Documente Academic
Documente Profesional
Documente Cultură
Max Deutsch
A Bradford Book
The MIT Press
Cambridge, Massachusetts
London, England
© 2015 Massachusetts Institute of Technology
All rights reserved. No part of this book may be reproduced in any form by any
electronic or mechanical means (including photocopying, recording, or information
storage and retrieval) without permission in writing from the publisher.
MIT Press books may be purchased at special quantity discounts for business or sales
promotional use. For information, please email special_sales@mitpress.mit.edu.
This book was set in Stone by the MIT Press. Printed and bound in the United States
of America.
Library of Congress Cataloging-in-Publication Data
Deutsch, Max, 1971–
The myth of the intuitive : experimental philosophy and philosophical method /
Max Deutsch.
pages cm
Includes bibliographical references and index.
ISBN 978-0-262-02895-0 (hardcover : alk. paper)
1. Methodology. 2. Philosophy—Research. 3. Intuition. I. Title.
B53.D484 2015
121’.3—dc23
2014034368
10 9 8 7 6 5 4 3 2 1
Contents
Acknowledgments vii
Introduction ix
Notes 163
References 183
Index 189
Acknowledgments
This book has been a long time coming, and my greatest debt is to those
who have given me the time I needed to finish it. First and foremost, I
thank my wife, Ann Baldoni, and my daughters, Rita Louise and Sadie Fran-
ces. To the rest of my supportive family, Hong Kong University and my
colleagues there, and my editor at MIT Press, Phil Laughlin: Thanks for
keeping the faith.
I first started thinking about experimental philosophy and the role of
intuitions in philosophical method in conversations (and heated argu-
ments) with the world’s best experimental philosophers. I thank Jonathan
Weinberg, Stephen Stich, and especially Ron Mallon for this early inspira-
tion. Later, Edouard Machery was kind enough to argue with me over issues
in the experimental philosophy of language. Much later, Herman Cappelen
visited Hong Kong and galvanized me with his own critical take on experi-
mental philosophy. Still later, my graduate student Phoebe Chan helped
me see things through fresh eyes. When the book was nearing completion,
I benefited from long discussions of my views with Dan Marshall, Jenny
Nado, Don Tontiplafol, and Lam Ka Ho.
Simple encouragement was needed just as often as philosophical input,
and I am grateful to have gotten plenty from many friends and family.
For this, thanks especially to Linda Jangaard, Jill Pipher, Stan Jonasson, Jeff
Hoffstein, Jim Baldoni, Nancy Baldoni, Emily Baldoni, John Madsen, Simon
Greenoff (the Lamma Socrates), Tara Goodwin, Ester Wensing, Kieran Col-
vert, Carolyn Primrose, Neil Ballantyne, John Weaver, Susannah Howard,
Liz Gower, Hans DeVries, Brad Christensen, Dave Hungerford, James Chow,
Ed Young, and Rajeev Balasubramanyam.
Many philosophers influenced my thinking about the ideas in this
book. Special thanks go to my favorite philosopher, Harry Deutsch, who
viii Acknowledgments
supported me every step of the way and always thought of the best objec-
tions. Thanks also to David Sosa, Patrick Hawley, Joe Lau, Jiwei Ci, Timothy
O’Leary, Chad Hansen, Alexandra Cook, Johanna Wolff, David McCarthy,
Jonathan Ichikawa, Timothy Williamson, Colin McGinn, Joel Pust, Kelly
Trogdon, Dan Robins, Michael Johnson, Paisley Livingston, Jay Newhard,
Michael Veber, John Collins, Michael Devitt, Eric Schwitzgebel, Hilary
Kornblith, and Wong Pak Hang.
Two anonymous reviewers for MIT Press read the manuscript and offered
detailed criticism. I hereby thank them for the opportunity to make the
book better. Thanks also to Judith Feldmann at the MIT Press for meticu-
lous copyediting.
I dedicate the book to the memory of my grandfathers: Robert Deutsch,
poet and scholar, and Olaf Jangaard, fisherman.
Introduction
have borne if Black had not been ready to take steps to ensure that he do it. (Frank-
furt 1969, 835–836)
The idea that xphiles have about this kind of maneuver in analytic philoso-
phy is this: Frankfurt is treating the intuition that Jones is morally respon-
sible as the evidence, or as the source of evidence, for the claim that he,
Jones, is morally responsible. And he, Frankfurt, is then using that claim—
that Jones is morally responsible—to object to the view that one is morally
responsible only if one could have done otherwise, since, as it seems in
Frankfurt’s story, Jones could not have done otherwise.
Assume for the moment that this is indeed what Frankfurt is up to in con-
structing the thought experiment involving Black and Jones. Now, one may
ask, should Frankfurt proceed in the way we are now supposing he does? In
particular, should he treat the intuition that Jones is morally responsible as
evidence for the claim that Jones is morally responsible? The answer seems
to depend on two issues: First, it depends on whether it really is intuitive
that Jones is morally responsible. Second, it depends on whether, if it is
intuitive, its intuitiveness counts as good evidence for the claim.
And here is where xphi comes in; for, as xphiles have rightly pointed
out, both of these are empirical issues, testable by methods familiar from
the social sciences. To test the first, we could devise a survey that presented
subjects with Frankfurt’s thought experiment and asked them to indicate
whether they intuit that Jones is morally responsible. To test the second,
we could use the first survey but divide our subject pool along different
demographic lines. Depending on the demographic, and depending on the
results we get, we might then be in a position to say that intuitions about
the thought experiment should not be treated as evidence.
Imagine, for example, that we discover in this fashion that whether one
intuits that Jones is morally responsible depends on one’s socioeconomic
status. That, it seems, would be cause for skepticism about the evidential
worth of intuitions about the case. For yours or my socioeconomic status
seems completely irrelevant to whether Jones is morally responsible. If our
intuitions about whether he is are sensitive to this irrelevant factor, then
that is grounds for holding that our intuitions about the case are not good
evidence for what is true in the case.
Designing and running experiments such as these is exactly what xphiles
have been doing for the past fourteen years or so. There are now hundreds
of studies testing people’s intuitions about a wide range of philosophical
Introduction xi
Suppose that Smith and Jones have applied for a certain job. And suppose that Smith
has strong evidence for the following conjunctive proposition:
d. Jones is the man who will get the job, and Jones has ten coins in his pocket.
Smith’s evidence for (d) might be that the president of the company assured him
that Jones would in the end be selected, and that he, Smith, had counted the coins
in Jones’s pocket ten minutes ago. Proposition (d) entails:
e. The man who will get the job has ten coins in his pocket.
Let us suppose that Smith sees the entailment from (d) to (e), and accepts (e) on the
grounds of (d), for which he has strong evidence. In this case, Smith is clearly justi-
fied in believing that (e) is true.
But imagine, further, that unknown to Smith, he himself, not Jones, will get the
job. And, also, unknown to Smith, he himself has ten coins in his pocket. Proposi-
tion (e) is then true, though proposition (d), from which Smith inferred (e), is false.
In our example, then, all of the following are true: (i) (e) is true, (ii) Smith believes
that (e) is true, and (iii) Smith is justified in believing that (e) is true. But it is equally
xii Introduction
clear that Smith does not know that (e) is true; for (e) is true in virtue of the num-
ber of coins in Smith’s pocket, while Smith does not know how many coins are in
Smith’s pocket, and bases his belief in (e) on a count of the coins in Jones’s pocket,
whom he falsely believes to be the man who will get the job. (Gettier 1963, 122)
times and have found literally nothing that would indicate that intuiting
that Smith does not know matters in any way to his argument against the
JTB theory. There is certainly nothing there to suggest that Gettier him-
self thinks so. I think—and I think that metaphilosophical attitudes about
Gettier’s argument against the JTB theory illustrate this perfectly—that the
idea that philosophy relies on intuitions as evidence is a myth, an endur-
ing and fairly widely held, yet entirely false, belief about the methods of
philosophy.
In one way or another I will be elaborating on this idea throughout the
book. It is its central idea, and the idea from whence its title derives. But put
it aside for the moment and pretend that these strange things people say
about the role of intuitions in Gettier’s argument are true. If the intuitive-
ness of the judgment that Smith does not know plays a significant eviden-
tial role in the argument against the JTB theory, then of course it makes a
good deal of sense to find out whether it really is intuitive, to whom, how
the intuitions pattern, how strong they are, and so on.
This is one thing that the original xphi paper (Weinberg, Nichols, and
Stich 2001) made a start on. Well, sort of. They did not test intuitions about
the 10 Coins Case, specifically; they tested a different variation. And they
cast their net fairly widely, testing a whole host of epistemic intuitions, not
just intuitions about Gettier cases. Later in the book, I will return, several
times, to Weinberg et al.’s groundbreaking paper. For now, I want to focus
on just one of its elements, the result they got concerning intuitions about
Gettier cases.
The result they got is that not everyone shares the Gettier intuition.
Whether one intuits that an agent in a Gettier case knows appears to
depend on one’s cultural background. They put their variation to (English-
speaking) Western, East Asian, and South Asian subject groups and found
that, while a clear majority of Western subjects intuited otherwise, clear
majorities of subjects in their Asian pools intuited that the agent in their
variation really does know. That is, majorities of their Asian subjects intu-
ited in a way consistent with the JTB theory, while majorities of their West-
ern subjects, as expected, intuited in a way inconsistent with it.
It thus appears from their results that intuitions about Gettier cases vary
along a dimension they should not if we are going to treat such intuitions
as evidence for the claims that are their contents. Earlier, I imagined col-
lecting data to the effect that socioeconomic status affects intuitions about
xiv Introduction
And this is just the tip of the iceberg. Frankfurt’s paper contains several dif-
ferent yet related arguments for the judgment that Jones is morally respon-
sible (and against the view that one is morally responsible for an act only if
one could have done otherwise).
What do these arguments have to do with who intuits what? In a
word: nothing. It is simply a mistake to think that Frankfurt’s argument
xvi Introduction
the book is to give negative xphi a fair hearing. What, exactly, is the form of
the argument against more traditional modes of philosophizing that nega-
tive xphiles make? Does this argument, which I will call the negative xphi
critique, really depend, as I have suggested, on false claims about the eviden-
tial role of intuitions? Are there important methodological lessons to draw
from negative xphi, even if, as I will argue, the negative xphi critique misses
its target by miles?
Importantly, xphiles are not wholly to blame for misattributing the
mythical method of appealing to intuitions to analytic philosophers.
Somehow, the idea that intuitions about thought experiments and cases
count, or are treated as counting, as an important source of evidence for
the truth about the cases is a popular metaphilosophical view, even among
non-xphile analytic philosophers, and even though the view is false, as
I will argue throughout the rest of the book. Still, xphiles have recently
played a large role in perpetuating the myth of the intuitive. My hope is
that this book will make them at least reconsider their reasons for doing
so. Xphiles who criticize analytic philosophy for its reliance on intuitions
about thought experiments are making a mistake. This book explains why.
I hasten to add that I think that xphi is one of the most interesting and
valuable developments in recent philosophy. This book is a polemic against
a particular view of the relation between xphi and analytic philosophy,
namely the negative xphi view that some of the xphi data on intuitions
present a serious challenge to the arguments and methods of analytic phi-
losophy. This view is held by only a subset of those who self-describe as
xphiles, and I am by no means out to condemn everything that is or might
be called xphi. In fact, I think collecting data on people’s philosophical
intuitions in the careful and scientifically informed way that many xphiles
do is an extremely worthwhile project. Cross-cultural xphi, perhaps espe-
cially concerning moral intuitions, has real potential for fostering cross-cul-
tural understanding and respect. Intracultural xphi has a similar potential
benefit, for it is useful in all sorts of ways to know when, where, and why
we, within a single culture, differ in our basic philosophical outlook.
We should welcome the new kid on the block. Perhaps most obviously
because xphi has breathed new life into methodological discussions of ana-
lytic philosophy, inspiring many philosophers to reflect carefully on their
practice and describe it as accurately as they can. Even those of us who,
for what we take to be good, considered, philosophical reasons, are critical
xviii Introduction
failure of the negative xphi critique, much of the work being done in xphi’s
name is valuable and interesting. Hopefully, the discussion in chapter 1
will bolster this point. Chapter 1 also contains my rendition of the negative
xphi critique, a section in which I discuss a phenomenon I call pragmatic
distortion, which I take to pose a significant difficulty for getting accurate
data about people’s intuitions via xphile intuition surveys, and my (very
minimalistic) account of the nature of intuitions.
In chapter 2, I point out that there is a true reading of “Many philosophi-
cal arguments treat intuitions as evidence.” This goes a long way, I argue,
toward explaining the appeal of the myth of the intuitive. However, I also
argue that this true reading does not support taking the myth to be true,
nor, consequently, does it in any way justify the negative xphile critique.
In addition, chapter 2 discusses the functioning of philosophical counter-
examples, taking Gettier’s anti-JTB theory counterexamples and Kripke’s
counterexamples to descriptivist theories of reference as case studies. The
chapter also includes criticism of a variety of more or less general method-
ological claims, made by both xphiles and non-xphile philosophers.
The short chapter 3 describes and criticizes some of Timothy William-
son’s metaphilosophical views. Williamson is known for charging xphiles
(among others) with views that commit them to a form of skepticism he
calls judgment skepticism. I argue that this charge does not stick.
In chapter 4, I show that Gettier’s refutation of the JTB theory appeals to
reason and argument at every stage; there is no appeal to intuitions, and,
in particular, no appeal to or reliance on intuitions in his arguments for
his claims about his thought experiments. I draw attention, in chapter 4,
to the fact that much of the “Gettierology” literature includes additional
arguments for taking Gettier’s claims about his thought experiments to be
correct. I also offer a reply to an objection to the effect that this badly
misunderstands these arguments, which are, according to the objection,
abductive arguments that take as premises Gettier’s claims about his thought
experiments.
In chapter 5, I continue the “arguments, not intuitions” theme of chap-
ter 4, showing that Kripke’s claim about his famous Gödel Case is supported
by a variety of arguments, none of which have anything to do with who
intuits what. There is also a brief discussion in chapter 5 of other “intuition-
free” arguments in philosophy. In chapter 5, I also offer a solution to a
problem, introduced briefly in chapter 3, that I call the relocation problem,
1 Varieties of Xphi, Pragmatic Distortion, and the
No-Theory Theory of Intuitions
1.1 Introduction
One goal of this chapter is to present several xphi studies and describe how
their results either have been or might be used to challenge traditional ana-
lytic philosophical arguments or methods.1 Understanding the challenges
involves examining a few of the extant xphi studies in at least some detail.
However, xphi comes in different varieties, and the challenges they pose
differ considerably. In this book, I will focus mainly on that variety now
known as negative xphi, and the results of most of the studies I will describe
later have been or could be used to argue for the sorts of conclusions that
are characteristic of xphi’s negative branch.
Negative xphi distinguishes itself from more positive varieties by taking
a more pessimistic stance toward traditional philosophical argument and
method. Negative xphi’s stance is not simply that traditional philosophy
makes empirically undersupported or just outright mistaken claims about
what intuitions people have. There are now many xphi studies that, using
survey methods, attempt to shed light on whether various groups of peo-
ple have this or that intuition. Much of the xphi relevant to the free will
debate is of this kind, for example; one of the burning issues in the area is
whether people are “natural compatibilists” who have intuitions that sug-
gest an implicit commitment to compatibilism between determinism and
free will and/or moral responsibility. But, while negative xphi involves col-
lecting data about intuitions and “intuitional diversity” (that is, diversity in
intuitions between different groups of people), this is really only a means
to an end, which is both to call certain traditional philosophical arguments
into question and, most importantly, to raise a worry about the epistemic
value of the philosophical intuitions negative xphiles take to be involved in
2 Chapter 1
My plan, in this section, is to describe some of the varieties of xphi that are
now being practiced. However, I am not very interested in the kinds of fine-
grained divisions in which xphiles themselves are (naturally) interested, so
I will not be providing an exhaustive or detailed catalog.4 Rather, my aim
Varieties of Xphi 3
is to give a sense of the ways in which different varieties of xphi raise dif-
ferent sorts of challenges to more traditional philosophical argumentation
and methods. In the next section, I will abstract away from the details and
present the most interesting of these challenges, namely the negative xphi
critique. This critique, in one way or another, will be the impetus and target
of the material in the chapters that follow.
