LOW but more cautiously in HIGH.17
14 Alternatively, line 2 might have said, “The best explanation for 1 is that both of their utterances are true.”
Pure invariantist arguments against line 2 have indeed focused on finding a satisfying alternative explana-
tion of the utterances’ propriety.
15 Inspired by Kent Bach (2010), though I’m not sure he would endorse my way of putting it. The proposal
is modeled on his, but he has a different sort of case (a “third-person” case) in mind when he makes his pro-
posal. Compare also Hawthorne’s “belief removal model” (2004, p. 169).
16 It has recently been disputed whether belief is genuinely required for knowledge. For example,
Myers-Shulz and Schwitzgebel (2013) and Murray et al. (2013) present empirical evidence that, on the ordi-
nary concept of knowledge, knowledge doesn’t require belief. For an opposing view, see Rose and Schaffer
(2013), and Buckwalter et al. (2013).
17 See also Brown (2005, pp. 147–8).
172 John Turri
Another response is to reject line 2 on the grounds that Low Pro’s utterance is false
because he fails to meet the demanding standards of knowledge. Low Pro’s utterance
nevertheless seems proper because it’s close enough to the truth for practical purposes
(Unger 1975). One way of sharpening this proposal is to invoke the phenomenon of
loose talk (compare Conee 2005 and Davis 2007). It would be laborious to always be
perfectly precise. It’s convenient to speak loosely when present purposes don’t require
precision. Some contexts tolerate very liberal approximations of the truth whereas
others require great precision. Our variable strictness in ascribing knowledge follows
the same pattern, one might argue, which explains the shiftiness in our use of ‘knows’
while maintaining that the truth-conditions of knowledge ascriptions are invariant.
The loose-talk strategy faces two challenges. First, it seems too hospitable to skepti-
cism. Indeed, Unger deployed it as part of his overall case for a radical form of skepti-
cism, to help explain why we falsely say that we know many things. As Conee (2005,
p. 52) puts it, on this view “only the most conspicuous facts of current perception,
the clearest memories, triple-checked calculations, and the like will” enable knowl-
edge. Other things being equal, I prefer a less skeptical defense of pure invariantism.18
Second, knowledge-talk doesn’t fit the profile of loose talk (MacFarlane 2005, p. 784).
You say, “I’m going to the store. Do we have any coffee left?” and I respond, “No, it’s all
gone.” If you challenge me, “Actually, there are a few grounds left in the jar,” a natural
comeback on my part would be, “[Of course] I meant it was just about all gone [and
so you should buy some more].” By contrast, in typical cases where someone is chal-
lenged, “Actually, you don’t know that,” they don’t say, “the point is that I just about
know it,” or, “[Of course] I meant that I nearly knew it.”
Patrick Rysiew (2001, 2007) rejects line 2 for different reasons. His proposal is prem-
ised on the uncontroversial observation that an utterance communicates more than its
literal content. We presuppose that our conversational partners are cooperative and,
consequently, that they strive to make their speech relevant. This presumption enables
us to communicate information beyond what we literally say, a phenomenon which I’ll
call suggestion or conveyance. Knowledge attributions are no exception: they can also
suggest or convey information. On Rysiew’s view, High Pro speaks properly but falsely
when he says that he doesn’t know. High Pro says that he doesn’t know in order to avoid
suggesting certain other false things. And in the context it seems more important to
avoid the false suggestion than to speak the literal truth. This is why we intuit that his
utterance, although false, is proper.
The details of Rysiew’s influential proposal are important so let’s consider it more
carefully. Knowing P requires being in a good enough epistemic position relative to
P. For convenience Rysiew understands being in a good enough epistemic position
as being able to rule out the relevant alternatives, on some moderately strong but
18 Compare Stanley’s (2005, p. 84) remarks on a contextualist version of the loose-talk strategy: “This is not
a very satisfying way of ‘rescuing’ ordinary knowledge attributions. Indeed, one may wonder whether it has
any advantages over skepticism at all.”
non-skeptical and invariant understanding of ‘relevant alternatives’.19 We’re often able

to rule out all the relevant alternatives but we’re almost never able to rule out all alter-
natives. Let ‘C’ name one of these irrelevant alternatives that, on a specific occasion,
you can’t rule out. Despite being irrelevant, C might still be conversationally salient
(‘salient’ for short). And in a context where C is salient, saying ‘I know P’ suggests that
you can rule out C. It suggests this because C is salient, you’re presumed to be coop-
erative, and so it’s expected that you wouldn’t say ‘I know P’ unless you (wanted to
communicate that you) can rule out C. So in order to avoid falsely suggesting that you
can rule out C, you say ‘I don’t know P’. And since conversational salience is highly
context-sensitive, Rysiew argues, this pragmatic account can explain the shiftiness of
our knowledge attributions.
Rysiew’s proposal faces a challenge. It’s not generally true that when asserting R sug-
gests S, asserting not-R similarly suggests not-S. For example, if a colleague asks me
my opinion of a certain restaurant and I respond, “The best I can say about that place
is that the staff is polite,” this suggests that their food isn’t good. But if I instead say,
“The best I can say about that place isn’t that the staff is polite,” this does not suggest
that their food is good. To take another example, if my partner asks whether we’ll be
on time for the reception, and I say, “It’s possible,” this suggests that I am not confident
that we will be on time. But if I instead say, “It’s not possible,” this does not suggest that
I am confident that we will be on time—far from it!
In light of these points, reconsider Rysiew’s proposal. Suppose that saying ‘I know P’
suggests that you can rule out all the salient alternatives, even the epistemically irrel-
evant ones you can’t rule out, such as C. Now when asked whether you know P, one way
to avoid suggesting that you can rule out C is to not say anything at all. But that would
be rude. A politer way is to hedge with ‘Maybe I know, but C sure is hard to rule out’ or
‘I’m tempted to say “I know,” but then again, there is the possibility of C’. Those hedges
have the advantage of being true. Why opt for the false ‘I don’t know’ when you could
avoid suggesting the falsehood by expressing a readily available truth? This defies our
expectations of competent, cooperative speakers. The upshot is that although Rysiew’s
view predicts that High Pro won’t say ‘I know P’, it doesn’t explain why he goes so far as
to say ‘I don’t know P’.20
But does it really defy our expectations? As Brown (2006, p. 415ff) notes, it’s not
uncommon for speakers to say something literally false in order to convey a relevant
truth. Such speech often seems proper, as in the following example. It’s lunchtime.
You’re hungry and would like some company over a meal. “Have you had lunch?”
you ask me. “No, I haven’t had lunch,” I respond. Arguably what I say is literally false
because I have eaten lunch at some time in the past; but it conveys a relevant truth,
namely, that I haven’t eaten lunch today. I could easily have truthfully said ‘I haven’t
19 Rysiew’s approach isn’t wedded to relevant alternatives theory. This will be important below.
20 Compare Fantl and McGrath (2009, p. 41). Why achieve your conversational purpose “by lying” they
ask, when it could be achieved “just as well with the truth?” See also DeRose (2009, pp. 111–24).
174 John Turri
eaten lunch today’ and had the same effect, yet it still seems appropriate for me to pre-
fer the shorter false utterance instead. Presumably it’s appropriate because you aren’t
liable to be misled by my literal words into thinking that I had somehow, amazingly,
lived over thirty years without eating lunch even once; rather, you can be counted on
to infer that I intended to communicate that I hadn’t eaten lunch today. On similar
grounds, Rysiew and Brown might argue, it’s appropriate for High Pro to prefer ‘I don’t
know’ to true, but less convenient formulations such as ‘Maybe I know, but C sure is
hard to rule out’.
Brown is clearly right that there is precedent for this sort of explanation. But one
worry is that the precedent differs importantly from the High cases that it’s invoked to
help explain. For example, in HIGH FLIGHT the organ courier can’t be counted on to
avoid being misled by Stewart’s literal words. The presumption is that the courier will
believe Stewart when Stewart says ‘I don’t know’. And the courier has no special reason
to think that Stewart is merely trying to avoid falsely suggesting that he can rule out
a salient but irrelevant alternative. So whereas I know that you won’t be misled when
I say ‘I haven’t eaten lunch’, Stewart doesn’t have any reason to think that the organ cou-
rier won’t be misled when he says ‘I don’t know’.
Rysiew and Brown’s view faces another challenge: some high cases lack an irrelevant
but salient alternative, which their view requires in order to explain the relevant behav-
ior. The organ courier mentions no error possibility to Stewart. No one mentions that
Stewart’s itinerary might contain a misprint, or that the captain might mistakenly land
the plane in Pittsburgh, or anything else. LOW FLIGHT and HIGH FLIGHT seem to
feature exactly the same set of alternatives: it’s a direct flight to Detroit versus it’s not
a direct flight to Detroit. It’s totally implausible to suggest that it’s not a direct flight to
Detroit is relevant in LOW FLIGHT but irrelevant in HIGH FLIGHT. And it’s equally
implausible to suggest that it’s not a direct flight to Detroit is salient in HIGH FLIGHT
but not in LOW FLIGHT. Thus Rysiew and Brown’s proposal can’t explain why Stewart
says ‘I don’t know’.
Rysiew and Brown might respond that heightened stakes naturally prompt us to
start worrying about additional alternatives, even if they’re not mentioned. For exam-
ple, they might say it’s natural that in HIGH FLIGHT Stewart’s thoughts will turn to
the possibility that his itinerary contains a misprint, or the possibility that the pilot
will make an unnecessary, unauthorized stop. The problem with this response is that
it doesn’t seem essential to the case that Stewart begins worrying about such possibili-
ties. Even if we stipulate that he’s cool under pressure and doesn’t begin to worry about
those possibilities, it seems neither unnatural nor inappropriate for him to deny that
he knows.
At this point it’s worth recalling the generic version of Rysiew’s proposal, stated in
terms of strength of epistemic position rather than relevant alternatives. It’s uncontro-
versial that knowledge requires a true belief plus a strong enough epistemic position.
How strong? We’re not giving an analysis of knowledge, so it’s harmless to answer
‘strong enough to know’. But strong enough to know doesn’t entail strong enough for
everything. Some purposes might require a position stronger than what knowledge
strictly requires. If that’s the case, then we can explain High Pro’s behavior even with-
out a conversationally salient but epistemically irrelevant alternative, as follows. In
both LOW FLIGHT and HIGH FLIGHT Stewart’s epistemic position relative to it’s a
direct flight to Detroit is strong enough to know and strong enough to satisfy a stran-
ger’s idle curiosity. But in neither case is it strong enough for directing personnel in
life-and-death medical matters. So Stewart knows in both cases and Stewart speaks
falsely in HIGH FLIGHT when he says ‘I don’t know’. Nevertheless, he also speaks
appropriately because he communicates that his epistemic position isn’t strong enough
for the courier’s conversationally salient purposes.
Understood this way, Rysiew and Brown’s view contrasts interestingly with the
loose-talk proposal discussed earlier. Just as it was argued earlier that Low Pro speaks
falsely but appropriately because present purposes recommend loose usage, here it is
argued that High Pro speaks falsely but appropriately because present purposes rec-
ommend, as it were, guarded usage.
Overall, although Rysiew and Brown’s defense of pure invariantism is impressive, it
faces difficult challenges. I’ll raise two further concerns about their view before pre-
senting my own proposal in the next section.
First, it’s noteworthy that competent speakers in High cases respond similarly
whether you ask them ‘P?’ or ‘Do you know whether P?’ (see Turri 2010b, 2011).
Consider this exceedingly minor revision of HIGH FLIGHT.
Stewart is in the Atlanta airport, waiting to board his flight. Suddenly a man dressed
in a uniform and carrying a small, hardshell cooler comes rushing down the con-
course, stops in front of Stewart’s gate, and breathlessly says to Stewart, “I’m an organ
courier transporting a kidney to a patient in Detroit. I need a direct flight to Detroit,
or the kidney will spoil. Is this a direct flight to Detroit?” With his itinerary in hand,
Stewart answers, “Sorry, I don’t know. You should check with an airline official.”
The only difference here is that the courier asks Stewart ‘Is this a direct flight to
Detroit?’ instead of ‘Do you know whether this is a direct flight to Detroit?’ The very
same answer, ‘I don’t know’, serves equally well and seems to have the same effect on
the conversation in both versions of the case. It might be too much to ask for an identi-
cal explanation of both versions, but we should expect the explanation to be similar for
both. It’s not clear that Rysiew and Brown’s view can offer this.
Second, Rysiew says that it’s “essential” to his proposal that “our untutored intuitions
about the truth conditions of various sentences are generally insensitive to the seman-
tic/pragmatic distinction” (2007, p. 648; cf. p. 660 n.31). That is, our intuitive estimation
of speech as true or false doesn’t distinguish the truth-value of what we literally say, on
the one hand, from the truth-value of what we suggest, on the other. Instead we “tend to
identify” the truth-value of an utterance with the truth of “the most salient proposition
a speaker actually communicates” in saying it (Rysiew 2001, p. 487). Moreover, Rysiew
extends this point to explain why some mistakenly believe that they don’t know certain
176 John Turri
things, as follows (2001, pp. 502–3). Saying ‘I know that I’m not a brain in a vat’ would
falsely suggest that I can rule out the irrelevant possibility that I am a brain in a vat.21 As a
result, I not only refrain from saying ‘I know I’m not a brain in a vat’, but I also mistakenly
“come to believe” that I don’t know that I’m not a brain in a vat.
Suppose that attributing such confusion to us is essential to Rysiew’s view.22 Then one
might question whether the anti-skeptical preference that Rysiew and I share is prop-
erly motivated. G. E. Moore (1959, p. 193ff) pitted the obviousness of commonsense
knowledge attributions against the skeptic’s wherewithal. And many non-skeptical
epistemologists follow him in that regard. But if it’s correct that we begin doing epis-
temology with long habits of mistaking false knowledge ascriptions for true ones,
and true knowledge ascriptions for false ones, then that poisons the well of Moorean
data, thereby weakening Moore’s hand—perhaps even to the point where he no longer
knows that he has one.
The objections and concerns I’ve raised might not debilitate Rysiew and Brown’s
view. I don’t claim that they are insurmountable. But they are enough to motivate me
to look for an alternative response to Anti-PI. Of course, alternative responses needn’t
compete with one another. They could be complementary.
9.5. Indirect Speech Acts

The responses to line 2 covered in the last section are widely labeled, in Keith DeRose’s
memorable phrase, “warranted assertability maneuvers” or “WAMs” for short (DeRose
1999, p. 196ff; 2009, p. 83ff). Brown helpfully encapsulates the essence of a WAM:
At the core of a WAM is the idea that the intuitions about contextualist cases [e.g. High/Low
pairs] can be explained by appeal to the truth-value of the propositions pragmatically conveyed
by knowledge attributions, rather than the literal truth-value of those attributions. (Brown 2005,
p. 150)
A WAM is one way to develop a pragmatic account of such cases. But it isn’t the only
way. In the remainder of this section, I’ll lay the groundwork for a pragmatic account
that doesn’t fit the WAM model, based on speech-act theory.23
Orders and requests are ways to direct people. More specifically, they are speech acts
we perform in order to direct people. Sometimes we direct them overtly by saying things
like ‘pass the salt’ or ‘give me the information’. We could even use an explicit performative,
as in ‘I request that you pass the salt’ and ‘I hereby command you to give me the informa-
tion’. But explicit performatives are awkward and overt direction is impolite, so we usually
make requests indirectly. Often we do this by asking questions, as with ‘can you pass the
salt?’, ‘could you give me the information?’, ‘would you mind not stepping on my foot?’,
21 This possibility is irrelevant, on Rysiew’s view, because nobody accepts it (2001, p. 499).
22 I’m not convinced that Rysiew is right when he says that this is essential to his view, but set that aside.
23 The remainder of this section is heavily indebted to John Searle’s (1979, ch. 2) discussion of indirect
speech acts.
and ‘do you want to join us for dinner?’ Another way to make an indirect request is by stat-
ing that we have certain preferences, as with ‘I would like (you to pass me) the salt’, ‘I hope
you’re able to give me the information’, and ‘I need you to get off my foot’.
That these are all ways of indirectly requesting is supported by the fact that we can
felicitously add ‘please’ to what we say, as in ‘please, can you pass the salt?’ and ‘would
you please give me the information?’. It is also supported by the fact that we naturally
respond to the questions as if they were requests. We hear them as requests. If at the
dinner table you ask, “Can you pass the salt?” I pass you the salt without further ado,
just as I would if you directly requested it in the imperative mood. Similarly, if you ask,
“Can you pass the salt?” and I respond, “No,” you don’t suddenly start worrying that
my arms are paralyzed or that I’m lying to you. You understand me to be denying your
request, not answering the question itself.
Making a request is your primary purpose in asking, “Can you pass the salt?” That
you’re also asking a question is incidental. It’s mutual knowledge that we both already
know that I can pass the salt, so you’re neither seeking information nor trying to lead
me down a path of self-discovery. Moreover, if I responded directly to your literal
question by saying, “Yes, I am indeed able to do that,” or, “Why do you want to know?”
it would be interpreted as either humorous (if said while passing the salt) or uncoop-
erative (if said despite not passing the salt).
Call the performance of an indirect speech act indirection. We can distinguish dif-
ferent types of indirection. Conventional indirection is accomplished by using idioms,
which usage has established as indirectional devices. Examples of conventional indi-
rection are ‘how about passing me the salt?’ as a way of requesting the salt, or ‘I’ll be
keeping an eye on you’ as a way of warning or putting someone on notice.
Conversational indirection is accomplished by exploiting features specific to the con-
versational context, along with general communicative principles and background
knowledge. Here is an example:
man: Let’s go to the movies tonight.
woman: I have a lot to prepare for a major court case scheduled early tomorrow
morning.
The man makes a direct proposal. Normally the woman’s response would count as a
denial of the proposal. But that’s not because ‘I have a lot to prepare for a major court
case scheduled early tomorrow morning’ is conventionally associated with denying
proposals. Rather, it’s because, in the context, it’s clear that the best way to make sense
of her assertion is that she wants to communicate that she will not be going to the mov-
ies tonight with him. To accept the proposal, all she had to say was ‘sure’, but instead she
chose to assert that she had a time-consuming task to complete, which would usually
prevent her from having enough time to go to the movies. She wouldn’t have said that
unless she was politely declining the proposal.
Not all cases of indirection fall neatly into either conventional or conversational.
Many seem to fall somewhere in between. Questions involving ‘can’, ‘would’, and
178 John Turri
‘could’ are unlike idioms, in that they retain a literal compositional meaning, they
admit of direct responses to their literal content, and their literal translation into other
languages can preserve their indirectional potential.24 Yet they are also unlike ‘I have
to prepare for a major court case’, in that their default status is to be heard as requests.
Stage-setting is required to hear ‘can you pass the salt?’ primarily as a question about
your abilities rather than as a request to pass the salt. By contrast, stage-setting is
required to hear ‘I have to prepare for a major court case’ primarily as a denial rather
than an assertion.
Closely related to the point about how readily we hear certain formulations as indi-
rect requests, conventional and conversational indirection also differ in whether ask-
ing for clarification is felicitous. It would be positively odd for a competent speaker to
ask for clarification upon being asked, “How about passing me the salt?” Sincerely ask-
ing in turn, “Just to clarify, are you asking me to pass you the salt?” would come across
as completely obtuse. By contrast, in cases of conversational indirection, asking for
clarification is typically felicitous. In the example above, it would not be out of order
for the man to follow up with, “So is that a ‘no’?” or, “We’re not going, then, right?” On
this dimension, typical indirect requests featuring ‘can’, ‘would’, and the like behave
more like conventional than conversational indirection. Normally if you ask, “Can you
change the channel?” and I sincerely ask in turn, “Are you asking me to change the
channel?” you’re likely to think me annoyingly dimwitted.
(If we want a label for the in-between cases of indirection involving ‘can’, ‘would’, and
the like, let’s call them convensational indirection.)
We can deny indirect requests either directly or indirectly. Responding with a simple
‘no’ is blunt and potentially impolite, but still in order. Answering ‘no’ is heard not as a
commentary on the speaker’s abilities—i.e. not as answering the literal question—but
as a denial of the request. It’s natural to add ‘sorry’, which makes it more polite, though
still direct. And if a request is answered with a ‘no’, earnestly replying in turn, “That’s
just not true—you’re an able-bodied adult fully capable of passing the salt,” or, “That’s a
lie and you know it!” will come across as either incoherent (if the original request was
made directly in the imperative), or coherent but obtuse (if the original request was
made indirectly in the interrogative).
A very common way of indirectly denying an indirect request is to echo the verb of
the literal original question and add a negation. If you ask, “Can you give me the infor-
mation?,” then my response, “I can’t,” is felicitous. If you ask, “Do you want to help me
with this?,” then my response, “I don’t,” is similarly felicitous. It is also felicitous to echo
the entire original question, and add ‘no’, ‘sorry’, and other respectful niceties, as in ‘I am
sorry, but no, unfortunately I can’t give the information to you’, and ‘It hurts me to say so,
but no, I don’t want to help you with this.’ Taken literally, and setting aside the niceties,
these indirect denials are literal assertions about the speaker’s inability to do something
24 Compare to paradigm cases of idioms, such as ‘they tied the knot’, ‘keep an eye out’, and ‘how about the
weather lately?’
or about the believer’s mental state. But they are heard primarily as denials. Earnestly
replying in turn, “That’s just not true. You’re able to give me the information—it’s right
there in your hand, and all you have to do is hand it to me,” would come across as coher-
ent but obtuse and could be met with, “You’re right, I am able to. I’m just not going to.”
9.6. WAM! BAM! POW!

It’s time to relate our discussion of indirection back to the argument against pure invari-
antism, Anti-PI. The ultimate goal is to defend pure invariantism without courting
skepticism. The strategy for achieving this goal is to deny line 2, on the grounds that
High Pro speaks properly but falsely when he says ‘I don’t know’. The proposed tactic for
implementing this strategy is to appeal to High Pro’s primary illocutionary intention in
uttering those words, and thereby drive a wedge between our intuition that he speaks
properly, on the one hand, and the further claim that he speaks truthfully, on the other.
High Pro’s primary illocution is warranted, which is why his speech is proper.
Because I’m appealing to primary illocutionary warrant—or what we might call
primary oratorical warrant—and because I want to emphasize how my proposal both
resembles and differs from WAMs, I hereby name my proposal a POW.
POWs and WAMs are alike because they both appeal to a basic form of assessment
that comes naturally to us when we’re interpreting speech, but is clearly distinct from
the assessment of the truth-value of the literal propositional content of the speaker’s
direct speech act, which, it so happens, is an assertion in all the key cases in the lit-
erature (whence the title ‘warranted assertability maneuver’). Call any device that fits
this mold a basic assessment maneuver, or a BAM. POWs and WAMs are species of
the BAM genus. They distinguish themselves by appealing to different forms of basic
assessment. A WAM explains intuitions by appealing to the truth-value of propositions
pragmatically conveyed by knowledge ascriptions, rather than the literal truth-value
of the knowledge ascriptions themselves. By contrast, a POW explains the intuitions
by appealing to the aptness of the indirect speech acts performed by speakers making
knowledge ascriptions, rather than the literal truth-value of those ascriptions. Indirect
speech acts aren’t limited to speech acts with truth-evaluable, assertive propositional
content. WAMs generate their plausibility from the fact that pragmatically conveyed
propositions attract our attention and figure centrally in our evaluation of speech.
POWs generate their plausibility from the fact that primary illocutionary intentions
attract our attention and loom large in our evaluation of speech.
The POW I propose focuses on High Pro’s primary illocutionary intention to deny
his interlocutor’s request.
Normally the point of asking someone ‘Do you know whether P?’ is not to better
understand their epistemic standing regarding P. Rather, the questioner’s primary
illocutionary intention is to get told whether P. Asking ‘Do you know whether P?’ is a
way of indirectly asking ‘P?’ (The same applies to other know-wh questions featuring
when/how/why/where/what, as well as formulations substituting ‘if ’ for ‘whether’,) As
180 John Turri
evidence of this, notice that ‘Do you know whether P?’ is heard as asking whether P,
we respond to it precisely that way, and we expect people to respond to it that way. If
you ask, “Do you know whether the game is tonight?,” a response of ‘I do know’ could
be interpreted as either playful or humorous, but is otherwise interpreted as uncoop-
erative. And again, the same is true for other know-wh questions. If you ask, “Do you
know when the flight boards?,” a simple ‘Yes I do’ is uncooperative. Stage-setting is
required for us to hear ‘Do you know whether P?’ primarily as posing the question it
literally asks. Its default status is to be heard as the question ‘P?’ (The same applies to
‘Can/could/would you tell me whether P?’, ‘Might you know whether P?’, ‘If you don’t
mind, I’d appreciate it if you could tell me whether P’, etc.)
To ask someone ‘P?’ is to request information. Usually we don’t seek information
just for the sake of it. Often we seek it because we’re going to rely on it in planning or
evaluating courses of action.
Now let’s return to HIGH FLIGHT. When the courier asks Stewart, “Do you know
whether this is a direct flight to Detroit?,” he’s requesting Stewart to provide informa-
tion that will be relied on in making a serious decision. Making this request is the cou-
rier’s primary illocutionary intention. It should be obvious that this is what the courier
is doing. To confirm this observation, notice that the courier’s prompt exactly fits the
model of indirect requests. And it would have been natural for the courier to frame his
final sentence in any of these ways:
• Can you tell me whether P?

• Could you tell me whether P?
• Would you tell me whether P?
• Might you know whether P?
• If you don’t mind, I’d appreciate it if you could tell me whether P.
• It would be very helpful if you could tell me whether P.
Adding ‘please’ would also be perfectly natural and appropriate.

It’s highly unusual for ordinary passengers such as Stewart to be asked to be relied on
for information relating explicitly to such decisions. Thus it’s perfectly understandable
and appropriate for Stewart to not want to grant the courier’s request. Moreover, it’s not
obviously in Stewart’s self-interest to grant the request and undertake responsibility,
and he’s not morally required to do so either, because there are people nearby whose
job it is to provide that information. So Stewart is warranted in denying the request.
And that is exactly what Stewart does, through indirection. By saying ‘I don’t know’,
Stewart indirectly denies the courier’s indirect request, in the typical way: he echoes the
courier’s question and adds a negation. The courier says ‘Do you know . . .?’ and Stewart
replies with ‘I don’t know’.25 By contrast, had Stewart replied with ‘Yes, I do—it’s a direct
25 Stewart might also be encouraging the courier to find another source of information on the matter. It’s
natural, though not required, to interpret Stewart that way.
flight’, he would have thereby accepted the courier’s request and encouraged the cou-
rier to rely on him, thereby undertaking responsibility.
As evidence that Stewart is indirectly denying a request, notice how Stewart’s
response exactly fits the profile of indirect denial. I already noted the echoic formula-
tion. It would be perfectly natural for Stewart to express himself in this context by say-
ing any of the following:
• No.
• No, sorry.
• No, sorry, I don’t.
• No, I’m sorry, I wish I could help, but unfortunately I don’t know.
• I’m sorry, I’d like to help, but I’m not the person to ask about that.
It would also be odd for the courier to respond directly to the literal content of Stewart’s
assertion. If Stewart says, “No, sorry, I don’t know,” it would not be in order for the cou-
rier to respond, “Do you have any evidence for that assertion?” or, “I doubt that that’s
true.” The natural response is more like, “Okay, thanks.”
When evaluating Stewart’s speech, our immediate response is not to think that he
said something false. This is because our evaluation tracks, in the first instance, the
fact that his primary illocutionary intention is warranted (i.e. it tracks primary ora-
torical warrant). Stewart’s point is to deny the request, and denials are neither true nor
false. That Stewart denies the request by making an assertion is incidental and typically
ignored, just as it is incidental and typically ignored when we indirectly deny a request
by saying ‘I can’t’.
The proposal thus far has granted that Stewart denies the request by literally assert-
ing that he doesn’t know. This strikes me as the most plausible account of the situation.
But it’s worth noting that there is a more radical proposal in the neighborhood, namely,
that ‘I don’t know’ functions idiomatically as a way of denying a request for informa-
tion, which complements the fact that ‘Do you know’ functions idiomatically as a way
of requesting information. On this alternative POW, the only speech act Stewart per-
forms is a direct denial, by uttering an idiomatic expression. And whereas I have pro-
posed that Stewart incidentally makes a false assertion that we ignore, the more radical
POW denies that there even is an assertion to be ignored.
This brings me to the essential point in response to line 2 of the argument. Even if
High Pro speaks literally falsely under such circumstances, we should expect this to be
ignored and we should expect his speech to both be and seem proper.
The present proposal seems perfectly fitted to explain the asymmetry in ver-
bal behavior in Low versus High cases. As stakes rise, people tend to become more
unwilling to be relied upon and undertake responsibility for information crucial to
decision-making. And this tendency is not unreasonable, even when holding constant
their confidence and the strength of their epistemic position.
This POW avoids problems faced by Rysiew and Brown’s WAM. First, it doesn’t
court methodological danger from the skeptic’s corner. It isn’t part of my proposal that
182 John Turri
we mistakenly judge that a false knowledge ascription26 is true because it conveys true
information. Rather, we simply ignore the false ascription because it is incidental and
unimportant. Second, and relatedly, this POW doesn’t leave us wondering why Stewart
chose to say something false. Stewart’s utterance follows the typical format of indi-
rect denials. Third, it helps explain why people in HIGH cases respond similarly to
both ‘Do you know whether P?’ and ‘P?’ They respond similarly because ‘Do you know
whether P?’ is simply heard as an indirect request to answer ‘P?’ Moreover, this POW
does all this without interpreting High Pro’s speech as being based on misinformation,
and without claiming that High Pro’s confidence diminishes.
9.7. Conclusion
J. L. Austin (1956/7, p. 11 n.5) once wrote that when we’re investigating why we use cer-
tain words in certain situations, and why certain linguistic behavior is appropriate, we
should “forget, for once and for a while, that other curious question ‘Is it true?’ ” It’s not
clear how consistently Austin thinks we ought to avoid asking ‘Is it true?’, but avoiding
it altogether is certainly unwise. After all, truth often matters, even if it’s almost never
the only thing that matters. If our discussion here is any indication, we can glean some
guidance on when it would be wise to look beyond the question of truth, at least for
a while, namely, when the speakers themselves aren’t primarily concerned with per-
forming a truth-evaluable speech act. When their primary illocutionary intention is,
for example, to give an order or deny a request, the explanation of our reaction to their
speech shouldn’t be expected to track the truth of what they literally say. In such a case,
it’s a distinct possibility that the truth-value of their literal speech will be irrelevant
not only to their own take on the situation, but also to our intuitive assessment of their
behavior as well.
References
Austin, J. L. (1956/7). “A Plea for Excuses,” Proceedings of the Aristotelian Society 57, pp. 1–30.
Bach, K. (2010). “Knowledge In and Out of Context,” in J. K. Campbell, M. O’Rourke, and H. S.
Silverstein (eds), Knowledge and Skepticism. Cambridge, MA: MIT Press, pp. 137–63.
Beebe, J. (2013). “A Knobe Effect for Belief Ascriptions,” Review of Philosophical Psychology 4,
pp. 235-58.
——., and Buckwalter, W. (2010). “The Epistemic Side-Effect Effect,” Mind and Language 25, pp.
474–98.
26 I say knowledge “ascription” even though it’s literally a knowledge denial (i.e. an assertion that knowl-
edge is not present), because I want to avoid confusion between this sort of denial, on the one hand, and
denying a request, on the other.
——., and Jensen, M. (2012). “Surprising Connections Between Knowledge and Action: The
Robustness of the Epistemic Side-Effect Effect,” Philosophical Psychology 25, pp. 689–715.
BonJour, L. (2010). “The Myth of Knowledge,” Philosophical Perspectives 24, pp. 57–83.
Brown, J. (2005). “Comparing Contextualism and Invariantism on the Correctness of
Contextualist Intuitions,” Grazer Philosophische Studien 65, pp. 136–64.
——. (2006). “Contextualism and Warranted Assertibility Manoeuvres,” Philosophical Studies
130, pp. 407–35.
Buckwalter, W., Rose, D., and Turri, J. (2013). “Belief through Thick and Thin,” Noûs. DOI: 10.1111/
nous.12048.
Cohen, S. (1999). “Contextualism, Skepticism, and the Structure of Reasons,” Philosophical
——. (2005). “Knowledge, Speaker and Subject,” Philosophical Quarterly 55, pp. 199–212.
Conee, E. (2013). “Contextualism Contested,” in M. Steup, J. Turri, and E. Sosa (eds),
Contemporary Debates in Epistemology, 2nd ed. Malden, MA: Wiley-Blackwell, pp. 60–9.
Davis, W. (2007). “Knowledge Claims and Context: Loose Use,” Philosophical Studies 132, pp.
395–438.
DeRose, K. (1999). “Contextualism: An Explanation and Defense,” in J. Greco and E. Sosa (eds),
The Blackwell Guide to Epistemology. Malden, MA: Blackwell.
——. (2009). The Case for Contextualism: Knowledge, Skepticism, and Context, Volume 1.
Fantl, J., and McGrath, M. (2009). Knowledge in an Uncertain World. Oxford: Oxford
University Press.
Feltz, A., and Zarpentine, C. (2010). “Do You Know More When It Matters Less?” Philosophical
Psychology 23, pp. 683–706.
Greco, J. (2010). Achieving Knowledge: A Virtue-Theoretic Account of Epistemic Normativity.
Cambridge: Cambridge University Press.
Grimm, S. (2011). “On Intellectualism in Epistemology,” Mind 120, pp. 705–33.
Hawthorne, J. (2004). Knowledge and Lotteries. Oxford: Oxford University Press.
Lewis, D. (1996). “Elusive Knowledge,” Australasian Journal of Philosophy 74, pp. 549–67.
MacFarlane, J. (2005). “The Assessment Sensitivity of Knowledge Attributions,” in T. Gendler
and J. Hawthorne (eds), Oxford Studies in Epistemology. Oxford: Oxford University Press, pp.
190–205.
McGrath, M. (2010). “Contextualism and Intellectualism,” Philosophical Perspectives 24, pp.
383–405.
Moore, G. E. (1959). Philosophical Papers. New York: Collier Books.
Murray, D., Sytsma, J., and Livengood, J. (2013). “God Knows (But Does God Believe?),”
Philosophical Studies 166, pp. 83–107.
Myers-Schulz, B., and Schwitzgebel, E. (2013). “Knowing That P Without Believing That P,” Noûs
47, pp. 371–84.
Pinillos, Á. (2012). “Knowledge, Experiments and Practical Interests,” in J. Brown and M. Gerken
(eds), New Essays on Knowledge Ascriptions. Oxford: Oxford University Press.
Reed, B. (2007). “The Long Road to Skepticism,” Journal of Philosophy 104, pp. 236–62.
——. (2010). “A Defense of Stable Invariantism,” Noûs 44, pp. 224–44.
Rose, D., and Schaffer, J. (2013). “Knowledge Entails Dispositional Belief,” Philosophical Studies
166, S19–S50.
184 John Turri
Rysiew, P. (2001). “The Context-Sensitivity of Knowledge Attributions,” Noûs 35, 4, pp. 477–514.
——. (2007). “Speaking of Knowledge,” Noûs 41, pp. 627–62.
Schaffer, J. (2006). “The Irrelevance of the Subject,” Philosophical Studies 127, pp. 87–107.
——., and Knobe, J. (2012). “Contrastive Knowledge Surveyed,” Noûs 46, pp. 675–708
Searle, J. (1979). Expression and Meaning. Cambridge: Cambridge University Press.
Sripada, C., and Stanley, J. (2012). “Empirical Tests of Interest-Relative Invariantism,” Episteme
9, pp. 3–26.
Stanley, J. (2005). Knowledge and Practical Interests. Oxford: Oxford University Press.
Starmans, C., and Friedman, O. (2012). “The Folk Conception of Knowledge,” Cognition 124, pp.
272–83.
Stine, G. (1976). “Skepticism, Relevant Alternatives, and Deductive Closure,” Philosophical
Studies 29, pp. 249–61. Reprinted (1st ed. 2000; 2nd ed. 2008), in E. Sosa and K. Jaegwon (eds),
Epistemology: An Anthology. Malden, MA: Blackwell.
76, pp. 138–55.
Turri, J. (2010a). “Epistemic Invariantism and Speech Act Contextualism,” Philosophical Review
119, pp. 77–95.
——. (2010b). “Prompting Challenges,” Analysis 70, pp. 456–62.
——. (2011). “The Express Knowledge Account of Assertion,” Australasian Journal of Philosophy
89, pp. 37–45.
——. (2012). “Is Knowledge Justified True Belief?,” Synthese 184, pp. 247–59.
——. (2013). “A Conspicuous Art: Putting Gettier to the Test,” Philosophers’ Imprint 13,
pp. 1–16. <http://hdl.handle.net/2027/spo.3521354.0013.010> (last accessed November
20, 2013).
——. (Under review). “The Problem of ESEE Knowledge.”
——., and Friedman, O. (Forthcoming). “The Folk Epistemology of Lotteries,” in J. Beebe (ed.),
Advances in Experimental Epistemology. Continuum.
Unger, P. (1975). Ignorance: A Case for Skepticism. Oxford: Oxford University Press.
Weinberg, J., Nichols, S., and Stich, S. (2001). “Normativity and Epistemic Intuitions,”
Williamson, T. (2005). “Contextualism, Subject-Sensitive Invariantism and Knowledge of
Knowledge,” Philosophical Quarterly 55, pp. 212–35.
PA RT I I I
Challenges
10
The Challenge of Sticking with
Intuitions through Thick and Thin
Jonathan M. Weinberg and Joshua Alexander†
Our disagreement about the nature and epistemic authority of intuitions is at root
a battle for the preservation of philosophy as an autonomous field of inquiry.
Gary Gutting (1998)
10.1. Introduction
Philosophical discussions often involve appeals to verdicts about particular cases,
sometimes actual, more often hypothetical, and usually with little or no substantive
argument in their defense. Philosophers—including ones on both sides of debates
over the standing of this practice—have very often called the basis for such appeals
“intuitions.” But, what might such “intuitions” be such that they could legitimately
serve such purposes? Answers vary, ranging from “thin” conceptions that identify
intuitions as merely instances of some fairly generic and epistemologically uncon-
troversial category of mental states or episodes to “thick” conceptions that add to
this thin base certain semantic, phenomenological, etiological, or methodological
conditions.
Thick conceptions have become increasingly popular in recent years, in part because
they offer a way of responding to recent empirical challenges to our intuition-deploying
practices. The basic idea is to insulate our intuition-deploying practices from these
† Authorship is equal. We would like to thank Bryce Huebner, Kaija Mortensen, two anonymous referees
from Oxford University Press, members of the philosophy departments at Florida International University,
The University of Illinois at Urbana-Champaign, York University, Marist College, and Siena College, and
participants at the 2011 Metro Experimental Research Group’s Experimental Philosophy Conference,
the Experimental Philosophy Society Meeting at the 2011 Pacific Division Meeting of the American
Philosophical Association, and the 2011 Joint Meeting of the Society for Philosophy and Psychology and the
European Society for Philosophy and Psychology for valuable feedback on earlier versions of this chapter.
188 Jonathan M. Weinberg and Joshua Alexander
challenges by arguing that, whatever it is that experimental philosophers have been

studying, they haven’t been studying the right kinds of mental states or episodes—that
is, the kinds of mental states or episodes at play in our intuition-deploying practices.
This puts philosophers in the position of being able to argue that the results of these
studies shed little insight on the standing of our actual intuition-deploying practices.
Although adopting a thick conception of philosophical intuitions may thus seem to
put philosophers in a position to defend against recent empirical challenges, we will
contend the apparent strength of thick conceptions is only apparent. Thick concep-
tions turn out to have their own methodological problems—some actually fail to prop-
erly immunize our intuition-deploying practices from the kinds of problems raised
by recent empirical challenges, others expose our intuition-deploying practices to dif-
ferent kinds of empirical challenges, and still others leave us in the methodologically
untenable position of being unable to determine when anyone is doing philosophy
correctly.
10.2. Philosophical Intuitions