One issue about categorization that I will mention here in order to put
it to one side is the extent to which some examples of xphi have nothing
much to do with intuitions or their role in traditional philosophical argu-
ment. The fact is that, in one way or another, most current xphi does touch
on intuitions and their role. Almost everything written under its banner
thus far contains reports of survey-style experimentation on people’s philo-
sophical intuitions. However, xphi is a burgeoning movement, and many
different sorts of projects are now billing themselves as xphi. Whether
“xphi” is defined narrowly as concerned with experiments on intuitions,
or more broadly to include other sorts of experiments, matters very little
to me. However, the examples of xphi that I will discuss, and the problems
I will raise for these examples, all have to do with the intuition-centered
variety of xphi.
My discussion of the varieties of (intuition-centered) xphi and the dif-
fering challenges they raise will be woven into presentations of the stud-
ies themselves. In every case, my descriptions of the studies will be fairly
brief and I will not linger long over the studies’ experimental design; nor
will I discuss in detail effect sizes or how the data reported in the studies
were gathered and analyzed. These are important issues, and some critics of
xphi think that its main weakness is methodological. The surveys’ designs
are problematic, these critics say, or confounds make the results difficult
to interpret, or this or that statistical test ought or ought not be used in
data analysis. These criticisms must be addressed by xphiles, but (with the
exception having to do with the semantics–pragmatics distinction that I
will discuss later in section 1.3) they are criticisms that have little to do with
the relation between xphi methods and traditional philosophical ones.
Xphi methods would face these criticisms regardless of how relevant xphi
is to traditional philosophical investigation. Since this book concerns the
relation between xphi and more traditional styles of philosophizing, these
methodological criticisms of xphi have little bearing on its main topic. I
will therefore mostly ignore them.5
4 Chapter 1
What unifies Gettier cases and Gettier intuitions into single categories?
When I teach undergraduate epistemology courses, I define Gettier cases
as those in which an agent justifiably and truly believes that p but fails to
know that p. To say that there are Gettier cases is, on this simple definition,
to imply that the content of the Gettier intuition is true. So long as the Get-
tier intuition is true, as I believe it is, there is nothing wrong with saying
that there are Gettier cases, understood as defined by the simple definition.
However, part of what is at issue in debates over the role of intuitions in
philosophical argument is whether intuitions, or the having of them, are
treated as evidence for the truth of their contents. Fruitfully discussing this
issue requires neutral descriptions of the “cases” the intuitions concern,
including Gettier cases.
One feature that every Gettier case appears to share is that the agent in
such a case justifiably believes p, and p is true, but it is, in some sense, a mat-
ter of luck that the agent truly believes that p. In Gettier’s 10 Coins Case, for
example, Smith gets it right and believes, truly and with justification, that
the man who will get the job has ten coins in his pocket, but it is clear that
his getting it right is due to luck. By sheer coincidence, Smith himself hap-
pens to have ten coins in his pocket (and, though presumably not by sheer
coincidence, will himself get the job). Later, when it comes time to respond
to xphi challenges, I will return to this structural feature of Gettier cases,
for the reply that I will make to those challenges, as they apply to Gettier’s
argument against the JTB theory of knowledge, exploits the fact that Gettier
cases, along with many other cases and thought experiments in philoso-
phy, often have these unifying structural features. For now, we can char-
acterize Gettier cases relatively neutrally as those in which an agent has a
justified true belief that p, but the belief is only luckily true. I will say more in
subsequent chapters about epistemic luck and its relevance to understand-
ing the functioning of Gettier’s cases, and, beyond this, to understanding
the functioning of cases and counterexamples in philosophical argument
more generally.
I have said that xphiles appear to have discovered that not everyone
shares the Gettier intuition. But the xphiles in question did more than
just this. There have always been those Frank Jackson (2011) describes as
“Gettier holdouts”—people who have professed to lack the intuition many
philosophers report having when they consider the cases. The interesting
twist that xphiles have added is that whether one has the Gettier intuition
6 Chapter 1
Bob has a friend, Jill, who has driven a Buick for many years. Bob therefore thinks
that Jill drives an American car. He is not aware, however, that her Buick has recently
been stolen, and he is also not aware that Jill has replaced it with a Pontiac, which is
a different kind of American car. (Weinberg et al. 2001, 444)
Subjects were then asked to say whether they thought Bob really knows that
Jill drives an American car or only believes that she does. This was a “forced
choice” question and the only answer options were these two. Table 1.1
displays the results reported in Weinberg et al. 2001.
How do these results bear on Gettier’s argument against the JTB theory
of knowledge? Well, in arguing against the JTB theory, Gettier appeals to
cases, such as the 10 Coins Case, that are similar to the case Weinberg et
al. tested. One way to pose the challenge raised by the results is to say that
whether a person will intuit that an agent in a Gettier case knows the rel-
evant proposition appears to depend on a factor that is irrelevant to the truth
of that intuition. This should lead us, or so one might think, to question
whether we are properly justified in intuiting, and then believing, what
we do about the cases. The intuitions seem sensitive to something—in this
case, culture—that has nothing to do with the truth of those intuitions. It
seems implausible in the extreme that whether Bob, from the Weinberg
et al. 2001 Gettier case, knows that Jill owns an American car, or Smith,
from Gettier’s 10 Coins Case, knows that the man who will get the job
has ten coins in his pocket, are matters that are culturally relative. Given
that there is a single, nonrelative answer to whether the agents in the cases
know, but given also that what one intuits the answer to be depends on a
Table 1.1
Peter is in his locked apartment reading, and is about to have a shower. He puts his
book down on the coffee table, and takes off his black plastic watch and leaves it on
the coffee table. Then he goes into the bathroom. As Peter’s shower begins, a bur-
glar silently breaks into the apartment. The burglar takes Peter’s black plastic watch,
replaces it with an identical black plastic watch, and then leaves. Peter is still in the
shower, and did not hear anything. (Starmans and Friedman 2012, 274)
Subjects were then asked several questions about the vignette, one of which
asked subjects to fill in the blank in the following “Gettier question”:
Peter ________ (Really knows/Only thinks) that there is a watch on the table.
The vice president of a company went to the chairman of the board and said, “We
are thinking of starting a new program. It will help us increase profits, but it will also
harm [help] the environment.” The chairman of the board answered, “I don’t care at
all about harming [helping] the environment. I just want to make as much profit as
I can. Let’s start the new program.” They started the new program. Sure enough, the
environment was harmed [helped]. (Knobe 2003, 191)
against the JTB theory stemming from Weinberg et al. 2001. There, the
upshot was that intuitions about Gettier cases ought not be trusted. The
interpretation of Knobe’s results we are now considering treats intuitions
about his scenarios differently. For to take Knobe’s results to reveal some-
thing important about intentional action itself is to take the intuitions
Knobe uncovered at face value, as revealing what is true in those scenarios.
It is not to impugn a method that treats intuitions as evidence; it is rather to
highlight the fact that the application of this method might be thoroughly
empirical.
Interestingly, there is a way of interpreting Knobe’s study that takes it
to be a relatively clear example of negative xphi. As I said, many philoso-
phers believe that evaluative facts about outcomes cannot have any bearing
on whether the action that produced that outcome was performed inten-
tionally. Suppose that is right. Then, what we have, given Knobe’s results,
is precisely variation in intuitions along a dimension—in this case, facts
about the evaluative properties of outcomes—that does not matter to the
truth of the intuitions. So, despite Knobe’s own intentions, his results could
quite easily be adopted for use in a negative xphile argument, one whose
conclusion is that some intuitions about intentional action, such as those
elicited by considering the Harm and Help scenarios, cannot be relied on as
evidence in theorizing about intentional action.
Gödel, the man who stole the proof. Such speakers misattribute the proof’s
discovery to him.
Or so Kripke says. In their 2004 paper, “Semantics, Cross-Cultural Style,”
Edouard Machery, Ron Mallon, Shaun Nichols, and Stephen Stich argue
that the alleged antidescriptivist upshot of Kripke’s Gödel Case is chal-
lenged by cross-cultural experiments they performed on people’s intuitions
about the reference of proper names. In one of their experiments, Machery
et al. presented subjects with the following vignette, closely modeled on
Kripke’s original Gödel Case:
Machery et al. 2004 Gödel Case
Suppose that John has learned in college that Gödel is the man who proved an
important mathematical theorem, called the incompleteness of arithmetic. John is
quite good at mathematics and he can give an accurate statement of the incomplete-
ness theorem, which he attributes to Gödel as the discoverer. But this is the only
thing that he has heard about Gödel. Now suppose that Gödel was not the author of
this theorem. A man called “Schmidt” whose body was found in Vienna under mys-
terious circumstances many years ago, actually did the work in question. His friend
Gödel somehow got hold of the manuscript and claimed credit for the work, which
was thereafter attributed to Gödel. Thus he has been known as the man who proved
the incompleteness of arithmetic. Most people who have heard the name “Gödel”
are like John; the claim that Gödel discovered the incompleteness theorem is the
only thing they have ever heard about Gödel. When John uses the name “Gödel,”
is he talking about:
or
(B) the person who got hold of the manuscript and claimed credit for the work?
(Machery et al. 2004, B6)
What they found is that Westerners were far more likely than East Asians to
give the “Kripkean” answer, namely (B). Figure 1.1 gives the percentages of
(B) answers given by the different subject groups.9
In certain places, Machery et al. treat their results as a challenge to
Kripke’s antidescriptivist arguments simply because the results suggest that
descriptivism is not as counterintuitive as was once believed. Since plenty
of East Asian English-speaking subjects appear to have intuitions consis-
tent with descriptivism, it cannot be claimed that antidescriptivism is the
intuitive view. This, it seems to me, is to take the results to have primarily
positive xphi implications; the goal is simply to show that antidescriptivist
intuitions are not universally shared.
Varieties of Xphi 13
100
90
Pr opor t ion of Kr ipk e a n An sw e r s
80
70
60 58
55
Westerners
50
Easterners
40
32
29
30
20
10
0
Gödel Case Tsu Ch'ung Chih Case
Figure 1.1
From Edouard Machery, “Expertise and Intuitions about Reference,” Theoria 72
(2011): 37–54. Used with permission.
However, in other places, Machery et al. make it clear that they intend
their work to be an example of negative xphi. And negative xphiles, remem-
ber, are not content simply to show, via empirical study, that this or that
judgment is or is not the intuitive one. Instead, negative xphiles aim to
show that certain intuitions vary in ways they should not, if they are to be
regarded as evidence for what they are intuitions about. Since intuitions
about reference vary with respect to a factor that should not matter to their
truth—namely culture—these intuitions cannot be trusted. This is the cen-
tral negative xphi message of Machery et al. 2004.
has, because of the implanted computer, highly reliable and accurate beliefs
about the temperature. But, Lehrer says, these beliefs do not amount to
knowledge on Mr. Truetemp’s part of what temperature it is. So reliabilism
is false.
In 2008, a group of experimental philosophers, Stacey Swain, Joshua
Alexander, and Jonathan Weinberg, put intuitions about Truetemp-style
cases to empirical test. Their interesting hypothesis was that intuitions
about such cases might vary with respect to the order in which other cases
were presented to subjects. In particular, they hypothesized that subjects
who encountered a clear case of non-knowledge before being presented with
a Truetemp case would be more likely to intuit that the agent in the True-
temp case knows than they would if presented with the Truetemp case first.
Likewise, they hypothesized that, if presented with a clear case of knowl-
edge first, subjects would then be less likely to intuit that the agent in the
Truetemp case knows.
Here are three cases used by Swain et al. 2008 to test these hypotheses.
The first is a Truetemp case closely modeled on Lehrer’s (1990) original; the
latter two are cases they used to test for the “order effect,” the first their
clear case of non-knowledge, the second their clear case of knowledge.
Swain, Alexander, and Weinberg 2008 Truetemp Case
One day Charles was knocked out by a falling rock; as a result his brain was “re-
wired” so that he is always right whenever he estimates the temperature where he
is. Charles is unaware that his brain has been altered in this way. A few weeks later,
this brain rewiring leads him to believe that it is 71 degrees in his room. Apart from
his estimation, he has no other reasons to think that it is 71 degrees. In fact, it is 71
degrees. (Swain et al., 2008, 154)
Dave likes to play a game with flipping a coin. He sometimes gets a “special feeling”
that the next flip will come out heads. When he gets this “special feeling,” he is right
about half the time, and wrong about half the time. Just before the next flip, Dave
gets that “special feeling,” and the feeling leads him to believe that the coin will land
heads. He flips the coin, and it does land heads. (154)
the article is correct: mixing the two products does create a poisonous gas. At noon,
Karen sees a janitor mixing Cleano Plus and Washaway and yells to him, “Get away!
Mixing those two products creates a poisonous gas!” (155)
Swain et al.’s hypotheses were borne out: their subjects’ intuitions about
whether Charles, in their Truetemp Case, knows that it is 71 degrees were
strongly influenced by which cases they encountered first. If they encoun-
tered Dave in the Clear Non-knowledge Case first, subjects were far more
likely to attribute knowledge to Charles in the Truetemp Case than they were
if they were instead first exposed to Karen in the Clear Knowledge Case.
Swain et al. draw a clear negative xphi conclusion: Intuitions about
Truetemp-style cases cannot be trusted or relied on as evidence, for they
are sensitive to a factor that is obviously irrelevant to their truth, namely
the order in which one encounters such cases when one is considering a
range of more or less similar cases. To the extent that Lehrer’s antireliabilist
argument depends on intuitions about his Truetemp case, the argument is
undermined, Swain et al. maintain.
what happened before it. So, if everything in this universe was exactly the same up
until John made his decision, then it had to happen that John would decide to have
French Fries. (Nichols and Knobe 2007, 669)
Then, one group of subjects was given the following, highly affect-laden
vignette:
Nichols and Knobe 2007 Determinism and Responsibility Case
In Universe A, a man named Bill has become attracted to his secretary, and he de-
cides that the only way to be with her is to kill his wife and 3 children. He knows
that it is impossible to escape from his house in the event of a fire. Before he leaves
on a business trip, he sets up a device in his basement that burns down the house
and kills his family. (670)
Subjects were then asked to answer the following question: Is Bill fully mor-
ally responsible for killing his wife and children? No less than 72 percent of
subjects who read the story about Bill intuited that, yes, Bill is fully morally
responsible for killing his wife and children.
However, another group of subjects, who were also given the descrip-
tion of Universe A, did not receive the vignette about Bill, but were instead
simply asked the following, not-affect-laden question: “In universe A, is
it possible for a person to be fully morally responsible for their actions?”
(670). And here, the results were dramatically different; a full 86 percent of
the subjects intuited that, no, it is not possible for a person in universe A to
be fully morally responsible for his or her actions.
Nichols and Knobe treat their experiment as an example of positive xphi.
Their goal is to figure out what intuitions people have about determinism
and moral responsibility, and why. As to why, one possibility they float is
that high affect distorts one’s ability to competently assess responsibility;
the negative emotions triggered by the story of Bill interfere with subjects’
ability to accurately ascribe responsibility.
However, as Nichols and Knobe themselves come close to recognizing
(see their discussion of the “philosophical implications” of their experi-
ment at 667–668), this explanation also points toward a negative xphile
conclusion, namely that intuitions about determinism and responsibility
should not be treated as evidence for the truth of compatibilism (or incom-
patibilism, for that matter) because these intuitions are sensitive to a factor
that is irrelevant to their truth. In this case, the truth-irrelevant factor is
the relative level of affect-inducing language used in framing the vignettes.