According to one rather influential way of thinking about philosophical practice,
philosophical theories are measured (at least in part) by our philosophical intuitions.
On this view, we advance philosophical theories on the basis of their ability to explain
our philosophical intuitions, defend their truth on the basis of their overall agreement
with our philosophical intuitions, and justify our philosophical beliefs on the basis of
their accordance with our philosophical intuitions. According to this way of think-
ing about philosophy, the role and corresponding epistemic status of philosophical
intuitions is similar to the role and corresponding epistemic status of perceptions (see,
e.g., Bealer 1998 and Sosa 1998). Like perceptions, philosophical intuitions provide a
non-inferential, defeasible justificatory foundation in at least the following way: a per-
son may appeal to an intuition as evidence without having to provide further evidence
for the intuition.1
Quite possibly the most famous example of this way of thinking about philosophical
practice is Edmund Gettier’s (1963) use of two thought experiments to help show that
a person’s justified true belief need not count as knowledge. Each of Gettier’s thought
experiments involves Smith who has deduced a true belief that q from a justified
false belief that p and, on that basis, formed a justified true belief that q. According to
Gettier, despite now having a justified true belief that q, Smith doesn’t know that q. Of
course, we aren’t just supposed to take Gettier’s word on this. Instead, it’s supposed to
seem obvious to us when we consider each case that Smith doesn’t know that q. And,
these shared judgments—our philosophical intuitions about the two hypothetical
1 This way of thinking about things leaves open whether treating intuitions (or perceptions) as evidence
involves treating psychological states (or propositions about psychological states) as evidence or treating the
contents of those psychological states as evidence.
Sticking with Intuitions through Thick and Thin 189
cases—are what are supposed to count as sufficient evidence that sometimes a person’s
justified true belief doesn’t count as knowledge.2
Examples like this abound. In fact, this way of thinking about the epistemic status
of philosophical intuitions is so widespread in contemporary philosophy that George
Bealer (1998) has called it part of our “standard justificatory procedure.”3 Other philos-
ophers from a wide range of metaphilosophical temperaments have gone further than
Bealer, claiming not only that philosophical intuitions are part of our standard justifi-
catory procedure, but also that this fact is part of what defines uniquely philosophical
methodology. For example, Janet Levin (2005) writes:
This procedure of rejecting or modifying theses in the face of intuitively convincing counterex-
amples has been characteristic, perhaps definitive, of philosophical argumentation throughout
its history.
And Alvin Goldman (2007) writes:

One thing that distinguishes philosophical methodology from the methodology of the sciences
is its extensive and avowed reliance on intuition. Especially when philosophers are engaged in
philosophical “analysis,” they often get preoccupied with intuitions.4
But what are philosophical intuitions? Answers vary, often reflecting different attitudes
about how we should go about trying to answer this question. Some philosophers
encourage us to consult our intuitions about intuitions; others recommend paying
special attention to what they introspectively seem to be from the first-person point
of view; still others advise looking closely at their role in philosophical practice. These
different approaches have produced a range of conceptions, ranging from what we will
call “thin” conceptions, which identify intuitions as merely instances of some fairly
generic and epistemologically uncontroversial category of mental states or episodes, to
“thick” conceptions, which add to this thin base certain semantic, phenomenological,
etiological, or methodological conditions.
Let’s start with the thinnest conception, according to which philosophical intui-
tions are simply instances of some fairly generic and epistemologically uncontroversial
category of mental states. Timothy Williamson (2007) has been the most visible and
vigorous proponent of this way of thinking about philosophical intuitions, arguing
2 We are starting off uncommitted about the precise psychological nature of intuitions, but for ease of
exposition (and following much extant practice) we will use terms like “intuition,” “intellectual seeming,”
and “judgment” more or less interchangeably.
3 Bealer not only makes the descriptive, sociological claim that philosophical intuitions are in fact a
standard part of our justificatory practices. He also argues for the normative claim that they must be. The
normative claim is part of his famous argument for the incoherence of empiricism. The basic idea is that
empiricists cannot defend their own methodological commitments using any set of justificatory resources
that don’t include philosophical intuitions.
4 One of the most interesting aspects of Goldman’s view about philosophy’s intuition-deploying practices
is his subsequent endorsement of a view he calls “mentalism.” According to this view, our philosophical
intuitions tell us less about the world (or at least the non-psychological world) than they do about the ways in
which we think about the world.
that more substantive conceptions leave out much of what is appealed to as evidence
when philosophers claim “the intuitiveness of some theories as a virtue and the coun-
terintuitiveness of others as a vice” (p. 217).5 The idea is that substantive conceptions
discriminate, and, in doing so, fail to reflect the wide variety of things that answer to
the name ‘intuition’ in philosophical practice. Williamson encourages a more toler-
ant view, according to which intuitions are (at most) just judgments or inclinations to
judge. This kind of maximal inclusiveness promises not only to better reflect what it is
that philosophers actually do, it might also seem to insulate philosophical intuitions
from skeptical challenge. The more that counts as intuitional evidence, the more like a
dead end global intuitional skepticism becomes.
The advantages of maximal inclusiveness, however, come at the cost of evidential
diversity. In the attempt to treat as much as possible as intuitional evidence, we risk dis-
counting important differences between various kinds (e.g., memorial, introspective,
perceptual, inferential, testimonial, and even more narrowly construed intuitional
kinds) of philosophical evidence—differences that figure significantly into questions
of reliability, defeasibility, and conflict resolution. Different kinds of evidence have
different strengths, and different weaknesses, and a responsible methodology for any
field of inquiry should attend closely to those differences, not paper them over. But,
even setting aside concerns about a lack of evidential diversity, there is another reason
to worry that maximal inclusiveness has dangerous vices to go along with its virtues.
As we will see, many contemporary philosophers want to be able to say that not all
intuitions are created equal—that is, that not everything that is appealed to as intui-
tional evidence is genuine intuitional evidence. And, we might worry that maximally
inclusive conceptions leave us without the resources needed to mark this difference.
In light of these sorts of considerations motivating some degree of intra-intuitive
discrimination, many philosophers have been tempted to adopt a thicker, more dis-
criminating conception of intuitional evidence. For a time, the most popular way
to do this was to argue that genuine philosophical intuitions have a special kind of
propositional content or distinctive phenomenology. Some philosophers argued that
philosophical intuitions must have modally strong propositional content (e.g., BonJour
1998); other philosophers argued that philosophical intuitions must have abstract
propositional content (e.g., Sosa 1998); and still others argued that philosophical intui-
tions must have the appearance of necessity (e.g., Bealer 1998). The general idea was to
locate some distinctive feature in the content or appearance of genuine philosophical
intuitions that could be used to pick them out of the crowd.
5 David Lewis (1983), Peter van Inwagen (1997), Jonathan Ichikawa (MS), Herman Cappelen (2012), and
Wesley Buckwalter and Stephen Stich (2014) also endorse this kind of maximal inclusiveness. Williamson
(2007) actually suggests that the practice of appealing to philosophical intuitions as evidence rests on a mis-
understanding of the nature of philosophical evidence, arguing, “Philosophers might be better off not using
the word ‘intuition’ and its cognates. Their main current function is not to answer questions about the nature
of evidence on offer but to fudge them, by appearing to provide answers without really doing so” (2007,
p. 220). See Joshua Alexander (2010) for a reply to Williamson’s argument against the philosophical signifi-
cance of philosophical intuitions.
In recent years, fashion has shifted away from attempts to identify genuine philo-
sophical intuitions on the basis of specific semantic or phenomenological characteris-
tics, although as we will see phenomenological characterizations are beginning to see
a revival of sorts. More popular now are approaches that focus on etiology and meth-
odology—that is, on where philosophical intuitions come from and what we do with
them once they are here. Kirk Ludwig (2007) provides an admirably clear and straight-
forward articulation of the etiological approach, arguing that genuine philosophical
intuitions are judgments made only on the basis of conceptual competence:6
For terminological clarity, I will use “intuition” to mean an occurrent judgment formed solely
on the basis of competence in the concepts involved in response to a question about a scenario,
or simply an occurrent judgment formed solely on the basis of competence in the concepts
involved in it (in response, we might say, to the null scenario). Intuitions in this sense are what
we seek to elicit in response to questions about scenarios in thought experiments. (pp. 135–6)
Ludwig goes on to argue that not all judgments generated in response to philosophical
thought experiments express our conceptual competence. Some, he suggests, are:
judgments based on well-entrenched empirical beliefs, judgments based on what would be
standardly implicated by the sentence that expresses the judgment, judgments based on reading
more or less into the scenario than is intended by the experimenter, judgments based on infor-
mation carried by the linguistic vehicle for it as opposed to what it means, and judgments made
in circumstances in which none should be made because the idealization that the terms involved
are semantically complete breaks down dramatically, as in borderline cases of “bald” or “heap”
or “person,” or, generally speaking, judgments whose etiology is not solely the competence of the
subject in use of the concepts in response to the scenario as described. (pp. 137–8)
Since not all judgments generated in response to philosophical thought experiments

express our conceptual competence, not all judgments are genuine philosophical intui-
tions. And, our task becomes to “try to factor out the contribution of competencies from
the other factors” (Ludwig 2007, pp. 144–5) when determining whether a particular
response counts as a genuine philosophical intuition (see, also, BonJour 1998, Sosa 1998,
and Henderson and Horgan 2001). This last point suggests a place where the etiological
gives way to the methodological. More is needed than simply a call for proper etiologi-
cal pedigree. What is needed is some way of determining when this call has been met.
Antti Kauppinen (2007) suggests that what we need is reflection: careful examination
and evaluation not only of our judgments about hypothetical cases, but also of the cases
themselves and what influence philosophical commitments might have on what details
we find relevant in those cases. Only this kind of careful, paradigmatically philosophical
reflection will help us pick out genuine philosophical intuitions.
6 Others in the etiological camp use somewhat different language, suggesting that genuine philosophi-
cal intuitions are judgments (or inclinations to judge) based on conceptual understanding (e.g., Sosa 1998),
determinate concept possession (e.g., Bealer 1998), or what competent users of the relevant concepts could
say if they considered the case in sufficiently ideal conditions where their judgments were influenced only by
semantic considerations (e.g., Kauppinen 2007).
10.3. The Challenge of Sticking with Thin

Conceptions
Discrimination always requires some defense, something that shows that our views are
discrimating rather than discriminatory, and some contemporary philosophers defend
their turn away from maximal inclusiveness by suggesting that less inclusive concep-
tions of philosophical intuition seem better able to protect our intuition-deploying
practices from a recent empirical challenge to those practices.7 Here’s the empirical
challenge—which we have elsewhere called the restrictionist challenge—in a nutshell.
Our intuition-deploying practices are in good standing only to the extent to which
philosophical intuitions are appropriately sensitive—sensitive only to those things that
are relevant (or, at least, those things that philosophers commonly take to be relevant)
to the truth or falsity of the claims for which the evidence is supposed to provide evi-
dence. The problem is that recent empirical work suggests that at least some philo-
sophical intuitions are problematically sensitive—sensitive to things (e.g., ethnicity,
gender, affectivity, and presentation order) that haven’t traditionally been thought to
be relevant to the truth or falsity of philosophical claims (see, for example, Weinberg
et al. 2001; Buckwalter and Stich 2014; Nichols and Knobe 2007, and Swain et al. 2008).
This problematic sensitivity is made worse by the fact that we currently lack the means
to either correct for it or predict from the armchair when or where else it will appear.
We find ourselves in the untenable epistemic position of suspecting that some intui-
tional evidence isn’t trustworthy without being able to reliably predict what intuitional
evidence isn’t trustworthy.8
What makes the restrictionist challenge so challenging is that almost any way of
responding to it—at least, any way that begins by accepting the empirical results them-
selves—seems to involve something of a sea change in how we think about the aims
or methods of philosophical inquiry. Alvin Goldman (2007), for example, is able to
accommodate intuitional diversity by suggesting that philosophical inquiry aims at
7 There have been other challenges to our intuition-deploying practices in recent years. Hilary Kornblith
(1998) has argued that, while philosophical intuitions count as evidence for something, they don’t count as
evidence for anything that ought to be of interest to philosophers. Robert Cummins (1998) has argued that
philosophical intuitions cannot be treated as evidence because we cannot determine antecedently and inde-
pendently whether or not they are reliable guides to the truth. And, Michael Devitt (1994) and Catherine
Elgin (1996) have argued that philosophical intuitions cannot be treated as evidence because they are fallible.
These challenges have tended to rest on too narrow a conception of philosophy (e.g., Kornblith’s challenge),
to demand of us the epistemically impossible (e.g., Cummins’ challenge), or to be so strong that they deem
all putative evidence untrustworthy (e.g., Devitt and Elgin’s challenge).
8 It is important to see that the challenge does not require taking on a claim as strong as the full-out unreli-
ability of intuitions. It is generally enough to raise serious concerns about a putative source of evidence to
show it has some substantial liability to error, even if its overall rate of reliability is well above a coin flip. (For
example, the conventional reporting norm in the social sciences is p < 0.05—not p < 0.5!) We articulate the
overall dialectic of the restrictionist challenge in greater detail in Alexander and Weinberg (2007), Weinberg
(2007), and Alexander (2012). We also have a more in-depth discussion of the relevant notion of “unreliabil-
ity” at stake in these debates in Alexander and Weinberg (Forthcoming). This aspect of the dialectic is not
one that is in dispute with our targets here.
helping us better understand our individual or shared concepts. On this view, intui-
tional diversity reflects conceptual diversity, and not all forms of conceptual diversity
are necessarily problematic. But, this way of responding to the restrictionist challenge
trades a view of philosophy as telling us something about the world for a view of phi-
losophy as telling us something about the ways in which we think about the world,
and invites a shift towards certain kinds of philosophical relativism anathema to many
traditional philosophers.9 Nevertheless, resisting this trade comes with its own meth-
odological costs, involving (at the very least) local methodological restrictions on our
intuition-deploying practices together with a global shift in how we think about and
approach these practices. Quarantine only works if we understand the dimensions of
an outbreak, and even then only education helps restore and maintain public health, so
this way of responding to the empirical challenge not only places limits on what intui-
tional evidence can be appealed to but also conditions on when intuitional evidence
can be appealed to, and here the suggestion is that our intuition-deploying practices
must be informed by an understanding of the relevant psychology, cognitive science,
and empirically-informed philosophy of mind.
It is perhaps unsurprising, then, that a more popular way of responding to the
restrictionist challenge has become to contest the empirical results themselves, or at
least to try to minimize their relevance.10 And, here is where thick conceptions of phil-
osophical intuition might seem to have a dialectical advantage over thin conceptions
(though we do not claim to have exhausted the resources available to thin concep-
tions here). If not all putative intuitional evidence is genuine intuitional evidence, and
if it can be shown that the restrictionist challenge is based on something other than
a close study of genuine intuitional evidence, then philosophers are in a position to
argue that the restrictionist challenge is actually silent on the standing of philosophy’s
intuitions-deploying practices.
Bealer (1998) provides an early version of this way of defending philosophy’s
intuition-deploying practices from empirical challenge:
Many philosophers believe that the empirical findings of cognitive psychologists . . . cast doubt
on [intuitions’] epistemic worth. But, in fact, although these studies bear on “intuition” in an
indiscriminate use of the term, they evidently tell us little about the notion of intuition we have
9 See Gonnerman and Weinberg (Forthcoming) for a distinct critique of this methodology.
10 Here we focus only on one way of trying to do this; there are others. Some philosophers have responded
by questioning the studies themselves (see, e.g., Cullen 2010). But criticizing the experiments and the
interpretation of their results is a scientifically substantive project and so this means more empirical work
for philosophers and not less. And, this means that at least the second part of the restrictionist challenge
remains, namely, to engage in more careful empirical study of philosophical intuitions. Other philosophers
have mounted what we have elsewhere called the expertise defense, arguing that philosophers shouldn’t be
interested in folk philosophical intuitions (see, e.g., Hales 2006, Ludwig 2007, and Williamson 2007; and,
for critical discussion, see Weinberg et al. 2010 and Alexander 2012). But, who has expertise about what
and under what circumstances turns out to be a rather complicated empirical question. It seems that only
certain kinds of training help improve task performance and, even then, only for certain kinds of tasks, and
there is reason to worry that philosophical training is not the right kind of training and that philosophical
thought-experimenting is not the right kind of task.
been discussing, which is relevant to the justificatory practices in logic, mathematics, philoso-
phy, and linguistics. As far as I have been able to determine, empirical investigators have not
attempted to study intuitions in the relevant sense. (p. 213)
Although not directed at the restrictionist challenge (it was instead directed at ear-
lier empirical challenges stemming from work by folks like Richard Nisbett, Daniel
Kahneman, and Amos Tversky), this strategy has found new voice among philoso-
phers concerned more directly with the restrictionist challenge. When discussing the
methods employed by many experimental philosophers, for example, Kirk Ludwig
(2007) writes:
The first point to make is that . . . responses to surveys about scenarios used in thought experi-
ments are not ipso facto intuitions, that is, they are not ipso facto judgments which express solely
the subject’s competence in the deployment of the concepts involved in them in response to the
scenario . . . . The task when presented with responses which we know are not (at least all) intui-
tions is to try to factor out the contribution of competencies from the other factors. This requires
an understanding of what the various factors are that may influence responses and enough
information about each subject to be able to say with some confidence what factors are at work.
It is clear that in the circumstances in which these surveys are conducted we do not have this
kind of information. (pp. 144–5)
In a similar vein, Antti Kauppinen (2007) writes:

My first criticism of the positive thesis of experimentalism—that surveys are a reliable source
of evidence for philosophically relevant claims about folk concepts—can then be formulated
as follows: the actual studies conducted so far have failed to rule out competence failures, per-
formance failures, and the potential influence of pragmatic factors, and as such do not yield
the sort of results that could support or raise doubts about philosophical appeals to conceptual
intuitions. (p. 105)
Simon Cullen (2010) explains the methodological burden being levied here:

[W]‌e must distinguish intuitions from survey responses . . . . Survey responses are a kind of
behaviour generated by several different kinds of inputs, only one of which might be a [philo-
sophical intuition]. Other inputs are the background beliefs a respondent draws on when inter-
preting a vignette; her beliefs about what the researchers are interested in; her sensitivity to the
conversational norms; and so forth. (pp. 277–8)
And John Bengson (2012) argues that this burden has not been met:
[T]‌hese attacks neglect a considerable gap between the answers elicited by the relevant empiri-
cal studies and the intuitions about which naysayers naysay. It cannot innocently be assumed
that subjects’ answers expressed how things struck them—what intuitions they had, if any. The
point is simple, but not insignificant. For, I will argue, it implies that we are at the present time
unwarranted in drawing any negative conclusions about intuitions from the relevant empirical
studies. (p. 496)
The common idea seems to be this. Only very special kinds of things count as genuine
philosophical intuitions and, owing to the limits of their methods, experimental phi-
losophers haven’t been in touch with these kinds of things. Or, put more conservatively,
experimental philosophers can’t be certain that they’ve been studying the right kinds
of things. Since we can’t be certain that they’ve been studying the kind of intuitional
evidence that is relevant to philosophy’s intuition-deploying practices, the results of
these studies give little insight into the standing of those practices.
10.4. The Challenge of Sticking with Thick

Intuitions
Let’s call this way of responding to the restrictionist challenge by appealing to a thick
account of genuine philosophical intuition, the thickness defense. It is worth taking a
minute to unpack the commitments that any proponent of a thickness defense must
sign on to, since then we can investigate whether those commitments are ones that are
likely to be satisfied. There are generally two kinds of conditions that must hold for any
given version of the thickness defense to succeed in providing a response to the restric-
tionist challenge. The first kind we will call veritist conditions, as they concern the pro-
pensity of genuine philosophical intuitions to track (or diverge from) the truth. Here,
we will consider two different veritist conditions; there may be others. The first we will
call the immunity condition: genuine philosophical intuitions are not themselves sus-
ceptible, to any worrisome degree, to the sorts of problematic effects that form the basis
of the restrictionist challenge just rehearsed. The qualifier “to any worrisome degree”
is important, since all that needs to be required is a methodologically sufficient immu-
nity to these sorts of errors, and not some sort of quasi-Cartesian absolute immunity.
The immunity condition is about the sorts of effects already explored by experimen-
tal philosophers, but there may well be other sorts of problematic effects beyond the
handful that have been investigated so far. Now, it truly would be a skeptical argument
of a fairly traditional sort to attempt to impose a requirement that proponents of the
thickness defense rule out even the mere possibility of some other, as-yet-unexplored
and as-yet-unspecified sources of error. And so it is important to see that this is not at
all the kind of argument that restrictionists are trafficking in. What we can legitimately
require, however, is that the specific sorts of factors that are meant to distinguish genu-
ine philosophical intuitions from “faux” philosophical intuitions—what we will call
thickening factors—do not seem at all likely to generate their own distinct problem-
atic effects. Thickening factors are meant to save traditional armchair philosophical
practices; they shouldn’t raise their own new dangers for it. Should a thickening factor
seem likely to lead to its own form of troubles, then a version of the thickness defense
employing that thickening factor would be ultimately fruitless, merely leading restric-
tionists to swap brands of ammunition. Any given version of the thickness defense, if
it is to cure traditional armchair philosophical practices, must thus first do no harm to

those practices; we thus call this second veritist condition the hippocraticity condition.
Where the immunity and hippocraticity conditions both pertain to the
truth-tracking or truth-diverging properties of alleged genuine philosophical intui-
tions, there is another set of conditions that must be met by any version of the thickness
defense. We will call these methodological conditions—can the proposed thickening
factors, regardless of their veritistic properties, be sufficiently put to work in philo-
sophical practice? There are at least two methodological conditions that must be met
by any version of the thickness defense. The first condition concerns whether agents
can actually distinguish between genuine and faux philosophical intuitions. After all,
here’s a candidate thickening factor that clearly will satisfy any veritist conditions: gen-
uine philosophical intuitions are all and only the ones that are true. The problem is that
proposing a verity condition as a thickening factor would obviously be a non-starter
without some further means at hand for separating true philosophical intuitions from
false.11 Thickening factors, thus, need to be detectable, and not just in principle or the-
ory, but in practice and—most of all—from the armchair. We call this the manifestabil-
ity condition.
While immunity, hippocraticity, and manifestability are rather substantive condi-
tions of adequacy, it is still possible for a version of the thickness defense to satisfy all
three of these conditions while still failing to provide a satisfactory defense of current
armchair philosophical practices. This is possible in cases where the thickening fac-
tor is one that, while veritistically virtuous and potentially manifestable, nonetheless
does not currently play a sufficiently universal role in philosophical practice. Suppose
it turned out, just to illustrate with a toy example, that intuiting while lying prone in a
very brightly lit, bare room provided maximal stability to our intuitions. It would be
easy to tell when we are in such a situation, and we could tell from the armchair that we
were in it (well, ex hypothesi we’d be telling it from the floor, but you get the idea). The
problem is that since philosophers do not currently pursue this technique, this version
of the thickness defense would still require some substantial changes to the norms of
philosophical cognition and argumentation. Now, this commitment need not apply
to everyone running a thickness defense, since it is open to defenders of armchair
philosophical practices to deny that, in defending those practices, they are also look-
ing to defend the status quo cathedrarum. Bealer, for example, sometimes protests that
he is only defending the possibility of armchair philosophy. But we think that many,
and perhaps most, proponents of the thickness defense do not take themselves to be
defending just some merely possible version of armchair philosophical methodology.
They take themselves to be defending what philosophers have been doing in recent
11 It is important to note that we aren’t objecting to the idea that truth might be a necessary condition on
genuine philosophical intuitions, although we are rather skeptical that it is; our concern is simply that the
condition can’t do any work to meet the restrictionist challenge unless we have some way of identifying
which of our philosophical intuitions are true.
decades, and continue to be doing today. These proponents of the thickness defense, at
least, are committed to what we would call the current practice condition.12
Although we will not deploy the current practice condition for any heavy lifting
in the discussion that follows, we have two reasons for having wanted to articulate it
briefly here. First, we think proponents of traditional armchair philosophical practice
should be more cognizant than they often are of the fact that many ways of respond-
ing to the restrictionist challenge are rather more revisionary than they seem to com-
monly take them to be, and we will flag a few instances of this phenomenon below.
Second, it helps illustrate that the three main conditions that we are looking to use in
our evaluation of the thickness defense—immunity, hippocraticity, and manifestabil-
ity—are likely not exhaustive. There are probably other substantive empirical com-
mitments that proponents of the thickness defense need to have a right to, if they are
to succeed in their response to the restrictionist challenge.13 It would be wise for them
to start paying attention to these commitments, and articulating and defending them
themselves. We hope, even if they don’t agree with our critical evaluations below, that
they will find our own initial articulation of these commitments helpful. We take it that
both restrictionists and the defenders of the armchair share the desire to gain as clear a
picture as possible of how well-functioning philosophical practices can work, and thus
also share an interest in not merely scoring unilluminating debating victories, but in
learning from the moves and countermoves of the debate.
Our project here is to score a number of prominent versions of the thickness defense
in terms of these three major commitments: immunity, hippocraticity, and manifes-
tability. We will do so as follows. For each proposal, we will evaluate its status for each
commitment using a very rough scale of {highly likely; open, but probable; wide open;
open, but improbable; highly unlikely}. Think of the first and last of these values as
being likely enough that one should, all things considered, at least pro tempore believe
the claim or its negation. The middle three values, on the other hand, do not involve
probability sufficient to license believing the claim or its negation, although “open,
but probable” and “open, but improbable” mark a current tendency in the state of evi-
dence. Since proponents of any version of the thickness defense need all three of these
commitments to be true, we will focus our evaluation on the conjunction of all three
commitments taken together. Here, we will combine these evaluations not by anything
like, say, averaging them, but rather by taking the score of the worst member of the
set to provide the score of the set on the whole. After all, the fact that Pr(A&B&C)
cannot be any greater than MIN({Pr(A), Pr(B), Pr(C)}) is itself trivial. And on a fairly
12 And we think they are right to be. Speaking just for ourselves, we are happy to stipulate the metaphysical
possibility of a community using the target armchair practices in a methodologically unproblematic way.
The extent to which we actually resemble the members of such a community is, of course, the tricky empiri-
cal question.
13 The same holds true for other attempts to defend armchair philosophical practice from the restrictionist
challenge. See, for example, recent discussions of philosophical expertise (Williamson 2007; Weinberg et al.
2010; Williamson 2011).
weak set of further assumptions, this will actually turn out to be a rather generous way
to produce an overall evaluation: unless at least two of Pr(A), Pr(B), and Pr(C) are 1
(and none are 0) it will turn out that Pr(A&B&C) will be strictly less than MIN({Pr(A),
Pr(B), Pr(C)}). In the end, then, each proposal can itself be evaluated as highly likely,
highly unlikely, or as a hypothesis of greater or lesser plausibility.
After scoring the individual proposals, we can consider how things stand in the
debate between the restrictionists and the defenders of the armchair. If any version
of the thickness defense comes out as highly likely, then that would constitute a pretty
clear victory for the thickness defense. Conversely, if all of them were to come out as
highly unlikely, then that would also constitute a pretty telling rejection of this whole
line of response to the restrictionist challenge. (Perhaps further investigations could
revise either such result.) But what if some versions of the defense come out as being
only a more or less plausible sort of hypothesis, and with no versions of the defense
doing any better than that? It’s important to keep clearly in mind that such an outcome
would make for a victory for the restrictionist challenge as well, even if a less satisfy-
ingly resounding one. After all, it is a core part of the restrictionist challenge that the
particular boundaries of where our intuition-deploying practices are unproblematic
can only be discerned from outside of the armchair, by means that involve substantial
empirical investigation. So, if all versions of the thickness defense stand at this time as,
at best, under-substantiated empirical hypotheses, then this kind of further empiri-
cal investigation is exactly what is needed to try to resolve their status. And the less
plausible the hypothesis, the more that this is so, for obvious reasons. But even should
there be at least one version of the thickness defense that is modestly plausible—even
if still not yet sufficiently confirmed—all that this means is that the defenders of arm-
chair philosophical practice might have an interesting line of investigation that they
could (and should!) pursue. That there are paths that might lead to results that block
the restrictionist challenge does not at all entail that such results really are there to be
had. Only a proper investigation would tell.
While there are as many different versions of the thickness defense as there are
potential thickening factors, in what follows we will focus on the three most prominent
ways of setting up the thickness defense: phenomenological approaches that focus on
introspectively accessible aspects of intuitional experience; etiological approaches that
focus on conceptual competence; and methodological approaches that focus on philo-
sophical reflection. Each approach is committed to the idea that genuine philosophi-
cal intuitions are sufficiently immune from the kinds of problematic effects that have
served as the basis of the restrictionist challenge and from the kinds of problematic
effects that would only provide experimental restrictionists with a different basis for
that challenge, and to the idea that we can distinguish genuine philosophical intui-
tions from faux philosophical intuitions on the basis of nothing more than armchair
reflection. They disagree only about how best to separate the intuitional wheat from
chaff, with one focusing on how genuine philosophical intuitions strike us, another
focusing on where they come from, and yet another focusing on what we do with them
once they are here. The question is whether any of these ways of setting up the thick-
ness defense gives us sufficient reason to believe at this time that genuine philosophi-
cal intuitions are appropriately immune, hippocratic, and manifest. We’ll take each
in turn.
A. Phenomenological approaches

Let’s start with George Bealer’s version of the phenomenological approach, perhaps
the most famous version of the thickness defense on offer. According to him, genuine
philosophical intuitions have a special phenomenological mark, which he character-
izes in terms like “the appearance of necessity.” One problem that we immediately run
into concerns the conditions of immunity and hippocraticity; Bealer simply hasn’t
given us any reason to think that genuine philosophical intuitions, understood in phe-
nomenological terms, are immune from the kinds of problematic effects that form the
basis of the restrictionist challenge or from other kinds of problematic effects. That
one very particular aspect of a process (its phenomenological feel) is correlated with
another (its immunity to order effects, etc.) is a remarkably strong claim, one that
would require significant empirical work to substantiate.14 And, while we invite propo-
nents of this kind of thickness defense to take up this work, we again want to stress that
the need for this kind of empirical work is itself a victory of sorts for the restrictionist
challenge since part of that challenge is precisely that empirical work is going to be
needed in order to establish the standing of philosophy’s intuition-deploying practices.
The real issue with Bealer’s version of the phenomenological approach, though, is
with manifestability. The problem is that it turns out there is a great deal of disagree-
ment, at the very highest levels of the profession and even among philosophers who
are friendly to armchair methods, about whether any genuine philosophical intuitions
display this special phenomenological mark (see, for example, Sosa 1998; Lynch 2006;
Goldman 2007; Williamson 2007, and Pust 2012). Part of the problem is that it is not
particularly clear what it means for a proposition to have the appearance of necessity
(see, for example, Pust 2012). Does it seem to us that the proposition is necessarily
true or does it seem necessary to us that the proposition is true? But, problems would
remain even if we were able to make clear what it means for a proposition to have
the appearance of necessity. Not all genuine philosophical intuitions, it seems, have
a particular way of seeming, which means that focusing on the fact that some things
strike us one way, while others strike us in other ways, doesn’t provide us a way to
determine which of these things are genuine philosophical intuitions. So, we have a
fairly complete failure of manifestability here. Since there is disagreement from the
armchair about whether any genuine philosophical intuitions have the appearance of
14 Depending on what it means for something to “strike us in a certain way,” there is some work that sug-
gests that some philosophical intuitions are immune from at least some kinds of problematic effects. So, for
example, Jen Wright (2010) has found that confidence is a good indicator of relative immunity from presen-
tation order effects.
necessity, it is impossible to use this phenomenological mark to determine from the

armchair which philosophical intuitions are genuine. This also explains why it is so
hard to provide a definitive evaluation of Bealer’s version of the phenomenological
approach in terms of immunity and hippocraticity: unable to identify the class of cog-
nitions in question, we struggle to be able to evaluate whether cognitions in this class
are or are not likely to be susceptible to various sorts of effects! But, even setting this
aside, the fact that Bealer’s approach runs into problems with manifestability is enough
by itself to make the approach highly unlikely to be successful.
What about Bengson’s version of the phenomenological approach, according to
which genuine philosophical intuitions “strike us in a certain way”? We take Bengson’s
formulation to pick out something like the idea, fairly common in the literature on
intuitions, that intuitions have a phenomenology of a proposition’s seeming to be true,
in a way that feels similar in this regard to perceptual seemings, but not typically sim-
ilar in terms of the sense-modality aspects of perceptual seemings. (For this reason
philosophers often refer to them as intellectual seemings.) Bengson raises the possi-
bility that the participants in experimental-philosophy studies are largely providing
answers that are not driven by any such seemings, and clearly intends this possibility
to make trouble for restrictionists. It is thus an excellent candidate for consideration in
our framework here: Bengson is offering a thickness defense, with the phenomenology
of seeming as his candidate thickening factor. And we can thus score his proposal in
terms of the three core commitments under consideration here.
First, Bengson’s version of the phenomenological approach makes it very likely that
genuine philosophical intuitions will be manifest. We take it to be the case that intui-
tiveness, in the attenuated sense he is considering it, is something that is pretty well
accessible to introspection. It seems to us highly plausible that this phenomenology
can be distinguished reasonably well from the phenomenology of what Bengson calls
“blind or stray answers”—answers that “express hypotheses, guesses, emotional reac-
tions, inferences or other conclusions.”15
Problems arise, however, when we consider the two veritist commitments. Bengson
is claiming that, if ordinary subjects restricted their judgments just to those with the
appropriate phenomenology of seeming, all these unfortunate effects that form the
basis of the restrictionist challenge would disappear. He does offer conjectures about
some interesting possibilities about ways in which experimental-philosophy studies
might be going astray. Our core diagnosis of where he goes wrong, though, is in think-
ing that mere possibilities—even empirically real ones, which we happily grant that
Bengson’s clearly are—can carry much weight in these sorts of debates.16 His claims are
15 Actually, there might be reason to worry that this is too generous since it isn’t clear how Bengson thinks
that we can introspectively peel off “emotional reactions” from seemings more generally, but we don’t want
to press this worry here. See also Huebner (MS) for a response to Bengson that could be understood in our
terms as offering him a low score in terms of manifestability.
16 It might be tempting to think that all Bengson needs to do to rebut the restrictionist challenge is to argue
that the class of responses is not coextensive with the class of intuitions being used by traditional armchair
typically about things that “sometimes” happen; for example, he frames two of his most
central claims as, “Sometimes things strike us a certain way; other times they do not,
even though we may still be able to give an answer to a question about whether things
are that way, when prompted,” and “Sometimes things strike us a certain way, even
though we may answer that things are not that way, but rather some other way, when
prompted” (Bengson 2012, p. 511). These claims, as “sometimes” claims, are clearly
true. But they go no distance towards securing what Bengson actually needs: that these
things happen to a large enough extent and in right distribution to explain away all of
the restrictionist’s favorite experimental findings.
Bengson comes close to seeing this when he examines the plausibility of moving
from the idea that
[AEI] A subject’s prompted answer expresses a subject’s intuition.
To the idea that:
[AEI*] A subject’s prompted answer usually or typically expresses a subject’s
intuition.
Bengson, who thinks that restrictionists are committed to either AEI or AEI*, gives
fairly good reasons for thinking that moving to AEI* isn’t particularly helpful, namely,
that there’s no particular reason, given the experimental designs currently used, to
think that “seeming-less” answers are vastly less common than “seeming-ful” answers.
We do not wish to defend AEI* here—we agree with him that it’s not a good commit-
ment to have. The problem for Bengson is both AEI and AEI* are much stronger com-
mitments than restrictionists actually need. It seems to us that all they need, in order to
respond to Bengson’s version of the thickness defense, is something like this:
[AEIE]
The undesirable effects are not themselves completely localized to
the seeming-less, “blind” answers, but also manifest in seeming-ful,
intuition-based answers.
AEIE is strictly weaker than AEI*. On the one hand, it is likely entailed by AEI*. If
pretty much all the data in the experimental-philosophy surveys were coming from
Bengsonian intuitions, then it would be highly likely that the observed effects are pre-
sent in such intuitions. But, on the other hand, AEIE does not entail AEI*. Suppose,
for example, that the answers themselves reflect a 50–50 mix of intuitions and stray
philosophers. This wouldn’t end the debate, since it would be fair to ask why we should think that the thick
philosophical intuitions used by traditional armchair philosophers are epistemically virtuous, but it might
seem to at least turn aside the restrictionist challenge. The problem is that mere non-coextensiveness is not
enough. Philosophers looking to defend traditional armchair practices need the much stronger claim that
the class of genuine philosophical intuitions is so distinct from the experimentally examined class of cogni-
tions, and distinct in the right ways, that no inductive argument remains from the putative problems of the
latter to any putative problems of the former. (We would like to thank an anonymous reviewer for helpful
suggestions here.)
answers, but the patterning of the unwanted effects is identical in both sets. Then AEI*
would be false, even though AEIE would still be true.
Bengson, unfortunately, doesn’t consider AEIE or anything like it, and so doesn’t
offer any counter-arguments to it.17 And AEIE is, at this time, itself highly likely. For
the dominant view about intuitive cognition in scientific psychology at this time
is that of the “dual process” or “two system” models of reasoning, judgment, and
decision-making.18 One of these systems is commonly described as typically involv-
ing fast, apparently effortless cognition, based more on similarity and associations,
whereas the other system involves slower, effortful, rule-based cognition. On this
model of human cognition, the first system, “System 1” is both the source of (at least
many) intuitive seemings, and also susceptible to many effects of the sorts that restric-
tionists have worried may afflict philosophical intuitions. The fact that these charac-
teristics—intuitiveness and instability—seem very frequently to travel together in the
human mind provides solid support for AEIE.
First, it is just not the case that the psychologists investigating System 1 have gener-
ally been just considering seeming-less cognitions. For example, in a paper based on
his Nobel acceptance address, Daniel Kahneman (2003) observes that, “A core prop-
erty of many intuitive thoughts is that under appropriate circumstances, they come to
mind spontaneously and effortlessly, like percepts” (p. 699), and also frequently analo-
gizes intuitions to sensory impressions. And Jonathan Evans (2003) writes, in an arti-
cle intended as an introduction to some of these issues, that “Dual-process theorists
generally agree that System 1 processes are rapid, parallel and automatic in nature: only
their final product is posted in consciousness” (p. 454, our emphasis). In our experiences
with this literature, these kinds of formulations are fairly common. The workings of
System 1 are not conscious, but their products are consciously available.
Second, System 1 has many traits of exactly the sort that Bengson is hypothesiz-
ing are not involved in genuine intuitions. For starters, System 1 cognition is largely
association-based, and for that reason it is also highly susceptible to contextual fac-
tors. Kahneman (2003) summarizes a large body of his research concerning the strong
presence of framing effects in intuitive cognition, arguing that framing has a big
influence on accessibility, and that “intuitive decisions will be shaped by the factors
17 He does consider a different defense of AEI*, which would involve the claim that “non-intuitive
responses such as hypotheses, guesses, emotional reactions, and inferences are (usually are, typically are)
random, arbitrary, or nonsystematic” (p. 512). He contends that that claim is not likely to be true, and we
are again in agreement with him. But that is very different from—indeed, pointed in the opposite direction
from—the claim that we are considering: that many of the systematic effects manifest in “stray” answers are
also manifest in seemings. This same problem applies to much of the rest of his discussion there, where he
proposes various possible explanations for how non-seemingy cognitions could display the restrictionist
effects; namely, he does not do anything to make probable that seemingy cognitions will not display such
effects as well.
18 Richard Samuels (2009) has recently argued that they are perhaps better viewed as two families of a
much larger number of systems, rather than two systems proper; see also, e.g., Gilbert (1999). Nothing in this
discussion turns on that distinction.
that determine the accessibility of different features of the situation. Highly accessi-
ble features will influence decisions, while features of low accessibility will be largely
ignored. Unfortunately, there is no reason to believe that the most accessible features
are also the most relevant to a good decision” (p. 710). (See also, e.g., Morewedge and
Kahneman (2010, p. 437) for a particular discussion of how processes of “associative
coherence” lead to both anchoring and framing effects.) We would conjecture that the
reliance of System 1 on association, as well as its use on learned prototypes of catego-
ries (see Kahneman 2003, p. 712ff), would also be a potential source of demographic
variation: members of different cultural groups may well form subtly distinct asso-
ciations, as well as form prototypes from somewhat different stockpiles of individual
cases. Finally, many canonical descriptions of the processes in System 1 also include
that they are sensitive to emotion, perhaps even including an “affect heuristic” (Slovic
et al. 2002).19
In short, according to this widely held view, the outputs of System 1 processes are
typically seeming-ful, and also typically sensitive to exactly those sorts of factors that
restrictionists are worried that philosophical intuitions may be sensitive to as well.
There just is not much reason at this time to think that a cognitive state or episode’s
having the phenomenology of seeming-ness correlates in any useful way with being
shielded from the influence of context, demography, or affect. Accordingly, restric-
tionists do not need at this time to lose any sleep about the status of commitments
like AEIE.
It’s important to see why it is things like AEIE, and not Bengson’s AEI or AEI*, that
are at stake in this debate. Here’s a totally made-up case that might help to illustrate the
basic problem. Suppose someone wanted to run with a version of the thickness defense
in which the thickening factor is being intuited on a Sunday. Adherents to such “phil-
osophical sabbaticism” (or, perhaps, “seventh-day rationalism”) could make a very
strong case that very little of the existing experimental philosophy work has examined
what would be, by their lights, genuine intuitional evidence. After all, a very large pro-
portion of existing restrictionist work was done in a classroom setting, during or after
an actual meeting of a university class, and it is thus enormously probable that none
of that data was collected on a Sunday. In this regard, the sabbaticists are in an even
better situation, in terms of the structure of Bengson’s argument, than Bengson him-
self is. But we hope it is utterly obvious why sabbaticism is a complete non-starter as
a response to restrictionism: there’s no reason at all to think that our cognition works
differently in any particular way on Sunday than it does on any other day of the week.
In terms of our evaluative framework, sabbaticism would do great in terms of mani-
festability—everyone can tell pretty easily what day of the week it is!—but would be an
abject failure in terms of immunity. The moral here is that the success of a version of
19 Bengson at one point in his paper contrasts heuristic and intuitive cognition (2012, §5.4). This seems not
to be the view held by most researchers in this area, in that many heuristics, especially those not deliberately
deployed, are located in System 1.
the thickness defense does not depend solely on whether the allegedly genuine intui-
tions turn out to be in some way or other different from whatever is being studied by
the restrictionist experimental philosophers. Instead, specific versions of the thickness
defense can be successful only if they identify some factor that distinguishes the alleg-
edly genuine intuitions, and makes it likely that they are psychologically distinct in a
way that renders them sufficiently shielded from the sorts of errors that form the basis
of the restrictionist challenge.
Now, sabbaticism is offered as an illustration of a problem with Bengson’s argument,
but we do want to stress that it is not an analogy to Bengson’s argument. Sabbaticism is
silly; Bengson’s proposal is most definitely not. In fact, it seems to us to be a legitimate
scientific hypothesis, one that is absolutely worthy of further investigation, and we
hope that he will pursue such an investigation. There may well be yet-unexplored divi-
sions within System 1, such that some particular sorts of effects like the ones proposed
by restrictionists do not, as a matter of fact, manifest in seeming-ful cognition. The
current state of the literature does not, we think, rule this hypothesis out. But Bengson
has failed, we fear, to recognize that it is a hypothesis that cannot do the dialectical
work he wants it to do, unless and until there is some substantial evidence that it is
at least somewhat likely to be true. As we noted above, Bengson tends in his paper
to frame matters in terms of possibilities. But once we start actually looking at prob-
abilities, his version of the thickness defense looks rather implausible, although, again,
eminently worthy of further investigation. It is, we think, an open, although improb-
able hypothesis at this time.
B. Etiological approaches