Varieties of Xphi 17
I assume that most negative xphiles will say that several traditional argu-
ments serve as instances of “argument A,” Gettier’s anti-JTB argument,
Kripke’s antidescriptivist argument, and Lehrer’s antireliabilism argument
included. However, a question remains about how far the critique extends;
that is, there is a question about how much of that part of traditional
philosophy that is alleged to rely on intuitions about cases is called into
question by the discoveries xphiles have been making. Should we say, for
example, that the discovered variability poses a threat only to those spe-
cific arguments that are alleged to appeal to the specific intuitions that
have been tested (and about which truth-irrelevant variability has been dis-
covered)? Or are we, right now, in a position to say that there is a threat
to an entire philosophical practice (the supposed practice of appealing to
intuitions about cases) or that a whole source of essential philosophical evi-
dence (the deliverances of intuition) has been revealed not to be a genuine
source of evidence?
Negative xphiles tend to say that the critique has a broader target than
just the arguments (allegedly) based on the specific intuitions that have
been tested, and for which the relevant sort of variability has been discov-
ered. Swain et al. are admirably clear on this issue:
[We] found that intuitions about the Truetemp Case vary depending on whether,
and which, other cases are presented before it. Such variability calls into question
the legitimacy of using the intuitions generated by the Truetemp Case as evidence
against reliabilism. But it is unclear what about this case makes it susceptible to
these effects, which raises questions about the reliance on intuitions about thought-
experiments more generally, especially given that this is not the only case called into
question by empirical research. We take the growing body of empirical data impugn-
20 Chapter 1
ing various intuitions to present a real challenge for philosophers who wish to rely
on intuitions as evidence. (Swain et al. 2008, 153)14
The issue of just how much skepticism about the use of intuitions as evi-
dence in philosophy is warranted, given the negative xphile results, is an
interesting one, but, for my purposes, the critique need have no broader a tar-
get than just the arguments that are said to appeal to intuitions that negative
xphiles have actually tested. For I will argue that the negative xphile critique
fails to hit even the relatively narrow targets it needs to hit if it is to do any-
thing like call entire practices, methods, or sources of evidence into question.
The reason it fails to hit even the narrow targets is that, as a matter
of fact, the traditional philosophical arguments toward which negative
xphiles direct their critique do not, in any relevant sense, depend on treating
intuitions as evidence. So, obviously, even if the critique shows that there
is some sense in which some intuitions cannot be justifiably regarded as
evidence, that will not matter to whether the arguments are good, cogent,
compelling arguments for their conclusions.
For example, Lehrer’s (1990) argument against reliabilism, the focus of
Swain et al. 2008, does not treat intuitions about the Truetemp Case as evi-
dence—not, at any rate, in any sense would make Swain et al.’s apparent
discovery of truth-irrelevant variability with respect to such intuitions a
threat or challenge to the argument. Of course, Lehrer’s argument depends
on it being true that agents in Truetemp cases do not know the relevant
propositions. But that is a different matter. What is true in the cases is one
thing; whether people intuit what is true in the cases is another, completely
different thing. Furthermore, if one actually bothers to read Lehrer’s text,
one will see that he offers not just an argument against reliabilism but also a
(subsidiary) argument for the view that Mr. Truetemp does not know. There
simply is no appeal to intuitions as evidence in Lehrer’s presentation of the
original Truetemp Case.15
This is to jump ahead a little, since my intention in this section is mainly
to lay out the negative xphile critique in a perspicuous way. My full reply to
the critique will be developed as the book unfolds. For now, suffice it to say
that even if negative xphiles are right that some philosophical intuitions vary
along truth-irrelevant dimensions, this will matter to traditional philosophi-
cal practice only if that practice involves treating intuitions as evidence. But
negative xphiles are wrong—badly wrong—about the practice; it does not
treat intuitions as evidence in any sense that will sustain their critique.
Varieties of Xphi 21
In fact, there is some reason to be skeptical even of the claim that nega-
tive xphiles have discovered any instances of truth-irrelevant variability in
intuitions in the first place. The reason is that, in a fair number of cases,
the questions they ask their subjects in their surveys, or the answer choices
they provide, are affected by a phenomenon I will call pragmatic distor-
tion. And where there is pragmatic distortion, drawing conclusions about
what intuitions subjects have in response to xphi survey questions is not
at all straightforward. This is a problem; for, as we have seen, negative xphi
involves attributing varying philosophical intuitions to various subject
groups. If, because of pragmatic distortion, negative xphiles cannot univo-
cally interpret their subjects’ responses, then their claim to have discovered
truth-irrelevant variability in those responses is undermined.
In the philosophy of language, it is standard to draw a distinction
between the semantic meaning of a sentence, S, and those meanings that
are not semantic (i.e., not encoded by the meanings of the words of S along
with the way these word-level meanings are “put together”) but are, or can
be, communicated by utterances (or inscriptions) of S. These other contents
or meanings are pragmatic meanings of utterances of S. Whether S is true
depends on whether its semantic meaning is true. The truth or falsity of a
merely pragmatic meaning does not have implications with respect to the
truth or falsity of the sentence, S, itself. A common example is irony. If I
utter the sentence “Frank is a genius” in a tone dripping with irony, I might
communicate to an audience that Frank is not a genius, but the sentence I
utter is not true if Frank is not a genius; it is true if (and only if) Frank is a
genius.
However, it is not always obvious when a meaning is merely a pragmatic
meaning and when it is instead semantic. Hence, sometimes people are
misled; they mistake the truth or falsity of a pragmatic meaning for the
truth or falsity of a sentence’s semantic meaning. And this has implica-
tions for what conclusions we can draw from the results of xphi intuition
surveys. If the questions asked in the surveys, or the answer choices offered,
give rise to pragmatic meanings, then, before we can conclude that there is
any genuine variability in intuitions, we must first rule out the possibility
that different subjects are reacting to different meanings, some to prag-
matic meanings, and some to semantic ones.
22 Chapter 1
was. In the Gettier cases, it was, in the Weinberg et al. 2001 study, the judg-
ment that Bob really knows that Jill owns a Ford, and, in the Starmans and
Friedman 2012 study, it was the judgment that Peter really knows that there is
a watch on the table. In the Machery et al. 2004 study on Gödel cases, it was
the judgment that John uses “Gödel” to talk about the man who stole the proof.
In the Knobe 2003 study, it was the judgment that the chairman intentionally
helped/harmed the environment. And so on. And those are just a few of the
intuitive judgments that xphiles have tested.
Of course, many judgments of a similar kind are identifiable in tradi-
tional philosophical argument as well. Indeed, the examples that xphiles
have tested are, in most cases, drawn more or less directly from more tradi-
tional philosophical texts. The judgment that two agents might be microphys-
ically identical but instantiate different mental states; the judgment that there
are agents who are not free to do otherwise than perform a certain act, and yet are
morally responsible for that act; the judgment that it is not morally permissible
to push a fat man off a bridge to his death even if doing so will stop a speed-
ing trolley and save the lives of five people tied to its track—these are further
examples of judgments that philosophers working in various areas have
labeled as intuitions.22 Demonstrating these judgments is possible, and
knowing what is thereby demonstrated is sufficient for understanding and
assessing the claim that these judgments are used as evidence in traditional
philosophical argument; a theory of the nature of intuitions is simply not
needed. The term “these judgments” in the previous sentence has a rea-
sonably clear reference. Call those judgments “intuitions.” “Are intuitions
used as evidence?” or “What is the role of intuitions in philosophical argu-
ment?” now have reasonably clear senses and may be fruitfully discussed.
No theory of intuition-hood is required.
I do not mean to imply that there is nothing that unifies these examples
of intuitive judgments. They are all judgments about hypothetical cases or
thought experiments, for example. And they are all judgments to the effect
that something described in the case or thought experiment has or lacks
some philosophically significant property. That all of the example judg-
ments are of this kind provides a way of lengthening the short list I gave
above. If a judgment concerns a hypothetical case or thought experiment,
and concerns, more specifically, whether something described in the case or
thought experiment has or lacks some philosophically significant property,
then it is likely an intuition in the sense relevant to the debate over the role
26 Chapter 1
philosophy are dubious about the prospects for precisely this sort of “arm-
chair” conceptual analysis. One would hope that their criticism could be
made and understood without first engaging in a lot painstaking concep-
tual analysis concerning the nature of intuitions. As I have already said,
I think it can, and that is a good thing, since, as I have also already said,
conceptually analyzing intuitions is difficult.
Even simple, allegedly necessary properties, such as being spontaneous
or noninferential, are problematic. For example, it is hard to believe that
the issues over the argumentative role of intuitions would be settled in
one fell swoop if it turned out that virtually no philosophical judgments
are made spontaneously. Being noninferential seems an essential element
of nearly everyone’s account, but problems arise here too. John Rawls’s
method in political philosophy is supposed to depend on intuitions, for
example, but he explicitly says that the relevant judgments are our consid-
ered moral judgments. In addition, as Williamson and others have pointed
out, many ordinary perceptual judgments are noninferential, at least in the
sense of not being the products of conscious, explicit inference.
Or consider the disagreement between George Bealer, on the one hand,
and Timothy Williamson, Peter Van Inwagen and David Lewis, on the
other. Bealer (1996) proposes that an intuition is an “intellectual seeming,”
which he takes to be a sui generis mental state, not analyzable in more famil-
iar mental state terms, such as belief or judgment. Bealer likens intuitions
to “perceptual seemings,” in which it perceptually appears to one that p.
Such perceptual seemings are not amenable to analysis in terms of belief or
judgment, or dispositions to believe or judge. For one thing, such analyses
get their phenomenology wrong. There is something it is like to experience
the bent stick illusion, in which a straight stick submerged in water visu-
ally appears to be bent, something that cannot be fully captured in terms
of the perceiver’s beliefs or dispositions to believe. Likewise, according to
Bealer, we experience purely intellectual seemings—intuitions—in which
it appears to us that p, but not under any perceptual mode. But these, too,
cannot be analyzed in terms of beliefs or judgments. In addition, Bealer
thinks that philosophical intuitions present their contents as necessarily true.
Williamson, on the other hand, taking his cue from Lewis (1983) and
Van Inwagen (1997), who, in the former case, identifies intuitions with
beliefs, and, in the latter case, with beliefs or inclinations to believe, denies
that that he undergoes any Bealerian intellectual seemings and claims that,
28 Chapter 1
role of intuitions. It offers not too much of an account because it does not
invite the potentially endless cycle of counterexample-and-theory-revision
endemic to many attempts at conceptual analysis.
Since it has these virtues, I recommend it to all participants in the debate
over intuitions and their role. In particular, I think xphiles, especially nega-
tive xphiles, would be wise to adopt the no-theory theory. This is because
they have sometimes been accused, with some justification, of not speci-
fying a suitable or clear target for their critique. Sometimes, this accusa-
tion hinges on the fact that xphiles have adopted an analysis of intuitions
that is, according to the accusers, flawed. For example, many xphile papers
describe intuitive judgments as spontaneous or relatively spontaneous. If
one does not think that the speed with which one judges could possibly
matter very much to the sorts of judgments made in philosophy or to the
argumentative role these judgments play, then one will be inclined to shrug
off challenges in xphile papers in which this feature of intuitions is empha-
sized. That reaction is easily avoided by adopting the no-theory theory. The
same is true for disputes over whether intuitions are noninferential or non-
perceptual, involve intellectual seemings, present their contents as necessarily
true, are the result of conceptual competence, are fallible, or some combination
of these. Adopting the no-theory theory allows someone interested in the
role of intuitions in philosophy simply to sidestep these questions.
However, the no-theory theory does assume that we have examples of
intuitive judgments about which those party to the debate over their role
can agree; that is, these parties can agree that the examples are examples of
intuitive judgments. In a recent book arguing that philosophy does not rely
on intuitions, Herman Cappelen (2012) appears to deny this. He objects to
an attempt, by Anna-Sara Malmgren (2011), to identify intuitive judgments
by appeal to paradigmatic examples. In the relevant section of his book,
Cappelen discusses the issue of whether the ways in which philosophers
speak of intuitions amounts to speaking of some single, coherent, and uni-
fied kind, or whether, instead, philosophers’ use of “intuition” and cognate
terms is so all over the map that there is no one thing that philosophers
mean when engaging in “intuition-talk.” Cappelen makes a good case for
the latter; the way philosophers talk about intuitions is a terrific mess. And,
to some extent, Malmgren (2011) can be read as denying this.24 However,
on its own, the point about the disparateness, or perhaps even incoherence,
of “intuition talk” does not count against the possibility of identifying
Varieties of Xphi 31
examples of intuitive judgments. People can say all sorts of disparate and
even inconsistent things about what makes something an example of a
certain kind without that preventing the identification of things that are
supposed to be instances of the kind. And there are instances of judgments
that most philosophers will agree are supposed to count as intuitive. I listed
several such examples earlier, and Malmgren lists her own.
Cappelen (2012) raises two objections. First, he says that, in fact, there is
no agreement over the examples or paradigms. His main argument for this
is that, in recent methodological discussions of Gettier cases, disagreement
emerges over exactly what the intuitive judgment relative to the cases is
supposed to be. Since “the Gettier judgment” is supposed to be one of our
clearest paradigms, Cappelen concludes that there are no clear paradigms.
Second, Cappelen says that even if several paradigms can be adduced, with-
out some way to expand the list, the paradigms are useless.
The recent methodological discussions of Gettier cases involve an
attempt to generalize over specific judgments about specific Gettier cases.
Part of what is at issue in those discussions is whether there is some general
form of judgment such that each specific Gettier judgment counts as an
instance of the form.25 When I say Gettier judgments are examples of intui-
tive judgments, I mean the specific judgments about the specific cases. Pre-
sumably, this is also what Malmgren means (though she too tries to specify
the general form of Gettier judgments). There cannot be any serious dispute
over what, in Gettier’s original paper, the specific cases are, nor can there be
any serious dispute over which judgments philosophers think are the intui-
tive ones, relative to those cases. These specific judgments can be taken as
paradigms of the intuitive. So Cappellen’s first objection misses its mark.
Cappellen’s second objection is that identifying the paradigms is useless
if there is no way to expand the list, but that expanding the list cannot be
done except by appealing to the sorts of controversial features that show up
in conceptual analyses of intuition and over which philosophers disagree.
But useless for what? For Cappellen, it seems that identifying the paradigms
is useless if one wants to make sense of the ways in which philosophers talk
of intuitions. But, as I suggested above, identifying paradigmatic intuitive
judgments makes sense of intuition talk in the following minimal way: it
identifies the range of judgments about which philosophers make their dis-
parate and sometimes inconsistent claims. In any case, there does appear to
be a relatively uncontroversial commonality between these judgments, and
32 Chapter 1
so there is a way to “project out,” as Cappellen puts it, from the paradigms:
the examples are all judgments about hypothetical cases and thought
experiments. So Cappellen’s second objection also misses its mark.
The no-theory theory makes enough sense of intuition talk for the main
methodological question of this book—What is the argumentative role of
intuitions in philosophy?—to be asked, and its answer pursued. That is all
that I need or require. I am officially agnostic about whether being nonin-
ferential, for example, is necessary for being intuitive. For my purposes in
this book, questions such as this need not be settled. It is enough to have
examples of the sorts of judgments that many philosophers would agree
are intuitive, even though they might disagree over what it means, exactly,
to describe them as such. One caveat: at certain points in the book, it will
be necessary to consider whether certain assumptions about the nature of
intuitions are being made by others in order to argue for this or that conclu-
sion about their argumentative role in philosophy. For example, as I will
argue later, part of the force of the negative xphi critique derives from the
assumption that traditional philosophy treats intuitive judgments as nonin-
ferentially justified. When I use “intuitive” or “intuitive judgment” in a way
that goes beyond the sense given to these terms by the no-theory theory, I
will flag these uses and explain the rationale behind them.
2 Intuitions and Counterexamples
2.1 Introduction
In the previous chapter, I argued that the phenomenon of pragmatic dis-
tortion casts some doubt on whether xphiles have discovered genuine
instances of truth-irrelevant variability in philosophical intuitions. How-
ever, even if empirical research has, so far, failed to properly confirm it, it
is, I think, a fair bet that better methods and more research will, at the end
of the day, reveal that intuitions about a variety of philosophical cases and
thought experiments really do differ along truth-irrelevant demographic
lines. Although I am skeptical of its philosophical significance, even I am
prepared to admit that we have, right now, fairly conclusive evidence that
people’s intuitions about the reference of proper names are culturally vari-
able. And I think that the evidence for saying that there is such variability
in intuitions about Gettier cases is strong, though not quite conclusive. But
I am also willing to bet that the evidence will grow stronger in the near
future.
The interesting question to ask is what will or should follow for philoso-
phy and its methods as the evidence for truth-irrelevant variability in intu-
itions mounts. As I described their position in chapter 1, negative xphiles
think that if there is such variability (and they think there is strong evidence
for such variability already) then that poses a serious challenge to some of
the methods and arguments of contemporary analytic philosophy. They
think that the variability results show that we should suspend judgment on
some of the claims made in contemporary analytic philosophy concerning
what is true in a variety of thought experiments and hypothetical cases.
This, you will recall, is the conclusion of what I described in the last chapter
as the negative xphi critique. I have said that I think the conclusion of the
negative xphi critique is mistaken; now it is time to begin arguing for this.
34 Chapter 2
However, in the only and quite specific sense that negative xphiles can
intend (EC), given what their variability results are results about, (EC) is
not true; that is, in that specific sense, it is not true that many philosophical
arguments depend on treating intuitions about thought experiments and
cases as evidence.
In my experience, saying this, even in the qualified and careful way I just
have, comes as a surprise, and not just to xphiles. The myth of the intuitive
is now an entrenched metaphilosophical view, so much so that “philoso-
phers treat intuitions as evidence” or “arguments in philosophy rely on
Intuitions and Counterexamples 35
Many philosophers, analytic and xphile alike, simply assume that (EC) is
univocally true. Some would go so far as to say that, ultimately, all philo-
sophical arguments depend on treating intuitions as evidence, and that
there is no troubling or suspicious ambiguity in this assumption. I am going
to argue that these ideas are mistaken, but there is a problem with deny-
ing (EC) outright. The problem is that there is a state/content ambiguity in
“intuition” that makes (EC) itself ambiguous. On one reading, the claim is
true. On the reading xphiles intend, however, it is false.
The negative xphi critique is unfairly bolstered by the fact that there is a
true reading of (EC). Since it has a true reading, it can be difficult to see how
one might object to the negative xphi critique by denying it. I think this
goes a long way toward explaining why the line against negative xphi that I
will develop here has been tried only rarely.1 For someone who has the true
reading in mind, the claim that “Many philosophical arguments depend on
treating intuitions as evidence”—(EC), that is—seems unassailable.
What are the true and false readings, and on what grounds do I base my
claim that the true reading is true and the false reading false? Let me work
up to the answer by elaborating on the view that there is a state/content
ambiguity in “intuition.”
Other related terms display a similar ambiguity. “Judgment,” for exam-
ple, may refer to the act or state of judging, or to the (propositional) content
36 Chapter 2
of such an act or state. “Belief” is ambiguous in this way too. It may refer
to a certain kind of psychological state, or to the content of a state of that
type. The state/content ambiguity of such terms affects the interpretation
of the phrases and sentences in which they appear. “The judgment that 2
+ 1 = 3” may be used to denote the act of judging that 2 +1 = 3, or the con-
tent of that act, namely that 2 + 1 = 3. “The belief that there is life on other
planets” picks out a “propositional attitude” of a specific type, but it can
also be used to speak directly about the content of the attitude, as in “The
belief that there is life on other planets is likely false,” where the intended
meaning is simply that it is likely false that there is life on other planets.
If “intuition” is ambiguous in this way, and can be used on some occa-
sions to refer to the act or state of intuiting and on others to the content
of such an act or state, then phrases and sentences involving the term will
inherit this ambiguity. “The intuition that p,” for example, may be used to
refer the act or state of intuiting that p, or simply to p. William Lycan (1988)
draws a distinction between intuitings, conceived as psychological events of
a certain type, namely those we describe as “intuiting that something is so,”
and intuiteds, conceived as the propositional contents of these psychologi-
cal events. “Intuition,” Lycan says, is ambiguous between an understanding
that takes it to refer to an intuiting and one that takes it to refer to an intu-
ited.2 I agree; there is a clear ambiguity here. Given its existence, we must
take care in evaluating claims about the role of intuitions in philosophy.
Doing so will require carefully disambiguating their state versus content
(intuiting versus intuited) interpretations.
Take the case of (EC): “Many philosophical arguments treat intuitions
as evidence.” One can understand this as asserting either (EC1) or (EC2):
(EC1) Many philosophical arguments treat the fact that certain contents
are intuitive as evidence for those very contents.
There are several reasons for taking Gettier and Kripke as case studies. First,
as we saw in chapter 1, both Gettier’s argument against the JTB theory of
knowledge and Kripke’s argument against the descriptivist theory of refer-
ence for proper names are targets of recent xphi critiques. Second, although
they concern issues in separate areas of philosophy (Gettier’s argument is
epistemological while Kripke’s is semantic), both arguments share a struc-
ture, a structure they in turn share with an enormous range of other argu-
ments in every branch of analytic philosophy. This shared argumentative
structure allows methodological conclusions about Gettier’s and Kripke’s
arguments to generalize to arguments in metaphysics, ethics, action theory,
philosophy of mind, and so on. Third, it is quite common to hear mislead-
ing claims about the role that intuitions play in Gettier’s and Kripke’s argu-
ments. If it can be shown how, in the cases of these arguments specifically,
the claims are misleading, that ought to go a long way toward convincing
40 Chapter 2
made by their inventors or discoverers. One might object that this focus is
too narrow; if we want an accurate picture of philosophical practice as a
whole, should not the scope of the investigation be broader? Perhaps it is
proper to include textbook or “pop philosophy” presentations of famous
philosophical thought experiments, or even the ways in which philoso-
phers render famous cases in informal conversations with other philoso-
phers at the dinner table or the bar.
This objection strikes me as wrongheaded. The important methodologi-
cal question is: What methods are employed by good philosophers, ones
who, by fairly wide consensus, have made interesting and important philo-
sophical progress, increasing, in a significant way, our body of philosophical
knowledge? Textbooks and works of pop philosophy are mostly rehashings
of much harder work already done by others. Philosophical conversations
at the bar often rely on thumbnail sketches and various other shortcuts that
do not reflect core methods of the discipline. The core methods of the dis-
cipline, and of any discipline, are reflected most clearly by the most clearly
successful examples of discovery and progress in the discipline. A focus on
such examples in philosophy is thus entirely appropriate.
In any case, surely the best place to look for insight into the function-
ing of specific thought experiments or counterexamples is to the original
presentations. Could there be a better source for insight into how Gettier
refuted the JTB theory of knowledge, or Kripke refuted the descriptivist the-
ory of reference for proper names, than Gettier’s and Kripke’s own work on
the subject? The idea that pop philosophy or textbook renditions deserve
equal consideration seems misguided; whether these sorts of renditions
accurately reveal the methods Gettier and Kripke employ depends on what
methods Gettier and Kripke employed, as revealed in their own work.9
Before moving on, it will be useful to remind ourselves of Gettier’s and
Kripke’s targets. What are they arguing against? Gettier targets the JTB
theory of knowledge, one consequence of which is the generalization that
every justified true believer is a knower. Though this is strictly a conse-
quence of the theory, not the theory itself, it will be convenient to label
it simply as the “JTB theory of knowledge.” Somewhat more formally, the
targeted generalization is the following:
For every subject, S, and every proposition, p, if S justifiably and truly be-
lieves that p, then S knows that p.
42 Chapter 2
For every proper name, N, and every object, x, if users of N associate definite
descriptions with N that are satisfied by x, then N refers to x.
The question now is: How do Gettier and Kripke argue against these gener-
alizations? An answer that no philosopher would dispute is that they do so
by presenting (alleged) counterexamples, in the form of hypothetical cases,
to the generalizations.10 In fact, Gettier and Kripke each present a variety of
counterexamples against his respective target. I will focus here on just one
counterexample from Kripke, his famous Gödel Case, which will be familiar
from chapter 1’s discussion of Machery et al. 2004, and one counterex-
ample from Gettier. Since the way in which philosophers actually argue, as
opposed to how various commentators take them to argue, matters to my
criticism of negative xphi, I will focus on one of Gettier’s original (1963)
cases, the well-known 10 Coins Case (presented in the introduction), not
the variation on Lehrer’s (1965) Nogot/Ford Case used in the study reported
in Weinberg et al. 2001.
Though it is uncontroversial that Gettier’s and Kripke’s arguments
involve counterexamples based on hypothetical cases, such as the Gödel
and 10 Coins cases, it strikes me as very poorly understood just what is and
is not involved in presenting this sort of counterexample to a philosophical
theory or generalization. It is common, even outside of experimental phi-
losophy circles, to misrepresent putative refutations-by-counterexample as
depending on the intuitiveness of the counterexample, and thus as depend-
ing on intuitions in the (EC1) “state” sense.
In the hope of correcting this common misrepresentation, I will first, in
section 2.3.1, quote Gettier and Kripke at length, presenting their counter-
examples in their own words. This will show that there is nothing in their
presentations to suggest that either Gettier or Kripke appeals to intuitions
in the “state” sense as crucial evidence for any of their arguments’ premises.
Then, in section 2.3.2, I will describe how commentators on Gettier and
Intuitions and Counterexamples 43
Suppose that Smith and Jones have applied for a certain job. And suppose that Smith
has strong evidence for the following conjunctive proposition:
d. Jones is the man who will get the job, and Jones has ten coins in his pocket.
Smith’s evidence for (d) might be that the president of the company assured him
that Jones would in the end be selected, and that he, Smith, had counted the coins
in Jones’s pocket ten minutes ago. Proposition (d) entails:
e. The man who will get the job has ten coins in his pocket.
Let us suppose that Smith sees the entailment from (d) to (e), and accepts (e) on the
grounds of (d), for which he has strong evidence. In this case, Smith is clearly justi-
fied in believing that (e) is true.
But imagine, further, that unknown to Smith, he himself, not Jones, will get the
job. And, also, unknown to Smith, he himself has ten coins in his pocket. Proposi-
tion (e) is then true, though proposition (d), from which Smith inferred (e), is false.
In our example, then, all of the following are true: (i) (e) is true, (ii) Smith believes
that (e) is true, and (iii) Smith is justified in believing that (e) is true. But it is equally
clear that Smith does not know that (e) is true; for (e) is true in virtue of the num-
ber of coins in Smith’s pocket, while Smith does not know how many coins are in
Smith’s pocket, and bases his belief in (e) on a count of the coins in Jones’s pocket,
whom he falsely believes to be the man who will get the job. (Gettier 1963, 122)
Let’s take a simple case. In the case of Gödel that’s practically the only thing many
people have heard about him—that he discovered the incompleteness of arithmetic.
Does it follow that whoever discovered the incompleteness of arithmetic is the refer-
ent of “Gödel”?
Imagine the following blatantly fictional situation. (I hope Professor Gödel is
not present.) Suppose that Gödel was not in fact the author of this theorem. A man
named “Schmidt,” whose body was found in Vienna under mysterious circumstanc-
es many years ago, actually did the work in question. His friend Gödel somehow
got hold of the manuscript and it was thereafter attributed to Gödel. On the view in
question, then, when our ordinary man uses the name “Gödel,” he really means to
refer to Schmidt, because Schmidt is the unique person satisfying the description,
“the man who discovered the incompleteness of arithmetic.” Of course you might
try changing it to “the man who published the discovery of the incompleteness of
arithmetic.” By changing the story a little further one can make even this formula-
tion false. Anyway, most people might not even know whether the thing was pub-
lished or got around by word of mouth. Let’s stick to “the man who discovered the
incompleteness of arithmetic.” So, since the man who discovered the incomplete-
ness of arithmetic is in fact Schmidt, we, when we talk about “Gödel,” are in fact
always referring to Schmidt. But it seems to me that we are not. We simply are not.
(Kripke 1980, 83–84)
about Gödel and Schmidt, we “simply are not” referring to Schmidt when
we use the name “Gödel” (Kripke 1980, 83). It is not that it is intuitive
that we are not talking about Schmidt; it is that we are not talking about
Schmidt, period.13 Facts about which propositions are intuitive appear to
play no role at all. Instead, in both cases, it looks as though Gettier and
Kripke are appealing to facts that are not purely psychological—the fact
that Smith does not know (e) in Gettier’s case, and the fact that we simply
are not referring to Schmidt when using “Gödel” in Kripke’s case—in order
to refute their target theories.
Looking back at the original presentations makes it puzzling how one
could be left with the impression that Gettier and Kripke adopt a method
that clearly treats intuitions in the “state” sense as evidence. Not once does
either say that he is treating the intuitiveness of his counterexample as evi-
dence for the counterexample’s truth. In fact, the question of (evidence for)
the truth of the counterexamples, to the extent that it occurs to Gettier and
Kripke, does not seem to strike them as something that might be seriously
disputed. The original presentations rather make it appear as though each
argument is simply the rejection of a generalization of the form “All Fs are
Gs” on the basis of a counterexample of the form “There is an F that is not
a G.” There is certainly no explicit appeal to a premise of the form “It is
intuitive that there is an F that is not a G.” So why believe, as perhaps some
xphiles do, that the best representation of the arguments’ form requires
such a premise?14
Perhaps the answer is that representing the arguments that way would at
least make sense of xphi methodology. For if Gettier or Kripke had appealed
to the intuitiveness of the counterexamples as evidence that the counter-
examples are genuine, that would provide the xphi intuition surveys with
their point. For then xphiles could say they are merely testing a hypothesis
that Gettier and Kripke themselves take to be essential to their refutations.
It would make sense to find out whether the counterexamples really are
intuitive, if Gettier or Kripke had said that they must be. But, in fact, nei-
ther says anything of the kind.
(G) Gettier refuted the JTB theory by presenting cases in which a subject
has a justified true belief that p, but it is intuitive (or intuitively true) that the
subject does not know that p.
Like (G), Bealer’s description strongly suggests (though does not come right
out and say) that a condition on Gettier having refuted the JTB theory is
that “our intuitions” are a certain way. This is wrong. If Gettier provided
evidence against the JTB theory, this evidence is that “situations like those
described in the Gettier literature are possible and that the relevant people in
Intuitions and Counterexamples 47
those situations would not know the things at issue” (1996, 122). Whether
we have intuitions to that effect simply does not matter to whether Get-
tier’s counterexamples are genuine.
As an example of a (K)-like misrepresentation of Kripke, consider Mach-
ery et al.’s (2004) description of Kripke versus the descriptivist theory of
reference:
Despite numerous disagreements, philosophers agree that theories of reference for
names have to be consistent with our intuitions regarding who or what the names
refer to. Thus, the common wisdom in philosophy is that Kripke (1972/1980) has
refuted the traditional descriptivist theories of reference by producing some famous
stories which elicit intuitions that are inconsistent with these theories. (2004, B2; empha-
sis added)
intuitive to other cultural groups. This leads them to suppose that they
have evidence that, even by philosophy’s own standards, Gettier’s alleged
counterexamples to the JTB theory of knowledge, and Kripke’s alleged
counterexamples to the descriptivist theory of reference, are not genuine.
But this mistakes intuitiveness for genuineness.