Etiological approaches would distinguish genuine from ersatz intuitions by their ori-
gins. There’s really only one extant flavor of this kind of approach at this time, and it
is the view like Kirk Ludwig’s described earlier, that genuine philosophical intuitions
are intuitions made only on the basis of conceptual competence. To see how concep-
tual competence versions of the thickness defense fair, let’s start with the immunity
condition.20 It turns out that conceptual-competence approaches fair somewhat bet-
ter vis-à-vis the immunity condition than do phenomenological approaches. After
all, it is plausible to think (if not yet empirically demonstrated) that issues of concep-
tual performance might be responsible for the kinds of problematic effects (e.g., order
and framing effects) that form the basis of the restrictionist challenge. And, nothing
that we know of from psychology or cognitive science seems to directly undermine
the plausibility of this suggestion. But, it is important to keep in mind that whether
20 There might be another reason for thinking that a conceptual-competence approach has some trac-
tion against the restrictionist challenge, namely, that philosophers in particular gain some special sort of
conceptual competence by virtue of their philosophical training; see n.10 above, on the ‘expertise defense’.
At the end of the day, we think that the expertise defense faces highly similar problems to the thickness
defense: they both rely on strong empirical claims for which there is currently scant evidence.
or not conceptual-competence approaches will sufficiently immunize genuine philo-

sophical intuitions depends crucially on what theory of concepts we are working with.
Many theories attribute structures to the mind that are themselves contextually sensi-
tive and widely variable across both individuals and populations—for example, theo-
ries according to which our concepts contain representations of exemplars (see, e.g.,
Smith 1999) or theories according to which sensitivity to contextual factors is practi-
cally a defining characteristic (see, e.g., the discussion of proxytypes in Prinz 2002).
Of course, philosophers might have an alternative theory of concepts in mind, per-
haps one that would impose context-free sets of necessary and sufficient conditions
on category membership. But, it is important to point out both that these kinds of
theories haven’t proved particularly popular to folks working on concepts in psychol-
ogy or cognitive science and that the truth of such theories will depend on how the
literature on concepts plays out. As such, if philosophers want to defend their armchair
practices by appealing to this kind of theory of concepts, they need to recognize that
doing so involves a substantial empirical commitment about the nature and structure
of concepts, one that doesn’t have much empirical support from psychology or cogni-
tive science.
Along similar lines, there is reason to worry that conceptual-competence approaches
aren’t going to be particularly hippocratic, either. To see why, it is important to note
that in order for these approaches to even get off of the ground as a defense of our
intuition-deploying practices, it must be that conceptual competences have some sort
of rich structure; there is no way for a conceptual atomist to even begin to run this
line of argument. Proponents of this view will require concepts to have theory-like
structures. But if some sort of internalized theory determines conceptual content, then
we have good reason to worry about the effects of theory contamination on our con-
ceptual judgments. If the conceptual theories are acquired or replaced by tacit learn-
ing processes (cf. Reber 1993) or by the modularization of explicitly formed theories
(cf. Karmiloff-Smith 1993), then we would face a significant problem for philosophy’s
intuition-deploying practices. We advance philosophical theories on the basis of their
ability to explain our philosophical intuitions, defend their truth on the basis of their
overall agreement with our philosophical intuitions, and justify our philosophical
beliefs on the basis of their accordance with our philosophical intuitions. If those same
theories play a role in shaping our conceptual judgments—if we have our intuitions
because of our theories, and not the other way around—we risk a most vicious kind
of epistemic circularity (for discussion, see, e.g., Cummins 1998). Because this risk is
rather more hypothetical, though, given that these philosophers haven’t even put a real
conceptual-competence account of the right sort on the table, it is probably appropri-
ate to err on the side of charity in our evaluation, and so we are willing to say that it is
an open hypothesis whether conceptual-competence approaches are going to be par-
ticularly hippocratic.
Things get worse. Even if genuine philosophical intuitions were to come out as both
immune and hippocratic on the conceptual-competence approach, there would still
be reason to worry that the conceptual-competence approach—and indeed etiological

approaches more generally—won’t be able to render genuine philosophical intuitions
particularly manifest. It is going to be quite hard, indeed maybe even impossible, to
introspectively determine when our own philosophical intuitions are the product of
our conceptual competence, and when instead they have been influenced by other fac-
tors. For some other factors, especially pragmatics, we do have at least some tests that
can perhaps be performed in the armchair. But for many others we don’t, and indeed
many philosophers (e.g., Henderson and Horgan 2001; Lynch 2006) have particularly
emphasized just how opaque to us the origins of our intuitions are, in a manner well
consistent with general cognitive psychological doubts (e.g., Nisbett and Wilson 1977;
Wilson 2002) about the reliability of introspection to get at origins of or processes
underlying our actions and thoughts. And we also noted above how research on System
1 cognition generally concurs with this view about the unavailability of the underlying
processes to consciousness. In short, we are just not at all likely to be well suited to
tell—especially with only the limited resources of the armchair—which of our philo-
sophical intuitions are genuine. All of this makes it rather unlikely that genuine philo-
sophical intuitions are going to be manifest on the conceptual-competence approach.
We would also caution that these same reasons would likely apply to any other etio-
logical approach to the thickness defense, were someone to try to formulate one.
C. Methodological Approaches

According to methodological approaches, genuine philosophical intuitions are the
products of careful philosophical reflection.21 To see how this approach fares, let’s
start with manifestability. We take it that the condition of manifestability is in fairly
good shape here, in that reflection is an armchair-pursuable activity par excellence.22
Nonetheless, we will note two worries, without pressing them further. First, propo-
nents of any reflection-based approach owe us some sort of story about how much
reflection is enough, and how we can tell. Second, although one can engage in reflec-
tion from the armchair, it may sometimes be the case that correction can only come
by reflecting on information gathered from outside the armchair. For example, no
amount of reflecting on our intuitions about space and time was ever going to teach us
that space–time is not Euclidean! It might be that reflection can indeed help improve
people’s intuitive judgments, without it being the case that armchair reflection is what
is needed. Nonetheless, we are willing to grant that it is fairly likely that genuine philo-
sophical intuitions will be manifest on this approach.
The problem with reflection-based approaches is that they score rather more poorly
on both immunity and hippocraticity.23 Although it is at least plausible that the greater
21 In principle there could be other methodological proposals, such as the role of peer review, or the appli-
cations of formal tools, and so on. But the method of reflection is the only current contender in the literature.
22 It is important to note that some ways of incorporating empirical evidence are consistent with what
several philosophers (such as Timothy Williamson and Daniel Nolan) consider to be armchair practices.
23 Hilary Kornblith (2002, 2010) nicely describes further problems associated with reflection.
reflection would help with at least some of the effects (see, e.g., Smith and Levin 1996,
on need for cognition and framing), there is no reason at this time to think that it will
help with all of them. Depending on the exact source of the demographic variations,
reflection might help; in particular, if it turns out that the demographic differences
are due to differences in motivation to reflect in the first place, then making every-
one reflect more might bring everyone to a convergent solution (for discussion of this
point, see Nagel (2012)). But, similar to what we saw with Bengson’s proposal above, all
we have are a string of possibilities, currently with at best very little evidence on their
behalf. Moreover, there is also some recent evidence suggesting that, in at least some
cases, further reflection does not make some effects go away.24 At best, then, the reflec-
tion version of the thickness defense makes it rather unlikely that genuine philosophi-
cal intuitions will be appropriately immune.25
Furthermore, work in psychology and cognitive science suggests that for some influ-
ences and under some conditions, reflection might intensify their effects. For example,
Petty et al. (2001) found that although more reflective subjects, when evaluating evi-
dence in a “chunked” setting (that is, where the evidence comes in distinct packages),
were less likely to show recency effects, they were more likely to show primacy effects.
And LeBoeuf and Shafir (2003) report that:
Interestingly, research on persuasion suggests that high-NC participants may sometimes be
more affected by framing than their low-NC counterparts. For example, high-NC, but not
low-NC, participants find arguments framed to match their moods more persuasive than argu-
ments not so framed (Wegener et al., 1994). Similarly, negative frames are more persuasive than
positive frames in encouraging detection-related health behaviors among those high, but not
those low, in NC (Rothman et al., 1999). Note also that verbalization during decision making can
hurt performance (Schooler & Melcher, 1995; Schooler et al., 1993), and that justification provi-
sion can lead to the deterioration of decisions relative to experts’ standards (Wilson & Schooler,
1991). Taken together, these studies paint a fairly pessimistic view of the power of extra thought
to improve the quality of many types of decisions.” (p. 89)26
24 See, e.g., Weinberg et al. 2012. See also Huebner (MS) for some nice discussion of the way in which moti-
vated cognition can end up leading reflection or argumentation to exacerbate confirmation biases.
25 This point can be put even more forcefully. Suppose that we are particularly generous to reflection-based
approaches, including the kind of approaches mentioned in n.20, and allowed that for each source of restric-
tionist troublemaking, there was an even slightly better than even chance that a reflection-based proposal
was true. (This would in fact be very generous since each proposal would involve making a pretty strong
empirical claim and our priors should start out well below 0.5.) The problem is that for reflection-based
approaches on the whole to succeed, it is not enough that merely some subset of restrictionist troubles goes
away. Reflection-based approaches need all of these troubles to go away, or at least, near enough. But then
the probability that this kind of defense works becomes something like (0.6)n where n is the number of doc-
umented troublemakers. We know that the number of potential troublemakers is at least four, and it is surely
implausible to think that we’ve actually already exhausted the number of troublemakers. But this means that
the probability that this kind of defense is going to work is not going to be very high.
26 ‘NC’ refers to a person’s need for cognition, a measure that corresponds to the likelihood that a person’s
intrinsic motivation to give a great deal of care and attention to cognitive tasks with which they are pre-
sented. See Cacioppo and Petty (1982).
Also relevant to the evaluation of immunity is their summary claim that, while “deeper
thinking is likely to prove critical in avoiding some errors, it is unlikely to play a signifi-
cant role in avoiding others” (p. 90). This is entirely consistent with the restrictionists’
general argument: determining just where intuitions are susceptible to what sorts of
effects, and where that susceptibility can be reduced or even eliminated—and where
it may instead be exacerbated—is a complicated question for empirical science. The
armchair simply lacks the resources to sort the signal from the noise in the tangled
instrument that is intuitive human judgment. Given a fair amount of evidence indicat-
ing that reflection can occasionally induce epistemic harms, it is unlikely that genuine
philosophical intuitions are hippocratic on the reflection version of the methodologi-
cal version of the thickness defense.
10.5. Conclusion
Let’s take stock. Philosophers need to recognize that that this debate is one that has to
play out not in terms of mere possibilities, even legitimately real empirical possibilities,
but in terms of probabilities, and what the current state of scientific evidence does—
and largely doesn’t—make likely. Every move, both for and against the armchair, has to
be measured against what commitments it incurs, and how well those commitments
actually seem to fare. And it turns out that none of the most popular ways of setting
up the thickness defense end up satisfying all three empirical conditions of adequacy
set out for a successful response to the restrictionist challenge. Among the phenom-
enological proposals, Bealer’s approach fails most clearly because of its poor score on
manifestability, whereas Bengson’s succeeds better in that regard, while scoring as at
best open-but-implausible on the veritist commitments. The conceptualist version of
the etiological approach scores middling at best on the veritist conditions, while com-
pletely failing the manifestability condition. And, the reflectionist version of the meth-
odological approaches also scores poorly on both veritist commitments. See the chart
below for our final scorecard for each version of the thickness defense considered here.
Now, surely, it is possible that some other specific version of the thickness defense
could fare better. Nonetheless, there is a discernible pattern in the distinct shortcom-
ings and limited-at-best successes of the proposals that are on the table. We would close
with some speculation as to a good reason why there might in fact be a more general
problem that will beset any thickist attempt to meet the restrictionist challenge from
the armchair. It turns out that there is a tension between the condition of immunity and
the conditions of hippocraticity and manifestability that might explain why the three
conditions can’t be simultaneously realized. In order to satisfy immunity, there needs
to be a stark difference between genuine philosophical intuitions on the one hand, and
ersatz philosophical intuitions on the other—a difference that explains both why faux
philosophical intuitions are susceptible to the kinds of problematic effects that form
the basis of the restrictionist challenge, and why genuine philosophical intuitions are
Table 10.1 Scorecard for different versions of the thickness defense
Immunity Hippocraticity Manifestability Overall grade
Phenomenological Open, but Open, but Highly Highly

(Bealerian) unlikely unlikely unlikely unlikely
Phenomenological Open, but Open, but Highly likely Open, but
(Bengsonian) unlikely unlikely unlikely
Etiological Open, but Open Highly Highly
(Conceptual) unlikely unlikely unlikely
Methodological Highly Highly Highly likely Highly
(Reflection) unlikely unlikely unlikely
not. But the more genuine philosophical intuitions are a different cognitive creature
altogether, the more likely it is that they will have their own peculiar epistemic pathol-
ogies, and thus violate hippocraticity. Additionally, in order to satisfy immunity, the
difference between genuine philosophical intuitions and faux philosophical intuitions
should not be one that folks can easily identify. After all, if folks could easily identify
the difference between genuine and ersatz philosophical intuitions, then there is less
reason to suppose that faux philosophical intuitions are the ones showing up in the
experimental studies that form the basis of the restrictionist challenge. But, the more
opaque we make the distinction between genuine philosophical intuitions and faux
philosophical intuitions the harder it will be to satisfy manifestability, as we saw with
the appeal to conceptual competence above.
Where does this leave us? It seems that philosophers face a dilemma on a sliding
scale: the thinner their conception of philosophical intuition, the more clearly they are
challenged by recent work in experimental philosophy; the thicker their conception
of philosophical intuition, the more likely they are to incur different kinds of meth-
odological problems and the harder it will be to tell when anyone is doing philoso-
phy correctly. It is very unlikely that a course through these contesting forces could
be navigable from the armchair. If there’s a useful distinction to be deployed between
genuine and ersatz intuitions, it’s going to turn out to be one that we will need not just
philosophical but also substantive scientific work to chart out.
References
Alexander, J. (2010). “Is Experimental Philosophy Philosophically Significant?,” Philosophical
——. (2012). Experimental Philosophy: An Introduction. Cambridge: Polity Press.
——., and Weinberg, J. (2007). “Analytic Epistemology and Experimental Philosophy,”
Philosophy Compass 2, pp. 56–80.
——. (Forthcoming). “The ‘Unreliability’ of Epistemic Intuitions,” in E. Machery and E. O’Neill
(eds), Current Controversies in Experimental Philosophy. London: Routledge.
Bealer, G. (1998). “Intuition and the Autonomy of Philosophy,” in M. DePaul and W. Ramsey
(eds), Rethinking Intuition: The Psychology of Intuition and its Role in Philosophical Inquiry.
Lanham, MD: Rowman & Littlefield, pp. 201–40.
Bengson, J. (2012). “Experimental Attacks on Intuitions and Answers,” Philosophy and
Buckwalter, W., and Stich, S. (2014). “Gender and Philosophical Intuition,” in J. Knobe and S.
Nichols (eds), Experimental Philosophy, Volume 2. Oxford: Oxford University Press, pp. 307–46.
Cacioppo, J., and Petty, R. (1982). “The Need for Cognition,” Journal of Personality and Social
Cappelen, H. (2012). Philosophy without Intuitions. Oxford: Oxford University Press.
Cullen, S. (2010). “Survey-Driven Romanticism,” Review of Philosophy and Psychology 1, pp.
275–96.
Cummins, R. (1998). “Reflection on Reflective Equilibrium,” In M. DePaul and W. Ramsey (eds),
Rethinking Intuition: The Psychology of Intuition and its Role in Philosophical Inquiry. Lanham,
MD: Rowman & Littlefield, pp. 113–28.
Devitt, M. (1994). “The Methodology of Naturalistic Semantics,” Journal of Philosophy 91, pp.
545–72.
Elgin, C. (1996). Considered Judgment. Princeton, NJ: Princeton University Press.
Evans, J. (2003). “In Two Minds: Dual-Process Accounts of Reasoning,” Trends in Cognitive
Sciences 7, pp. 454–59.
Gilbert, D. T. (1999). “What the Mind’s Not,” in S. Chaiken and Y. Trope (eds), Dual Process
Theories in Social Psychology. New York: Guilford.
Gonnerman, C. and Weinberg, J. (Forthcoming). “Why Messy Minds Make for Messier
Armchairs: Goldman and The Epistemology of Categorizer Pluralism,” in H. Kornblith (ed.),
Goldman and His Critics.
Gutting, G. (1998). “Rethinking Intuition: A Historical and Metaphilosophical Account,” in
M. DePaul and W. Ramsey (eds), Rethinking Intuition. Lanham, MD: Rowman & Littlefield,
pp. 3–16.
Hales, S. (2006). Relativism and the Foundations of Philosophy. Cambridge, MA: MIT Press.
Henderson, D., and Horgan, T. (2001). “The A Priori Isn’t All It Is Cracked Up To Be, But It Is
Something,” Philosophical Topics 29, pp. 219–50.
Huebner, B. (MS). “Constructing Philosophical Intuitions.”
Ichikawa, J. J. (MS). “Intuitions and Begging the Question,” available at: <http://www.jonathani-
chikawa.net/Home/research/wip/ibq.pdf> [last accessed October 9, 2013].
Kahneman, D. (2003). “A Perspective on Judgment and Choice,” American Psychologist 58, pp.
697–720.
Karmiloff-Smith, A. (1993). “Beyond Modularity: A Developmental Perspective on Cognitive
Science,” European Journal of Disorders of Communication 28, pp. 95–105.
Kauppinen, A. (2007). “The Rise and Fall of Experimental Philosophy,” Philosophical
Explorations 10, pp. 95–118.
Kornblith, H. (1998). “The Role of Intuitions in Philosophical Inquiry: An Account with No

Unnatural Ingredients,” in M. DePaul and W. Ramsey (eds), Rethinking Intuition. Lanham,
——. (2002). Knowledge and Its Place in Nature. Oxford: Oxford University Press.
——. (2010). “What Reflective Endorsement Cannot Do,” Philosophy and Phenomenological
Research 80, pp. 1–19.
LeBoeuf, R., and Shafir, E. (2003). “Deep Thoughts and Shallow Frames: On the Susceptibility to
Framing Effects,” Journal of Behavioral Decision Making 16, pp. 77–92.
Levin, J. (2005). “The Evidential Status of Philosophical Intuition,” Philosophical Studies 121, pp.
193–224.
Lewis, D. (1983). Philosophical Papers, Volume I. New York: Oxford University Press.
Ludwig, K. (2007). “The Epistemology of Thought Experiments: First Person versus Third
Person Approaches,” Midwest Studies in Philosophy 31, pp. 128–59.
Lynch, M. (2006). “Trusting Intuitions,” in P. Greenough and M. Lynch (eds), Truth and Realism.
Morewedge, C., and Kahneman, D. (2010). “Associative Processes in Intuitive Judgment,” Trends
in Cognitive Sciences 14, pp. 435–40.
Nagel, J. (2012). “Intuitions and Experiments: A Defense of the Case Method in Epistemology,”
Nichols, S., and Knobe, J. (2007). “Moral Responsibility and Determinism: The Cognitive
Science of Folk Intuitions,” Noûs 41, pp. 663–85.
Nisbett, R., and Wilson, T. (1977). “Telling More Than We Can Know: Verbal Reports on Mental
Processes,” Psychological Review 84, pp. 231–59.
Petty, R., Tormala, Z., Hawkins, C., and Wegener, D. (2001). “Motivated to Think and Order
Effects in Persuasion: The Moderating Role of Chunking,” Personality and Social Psychology
Bulletin 27, pp. 332–44.
Prinz, J. (2002). Furnishing the Mind: Concepts and their Perceptual Basis. Cambridge,
MA: MIT Press.
Pust, J. (2012). “Intuition,” in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy, available
at: <http://plato.stanford.edu/archives/win2012/entries/intuition/> [last accessed October
7, 2013].
Reber, A. (1993). Implicit Learning and Tacit Knowledge: An Essay on the Cognitive Unconscious.
Rothman, A., Martino, S., Bedell, B., Detweiler, J., and Salovey, P. (1999). “The Systematic
Influence of Gain- and Loss-framed Messages on Interest in and Use of Different Types of
Health Behavior,” Personality and Social Psychology Bulletin 25, pp. 1355–69.
Samuels, R. (2009). “The Magic Number Two, Plus or Minus: Dual Process Theory as a Theory of
Cognitive Kinds,” in K. Frankish and J. Evans (eds), In Two Minds: Dual Process and Beyond.
Schooler, J., and Melcher, J. (1995). “The Ineffability of Insight,” in S. Smith and T. Ward (eds),
The Creative Cognition Approach. Cambridge, MA: MIT Press, pp. 97–133.
——., Ohlsson, S., and Brooks, K. (1993). “Thoughts Beyond Words: When Language
Overshadows Insight,” Journal of Experimental Psychology: General 122, pp. 166–83.
Slovic, P., Finucane, M., Peters, E., and MacGregor, D. (2002). “The Affect Heuristic,” in
T. Gilovich, D. Giffin, and D. Kahneman (eds), Heuristics and Biases: The Psychology of
Intuitive Judgment. Cambridge: Cambridge University Press, pp. 397–420.
Smith, E. (1999). “The Exemplar View,” in E. Margolis and S. Laurence (eds), Concepts: Core
Readings. Cambridge, MA: MIT Press.
Smith, S., and Levin, I. (1996). “Need For Cognition and Choice Framing Effects,” Journal of
Behavioral Decision Making 9, pp. 283–90.
Sosa, E. (1998). “Minimal Intuition,” in M. DePaul and W. Ramsey (eds), Rethinking Intuition: The
Psychology of Intuition and its Role in Philosophical Inquiry. Lanham, MD: Rowman &
Littlefield, pp. 257–70.
76, pp. 138–55.
van Inwagen, P. (1997). “Materialism and the Psychological—Continuity Account of Personal
Identity,” Philosophical Perspectives 11, pp. 305–19.
Wegener, D., Petty, R., and Klein, D. (1994). “Effects of Mood on High Elaboration Attitude
Change: The Mediating Role of Likelihood Judgments,” European Journal of Social Psychology
24, pp. 25–43.
Weinberg, J. (2007). “How To Challenge Intuitions Empirically Without Risking Skepticism,”
——., Gonnerman, C., Buckner, C., and Alexander, J. (2010). “Are Philosophers Expert
Intuiters?,” Philosophical Psychology 23, pp. 331–55.
——., Alexander, J., Gonnerman, C., and Reuter, S. (2012). “Restriction and Reflection: Challenge
Deflected, or Simply Redirected,” The Monist 95, pp. 200–22.
——., Nichols, S., and Stich, S. (2001). “Normativity and Epistemic Intuitions,” Philosophical
Topics 29, pp. 429–60.
Williamson, T. (2007). The Philosophy of Philosophy. Malden, MA and Oxford: Blackwell.
——. (2011). “Philosophical Expertise and the Burden of Proof,” Metaphilosophy 42, pp.
215–29.
Wilson, T. (2002). Strangers to Ourselves: Discovering the Adaptive Unconscious. Cambridge,
MA: Harvard University Press.
——., and Schooler, J. (1991). “Thinking Too Much: Introspection Can Reduce the Quality of
Preferences and Decisions,” Journal of Personality and Social Psychology 60, pp. 181–92.
Wright, J. (2010). “On Intuitional Stability: The Clear, the Strong, and the Paradigmatic,”
Cognition 115, pp. 419–503.
11
Sceptical Intuitions
Duncan Pritchard†
11.1. Introductory Remarks

I will be approaching the topic of intuitions in a tangential fashion. My initial focus will be
on the debate regarding the problem of philosophical scepticism, and the role that intui-
tions play in that debate. In particular, I will be looking at the specific way in which Barry
Stroud motivates the radical sceptical challenge, and considering the superficially similar,
but fundamentally different, responses to the radical sceptical challenge that can be found
in the work of J. L. Austin and Ludwig Wittgenstein. As we will see, this debate demon-
strates something interesting about the nature of intuitions and the role that they can play
in philosophical inquiry. More specifically, I will be arguing that it demonstrates that we
need to think of the philosophical use of intuitions in such a way that it can require a signif-
icant level of expertise. Finally, I will be contending that this conception of the philosophi-
cal use of intuitions has a bearing on the negative programme in experimental philosophy.
11.2. The Radical Sceptical Paradox

Radical scepticism is one of the age-old problems of philosophy. While it comes in a
number of forms, we will here for simplicity’s sake focus on how this problem is stand-
ardly understood in the contemporary literature.1 This version, in essence, has the fol-
lowing general structure:
† Thanks to David Bloor, Cameron Boult, Adam Carter, Stew Cohen, Axel Gelfert, Georgi Gardiner,
Mikkel Gerken, Sandy Goldberg, Alvin Goldman, Peter Graham, John Greco, Allan Hazlett, David
Henderson, Jesper Kallestrup, Chris Kelp, Klemens Kappel, Hilary Kornblith, Joseph Kuntz, Martin Kusch,
Mike Lynch, Ram Neta, Chris Ranalli, Shane Ryan, Ernie Sosa, and Stefan Tolksdorf for helpful discussion
of the topics covered in this paper. Special thanks to Darrell Rowbottom, and to two anonymous referees for
Oxford University Press, who provided detailed comments on an earlier incarnation. This chapter was for
the most part written while I was in receipt of a Philip Leverhulme Prize.
1 See Pritchard (2002) for a survey of recent work on the radical sceptical problem, with particular focus
on this specific formulation of that problem.
214 Duncan Pritchard
The Template Radical Sceptical Argument

(SP1) I am unable to know that a certain radical sceptical hypothesis doesn’t obtain.
(SP2) But I need to be able to know that this radical sceptical hypothesis doesn’t
obtain if I am to have much of the everyday knowledge which I typically
attribute to myself.
(SC) So I am unable to have much of the everyday knowledge which I typically
attribute to myself.
We can get a better sense of what this type of scepticism involves by plugging in a
particular radical sceptical hypothesis and seeing what effect it has on what is puta-
tively an epistemically unimpeachable item of ordinary knowledge. For our sceptical
hypothesis, we will take the famous ‘brain-in-a-vat’ (‘BIV’) scenario, wherein the agent
is ‘fed’ experiences as if she is living a normal life (seeing friends, playing sport, going
to work, and so on) when in fact she is a disembodied brain floating in a vat of nutrients
that is hooked-up to supercomputers which are generating these fake experiences.2
For the putatively epistemically unimpeachable item of ordinary knowledge we will
take an agent’s knowledge, in normal circumstances, that she has hands. Clearly, if, in
normal circumstances, one fails to know a proposition like this, then there isn’t much
that one knows.
We are now in a position to run a specific sceptical argument based on the
template above:
The BIV Radical Sceptical Argument
(SP1*) I am unable to know that I am not a BIV.
(SP2*) But I need to be able to know that I am not a BIV if I am to know that
I have hands.
(SC*) So I am unable to know that I have hands.
Although (SC*) falls short of the radical sceptical conclusion, given the fact that know-
ing that one has hands is held to be in ordinary circumstances the kind of thing that
one must know if one knows anything of substance about the world, with the conclu-
sion of this specific radical sceptical argument we are only a short logical step from a
more general radical sceptical conclusion.
There are several features of this formulation of the radical sceptical argument
which are deserving of comment, not least the extent to which this broadly ‘Cartesian’
way of thinking about radical scepticism (due to how it essentially trades on radical
sceptical hypotheses) differs from other conceptions of the radical sceptical problem.3
2 Cf. Putnam (1981, ch. 1). For an overview of some of the philosophical issues with regard to this sceptical
scenario, see Brueckner (2004).
3 Such as Pyrrhonian scepticism, which, among other things, differs from the sceptical argument just
given in that it is not even expressed in the form of an argument. I discuss Pyrrhonian scepticism, and
Sceptical Intuitions 215
Our current concern, however, is not with the details of this formulation of the radical
sceptical problem, but rather the fact that this formulation has been held by many to
capture the sense in which the problem of radical scepticism—like other ‘deep’ prob-
lems of philosophy, such as the problem of vagueness or the problem of free will—
constitutes a philosophical paradox. That is, it is held to be a philosophical problem
which arises out of a deep tension within our own naturally derived concepts, such
that the only way of responding to this difficulty is by arguing for claims which are to
some extent counterintuitive. This is not to say, of course, that the problem of radical
scepticism, qua paradox, is beyond solution. But paradoxes, if they are bona fide at any
rate (an issue that we will come back to), are more difficult to resolve than ordinary
philosophical problems.
In terms of the contemporary debate about scepticism, it is Barry Stroud’s seminal
book, The Significance of Philosophical Scepticism, which brought home to the phil-
osophical community the importance of thinking of radical scepticism in this way.
Immediately prior to Stroud’s book, it was common within philosophy to treat the
problem of radical scepticism as clearly a pseudo-problem, and hence not as repre-
senting a paradox at all. But what Stroud so convincingly argued in this book was that
the standard responses to radical scepticism of the time—he considered anti-sceptical
approaches from such diverse philosophical quarters as ordinary language philoso-
phy, naturalized epistemology, and verificationism—fail precisely because they don’t
take into the account the way in which scepticism seems to naturally fall out of our
ordinary ways of epistemic evaluation. As Stroud (1984, pp. 81–2) famously put it, scep-
ticism arises out of “platitudes that we would all accept.” 4
Now this might initially sound puzzling, since the argument given above does not
seem to entirely consist of mere platitudes. Indeed, it seems to consist, at least in part,
of controversial claims which are at odds with our ordinary practices of epistemic
evaluation.
Take (SP1) first. While many philosophers would grant that intuition points to the
truth of this proposition, it is certainly not a claim that one would expect to find a
non-philosopher making. After all, sceptical hypotheses are simply not considered
in normal contexts of epistemic evaluation. The possibility that one might be a BIV
its relationship to the broadly ‘Cartesian’ scepticism that is our focus, elsewhere. See especially Pritchard
(2000; 2005, ch. 8; 2011b). See also n.6, where I briefly remark on the very different sceptical problem of the
criterion.
4 It is important to distinguish this key element in Stroud’s writings on radical scepticism from a second
strand in his thinking on this topic which is superficially similar and in some respects closely related. This
second strand is Stroud’s metaepistemological scepticism about the viability of adequately conducting the
epistemological project. Although there are some important points of contact between Stroud’s motivation
for offering this metaepistemological scepticism and his motivation for defending the thesis that scepticism
is a genuine paradox, since they are distinct problems they require distinct treatments, and I will not be
discussing this second strand in Stroud’s work on radical scepticism here. For two key exchanges regarding
Stroud’s metaepistemological scepticism, see Sosa (1994) and Stroud (1994); and Cassam (2009) and Stroud
(2009). See also Stroud (1989).
undergoing the relevant deception is just not ever taken seriously in quotidian epis-
temic contexts. So even if one can convince someone that (SP1) is true, it will be a fur-
ther matter to show that it falls out of our ordinary practices of epistemic evaluation.
The situation with (SP2) is even more problematic on this score.5 This is because at
first pass not only is it not intuitively true, but also—worse—it actually seems to be
intuitively false. For on the face of it, it just seems plainly bizarre to hold that in order
to know the kind of propositions that we typically ascribe to ourselves in normal con-
texts that we should be able to know the denials of radical sceptical hypotheses. Why
should I have to know that I am not a BIV in order to know something as mundane as
that I have hands, particularly when I can see my hands right in front of me? That this
premise should be regarded as a ‘platitude’ therefore seems clearly misguided.
And note that it doesn’t really make any difference whether one focuses on the gen-
eral form of the sceptical argument (the ‘template’ radical sceptical argument, as it is
described above) as opposed to a particular version of that argument, such as the one
which appeals to the BIV sceptical hypothesis. What makes (SP1) dubious applies just
as well to (SP1*), and the same goes for the pairing of (SP2) and (SP2*).
It is unsurprising, then, that immediately prior to Stroud’s book the problem of radi-
cal scepticism—even when it was taken seriously as a bona fide philosophical problem
(which it often wasn’t)—wasn’t regarded as a paradox, in the sense that it could be
generated from apparently uncontentious premises.6 As a result, the radical sceptical
problem, although it has of course been taken very seriously during various periods in
the history of philosophy, was not regarded as a pressing or important philosophical
difficultly during this time.
Stroud countered this consensus against the sceptical paradox by carefully show-
ing how this problem is in fact generated by our ordinary epistemic practices, even if
there is nothing within our ordinary epistemic practices which would appear to sus-
tain premises like (SP1) and (SP2). He argued that while it is undoubtedly true that in
ordinary contexts of epistemic evaluation we do not even consider radical sceptical
error-possibilities, much less regard them as having a status such that we need to be
able to know them to be false if we are to possess the kind of knowledge which we typi-
cally ascribe to ourselves in these contexts, it is nonetheless true that the radical scepti-
cal problem arises out of our ordinary epistemic practices.
Perhaps the best way to make sense of Stroud’s claim in this regard is to consider
how he replies to J. L. Austin’s famous attack on radical scepticism in his seminal
5 Note that Stroud in effect appeals to a different second premise in setting up his version of the sceptical
argument; one that is logically stronger and which Stroud argues is endorsed by Descartes in his Meditations.
See Stroud (1984, ch. 1). Nothing hangs on this difference, so for our purposes we can focus on the weaker,
and more widely endorsed, formulation of the second premise.
6 I am here bracketing Chisholm’s important work on the problem of the criterion (e.g., Chisholm 1973).
While this problem is undoubtedly a sceptical paradox in the relevant sense—and was regarded as such, by
Chisholm and some of his contemporaries—it is very different from the kind of sceptical paradox currently
under consideration, which essentially turns on radical sceptical hypotheses. I am grateful to an anonymous
referee for Oxford University Press for pressing me on this issue.
paper “Other Minds” (Austin 1961).7 Key to Austin’s attack on scepticism is to high-
light how the sceptic is employing standards for rational evaluation which aren’t mir-
rored in everyday life. For while it is undeniable that in ordinary epistemic contexts we
require agents to be able to rule out (i.e., know to be false) a range of error-possibilities
which could undermine their knowledge of the target proposition, this range of
error-possibilities does not extend to radical sceptical hypotheses like the BIV hypoth-
esis. As Austin argues, in order to know that the creature before you is a goldfinch, you
might well need to be able to rule out the possibility that it is another sort of bird, such
as a woodpecker, but you don’t need to be able to rule out the possibility that it is a hol-
ogram of a goldfinch, or a stuffed goldfinch. As he puts the point in a famous passage:
Enough is enough: it doesn’t mean everything. Enough means enough to show that (within rea-
son, and for the present intents and purposes) it ‘can’t’ be anything else, there is no room for an
alternative, competing, description of it. It does not mean, for example, enough to show it isn’t a
stuffed goldfinch. (Austin 1961, p. 84)
By contrasting the sceptic’s process of epistemic evaluation with our everyday pro-
cess of epistemic evaluation, Austin is trying to persuade us that the sceptical usage
is somehow illegitimate, in that it is a perversion of our normal practices of epistemic
evaluation. At the very least, insofar as we are convinced that the kind of epistemic
evaluation employed by the sceptic is very different from our everyday process of epis-
temic evaluation, then it seems to follow that scepticism can’t be a paradox in the sense
outlined earlier.
It is this last point that Stroud is most keen to counter. He grants that Austin is quite
right to claim that in ordinary epistemic contexts we do not require agents to rule out
sceptical error-possibilities, but he maintains that this fact alone doesn’t settle the issue
of whether scepticism falls out of our ordinary epistemic practices, such that it is a
genuine paradox which trades on a deep tension in our naturally derived epistemic
concepts. In particular, he argues that given that ordinary epistemic contexts are sub-
ject to various kinds of merely practical constraints—e.g., limited time, lack of imag-
ination on the part of the participants, and so on—the failure to consider sceptical
error-possibilities in these normal epistemic contexts does not itself exclude the possi-
bility that the sceptic’s consideration of these error-possibilities is licensed by our ordi-
nary epistemic practices. In particular, he argues that the sceptic’s system of epistemic
evaluation is licensed by our ordinary system of epistemic evaluation on account of the
fact that the former is simply a ‘purified’ version of the latter. That is, if we employ our
ordinary practices of epistemic evaluation with due diligence and set aside all purely
practical limitations, then what we end up with is the system of epistemic evaluation
employed by the sceptic, one that requires (in line with (SP2)) that agents must be able
to rule out radical sceptical hypotheses if they are to have the everyday knowledge that
they standardly attribute to themselves.
7
See especially Stroud (1984, ch. 2).
For our purposes we can set aside the question of whether Stroud is right that the
sceptical system of epistemic evaluation is just a purified version of the system of epis-
temic evaluation that we ordinarily employ. What is more salient is rather the point
that in order to demonstrate that scepticism is not a paradox it is not enough to show
that our quotidian practices of epistemic evaluation are different from the sceptical
practices. Instead, one must further demonstrate that one cannot derive the sceptical
system of epistemic evaluation from our ordinary system of epistemic evaluation in
the way that Stroud claims.8
11.3. Wittgenstein on Radical Scepticism