By “genuine counterexample,” I mean simply a possible case that falsi-
fies the generalization or theory in question. If Gettier has discovered pos-
sible cases of agents who justifiably and truly believe p, but fail to know p,
such as the Smith character from the 10 Coins Case, then he has refuted
the JTB theory of knowledge. Likewise, if Kripke has discovered possible
cases in which a name refers to x even though the definite descriptions
users associate with the name are not satisfied by x, as appears to be true in
his Gödel Case, then Kripke has refuted the descriptivist theory of reference
for proper names. Some people, having read and been convinced by Get-
tier’s and Kripke’s writings, have what they would describe as “intuitions”
whose contents are precisely the propositions that Gettier and Kripke cite
against their targets—intuitions that, if true, imply the falsity of the JTB
theory of knowledge and the descriptivist theory of reference. Thus, there
are, for such people at any rate, “intuitive counterexamples” to these theo-
ries. But the existence of such people and their intuitions is independent
of the existence of genuine counterexamples to the theories. There could
be genuine counterexamples, even if no proposition qualified as an intui-
tive counterexample.16 And, of course, there could be intuitive propositions
that, if true, would falsify a theory, but which are not true and, hence, not
genuine counterexamples to the theory.
When discussing the 10 Coins and Gödel cases above, I urged that, to
the extent that there is a method common to the way in which Gettier
and Kripke argue against their targets, the method is simply that of reject-
ing a generalization of the form “All Fs are Gs,” on the grounds that there
is at least one F that is not a G. There is nothing objectionable about this
method; rational enquirers everywhere attempt to falsify generalizations by
unearthing counterexamples.
There is some acknowledgment by xphiles, incredulous though it is, of
this “no intuitions” account of refutation-by-counterexample in analytic
philosophy. Alexander and Weinberg (2007), commenting on a “no intu-
itions” account due to Timothy Williamson (2004, 2007) that is similar in
some respects to mine, write,
Intuitions and Counterexamples 49
Timothy Williamson has also developed a more radical response to the restrictionist
[i.e. negative xphi] threat: rejecting the picture of philosophical practice as depend-
ing on intuitions at all! He argues that our evidence … is not any sort of mental
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 evi-
dence, 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. (2007, 72)
to refute it. But Kripke’s target is not the view that, necessarily, a name refers
to whichever object is denoted by the descriptions users associate with
the name. No descriptivist would accept this necessary generalization as
a consequence of her view, nor would Kripke accept that the alternative
to descriptivism he proposes, the so-called causal-historical view of refer-
ence, implies that names necessarily fail to have their references fixed by
description. Kripke admits that descriptive names are possible. In fact, he
thinks they are actual, though very rare. So how can a purely hypothetical
case, such as the Gödel Case, suffice to refute the descriptivist theory of
reference?
The simplest plausible answer is that there is a necessary generalization
that descriptivists would accept as a consequence of their view, which is
refuted by hypothetical cases such as the Gödel Case. For reasons we have
already been through, this generalization is not what I labeled the “descrip-
tivist theory of reference” above, but it is in the neighborhood. The relevant
generalization involves the notion of an ordinary proper name, where it is
assumed that this category can be specified independently of whichever
theory of reference correctly applies to such names. Perhaps, for example,
one could draw up a list of the names that appear in a language, “Gettier,”
“Kripke,” “Gödel,” “Smith,” and so on, and define an ordinary proper name
as one that functions semantically just as the majority of names on the list
function. Descriptivists would, it seems to me, accept that, on their view, it
is necessary that, if a name is an ordinary proper name, in the sense defined,
then its reference is fixed by description. This allows for the possibility of
nonordinary names that do not have their references fixed by description
and, at the same time, opens the door to refutation by purely hypotheti-
cal counterexample. For if “Gödel,” as used in Kripke’s Gödel Case, is an
ordinary proper name, but does not have its reference fixed by description,
then the purely hypothetical Gödel Case refutes the descriptivist theory of
reference.
Even if, as I have now admitted, the more accurate renditions of Get-
tier’s and Kripke’s putative refutations take them to involve (allegedly)
necessary generalizations and purely hypothetical counterexamples, it is
unclear how this might support the claim that there is some crucial meth-
odological difference between Gettier’s and Kripke’s arguments and cases in
which a contingent generalization is rejected by an actual, not hypotheti-
cal, counterexample.18 In any event, the fact that an argument appeals to
52 Chapter 2
what is true with respect to a hypothetical case surely does not imply that it
depends evidentially on facts about people’s psychological states. And my
objection to the view that Gettier’s and Kripke’s arguments depend on such
facts does not turn on supposing that the generalizations they target are
contingent or that the counterexamples are not hypothetical.
From his talk of “mental responses” being used as “evidence for the correct
answer,” it is clear that Goldman thinks that philosophers think of intu-
itions in the “state” sense as essential evidence for their theories. Indeed,
he appears to think this is just obvious. I have been arguing that this is
a mistake. In contrast with Goldman, I think it is very difficult to find
a case of an argument in philosophy that depends on “people’s mental
responses” as evidence. Gettier’s and Kripke’s arguments, which Goldman
clearly has in mind, are not such cases, as I have argued. So where are we
supposed to find this “extensive and avowed reliance on intuition”? If we
focus on just that part of philosophy that involves attempts to refute gen-
eralizations on the basis of counterexamples, it strikes me that Gettier’s
and Kripke’s arguments are paradigmatic instances of the type. There is
surely no obvious reason why the method of offering counterexamples to
the epistemological or semantic theories that interest Gettier and Kripke
should be different in some important respect from the method of offering
counterexamples to metaphysical or ethical theories. It would be peculiar
if, while Kripke’s counterexamples to the descriptivist theory, for example,
require no backing via an appeal to their intuitiveness, counterexamples to,
say, utilitarianism do require it. If I am right, and the generalization versus
counterexample method we see in the work of Gettier and Kripke does not
evidentially depend on the intuitiveness of any principle or proposition,
then it is safe to assume that the method, as it appears in the work of others
in different areas, does not evidentially depend on this either.
Methodological confusions infect not just implicit construals of coun-
terexamples in accounts of the functioning of specific philosophical argu-
ments, but also in explicit definitions of counterexamples. As an example,
consider Brian Weatherson’s (2003) recent definition of “counterexample”:
“Let us say that a counterexample to the theory that all Fs are Gs is a pos-
sible situation such that most people have an intuition that some particular
thing in the story is an F but not a G” (2003, 2; emphasis added). The “let
54 Chapter 2
There are two misleading ideas here. One of these, which will have to wait
until chapter 5 for full debunking, is the idea that Socrates, as represen-
tative for a long-standing methodological style still present, DePaul and
Ramsey say, in contemporary analytic philosophy, is interested primarily in
concepts. This is wrong, both of Socrates and of contemporary analytic phi-
losophers. Gettier is not interested in epistemological concepts, at least not
primarily, and Kripke is not primarily interested semantic concepts. Rather,
both are primarily interested in the phenomena that certain epistemologi-
cal and semantic concepts characterize, knowledge in Gettier’s case and ref-
erence in Kripke’s. As for Socrates, although I am no expert, on my reading
of the dialogues, Plato’s protagonist is interested in things like piety, justice,
and knowledge—themselves, not merely our concepts of these things.
Intuitions and Counterexamples 55
The other misleading suggestion in the quote from DePaul and Ramsey
is the suggestion that it is a condition on there being a counterexample to
a “proposed definition” that the definition “yields a result that conflicts
with our intuitions.” This is the same mistake Weatherson makes. It is a
mistake to say that a definition of justice faces counterexamples only if
it yields results that conflict with our intuitions. Rather, such a definition
faces counterexamples if, and only if, there are examples of injustice that
count as just by the definition, or examples of justice that do not count as
just by the definition. The definition’s accuracy is independent of whether,
and how many, people have intuitions that conflict with the definition.
Once again we see unwarranted assumptions about the argumentative
role of intuitions in the “state” sense creeping into philosophers’ concep-
tions of their own method. It is no surprise, then, to find xphiles mak-
ing these same assumptions. Indeed, most of the misleading quotations
in the preceding paragraphs come from analytic, not experimental, phi-
losophers. So xphiles can hardly be blamed for assuming that analytic phi-
losophers adopt methods that involve empirical speculation about people’s
intuitions. In fact, as eminent a philosopher as Frank Jackson (2000) is on
record urging that “serious opinion polls on people’s responses to various
[philosophical] cases,” of just the sort that xphiles have been busy conduct-
ing, are a necessary component of philosophical theorizing (2000, 36–37).
Whether xphiles are to blame for it or not, however, it is a mistake. An
opinion poll might reveal whether or not a given proposition is intuitive or
widely believed, but, if I am right, and intuitions (in the “state” sense) are
not, and need not be, treated as evidence in philosophical theorizing, then
it is at best unclear how the results of xphi intuition surveys are supposed to
pose a challenge to philosophy as it has been traditionally practiced.
the fuss I will soon make about the arguments that philosophers give for
their judgments about thought experiments and cases simply relocates the
inevitable appeal to intuitions in the “state” sense.
In chapter 3, I call this problem the relocation problem, and I sketch a solu-
tion, one more fully developed later, in chapter 5. The bulk of chapter 3 is
given over to examining Williamson’s case for the view that arguments for
judgments about thought experiments and cases are not required, because
demanding them leads to a debilitating form of skepticism.
3 The Relocation Problem and Williamson on “Judgment
Skepticism”
3.1 Introduction
The main goal of this chapter is to assess one of Timothy Williamson’s com-
plaints against negative xphi, namely that it leads to an untenable form
of skepticism he calls “judgment skepticism.” As I read him, Williamson
takes what I called “the evidence-for-the-evidence question” at the end of
chapter 2—the question of how we know that certain philosophical coun-
terexamples are genuine—as illegitimate, or at least as illegitimate if viewed
as a demand for an inference that might support the conclusion that this or
that counterexample is genuine. On Williamson’s view, part of the problem
with negative xphi is precisely that it raises the evidence-for-the-evidence
question. In effect, negative xphiles reject one possible answer, namely that
it is our intuitions that serve as the evidence-for-the evidence, and leave the
question of what does serve wide open. But Williamson does not think that
our knowledge of, for example, the counterexamples used to refute the JTB
theory of knowledge comes via argument or inference. Instead it is a basic
kind of philosophical evidence—evidence that requires no evidence of its
own. Furthermore, seeking inferential reasons for supposing that proposi-
tions of this sort are true, leads, Williamson says, to judgment skepticism.1
I will argue that these views of Williamson’s are wrong. It is standard
practice in philosophy to argue for judgments about the thought experi-
ments and cases from which many philosophical counterexamples derive.
So, as a matter of fact, such judgments are not treated as a basic kind of
evidence. In my view, the idea that judgments about thought experiments
and cases count as basic evidence, or are noninferentially justified, is a lin-
gering remnant of the myth of the intuitive in Williamson’s thinking. In
some ways, Williamson qualifies as a powerful opponent of the myth; for
60 Chapter 3
example, he is hostile, and rightly so, to the notion that intuitions in the
“state” sense play an evidential role in philosophical reasoning. Despite this,
some elements of his overall methodological stance strike me as wedded,
still, to some of the views in that package of mistaken views that constitute
the myth, the view that thought experimentation involves noninferential
judgment being one of these. Furthermore, I will argue that the charge of
risking an overarching skepticism about judgment does not stick; nothing
in the negative xphi critique opens negative xphiles to such a charge, and,
more generally, there is nothing wrong with raising the question of what
sorts of inferential reasons we might have for supposing that this or that
judgment about a thought experiment or case is true. In particular, raising
this question does not lead to a destructive form of skepticism. In other
words, the evidence-for-the-evidence question is perfectly legitimate.
As I admitted at the end of chapter 2, sometimes we philosophers need
to determine whether an alleged counterexample is the real McCoy. I
argued in chapter 2 that there might be counterexamples to a philosophical
theory regardless of whether they count as the contents of anyone’s intu-
itions. Gettier refuted the JTB theory of knowledge, if his judgment about
the 10 Coins Case is correct; and his judgment might be correct, regardless
of whether anyone shares the judgment. But there is a legitimate question
left open here: how do we know that Gettier refuted his target theory, if,
indeed, he did? If the counterexample is itself properly regarded as evidence
against some more general theory, what is the evidence-for-the-evidence;
that is, what justifies the judgment that it is a genuine counterexample?
I reject the answer that we know this by knowing whether facts concern-
ing intuitions in the “state” sense obtain. Instead, as I have said, philoso-
phers argue for their judgments about counterexamples, and (the premises
of) these arguments are (and are treated as) the evidence for such judg-
ments. In chapters 4 and 5 I will demonstrate this by presenting many clear
examples of philosophers arguing that their counterexamples are genuine.
However, the view that it is arguments, not intuitions, that serve as the
evidence for the genuineness of philosophical counterexamples raises the
following question: what is the evidence for the premises of these argu-
ments? Before turning to Williamson in section 3.3, I briefly, in section 3.2,
reply to the worry raised by this question. The issue is then dealt with more
fully in chapter 5.
The Relocation Problem 61
My case for the claim that philosophy’s methods are free of appeals to
intuitions begins, in the next chapter, with a discussion of Gettier judg-
ments because these are thought by many philosophers, xphile and non-,
to be the clearest cases we have in philosophy of judgments that are brutely
intuitive, where to be “brutely intuitive” is to be held and justified on the
basis of intuition alone. If it turns out that not even Gettier judgments are
brutely intuitive, that will go a long distance toward showing that intu-
itions play a far less prominent role in philosophy than many philosophers
take them to.
An additional reason for beginning with Gettier judgments is that such
judgments serve as a main springboard for Timothy Williamson’s (2007)
work on philosophical method. I have said that, despite broad similarities,
Williamson and I differ in our methodological outlook. One important dif-
ference concerns our attitudes toward the cross-cultural results regarding
Gettier judgments reported in Weinberg et al. 2001.
64 Chapter 3
whatever we used to do the original justifying, and then justify our justi-
fications for the original justifications, and so on. Yet, despite the seeming
seriousness of the challenge, Williamson suggests that we are free to ignore
it. On his view, the negative xphi challenge rests on mistaken implicit views
about the nature of evidence, views that lead to a wide-ranging and implau-
sible judgment skepticism.
Williamson describes this peculiar form of skepticism in chapter 7 (“Evi-
dence in Philosophy”) of The Philosophy of Philosophy. Unlike other, more
familiar varieties of skepticism, which target “the distinctive features of
perception, memory, testimony, or inference,” judgment skepticism targets
“our practices of applying concepts in judgment” (Williamson 2007, 220).
Williamson warns that “although, in practice, judgment skeptics are skep-
tical only of a few judgments or concepts at a time, the underlying forms
of argument are far more general” (224). In Williamson’s view, judgment
skeptics have no plausible way to avoid the view that we never knowledg-
ably apply concepts in judgment, but this is a quick road to “total intel-
lectual paralysis,” and seems self-defeating on its face. Williamson likens
judgment-skeptical arguments to a bomb “which, if it detonates properly,
will blow up the bombers and those they hope to promote together with
everyone else” (224).
Although he names no xphile names in his chapter 7, it is clear that
Williamson regards negative xphile arguments as resting on these explo-
sive judgment-skeptical forms. Among the specific examples of judgment
skeptics Williamson cites, which also include “eliminativists” about the
propositional attitudes and so-called mereological nihilists, who, on meta-
physical grounds, deny the existence of macroscopic physical objects such
as baseballs and mountains, is a hypothetical philosopher who thinks that
“the [Western] Gettier judgment is mere cultural prejudice” (211). This is a
strong hint that Williamson has Weinberg et al. 2001 in mind and takes it to
represent a judgment-skeptical position.3 Earlier, in his chapter 6 (“Thought
Experiments”), Williamson mentions Weinberg et al. 2001 explicitly, criti-
cizing that paper for making too much of cross-cultural differences in folk
judgments about Gettier cases. (According to Williamson, the authoritative
judgments are those made by experts, i.e., those with some training in phi-
losophy. I will return to this idea in chapter 6; it marks a further difference
between Williamson and me.) It is thus natural to read Williamson’s chap-
ter 7, in part, as raising a further objection to the negative xphile position:
that position rests on judgment skepticism.
66 Chapter 3
in this instance of Western, East Asian, and Indian judges—shows that the
content of the Western Gettier judgment cannot be justifiably regarded as
evidence.
However, this is a superficial and inaccurate picture of the structure of
the negative xphile argument, and it works even less well, it seems to me,
in picturing the structure of arguments for eliminativism or mereological
nihilism.8 The problem, I think, is that Williamson has wrongly assumed
that there must be some common thread between the views he describes
as judgment skeptical, when in fact very different philosophical concerns
underlie these various positions. For example, those who adopt mereologi-
cal nihilism do so for far different reasons than those that lead negative
xphiles to adopt the negative xphile position. In the former case, the rea-
sons have to do with the metaphysics of causation and worries about over-
determination; in the latter case, the reasons have to do with the existence
of variability in intuitions about hypothetical examples. There is no over-
arching view about the nature of evidence, such as Evidence Neutrality,
that unifies these reasons or the views they are meant to support. In any
case, there need not be; mereological nihilists and negative xphiles can
perfectly well argue for their views without relying on any general claims
about the nature of evidence.
As emphasized in chapter 2, in the case of the argument for the nega-
tive xphile position, what is crucial is not disagreement per se, but rather
disagreement of a certain type, namely disagreement in intuitive judgments
and along a dimension (e.g., culture) that ought not to matter to the truth
of the intuitions. (In chapter 1, this was described an as variability of intu-
itions along “truth-irrelevant” lines.) No negative xphile need agree with
Evidence Neutrality in order to make this sort of argument.
Negative xphiles are wrong for a different reason, one that has nothing
to do with their views about the nature of evidence. Instead, their mistake
concerns what sort of evidence is regarded as available by philosophers who
argue by appeal to intuitive propositions. They are wrong in thinking that
the putative evidence for the content of an intuitive philosophical judg-
ment is that the judgment is intuitive. This is the fundamental mistake, the
mistake, as I characterized it in chapter 2, of taking (analytic philosophers
to suppose) that intuitions in the “state” sense are evidence for their con-
tents. Indeed, I have already argued, the negative xphile critique depends
not only on taking analytic philosophers to rely on intuitions in the state
The Relocation Problem 71
sense as evidence, but also on taking them to treat intuitions in the “state”
sense as an essential source of evidence for the contents of those intuitions.
However, as I will show in the next chapter, in many instances, and cer-
tainly with respect to Gettier’s 10 Coins Case and Kripke’s Gödel Case, there
are further arguments, as opposed to mere intuitiveness, that give evidential
support to specific judgments concerning these cases. If I am right, that
is a good thing for the overall anti-xphi stance I share with Williamson.
For, even if Williamson were right that the falsity of Evidence Neutrality
suffices to show that negative xphiles have no reason to deny that we may
justifiably regard the contents of our intuitions as evidence, the question
would remain of whether we can assert that we may justifiably regard them
as such. This is left something of a mystery in Williamson’s book. We judge
as we do about Gettier cases, for example, but is there reason to think we
are right to make these judgments? That is, is there any reason to think
that the judgments are true? Williamson does not answer these questions
and in fact appears to believe that they need not be answered. I think that
they can and should be answered, that the evidence-for-the-evidence ques-
tion, as I described it earlier, can and should be answered as well, and, most
importantly, that intuitions and intuitiveness play no significant role in
answering it.
4 The Evidence for the Evidence: Arguing for Gettier
Judgments
4.1 Introduction
to license any conclusion about whether and to what extent the believers
or disbelievers in p are justified in believing or disbelieving p.3 To take a
relatively straightforward example, creationists and evolutionists disagree
over the truth of the theory of evolution. But no one would for a moment
think that this fact alone—that is, the fact that there exists this disagree-
ment—shows that there is some epistemological flaw in the belief in evo-
lution (or in creationism, for that matter). The important issue is whether
belief (or disbelief) in the theory of evolution is justifiable. If it is—if, for
example, there are strong, compelling arguments that the theory is true—
then that there is group of people who reject the theory does not matter
even a little bit to the theory’s epistemic credentials. If one is in possession
of a good argument in favor of p, then it is epistemically unimportant that
p is rejected by someone or some group of people.
These points are fairly obvious in cases of simple disagreement in belief.
Such disagreement is unproblematic because it can in principle be resolved
by appeal to reasons and arguments, and these reasons and arguments will
almost never cite psychological facts about how many or which sorts of
people believe or disbelieve the proposition in question. I think that the
same is true of a great many intuitive disagreements; just as in cases of
simple disagreement in belief, intuitive disagreements can and should be
resolved by appeal to reasons and arguments. And, as in cases of disagree-
ment in belief, if one side has a good argument for the intuition that p, then
the fact that others do not intuit p or intuit not-p is epistemically unimport-
ant. To regard intuitive clashes as different, to think that such clashes really
do have serious epistemic consequences, is to treat intuition as an essential
source of evidence regarding the relevant proposition, or perhaps even as
the only source. For, if there were other evidence from a different source
that provided all that was required to decide the issue, the case would col-
lapse into a case of simple disagreement in belief and the only question to
ask would be the straightforward question of which side of the clash has
more evidence in favor of it.
Recognizing that negative xphiles take analytic philosophers to treat
intuitions about thought experiments and cases as the only evidence, or
at least as essential evidence, for what is true in those thought experiments
and cases is important for understanding the overall metaphilosophical
dialectic. My own view is that philosophical arguments never appeal to
the intuitiveness of a judgment about a case in order to justify belief in
The Evidence for the Evidence 77
In the next section, I present some of the arguments that have been given
for Gettier judgments in particular. In chapter 5, I will present other thought
experiments and the accompanying arguments that have been presented for
judgments about them. It is important for recognizing the mistaken pre-
suppositions of negative xphi, but also, more positively, for understanding
philosophical method, that we look to actual cases to determine how philos-
ophers reason about thought experiments. As I will show, failing to provide
an argument, or at least a set of considerations, in favor of a judgment about
a thought experiment must be very rare. In any case, there are plenty of
counterexamples—many cases in which intuitive judgments about thought
experiments are supplemented with reasons and arguments.
The first subsection below, 4.3.1, describes and comments on the argument
that Gettier himself gives for the judgments he makes about the original
two cases he presents in “Is Justified True Belief Knowledge?” The second
subsection below, 4.3.2, describes arguments that have been given by oth-
ers, both for judgments about those original two cases, as well as arguments
for Gettier judgments more generally. My claim—that one never finds cases
of intuitions about thought experiments that do not receive supporting
arguments—should be kept in mind. It is an empirical claim, and fully sub-
stantiating it will take more research and case studies.6 The claim is also one
concerning the whole of analytic philosophy, and my scope will be limited
in the remainder of this chapter to epistemology; in fact, it will be limited
to that small part of epistemology that some philosophers derisively call
“Gettierology.”7 Nevertheless, I think of the examples I will give as present-
ing a challenge to xphiles who think they see heavy reliance on intuition
in analytic philosophy: given that, in several central cases, including cases
involving Gettier judgments, appeals to intuition appear to be entirely
absent, show us cases in which such appeals are clearly meant to be doing
the serious work you say they are.
In the last section of the chapter, 4.4, I argue that if we focus not on the
reception of a thought experiment or a hypothetical case counterexample
but instead on its invention or discovery, we will find far less plausible the
idea that intuitions play a crucial role in arguments involving these thought
experiments and cases. It is true perhaps that, as readers of philosophy, we
The Evidence for the Evidence 79
typically run. One remarkable thing about the paper in which Gettier pre-
sented his famous cases is its length. It is short even given the quite short
average length of papers in Analysis, where it was first published. Most
papers in analytic philosophy are considerably longer and some of this
length is given over, as Cappelen ably demonstrates, to arguing for judg-
ments made about thought experiments. Gettier’s paper does contain con-
siderations that are meant to lend support to the judgments about the cases
described therein, but there is something to the idea that the main function
of the paper is simply to present the cases.
Despite the fact that Gettier’s refutation of the JTB theory fails, in this
way, to be paradigmatic or representative, it is still true that if there is an
appeal to intuitions as evidence anywhere in analytic philosophy, then it is
present in Gettier’s refutation of the JTB theory. So the stakes are high, but
really only for the negative xphile critique. If Gettier’s refutation is a clear
example of an argument that appeals to intuitions as evidence, that would
show only that a fairly unrepresentative argument in analytic philosophy is
susceptible to that critique; other arguments may be immune. However, if
it turns out that not even Gettier’s refutation is a clear example of an argu-
ment that appeals to intuitions as evidence, then that would show, I think,
that we are very unlikely to find such appeals elsewhere.
It is this latter circumstance that is the actual one: there is no appeal
to intuitions as evidence in Gettier’s refutation. Four points are relevant
here. First, as I showed in chapter 2, the best way to understand Gettier’s
refutation is to view it as involving the presentation of counterexamples to
the JTB theory (or, more precisely, to a generalization implied by the JTB
theory). The evidence against the JTB theory is simply the counterexamples
themselves. One can ask the further, evidence-for-the-evidence question,
namely, what justifies the view that Gettier’s alleged counterexamples are
genuine? But this really is a further question—whether Gettier succeeded in
refuting the JTB theory depends only on whether the counterexamples he
offers are genuine and not at all on whether Gettier, also and in addition
to providing counterexamples, offered evidence for his evidence, that is,
evidence that the counterexamples are genuine.
Second, there are grounds for saying that the exceptionally clear way in
which Gettier presents his counterexamples makes arguing for their genu-
ineness superfluous. For example, I would say that the fact that Gettier’s
Smith character in his 10 Coins Case does not know that the man who will
The Evidence for the Evidence 81
get the job has ten coins in his pocket is obvious, and would be to anyone
who correctly understands the structure of the case, a structure that Gettier
makes very easy for any minimally competent reader to grasp. It is clear,
as Gettier himself puts it, that Smith does not know, and clear also that he
does not know even while justifiably and truly believing. Xphiles might
complain that these adjectives, “clear,” “obvious,” and related terminology
are just more “intuition-talk,” but that is wrong. Being obvious or clear is
different from being intuitive. In ordinary English, to describe a point or
claim as clear or obvious is to comment on how easily one can recognize
that the claim or point is true, namely very, very easily indeed. That might
be something a theorist of intuitions would want to say about intuitive
claims, but the idea that the ease with which one may recognize that a
claim is true counts as evidence in favor of the claim is an odd one. To say
that a claim or point is obvious or clear is to presuppose that it is true. It is
not to cite something—clarity or obviousness—as evidence for the truth of
the claim or point. So when Gettier claims that it is “clear” that his Smith
character does not know that the man who will get the job has ten coins
in his pocket, he is not adducing evidence for the claim, let alone evidence
that consists in his or other people’s intuitions. He is instead pointing to
the ease with which one can recognize that the claim is true. It is easy to
recognize this because the case is structured so simply and clearly, the cru-
cial bit of structure being this: Smith bases his belief, in part, on the count
of coins in the pocket of a man who will not get the job, namely Jones.
Anyone for whom this registers should be able to recognize that Smith’s
justified true belief does not, therefore, add up to knowledge. It is clear that
it does not, and that clarity is partly a function of the way the case is struc-
tured and presented.
But only partly, and this brings me to the third point: Gettier provides
a reason for thinking that his judgment about the 10 Coins Case is true. He
answers the evidence-for-the-evidence question concerning the 10 Coins
Case as follows:
But it is equally clear that Smith does not know that (e) is true; for (e) is true in virtue
of the number of coins in Smith’s pocket, while Smith does not know how many coins
are in Smith’s pocket, and bases his belief in (e) on a count of the coins in Jones’s pock-
et, whom he falsely believes to be the man who will get the job. (Gettier 1963, 122)
This reason, brief though it is, counts as an argument for the crucial judg-
ment that Gettier makes about the case, namely the judgment that Smith
82 Chapter 4
Smith is justified in believing (h), since (h) follows from his justified belief
that Jones owns a Ford and Smith sees this implication and forms his belief
in (h) precisely because he sees the implication between the proposition
that Jones owns a Ford and (h). Gettier then argues that if two further con-
ditions hold, then Smith’s belief in (h), though true and justified, is not
knowledge:
But imagine now that two further conditions hold. First Jones does not own a Ford,
but is at present driving a rented car. And secondly, by the sheerest coincidence, and
entirely unknown to Smith, the place mentioned in proposition (h) happens really
to be the place where Brown is. If these two conditions hold, then Smith does not
know that (h) is true, even though (i) (h) is true, (ii) Smith does believe that (h) is
true, and (iii) Smith is justified in believing that (h) is true. (Gettier 1963, 123)
The Evidence for the Evidence 83
But there is just no evidence anywhere in the paper that the purpose of Get-
tier’s two thought experiments (the 10 Coins and Brown in Barcelona cases)
is to “generate intuitions.” None whatsoever. Alexander and Weinberg con-
tinue to misdescribe Gettier’s method in the rest of the passage:
Gettier’s thought-experiments involve a person who has deduced a true belief 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, the person
lacks knowledge that q. Purportedly, when we consider this case, we will have the
intuition that the person whose epistemic position is detailed in the thought-exper-
iment does not know that q. Further, this is to count as sufficient evidence against
the claim that a person knows that p just in case that person’s true belief is justified.
(Alexander and Weinberg 2007, 57; emphasis added)
This is clearly an argument for the claim that Smith does not know that
(e) is true. The conclusion is stated first, before the semicolon, and what
follows the semicolon is the argument’s premise. In fact, the premise is
helpfully preceded by the premise-indicator term, “for”; Smith does not
86 Chapter 4
know (e) for the reason Gettier cites. Nagel seems to think that, even if this
is an argument, it is neither explicit nor is it one “showing exactly” why
Smith does not know that (e) is true. In my view, it is both explicit and does
explain why Smith does not know—explains it exactly, even.
Is this a quibble? Perhaps there is a fairly important methodological
point here: if what we expect when looking for arguments for judgments
about thought experiments and counterexamples are “explicit” arguments
“showing exactly” why a judgment is true, we might find fewer examples
than we would if our criteria were a bit laxer. I do not think it is fair to
characterize Gettier’s arguments for his judgments about his cases as less
than explicit or exact (his arguments are short, but that is different), but,
in general, expecting explicit and exact arguments seems too demanding.
Sometimes, surely, a consideration or two merely suggesting, as opposed
to entailing, that a judgment is true will suffice. In any case, the idea that
judgments about thought experiments receive support only from intu-
itions would not be made more plausible or acceptable if it turned out that
many such judgments fail to be presented with explicit and exact argu-
ments to back them up. That would prove only that the arguments for
these sorts of judgments are not of a particular sort, not that there are no
such arguments at all.
It is true that Gettier does not offer any “positive analysis of knowledge
of his own,” but offering such an analysis is not the aim of Gettier’s short
paper; the aim is rather to argue against the JTB theory. Failing to offer a
positive analysis can hardly be taken as grounds for supposing that he does
not know or cannot explain why his Smith character lacks knowledge, since
knowing or being able to explain this does not require a positive analysis.
Whether he fails to specify “any necessary conditions on knowledge”
that are lacking in his cases, as Nagel claims, is open to debate. It is true that
Gettier does not specify such conditions by laying them out explicitly and
describing them as necessary conditions on knowledge generally. On the
other hand, he does offer a reason, or set of reasons, for thinking that, in
each case, Smith does not know. So, a plausible interpretation of what Get-
tier is doing in his arguments for his judgments about his cases is that he is
indicating (if not strictly specifying) that there are conditions on knowing
that Smith, in each case, has failed to meet.