On this point it is interesting to contrast Austin’s approach to scepticism with the super-
ficially similar, but in fact fundamentally dissimilar, approach to the problem offered
by Wittgenstein in On Certainty. In this work, Wittgenstein is also keen to highlight
the differences between the way in which the sceptic conducts epistemic evaluations
and how epistemic evaluations are actually conducted in ordinary life. In particular,
he argues that our ordinary practice of rational evaluation is essentially local, in that all
rational evaluation takes place relative to certain core commitments, which themselves
are immune to rational evaluation, what Wittgenstein refers to as ‘hinges’:
. . . the questions that we raise and our doubts depend upon the fact that some propositions are
exempt from doubt, are as it were like hinges on which those turn. (1969, §341)
So far, Wittgenstein’s point is essentially the same as Austin’s. Note, however, how this
passage continues:
That is to say, it belongs to the logic of our scientific investigations that certain things are in deed
not doubted.
But it isn’t that the situation is like this: We just can’t investigate everything, and for that rea-
son we are forced to rest content with assumption. If I want the door to turn, the hinges must stay
put. (1969, §342–3)
The idea in play here is crucial to understanding Wittgenstein’s approach to radical

scepticism. For his claim is not merely that there is a radical difference between the
8 There may be other ways to motivate the sceptical paradox while granting that the premises that make
up that paradox do not reflect our ordinary epistemic practices. Craig (1990), for example, offers an influen-
tial account of the concept of knowledge—what we might broadly class as a ‘genealogical’ account—which
elucidates the basic human need that this concept serves. One consequence of the specific genealogical pro-
posal that Craig offers is that the extension that we would expect our concept of knowledge to have given
the basic human need that it answers to will not be the extension that it in fact has. Although the account
which Craig offers does not have any obvious sceptical implications, one could imagine a version of this
sort of proposal which exploited the gap between the actual extension of the concept of knowledge and the
expected extension, given the purpose that it is meant to serve, to explain why radical scepticism is rooted in
our everyday epistemic practices even if not fully displayed within them. For some key discussions of Craig’s
proposal, see Lane (1999); Williams (2002); Neta (2006); Fricker (2007, 2010); Kusch (2009, 2011); Kappel
(2010); and Kornblith (2011). See also Pritchard (2011a, 2012a).
essentially local practices of rational evaluation employed in quotidian epistemic con-

texts and the epistemically unconstrained practices employed in the sceptical context.
Instead, his point is much stronger—viz., that it is a matter of logic that reasons should
be essentially local in the way that he describes, such that the very idea of a fully gen-
eral rational evaluation of the sort that the sceptic envisages (i.e., a rational evaluation
which lacks hinge commitments which are immune to rational evaluation) is simply
incoherent. Indeed, what applies to the sceptic here also applies to the anti-sceptic—
such as Moore (1925, 1939), one of the main figures discussed in Wittgenstein (1969)—
who tries to offer a positive fully general rational evaluation of our beliefs. According
to Wittgenstein, such fully general rational evaluations are impossible, regardless the
philosophical end to which they are put.
If Wittgenstein is right then Stroud’s account of why radical scepticism poses a par-
adox is blocked, and hence, Wittgenstein achieves something with his anti-sceptical
proposal which Austin’s anti-sceptical approach failed to achieve. In particular, what
Wittgenstein would be demonstrating is that the sceptical practices of epistemic evalu-
ation are not just different (in degree) from our quotidian practices of epistemic evalu-
ation, but in fact fundamentally distinct (i.e., different in kind) and, worse, incoherent.
That they are fundamentally distinct deprives Stroud of the logical space within which
he can make a case for the claim that the sceptical practices of epistemic evaluation,
while different from our ordinary practices of epistemic evaluation, are nonetheless
underwritten by them by being simply purified versions of these ordinary practices.9
What is at stake here? Why should it matter whether the radical sceptical problem
constitutes a paradox? I said earlier that the sceptical problem is easier to resolve if it
is not a genuine paradox. We are now in a better position to understand why this is the
case. For notice that if scepticism does not arise out of our ordinary epistemic prac-
tices, then it is much easier to convict the radical sceptic of making a theoretical error.
The sceptic is, it seems, offering us a revisionistic epistemology which generates an
intellectually objectionable conclusion, and we might naturally be suspicious of why
we should adopt such theoretical revisionism given that it has this effect.
The discussion of Wittgenstein’s way of responding to the sceptical problem brings
this point out very effectively. For if Wittgenstein is right then the sceptical problem is
simply the product of a faulty epistemological picture which endorses the possibility
of fully general rational evaluations. Not only do our ordinary epistemic practices not
display such fully general rational evaluations, but a closer inspection of the nature of
rational evaluation is meant to reveal to us that there could be no such thing as a fully
general rational evaluation. It is not then as if we could even make sense of aspiring to
conduct the kind of rational evaluation of our beliefs demanded by the sceptic. The
theoretical picture being offered to us by the sceptic (and, for that matter, by a certain
9 For further discussion of this aspect of Wittgenstein’s anti-scepticism, see Pritchard (2011c, 2012c, and
Forthcoming). For a seminal discussion of the general idea in play here—viz., that Wittgenstein is trying to
show that radical scepticism is the product of theory rather than commonsense—see Williams (1995).
kind of anti-sceptical epistemologist) is thus simply wrong and so should be rejected

out of hand.
In contrast, if radical scepticism does pose a paradox then, as Stroud makes very
clear, it is going to be very hard to offer an intellectually compelling response to that
problem, since any such response would be forced to deny “platitudes that we would
all accept,” and that can hardly be intellectually appealing.10 Moreover, if Stroud is right
that the radical sceptical problem can pose a paradox even if there is only an indirect
connection between our ordinary practices of epistemic evaluation and the practices
of epistemic evaluation appealed to by the radical sceptic, then simply noting the dif-
ference between these two practices of epistemic evaluation will not suffice to ensure
that you don’t have to deny a platitude in responding to the sceptical problem.11
One final point is in order on this score, and this is the sense in which Wittgenstein’s
response to the problem of radical scepticism can be described as ‘quietist’. For note
that his goal in this regard is to show that the sceptical problem does not fall out of
our ordinary epistemic practices, as someone like Stroud would allege, but is rather
the product of faulty philosophizing. For Wittgenstein, then, we shouldn’t think of the
problem of radical scepticism as arising out of our ordinary epistemic practices and
then being ‘solved’ by philosophical reflection. Rather, we need to understand that it is
a problem that is generated by philosophical reflection and completely detached from
our ordinary epistemic practices. Once we understand that, so the line goes, there is
then no problem for philosophy to solve.12
11.4. Sceptical Intuitions

Part of what makes quietism an appealing option when it comes to a debate like radi-
cal scepticism is the fact that while this is meant to be a paradox, it is at the same time
granted by all concerned that it is not obviously a paradox. That is, someone like Stroud
will not dispute that ordinary folk will not recognize that they are faced with a paradox
when first presented with the sceptical argument. Instead, the idea is that with a certain
degree of philosophical ingenuity one can get ordinary folk to realize that, on reflection,
10 Another way of expressing this point—due to Schiffer (1996)—is that if radical scepticism poses a para-
dox then there is no ‘happy face’ solution to this problem, only a ‘sad face’ solution, because it would mean
that there is a “deep-seated incoherence” within our concept of knowledge (p. 330).
11 Note as well that there is an important dialectical shift when we are dealing with a philosophical prob-
lem qua paradox. This is because paradoxes expose fundamental tensions within our own concepts, and
hence constitute problems for everyone. In contrast, if one can show that a putative paradox is merely the
product of a theoretical commitment, then it is easier to disengage oneself from the problem by treating it
as a difficulty for a particular philosophical stance. For an excellent discussion of this point, and its implica-
tions for the radical sceptical problem, see Wright (1991).
12 Interestingly, while many commentators have noted quietistic themes in Wittgenstein’s earlier work,
particularly the Philosophical Investigations (1953), there hasn’t been much discussion of the quietism that
seems to be clearly on display in On Certainty (1969) (which consists of his final notebooks). For an excellent
recent discussion of Wittgenstein’s quietism, see McDowell (2009).
there is a paradox in play here. But of course this feature of the debate naturally invites
the quietistic thought that perhaps this ‘paradox’ is generated by philosophical theory
rather than being the result of a deep tension in our ordinary epistemic concepts.
We can see this point in action by comparing the radical sceptical problem with
other philosophical puzzles which clearly are paradoxical in nature, and which don’t
require any great philosophical expertise to set-up. Take, for example, the problem of
vagueness. Here, for instance, is a particular variant of this problem:
( VP) Removing one hair from a non-bald person’s head cannot make them bald.
(VC) So, one can remove single hairs from a non-bald person’s head indefinitely
without that person becoming bald.
The premise seems undeniable. How could removing a single hair on a non-bald per-
son’s head make them bald? But, equally, if this is true then it ought to be the case
that any number of iterations of removing a hair from a non-bald person’s head will
not make them bald too. But then the conclusion of this argument follows deductively
from the premise and we have our paradox. As a result, in responding to this problem
it thus seems unavoidable that we will be forced as theorists to argue for a claim which
is counter to intuition, and hence all responses to the problem of vagueness are to that
extent at least problematic.
The problem of radical scepticism, however—even if we grant with Stroud that it is a
paradox—is not like the problem of vagueness. We just set up the paradox of vagueness
by appealing only to familiar concepts and principles, and by stating a premise which
most people who are new to philosophy will recognize as obviously true. In this sense,
no great philosophical expertise is required to get people to see that there this is some-
thing paradoxical about vagueness; you just need to know a good example to illustrate
the problem.
But as we noted in section 11.2, one can’t motivate the radical sceptical paradox in
such a straightforward fashion, because the premises of this ‘paradox’ will strike the
philosophically uninitiated as obviously false. Instead, it takes a skilled epistemologist
to show folk that implicit within our everyday epistemic practices are commitments
which can be employed to generate the sceptical problem. But this difference between
these two ‘paradoxes’ raises a natural worry—viz., that while the problem of vagueness
is a genuine paradox which involves only appeal to intuition, the problem of radical
scepticism is not a genuine problem since it cannot be generated by intuition alone but
requires in addition a substantial contribution from philosophical theory.
Although there is clearly a distinction to be drawn between the kind of philosophical
problem posed by a problem like vagueness and that posed by radical scepticism, I think
we should be wary about drawing the contrast in quite this way such that only the former
is a genuine paradox. The first point to note is that it is somewhat tendentious to claim
that philosophical theory is being employed in order to make a case for treating the radi-
cal sceptical problem as a paradox. This is certainly not how someone like Stroud would
describe what he is doing, after all. In particular, he would claim that all that is happening
here is that a philosopher is using her analytical skills in order to uncover the relevant
epistemic commitments which are implicit in our everyday epistemic practices.
This leads us to a second, and more fundamental, point, one that is particularly sali-
ent for our purposes. Let us imagine someone who has been convinced by Stroud’s
arguments that the problem of radical scepticism really does arise out of our ordinary
epistemic practices, and hence is a paradox. We can delineate two stages in this agent’s
response to the sceptical problem, so presented. In the first, she does not regard the
premises of the argument as intuitive at all, but as rather claims which run counter to
her instinctive judgements regarding the correct usage of the relevant epistemic con-
cepts. In the second, she spontaneously judges the premises to be true and regards
them as reflecting a correct usage of the relevant epistemic concepts.
The critical question for us is whether we should regard the agent’s judgements in
the second case as genuine intuitions, given what has taken place in the interim to
convince her to form these judgements. If we are to hold that the problem of radical
scepticism, even if Stroud’s account of it as being implicit within our everyday epis-
temic practices is right, is not a paradox because it takes theoretical skill to get us to
recognize the truth of the premises, then presumably this is because the relevant theo-
retical input entails that the judgements in play are not genuine intuitions. Put another
way, the claim is that the judgements made by the agent in the first case just described
were genuine intuitions about the truth of the premises, but that the judgements made
in the second case are in fact the product of philosophical theory and hence not intui-
tions.13 But this is clearly problematic, since philosophers like Stroud (and many others
for that matter) are obviously committed to thinking of the judgements made in the
second case as the real intuitions.
It seems then that there is a tension here. On the one hand, there is a natural view
about intuitions, such that they are uninformed by philosophical theory in any sense,
and hence constitute a raw intellectual response to an intellectual stimulus. On the
other hand, there is actual philosophical practice—at least as regards the fundamen-
tal philosophical debate regarding radical scepticism—which understands the target
‘intuitions’ in such a way that they can be the product of a significant degree of philo-
sophical scene-setting.
11.5. Intuitions and Philosophical Methodology

What I am here calling the ‘natural view’ about intuitions is effectively one on which
intuitions are mere intellectual seemings, unguided by any sort of expertise or
13 Note that the contrast here is not between pre-theoretical intuitive judgement and intuitive judgement
that is informed by theory, but rather between pre-theoretical intuitive judgement and intuitive judgement
that is informed by philosophical theory. After all, one might view our folk epistemology as being itself a kind
of theory, and hence even our ‘pre-theoretical’ intuitive judgements with regard to the sceptical problem
could be to this extent the products of theory. That wouldn’t make them the products of philosophical theory
though. I am grateful to Darrell Rowbottom for pressing me on this issue.
theoretical knowledge which would prompt us to ‘see’ things differently. This account
of intuitions has no trouble explaining why the problem of vagueness is a paradox,
since anyone who understands the propositions at issue will find themselves inclined
to assent to their truth (even once they realize that the claims that make up this para-
dox cannot be collectively true). But it struggles to account for what Stroud wants to
take as being the reflectively informed intuitions which drive the sceptical paradox.
On this view, such judgements would presumably not count as intuitions at all. And
yet, even though it takes some work to get people to ‘see’ things in the appropriate way,
the judgements in question do seem to be naturally thought of as intuitive judgements.
In particular, like intuitive judgements more generally, they are non-inferential and
spontaneously formed.
I think that part of the obstacle to treating the reflection-informed judgements as
intuitions is the analogy with perceptual seemings that tends to inform our concep-
tion of intuition. It is often noted, for example, that just as the lines in the Müller-Lyer
illusion continue to seem of different length even once one becomes aware that they
are in fact the same length, so our judgement that a certain proposition is intuitive
can remain even once theory has convinced us that this proposition must be false.14
It is certainly true that our intuitive judgements often have this feature, and the prob-
lem of vagueness is a case in point. Even once one becomes convinced of a theoretical
response to this particular problem, such that (say) one denies the premise that we saw
generating this puzzle in the formulation offered above, one will still nonetheless tend
to find this premise intuitive.
But if I am right that not all philosophical problems are like that of vagueness in
this respect, in that they appeal to intuitions that require some setting-up on the part
of the philosopher proposing the problem, then we need to complicate our picture
of how intellectual seemings are analogous to perceptual seemings. For it seems that
the problem of radical scepticism demonstrates the possibility that what once seemed
intuitive can over time be overturned such that what was once intuitive is no longer
intuitive; indeed, such that the very opposite of what was once intuitive is now intuitive.
At the very least, if we want to understand the role that intuition is meant to play in
philosophical methodology then we need to create the logical space for this possibility.
I suggest that it is not the perceptual analogy that is the problem, but rather a far too
rigid application of that analogy. For clearly what we want to say is that when it comes
to a debate like radical scepticism it is possible for someone who is first engaging with
the debate to not see clearly what is at issue (even though they think that they do). The
task of the philosopher, on any model of philosophy which can accommodate some-
one like Stroud anyway, is to help this person come to see the debate aright, and in
doing so appreciate the intuitive force of the premises in play.
Indeed, I don’t think the problem of radical scepticism is unique in this respect (if
it were, then that would obviously be grounds to be suspicious about this debate).
See, for example, Sosa (2007a, ch. 3).

14
Consider, for instance, what happens when one introduces philosophy students to
Robert Nozick’s (1974) ‘experience machine’ for the first time. This is a machine that
creates an artificial life for the subject which is experientially indistinguishable from
‘real’ life, in the sense that once one is in the machine one can’t tell that one’s experi-
ences are in fact artificially generated. Let us stipulate that life inside the machine is
significantly more pleasurable than normal life outside the machine. Here is the philo-
sophical question: should one prefer an artificial life inside the machine, with all its
additional attendant pleasure, to a real life outside the machine with all its attendant
trials and tribulations?
My experience as someone who has often taught this example to students who are
encountering philosophy for the first time is that insofar as the students have any ini-
tial opinions on this matter at all, then they tend to intuitively regard the life in the
experience machine as at least no worse than the real life, and often preferable to the
real life. Significantly, however, this judgement tends not to be stable. For example, if
one asks the students whether they would be happy for their children to live their lives
in the experience machine then most opt for the real life outside the machine, even
though they recognize that there is a tension between this judgement and their previ-
ous judgement about the desirability of the life in the machine. Relatedly, if one makes
explicit that entering the machine is a one-way ticket—perhaps because one’s body
becomes unusable thereafter as part of the ‘re-orientation’ process—then again stu-
dents’ intuitions tend to shift towards regarding the life outside the machine as being
preferable to the life inside the machine. In fact, once one has explored the example in
some detail then the groundswell of opinion tends to be in favour of treating the real
life outside the machine as better than the artificial life inside the machine.
Here, then, we have a case in which people’s initial verdicts about a scenario change
over time in response to questioning and further reflection. Now we could describe
this process as simply being one in which the philosopher impresses her intuitions on
the audience, and indeed it may well be the case that this is what is happening (I will
return to this issue). But what is meant to be taking place here is that the philosopher
is getting the audience to see that their thinking about an issue is muddled, and to
enable them to see this issue more clearly. Crucially, however, the judgements that the
agent makes who does see the matter clearly are meant to be no less intuitive than were
their initial, and conflicting, intuitive judgements in this regard. The analogy with the
debate regarding radical scepticism as Stroud describes it should be clear.
If we take this way of thinking about philosophy seriously—and I am suggesting
that we should—then we will need a conception of intuition which is consistent with
it. Although this constraint on a theory of intuition might initially look to have sub-
stantive implications for our thinking about intuition, I think that on closer inspec-
tion this is not nearly as restrictive a constraint as it might first appear. First off, notice
even the kind of deflationary accounts of intuition offered by such people as Timothy
Williamson (e.g., 2004, 2007), such that intuitions are just inclinations to judge
which have no distinctive epistemological role, could be perfectly consistent with this
constraint. All that matters for our purposes is that we are able to capture the sense in
which these judgements can be epistemically improved via philosophical engagement
while remaining intuitive judgements.15
Equally, this constraint is compatible with an account of intuitions on which they
have a distinctive aetiology, such as being triggered by understanding (see, e.g., Sosa
1998).16 All that would be required is that we ensure that the aetiological story in play
allows for the possibility of an improvement in one’s intuitive judgement, as presuma-
bly it will. To take the specific case of understanding, one would presumably just regard
the revised intuitive judgment as being the product of a greater level of understanding.
Perhaps the only prima facie tension between this constraint on a theory of intuition
and a particular type of theory of intuition arises when we consider views on which
intuition is understood as a kind of intellectual seeming (e.g., Bealer 1987, 1996a, 1996b,
1998; BonJour 1998). After all, one might regard such a view as treating only the ini-
tial judgement as a genuine intuition, since only this judgement qualifies as the ‘pure’
intellectual seeming. But even proponents of this kind of view do not appear to be
inclined towards this kind of restrictive line, and there seems nothing inherent within
the intellectual-seeming view that would necessitate that it should be so restrictive.17
So I don’t think that this conception of the role of intuition in philosophical method-
ology need create any particular problems for the theory of intuition, at least on closer
inspection. Where it does have a bearing is on how we should respond to a certain style
of attack on philosophical methodology that can be found in the recent literature.
11.6. The Negative Programme in Experimental

Philosophy
One of the most significant movements in contemporary philosophy has been experi-
mental philosophy, where experimental work has been done surveying the intuitive
judgements made by non-philosophers to examples that are relevant to philosophy. So,
for example, where philosophers have argued that it is intuitive to respond to a certain
described scenario in a certain way, experimental philosophers have done a systematic
survey of non-philosophers’ immediate responses to this scenario to see how this cor-
responds to the intuitional ‘data’ offered by the philosopher.
15 Indeed, see Williamson (2007, p. 191) for a clear defence of the idea that the philosophical use of intui-
tions requires a significant level of expertise. (Note, however, that Williamson is motivating this as a general
claim about philosophical methodology, while I am here only arguing that this applies within certain philo-
sophical debates, like that concerning radical scepticism.)
16 Indeed, a recurrent motif of Sosa’s work on intuition is that skill can be required to elicit certain intui-
tions. See Sosa (1998, 2007a, ch. 3, 2007b, 2007c, and 2009).
17 Indeed, the most prominent advocate of this account of intuition seems inclined to treat the initial
judgements as not being, strictly speaking, intuitive judgements at all. See, for example, Bealer (1998).
Following Jonathan Weinberg (2010), we can draw a distinction between a ‘posi-

tive’ and a ‘negative’ programme in experimental philosophy.18 Whereas the former
is meant merely to offer supplementary data for philosophers to appeal to, the latter
is meant to undermine philosophers’ attempts to appeal to intuitions. It is the latter
programme that I want to focus on here.
One way in which the negative programme has been motivated has been by appeal
to a significant mismatch between the experimental data that has been uncovered and
the corresponding intuitional data offered by philosophers. In particular, where it can
be shown that the mismatch is systematic—in that people from, say, different cultures
to those from which the intuitional data is usually drawn offer different responses to
cases—then the case for the negative programme can seem very strong.19 Of course,
there have been disputes about the nature of the experimental data in question—e.g.,
how the questions were posed, and so on—and a lot of the focus with regard to the
negative programme in experimental philosophy has been on this issue.20 This is not
my concern here, however.
Rather, my concern is whether there is something wrong-headed about equating
the intuitional data appealed to by philosophers to the immediate responses to cases
offered by the philosophically uninformed. The case of the sceptical paradox (or, for
that matter, Nozick’s “experience machine”) seems to indicate otherwise. But if that’s
right then nothing need follow from the fact that the experimental data radically con-
flicts with the intuitional data offered by philosophers. What would be key would
be the type of philosophical intuition that is being appealed to. If it is an intuition of
the sort that is present in the paradox of vagueness then this mismatch might well be
significant, since here the intuitions do indeed reflect an immediate philosophically
uninformed response to (in this case) a proposition. If in this case the ‘folk’ turned out
not to have the same intuitions as the philosophers then this would surely be cause for
concern. But when it comes to debates like radical scepticism where the relevant phil-
osophical intuition is putatively informed by philosophical expertise, then the mere
existence of this mismatch between ‘folk’ and ‘philosophical’ intuitive judgements is
far from being philosophically problematic. Indeed, it is entirely to be expected.
Does this mean that the negative programme in experimental philosophy is poten-
tially in jeopardy? I don’t think so.
First, notice that at most all that follows from the conception of intuition offered
above is that we shouldn’t expect the experimental data in certain cases to correspond
to the intuitional data offered by philosophers. If the range of those ‘certain cases’ is
relatively slim, then the negative programme can still inflict a great deal of damage on
the philosophical appeal to intuition. In particular, notice that while flagging up the
particular way in which intuition is appealed to with regard to the problem of radical
18
Weinberg (2010, p. 823) in turn credits this distinction to Farid Masrour.
19
See, for example, Nichols et al. (2003) and Swain et al. (2008).
20
See, for example, Kauppinen (2007); Sosa (2007b); Levin (2009); Cullen (2010); and Deutsch (2010).
scepticism we also made a point of demonstrating how the philosophical use of intui-
tion can often involve an appeal to a subject’s immediate response to a case (as in the
case of the paradox of vagueness). The idea that the philosopher’s use of intuition can
sometimes involve expertise is thus not a panacea which can offer a blanket cure to the
very specific sceptical challenge posed by the negative programme.
Second, even if it is true that the philosophical use of intuition is such that in cer-
tain cases a great deal of skill is required in order to elicit the relevant intuition, it still
doesn’t follow that philosophers should be unconcerned about the challenge posed
by the proponent of the negative programme. The contention would instead be that
the point of attack for the proponent of this programme shouldn’t come from a mere
consideration of the immediate responses of subjects to certain cases, but should
rather confront head-on the philosopher’s appeal to expertise. For if it can be shown
that this appeal to expertise is illusory, then that could cause a serious headache for
philosophers.
There are at least two ways in which proponents of the negative programme might
go about doing this. One way would be to make use of recent work in experimental
psychology on this very issue, since there is a growing literature regarding studies
which show, for example, that subjects can often overestimate their level of exper-
tise, particularly when it comes to failing to realize just how narrow their specialized
field of competence is.21 A second possibility, which could dovetail with the first,
is to show that there is something inherently dubious about the process by which
subjects are ‘trained’ to acquire philosophical ‘intuitions’. This second issue is par-
ticularly pressing given that the philosophical use of intuition appears to face the
so-called ‘calibration problem’. For whereas other epistemic sources, such as per-
ceptual observation, can be calibrated for their reliability by comparing their results
against outcomes gained from other sources, this doesn’t seem to be available in the
case of philosophical intuition, since this seems only to be testable by appeal to other
philosophical intuitions.22
Furthermore, notice that the particular way in which philosophical intuitions are
appealed to in a debate like radical scepticism can itself be a source of scepticism
in this regard. For insofar as philosophical expertise is required in order to set-up
a philosophical problem, then that surely provides an impetus towards a quietistic
response to the problem which maintains that the ‘paradox’ in play is in fact just the
product of faulty—and dispensable—philosophical theory. Indeed, this was just how
the issue of radical scepticism played out above if one opted for the Wittgensteinian
response to the problem. It should be clear that any philosophical problem which
requires philosophical expertise to set up would be prima facie amenable to a
21 Weinberg (2010, §III) offers an interesting discussion of the relevant empirical literature, and how
it might be brought to bear in support of the negative programme in experimental philosophy. See also
Weinberg et al. (2010).
22 This problem is usually credited to Cummins (1998). For some key discussions of this problem, see
Weatherson (2003); Nagel (2007); and Weinberg et al. (2012).
corresponding quietistic response. Empirical and theoretical concerns about the

philosopher’s appeal to expertise would thus find common cause with the quietist’s
concerns about how philosophy can be the source, rather than the resolution, of phil-
osophical problems.
So while I hold that a philosophical problem like radical scepticism demonstrates
that we need to re-think the role that intuitions play in philosophical methodology,
and that this creates problems for a crude rendering of the negative programme in
experimental philosophy, I certainly don’t think that moving to this reoriented con-
ception of the role of intuitions in philosophical methodology completely undermines
the challenge posed by this branch of experimental philosophy. The reason for this is
that such a reoriented conception exposes a key source of vulnerability in the philo-
sophical project, and thus highlights just where the proponent of the negative pro-
gramme needs to re-target their attack.23
11.7. Concluding Remarks

My concern in this paper has been to show what the debate regarding radical scepti-
cism reveals about the nature of the philosopher’s appeal to intuition. I have argued
that we need to make room for the possibility that the intuitional judgements rele-
vant to philosophy can sometimes be quite significantly informed by expertise. I have
noted that this poses a prima facie problem for the negative programme in experimen-
tal philosophy, in that it means that one cannot conclude from the fact that there is a
significant mismatch between the intuitional data appealed to by philosophers and
the experimental data, that there is thereby a problem posed for the philosophical use
of intuition. But I have also argued that, at best, this only provides philosophy with
a provisional, and possibly quite narrow, defence of its use of intuition. On the one
hand, it is potentially a narrow defence because it is still the case that quite a lot of the
philosophical use of intuition might well be susceptible to the critique offered by the
negative programme. On the other hand, it is a provisional defence in that in effect it
highlights what the proper target of the critique offered by the negative programme
in experimental philosophy should be, namely the appeal to philosophical expertise,
something which can itself be called into question on experimental grounds. On this
score, as we have seen, there is further support to be found from philosophers them-
selves, particularly from those approaching philosophical problems from quietistic
quarters. In particular, the very debates that motivate the idea that we need to under-
stand the philosophical use of intuition as requiring philosophical expertise are also
the ones most susceptible to a quietistic critique.
23 For further discussion of the implications of the negative programme in experimental philosophy for
contemporary epistemology, see Pritchard (2012b).
References
Austin, J. L. (1961). “Other Minds,” in J. O. Urmson and G. J. Warnock (eds), J. L. Austin:
Philosophical Papers. Oxford: Oxford University Press, pp. 76–116.
Bealer, G. (1987). “The Philosophical Limits of Scientific Essentialism,” in J. E. Tomberlin (ed.),
Philosophical Perspectives 1: Metaphysics. Atascadero, CA: Ridgeview, pp. 289–365.
——. (1996a). “A Priori Knowledge and the Scope of Philosophy,” Philosophical Studies 81, pp.
121–42.
——. (1996b). “On the Possibility of Philosophical Knowledge,” in J. E. Tomberlin (ed.),
Philosophical Perspectives 10: Metaphysics. Atascadero, CA: Ridgeview, pp. 1–34.
——. (1998). “Intuition and the Autonomy of Philosophy,” in M. R. DePaul and W. Ramsey (eds),
Brueckner, T. (2004). “Brains in a Vat,” in E. N. Zalta (ed.), The Stanford Encyclopedia of
Philosophy, available at: <http://plato.stanford.edu/entries/skepticism-content-externalism
/> [last accessed 22 October, 2013].
Cassam, Q. (2009). “Knowing and Seeing: Responding to Stroud’s Dilemma,” European Journal
of Philosophy 17, pp. 571–89.
Chisholm, R. (1973). The Problem of the Criterion. Milwaukee, MN: Marquette University Press.
Craig, E. (1990). Knowledge and the State of Nature: An Essay in Conceptual Synthesis.
Cullen, S. (2010). “Survey-Driven Romanticism,” Review of Philosophy and Psychology 1, pp.
275–96.
Cummins, R. (1998). “Reflection on Reflective Equilibrium,” in M. DePaul and W. Ramsey (eds),
Deutsch, M. (2010). “Intuitions, Counter-Examples, and Experimental Philosophy,” Review of
Philosophy and Psychology 1, pp. 447–60.
Fricker, M. (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford
University Press.
——. (2010). “Scepticism and the Genealogy of Knowledge: Situating Epistemology in Time,”
in A. Haddock, A. Millar, and D. H. Pritchard (eds), Social Epistemology. Oxford: Oxford
University Press, pp. 51–68.
Kappel, K. (2010). “On Saying that Someone Knows: Themes from Craig,” in A. Haddock, A.
Millar, and D. H. Pritchard (eds), Social Epistemology. Oxford: Oxford University Press,
pp. 69–88.
Kauppinen, A. (2007). “The Rise and Fall of Experimental Philosophy,” Philosophical
Kornblith, H. (2011). “Why Should We Care About the Concept of Knowledge?,” Episteme 8
pp. 38–52.
Kusch, M. (2009). “Testimony and the Value of Knowledge,” in A. Haddock, A. Millar, and D. H.
Pritchard (eds), Epistemic Value. Oxford: Oxford University Press, pp. 60–94.
——. (2011). “Knowledge and Certainties in the Epistemic State of Nature,” Episteme 8, pp. 6–23.
Lane, M. (1999). “States of Nature, Epistemic and Political,” Proceedings of the Aristotelian Society
99, pp. 211–24.
Levin, J. (2009). “Experimental Philosophy,” Analysis 69, pp. 761–9.
McDowell, J. (2009). “Wittgensteinian ‘Quietism’ ,” Common Knowledge 15, pp. 365–72.
Moore, G. E. (1925). “A Defence of Common Sense,” in J. H. Muirhead (ed.), Contemporary
British Philosophy, 2nd series. London: Allen & Unwin, pp. 193-223.
——. (1939). “Proof of an External World,” Proceedings of the British Academy 25, pp. 273–300.
Nagel, J. (2007). “Epistemic Intuitions” Philosophy Compass 2, pp. 792–819.
Neta, R. (2006). “Epistemology Factualized: New Contractarian Foundations for Epistemology,”
Synthese 150, pp. 247–80.
Ethno-Epistemology,” in S. Luper (ed.), The Skeptics. Aldershot: Ashgate, pp. 227–47.
Nozick, R. (1974). Anarchy, State, and Utopia. New York: Basic Books.
Pritchard, D. H. (2000). “Doubt Undogmatized: Pyrrhonian Scepticism, Epistemological
Externalism, and the ‘Metaepistemological’ Challenge,” Principia—Revista Internacional de
Epistemologia 4, pp. 187–214.
——. (2002). “Recent Work on Radical Skepticism,” American Philosophical Quarterly 39, pp.
215–57.
——. (2005). Epistemic Luck. Oxford: Oxford University Press.
——. (2011a). “The Genealogy of the Concept of Knowledge and Anti-Luck Virtue Epistemology,”
in S. Tolksdorf (ed.), Der Begriff des Wissens/Conceptions of Knowledge. Berlin: de Gruyter, pp.
159–78.
——. (2011b). “Wittgensteinian Pyrrhonism,” in D. Machuca (ed.), Pyrrhonism in Ancient,
Modern, and Contemporary Philosophy. Dordrecht: Springer, pp. 193–202.
——. (2011c). “Wittgenstein on Scepticism,” in O. Kuusela and M. McGinn (eds), The Oxford
Handbook on Wittgenstein. Oxford: Oxford University Press, pp. 521–47.
——. (2012a). “Anti-Luck Virtue Epistemology,” Journal of Philosophy 109, pp. 247–79.
——. (2012b). “The Methodology of Epistemology,” The Harvard Review of Philosophy 18, pp.
91–108.
——. (2012c). “Wittgenstein and the Groundlessness of Our Believing,” Synthese 189, pp.
255–72.
——. (Forthcoming). “Wittgenstein on Hinges and Radical Scepticism,” in H.-J. Glock and J.
Hyman (eds), On Certainty: Blackwell Companion to Wittgenstein. Oxford: Blackwell.
Putnam, H. (1981). Reason, Truth and History. Cambridge: Cambridge University Press.
Schiffer, S. (1996). “Contextualist Solutions to Skepticism,” Proceedings of the Aristotelian Society
96, pp. 317–33.
Sosa, E. (1994). “Philosophical Scepticism and Epistemic Circularity,” Proceedings of the
——. (1998). “Minimal Intuition,” in M. DePaul and W. Ramsey (eds), Rethinking Intuition: The
Psychology of Intuition and its Role in Philosophical Inquiry. Lanham, MD: Rowman &
Littlefield, pp. 257–70.
——. (2007a). A Virtue Epistemology: Apt Belief and Reflective Knowledge, Volume I.
——. (2007b). “Experimental Philosophy and Philosophical Intuition,” Philosophical Studies 132,
pp. 99–107.
——. (2007c). “Intuitions: Their Nature and Epistemic Efficacy,” Grazer Philosophische Studien
74, pp. 51–67.
——. (2009). “A Defence of the Use of Intuitions in Philosophy,” in M. A. Bishop and D. Murphy
(eds), Stich and His Critics. Malden, MA: Wiley-Blackwell, pp. 101–12.
Stroud, B. (1984). The Significance of Philosophical Scepticism. Oxford: Clarendon Press.
——. (1989). “Understanding Human Knowledge in General,” in M. Clay and K. Lehrer (eds),
Knowledge and Skepticism. Boulder, CO: Westview Press, pp. 31–50.
——. (1994). “Scepticism, ‘Externalism’, and the Goal of Epistemology,” Proceedings of the
——. (2009). “Explaining Perceptual Knowledge: Response to Quassim Cassam,” European
Journal of Philosophy 17, pp. 590–96.
76, pp. 138–55.
Weatherson, B. (2003). “What Good are Counterexamples?,” Philosophical Studies 115,
pp. 1–31.
Weinberg, J. (2010). “Experimental Epistemology,” in S. Bernecker and D. H. Pritchard (eds),
Routledge Companion to Epistemology. London: Routledge, pp. 823–35.
——., Crowley, S. J., Gonnerman, C., Swain, S., and Vandewalker, I. (2012). “Intuition and
Calibration,” Essays in Philosophy 13, pp. 256–83.
——., Gonnerman, C., Buckner, C., and Alexander, J. (2010). “Are Philosophers Expert
Williams, B. (2002). Truth and Truthfulness: An Essay in Genealogy. Princeton, NJ: Princeton
University Press.
Williams, M. (1995). Unnatural Doubts: Epistemological Realism and the Basis of Skepticism.
Princeton, NJ: Princeton University Press.
Williamson, T. (2004). “Philosophical ‘Intuitions’ and Scepticism about Judgement,” Dialectica
58, pp. 109–53.
Wittgenstein, L. (1953). Philosophical Investigations, G. E. M. Anscombe and R. Rhees (eds) and
G. E. M. Anscombe (trans). Oxford: Blackwell.
——. (1969). On Certainty, G. E. M. Anscombe and G. H. von Wright (eds) and G. E.
M. Anscombe and D. Paul (trans). Oxford: Blackwell.
Wright, C. J. G. (1991). “Scepticism and Dreaming: Imploding the Demon,” Mind 397, pp. 87–115.
12
Who Needs Intuitions? Two
Experimentalist Critiques
Jonathan Jenkins Ichikawa†
12.1. Introduction
Here is a familiar situation: some philosophers are considering whether some philo-
sophical theory X is true. One of them invokes a thought experiment to argue that it
is not. “Here is an imaginary case S,” she says. “Intuitively, it is a case in which p. But
according to theory X, S is a case in which not-p. So X is false.” In many cases, argu-
ments with this sort of shape have been widely accepted, and have provided the basis
for the mainstream rejection of the relevant theories. Examples of apparent arguments
of this form are easy to come by: Edmund Gettier’s (1963) refutation of the identity of
knowledge with justified true belief is one; another is Hilary Putnam’s (1973) refutation
of semantic internalism; a third is Saul Kripke’s (1980) refutation of the descriptivist
theory of names. These instances of the argument-form are generally, if not univer-
sally, accepted among analytic philosophers engaging in broadly armchair methods.
Other arguments of the same general form are more controversial: consider David
Chalmers’s (1996) argument from the intuitive possibility of phenomenal zombies to
the denial of physicalism about the mind, or Judith Jarvis Thompson’s (1971) argument
from an intuitive verdict about an unwelcome violin player to the permissibility of
abortion. Appeal to intuition appears ubiquitous in armchair philosophy.1
† The material of this chapter has been kicking around for some time; its first incarnation was in my
Ph.D. dissertation in 2008. I’m grateful, for helpful advice and discussions, to my Ph.D. supervisor, Ernest Sosa,
and to the rest of my supervisory committee: Brian Weatherson, Jason Stanley, Alvin Goldman, and Tamar Szabó
Gendler. Thanks also to Derek Ball, Herman Cappelen, Yuri Cath, Juan Comesaña, Benjamin Jarvis, Carrie
Ichikawa Jenkins, Julia Langkau, Aidan McGlynn, Stephen Stich, Jonathan Weinberg, Timothy Williamson, and
two anonymous referees for helpful advice and comments at diverse stages in the life of this chapter, and for
useful conversations on related topics. Some of the work for this chapter was done as part of the AHRC-funded
project on intuitions and philosophical methodology at the Arché Research Centre at St Andrews.
1 What is ‘armchair philosophy’ exactly? I can here do little beyond pointing to exemplars, like those in
this paragraph. I am not at all convinced that all instances of ‘armchair philosophy’ have anything interesting
Who Needs Intuitions? Two Experimentalist Critiques 233
In the last decade, arguments of this sort of form have come under a particular
sort of philosophical scrutiny.2 Insofar as philosophical argument relies on philo-
sophical intuition, it is natural to wonder whether such intuitions are reliable guides
to truth. The experimentalist critique of traditional methodology comprises the
extensive recent literature challenging the invocation of intuitions in philosophy on
empirical grounds. So-called ‘experimental philosophers’ have produced data that
purport to call the trustworthiness of philosophical intuition into question.3 Given
the apparent importance of philosophical intuition in traditional philosophical
methodology, these studies are sometimes thought to provide serious worries for
the latter.
However, it has recently become increasingly fashionable to resist this
intuition-emphasizing interpretation of traditional philosophical methodology.4 In
previous work, I have sided with the resistance. My present project is to evaluate the
significance of the experimentalist critique, once naïve ideas about the importance of
intuition have been rejected. In section 12.2, I review and endorse some recent sugges-
tions that the role of intuition in philosophical methodology has been exaggerated.
In section 12.3, I consider whether this constitutes a quick objection to the central
tenets of the experimentalist critique; I agree with recent experimentalists that there
is an important sense in which it does not. This motivates the central project of sec-
tion 12.4: drawing an underappreciated distinction between two very different sorts of
experimentalist worries. One experimentalist worry—the one emphasized in section
12.3, and typified by projects such as Jonathan Weinberg (2007)—is indeed orthogo-
nal to questions about the role of intuition in philosophical methodology. In section
12.5, I argue that the other experimentalist worry—more prevalent in experimentalist
papers like Weinberg et al. (2008), and emphasizing important connections to earlier
projects like Stich (1990)—looks, at least on its surface, to depend on an important role
for intuitions in standard philosophical method. I will consider alternate recastings of
this critique in sections 12.6–12.9; I conclude that none are cause for serious pessimism
about armchair philosophy.
in common that is not so general as to include all cognition. This is a central theme of Timothy Williamson
(2007), and of Jonathan Ichikawa and Benjamin Jarvis (2013). Since my ultimate project in this chapter is
to rebut a critique that is targeted at armchair philosophy, I take it the imprecision in the latter notion is the
critique’s problem, not mine.
2 The experimental critique represents an intensified form of a critique that was already present in the
literature; some important precursors can be found in Gilbert Harman (1977); Stephen Stich (1990); Robert
Cummins (1998); and Jaakko Hintikka (1999).
3 Not all experimental philosophers press the experimentalist critique; some do empirical work simply
because they are interested in what uncontroversially are empirical questions. See e.g. Joshua Knobe (2007),
emphasizing questions about the human mind. This chapter is concerned only with the experimentalist
critique—what Joshua Alexander et al. (2010) call the ‘negative program’ of experimental philosophy.
4 The most forceful presentation of the resistance is Herman Cappelen (2012); see also Max Deutsch
(2009, 2010); Williamson (2004); Ichikawa (2014), and Ichikawa and Jarvis (2013).
234 Jonathan Jenkins Ichikawa
12.2. Does Armchair Philosophy Need Intuition?