Of course, later commentators on Gettier’s paper have attempted to
explain in a more general way why Gettier’s Smith character does not know,
The Evidence for the Evidence 87
of explaining exactly why they are cases of justified true belief without
knowledge.12
However, as I said above, I think Gettier does offer arguments intended
to explain exactly why his Smith character, in both the 10 Coins and Brown
in Barcelona cases, fails to know. According to Gettier, Smith’s belief, in
each case, fails to satisfy a necessary condition on knowing. Gettier does
not state a general version of the condition, but it is all but explicit in the
passages, quoted above, in which Gettier is arguing for the conclusion that
Smith does not know. In each case, Smith’s justified belief is only luckily
true. It is the presence of this sort of “epistemic luck,” which, in each case,
and according to Gettier, disqualifies Smith’s justified belief as knowledge.
So the purported necessary condition is just this:
If S’s belief that p is (also) knowledge that p, then S’s belief that p is not
luckily true.
The passages in which Gettier is arguing that Smith does not know make
it clear that epistemic luck is meant to explain Smith’s lack of knowledge.
In the 10 Coins Case, Smith’s belief is true “in virtue of” a fact that Smith
“does not know,” so, if he gets things right, as he does, that is just luck.
Smith, in the 10 Coins Case, bases his belief on the count of the coins in
the pocket of a man who will not get the job, so, again, getting things right
involves a lucky coincidence. In the Brown in Barcelona Case, the fact that
makes Smith’s belief true (the fact that Brown is in Barcelona) obtains only
by the “sheerest coincidence, and entirely unknown to Smith”—luck again.
A more detailed picture of what goes wrong in Gettier cases, a more
detailed picture, that is, of why agents in Gettier cases lack knowledge, is
desirable. For example, it would be nice to have an analysis of what it is to
be epistemically lucky, and Gettier does not provide this in his paper (nor,
of course, does he use this terminology). He gives a pair of examples of an
agent, Smith, whose justified beliefs are merely luckily true, but there is
no definition, or general account, of the phenomenon of epistemic luck.
Failing to give an account of epistemic luck does not mean that Gettier’s
argument fails to explain exactly why Smith lacks knowledge, however.
Exact explanations can be in terms of concepts for which we lack analyses.
In any case, contemporary analytic epistemologists are making interesting
progress on analyzing epistemic luck,13 and some of this work can perhaps
The Evidence for the Evidence 89
be put to use in fleshing out Gettier’s argument. But none of this, nor even
whether Gettier was on the right track in diagnosing his own cases, mat-
ters to the important underlying methodological issue, which is just this:
does Gettier argue for his judgments about his cases, or does he instead rely
solely on their intuitiveness? I hope to have shown in this section that the
answer is very clearly that Gettier argues for his judgments. He does so via
an all but explicitly proposed necessary condition on knowledge, one bar-
ring luckily true beliefs from qualifying as knowledge.14
Smith believes (1) are true, but the grounds on which he accepts these grounds, viz.
that Brown knows them, are false. (Clark 1963, 448)
This is an argument that Smith fails to know (1). (Never mind what (1) is
and ignore the details of the case. The purpose of quoting this passage is
simply to bring out the fact that Clark argues for his Gettier judgment.) He
cannot know it because it is based on a “wild guess” of Brown’s that just
“happens to be right.” This claim, which echoes Gettier’s claims about the
effect of epistemic luck, is then connected to Clark’s general explanation of
Gettier judgments: The luck involved in the case generates a false ground
for Smith’s belief in (1).
Turn now to Goldman 1967, which argues that, in many instances, justi-
fied true believers in Gettier cases fail to know because the facts that cause
them to believe are not appropriately causally connected to the facts that
make their beliefs true. Goldman’s explicit diagnosis of Gettier’s original
cases is that Gettier’s Smith character fails to know because a necessary
causal condition on knowledge is not satisfied in those cases. Goldman’s
discussion of his diagnosis and of its contrast with Clark’s proposal is a
remarkably clear example of a philosopher arguing for Gettier judgments.
Here is a passage from Goldman 1967 (q is the proposition that Jones owns
a Ford, while p is the proposition that Jones owns a Ford or Brown is in Bar-
celona; Goldman’s discussion in the passage concerns Gettier’s Brown in
Barcelona case):
Michael Clark, for example, points to the fact that q is false and suggests this as the
reason why Smith cannot be said to know p. … I shall make another hypothesis to
account for the fact that Smith cannot be said to know p, and I shall generalize this
into a new analysis of “S knows p.”
Notice that what makes p true is the fact that Brown is in Barcelona, but that this
fact has nothing to do with Smith’s believing that p. That is, there is no causal con-
nection between the fact that Brown is in Barcelona and Smith’s believing p. … Thus
one thing that seems to be missing in this example is a causal connection between
the fact that makes p true (or simply: the fact that p) and Smith’s belief of p. The re-
quirement of such a causal connection is what I wish to add to the traditional analysis.
(Goldman 1967, 358; emphasis in original)
unless one is wedded to the idea that Gettier’s demonstration of the falsity
of the JTB theory depends on intuitions as evidence and that this is the
standard interpretation of his demonstration in post-1963 epistemology. If
one is wedded to that idea, this passage is surprising, but there are count-
less other such passages in the post-1963 literature, ones that show, just like
this one from Lehrer and Paxson 1969, that epistemologists took Gettier to
be arguing against the JTB theory, and arguing for his judgments about his
cases. Here are Lehrer and Paxson on what Gettier did and how he did it:
This analysis of nonbasic knowledge [the JTB theory] is, of course, defective. As Ed-
mund Gettier has shown, there are examples in which some false statement p entails
and hence completely justifies S in believing that h, and such that, though S cor-
rectly believes that h, his being correct is mostly a matter of luck. Consequently, S
lacks knowledge, contrary to the above analysis. (Lehrer and Paxson 1969, 227)
Note that Lehrer and Paxson attribute a reason to Gettier for thinking that
the agents in his cases do not know: their true beliefs, though justified, are
“mostly a matter of luck.”
Later, in motivating their defeasibility analysis, Lehrer and Paxson, like
Clark and Goldman before them, look for an explanation for what the epis-
temic luck in the various Gettier cases consists in:
In the examples referred to above, there is some true statement that would defeat
any justification of S for believing that h. In the case of the pyromaniac [a Gettier
case due to Skyrms 1967], his justification is defeated by the true statement that
striking the match will not cause it to ignite. This defeats his justification for believ-
ing that the match will ignite upon his striking it. (Lehrer and Paxson 1969, 227)
They then propose adding a defeasibility condition to the JTB analysis and
spend the bulk of the rest of the paper trying to arrive at the correct account
of epistemic defeat. None of this attempt involves appealing to intuitions as
evidence. Lehrer and Paxson’s paper contains a good number of relatively
complicated thought experiments but the various conclusions they draw
about these thought experiments are argued for; nothing rests or hangs on
what strikes them as intuitive.
At one point in Lehrer and Paxson 1969, there is a comparison between
two thought experiments. In one, according to Lehrer and Paxson, there is
justified true belief and also knowledge;16 in the other, there is justified true
belief without knowledge (a variation on one of Gettier’s original cases). But
the two cases are similar in that there at least appears to be a “defeater” in
each case, that is, a true proposition, which, were it conjoined to the agent’s
94 Chapter 4
One comment about the how this bears on what I earlier called the relo-
cation problem. The arguments for Gettier judgments I canvassed in this
section appeal to premises, of course. So what justifies these premises? For
example, why should we think, with Gettier, that the sort of epistemic luck
involved in his cases is what prevents the justified true beliefs in them from
qualifying as knowledge? Will we not reach a stage at which we will have to
appeal to brute intuition? These are reasonable questions, and my answers
to them will not be fully complete until I return to the relocation problem
in chapter 5.
However, for now, I want to draw attention to the following conse-
quence of having relocated the appeal to intuition elsewhere, “further up”
on the justificatory chain, as it were, even if that were all that could be
accomplished. The consequence is that the current xphi data, the data that
has been gathered so far, is entirely irrelevant to whether we know that this
or that philosophical judgment is true. For if intuitions come into play, but
only further up the justificatory chain—for example, with respect to a pro-
posed necessary condition on knowledge that bans certain sorts of luckily
true beliefs—then the only relevant data would be data concerning people’s
intuitions about whether that is, indeed, a necessary condition on knowl-
edge. But of course there is no empirical data on that question; xphiles
do not have data about people’s judgments about more general epistemic
principles. Hence, if the appeal to intuition has been relocated, then there
is currently no reason based on empirical data about intuitions to be at all
skeptical about any philosophical judgment or principle. My view is that
it is a mistake to think that intuitions enter the picture at any stage, and I
will argue for this more fully in the next chapter. However, relocating the
appeal to intuition, even if that were all that pointing to the various and
sundry arguments that have been given for Gettier judgments and other
judgments about thought experiments could accomplish, would stop the
negative xphi challenge dead in its tracks.
Here is an objection to my claim that there are arguments for Gettier judg-
ments, both in Gettier 1963 and in many post-1963 examples of Getti-
erology besides:18 There are arguments there, certainly, but conceiving of
96 Chapter 4
to intuitive judgments for these reasons, but I think this is a mistake. The
mistake involves taking the wrong perspective on thought experiments.
The perspective that is usually taken is that of a consumer of a thought
experiment. When thinking about philosophical arguments, we, for what-
ever reason, think of our reactions to these arguments, and to the cases,
counterexamples, and thought experiments they involve. Now, it may be
that many of us do make relatively spontaneous, nonreflective, and nonin-
ferential judgments when encountering a thought experiment in someone
else’s work. If we instead take the perspective of the producer (or inventor or
discoverer) of a thought experiment and ask how the producer’s judgments
about his or her own thought experiments are generated, the picture is very
different. For, as many analytic philosophers know, inventing an illuminat-
ing and fruitful thought experiment, or constructing a good counterexam-
ple to an even somewhat plausible theory, is difficult business—so difficult
that to describe the judgments that one elicits in oneself, having invented
the thought experiment, or constructed the counterexample, as spontane-
ous, nonreflective, or noninferential seems plainly absurd.
This general point about the different perspectives one can take on
thought experiments can be put in terms specific to Gettier 1963 and the
Gettier cases and Gettier judgments contained therein: perhaps many read-
ers of Gettier 1963 intuit that Gettier’s Smith character does not know, but
Gettier himself did not intuit this. For Gettier, arriving at that judgment took a
considerable amount of ingenuity, careful thought, and inference. To cook
up his counterexamples to the JTB theory, he needed a recipe, and it is to
that recipe that we should look in trying to get clear on the methods he
used to refute the JTB theory.
Gettier makes the recipe fairly clear, and I described its main ingredi-
ent (epistemic luck) in the previous section when discussing Gettier’s argu-
ments for his judgments about his cases. The point I want to make here is
that he did not pull the counterexamples out of thin air—of course he did
not. No one pulls philosophical thought experiments or counterexamples
out of thin air. The question of whether they are effective thought experi-
ments or genuine counterexamples ought to be answered by appeal to the
sorts of considerations that led to their production, not by citing facts about
how they happen to have been received.
This, then, is the second reply to the order of explanation objection: if
we take the perspective of the inventor of a thought experiment, or the
The Evidence for the Evidence 99
5.1 Introduction
what does and does not get appealed to as evidence in them. It strikes me,
as it strikes Cappelen, as ironic that this method has never been adopted by
negative xphiles. They simply assume that philosophers treat intuitions as
evidence and then proceed with their critique. The irony is that the empiri-
cal evidence concerning first-order philosophical practice reveals their own
intuitional diversity evidence to be largely irrelevant to philosophy as it is
actually practiced.
For my purposes in this book, the most important examples to examine
are those for which there exist xphi data showing that the relevant intuitions
exhibit the kind of truth-irrelevant diversity that negative xphiles believe
to be a serious problem. Because there is now a fair bit of data concerning
intuitional diversity relative to Gettier cases, I expended a fair bit of energy
in the last chapter discussing the arguments that have been given by epis-
temologists for Gettier judgments. In this chapter, I will show that Kripke’s
antidescriptivist judgment about the Gödel Case is supported by argument
as well, showing, thereby, that diversity of intuition about the Gödel Case
is irrelevant to the question of whether we have good evidence for Kripke’s
judgment about that case (and so, to that extent, against descriptivism). I
will then briefly discuss the arguments that have been given by Keith Leh-
rer for his judgment about the Truetemp Case. (Briefly, because, in this case,
the arguments have already been re-presented by Cappelen.)
Are there any arguments in philosophy that rest on a brute intuition
about a case? If there were, that would weaken the empirical case for the
view that intuitions about cases are not treated as essential evidence con-
cerning what is true in them. However, if there are such arguments, they are
very difficult to come by. I do not know of a single clear example, and there
are in this book, and in Cappelen’s, a great many examples of arguments
that involve thought experiments and cases but do not rest on brute intu-
itions about them. However, in informal conversations with philosophers
who incline toward the myth of the intuitive, I sometimes hear favorite
examples of arguments that they allege rest on a brute intuition about a
case. I briefly discuss three of these favorites in the next section, ones that
seem to come up again and again: Frank Jackson’s knowledge argument,
John Searle’s Chinese room argument, and Donald Davidson’s Swampman
argument. However, as I will show, these examples are not what they are
purported to be; they are examples in which one does find supporting argu-
ments, right there in the surrounding text of the original presentations.
More Evidence for the Evidence 103
One question that I should revisit here, before presenting more examples
of philosophers arguing for their judgments about cases, is the question
of how large an evidential role intuition is supposed to play, according to
those who think it has any such role to play at all. Back in chapter 1, I
emphasized the fact that the negative xphi critique represents a significant
challenge to philosophy and its methods only if intuitions about cases are
104 Chapter 5
treated as essential evidence for what is true in the cases. Only then would
the variability results lead us to suspend judgment about what is true in
the cases.
But there is a range of metaphilosophical positions one might take
regarding the evidential role of intuition. One might think, for example,
that intuitions about a given case are treated as a part, but only a nones-
sential part, of the evidence for what is true in the case. Simply pointing
out that judgments about cases almost always receive backing arguments
does not rebut this view, since the view is obviously compatible with the
fact that there are backing arguments.
However, while the existence of backing arguments is compatible with
the philosophers presenting these arguments also taking intuitions to pro-
vide some evidential support for the relevant judgments, the real question
is whether there is any evidence that philosophers actually do endorse a
view to the effect that intuitions are nonessential evidence for the truth
about thought experiments and cases. If philosophers, in their reasoning
about cases, came right out and said, “Intuitions are some, though not all,
of the evidence for this judgment about this case,” then there would be
such grounds. But they seem never to say such a thing, not, anyway, in
first-order philosophizing. I am open to the possibility that such a view is
implicit in their practice, but is there any evidence that this possibility is
actual? My admittedly limited survey of arguments involving appeals to
thought experiments and cases suggests not. There is no reason to think
that the philosophers who present these arguments are implicitly suggest-
ing that the intuitiveness of a judgment about a given case is even part of
the evidence for the truth of the judgment.
In fact, since the philosophers in question usually argue for such judg-
ments, we have some reason to deny that there is any such implicit appeal
to intuitions. For, as Cappelen points out, if someone, philosopher or oth-
erwise, presents an argument for p, and says and does no more by way
of indicating why the rest of us should believe p, then we are justified in
believing that the argument encodes what its presenter takes to be the evi-
dence for p, not just some of it. When we add that the contexts of the pre-
sentations are philosophical papers and books, wherein we should expect
the philosopher who wants to convince us of p to explicitly present his or
her fullest and best case for p, the idea that there might nevertheless be
some implicit appeal to intuitions becomes extremely doubtful.