It is a commonplace that armchair philosophy relies on intuition, but is it true?
‘Intuition’, like attitude names generally, is polysemous. For example, the word
‘belief ’ can pick out a particular mental state or episode—“her belief is a belief that
everything happens for a reason”—or it can refer to the propositional content of such
a state—“her belief is that everything happens for a reason.” Likewise, when Dave
Chalmers has the intuition that zombies are possible, we can use ‘intuition’ to refer to
a state of Chalmers’s mind, as in “the intuition explains his attraction to dualism”—or
we can use it to refer to the content of this state, as in “the intuition entails the falsity
of physicalism.” William Lycan (1998) calls ‘intuitions’ in the former sense ‘intuitings’,
and ‘intuitions’ in the latter sense ‘intuiteds’. How should we interpret the claim that
armchair philosophy relies on intuitions? Is it a claim of reliance on attitudes, or reli-
ance on contents?
A claim of reliance on intuitive contents would be hard to deny. If we admit that
there are such things as intuitions, and that often, we have intuitions with contents
that are used as premises in armchair philosophical arguments, then we’ve already
admitted that armchair philosophy relies on ‘intuition’ in this sense. For instance, it
is overwhelmingly plausible that some armchair philosophy relies on the fact that
‘Gödel’ refers to Gödel, even if it turns out he didn’t really prove the incompleteness of
arithmetic.5 It is also overwhelmingly plausible that many of us have the intuition that
‘Gödel’ refers to Gödel, even if it turns out he didn’t really prove the incompleteness of
arithmetic.6 So we often rely on intuitive contents. This verdict is as uninteresting as
it is obvious; philosophers rely on intuitions in this sense the same way philosophers
(along with everybody else) rely on beliefs—this shows us nothing interesting about
the epistemology or the appropriate methodology of philosophy.
The more interesting claim is that philosophers rely on those mental states that
are intuitions. At least some defenders and critics of armchair philosophy alike have
understood it as so relying. For example, Weinberg et al. (2008) write:
The family of strategies that we want to focus on all accord a central role to what we will call
epistemic intuitions. Thus we will call this family of strategies Intuition Driven Romanticism (or
IDR). As we use the notion, an epistemic intuition is simply a spontaneous judgment about the
epistemic properties of some specific case—a judgment for which the person making the judg-
ment may be able to offer no plausible justification. To count as an Intuition Driven Romantic
strategy for discovering or testing epistemic norms, the following three conditions must be satis-
fied: (i) The strategy must take epistemic intuitions as data or input. (It can also exploit various
other sorts of data.) (p. 19, my emphasis)
5 Kripke (1980).
6 It is uncontroversial that many of us have this intuition, but Edouard Machery et al. (2004) argue that
our intuitions may be more esoteric than has generally been assumed; Ron Mallon et al. (2009) argue that
this undermines many philosophical arguments. Against the former, see Genoveva Martí (2009); against
the latter, see Ichikawa et al. (2012).
I take this passage as committing their target to reliance on ‘intuition’ in the mental-
state sense. Intuition is here identified with judgment. Of course, ‘judgment’ is suscep-
tible to the same readings as ‘intuition’ is, but the psychological qualifier ‘spontaneous’
forces the mental-state reading.
Weinberg et al. (2008) are concerned to cast doubt on armchair philosophy. But
some defenders of the armchair, too, have committed to the significance of intuition in
this more interesting sense. Here, for instance, is Joel Pust (2000):
Here is a case (derived from Lehrer . . . ) from that massive literature:
Nogot’s Ford. Suppose your friend Nogot comes over to your house to show you the new Ford
automobile he has just purchased. [standard Gettier story omitted] . . . . Do you know that a
friend of yours owns a Ford?
Most philosophers take the fact that they have the intuition that S does not know that p in this
case to show that S does not know that p. (p. 5)
Pust closes here with a sociological claim: most philosophers take the fact that they have
a particular intuition to demonstrate a philosophical thesis. He offers no defense of this
empirical claim, apparently considering it obvious. It is certainly true that most philoso-
phers take S not to know that p in this case; and it is also certainly true that most philoso-
phers have the intuition that S does not know that p in this case. But should we accept Pust’s
claim that most philosophers take the fact about their own mental states to show that the
fact about S is true? I suggest not. Suppose a philosopher is asked to defend the judgment
about S. The appropriate response would be to cite, for instance, the fact that S’s belief that p
was derived from a falsehood, or that he was lucky to have gotten his belief right. It is prob-
ably true that, upon sufficient questioning, many philosophers might ultimately exclaim,
“I just have an intuition!” But it is not obvious that this must be an attempt to explicate the
evidence; it might well just be an attempt to end the dialectical train of inquiry. Certainly
much more can and should be said here; see Cappelen (2012) and Part III of Ichikawa and
Jarvis (2013) for more comprehensive arguments for the view of this paragraph.
Very shortly after the passage quoted above is another, more telling, passage from
Pust that can provide insight into the mainstream diagnosis of ascribing importance to
the mental state, intuitions:
The analysis of justified belief proceeds in exactly the same fashion. A theory is proposed . . . and
it is tested by its ability to account for intuitive judgments regarding the justifiedness or unjusti-
fiedness of particular actual and hypothetical beliefs. That this is so is recognized by many phi-
losophers who reflect on their practice. For example, the epistemologist John Pollock claims that
in epistemological analysis:
[O]‌ur basic data concerns what inferences we would or would not be permitted to make
under various circumstances, real or imaginary. This data concerns individual cases and our
task as epistemologists is to construct a general theory that accommodates it. (p. 5, emphasis
in original)
Pust cites Pollock as an example of an epistemologist who reflects on his own practice,
and recognizes the crucial role that intuitions play in it; yet the quotation Pust selects
does not include the word ‘intuition’ or any of its cognates—nor, indeed, does Pollock
describe any psychological states as data. According to Pollock, the basic data are the
acceptability of inferences. But Pust takes away the moral that Pollock recognizes that
the basic data are intuitive judgments. I can see this only as a kind of equivocation on
the two senses of ‘intuition’ outlined above.
According to a psychologistic reading, the intuition that p is evidence for q (or data
in favor of q, or something that shows that q, etc.) just in case the proposition that I
have the intuition that p is important evidence (etc.) for q. Pust clearly has in mind—at
least sometimes—this psychologistic reading. He is sometimes explicit, as when he
says, as quoted above, that “most philosophers take the fact that they have the intuition
that [q] to show that [q].” The reading is also explicitly endorsed by Weinberg et al.
(2008), who characterize intuition-driven romanticism as a methodology that takes
facts about intuitions as inputs, and generates philosophical theories on the basis of
these psychological data.
But we have seen there is a more modest reading available for “the intuition that p is
evidence for q.” It may mean merely something like the claim that the intuited proposi-
tion—namely, p—is evidence for q. It is on the weaker reading that the Pollock quote
given by Pust plausibly lends credibility to the claim that Pollock treats intuitions are
evidence in his epistemology. That such-and-such is permissible—an intuited proposi-
tion—is the starting point for Pollock’s theorizing.
It’s undeniable that philosophers rely on many propositions that are intuitive; but
it does not follow that philosophers rely on psychological states, intuitions, and allow
them to play evidential roles. Neither does it follow that intuitive propositions are
either available or relied upon because they are intuitive. So let’s distinguish three kinds
of metaphilosophical claims:
1. Intuited contents are (often) taken as important evidence/reasons/data/input in
armchair philosophy.
2. Intuited contents are (often) taken as important evidence/reasons/data/input in
armchair philosophy because they are intuited.
3. Intuition states, or facts about intuition states, are (often) taken as important
evidence/reasons/data/input in armchair philosophy.
Once these three claims about intuitions are distinguished, it becomes much less obvi-
ous whether there is any generally widespread commitment to the stronger claims
about the roles of intuition in philosophy. Even many of those who have done the most
to emphasize the role of intuitions are not obviously committed to ‘intuition’ in the
state sense as playing such a role; George Bealer (1998), for instance, takes care to dis-
tinguish his central position from (3):
When I say that intuitions are used as evidence, I of course mean that the contents of the intui-
tions count as evidence. When one has an intuition, however, often one is introspectively aware
that one is having that intuition. On such an occasion, one would then have a bit of introspective
evidence as well, namely, that one is having that intuition. Consider an example. I am presently
intuiting that if P then not not P. Accordingly, the content of this intuition—that if P then not
not P—counts as a bit of my evidence; I may use this logical proposition as evidence (as a reason)
for various other things. In addition to having the indicated intuition, I am also introspectively
aware of having the intuition. Accordingly, the content of this introspection—that I am having
the intuition that if P then not not P—also counts as a bit of my evidence; I may use this proposi-
tion about my intellectual state as evidence (as a reason) for various other things. (p. 205)
It is not entirely obvious from this passage what Bealer’s considered attitude toward
(3) is; it is clear that (3) does not capture his central view about intuition, but it is not
clear whether he thinks it is true; whether he does depend on which ‘various other
things’ he has in mind.7
Explicit commitments to (3) by armchair philosophers are hard to come by, setting
aside one notable class of exceptions. Some philosophers explicitly sign on to a project
in which facts about intuitions are important evidence, and are right to do so. I have
in mind philosophers who study psychological matters like concepts (or intuitions
themselves). So rather than, for instance, studying the nature and grounds of human
knowledge, one might be more interested in questions about epistemic concepts. Alvin
Goldman (2007) is an exemplar of this approach.8 The project of articulating the nature
and application of epistemic concepts, which has a perfectly obvious and respectable
use for psychological facts about intuitions, is not the project that presently concerns
me in this chapter.
When the subject matter is non-psychological, this sort of emphasis on psycho-
logical intuitions does appear misplaced. And it’s not at all clear that much actual
practice does include such an emphasis. Cappelen (2012), Williamson (2007), and
Deutsch (2010) defend both these claims; their arguments seem to me largely cor-
rect. I see no reason to accept (3). Contrary to the suggestion of Weinberg et al.
(2008), very little contemporary analytic philosophy takes psychological intuitions
as central inputs.
A thorough consideration of (2) is beyond my present scope,9 but my own view is
that (2) and (3)—the only principles on the table that provide an important role to
intuition states—are both false, at least when used to describe the best examples of
armchair philosophy in the literature. I shall call any view that endorses (2) or (3) one
in which intuitions play a central role in armchair philosophy; it will be a working
assumption for much of this chapter that intuitions do not play a central role.
7 Does Bealer endorse (2)? It’s not entirely clear. In his (2002) he glosses his view as the claim that ‘by
virtue of having an intuition that p, one has a prima-facie reason or prima-facie evidence for p’ (p. 74). This is
not a statement of (2), which would say that in the relevant circumstances, one would have p as evidence, not
merely have evidence for p. Still, it may be that (2) is the best view to attribute, given this quotation and the
one cited in the main text.
8 An interesting historical note: Weinberg, et al. (2008) was originally published (in 2001) as part of a
special issue on the Philosophy of Alvin Goldman; Goldman’s (2001) reply to Weinberg, Nichols, and Stich
consists largely in the clarification that his work is to be understood as engaging with the mentalistic project
discussed here.
9 Chs. 12–13 of Ichikawa and Jarvis (2013) give an extended argument against (2).
So it is at best less than obvious that those psychological states that are intuitions
have an important role to play in the epistemology of armchair philosophy. What bear-
ing has this on the experimentalist critique?
12.3. Does the Experimentalist Critique Need

Armchair Philosophy to Need Intuition?
Prima facie, the experimentalist critique looks to be one that challenges intuitions. So
it is easy to think that it constitutes a significant challenge to armchair methodology
only if the latter relies on such intuitions.
Weinberg et al. (2008), in the passage quoted above, clearly presses a version of the
experimentalist critique that commits to the use of psychological data as inputs into
the mechanism of theory construction. Much of the ensuing literature in the critical
experimentalist spirit has followed suit in emphasizing intuitions. So it is not surpris-
ing that, for instance, Herman Cappelen takes the rejection of the centrality of intui-
tion to neutralize the experimentalist critique:
The Big Objection to experimental philosophy is easy to state and should be obvious: philoso-
phers don’t rely on intuitions about thought experiments, so studies of intuitions people have
about thought experiments have no direct relevance for philosophical arguments or theoriz-
ing. . . . In short: If philosophers don’t rely on intuitions, then the project of checking people’s
intuitions is philosophically pointless. (Cappelen 2012, pp. 221–2)
Max Deutsch (2009, 2010) defends similar claims; for example:

Philosophers need not assume that their own intuitions about cases are universal. So surveys
showing them that they are not universal are irrelevant. Majority opinion does not determine
the truth, or constitute the primary source of evidence in philosophy, and despite appeals to
‘what we would say’ about cases, majority opinion has never been thought to play these roles in
philosophical argument. (Deutsch 2009, p. 465)
Experimentalist critics are not, generally speaking, much impressed by this sort of
move;10 a central aim of the present section is to explain why not. Although I’m in
quite a lot of agreement with the line pressed by Cappelen and Deutsch, there is,
I think, something to be said for limiting its significance against the experimentalist
critique.
Alexander and Weinberg (2007) consider Williamson’s rejection of the idea that
psychologistic facts about intuitions are given strong evidential significance in arm-
chair philosophy. They offer two responses in succession. The first, I think, largely
misses the point:
Timothy Williamson has also developed a more radical response to the restrictionist
threat: rejecting the picture of philosophical practice as depending on intuitions at all! He argues
10
See e.g. Alexander (2010), Alexander and Weinberg (2007, p. 72), Weinberg (2009).
that our evidence, in considering [philosophical thought experiments] . . . is not any sort of men-
tal seeming, but the facts in the world. He compares philosophical practice to scientific practice,
where we do not take the perceptual seemings of the scientists as our evidence, but the facts
about what they observed. Similarly, then, we should construe Gettier’s evidence to be not his
intellectual seeming that his case is not an instance of knowledge, but rather the modal fact itself
that such a case is not an instance of knowledge . . . .
But we do not think that Williamson’s arguments can provide much solace for traditional ana-
lytic philosophers. For the results of experimental philosophers are not themselves framed in
terms of intuitions, but in terms of the counterfactual judgments of various subjects under vari-
ous circumstances. Although the results are often glossed in terms of intuitions to follow stand-
ard philosophical usage, inspection of the experimental materials reveals little talk of intuitions
and mostly the direct evaluation of claims. (p. 72)
It is true that part of Williamson’s project consists in a claim about the nature of intui-
tions: they’re mere judgments. But this is not closely related to his strategy outlined in
the first quoted paragraph: that of denying that philosophical evidence is ultimately
psychological. As such, the choice of psychologistic terms—intuitions, counterfactual
judgments, or whatever you like—is irrelevant to the question of the sources of our
philosophical knowledge. Alexander and Weinberg say that the experimentalist results
can be described as measuring ‘counterfactual judgments’ just as well as they meas-
ure ‘intuitions’—but each is equally psychological, and each is, on the Williamsonian
approach, equally denied status as the philosopher’s ultimate evidence. Gettier’s evi-
dence, on Williamson’s view, is neither his intellectual seeming as though there is no
knowledge in the relevant case, nor his state of counterfactual judgment that there
would be no knowledge in the relevant case; it is the fact itself: that there is no knowl-
edge in the relevant case.
If we keep this fact in mind, it’s easy to see that the last quoted sentence, in its relevant
interpretation, is false. Experimental philosophers do not typically make direct evalua-
tions of the relevant claims; their data is behavioral and psychological: how do subjects
react—what intuitions do they have, or what judgments do they make, or which option
do they circle, and sometimes: how confident are they about their answers? In theory,
one could devise an experimental philosophy study that directly investigated the rel-
evant claims. For instance, one might take subjects from various ethnicities and sub-
ject them to Gettier cases. They’d be given misleading evidence, such that they formed
false beliefs, and then prompted in such a way so as to induce them to infer truths from
those falsehoods. One might then observe whether members of one ethnicity were
more likely to have knowledge than the other. Such would be a cross-cultural study of
Gettier cases—are Western subjects more or less likely than East Asian subjects to have
knowledge in Gettier cases?—that involved direct evaluation of philosophically rel-
evant claims. But experiments like this one are, of course, far from the norm in experi-
mental philosophy.
Alexander and Weinberg go on immediately from the previous passage to make a
second argument; this one, I think, is much more germane.
The restrictionist challenge does not need to turn on a (potentially mistaken) psychologization
of philosophers’ evidence; that it does not turn on that skeptical move hopefully helps make
clear that it is not itself a skeptical challenge. In terms that Williamson should be happy with, the
challenge reveals that at the present time philosophers may just not know what their evidence
really is. And the true extent of their evidence is not, we think, something that they will be able to
learn from their armchairs. (p. 72)
The idea here seems to be that the experimentalist critique provides philosophers with
reasons to doubt the epistemic status of the premises on which they rely, and that these
challenges demand a fairly significant shift toward a more empirical methodology.
This is a recognizable and respectable form of critique, and one worth taking seriously.
And Alexander and Weinberg are correct to say that it does not depend on whether
philosophy proceeds from psychologistic evidence or sources of evidence. The form of
this sort of critique is perfectly general.
Is it too general? There are two sorts of reasons one might be worried. One is that
the skeptical implications of the concern extend too far, and will undermine all of our
methods—not just within armchair philosophy. I won’t go into this important ques-
tion here.11 A second worry about generality is that it doesn’t make sufficient sense
and contact with the original presentations of the experimentalist critique. Why do so
many papers, like the ones cited above, seem to commit to a psychologistic conception
of philosophical evidence, if such is irrelevant to the critique itself?
The answer to this question is that ‘the experimentalist critique’ has shifted over
time. There are now two importantly distinct families of critique in the experimentalist
tradition. Although they have not been clearly distinguished, even by their practition-
ers, they seem to me to have little to do with one another. We can make philosophical
progress by distinguishing them rather sharply.
12.4. Which Experimentalist Critique?

The passage from Alexander and Weinberg suggests a general form of the experimen-
talist critique: experimental data provides us with reason not to trust our armchair
judgments about philosophical thought experiments—whether you call them ‘intui-
tions’ or not, and whether you think they proceed from psychologistic premises or not.
Empirical data about human reliability can provide undercutting defeaters for judg-
ments of any kind. Call this the defeater critique. The defeater critique sees traditional
armchair methods as insufficiently self-reflective; armchair philosophers proceed
without enough regard for their abilities to discern relevant facts. Twentieth-century
analytic philosophy, on this view, proceeded in something of a dogmatic slumber.
When epistemologists first encountered Gettier cases, it seemed obvious to them
11 See e.g. Weinberg (2007); Williamson (2004); and Sosa (2007). For my take on this issue, see Ichikawa
(2012).
that they were cases of ignorance rather than knowledge; but maybe they should have
slowed down a bit before just assuming they were right.
Empirical data can certainly bear on this sort of claim, and it’s not too hard to imag-
ine uncovering rather strong reasons for worry. It will come as no surprise to any-
one that our philosophical capacities are fallible—sometimes we make mistakes—but
many specific ways in which we are fallible could well turn out to be surprises. And at
least in theory, these surprises could easily bring with them radical methodological
consequences for armchair philosophy. Consider the various sorts of fallacies to which
we humans are sometimes susceptible. The extent to which we are subject to these fal-
lacies is an empirical question; so too is the question, under what circumstances are we
better and worse at avoiding them.
To take a rather extreme example for the sake of illustration, we might discover
through empirical investigation that armchairs like the one I’m now sitting in bring
out the worst in our cognitive abilities. Something about the experience of sitting in an
armchair affects our brains in a way that causes us to be worse at philosophy than we’d
be if we were standing, or sitting on a bench. This would be a very strong vindication of
the defeater critique. Empirical evidence to this effect would certainly undermine our
rational confidence in much philosophy, and it would motivate a change in methodol-
ogy. It would enjoin us literally to leave the armchair.
The actual data, of course, are not nearly as clean as in this hypothetical case—
which is why the soundness of the defeater critique is controversial—but the model
is the same. Experimentalists uncover data about a certain kind of unreliability, and
suggest that it is infecting philosophical judgment in a way that should undermine
our confidence in our own judgments. For example, Swain et al. (2008) uncover data
that philosophical intuitions are unduly influenced by the order in which cases are
considered: subjects are more likely to attribute knowledge in a tricky case—a fake
barn case or a Truetemp case—if they’ve recently been asked about an obvious case
of ignorance than they are if they’ve just been asked about an obvious case of knowl-
edge. That casual judgment is susceptible to such irrelevant features should not be
particularly surprising; there is strong independent reason to believe that humans
are susceptible to many such kinds of performance errors.12 Nevertheless, the point
that philosophers must not blindly stick to whatever philosophical intuitions they
find themselves attracted to is surely right. We should be circumspect in our philo-
sophical judgments, especially in situations where we are particularly prone to error.
Here, empirical data can surely help us to improve our epistemic positions, by helping
us to identify the fallacies and biases to which we are prone. Insofar as the relevant
discoveries are distinctively empirical, and insofar as they constitute a significant rea-
son to worry about armchair methodology, we have empirical data that rationally
12 For a nice summary, see Stich (1990, pp. 4–9) citing Wason and Johnson-Laird (1970), Tversky and
Kahneman (1983), and others.
undermines confidence in our philosophical judgments. Many other experimentalist

arguments also fit this mold.13
The more modest form of the defeater critique runs on the basis of particular falla-
cies discovered by particular experiments: we’ve discovered that factors X, Y, and Z
negatively influence philosophers’ moral judgments, so, to the extent possible, phi-
losophers ought to try to avoid relying on moral judgments in the presence of X, Y, and
Z. Depending on the particular factors, doing so may itself require ‘leaving the arm-
chair’ and using empirical methods; even if it does not, we’ve already left the armchair
in a broader, methodological sense, since it was empirical research that taught us to be
careful around X, Y, and Z.
A less modest version of the defeater critique makes an additional inductive
step: since we’ve discovered, empirically, so many surprising factors—X, Y, and Z—
that interfere with our philosophical abilities, there’s good reason to suspect there
could be lots more similar factors that we haven’t found yet. We therefore should not
rest with our armchair judgments at all, until we’ve done quite a bit more empirical
work identifying what does and what does not impair us.
Many words have been written about the defeater critique; I’ll not give it too many
more just now. I am interested in distinguishing it from another, rather different,
experimentalist critique of armchair philosophy.
12.5. The Arbitrariness Critique

I have agreed with the suggestion of Alexander and Weinberg that there is a recogniz-
able empirical critique of armchair methodology that does not depend on the latter’s
granting of a substantive role for intuitions. I do not think, however, that this shows
that those who have attempted to defend armchair philosophy by resisting the psy-
chologization of philosophical evidence have been spinning idly. For there is another
experimentalist critique in the literature that operates quite differently, and indepen-
dently, from the defeater critique; this, which I call the arbitrariness critique, may well
depend on problematic views about the role of intuitions in philosophy.
The arbitrariness critique is emphasized most prominently in the work of Stephen
Stich, although other experimental philosophers have also pressed versions of it—
sometimes combining it, without perhaps quite realizing it, with the defeater critique.
Given the influential formative role Stich played in the experimental philosophy
movement, it is not surprising that his version of the critique should have held such
influence. However, I believe that the critical experimentalist literature, on the whole,
13 To offer just a couple of examples: Tamara Horowitz (1998) impressively employs Daniel Kahneman and
Amos Tversky’s Prospect Theory, an empirical theory about how humans reason with risks and probability,
to discredit certain deontological intuitions in normative ethics. Tamar Gendler (2007) offers a catalogue of
similar projects, and Gendler (2002) is her own attempt to discredit, on empirical grounds, a particular sort
of intuition about personal identity.
is helpfully thought of in broad terms as having shifted its focus from the arbitrariness
critique to the defeater critique.
What is the arbitrariness critique?
It is worthwhile to return to the earlier presentations of the experimentalist critique;
let us focus for a time on Weinberg et al. (2008).14 I’ve already remarked on one odd
feature of their argument, with respect to understanding them as offering the defeater
critique: the focus on psychologistic intuitions does not seem relevant. Empirical stud-
ies about biases and errors can cast doubt on all kinds of judgments, not merely those
that are products of processes that take intuitions as inputs. Another surprising feature
of Weinberg et al.’s project, from this point of view, is the focus on distinctively norma-
tive questions. Their article, “Normativity and Epistemic Intuitions,” would have rea-
son to focus on neither normativity nor intuition, were it a presentation of the defeater
critique. Their project is different.
Weinberg et al. focus in particular on armchair epistemology. Because of the empha-
sis on the normative, it is not obvious to what extent the arguments might generalize
to other areas in philosophy. Other normative domains, like ethics, may be treatable in
the obvious way, but it is not clear how or whether the arbitrariness critique, as I under-
stand it, could be extended to, for instance, metaphysics or the philosophy of language.
In the rest of this chapter, I will follow Weinberg et al. in their focus on normative epis-
temology. They explicitly concern themselves only with that branch of epistemology
that “attempts to establish norms to guide our epistemic efforts,” (2008, p. 18) answer-
ing the question: “How ought we to go about the business of belief formation and revi-
sion?” (p. 19).
The worry is particular to these normative questions. Weinberg et al. charge that an
intuition-based methodology has insufficient resources to answer them. Sure, we could
rely on our intuitions about whether a subject is believing appropriately, or whether a
state is knowledge. But intuitions like these are the products of minds that are heav-
ily influenced by idiosyncratic features of our languages and societies. We could have
grown up with any number of different sets of codes and norms; why think that the
ones we happen to have provide any guidance about what we ought to do? They write:
There might be a group of people who reason and form beliefs in ways that are significantly dif-
ferent from the way we do. Moreover, these people might also have epistemic intuitions that are
significantly different from ours. More specifically, they might have epistemic intuitions which,
when plugged into your favorite Intuition Driven Romantic black box yield the conclusion that
their strategies of reasoning and belief formation lead to epistemic states that are rational (or
justified, or of the sort that yield genuine knowledge—pick your favorite normative epistemic
notion here). If this is right, then it looks like the IDR strategy for answering normative epis-
temic questions might sanction any of a wide variety of regulative and valuational norms. And
that sounds like bad news for an advocate of the IDR strategy, since the strategy doesn’t tell us
14 This paper was originally published in 2001; it was the first widely read and discussed version of the
experimentalist critique, and played a substantial role in advancing the experimental philosophy movement.
what we really want to know. It doesn’t tell us how we should go about the business of forming
and revising our beliefs. One might, of course, insist that the normative principles that should be
followed are the ones that are generated when we put our intuitions into the IDR black box. But it
is less than obvious (to put it mildly) how this move could be defended. Why should we privilege
our intuitions rather than the intuitions of some other group? (p. 22)
It’s worth noticing that the arbitrariness critique, as given here, is pressed in a broadly
armchair manner in Stich’s (1990). Weinberg et al. add to Stich’s earlier critique by
going on to present evidence suggesting that the hypothetical people they discuss are
actual, thus neutralizing, in their minds, the objection that the possibility of such indi-
viduals is too far-fetched to treat seriously.15 The results are well-known, even if the
context is not: they suggest that there is evidence that East Asian subjects have system-
atically different intuitions about central cases in epistemology.
So this, at its core, is the arbitrariness critique. We could have had any number of
different intuitions. And if we had different ones than our actual ones, then proper
exercise of the methods we use to make normative epistemic judgments would pro-
duce different judgments—perhaps even conflicting ones. It is in an important sense
arbitrary which intuitions we happen to have, and so it is arbitrary which normative
systems we endorse. But this is a situation that should make us deeply uncomfortable;
epistemic norms are too important to be so susceptible to matters of chance. So we’d
better use some other method to decide which norms to accept.16
We could compare the situation to that in other domains in which we are led by
arbitrary factors to subscribe to particular principles, such as the acquisition of lan-
guage, or norms of etiquette. We come to general norms of etiquette by systematizing
the behaviors and judgments we happen to be exposed to; I know to mail paper wed-
ding invitations, rather than emailing or just inviting people orally—this knowledge
comes from my observation of how such things are usually arranged, what is praised,
what is censured, etc. These features are to a significant degree arbitrary; if I happened
to live in a culture that did things differently, I’d have internalized different norms.
This verdict is comfortable, in the case of etiquette, in a way that it is not in the case of
epistemology. A culture of people who think unjustified beliefs are great is, we think,
making a mistake; we do not feel the same way about a culture of people who think oral
invitations to weddings are great.
As I have presented it—and as Weinberg et al. have presented it—the arbitrariness
critique does depend critically on the psychologistic role of intuitions in normative the-
ory construction.17 If something other than intuition—something less arbitrary—does
the relevant work in our choosing of a particular set of norms, then the set of norms we
have arrived at is not arbitrary: it is the one countenanced by that non-arbitrary factor.
15 They attribute such an objection to John Pollock and Joseph Cruz (1999).
16 What other method? Weinberg et al. (2008) don’t tell us; their project is negative. Stich (1990) endorses
a kind of pragmatism that is meant to fill this role.
17 Weinberg et al. (2008) are explicit in admitting this—see, for instance, the list of conditions on
their p. 20.
So it appears as if the arbitrariness critique is susceptible to the denial of the central-

ity of intuition, in a way that the defeater critique is not. The moves of Alexander and
Weinberg (2007), cited in section 12.3, do not defend the arbitrariness critique from
the denial of a central role for intuitions.
Is there another move available for the arbitrariness critique? Is there a way to press
it without relying on a central role for intuitions? In the final sections, I will consider
several strategies. None, I think, should be particularly troubling for the practitioner of
armchair philosophy.
12.6. Xenophobia and Conflicting Values

We value knowledge; that’s part of the reason we have a word for it. Suppose there
was another culture, rather different from ours, whose members did not value knowl-
edge. Their epistemic practices are different from ours. Now it looks like an individual’s
own emphasis on knowledge is an accident of birth. Our epistemic evaluations are
informed by our epistemic concepts, which are a product of contingent features of our
upbringing. Nichols et al. (2003) write that:
Without some reason to think that what white, western, high [socioeconomic status] philoso-
phers call “knowledge” is any more valuable, desirable or useful than any of the other commodi-
ties that other groups call “knowledge” it is hard to see why we should care if we can’t have it.
(p. 245)
In a similar vein, Stich (1990) writes:

[U]‌nless one is inclined toward chauvinism or xenophobia in matters epistemic, it is hard to see
why one would much care that a cognitive process one was thinking of invoking (or renouncing)
accords with the set of evaluative notions that prevail in the society into which one happened to
be born. (p. 94)
Suppose we learned that in some society, the word best translated as ‘knowledge’ car-
ried a different meaning from the ordinary English ‘knowledge’—perhaps their word
means justified true belief (JTB). (On one interpretation of Nichols et al., their find-
ings suggest that certain idiolects of English are like this.) What reason, they ask, do
we have to prefer our criterion of epistemic evaluation (knowledge) to theirs (JTB)?
To prefer in the absence of any such reason, we’re told, would be unwarranted xeno-
phobia. Once again, we’re presented with some evidence that such individuals are not
merely possible but actual; but even if this is mistaken, surely we could have used a
concept of evaluation like that. So isn’t it in some sense arbitrary that we don’t? What
reason have we to prefer the notion we happened to end up with to any other?
The first thing to say in response is that it’s a rather substantive assumption that our
thinking about knowledge is in the relevant sense arbitrary. If it is an objective fact that
knowledge is an important feature of anyone’s intellectual life—if knowledge turns out
to be, in the Lewisean sense, a very natural property—then the possibility of a society
getting by without thinking of it may become a bit more dubious. For example, if one
is sympathetic to Timothy Williamson’s project, emphasizing the centrality of knowl-
edge for many diverse properties—evidence, belief, assertion, reference, justification,
etc.18—then one will be more puzzled by the very idea of a thriving society that doesn’t
care about knowledge. For my part, I am rather attracted by these Williamsonian
views, although I won’t go into them further now.
A second plausible response to this challenge seems to me to be the pluralist one
suggested by Ernest Sosa: what’s to stop you from valuing JTB? Nothing at all. We value
all sorts of things; value whatever you want. This is consistent with continuing to value
knowledge. Sosa writes:
The fact that we value one commodity, called ‘knowledge’ or ‘justification’ among us, is no obsta-
cle to our also valuing a different commodity, valued by some other community under that same
label. And it is also compatible with our learning to value that second commodity once we are
brought to understand it, even if we previously had no opinion on the matter. (2009, p. 109)
This response strikes me as entirely correct. But Stich challenges this pluralistic line,
saying first that a satisfactory epistemology should supply norms of permission, not
merely norms of valuation, and second, that pluralism about norms of valuation is
implausible. Stich writes:
Norms of valuing do play a role in traditional epistemological debates, but they are not the only
sorts of norms that epistemologists have considered. As we noted earlier, Goldman insists, quite
correctly, that justification rules (or “J-rules”) play a central role in both classical and contempo-
rary epistemology, and J-rules specify norms of permissibility, not norms of valuing. They “per-
mit or prohibit beliefs, directly or indirectly, as a function of some states, relations, or processes
of the cognizer” (Goldman 1986: 60). When we focus on these rules, the sort of pluralism that
Sosa suggests is much harder to sustain. If a rule, like the one cited a few paragraphs back, says
that ceteris paribus we ought to hold a belief if it is an instance of knowledge, and if ‘knowledge’
is interpreted in different ways by members of different groups, then Sosa’s pluralism leads to
inconsistency. There will be some beliefs which we ought to believe on one interpretation of
‘knowledge’ but not on the other. (2009, p. 235)
There are several avenues of possible resistance available to the armchair philosopher.
First, Sosa’s pluralism—broadly the view that all parties speak truly in their norma-
tive epistemic claims—does not, contrary to Stich’s assertion, lead to inconsistency on
the assumption that ‘knowledge’ refers to different properties in different cultures, and
there is a rule that we ought to believe what we know. Stich writes that a rule “says that
ceteris paribus we ought to hold a belief if it is an instance of knowledge.” Stich is part
of our culture, and speaks our language, so his word ‘knowledge’ means knowledge. So
this is a rule about knowledge. How is the presence of other cultures in which speakers
mean other things by ‘knowledge’ supposed to lead us to inconsistency?
Williamson (2000). For the view about reference, see Williamson (2007, ch. 8).
18
I suspect that Stich is imagining that our alien friends also aver sentences like “there
is a rule that ceteris paribus we ought to hold a belief if it is an instance of knowledge.”
This is not stipulated to hold in the hypothetical case; and perhaps more significantly,
we have, so far as I can tell, no data whatsoever that points to anything like this actually
being the case. But if we suppose that members of alien cultures who mean other than
knowledge by ‘knowledge’ do utter sentences like that one, then a pluralistic line that
tries to make everybody right does face some prima facie tension. Even still, there are a
number of options available.
One option is to suggest that other words in the alien language differ from the cor-
responding words in English, just as ‘knowledge’ does. Perhaps their ‘rule’ or ‘ought’
or ‘believe’ means something that renders their sentence in no tension with ours.
A related option is to exploit the context-sensitivity of the shared modal ‘ought’—giv-
ing each the English stable character, but letting it express a different modal relation in
its own conversational context.19
Even if we granted that our aliens utter the same norm-sentences we do, and that
only ‘knowledge’ takes a different meaning between our two languages, and that no
contextual variation is active to help them play nicely together, the pluralist line would
still not be inconsistent. For the norm-sentences take the form of ceteris paribus rules;
it is of course a hallmark of ceteris paribus rules that they tolerate exception. If my
neighbors use ‘knowledge’ to pick out JTB, which they value, they and I can all endorse
and share the relevant ceteris paribus permissibility rules (once we work out how to
talk to each other):
Ceteris paribus, believe that p if and only if doing so will result in knowledge that p.
Ceteris paribus, believe that p if and only if doing so will result in JTB that p.
(Here, as in the rest of this chapter, I am writing in English.)
Incidentally, it is worth pointing out that it is not clear that there are many actual
practical circumstances in which these rules advise an agent in divergent ways; the
person who tries to live by the one will look quite a lot like the person who tries to live
by the other. (Ceteris paribus, JTB is knowledge!) This suggests that the extent to which
there is genuine conflict among plausible epistemic norms of permissibility may not be
as great as Stich assumes.
Of course, this defense of pluralism only holds if we treat the norms as ceteris paribus
rules. Could a version of the critique insist on absolute rules? Not very effectively, for
absolute rules of this sort are highly implausible. It is all things considered permissible
to believe unknowledgeably, or to withhold belief that could have been knowledge, in
some circumstances, even though doing so will violate some ceteris paribus epistemic
rules. Demanding that one believe everything knowable would place implausible
19 C. S. Jenkins and Daniel Nolan (2010) argue that some apparent moral disagreements within English are
resolvable in this way.
demands of attention; it would prohibit building houses in favor of counting bricks. As

Sosa points out, this is so even if we limit the relevant domain to the epistemic:
Silly beliefs about trivial matters can attain the very highest levels of epistemic justification and
certain knowledge even if these are not beliefs that one should be bothering with, not even if
one’s concerns are purely epistemic. (2009, p. 110)
We face no new challenges by being pluralists about epistemic value. Stich gives a sec-
ond objection to Sosa’s pluralist suggestion:
Moreover, even in the case of norms of valuing Sosa’s pluralism can lead to problems. Sosa is
surely right to claim that someone who values owning money banks can also value owning river
banks. But if there is one of each on offer and the person’s resources are limited, she will have to
make a choice. Which one does she value more? (2009, p. 235)
I say, with Sosa, that there’s nothing stopping me from valuing the other societies’ epis-
temic goods in addition to my own, if I can learn to think about them. If some people
or societies value true belief, or justified belief, or justified true belief, or belief derived
from a generally reliable source, or certainty, but don’t even have a word for knowledge,
we can all get along just fine, and even learn to value one another’s preferred states too.
Stich replies, but which do you value more? Since we are finite creatures with finite
resources, we must choose among the things we value; the pluralist hasn’t told us how
to adjudicate between valuable things—and traditional armchair epistemic meth-
odology doesn’t obviously have the resources to identify the appropriate criterion. It
should be clear that this is exactly analogous to the question just raised about ceteris
paribus rules.
My answer here is simple: I agree with Stich that traditional armchair epistemol-
ogy does not obviously provide the resources to adjudicate between different valuable
states, or conflicting ceteris paribus rules. But I very much doubt it ever pretended to.
Suppose we set aside questions about differing epistemic concepts; even if knowledge
is the only epistemic game in town, we still have to decide whether to read the ency-
clopedia or walk the dog. Is it even obvious that the sorts of value in play here are com-
mensurable? If Stich thinks it is a great scandal that traditional epistemology provides
no clear advice on this matter, he has broader expectations for epistemology than I.
I have so far been assuming that the rival epistemic goods, though not identical to
our epistemic goods, were not antithetical to them. If we value knowledge, and our
neighbors value truth, JTB, or certainty, then there seems to me to be no obstacle to
our sharing their values, as explained above. This case to me seems analogous to this
one: I like opera. I can get along just fine with Emily who likes Puccini operas, Andrew
who likes theatrical performances in general, and even Martin who likes basketball.
Not only are we peaceful neighbors, but we can even learn to appreciate one another’s
particular preferences, and share them to a large extent.
But could there be a person or society with vastly different epistemic values—val-
ues that are not only non-identical to ours, but in direct tension with them? Maybe
they value false beliefs, or unjustified ones. Or maybe they’re Pyrrhonians, who value
the complete absence of belief. This is more like the case where I like opera and my
neighbor demands total silence—our values just plain conflict. (Of course, there is no
evidence on the table that there are people or societies like this.) In this extreme case,
I must reject the values of my alien neighbors. But this need be no xenophobia—socie-
ties like the ones imagined, if they are possible, will be dramatically unsuccessful ones.
There are at least clear, non-arbitrary pragmatic grounds on which we can confidently
judge our practices superior to theirs.
12.7. Arbitrariness and the Value of Knowledge

Here is another argument relating arbitrariness and value that can be found in
Stich’s work.
Suppose we successfully articulated what rules we must follow in order for our
beliefs to fall under our concept knowledge; why should we care about follow-
ing those rules? What value is there in complying with a standard that happens to be
reflected in our language and society? Call this the ‘so what’ objection: now we know
what knowledge is. So what?
The critic pressing this ‘so what’ challenge to normative epistemology may well con-
cede that, for example, armchair resources are sufficient for knowledge that the tradi-
tional judgment about a Gettier case—that the subject’s state does not fall under the
everyday concept knowledge—is correct. In so doing, of course, he admits that we
know that the Gettier subject does not know. (All parties must agree that S knows that
p if and only if S’s relation to p falls under the concept knowledge, and that this fact is
easily known by those of us who have the concept.) But the critic says: so what? Stich
writes:
The analytic epistemologist proposes that our choice between alternative cognitive processes
should be guided by the concepts of epistemic evaluation that are “embedded in everyday
thought and language.” But this proposal is quite pointless unless we value having cognitive
states or invoking cognitive processes that accord with these commonsense concepts. And it
is my contention that when they view the matter clearly, most people will not find it intrinsi-
cally valuable to have cognitive states or to invoke cognitive processes that are sanctioned by the
evaluative notions embedded in ordinary language. (1990, p. 93)
The idea, I take it, is that the interesting questions of epistemology are normative—
they’re supposed to help us to know what sorts of beliefs to pursue. Knowing what
beliefs are sanctioned by our commonsense epistemic-evaluative concepts is no help
in this normative enterprise unless we have some reason to value having beliefs that are
so sanctioned; this, Stich says, is implausible.
The referent of the everyday concept knowledge is knowledge. This is easy for me
to know; it follows straightforwardly from the fact that I am employing the everyday
notion in thinking that thought and writing that sentence. If we keep this in mind,
I think it should be clear that the value Stich attributes to the analytic epistemologist—
according with the standards of everyday thought and language—is optional as the epis-
temologist’s object of value.
Here is a traditional view: knowledge is valuable. The attempt to explain the value
of knowledge has occupied considerable attention from epistemologists since Plato.
Among the candidate explanations are suggestions like: knowledge is the norm
of assertion; knowledge is the norm of action; knowledge helps the subject achieve
his interests; knowledge is a more stable kind of true belief; knowledge is part of the
Platonic Good; knowledge is a successful achievement of a characteristically human
performance.
Does Stich’s argument cast doubt on the cogency of the project of treating knowl-
edge as valuable, and seeking the explanation for that value? I agree with Stich that it
would be an odd creature indeed who placed great value in the state of according with
the standards of everyday thought and language. Call this state Φ. Such a valuation is
not incoherent, but it does appear ill-motivated. It is no great defense of traditional
epistemology if it leaves the value of knowledge like that.
But one needn’t value Φ to value knowledge. Although in fact, in the actual world,
all and only people with Φ have knowledge, Φ and knowledge are, of course, not the
same property. They have different modal profiles; the biconditional that one has
knowledge iff one has Φ is only contingently true—there are possible worlds where
knowledge is not what it takes to accord with the standards of everyday thought and
language.
Etiquette norms (the ones around here) dictate that wedding invitations be sent
by post. Many of us value acting in accordance with those norms. There is at least
instrumental value in complying with the rules of etiquette in one’s society; perhaps
there is intrinsic social value as well. But the way in which we value mailing invi-
tations is contingent on the rule being as it is. The way we value epistemic norms
are different. Knowledge is valuable, regardless of what epistemic ideals happen
to be coded into our language. The disanalogy is especially apparent in divergent
counterfactuals:
If our social norm were to send wedding invitations by singing telegram, instead of
by post, there would be no etiquette value to sending them by post.
If our social norm were to have beliefs that are justified true beliefs, instead of knowl-
edge, there would be no epistemic value to knowing.
Many of us will accept the first but not the second. That second has some of the feel of:
If our social norm were to torture animals whenever we were bored, there would be
no moral value in refraining from torturing animals when we were bored.
I take it just about everybody who thinks there is actual moral value in refraining from
torturing animals rejects this one.
Another way to see this point is to observe how far Stich’s argument, if sound, would
generalize. Take whatever candidate for value that you like—desire-satisfaction, or
pleasure, or eudaimonia, or true belief, or whatever you find most plausible. Stich’s
counterpart can argue:
You propose that our choice between alternate courses of action should be guided by
what falls under our concept pleasure. But this proposal is quite useless unless we
value having states that accord with this commonsense concept. It is my contention
that when they view the matter clearly, most people will not find it intrinsically valu-
able to have states that are sanctioned by the pleasure concept that happens to be
embedded in ordinary language.
It is a mistake to argue, from the premise that it is implausible to value matching the
ordinary concept of X-NESS, to the conclusion that it is similarly implausible to value
x-ness.
12.8. Arbitrariness and the Interest of Knowledge

So hypothetical or actual diversity with respect to which mental states are valued, or
feature into rules, does not undermine the value of knowledge, or the plausibility of
epistemic rules that feature knowledge. But there is another way in which one might be
led from considerations of such diversity to an arbitrariness-based concern about nor-
mative epistemology: we might worry, not whether knowledge is a valuable state to be
in, but whether knowledge is a theoretically interesting state worthy of philosophical
study. (And so, mutatis mutandis, for various other epistemic, or more broadly philo-
sophical, notions.)
I have not found an explicit presentation of this kind of line in the extant lit-
erature pressing the arbitrariness critique, but it is perhaps a natural fallback to
the concerns about the value of knowledge itself, refuted in the previous section.
If knowledge is but one of many similar epistemic properties that have epistemic
value and figure into epistemic rules, does it warrant the philosophical energy arm-
chair philosophers have spent upon it? Are we being too narrow-minded—perhaps
even xenophobic—by studying the epistemic state we happen to talk about, at the
expense of others?
We do have limited cognitive resources, and so when we choose to study something,
it is at least often at the expense of studying something else. And the discovery of peo-
ple that seem to do pretty well, focusing on X instead of Y, might well give us reason
to shift some of our Y-attention to X: maybe they’re onto something. But this does not
constitute a serious threat to armchair philosophy, for three reasons.
First, pluralism in philosophical subject matters is just as plausible as is pluralism
about epistemic value. The issues here largely parallel those discussed in the discussion
of pluralism in section 12.6. Although epistemologists often are particularly interested
in knowledge, they need not be, and are not, exclusively so interested.20 The study of
knowledge just doesn’t preclude the study of anything else.
Second, armchair philosophy includes the resources for adjudicating between the
studies of various subject matters. Indeed, as in the case of the value of knowledge
itself, we face the question of how to weigh the value of studying knowledge against
other values—including the value of studying other philosophical topics—regard-
less of experimentalist concerns about cognitive diversity. It is interesting and worth-
while to study knowledge; it’s also interesting and worthwhile to study happiness (and
to study earthquakes, and to read novels . . . ). How do we decide where to focus our
energy? Williamson (2000) argues forcefully that knowledge is of strong theoretical
interest; of course this can be contested (e.g. by Mark Kaplan 2003). The central point
for present purposes is that it can be, and is, contested and evaluated within armchair
philosophy.
Third, the methodological objection is specific to its particular subject matter; diver-
sity with respect to knowledge challenges the value of studying knowledge; this objec-
tion should be evaluated entirely independently from one that challenges the value
of studying true belief, or justified belief, or safe belief (or, for that matter, virtuous
action, causation, reference, intentionality, parthood, etc.). So even if, contrary to what
I have just suggested, cognitive diversity implies an arbitrariness that undermines the
importance of studying knowledge, this is no objection to armchair philosophy on the
whole—or even to that subset of it which is normative—or even to that which is both
normative and epistemic. One is not challenged qua analytic philosopher by engag-
ing with an argument that a particular subject matter is uninteresting. It is well within
the realm of traditional methodology to argue that a philosophical subject matter is
unworthy of especial study.
I see no threat from arbitrariness against the value of studying knowledge.
Furthermore, there is no generalized threat from arbitrariness about the ability of the
armchair philosopher to focus her work in interesting areas. And to summarize the
conclusion of sections 12.6–12.8, there does not seem to me to be any obvious way to
press the arbitrariness critique in a way troublesome for the armchair philosopher
without reliance, as in Stich’s original presentations of it, on the centrality of intuition.
So rejecting centrality is an effective way of resisting the arbitrariness critique.
12.9. Conclusion
There is more than one ‘experimentalist critique’. At least two have developed in the
literature over recent years—I have called these the defeater critique, and the arbitrari-
ness critique. They ought to be clearly distinguished. Indeed, other than making use of
20 Two quick examples of epistemological projects which do not focus on knowledge are Crispin Wright
(1991, p. 88) and Laurence BonJour, in BonJour and Sosa (2003).
at least some of the same kinds of data, and targeting features of the same practice, they
have little in common with one another.
In particular, questions about the epistemic role of intuitions in the methodology of
armchair philosophy are irrelevant to the defeater critique. Experiments that demon-
strate limits to human philosophical abilities have relevance to human philosophical
practice, regardless of what roles intuitions do or do not play in that practice.
However, when it comes to the arbitrariness critique, the centrality of intuitions
seems to play a much more important role. I considered several ways of attempting to
advance the arbitrariness critique without assuming intuition centrality; none were
found troublesome. So if, as suggested in section 12.2, intuition centrality is false, then
the armchair philosopher needn’t be worried by the arbitrariness critique.
References
Alexander, J. (2010). “Is Experimental Philosophy Philosophically Significant?,” Philosophical
——., Mallon, R., and Weinberg, J. M. (2010). “Accentuate the Negative,” Review of Philosophy
and Psychology 1, pp. 297–314.
——., and Weinberg, J. (2007). “Analytic Epistemology and Experimental Philosophy,”
Philosophy Compass 2, pp. 56–80.
Bealer, G. (1998). “Intuition and the Autonomy of Philosophy,” in M. DePaul and W. Ramsey
Lanham, MD: Rowman & Littlefield.
——. (2002). “Modal Epistemology and the Rationalist Renaissance,” in T. Gendler and J.
Hawthorne (eds), Conceivability and Possibility. New York: Oxford University Press, pp.
71–125.
BonJour, L., and Sosa, E. (2003). Epistemic Justification: Internalism vs. Externalism, Foundations
vs. Virtues. Malden, MA and Oxford: Blackwell.
Cappelen, H. (2012). Philosophy without Intuitions: Oxford: Oxford University Press.
Chalmers, D. J. (1996). The Conscious Mind: In Search of a Fundamental Theory. New York: Oxford
University Press.
Cummins, R. (1998). “Reflections on Reflective Equilibrium,” in M. DePaul and W. Ramsey
Lanham, MD: Rowan & Littlefield, pp. 113–28.
Deutsch, M. (2009). “Experimental Philosophy and the Theory of Reference,” Mind and
Language 24, pp. 445–66.
——. (2010). “Intuitions, Counter-Examples, and Experimental Philosophy,” Review of
Philosophical Psychology 1, pp. 447–60.
Gendler, T. (2002). “Personal Identity and Thought Experiments,” The Philosophical Quarterly
52, pp. 34–54.
——. (2007). “Philosophical Thought Experiments, Intuitions, and Cognitive Equilibrium,”
Goldman, A. (2001). “Replies to Contributors,” Philosophical Topics 29, pp. 461–511.
Harman, G. (1977). The Nature of Morality. New York: Oxford University Press.
Hintikka, J. (1999). “The Emperor’s New Intuitions,” Journal of Philosophy 96, pp. 127–47.
Horowitz, T. (1998). “Philosophical Intuitions and Psychological Theory,” Ethics 108, pp. 367–85.
Ichikawa, J. J. (2012). “Experimentalist Pressure Against Traditional Methodology,” Philosophical
——. (2014). “Intuition in Contemporary Philosophy,” in L. M. Osbeck and B. S. Held (eds),
Rational Intuition: Philosophical Roots, Scientific Investigations. Cambridge: Cambridge
University Press.
——., and Jarvis, B. W. (2013). The Rules of Thought. Oxford: Oxford University Press.
——., Maitra, I., and Weatherson, B. (2012). “In Defense of a Kripkean Dogma,” Philosophy and
Jenkins, C. S., and Nolan, D. (2010). “Maximising, Satisficing and Context,” Noûs 44, pp.
451–68.
Kaplan, M. (2003). “Who Cares What You Know?,” Philosophical Quarterly 53, pp. 105–16.
Knobe, J. (2007). “Experimental Philosophy and Philosophical Significance,” Philosophical
Kripke, S. A. (1980). Naming and Necessity. Cambridge, MA: Harvard University Press.
Lycan, W. (1998). “Moral Facts and Moral Knowledge,” in W. Lycan (ed.), Judgement and
Justification. Cambridge: Cambridge University Press.
Machery, E., Mallon, R., Nichols, S., and Stich, S. (2004). “Semantics, Cross-Cultural Style,”
Cognition 92, B1–B12.
Mallon, R., Machery, E., Nichols, S., and Stich, S. (2009). “Against Arguments from Reference,”
Martí, G. (2009). “Against Semantic Multi-Culturalism,” Analysis 69, pp. 42–8.
Ethno-Epistemology,” in S. Luper (ed.), The Skeptics. Aldershot: Ashgate, pp. 227–47.
Pollock, J. L., and Cruz, J. (1999). Contemporary Theories of Knowledge, 2nd ed. Lanham, MD
and Oxford: Rowman & Littlefield.
Pust, J. (2000). Intuitions as Evidence. New York and London: Garland.
Putnam, H. (1973). “Meaning and Reference,” Journal of Philosophy 70, pp. 699–711.
132, pp. 99–107.
——. (2009). “A Defense of the Use of Intuitions in Philosophy,” in M. A. Bishop and D. Murphy
(eds), Stich and His Critics. Malden, MA: Wiley-Blackwell, pp. 101–12.
Stich, S. (1990). The Fragmentation of Reason: Preface to a Pragmatic Theory of Cognitive
Evaluation. Cambridge, MA and London: MIT Press.
——. (2009). “Replies,” in M. A. Bishop and D. Murphy (eds), Stich and His Critics. Malden,
MA: Wiley-Blackwell, pp. 190–252.
76, pp. 138–55.
Thompson, J. J. (1971). “A Defense of Abortion,” Philosophy and Public Affairs 1, pp. 47–66.
Tversky, A., and Kahneman, D. (1983). “Extensional Versus Intuitive Reasoning: The
Conjunction Fallacy in Probability Judgment,” Psychological Review 90, pp. 293–315.
Wason, P., and Johnson-Laird, P. (1970). “A Conflict Between Selecting and Evaluating
Information in an Inferential Task,” British Journal of Philosophy 61, pp. 509–15.
Weinberg, J. (2007). “How to Challenge Intuitions Empirically Without Risking Skepticism,”
——. (2009). “On Doing Better, Experimental-Style,” Philosophical Studies 145, pp. 455–64.
——., Nichols, S., and Stich, S. (2008). “Normativity and Epistemic Intuitions,” in J. Knobe and
S. Nichols (eds), Experimental Philosophy. New York: Oxford University Press, pp. 17–36.
First published as Weinberg, J., Nichols, S., and Stich, S. (2001). “Normativity and Epistemic
Intuitions,” Philosophical Topics 29, pp. 429–60.
——. (2004). “Philosophical ‘Intuitions’ and Scepticism About Judgement,” Dialectica 58, pp.
109–53.
Wright, C. (1991). “Scepticism and Dreaming: Imploding the Demon,” Mind 397, pp. 87–115.
13
Grasp of Essences versus
Intuitions: An Unequal Contest
E. J. Lowe
13.1.
One currently popular methodology of metaphysics has it that ‘intuitions’ play an evi-
dential role with respect to metaphysical claims somewhat analogous to that suppos-
edly played by experimental observations with respect to the hypotheses of theoretical
science. I believe that such a view is fundamentally misguided and leads inexorably
to an anti-realist conception of metaphysical claims. Instead, I shall defend a radi-
cally different and uncompromisingly realist methodology of metaphysics: one which
implies that any rational being, simply in virtue of being rational, is necessarily capa-
ble of grasping the essences of at least some mind-independent entities. The notion
of essence in play here is basically an Aristotelian one, whereby an entity’s essence
is captured by an account of what that entity is, or what it is (or would be) to be that
entity. In short, it is captured by a so-called ‘real definition’ of the entity in question.
Grasp of an entity’s essence—knowing what that entity is or would be—is not, how-
ever, to be construed simply in terms of the possession or mastery of a certain con-
cept or family of concepts relating to that entity. Concepts are just ways of thinking
of things and, as such, may or may not be adequate to the real natures—that is, to the
essences—of the things being thought of. Our ‘intuitions’, particularly as revealed by
aptly constructed thought experiments, may indeed cast interesting light on some
of our concepts, but they are not are cannot be reliable guides to mind-independent
truths of essence. A rational being must no doubt possess some suitable mental capac-
ity enabling it to attain knowledge of such truths, if indeed it does so—call it ‘rational
insight’, if you will—but the exercise of this capacity cannot consist in the eliciting of
intuitions and reflection on them. Intuitions as such are just psychological states of
rational subjects, but no such state can of itself constitute reliable evidence for the cor-
rectness of a rational subject’s judgement concerning the nature or essence of some
mind-independent entity. Rather, this kind of judgement, properly exercised, must be
Grasp of Essences vs. Intuitions: An Unequal Contest 257
thought of as having an inherent and autonomous—albeit by no means infallible—

power to reveal mind-independent truths of essence. Furthermore, in my view, this
kind of judgement is no less indispensable in theoretical science than it is in metaphys-
ics, so that the analogy between intuition and thought experiments, on the one hand,
and observation and ‘real’ experiments on the other is in fact as damaging to a realist
conception of science as it is to a realist conception of metaphysics. Neither of these
intellectual enterprises can proceed successfully without an exercise of the distinctive
rational capacity that enables us to grasp mind-independent truths of essence.
According to the position that I shall be defending, metaphysical inquiry has two
distinct but related dimensions: one a priori and the other a posteriori. The a priori
dimension concerns the realm of the possible, while the a posteriori dimension con-
cerns the realm of the actual. A key point to be recognized here is that no actual matter
of fact can be established solely by a posteriori means and without at least an implicit
reliance on modal truths which can be established only by a priori means. In this sense,
there can be no such thing as purely empirical knowledge—and this is why metaphysi-
cal inquiry is indispensable to natural science. An important consequence of this key
point is that a certain popular model of the relationship between empirical knowl-
edge and knowledge of modal truths, including truths of essence, is fundamentally
flawed. According to this model, which we owe in large measure to the highly influen-
tial work of Saul Kripke (1980), truths of essence are for the most part knowable only a
posteriori, putative examples being such truths as that water is H2O and that gold has
atomic number 79. In fact, however, the view that such truths are metaphysical neces-
sities whose necessity is discoverable only a posteriori rests, I believe, upon a confusion
between two quite different notions of essence: a ‘scientific’ notion, which is traceable
back to John Locke, and a properly modal notion, which is traceable back to Aristotle.
The question then remains as to precisely how we may acquire knowledge of
what is possible, including what is ‘merely’ possible (that is, possible but not actual).
Conceivability and ‘intuition’, as I have just indicated, are in my view at best unreliable
guides to possibility. The appeal to knowledge of counterfactual conditionals, advo-
cated recently by Timothy Williamson (2007), is no better, in my opinion. The solu-
tion, I believe, lies in seeing that all modal truths are grounded in truths of essence
rather than vice versa—as Kit Fine (1994) has argued—and that knowledge of essence
issues from the grasp of propositions of a special kind: those that express so-called ‘real
definitions’. On this model, knowledge of essence may fairly be described as a priori,
but at the same time we need to reconceive the very notions of the ‘a priori’ and the ‘a
posteriori’. A priori knowledge should not be thought of as knowledge that is entirely
independent of experience and as such available, in principle, even to a thinker who
has never engaged in empirical inquiry. Rather, knowledge in any domain should be
seen as advancing in a cyclical manner, by alternating stages of a priori and a poste-
riori inquiry, each dependent on its predecessor. In this sense, there is no such thing as
‘purely’ a priori knowledge, nor any such thing as ‘purely’ a posteriori knowledge. We
must as a consequence repudiate the Kantian conception of metaphysical knowledge
258 E. J. Lowe
as a repository of apodictically certain and permanently unassailable truths: in meta-

physics, as in natural science, we should be committed fallibilists. Indeed, in my view,
metaphysical fallibilism is a necessary condition of metaphysical realism, and it was
Kant’s mistaken prejudice against the former that precipitated his disastrous rejection
of the latter.
13.2.
It is a central principle of all standard modal logics that what is actual is therefore
possible—ab esse ad posse valet consequentia. It might be supposed that an epistemic
implication of this is that the securest way to establish that something is possible is
simply to establish that it is actually the case, and hence that at least some possibilities
can be established purely by appeal to empirical evidence—namely, those that can be
established as actualities purely by appeal to such evidence. However, that would be a
mistake. For the implication of the logical principle in question is that the possibility
of something’s being the case is a logically necessary condition—a pre-requisite—of its
actually being the case. Hence, empirical evidence can qualify as evidence that some-
thing, p, is actually the case only on condition that p is at least possibly the case. This
should be obvious, on reflection. Why, for example, do physicists not seek to discover
whether perpetual motion machines are possible by looking for empirical evidence
for their actual existence? Because they believe, justifiably, that they know already that
such machines are impossible and hence that nothing could count as genuine evidence
for the actual existence of such a machine. They dismiss any alleged evidence for their
actual existence as a case of either fraud or stupidity on the part of the person making
the allegation. And they are right to do so.
There is an important deeper implication of this. Clearly, in looking for empiri-
cal evidence we can search only the actual world. Empirical data consist entirely of
actual states of affairs, since only these can have the causal impact upon our senses
that is involved in sense-perception of any kind. Hence, a state of affairs that is
merely possible—possible but not actual—cannot be identified as such by purely
empirical means. The actual world is, of course, ipso facto a possible world—the
aforementioned logical principle assures us of this. But how do we know which
world, of all possible worlds, is the actual world? Trivially, we know that it is this
world, the world that actually obtains. But merely knowing this doesn’t help us to
determine whether or not some specific putative state of affairs, p, is actually the
case. We cannot simply appeal to the evidence of our senses to warrant the belief
that p is actually the case for, as we have seen, such evidence can qualify as evidence
for the truth of p only on condition that p is genuinely possible. Thus, when we are
inquiring into whether p is possible, the distinction between p’s being actual and its
being merely possible is epistemically irrelevant, since we are no better placed to
establish the former by purely empirical means than we are to establish the latter.
What emerges from these observations is that, in the epistemic domain, knowledge
of what is possible precedes and underpins knowledge of what is actual and that—
since empirical evidence can embrace only what is actual—knowledge of what is
possible must draw on evidence or reasoning of another kind and hence be a priori
in nature (see further Lowe 1998, pp. 1–27). Accordingly, to the extent that the nat-
ural sciences are empirical sciences, drawing on observational and experimental
evidence, together with inductive or abductive argument from such evidence, their
success as sources of knowledge about the actual world presupposes the possibility
of another successful kind of inquiry that is a priori in character: a kind of inquiry
whose object it is to establish, independently of what might actually be the case,
what is possibly the case. This kind of inquiry explores the space of possible worlds,
attempting to delimit that space by determining both what is possible and what is
co-possible with what else. This form of inquiry is pure metaphysics, in the proper
and original sense of the term.
Pure metaphysics, then, is that form of inquiry whose object it is to explore and
delimit the space of possibility—to determine which worlds are genuinely possible—
and this is a form of inquiry which precedes and is presupposed by any form of empiri-
cal inquiry, such as the natural sciences. But pure metaphysics cannot establish which
of all the possible worlds is actual—which such world is this world. For that, appeal
must be made to empirical evidence, the best of which is provided by the mature natu-
ral sciences. Determining what is actually the case, then, requires a combination of
metaphysics and natural science, which consequently complement each other rather
than being in competition. However, it must also be remarked that another key feature
of metaphysics is its degree of generality: it is concerned with what is co-possible with
what else, not just with what is possible simpliciter. Its scope is entire possible worlds,
not just small parts of them. No natural science has that kind of scope with respect
to the actual world. Hence, even with regard to the actual world, metaphysics has a
special role to play, which cannot be taken over by any of the natural sciences. Its role
is to mediate between those sciences, settling boundary disputes between them and
attempting, as far as is possible, to reconcile their findings. The underlying principle at
work here is the principle of the unity of truth: that all actual truths, being truths about
a single possible world, must be consistent with each other (see further Lowe 2006,
pp. 177–91). So, if pure metaphysics is aptly described as an inquiry into the space of
possibilities, then applied metaphysics—or ‘cosmology’, in the original sense of that
term—may aptly be described as the attempt, conducted with the aid of the various
natural sciences, to establish the fundamental structure of reality as a whole. In this
respect, metaphysics has an a posteriori as well as an a priori dimension.
13.3.
We can see with the help of the foregoing observations that a certain currently popu-
lar model of the relationship between empirical knowledge and knowledge of modal
260 E. J. Lowe
truths is fundamentally in error. According to this model, many modal truths—and

above all those concerning the concrete, physical world, as opposed to the abstract
domains of pure mathematics and logic—are knowable only a posteriori, in the fol-
lowing kind of way. First, it is established by a priori means that a certain kind of con-
ditional proposition is true—one whose antecedent is a non-modal proposition and
whose consequent is a modal one. Then is it established, by a posteriori means, that the
antecedent of this conditional proposition is in fact true—is true in actuality. Finally,
by the rule of modus ponens, the truth of this conditional proposition’s modal conse-
quent is inferred. The archetypal example of this method of proceeding is Saul Kripke’s
(1980) argument for the a posteriori necessary truth of such identity-propositions as
the proposition that Hesperus is Phosphorus and the proposition that water is H2O—
although whether the latter is properly described as an identity-proposition is a moot
point. In the Hesperus/Phosphorus case, the conditional proposition in question is the
proposition that if Hesperus is identical with Phosphorus, then it is necessarily the case
that Hesperus is identical with Phosphorus. This conditional proposition is, suppos-
edly, demonstrable a priori. Allegedly, however, that Hesperus is in fact identical with
Phosphorus may be and has been established by purely a posteriori means. Indeed, it
is alleged that it is only by such means that this actual truth may be established. Given
that it has been thus established, however, we are now, supposedly, in a position to infer
that it is necessarily the case—that it is the case in every possible world—that Hesperus
is identical with Phosphorus. But since, supposedly, this is the only way in which this
modal truth can be established, it follows that our knowledge of it can only be a pos-
teriori, on the grounds that this knowledge rests upon our unavoidably a posteriori
knowledge that Hesperus is in fact identical with Phosphorus.
One thing that is wrong with this model of our acquisition of modal knowledge about
the concrete, physical world is that it presumes, erroneously, that the antecedent of the
relevant conditional proposition can be established by purely a posteriori means and
without reliance on any preceding modal knowledge. This is false, as the very example
of Hesperus and Phosphorus demonstrates. Astronomers did indeed discover, with the
aid of careful observation, that Hesperus and Phosphorus—the Evening Star and the
Morning Star—are one and the same planet, namely, Venus. But they did this by discov-
ering that Hesperus and Phosphorus coincide in their orbits around the Sun and hence
that they occupy the same positions in space at the same time. However, the conclu-
sion that Hesperus and Phosphorus are therefore identical rests upon the presumption,
which is modal in character, that two distinct planets cannot exist in the same place at
the same time. No doubt this modal proposition is true, but its truth is not something
that can be established by purely a posteriori means. It would be absurd to suppose, for
instance, that it could be established inductively from many observations of planets,
none of which revealed any two planets coinciding with one another. Rather, this truth
is one that is known by appeal to metaphysical considerations and reasoning, resting
as it does on the general proposition that material objects of the same kind exclude one
another from the same place at any given time (see further Lowe 2008a).
Another problem with the foregoing model of our acquisition of modal knowl-
edge concerns the status of the conditional proposition that is called upon in any
given case. In the Hesperus/Phosphorus case, it is the proposition that if certain
entities are in fact identical, then they are necessarily identical. This conditional
proposition is itself supposed to be a necessary truth, but one that can be known a
priori. I have, of course, no quarrel with the idea that necessary truths can be known
a priori, but there is a special problem concerning conditional propositions of this
kind. This can be brought out more clearly with the water/H2O example. As I men-
tioned earlier, it is questionable whether, in this case, we are really dealing with an
identity-proposition, as opposed to something more like a composition-proposition.
The modal truth that we are alleged to have discovered a posteriori is perhaps best
characterized as the proposition that water is necessarily composed of H2O mole-
cules. Now, how are we supposed to know, a priori, that if water is composed of H2O
molecules, then it is necessarily composed of H2O molecules? If this were knowable
purely a priori, then surely the ancient Greeks could have known it. But they surely
couldn’t have known it, because they had—and could have had—absolutely no con-
ception of what an H2O molecule is or might be. Concept-formation is a deeply
historical process and it makes no real sense to suppose that the modern chemi-
cal concept of an H2O molecule could have arisen in the context of ancient Greek
thought. The only plausible suggestion, then, is that this conditional proposition—
that if water is composed of H2O molecules, then it is necessarily composed of H2O
molecules—is not itself knowable purely a priori, but is at best only inferable from
two other propositions, one of which is knowable a priori while the other is not. The
most likely propositions in question would plausibly be the following.
(1) For all X, if water is composed of X, then necessarily water is composed of X.
(2) Water is composed of H2O molecules.
The suggestion is that the general conditional (1) is knowable purely a priori while the
singular composition-proposition (2) is knowable purely a posteriori. Of course, we
have already seen, by implication, reason to doubt the latter claim—that (2) is know-
able purely a posteriori—since we have seen reason to doubt that any non-modal truth
about the actual world can be known by purely empirical means. (2) is no more immune
to this point than is the proposition that Hesperus is identical with Phosphorus. But of
more interest right now is the general conditional proposition (1) and, just as impor-
tantly, the inferential step from (1) and (2) to the specific conditional proposition
(3) If water is composed of H2O molecules, then water is necessarily composed of
H2O molecules.
As for (1), its status as a truth, let alone an a priori truth, is certainly questionable. What
entitles us to suppose that water has the same composition in every possible world in
which it has a composition? Notice that (1) does not presuppose that water does have a
composition in every possible world: it leaves open the possibility that, in at least some
262 E. J. Lowe
possible worlds, water is a non-composite entity. As for the ancient Greeks, many of
them supposed that water was an element and as such not composed of anything other
than itself. They, then, could only have regarded (1) as a trivial truth, since for them
the only acceptable value of the variable ‘X’ would be water. The fact that (1) has more
appeal to modern philosophers is, I suggest, a product of their understandable admira-
tion for the advances of modern chemistry. Modern chemistry has revealed that many
kinds of stuff, including water, have interesting molecular compositions, which serve
to explain many of the macroscopic features of such stuffs. Water’s chemical composi-
tion, H2O, helps to explain, for instance, why it is transparent, why it is weakly electri-
cally conductive, why it becomes less dense on freezing, and why it is liquid at room
temperature. However, all of these explanations appeal only to actual physical laws,
discovered a posteriori, and they are causal in character. As such, they have nothing to
do with whatever physical appearance and behaviour water might or might not have
in other, non-actual possible worlds, in which physical laws might be very different.
The mere fact that the chemical composition of water in this world serves very well
to explain its appearance and behaviour in this world provides us with no metaphysi-
cal reason whatever to suppose that it must have the same composition in every other
possible world. Indeed, if we do suppose this, then, since we must allow that in other
such worlds very different physical laws might obtain, we must also allow that water
might differ radically in its appearance and behaviour in different possible worlds: for
instance, that in some worlds it might have the appearance and behaviour that tar or
bronze does in this world. And yet the supposition that water could be so varied in its
possible appearance and behaviour seems absurd and, certainly, not something that
we should regard as uncontroversially true.
The modern philosophers who commit themselves to such seeming absurdities
appear to conflate two quite different conceptions of the essence of a material sub-
stance (see further Lowe 2011). One is the Lockean notion, according to which the
essence of a material substance consists in its ‘internal constitution’, describable
in terms of its atomic or molecular composition and structure (see, for example,
Locke 1975 [1700]: II, XXIII, §3). According to Locke, it was this constitution that
served to explain, causally, the observable, macroscopic features of a substance.
However, there is another and older notion of essence, to be found in the works of
Aristotle and his Scholastic successors, which was explicitly repudiated by Locke.
This was the properly modal notion of essence, according to which the essence of
a substance consisted in its substantial form, regarded as an absolutely necessary
and unalterable feature of it which distinguished it from any other substance and
thereby served as the criterion for our classifying this substance as being the very
substance that it is. Locke, as I say, expressly rejected this notion of essence and its
associated theory of classification, in favour of a nominalistic and conventionalist
classification based on human language and human interests. What many modern
philosophers have done, though, is to take Locke’s notion of essence and attempt to
confer upon it the modal and classificatory roles that are associated with Aristotle’s
notion of essence—a confusion that both Locke and Aristotle would look upon
with horror.
One other point is worth mentioning here (but for fuller discussion see Lowe 2007).
This concerns the inference from (1) for all X, if water is composed of X, then necessar-
ily water is composed of X and (2) water is composed of H2O molecules to (3) if water
is composed of H2O molecules, then water is necessarily composed of H2O molecules.
In (1), we see quantification into a position governed by a modal expression, namely,
in the consequent of the conditional. The inference can be regarded as valid only if it is
legitimate to substitute ‘H2O molecules’ for the variable ‘X’ in both of its bound occur-
rences in (1). But the legitimacy of this is certainly questionable. One may be prepared
to allow the substitution of a proper name for a bound variable in such modal contexts,
but the expression ‘H2O molecules’ is certainly not a proper name: it is an abbrevia-
tion for some such description as ‘molecules formed by the bonding of two hydrogen
atoms to a single oxygen atom’. And, notoriously, Kripke himself was adamant that
the logical behaviour of descriptions in modal contexts is very different from that of
proper names.
13.4.
I said earlier that the task of pure metaphysics is to determine what is possible, inde-
pendently of what might actually be the case, and that this requires a method of
knowledge-acquisition which is a priori. Here it is important to emphasize that the
kind of possibility now in question is so-called metaphysical possibility, which I take to
be objective and mind-independent in character, as well as being discoverable a priori.
Metaphysical possibility, then, is not to be confused with physical or natural possibility,
as determined by physical laws of nature (pace Shoemaker 1998). Nor, however, is it to
be confused with mere logical possibility, or freedom from contradiction (see further
Lowe 1998, pp. 1–27). Finally, it is not even to be confused with conceptual possibility,
since this is not suitably mind-independent. Of course, there are those who will say
that, having thus stipulated what metaphysical possibility is not, we have left nothing
for it to be. And there are others who will say that, even if we have left something for it
to be, we have left something that is not humanly knowable. I shall leave it to those who
make such claims to substantiate them, if they can, without implicitly relying on the
very notion of possibility that they purport to find problematic. Meanwhile, we who
accept that there is such a thing as metaphysical possibility and that it is not beyond
our epistemic grasp need to give some account of how it is that we can come to know
modal truths involving possibility of this kind.
I shall once again dismiss, without detailed examination here, appeals to conceiv-
ability or ‘intuition’ as guides to what is metaphysically possible (see further Gendler
and Hawthorne 2002), because these guides seem to be at best unreliable ones. After
all, perpetual motion machines are least conceivable, or else there would be no need to
264 E. J. Lowe
dismiss their possibility, and yet they are, plausibly, not only physically but also meta-
physically impossible. The same might be said of time-travel into the past. Equally, it
seems to me irrelevant to ask what our ‘intuitions’ are concerning the identity of the
person emerging from a hypothetical teletransportation device, when what is at issue
is whether it is metaphysically possible for a person to survive ‘transportation’ by such
means. Nor does it seem to me appropriate for us to try to settle such a question by
reflecting on our concept of a person: conceptual analysis may reveal something about
the structure of our thought, but there is no clear reason to suppose that it is a reliable
guide to the structure of modal reality, that is, the space of metaphysical possibilities.
Timothy Williamson (2007) has recently proposed a rather different model of modal
knowledge-acquisition, appealing to our naturally evolved capacity to make relatively
reliable counterfactual conditional judgements. This model is predicated upon the
thesis that modal propositions are explicable in terms of counterfactual conditionals—
for instance, that ‘It is necessarily the case that p’ is logically equivalent to ‘If p were
not the case, then a contradiction would obtain’ and, correlatively, that ‘It is possibly
the case that p’ is logically equivalent to ‘It is not the case that, if p were the case, then
a contradiction would obtain’. As it happens, I would want to dispute the claim that
these purported equivalences hold. In fact, I contend that counterfactual conditional
propositions are analysable in terms of modal propositions, not vice versa, although
I shall not go into the details here (but see further Lowe 1995). However, quite apart
from this, it is simply unclear how a naturally evolved capacity to make reliable judge-
ments concerning the truth or falsehood of ordinary counterfactual conditionals in
the everyday circumstances of life could be expected to remain reliable when dealing
with counterfactual conditionals of the very special kind invoked by Williamson, in
which the consequent refers to the obtaining of a contradiction. Moreover, the relevant
sense of ‘contradiction’, where metaphysical modality is concerned, could not simply
be that of a formal, logical contradiction: it would have to be that of a metaphysically
impossible state of affairs. But then we are back to where we started, with the question
of how we acquire knowledge that something is metaphysically necessary, possible, or
indeed impossible. Consequently, I see no real promise in Williamson’s approach to
modal epistemology.
Where I do see promise is in a very different approach to modality that has been
pioneered in recent years by Kit Fine (1994), although it has a long and distinguished
philosophical pedigree, being traceable back to Aristotle via the Scholastics. On this
approach, instead of trying to explicate the notion of essence in terms of that of modal-
ity, as on the Kripkean account of essence, the very reverse needs to be done. According
to Aristotelian essentialism, as I shall call it, the notion of metaphysical modality is to
be explicated in terms of the notion of essence. At the heart of Aristotelian essential-
ism is the notion of a real definition, understood as a special kind of proposition which
expresses, in a perspicuous fashion, the essence of some entity or kind of entity. For
example, the real definition of a circle, as a kind of geometrical figure, might be taken
to be this: a circle is the locus of a point moving continuously in a plane at a constant
distance from another fixed point. It is important to distinguish such a real definition—
called ‘real’ because it defines a thing, or res—from a verbal definition, which defines
the meaning of a word.
The correctness conditions of real and verbal definitions are quite different. One can
show a verbal definition to be incorrect by showing that it does not conform to the
actual usage of the word in question. But one can show a real definition to be incorrect
only by showing that it fails to express anything’s nature or essence—that is, fails to
express, concerning something, what that thing is. If I had said, earlier, that an ellipse,
rather than a circle, is the locus of a point moving continuously in a plane at a constant
distance from another fixed point, I would certainly have been in error: but not because
I had failed to supply a perfectly satisfactory real definition of something. There is such
a thing as the locus of a point moving continuously in a plane at a constant distance from
another fixed point, and my only mistake would have been to call this an ‘ellipse’ rather
than a ‘circle’. That mistake, however, is just a verbal one, concerning the correct appli-
cation of a word. It matters not at all whether we call circles ‘circles’ or ‘ellipses’. What
matters is whether we correctly understand what a certain kind of geometrical figure,
standardly called a ‘circle’, really is. Of course, a real definition has to be expressed in
words, selected from some language which the person framing it understands and uses
correctly. But none of this should distract us from the fact that what a real definition
aims to do, and can succeed in doing, is to express the real nature of some thing or
kind of thing. This real nature will be had by that thing quite independently of how we
describe it or think of it and there is no good reason to suppose that we are incapable
of correctly identifying such real natures, in at least some cases. For if we could never
correctly identify the real nature of anything—never know what a certain thing or kind
of thing is—then we could never know anything at all, which no one but the most
extreme sceptic would propose (without, of course, claiming to know it).
The connection between modality and essence is this. Those features of a thing
which belong to its real definition are necessary features of the thing, without which
it could not exist. For instance, a circle could not exist without having a centre (this
being the ‘fixed point’ mentioned in its real definition). Hence, it is a metaphysically
necessary truth that every circle has a centre. However, not every necessary feature of a
circle belongs to its real definition. For instance, it is a necessary feature of a circle that
any triangle all of whose vertices lie on the circle’s circumference and one side of which
is a diameter of the circle is a right-angled triangle (at least if we confine ourselves to
Euclidean geometry). But this is a demonstrable consequence of the real definitions of
a circle and a right-angled triangle. Hence, in this case, the metaphysically necessary
truth in question, although it does not reside in or issue from any single real defini-
tion, is nonetheless rooted in such definitions. The complete class of metaphysically
necessary truths is rooted in the complete class of real definitions, that is, in the natures
or essences of all things. Some further refinements are needed to render this account
entirely satisfactory, but this quick sketch suffices for present purposes (see further
Lowe 2008a, 2008b, 2012). Of course, I make no claim that we human beings are in
266 E. J. Lowe
a position to grasp and hence know the complete class of real definitions, only the
much more modest claim that we do grasp and know at least some of them. Some of
them we may grasp only imperfectly or partially, moreover, as when a child grasps only
that a triangle is a three-sided figure, without grasping more fully that it is a rectilinear
three-sided planar figure.
My proposed solution to the problem of modal knowledge—the question of how
we can come to know some modal truths—is that we can do so because this kind
of knowledge is ultimately grounded in a special kind of understanding: the kind of
understanding that is involved in grasping a real definition. A real definition, recall, is
just a proposition of a special kind and, hence, the claim that we can understand at least
some real definitions is no more problematic than the more general claim, which is
surely incontestable, that we can, at least sometimes, understand propositions. Anyone
who believes that we can think at all must accept that at least sometimes we can under-
stand propositions. But in this respect, real definitions are in no way special: they are,
after all, just propositions. What is special about them from an epistemic point of view
is their truth-conditions. For they have no truth-makers in the ordinary sense: noth-
ing has to exist in order for a real definition to be true (for more on truth-makers in
general, see Lowe and Rami 2009). This is illustrated by our very example of the real
definition of a circle. It is a true definition and yet, very probably, there actually exists
nothing that perfectly conforms to it. In grasping the real definition, we know what a
circle is or would be, but we do not necessarily know that there actually is any such fig-
ure existing in this, the actual world. Real definitions, accordingly, are precisely suited
to the epistemological task of pure metaphysics: by grasping them, we come to know
that certain states of affairs are possible, independently of knowing whether or not they
are actual. Another way of putting this is in terms of the familiar notion of ‘direction of
fit’. The actual world contains Xs just in case something exists in the actual world which
conforms to the real definition of an X. The real definition provides a recipe or formula
for a certain thing or kind of thing, which the world may or may not oblige us by real-
izing in actuality. It would be quite wrong, then, to think that a real definition, to be
satisfactory, must ‘fit’ some actually existing thing or kind of thing and consequently
that we must examine existing things to discover whether or not our real definitions
are ‘correct’. This is the mistake of the a posteriori essentialists, who claim that we must
discover by empirical means what the essence of, for example, water is. On the con-
trary, if we did not already know what water is, by grasping its essence—even if only
imperfectly—through at least an implicit knowledge of its real definition, we would be
quite unable to identify anything in the actual world as being water and consequently
would be quite unable to investigate any such thing empirically with a view, suppos-
edly, to ‘discovering’ its essence. The a posteriori essentialists get the cart entirely
before the horse, at least to the extent that they understand essences to determine the
modal features of things, as opposed merely to explaining, in a purely contingent and
causal fashion, why they have the actual observable appearance and behaviour that
they do. Once more, they confuse the Lockean notion of essence with the properly
modal, Aristotelian notion of essence, attributing to the former modal characteristics

which can really only belong to the latter.
13.5.
To draw matters to a close, I now need to refine and qualify one particular aspect of
the foregoing account. I shall do this very briefly, recognizing that a very much fuller
account of every aspect of my position is ultimately desirable—something that I hope
to provide at a later date. The particular point in question is this. When I say that pure
metaphysics engages in a priori inquiry, I do not mean to suggest that such inquiry is
open to any thinker to engage in quite independently of the possession by that thinker
of any empirical knowledge whatsoever. It would be a gross caricature of a priori
knowledge to describe it as knowledge attainable, even if only in principle, by a thinker
devoid of all empirical access to the actual world. Rather, a priori inquiry is distinctive
in that it proceeds in advance of some a posteriori knowledge, while still, in general,
relying on other such knowledge. A rather clear example of this is provided by modern
chemical knowledge of the elements, as characterized by their positions in the periodic
table and their associated nuclear compositions. Prior to the actual synthesis of vari-
ous transuranic elements—prior, indeed, to the actual existence, anywhere in the uni-
verse, of certain of these elements—chemists knew what they would be. That is to say,
they grasped the real definitions of certain as yet non-existent transuranic elements.
In speaking of the chemical elements here, I should emphasize that I am speaking of
certain kinds of atoms, not the gross material stuffs composed of those atoms. A gold
atom, for instance, has very different physical properties from the kind of gross mate-
rial stuff that we call gold. The latter is malleable, ductile, shiny, and yellow in colour,
but no gold atom has any of these properties. A gold atom undoubtedly necessarily
possesses 79 protons in its nucleus, since part of what it is to be a gold atom—part of its
real definition—is for it to have that many protons in its nucleus. None of this implies
that the gross material stuff that we call gold, and now know actually to be composed of
such atoms, is essentially and hence necessarily composed of such atoms.
This understood, what I am saying, to repeat, is that modern chemists knew, prior to
their synthesis in particle accelerators, that certain transuranic elements were possible,
because they knew what they would be, in terms of their nuclear composition. This was,
then, a priori knowledge on their part. But, clearly, it was not knowledge that was avail-
able entirely independently of previously acquired empirical knowledge. The chemists
in question had to know, for example, that atomic nuclei are composed of protons and
neutrons, and that the latter are subject to certain strong physical forces of attraction
and repulsion. And at least some of this knowledge was clearly acquired only through
empirical investigation. Typically what happens as a theoretical science advances is
that its practitioners engage in a cycle of a priori and a posteriori epistemic procedures,
first proposing certain hypotheses and then investigating the world empirically to
268 E. J. Lowe
discover whether it seems likely that entities of the kinds postulated in those hypoth-
eses do actually exist. If they are satisfied that the entities in question do exist, then
they may propose further hypotheses postulating yet other entities whose natures or
essences are defined partly in terms of those of previously discovered entities, as in the
case of the transuranic elements and their composition by protons and neutrons.
This, then, is why I said, in my opening remarks, that there is really no such thing
as ‘purely’ a priori knowledge, nor any such thing as ‘purely’ a posteriori knowledge,
and that, as a consequence, we must repudiate the Kantian conception of metaphysi-
cal knowledge as a repository of apodictically certain and permanently unassail-
able truths: in metaphysics, as in natural science, we should be committed fallibilists.
Indeed, in my view, metaphysical fallibilism is a necessary condition of metaphysical
realism, and—to repeat—it was Kant’s mistaken prejudice against the former that pre-
cipitated his disastrous rejection of the latter (see further Lowe 1998, pp. 1–27).
References
Fine, K. (1994). “Essence and Modality,” in J. E. Tomberlin (ed.), Philosophical Perspectives,
8: Logic and Language. Atascadero, CA: Ridgeview.
Gendler, T. S., and Hawthorne, J. (eds) (2002). Conceivability and Possibility.
Kripke, S. A. (1980). Naming and Necessity. Oxford: Blackwell.
Locke, J. (1975 [1700]). An Essay Concerning Human Understanding, P. H. Nidditch. (ed.)
Lowe, E. J. (1995). “The Truth About Counterfactuals,” The Philosophical Quarterly 45, pp. 41–59.
——. (1998). The Possibility of Metaphysics: Substance, Identity, and Time. Oxford: Clarendon Press.
——. (2006). The Four-Category Ontology: A Metaphysical Foundation for Natural Science.
——. (2007). “A Problem for A Posteriori Essentialism Concerning Natural Kinds,” Analysis 67,
pp. 286–92.
——. (2008a). “Two Notions of Being: Entity and Essence,” in R. Le Poidevin (ed.),
Being: Developments in Contemporary Metaphysics. Cambridge: Cambridge University Press,
pp. 23–48.
——. (2008b). “How Are Identity Conditions Grounded?,” in C. Kanzian (ed.), Persistence.
Frankfurt: Ontos Verlag, pp. 73–89.
——. (2011). “Locke on Real Essence and Water as a Natural Kind: A Qualified Defence,”
Proceedings of the Aristotelian Society, Supplementary Volume 85, pp. 1–19.
——. (2012). “What is the Source of Our Knowledge of Modal Truths?,” Mind 121, pp. 919–50.
——., and Rami, A. (2009). Truth and Truth-Making. Stocksfield: Acumen.
Shoemaker, S. (1998). “Causal and Metaphysical Necessity,” Pacific Philosophical Quarterly 79,
pp. 59–77.
14
X-Phi without Intuitions?
Herman Cappelen†
14.1. Introduction
One central purpose of Experimental Philosophy (hereafter, x-phi) is to criticize the
alleged reliance on intuitions in contemporary philosophy. In my book Philosophy
without Intuitions (2012; hereafter, PWI), I argue that philosophers don’t rely on intui-
tions.1 If those arguments are good, experimental philosophy has been engaged in an
attack on a straw man.2 The goal of this chapter is to bolster the criticism of x-phi in the
light of responses.
Proponents of x-phi typically respond to the kinds of arguments adduced in PWI in
one of two ways:
The X-Phi-Doesn’t-Need-Intuitions Reply. Experimental philosophy doesn’t
(need to) target intuitions: its criticism of philosophical practice stands even if it isn’t
the case that philosophers rely on intuitions.
As a heuristic, think of the reply along these lines: we can go back to influential papers
in experimental philosophy and replace ‘intuition’ with a different term. So, in passages
like, “Experimental philosophers have begun conducting empirical research to find
out what intuitions are generated in response to certain cases. But rather than support-
ing and explaining the practice of appealing to intuitions as evidence, the results of this
† Thanks to Stewart Cohen, Josh Dever, and two anonymous readers at Oxford University Press for help-
ful feedback.
1 I also take this to be one of the central messages of Timothy Williamson’s The Philosophy of Philosophy. It
is also a point made forcefully by Max Deutsch (2009, 2010).
2 The x-phi movement is sometimes described as divided into a negative and a positive or constructive
camp. For the constructive experimental philosophers, the study of intuitions can help us do philosophy.
The focus in this chapter is on the negative camp, but it is worth noting that the positive version of the pro-
gramme is equally committed to the assumption that philosophers rely on intuitions as evidence. It endorses
that assumption and then tries to make a positive contribution by discovering interesting facts about the
patterns of intuitive reactions. As a result, the objections in this chapter (and in PWI) are, if effective at all,
equally effective against both camps.
270 Herman Cappelen
research challenge the legitimacy of appealing to intuitions” (Swain et al. 2008, p. 140,
italics added), we can delete ‘intuition’, replace it with some other term, ‘XYZ’, and:
a) The central claims are still true.
b) X-phi surveys are relevant to XYZ.
The second strategy for responding to the kind of criticism found in PWI is the
Direct Reply:
The Direct Reply. The arguments in PWI fail—philosophers do rely on intuitions.
My own view, argued for extensively in Part I of PWI, is that the word ‘intuition’ is
such a semantic and pragmatic mess that those interested in the philosophy of phi-
losophy are better off if positions and arguments are not articulated using that and
cognate terms. As a result, I will treat all of the proposals below as instances of the
X-Phi-Doesn’t-Need-Intuitions Reply (even though some of the proponents of x-phi
will insist on using the word ‘intuition’ for some of the phenomena appealed to).
14.2. The X-Phi-Doesn’t-Need-Intuitions Reply

To illustrate how integral ‘intuition’-talk is to x-phi it will be helpful to briefly remind
the reader of how proponents of the movement typically describe their motivations
and results. Here are some representative articulations of the x-phi project:
According to standard practice, a philosophical claim is prima facie good to the extent that it
accords with our intuitions, prima facie bad to the extent that it does not. Given that intuitions
about thought-experiments are standardly taken as reasons to accept or reject philosophical
theories, then we should be interested in finding out what the relevant intuitions are. (Swain
et al. 2008, p. 140, italics added)
According to the party line, the exploration of these appeals to intuitions is the goal
of x-phi:
A number of experimental philosophers in recent years (e.g., Machery, Mallon, Nichols, &
Stich, 2004; Swain, Alexander, & Weinberg, 2008; Weinberg, Nichols, & Stich, 2001; see also
Sinnott-Armstrong, 2008) have begun to challenge analytic philosophy’s longstanding practice of
deploying armchair intuitive judgments about cases. (Weinberg et al. 2010, p. 331, italics added)
The results, we are told, don’t look good for the longstanding practice:
Experimental philosophers have begun conducting empirical research to find out what intuitions
are generated in response to certain cases. But rather than supporting and explaining the prac-
tice of appealing to intuitions as evidence, the results of this research challenge the legitimacy of
appealing to intuitions. Weinberg, Nichols, and Stich revealed that epistemological intuitions
vary according to factors such as cultural and educational background; Machery et al. document
a similar cultural variation in semantic intuitions; and Nichols and Knobe have discovered that
the affective content of a thought-experiment can influence whether subjects have compatabilist
or incompatabilist intuitions. (Swain et al. 2008, p. 140, italics added)
X-Phi without Intuitions? 271
The topic of this chapter can be put very simply: suppose the arguments in PWI are
correct and appeals to intuitions play no role in philosophical practice. This looks like
a disaster for the proponents of the quoted passages. Their project is founded on a
false idea of what philosophy is and so their surveys have no relevance to what we
philosophers do.
It is when faced with this challenge that some experimental philosophers backtrack.
They claim that ‘intuition’-talk was nothing but an innocent terminological mistake—
they were simply following standard philosophical usage. Nothing substantive hangs
on this choice of words and there are easy ways to correct for the poor choice of words,
we’re told. Joshua Alexander and Jonathan Weinberg (2007) say:
Although the results are often glossed in terms of intuitions to follow standard philosophical
usage, inspection of the experimental materials reveals little talk of intuitions and mostly the
direct evaluation of claims. (p. 72, italics added)
We find that kind of backtracking even in popular presentations of x-phi. Here, for
example, is what Edouard Machery says:
. . . philosophers assume that judgments about philosophical cases or thought experiments (what
is often called, misleadingly, “intuitions”) are likely to be true or reliable. A huge amount of
work in experimental philosophy casts doubt on this view, by showing that these judgments are
influenced by irrelevant variables such as culture, age, order of presentation of cases, and so on.
(3:AM Magazine, interview3 )
The goal of this chapter is to convince the reader that experimental philosophers
underestimate the difficulties involved in the revision they here gesture at. They talk as
if what is required is nothing but a slight terminological revision, but it is not. In what
follows I consider eight options for how to reformulate the x-phi project. They all fail.
One preliminary: in important ways, this chapter is not self-contained. At several
points in what follows, I will refer the reader to arguments and evidence presented
in PWI. It would be pointless and repetitive to go over that ground again here so
at some crucial points arguments build on material in considerably more detail in
PWI. The goal here is to highlight those elements of PWI that are relevant to the
X-Phi-Doesn’t-Need-Intuitions-Reply.
Option 1: Replace ‘intuitions’ with philosophical ‘judgements’ or ‘beliefs’

This option, in its full generality, is clearly absurd, but the absurdity helps illuminate
the more general problem. So let’s consider the question of whether x-phi surveys
have relevance if their target is not the reliance on intuitions, but instead philosophers’
‘judgements’ or ‘beliefs’. To see the absurdity of this proposal, consider these philo-
sophical judgements I make:
3 “Without Concepts: Edouard Machery interviewed by Richard Marshall,” Available at <http://

www.3ammagazine.com/3am/without-concepts/> [last accessed October 23, 2013].
272 Herman Cappelen
• Quotation is governed by the axiom: ‘A’ quotes ‘‘A’’.

• Agreement and disagreement patterns do not provide evidence for relativism
about predicates of taste.
• A monadic truth predicate is more fundamental than a relational truth predicate.
• Disagreement diagnostics for shared content are better than same-saying diag-
nostics for shared content.
To conduct a survey about what, e.g., undergraduates in New Jersey and Hong Kong
believe about quotation or monadic truth would be pointless in the extreme. So, if
we find out, for example, that 8.2 per cent more students in HK believe in monadic
truth than in NJ, that isn’t helpful to my philosophical practice, the goal of which is
to find out whether truth is monadic or not. It is pointless because to engage with my
beliefs about monadic truth you have to engage with my reasoning and the evidence
that I have. The surveys don’t do that and so don’t speak to my philosophical concerns.
In what follows I will assume that all parties to this debate agree with this and that
Option 1 must be rejected. What is needed is obvious: a more specific characterization
of the kind of judgement in question. The next six proposals are all attempts to narrow in
on a subset of judgements that x-phi surveys target.
Option 2: Replace ‘intuitions’ with ‘philosophically relevant judgements

about cases’
According to Option 2, the relevance of x-phi studies is limited to philosophically rele-
vant beliefs (or judgements) about so-called ‘cases’ or ‘thought experiments’. Alexander
and Weinberg (2007) write:
Going back arguably at least to Frege (and, in some sense, all the way back to Socrates), it has
been a standard practice in analytic philosophy to employ intuitions generated in response to
thought-experiments as evidence in the evaluation of philosophical claims. (p. 56)
Purged of ‘intuition’-talk, this passage suggests that the intended target is not just any
philosophical judgement, but those that concern thought experiments. I’ll set aside
the difficulties of making clear what a ‘thought experiment’ is and how to understand
‘philosophically relevant’ (though I think those are important and overlooked prob-
lems, see PWI, §9.1). Instead, I’ll focus on a more immediate problem. Recall that
Option 1 was rejected because x-phi surveys failed to engage with my reasons or evi-
dence (for, e.g., the claim that truth is monadic). The same problem arises in connec-
tion with the option now under consideration. Consider Case N:
ase N. A little girl, Nora, takes a big bite of a strawberry ice cream and utters: “This
C
is delicious!”
While reflecting on Case N, I make the following judgement: Nora’s judgement is
monadically true. Again, I say this because I have thought very hard for many years
about the difference between the relativized truth predicate and monadic truth predi-
cate and now endorse a view according to which judgements involving the predicate
‘delicious’ are monadically true. The book Relativism and Monadic Truth (2009) is an
extended presentation and defence of this view. Again, I take it to be obvious that sur-
veys about how undergraduates in New Jersey and Hong Kong respond to the question
“Is Nora’s judgement monadically true?” have no philosophical significance. Even if
we found a difference of 8.2 per cent between NJ and HK undergraduates, it wouldn’t
move me and it would have no bearing on the debate about whether truth is monadic
or relativized.

about cases that we have no evidence for and don’t base on reasons’
Weinberg says:
Intuitions are odd critters: intellectual happenings in which it seems to us that something is the
case without arising from our inferring it from any reasons that it is so, or our sensorily perceiv-
ing that it is so, or our having a sense of remembering that it is so. (2007, p. 318)
Shaun Nichols et al. say:

As we use the notion, an epistemic intuition is simply a spontaneous judgment about the epis-
temic properties of some specific case—a judgment for which the person making the judgment
may be able to offer no plausible justification. (2001, p. 432).
Again, there is a way to purge these passages of ‘intuition’-talk and the remaining pro-
posal is what I suspect is at the core of x-phi’s picture (or caricature) of philosophical
practice: at the foundation of much philosophy is a set of case judgements made by
philosophers without evidence or reasons—they are made spontaneously, quickly, and
‘intuitively’. We read e.g. a Gettier case, or a Truetemp case, and—boom!—a judge-
ment about the case just comes to us. If it were true that philosophers relied on such
spontaneous responses as foundational evidence, then it might be very interesting to
learn that they vary widely with philosophically irrelevant factors.
The problem for x-phi is that this model of case judgements is false. In all of the cases
anyone has ever mentioned as examples of philosophical case judgements, the judgements
are made based on reasons and evidence and are assessed on that basis. This is a point at
which I have to refer readers to earlier work. PWI goes through many cases (or thought
experiments) in detail and finds none that fit the Option 3 caricature. I will provide only
one illustration here (not included in PWI). Consider Stewart Cohen’s (1999) famous air-
port case. This is a paradigm of how contemporary epistemologists appeal to judgements
about thought experiments. To see whether it is a legitimate target of Option-3-x-phi,
we have to investigate whether the targeted judgement is one that occurs spontaneously,
‘without reason’, etc. The case is first introduced in the context of an attempt to spell out
fallibilism. Cohen says: “Falliblism allows that we can know on the basis of non-entailing
reasons. But how good do the reasons have to be? Reflection on cases show that this can be
a difficult question to answer” (1999, p. 58). We are then presented with the famous case:
Mary and John are at the L.A. airport contemplating taking a certain flight to New York.
They want to know whether the flight has a layover in Chicago. They overhear someone ask
274 Herman Cappelen
a passenger Smith if he knows whether the flight stops in Chicago. Smith looks at the flight
itinerary he got from the travel agent and responds, “Yes I know—it does stop in Chicago.” It
turns out that Mary and John have a very important business contact they have to make at the
Chicago airport. Mary says, “How reliable is that itinerary? It could contain a misprint. They
could have changed the schedule at the last minute.” Mary and John agree that Smith doesn’t
really know that the plane will stop in Chicago. They decide to check with the airline agent.
(p. 58)
Note that Cohen does not then go on to offer up a spontaneous, non-reflective judge-
ment. On the contrary, he starts to give reasons for and against various assessments of
the case:
What should we say about this case? Smith claims to know that the flight stops in Chicago. Mary
and John deny that Smith knows this. Mary and John seem to be using a stricter standard than
Smith for how good one’s reasons have to be in order to know. Whose standard is correct? Let’s
consider several answers:
1) Mary and John’s stricter standard is too strong, i.e., Smith’s standard is correct and so
Smith can know the flight stops in Chicago (on the basis of consulting the itinerary).
Is this a good answer? If we say that contrary to what both Mary and John presuppose,
the weaker standard is correct, then we would have to say that their use of the word ‘know’
is incorrect. But then it is hard to see how Mary and John should describe their situation.
Certainly they are being prudent in refusing to rely on the itinerary. They have a very
important meeting in Chicago. Yet if Smith knows on the basis of the itinerary that the flight
stops in Chicago, what should they have said? “Okay, Smith knows that the flight stops in
Chicago, but still, we need to check further.” To my ear, it is hard to make sense of that claim.
Moreover if what is printed in the itinerary is a good enough reason for Smith to know,
then it is a good enough reason for John and Mary to know. Thus John and Mary should
have said, “Okay, we know the plane stops in Chicago, but still, we need to check further.”
(pp. 58–9)
Cohen then goes on to consider two more options. He considers the options carefully
and his conclusion is reached tentatively; the difficulty and complexity is emphasized
throughout. As I show in PWI, this is paradigmatic of how philosophers engage with
cases.4
The central point I want to emphasize is this: if a proponent of x-phi endorses
Option 3, philosophical judgements (or activities of some kind) are appropriate targets
for x-phi just in case they have certain features (they have to be, e.g., intellectual hap-
penings in which it seems to us that something is so even though we don’t have reasons for
judging that that it is so, etc.). We are, for each candidate target, owed some evidence
that it (the judgement or activity) has those features. X-phi practitioners never provide
such evidence and seem to not even recognize a burden to provide it. As the reader can
confirm by looking at the quoted Cohen passages above, even cursory glances at the
4
See PWI, Part II, where I provide ten additional detailed case studies.
paradigm targets make it extremely unlikely that such evidence will be forthcoming.
Philosophical engagement with cases simply doesn’t fit the Option 3 mold.
Option 4: Replace ‘intuitions’ with ‘philosophically relevant beliefs about

cases made quickly or using System-1 processing (as opposed to
System-2 processing)’
Option 4 is very close to Option 3, but it gains, at least in some eyes, respectability by
being framed in terminology borrowed from a branch of experimental psychology.
Here is Jennifer Nagel’s characterization of the kind of judgement good x-phi should
target (she is using ‘intuition’-vocabulary, but the context makes it easy to see how that
can be replaced with an ‘intuition’-free description of the relevant kind of judgement):
Both in philosophy and in psychology, intuitive judgments are seen in contrast to the judgments
we produce through deliberate reasoning. Because this point is emphasized in a particularly
clear fashion by Hugo Mercier and Dan Sperber, I follow the outlines of their view in what fol-
lows, and use their terms ‘intuitive’ and ‘reflective’ for the two contrasting kinds of judgment.
Mercier and Sperber describe intuitive judgments as generated by ‘processes that take place
inside individuals without being controlled by them’ (Mercier and Sperber, 2009, 153). The spon-
taneous inferences produced by these processes modify or update what we believe ‘without the
individual’s attending to what justifies this modification’ (ibid.). (2012, p. 498)
This kind of intuitive (System 1) judgement is contrasted with reflective (System
2) judgement. Here is Nagel’s helpful sketch of the latter:
In reflective judgment, by contrast, we engage in explicit reasoning and devote personal-level
attention to the grounds of the conclusions we reach. Because of the strict capacity limitations
on conscious attention, reflective thinking is sequential in character; where intuitive judgment
can integrate large amounts of information very rapidly in associative parallel processing, reflec-
tive judgment is restricted by the bottleneck of limited working memory space (for detailed
discussion, see Evans, 2007). However, what it lacks in speed, reflective judgment makes up in
flexibility (a point particularly emphasized in Stanovich, 2005). (2012, pp. 498–9)
Option 4 proposes that x-phi targets judgements of the first category, not the second.
My response is by now predictable: for anyone doing philosophy, it will come as a huge
surprise to be told that philosophers’ judgements about cases belong in the first and
not the second category. It is, for reasons given above, simply false that the parts of
philosophy Nagel and other practitioners of experimental philosophy target fall into
the first category (the one many psychologists label ‘intuitive’). No evidence has ever
been provided by any proponent of x-phi and recent studies of philosophical prac-
tice throw serious doubt on it.5 The passage from Cohen illustrates the general point.
Cohen thought about the airport case for years, it wasn’t quick and it is obvious from
the text that it doesn’t fit Nagel’s description of what x-phi targets. In sum: while there
is no doubt that System 1 judgements are an important psychological category that
5
See PWI, Part II.
276 Herman Cappelen
psychologists now have interesting insights into, this fails to have any relevance to phi-
losophers’ judgements about cases.6

about cases based on nothing but conceptual competence’
A natural thought: the problem with Options 3 and 4 is that the judgements in ques-
tion are claimed to have no evidence in their favour. What we are looking for, rather,
are judgements with a distinctive kind of evidence in their favour. One common view is
that the target judgements about cases are those based solely on conceptual competence.
Here is yet another of Weinberg’s characterizations of his target:
. . . no empirical evidence is required, because one is presumed to have stipulated all the contin-
gencies in the construction of the hypothetical, and one is thus applying only one’s mastery of
the concepts involved and not any empirical knowledge. (2007, p. 320)
So the target judgements are not without any justification, they are instead based on
a distinctive source of evidence: conceptual competence. This version of x-phi goes
hand in hand with the view of philosophy—and reflections on thought experiments in
particular—as an a priori, armchair, enterprise.
There is much to be said about this way of describing philosophical judgements, and
(as with Option 3) much of what I have to say is spelled out in considerably more detail
in PWI (see chs 7, 8, 9, and in particular 10). I will simply highlight three important
concerns:
1. An x-phi practitioner who advocates Option 5 needs to do a great deal of work
in order to establish that a particular claim is an appropriate target—i.e. that it
is ‘based on nothing but conceptual competence’. No proponent of x-phi has
ever tried to substantiate this about any of their targets.7 Nor have they provided
evidence that the original authors of the thought experiments treated them in
that way. To do that would require serious philosophical work: before prepar-
ing a survey, the x-phi practitioner would have to tell us what she means by
‘concept’, ‘competence’ and ‘based on nothing but’ and then show how, in the
target cases, the relevant judgement has (or is believed to have) these properties.
Significant parts of PWI are devoted to showing how prima facie implausible
this is as a description of typical x-phi targets. In c hapters 7, 8, 9, and 10, I first
develop diagnostics for when a philosopher is attempting to rely on nothing but
conceptual competence—i.e. for when a restriction of the form ‘don’t base the
6 I should note that Nagel also says, “Even if we need some mixture of intuitive and reflective cognition to
follow the story, it is possible that we use intuitive processing across the board in determining whether the
key mental state in the story is an instance of knowledge or mere belief ” (2012, p. 500). This indicates a kind
of mixed picture. I suspect that when this is worked out, the result is a version of Option 6 below.
7 And I don’t mean this as a form of hyperbole—it is literally true: no one has even tried.
judgement on anything but conceptual competence’ is in effect. I then consider

a series of case studies in chapter 8 and none of them fit that description.8
2. It might be illuminating to compare the way PWI and Timothy Williamson
(2007) argue against Option 5. Williamson (2007, ch. 4) considers a range of
variations on the idea that in order to possess a concept C (or understand a word
W) an agent has to assent to certain propositions involving C or be justified in
believing certain propositions involving C (or assent to certain sentences con-
taining W or be justified in assenting to certain sentences containing W). He
presents powerful arguments against such views. If Williamson is right, ‘con-
ceptual analysis’ as it is traditionally understood is impossible to engage in. The
search for conceptual justification is like the search for unicorns. This would
present a problem for the Option 5-version of x-phi: it would be targeting an
activity in which it is impossible to engage (in effect targeting nothing). PWI,
on the other hand, argues against Option 5 without assuming Williamson’s
strong conclusion (i.e. PWI doesn’t assume that the notion of a judgement rely-
ing on nothing but conceptual competence is empty). PWI’s line of argument goes
something like this: suppose for the sake of argument that it is possible to make
a judgement ‘relying on nothing but conceptual competence’. If that were the
case: how would we know (or have reason to think) that someone is trying to
do that? Chapter 7 presents diagnostics and c hapter 8 uses those to investigate
a series of case studies. The conclusion is an unambiguously negative one: it is
a mistake to think that philosophers who reflect on thought experiments shun
evidence that goes beyond what can be reached by ‘relying on nothing but con-
ceptual competence’.9
3. Suppose a proponent of x-phi meets the challenges in (1)–(2) above. Suppose
she has established about a particular target judgement that it is based ‘solely
on conceptual competence’ (or that author intends for it to be so based). She
will then have to show how responses to surveys are relevant to an evaluation
of that judgement. It is hard to overemphasize how difficult a task this is. She
must, in effect, show that subjects ‘rely on nothing but conceptual competence’
when responding to the survey (and are not, for example, just making a spontane-
ous, System 1 judgement). Otherwise, the judgements of the survey respondents
and the judgements of the targeted philosophers aren’t of the same type (and so,
again, there is a mystery why the one should tell us anything about the other).
8 This is despite the fact that many philosophers—in moments of meta-reflection—might describe what
they do as some form of conceptual analysis. That is simply a mis-description of what they actually do.
A general theme of PWI is that we should not take philosophers’ meta-descriptions at face value (no more
than we should take mathematicians’ or economists’ descriptions of the nature of their disciplines at face
value).
9 One advantage of the PWI strategy over Williamson’s: suppose Williamson is right. An advocate of
Option-5-x-phi could grant that and say that their target is the activity of trying to do philosophy as a form of
conceptual analysis (i.e., trying to do what Williamson argues is impossible). PWI shows that they don’t.
278 Herman Cappelen
No x-phi survey has ever even attempted to produce evidence of this kind of
match.10
Before leaving Option 5 it is worth drawing attention to a little argument that can be
extracted from the Weinberg quote that started this section. Recall Weinberg (2007)
saying, “no empirical evidence is required, because one is presumed to have stipu-
lated all the contingencies in the construction of the hypothetical . . . .” The argument
seems to be of the form: “The reasoning must be based just on conceptual competence,
because we’ve been given so much starting information to reason from.” But that’s a
bizarre line of thought: surely, the amount of detail built into the case is prima facie
evidence that the reasoning isn’t by conceptual competence, since it can be reasoning
from those details. Compare: I study some organism closely in the lab and, on the basis
of the study, conclude that it’s a mosquito. It doesn’t follow that my conclusion must
have been based on nothing but conceptual competence, because I had so much data
from the lab that there was no further useful data that could have been added to it.11

about cases for which we have too little “standard” evidence’
You might think that what x-phi needs as a target is a set of judgements that the agent
has thought a lot about, and has arguments for and against, but where there is still a lit-
tle bit of room for ‘something else’. This ‘something else’ leaves room for x-phi surveys.
Weinberg (Forthcoming) makes a proposal along these lines. He suggests that there is
a set of judgements with properties (I1)–(I3):12, 13
• I1: The judger takes herself already to have evidence for p, in a manner substan-
tial enough to license p prima facie as a justifier without need of further offered
justification.
• I2: The justification the judger takes herself to have for p “is not itself wholly based
on perceptual input, memory, testimony, or inference from such judgements”
(note that this leaves open the possibility that the intuitive judgement may par-
tially involve such sources—thereby distinguishing this proposal from Option 3).
• I3: The judger assumes that there is some non-standard justification for p that
complements the justification she has from ‘standard sources’.
So, roughly, Weinberg’s idea is that there are important claims about cases in which
the agent has a confidence level that exceeds what the ‘standard’ evidence justifies. The
10 For an elaboration of these points, see PWI, ch. 10.

11 Thanks to Josh Dever for drawing my attention to this problem-argument in the passage from Weinberg.
12 Weinberg also thinks this is how ‘intuition’ should be characterized, so he takes this as a defence of
the claim that philosophers rely on intuitions. Since the current chapter assumes we are better off purging
our discussions of ‘intuition’ vocabulary, I ignore this aspect of Weinberg’s proposal (though see Cappelen
(Forthcoming) for a refutation of this aspect of Weinberg’s proposal).
13 It’s unclear whether these are necessary or sufficient conditions.
agent also believes that some ‘non-standard source of evidence’ will fill the evidential
gap and justify the confidence level. Those are the judgements targeted by x-phi sur-
veys. One alleged advantage of Option 6 is that it avoids the objection raised against
Options 3 and 4, i.e. that philosophers rely on a variety of sources of evidence for their
judgements about cases. This becomes irrelevant since the x-phi fan can say, “Yes, of
course Cohen and others in your case studies present reasons and evidence for their
claims, but so what? I didn’t mean to rule that out—I just want a case where there’s also
a mysterious ‘something else’ and that’s x-phi’s target.”
This option suffers from the same problems as Option 5. On the current proposal,
the x-phi practitioner assumes that the target claim relies, in part, on some unspecified
‘non-standard’ evidential sources. The ‘unspecified’ part here is in effect just rhetoric
since the only option for ‘non-standard evidential source’ that is live in the current
debate is ‘reliance on conceptual competence’. Absent some alternative, all the con-
cerns raised in connection with Option 5 apply (or, if you insist on talking about ‘some
unspecified evidential sources’ replace ‘conceptual competence’ in the objection to
Option 5 with ‘some unspecified non-standard evidential sources’ and all the same
worries will apply).
Rather than repeat all of the points made in response to Option 5, I want to
re-emphasize one of the objections. Here is the concern articulated using ‘non-standard
evidential source’ instead of ‘conceptual competence’: suppose some target has been
properly identified. An x-phi practitioner would have to commit to an account of what
this ‘non-standard source of evidence’ is, and in the light of that she would have to
show that responses to surveys reveal something significant about this non-standard
source. This requires showing that those who respond to the surveys base their judge-
ments on this non-standard source (and not, for example, making a spontaneous,
System 1 judgement). No attempt has yet been made to establish that.
Option 7: Replace ‘intuitions’ with ‘judgements of philosophically relevant

beliefs about esoteric cases’
In one of his papers, Weinberg makes yet another attempt at characterizing the target
of x-phi. He suggests that the primary target of x-phi is the practice of making judge-
ments about ‘esoteric, unusual, farfetched’ cases. Weinberg (2007) says:
Intuitions may be fine as a class, taken on the whole, and the opponent has neither the need nor
the desire to attack that whole class. But philosophers do not invoke a vast undifferentiated mass
of intuitions in defense of their claims—rather, we cite particular intuitions about particular
hypothetical cases. And the opponent is concerned that some significant number of these cases
may be far less than ideal for this sort of appeal. For the practice appears to set no constraints
on how esoteric, unusual, farfetched, or generally outlandish any given case may be. Everyone
is familiar with the likes of Davidson’s Swampman and Searle’s Chinese room, but one can look
at the very recent literature and find the likes of double-lesioned testifiers, new evil demons, and
fissioning/fusioning/teleporting pairs (or are they?) of persons. So this anything-goes aspect of
the practice is what makes it particularly ripe for the opponents’ challenge. (p. 321)
280 Herman Cappelen
Purged of ‘intuition’-talk something like the following claim remains:

JW1: Philosophers’ practice of making judgements about esoteric, unusual and far-
fetched cases is particularly unreliable and dubious. They are the judgements that
are vulnerable to the experimentalist’s challenge.
I find the appeal to the esoteric, unusual and farfetched puzzling since these features
are not in any way correlated with the degree of difficulty of a judgement. Lots of
strange and unusual cases are very easy to judge in a reliable way. Here is a very eso-
teric, unusual, and farfetched case:
Easy Esoteric and Farfetched Case: Suppose there are two pink elephants in my
office. Then yet another pink elephant comes into my office (and the first two pink
elephants stay in the room). Question: How many pink elephants are in my office?
Presumably, Weinberg does not think that if a philosopher reflects on this case, she is
engaged in the dubious and suspicious part of our philosophical practice. The experi-
mental challenge better not target this ‘thought experiment’. The flipside of this is that
many of the normal (not farfetched, not unusual, not esoteric) cases are very hard.
John Perry’s famous paper, “The Problem of the Essential Indexical” (1979), starts out
with a perfectly ordinary case—a kind of event that happens all of the time. One of the
main points emphasized by Tyler Burge in “Individualism and the Mental” (1979) is
the ordinariness of his cases—how they are ubiquitous.14 But, as the last thirty years
of philosophy have proved, these are extremely hard cases to make judgements about.
I suspect the charitable interpretation of Weinberg is that he is focusing on esoteric,
unusual, and farfetched cases that are difficult and that it is the great degree of difficulty
that distinguishes his target, not that they are esoteric, unusual, or farfetched. So, let’s
consider JW2:
JW2: Philosophers’ practice of making judgements about very difficult cases is par-
ticularly unreliable. It is such judgements that are vulnerable to the experimentalist’s
challenge.
JW2 moves dangerously close to triviality. Of course we are less reliable when we make
very difficult judgements. That’s true in general—it is not a surprising feature of philo-
sophical practice. When we try to answer very hard questions, we are not as reliable as
we are when we try to answer easy questions. Surveys of undergraduate responses to
vignettes won’t make finding the answer any easier.
Option 8: Define ‘experimental philosopher’ as ‘any philosopher who does

relevant experiments’
For a discussion of Perry’s and Burge’s cases and their relevance to this debate, see PWI, ch. 8.1–2.
14
This option constitutes a radical change in strategy: in a recent paper, David Rose and
David Danks propose experimental philosophers abandon what they call ‘the narrow
conception of experimental philosophy’:
. . . discussions of experimental philosophy have often been ambiguous about what exactly
experimental philosophy is. For example, Knobe and Nichols, in the introduction to their
Experimental Philosophy volume, propose that “experimental philosophers [are those who]
proceed by conducting experimental investigations of the psychological processes underlying
people’s intuitions about central philosophical issues” (p. 3). Similarly, Nadelhoffer and Nahmias
(2007, 123) write: “Experimental philosophy is . . . a recent movement whose participants use the
methods of experimental psychology to probe the way people make judgments [i.e., have intui-
tions] that bear on debates in philosophy.” Liao et al. (2011, 2) observe that “a number of philos-
ophers have conducted empirical studies that survey people’s intuitions about various subject
matters in philosophy,” and describe their own work as using “this method of experimental phi-
losophy.” These quotes and other writings suggest a narrow conception of experimental philoso-
phy: experimental philosophy involves philosophers conducting psychological experiments for
which the primary target is intuitions or judgments. (2013, pp. 513–14)
In place of this narrow conception, they propose a broad conception:

We should instead adopt a broad conception of experimental philosophy: experimental philoso-
phy is simply an instantiation of the long tradition of philosophical naturalism—the view that
empirical data are relevant to certain philosophical questions—coupled with actually conduct-
ing some of the relevant experiments, as necessary. (p. 515)
In a way, Option 8 is sympathetic to the spirit of this chapter and to the line of argument
in PWI. Of course, no one should object to the idea that philosophers do experiments
(consider, for example, Barry Smith’s extensive experimentation while writing about
the philosophy of wine15—who could object?). What is important to emphasize is that
this is compatible with everything written by experimental philosophers being false,
attacking a straw man, and having no relevance whatsoever to the debates to which they
were supposed to be relevant. The redefinition of ‘x-phi’ goes no way toward defending
work that has already been done by experimental philosophers against the objections
in, e.g., PWI and Williamson (2007).
Joshua Knobe (in an email exchange) points out to me that what I just said has to be
interpreted with care. Knobe suggests we distinguish between the metaphilosophy of
x-phi and first-order x-phi (or the practice of x-phi), and says:
The metaphilosophical writings of experimental philosophers keep emphasizing this notion of
‘philosophical intuition’, but if you look at the majority of the actual experimental work being
conducted, what you find is that only a small percentage of it actually depends on anything of
15 For examples of this version of experimental philosophy, see for example the events page of the London
Experimental Oenology Seminars at <http://experimentaloenology.wordpress.com/past-activities/> [last
282 Herman Cappelen
the kind. Most of it is just straightforward attempts to contribute to the interdisciplinary field of
cognitive science.
Think of the following as the Knobe-defence of x-phi: the surveys we do are intrinsi-

cally interesting and they might throw some light on some aspects of human cogni-
tion. I grant that the metaphilosophy of x-phi presented in the kinds of passages you
quote is mistaken. The received metaphilosophy of x-phi—the spin that has made it
seem philosophically significant—is mistaken. This chapter (and the criticism in PWI)
is focused exclusively on the metaphilosophy of x-phi. The claim I have made is this: if
we take x-phi to be as characterized by the metaphilosophers of x-phi, then it is a failed
and misguided research project. The Knobe-reply (as I interpret it) grants this, and
adds: okay, but that just shows that the ‘official’ metaphilosophy of x-phi is wrong. Here
is a helpful analogy (also due to Knobe): just as PWI argues that a great deal of (non-
x-phi) metaphilosophy (e.g. Bealer, Sosa, Goldman) mischaracterizes (non-x-phi)
philosophical practice, proponents of the Knobe-reply think that the metaphilosophy
of x-phi mischaracterizes x-phi practice. I tend to agree with Knobe and I wish I had
highlighted that option in PWI. My goal was to point out the mistaken meta-remarks
about x-phi. Of course, proponents of the Knobe defence are left without a metaphi-
losophy. They owe us an answer to the question: what is the relevance of this work to
philosophy? If you favour the Knobe-reply, think of this paper as a challenge: articulate
the philosophical relevance of survey responses without resorting to one of the options
dismissed above.
A final remark about the Knobe reply: it was the meta-philosophy of x-phi—i.e. the
framing of the project in terms of an attack on traditional philosophical methodol-
ogy—that made the movement exciting. That spin is primarily responsible for the
enormous attention paid to the movement. I think there would be much less enthu-
siasm for the project if it was simply some philosophers who did some experiments
in collaboration with psychologists, and left it entirely open what, if any philosophical
significance those experiments have.
14.3. Comparison with the ‘Expertise Reply’

A comparison with the so-called expertise reply to x-phi (and what Weinberg and
Alexander (2014) call ‘The Methodological Reply’) might help illuminate motivations
behind this paper.
The Expertise Reply: The problem with x-phi isn’t that it studies intuitions, but rather
that it studies the intuitions of the wrong subjects. The mistake, according to this reply,
was to focus on undergraduates. If, instead, the surveys were of professional philoso-
phers, the practice would be valuable and philosophically relevant.
The Methodological Reply: “According to methodological approaches, genuine phil-
osophical intuitions are the products of careful philosophical reflection” (p. 206).
These lines of reply to x-phi are misguided because they buy into the same mis-
taken picture of philosophy that spawned x-phi in the first place. The picture is that we
engage in philosophical reflection and training and then when presented with survey
questions, the responses are of a fancy kind—reflective survey replies. But this is not
how philosophy is done. Philosophers engage in careful reflection and reason-giving,
but they don’t do that to become more sophisticated survey respondents. X-phi’s basic
mistake is to treat philosophizing as an undergraduate multiple-choice exam: yes/no/
maybe to p? That is not how philosophy is done.
14.4. Objection: Don’t you care about cognitive

biases?
Even taking onboard the above criticisms, there’s sometimes a lingering sense that a
wholesale dismissal of x-phi is unreasonable. After all, on one construal they are sim-
ply trying to discover cognitive biases. Do I mean to reject all attempts to discover
biases that philosophers might suffer from? In response to this concern, it might help
to sketch the kind of dialogue I sometimes have with proponents of x-phi. In what
follows, X-P is someone trying to explain to me the value of x-phi surveys and HC
presents my reply.
x-p: Don’t you care about cognitive biases?
HC: Yes, I do. The literature on heuristics and biases is very interesting.
X-P: Good, then what we experimental philosophers have found should interest
you: philosophers are subject to all kinds of surprising biases, e.g. ordering
effects.
HC: What do you mean by ‘philosophers’? You mean philosophers when they do
philosophy or in some other capacity?
X-P: Well, we haven’t done any experiments on philosophers yet, but what the
evidence points to is this: when philosophers give responses to survey ques-
tions about vignettes, they are subject e.g. to ordering effects.
HC: Okay, but that’s not philosophers doing philosophy. Doing philosophy is
thinking about and then giving reasons for a judgement about a case and the
judgement is evaluated based on the quality of the reasons given. No part of
philosophy is like responding to survey questions about vignettes. So even
if this data were right, it wouldn’t tell us about biases in philosophical prac-
tice. It would tell us that when philosophers respond to vignettes, they are
subject to ordering effects, but that’s at best an unsurprising factoid of no
philosophical value.
X-P: I disagree: lots of philosophers just make quick, brute, immediate, spon-
taneous, judgements about cases. Those judgements play a huge role in
284 Herman Cappelen
philosophy and the biases we have uncovered will help us understand that
part of philosophical practice.
HC: That’s an empirical claim about philosophy. You have, literally, nothing to
back that up and there’s overwhelming data against it (see, e.g., PWI, Part II).
Now, at this point in the dialectic we can go through the kind of arguments spelled out
above, but the point about biases can then be picked up again:
X-P: So is your view that we should not care about or try to find biases in phil-
osophical practice? We should just ignore them? Isn’t that an irresponsible
intellectual attitude?
HC: That’s not my view. There are lots of biases in philosophy. Where you and
I differ is on how to discover and evaluate those. I think we discover those
by doing philosophy—by finding assumptions and presuppositions that
bias philosophical reflection. Construed that way, I see myself as spending
a great deal of my time as a philosopher discovering philosophical biases.
The discussion at the core of this paper is a good illustration: I think much
metaphilosophy is biased in favour of the view that philosophical practice
is intuition-based. That bias has distorted metaphilosophical reflections. But
discovering that bias, and justifying opposition to it, doesn’t involve surveys.
It is about doing philosophy like we were trained to do it: it involves thinking
hard about arguments, uncovering hidden assumptions and then question-
ing them. Note that this doesn’t mean that I don’t think there are all kinds
of non-philosophical factors that at some level influence the practice of phi-
losophy. Proponents of x-phi provide a great illustration. It’s striking, to put it
mildly, that so many prominent experimental philosophers have either stud-
ied under or worked with Stephen Stich (these include Edouard Machery,
Jonathan Weinberg, Ron Mallon, and Joshua Alexander). It is likely that these
close personal connections make them biased in favour of the movement
Stich founded. That, however, is obviously irrelevant to an evaluation of the
x-phi movement or to the more specific claims they make.16
14.5. Conclusion
The offhand remarks we get from x-phi lovers about how to do x-phi without intui-
tions are unhelpful. They give the impression that it’s a tiny adjustment hardly worth
16
Stewart Cohen suggested that in the spirit of reconciliation, this piece of dialogue be added:
x-p: Okay, maybe x-phi surveys are irrelevant to finished work by professional philosophers, but, surely,
even professional philosophers occasionally engaged in quick, non-reflective philosophical reflections
and conversations—can x-phi studies be relevant to what they do during those informal moments?
HC: Maybe, it’s an empirical question—one would need to do empirical studies to show the relevance, it
can’t be just a priori assumed. And one would then need evidence of the kind of effect this had on the
finished work.
mentioning other than in a parenthetical remark. That’s deeply misleading: the issues

raised are complicated and the threat to x-phi is serious. In sum, a foundational prob-
lem threatens x-phi: the movement has been based on a speculative, non-empirical
approach to its subject matter. Before you can claim that a survey (or a study of any
kind) shows something important about an activity or a group of people, you need
to carefully study what those people do and what that activity is. Proponents of x-phi
have bought into a caricature of philosophy and they need to do more careful empirical
work to identify a genuine target. Here is an analogy of what has been going on: sup-
pose experimental philosophers decided to investigate philosophers’ diet, rather than
their judgements about thought experiments. We can all agree that the following
would be an extraordinarily ill-conceived strategy: pick some chemical, XYZ, and then
conduct various experiments on it before checking whether philosophers consume
XYZ. Even if the philosophy-diet-experimenters discover that XYZ is unhealthy and
dangerous, nothing follows about the health of philosophers or their dietary habits.
Why does any of this matter? Can’t those of us who don’t care for x-phi just leave the
x-phi lovers alone to do their business? Is it a harmful activity? My answer to this last
question is a tentative ‘yes’. Not only does x-phi promote a false picture of philosophy,
but, insofar as the programme is successful, it will change philosophy. Suppose I am
right in my claim that philosophers don’t rely on intuitions or anything that can be
measured by responses to x-phi surveys. With the institutionalization of x-phi, interest
in and reliance on surveys have gradually become integral parts of professional philos-
ophy. So though x-phi was born in sin, so to speak, its institutional foothold has made
it a truism that philosophers care about survey responses, since they (i.e. experimental
philosophers) do. I agree with one of Stich’s motivating thoughts: philosophy should
not be based on the kinds of judgements people make when responding to surveys.
That is an awful (borderline absurd) way to do philosophy. The institutionalization
of x-phi has made it the case that many philosophers now think those kinds of judge-
ments are important. They have, in effect, created the practice they set out to undermine.
If that is right, then even x-phi lovers should agree that their influence is damaging.
References
Alexander, J., and Weinberg, A. M. (2007). “Analytic Epistemology and Experimental
Philosophy,” Philosophy Compass 2, pp. 56–80.
Burge, T. (1979). “Individualism and the Mental,” Midwest Studies in Philosophy 4, pp. 73–121.
Cappelen, H. (2012). Philosophy without Intuitions. Oxford: Oxford University Press.
——. (Forthcoming). “Reply to Bengson, Chalmers, Weatherson and Weinberg,” Philosophical
Studies.
——., and Hawthorne, J. (2009). Relativism and Monadic Truth. Oxford: Oxford University Press.
Cohen, S. (1999). “Contextualism, Skepticism and the Structure of Reasons,” Philosophical
Deutsch, M. (2009). “Experimental Philosophy and the Theory of Reference,” Mind and
Language 24, pp. 445–66.
286 Herman Cappelen
Deutsch, M. (2010). “Intuitions, Counter-Examples, and Experimental Philosophy,” Review of

Philosophical Psychology 1, pp. 447–60.
Evans, J. (2007). “Dual-processing Accounts of Reasoning, Judgment, and Social Cognition,”
Annual Review of Psychology 59, pp. 255–78.
Knobe, J., and Nichols, S. (eds) (2008). “An Experimental Philosophy Manifesto,” in
Experimental Philosophy. Oxford: Oxford University Press.
Liao, S. M., Weigmann, A., Alexander, J., and Vong, G. (2011). “Putting the Trolley in Order:
Experimental Philosophy and the Loop Case,” Philosophical Psychology 5, pp. 661–71.
Mercier, H., and Sperber, D. (2009). “Intuitive and Reflective Inferences,” in J. Evans and K.
Frankish (eds), In Two Minds: Dual Processes and Beyond. New York: Oxford University Press,
pp. 149–70.
Nadelhoffer, T., and Nahmias, E. (2007). “The Past and Future of Experimental Philosophy,”
Philosophical Explorations 10, pp. 123–49.
Nagel, J. (2012). “Intuitions and Experiments: A Defense of the Case Method in Epistemology,”
Nichols, S., Stich, S., and Weinberg, J. M. (2001). “Normativity and Epistemic Intuitions,”
Perry, J. (1979). “The Problem of the Essential Indexical,” Noûs 13, pp. 3–21.
Rose, D., and Danks D. (2013). “In Defense of a Broad Conception of Experimental Philosophy,”
Metaphilosophy 44, pp. 512–32.
Stanovich, K. E. (2005). The Robot’s Rebellion: Finding Meaning in the Age of Darwin. Chicago:
University of Chicago Press.
Swain, S., Joshua A., and Weinberg, J. M. (2008). “The Instability of Philosophical
76, pp. 138–55.
Weinberg, J. M. (2007). “How to Challenge Intuitions Empirically Without Risking Skepticism,”
——. (Forthcoming). “Cappelen Between Rock and a Hard Place,” Philosophical Studies.
Weinberg, J. M., Gonnerman, C., Buckner, C., and Alexander J. (2010). “Are Philosophers Expert
Weinberg, J. M., and Alexander, J. (2014). “The Challenge of Sticking with Intuitions through
Thick and Thin,” in A. R. Booth and D. P. Rowbottom (eds), Intuitions. Oxford: Oxford
University Press, pp. 187–212.
Index
a priori 2–4, 31–32, 47, 53, 56, 58–59, 63, 65–67, competence 3, 46–49, 71, 93–94, 97–98, 105, 142,
87, 89, 91–97, 99–105, 107, 109, 111–115, 122, 191, 194, 198, 204–206, 209, 227, 276–279
134, 163, 210, 229, 257, 259–263, 267–268, counterexamples 60, 76, 115, 231
276, 284 counterfactuals 2, 250, 268
action:
mental 9, 20, 21, 26, 33–34 definitions 2, 103, 120, 157, 257, 265–267
agency 28, 37, 40–41, 47–48, 78, 146 deliberation 40, 46
analytic 2–3, 6, 92, 94–97, 102–103, 114, 142, 147, DePaul, M. 1, 6, 10, 31–32, 67, 91, 113–114, 133, 141,
209, 222, 232, 237, 240, 249–250, 252–253, 165, 210–212, 229–231, 253
270, 272, 285 Descartes, R. 11–12, 15, 32, 100, 216
Anscombe, E. 3, 22, 31, 56, 67, 231 disposition 3, 10–11, 31, 49, 69, 79, 81–84, 86–89,
argument: 101, 183
pumps 4, 119, 121, 123, 125, 127–129, 131, 133 dualism 234, 280, 285
Aristotle 79, 95, 257, 262–264
armchair 3, 52–54, 58, 66, 120, 125, 135, 169, 192, empiricism 4, 64–67, 89, 112, 114, 120, 123, 127,
195–201, 205–206, 208–210, 232–236, 238, 132, 134, 189
240–246, 248–249, 251–253, 270, 276 epistemic:
autonomy 2, 4, 31, 113, 161, 210, 229, 253 circularity 63, 67, 205, 230
justification 3, 14, 16, 41, 49, 100, 163, 210, 229,
Bealer, G. 2, 10, 12, 31, 70–73, 83, 89, 93, 95, 234, 246, 248, 253
97–98, 107, 113, 188–191, 193, 196, 199–200, norms 234, 244, 247, 250
208–210, 225, 229, 236–237, 253, 282 status 16, 33, 59, 87, 90, 113, 149, 154, 162,
begging the question 114, 210 188–189, 210, 240, 254
bias 44, 110, 135, 137–138, 144, 146–147, 207, 212, epistemology 4–5, 32–34, 39, 56, 64, 67, 89, 92,
241, 243, 283–284 95, 113–114, 120, 134, 146, 149, 154, 162, 165,
Boghossian, P. 10, 15, 17–24, 26, 31 168, 170, 183–184, 209–211, 215, 219, 222,
Brown, J. R. 4, 84, 119–127, 129–130, 133, 146 228–231, 234, 236, 238, 243–244, 249–251,
253–254, 264, 286
Carroll, L. 10, 19, 32 essence 6, 28, 176, 213, 256–257, 259, 261–268
Carrollian argument 17–18, 24, 26, 29–30 evidence 3, 6, 9, 12–14, 19, 32, 34, 50–55, 57,
category 25, 28, 94, 100, 106, 187, 189, 205, 268, 59–63, 65–67, 69–73, 75–78, 80, 85–86, 93,
275 96–98, 100–104, 107–108, 110–111, 114, 120,
causal-requirement 51–56, 58–59 128, 136–137, 142, 145, 150, 154, 156, 158–159,
causation 23, 32, 56–57, 67, 252 161–163, 166, 168–171, 180–181, 188–190,
clarity 49, 99, 102–103, 191 192–195, 203–204, 206–208, 235–242,
cognition 90, 146, 184, 196, 200, 202–204, 244–246, 249, 254, 256, 258–259, 269–279,
206–207, 210–212, 233, 254, 276, 286 283–284
cognitive science 146, 193, 204–205, 207, evolution 78, 109, 134
210–211, 282 experimental philosophy (see also X-phi) 2,
coherence 154–156, 162–164, 203, 220 5, 89–90, 114, 135–136, 141, 143–147, 169,
concepts 1–2, 6, 31, 35, 49, 72–73, 91, 95, 97, 99, 188, 200–201, 209, 213, 227, 229–230, 239,
101, 103–108, 110–111, 113, 115, 120, 133–134, 253–255, 269, 271, 281, 285–286
144, 158, 191, 193–194, 205, 211–212, 215, 217, experimentalist critique 5–6, 33, 232–233, 235,
220–222, 237, 245, 248–249, 256, 271, 276 237–243, 245, 247, 249, 251–253, 255
conceptual analysis 1–2, 88–90, 93, 95–96, expertise 6, 17, 138–139, 141–143, 146, 193, 197,
105–106, 113, 264, 277 204, 212–213, 221–222, 225–228, 282
conditionals 87, 257, 264 explanation 29, 37, 46, 69, 71–72, 79, 88–90,
consciousness 10–11, 32, 34, 49, 55, 70–71, 87, 102, 109, 120, 122, 132–133, 153, 162, 170–171,
202, 206 174–175, 182–183, 202, 250, 262
contextualism 166–169, 183–184, 285 externalism 67, 110, 229–231, 253
288 Index
folk: metaphysics 2, 6, 90, 95, 113, 120, 134, 146, 229,

physics 78, 88 243, 256–257, 259, 263, 266–268
psychology 88, 205, 209, 211 metasemantics 73, 75, 77–79, 82, 85
semantics 77–79 methodology 2, 4–6, 93–95, 104, 105, 109, 115,
Fodor, J. 74–75, 78, 82, 90 145, 189–190, 193, 196, 210, 222–223, 225,
foundationalism 154, 159 228, 230, 232–234, 236, 240–243, 252–254,
256
Galileo 121, 139 modality 134, 200, 264–265, 268
Gettier, E. 101, 113, 184, 188, 210, 232, 253 moral:
gettier case 72, 140–141, 239–240, 249, 273 epistemology 4–5, 33, 254
Gutting, G. 1–2, 187, 210 intuitionism 4–5, 33, 88, 148, 151, 153, 156–157,
159, 161, 163–164
habits 176, 285 Muller-Lyer:
Harman, G. 26–27, 33, 90, 148, 159, 161, 163, 233, illusion 41, 223
254 lines 40, 46
Humean 23, 30, 32
natural kind 73, 77, 83–84, 105–106, 268
inference 3, 10, 14–26, 28–30, 44, 58, 65–66, naturalism 120, 122, 153, 289
94, 120, 151, 153–154, 156, 161–162, 200, 202, necessity 90, 93, 102, 105, 114, 126, 144, 146, 190,
235–236, 263, 275, 278, 286 199–200, 254, 257, 268
intellectualism 167, 183 Newton, I. 136, 138–139
internalism 10, 14–20, 24, 67, 110, 152, 232, 253 Norton, J. 4, 119–120, 122, 127–128, 130–134, 146
introspection 52–54, 56–58, 61–62, 200, 206,
212, 237 observation 12, 26–29, 57, 119, 143–144, 170, 172,
intuition: 180, 227, 244, 256–257, 259–260
epistemic 44, 49, 114, 147, 184, 230, 243, 255, ontology (see also metaphysics) 13, 268
273, 286 ordinary language 2, 144, 170, 215, 249, 251
linguistic 3, 5, 86, 165, 168
philosophical 33, 90, 133, 147, 163, 190, 192, perception 2–3, 9–10, 20–23, 25–27, 32–33, 36,
195–196, 199–200, 202–206, 208–209, 211, 40–41, 44, 46, 57–58, 62, 67, 70–71, 119,
227, 230,233, 241, 254–255 124–126, 135–136, 139–140, 159, 172, 188, 258
rational 46–47, 49–50, 53–55, 58–59, 61–62, phenomenology 32–35, 95, 190, 200, 203
64–66, 94, 254 Plato 32, 67, 95, 100, 108, 114, 123–124, 126, 211,
role of 3, 15, 17, 19, 27, 32–33, 35, 47, 113–114, 119, 229, 250
122, 210–211, 229, 238, 244, 253 possible world 87, 132, 250, 258–261
sceptical 213, 215, 217, 219, 221, 223, 227, 231 possibility 5, 26–27, 34, 43, 53, 60, 70–71, 74,
invariantism 5, 165–173, 175, 177, 179, 181, 183–184 80, 89, 95, 106, 111, 120, 127, 133–134, 154,
immediacy 4, 91–92, 94–95, 100 169, 173–174, 176, 182, 195, 197, 200, 215, 217,
imagination 10, 119 219, 223, 225, 227–229, 232, 244–245, 253,
257–259, 261, 263–264, 268, 278
Kant, I. 92–93, 257–258, 268 pluralism 163, 210, 246–248, 251
Knobe, J. 89, 135, 146, 169, 182, 184, 192, 210–211, properties 51, 57, 61, 79, 103, 110, 149, 196, 234,
233, 254–255, 270, 281–282, 286 246, 251, 267, 273, 276, 278
Kripke, S. 69, 75–77, 79–81, 90, 96–98, 114, 141, psychology 1, 14, 31–32, 47, 87–88, 90, 113–114,
144, 232, 234, 257, 260, 263–264, 268 125, 133, 136–137, 145–146, 152, 162, 164, 169,
182–183, 187, 193, 202, 204–206, 209–212,
latitudinarianism 42–43 227, 229–231, 253–254, 275, 282, 286
laws of nature 122, 133–134 pushmi-pullyu representations 15, 27–29, 31, 34
learning 4, 83, 103, 107, 120, 138–139, 141, 197, Putnam, H. 75, 108, 113–114, 214, 230, 232, 254
211, 246
Lewis, D. 100–101, 114, 167, 183, 190, 245 rationality 3, 14, 28–29, 45–46, 66–67, 140
Lowe, E.J. 6, 256, 259–260, 262–266, 268 rationalism 3–4, 50–53, 55, 57, 59, 61–63, 65, 67,
112, 114, 120, 127, 134, 203
memory 44, 58, 71, 138–139, 144, 159, 275, 278 reflection 10–11, 15–18, 22, 24–25, 46, 94–95,
mental state 15–16, 20–21, 23, 25–28, 30–31, 99, 134, 140, 157, 191, 198, 206–210, 212, 220,
37–38, 48, 52, 57, 61, 69, 87, 179, 189, 234–235, 223–224, 229, 253, 256, 258, 273, 276–277,
251, 276 282–284
Index 289
reflective equilibrium 3, 114, 155, 157, 164, 210, 168, 188, 191, 193, 211, 239, 253, 256–257,
229, 253 271–273, 277, 280
relativism 70, 164, 167, 193, 210, 272–273, 285
reliabilism 12, 60, 65, 67 universals 67, 123–127
science 4, 6, 31, 65, 76, 85–87, 90, 95, 112, 119–120, Van Inwagen, P. 101, 115, 190, 212
122, 124–126, 133–136, 140–141, 143, 146,
163–165, 189, 192–193, 204–205, 207–208, warrant 15–16, 18, 70–71, 93, 96, 100, 112, 149,
210–211, 256–259, 267–268, 282 152–154, 158–160, 162, 164, 176, 179–181, 183,
seeming: 194, 245, 251, 258
intellectual 38–39, 44, 47, 72, 93, 189, 223, 239 Weinberg, J. 5, 69–71, 90, 96, 102,
perceptual 41, 47, 223, 239 110–111, 114–115, 135, 147, 169, 184, 190,
semantics 67, 73, 75, 77–79, 82, 85, 90, 146–147, 192–193, 196–198, 209–210, 212, 226–227,
166, 210, 254 230–240, 242, 244–245, 253–255, 270,
scepticism 6, 35, 100, 213–231, 255 272–273, 276, 278–280, 282,
Sosa, E. 2–3, 12, 32, 34, 36, 48–49, 70–71, 90, 284–286
92–93, 95, 101, 111, 113–114, 183–184, 188, 190, Williamson, T. 2–3, 10, 14, 25, 70–71, 88, 90–94,
199, 212–213, 223, 225–226, 230, 232, 240, 96–97, 101–102, 109, 115, 158, 164, 170, 184,
246, 248, 252–254, 282 189–190, 193, 197, 199, 206, 212, 224–225,
speech acts 5, 165, 170, 172, 176, 179 231–233, 237–240, 246, 252, 255, 257, 264,
Stich, S. 69, 78, 80, 89–90, 96, 114, 146–147, 184, 268–269, 277, 281, 286
190, 192, 210, 212, 230–233, 237, 241–242, Wittgenstein, L. 84, 213, 218–220, 227,
244–252, 254–255, 270, 284–286 230–231
thought experiment 2, 4, 6, 67, 69, 75, 82, 86–88, X-Phi (see also experimental philosophy) 1–2,
96, 111, 114, 119–125, 127–131, 134–142, 147, 4–6, 269–279, 281–283, 285

Anthony Robert Booth - Darrell P. Rowbottom - Intuitions-Oxford University Press, USA (2014) PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Anthony Robert Booth - Darrell P. Rowbottom - Intuitions-Oxford University Press, USA (2014) PDF

Încărcat de

Drepturi de autor:

Formate disponibile

Intuitions

art I The Ontological and Epistemological

Part II Intuitions in Disciplines or Sub-Disciplines

Part III Challenges

11. Sceptical Intuitions 213

Joshua Alexander, Siena College

Nineteen ninety-eight saw the publication of a volume of essays entitled Rethinking

metaphilosophical and immediacy bundles. Finally, drawing on some of her earlier

how intuitional judgements are sometimes significantly informed by expertise via a

reflection on the transmission of justification through inference. According to infer-

1.2. What Are Intuitions?

(5) Intuition experiences possess presentational phenomenology; whenever you

If a < 1, then 2 – 2a > 0 (A)

√ 7 + √10 > √ 3 + √17 (B)

Reflection on example intuitions seems to me to provide some motivation for

1.3. Intuitions as Justifiers and as Evidence

Phenomenalism: if an intuition plays the justifier role, it does so in virtue of having a

1.4. Inferential Internalism

[b] (Simple Inferential Internalism): A deductive inference performed by S is

There are important differences among these formulations of inferential internalism.

10 It might be sufficient if reformulated as a principle about propositional justification. But, as indicated

11 How exactly to understand the becausal condition is an issue I will discuss below.

1.5. Boghossian’s Carrollian Argument

(iv) This particular inference from (1) and (2) to (3) is valid.

In Boghossian’s argument (iv) is the content of my intuition that (1) and (2) support

The main premise in Boghossian’s case against inferential internalism is this:

1.6. Intuition in Action

1.7. Objections and Elaborations

of receiving inconsistent directions—i.e. directions that cannot jointly be satisfied. It

judgments are pushmi-pullyu representations.33 And Timothy Bayne (2010) has

1.8. The Ground of Intuitive Guidance

Evans, G. (1982). The Varieties of Reference. Oxford: Oxford University Press.

Mandelbaum, M. (1955). Phenomenology of Moral Experience. Baltimore, MD: Johns Hopkins

Stratton-Lake, P. (2002). Ethical Intuitionism: Re-evaluations. Oxford: Oxford

2.2. What Is an Intuition? Seemings and

Experiencing sensorily as if p is anyhow distinct from its seeming to one as if p (i.e.,

2.3. Deliberating and Pondering

2.4. Sources and Foundations

being properly based rationally on a regress-stopper—namely, the sensory experi-

Here are some examples of such foundational seemings:

2.6. Irrational, Epistemically Ill-Formed Intuitions

2.7. What Distinguishes Intuitions with Probative

2.8. A Competence-Based Account

Intuitions are best understood as intellectual seemings or attractions. Unlike visual

2.9. Intuition and Perception

blindsight, we still need a guiding competence if the subject is to attain justification

2.10. Just How Can Understanding Function as a

3.2. The Causal-Requirement Argument

In Fumerton’s hands, skepticism about our armchair access to evidence sufficient to

The Causal-Requirement Argument—Version 2

5 I thank two anonymous referees for correction here.

3.3. The Reliability Argument

Therefore, rational intuition is reliable.

3.4. The Empirical-Case Argument

might be required in order to entertain a testimonial content, one’s justification for

4.2. Moderate Intuitionism: Motivations

4.3. Moderate Intuitionism: In Search of

4.4. Metasemantics and Constitutivity

4 Or event, or what have you.

4.5. Externalist Metasemantics and Intuition

Beginning with Saul Kripke (1972), who formulated it.

10 Cf. Stich 1996, pp. 41–2.

2.2. What Is an Intuition? Seemings and

2.7. What Distinguishes Intuitions with Probative

2.10. Just How Can Understanding Function as a

4.3. Moderate Intuitionism: In Search of

4.8. The Dispositional Theory and Moderate

6.4. Norton on Thought Experiments as

8.2. Intuitions as Evidence Favoring One Coherent

8.6. Paradigm Moral Intuitions Are Not Beliefs Nor