More Evidence for the Evidence 105
In any case, Kripke’s behavior, in the rest of Naming and Necessity, does
not explicitly suggest that he subscribes to the view that intuitions are evi-
dence for judgments about cases. He somewhat (but only somewhat) regu-
larly uses the word “intuition” and its cognates, but most of these uses are
most plausibly interpreted as indicating that Kripke takes the propositions
he characterizes in such terms as ones we might pretheoretically or uncritically
endorse. (This is related, clearly, to his use of “intuitive content” in the quo-
tation with which I began.) It would be strange to ascribe to Kripke a view
according to which the propositions we would pretheoretically or uncriti-
cally endorse are ones for which we have very heavy evidence. It is a strange
view to ascribe to anyone, but ascribing it to Kripke in particular would be
to forget the revolutionary character of many of the conclusions in Naming
and Necessity. At least some of what is revolutionary about Kripke’s book is
revolutionary in the way that good philosophy often is; that is, by showing
us that our preconceived, pretheoretical opinions are wrong.
Even if I am right that Kripke does not explicitly treat intuitions as evi-
dence, it could be that there are implicit appeals to intuitions in Kripke’s
arguments for his conclusions about, specifically, cases and thought experi-
ments (of which there are a great many in Naming and Necessity). I argued
in chapter 3 that, at least in the case of the Gödel Case, such an appeal is
not explicit; the passages in which Kripke presents the Gödel Case do not
explicitly say that intuitions about the case are meant to play some argu-
mentative role. This leaves it open that Kripke means his judgment about
the case to find implicit support in the fact that it is intuitive. However,
once again, the textual evidence speaks against this possibility, since what
one finds, when one bothers to look, is straightforward appeals to argu-
ment instead.
Immediately following his presentation of the Gödel Case, Kripke men-
tions three real-life (not hypothetical) cases that he takes to be similar to it.
He does this, he says, in order to discourage the impression that the circum-
stances imagined in his hypothetical case are far-fetched or even unusual.
Just as the imagined speakers in the Gödel Case do, we often use a name,
Kripke says, “on the basis of considerable misinformation” (Kripke 1980,
84). As examples, Kripke mentions uses of the names “Peano,” “Einstein,”
and “Columbus.” Kripke makes the same point with each example: in each,
it is clear that the misinformation that some speakers labor under when
using these names does not affect the referential facts; that is, speakers who,
More Evidence for the Evidence 107
for example, mistakenly believe that Peano discovered the so-called Peano
axioms (the discoverer of the axioms, Kripke reminds us, was Dedekind,
not Peano) do not, because of their mistake, fail to refer to Peano with their
uses of “Peano.” On the contrary, those uses refer to Peano, about whom
such speakers, because misinformed, are led to say false things (such as
that Peano discovered the Peano axioms). Similarly, although some peo-
ple believe that Einstein invented the atomic bomb and so associate “the
inventor of the atomic bomb” with “Einstein,” they still, and despite their
mistake, refer to Einstein with their uses of “Einstein.” Likewise for “Colum-
bus”: a good number of the “identifying descriptions” that get associated
with the name, “Columbus,” simply do not apply to Columbus, the man.
“Columbus was the first man to realize that the Earth is round” is false,
regardless of who utters it, and regardless, therefore, of which identifying
descriptions its utterer associates with “Columbus,” including, of course,
“the man who first realized that the Earth is round.”
I take Kripke’s comparison between the Gödel Case and the real-life cases
to be an argument for his judgment about the Gödel Case. If nothing else,
their placement in Naming and Necessity strongly suggests this. Coming, as
they do, immediately after the presentation of the Gödel Case, it would be
odd if Kripke did not take the facts about them to have some bearing on
what he says just before, in his discussion of the Gödel Case. Furthermore,
as I said above, Kripke explicitly says that the presentation of the real-life
cases is meant to counter a potential objection to his judgment about the
Gödel Case, namely the objection that the circumstances described in the
presentation of the Gödel Case are unusual in some way that would blunt
its antidescriptivist force. The idea here is that the similarities between the
hypothetical case and the real-life ones are so strong that, whatever the
correct characterization of the referential facts in the real-life cases happens
to be, an analogous characterization should apply to the hypothetical case.
There is also, and importantly, an assertion of—not an intuition about—
what the correct characterization in the real-life cases is: in the real-life
cases, speakers who are misinformed in various ways about the bearers of
the names they use nevertheless refer to those bearers with their uses. This
is strong evidence that the misinformed users of “Gödel,” in the hypotheti-
cal Gödel Case, refer to Gödel, not Schmidt.
It is not, however, intuitive evidence; it would be a mistake, I think,
to describe Kripke’s strategy, in his discussion of the real-life cases, as an
108 Chapter 5
associated with a name does in fact denote the bearer of that name (this,
Kripke points out, is true of “Gödel” and “the discoverer of incomplete-
ness” as we actually, in real life, use the terms—the Gödel Case, remember,
is “blatantly fictional”), the name refers to its bearer because the descrip-
tion denotes the bearer. If that were true, then, in the blatantly fictional
circumstances imagined in the Gödel Case, it should be true that “Gödel”
refers to Schmidt, not Gödel. That this is not so—that, in those imagined
circumstances, “Gödel” refers, still, to Gödel—is then argued for by appeal
to the real-life cases, in which we know already that certain descriptions
associated with certain names do not denote the bearers of those names.
Kripke’s argument for his judgment about the Gödel Case does not end
with the comparison to the real-life cases, however. In a footnote to his
discussion of the Gödel Case (Kripke 1980, 85, n. 36), Kripke notes that
if descriptivism were true, then certain sentences that express misconcep-
tions—his example is “Peano discovered the axioms of number theory”—
would express trivial truths instead. The point, cast as it is in terms of the
nature of the content of certain name-containing sentences, counts against
descriptivism about the meanings of names, but Kripke does not always
carefully separate the varieties of descriptivism, and it is clear, given that
the footnote’s context is a discussion focused on descriptivism about refer-
ence, that his intention is to say something against that variety as well.
To bring out its force as a point against descriptivism about reference, it
can be put in terms not of triviality, but of immunity from error. According
to descriptivism about reference, the reference of “Peano,” for speakers who
associate “the discoverer of the axioms of number theory” with the name,
will be whoever is uniquely identified by that description. That implies
that such speakers are immune to error when they utter Kripke’s example
sentence. Such speakers cannot but refer to the discoverer of the axioms of
number theory when using “Peano,” so they can only speak the truth if
they say “Peano was the discoverer of the axioms of number theory.”3 The
argument against this implication is straightforward: no speaker who utters
that sentence speaks truly; Dedekind, not Peano, discovered the relevant
axioms.
The application of the point to the Gödel Case is also straightforward: if
one is inclined, on descriptivist grounds, to say that the speakers in Kripke’s
imagined circumstances refer to Schmidt with their uses of “Gödel,” then
one will be forced to say that those speakers do not and cannot err when
110 Chapter 5
cases that are consonant with the picture, including, of course, the judg-
ment that “Gödel” refers to Gödel in the Gödel Case.
I hope, by now, that it is clear that Kripke’s judgment about the Gödel
Case is not something we are meant to accept on the grounds that it is
intuitive. Instead, Kripke offers a host of arguments for the judgment, rang-
ing from considerations that have a direct evidential bearing on the issue
(the facts about the real-life cases and the point about immunity from error,
e.g.) to considerations about meaning and reference that are more general
and pertain less directly to that specific judgment (the arguments against
meaning descriptivism and those in favor of the causal-historical picture of
reference, e.g.), but which count, nevertheless, as considerations that lend
support to its truth. The view that it is simply an intuition about the Gödel
Case that is the engine driving Kripke’s view of the case is a bad mistake. A
worse mistake is thinking that cross-cultural variability in intuitions about
the case, or ones similar to it, presents a significant challenge to either to
Kripke’s judgment about the case or to the methods he uses in arriving at
it. Instead, the judgment stands or falls with the quality of the arguments
marshaled in its defense.
The worst mistake is to think, as Machery, Mallon, Nichols, and Stich put
it in their most recent paper, that Kripke and other philosophers of language
think that “the correct method for determining the right theory of refer-
ence” is by “appeal to the intuitions of competent speakers about the refer-
ence of proper names (or other kinds of words) in actual and possible cases”
(Machery et al. 2013, 620; emphasis added). No. This is just plain wrong.
The methods—plural—for determining the right theory of reference involve
considering cases, of course, but the truth about these cases is argued for, not
intuited. No one could read Kripke’s presentation of the Gödel Case and its
surrounding discussion and sensibly believe that Kripke thinks otherwise.
such that his temperature beliefs do not count as knowledge. Purportedly, if we con-
sider this case, we will have the intuition that Mr. Truetemp does not know that it
is 104 degrees. Reliabilism’s inability to account for this intuition is supposed to be
reason to reject reliabilism. (Swain et al. 2008, 140)
It is very difficult to see how anyone could read this passage as anything
other than an attempt to provide reasons for the judgment that Mr. Tru-
etemp does not know that the temperature is 104 degrees. Unlike Cappelen,
I am, given my adoption of the no-theory theory of intuitions, prepared to
grant that the judgment is an intuition, or counts as such for some read-
ers, anyway. But it is very clear that, as far as Lehrer is concerned, this is
irrelevant, evidentially speaking. That Mr. Truetemp does not know is taken
to be the conclusion of an argument whose premise is that Mr. Truetemp
has “no idea why the thought [about the temperature] occurred to him or
that such thoughts are nearly always correct.” Lehrer even makes this argu-
mentative structure explicitly clear, by using the conclusion indicator word
“consequently.” This is not the behavior of a philosopher who means the
case for the judgment about the Truetemp Case to hang on whether that
judgment is intuitive.4
More Evidence for the Evidence 113
As Cappelen goes on to note, Lehrer not only offers this main argument,
contained in the quotation above, but also one meant to bolster it. Lehrer
modifies the original story so that the doctors responsible for implanting
the “tempucomp” do not themselves have any idea about its effects. This,
Lehrer insists, makes it even clearer that Mr. Truetemp does not know that
it is 104 degrees. Not only does Mr. Truetemp himself not have any inkling
about the reliability of his tempucomp-produced beliefs, but no one does,
and this, Lehrer insists, makes it impossible to truly describe Mr. Truetemp’s
belief as knowledge that the temperature is 104 degrees.5
But keep in mind that the point of drawing attention to the arguments
that philosophers give for their judgments about cases is a methodologi-
cal one. The question is not whether Lehrer’s arguments for his judgment
about the case are good arguments. The question is rather whether intu-
itions about the Truetemp Case play the evidential role that Swain et al.
(2008) assign to them. The answer to this fully empirical question is a
resounding “no.” There is nothing in Lehrer’s presentation of the case that
would in any way justify the view that his argument against reliabilism
turns on facts about what intuitions “we” do or do not have when consider-
ing the case. And that makes the sort of data about order effects reported in
Swain et al. 2008 simply irrelevant to the underlying epistemological issues
concerning the truth of reliabilism and the genuineness of the Truetemp
counterexample.
make a judgment about a case and say no more in its defense than that the
judgment is intuitive.
I think that appealing to brute intuitions about cases in philosophy must
be very rare. Many of the alleged examples of arguments that make such an
appeal are merely alleged. When one looks back at the original presenta-
tions of these arguments, one finds that the supposedly brute intuitions are
instead supported by reasons and argument.
In informal discussions with other philosophers about the role of intu-
itions in philosophy, I often ask for examples of arguments that they take
to appeal to brute intuitions about cases. I have yet to be presented with a
convincing one.
Fred, we come to discover, makes more color discriminations than the rest
of us. This is confirmed by a host of behavioral and physiological evi-
dence; Fred sees two distinct colors where we see just one, the one we call
“red.” We learn all the physical facts about Fred, but this will not teach us
everything there is to know about Fred. In particular, we will not find out
what it is like for Fred to see the extra color. If we were made, via surgery,
to have the type of visual system Fred has, we would learn, thereby, what
it is like to see the extra color. But this is something we would not know
before such an operation, despite knowing all the relevant physical facts
there are to know about Fred. So, physicalism is false.
More Evidence for the Evidence 115
5.2.3.2 The Chinese Room I often hear it claimed that John Searle’s
Chinese room argument (Searle 1980) rests on a brute intuition about a
thought experiment. But this, once again, is a bit of philosophical folklore;
Searle argues for his judgment about the Chinese room.
Again I will assume some familiarity with the details of the thought
experiment and with the larger argument in which the judgment figures,
the argument against the view that “an appropriately programmed com-
puter … can literally be said to understand and have other cognitive states”
(Searle 1980, 417; emphasis in original).
Many readers will recall that the crucial judgment that Searle makes
about the Chinese room case is that the man in the Chinese room does not
understand any Chinese. (The man, in the original 1980 presentation, is
Searle himself.) When the judgment first appears (1980, 418), Searle says
that its truth seems “quite obvious,” and he does not immediately back it up
with an argument (418). No doubt this has contributed to the impression
that Searle does not bother to argue for the judgment, but the fact is that
the argument is there; it simply appears a bit later.
The argument occurs in Searle’s counter to the “Systems Reply” and
begins with Searle pointing out that the answer to the question of what the
man has in the case of sentences of English (the man in the room, Searle
himself, is a native English speaker), but lacks in cases of sentences of Chi-
nese, is that he knows what the former, but not the latter, mean (418). That
might sound, to some ears, quite close to the claim that the man does not
understand any Chinese, but, in fact, the claims are different. The fact that
the man in the room does nothing that would recognizably count as learn-
ing the meanings of the Chinese symbols with which he is operating counts
in favor of the judgment that he does not understand those symbols. Searle
is clearly relying on this fact in arguing for the crucial judgment that the
man does not understand any Chinese.
Later, the argument is developed in the following way: Understanding
the sentences of a language involves knowing what they are about. The
man in the room understands English sentences with “hamburgers” and
“restaurants” in them partly in virtue of the fact that he knows that these
sentences are about hamburgers and restaurants. On the other hand, he
has no clue what the Chinese sentences with which he deals are about. For
example, he does not know that Chinese sentences with the Chinese word
for “hamburgers” in them are about hamburgers. He does not know, even,
More Evidence for the Evidence 117
which word of Chinese refers to hamburgers; indeed he does not know what
any of the Chinese words refer to. That implies, according to Searle, that
the man in the room does not understand the Chinese sentences. In other
words, the man’s lack of knowledge of various semantic facts about the Chi-
nese sentences is meant to ground an argument for the judgment that the
man does not understand those sentences. (See Searle 1980, 419.)
This argument for Searle’s judgment about the Chinese room case is
bolstered by an argument to the effect that the operations that the man
performs with the Chinese sentences (which, the reader will recall, Searle
repeatedly describes as the mere “manipulation of formal symbols”) are not
such that they would produce knowledge of the semantic facts necessary
for literal understanding. The man can perform those operations without
knowing the semantic facts about the sentences. The conclusion that the
man can perform those operations without understanding the sentences is
supposed by Searle to be just that—a conclusion from reasoning about the
conditions on literal understanding, not a brute intuition about what is
true in the Chinese room case.11
Many readers will recall that Davidson judges that Swampman does not mean
anything by the sounds it makes and does not have any thoughts (444). There is
a sense in which it is true that Davidson offers this judgment without offer-
ing very much by way argumentative backing. (Here, once more, I assume
that the reader is familiar, at least in outline, with some of the issues raised
in Davidson 1987.) But this is because, as is very clear from the surrounding
discussion, Davidson is taking it for granted that externalism about mental
content (which, Davidson thinks, implies that his judgment about Swamp-
man is true) is true. For Davidson, Swampman is more an illustration of the
118 Chapter 5
To reinforce the point, I will comment on one of the more recent exam-
ples of an esoteric, unusual case that Weinberg cites (see Weinberg 2007,
321), a case from Jennifer Lackey’s (2008) work on testimony.
Lackey maintains that a certain widely held theory of testimony gets
the epistemology of testimony almost entirely wrong. This widely held
theory assigns a central role to a speaker’s knowledge in explaining how a
hearer comes to know, via a speaker’s testimony, that something is the case.
Among the views proffered by this theory are views about the necessary
conditions on a hearer acquiring knowledge by testimony. One such view is
that a hearer cannot come to know that p from a speaker’s statement that p
unless the speaker herself knows that p. Lackey argues against this, and vari-
ous allied views, with a variety of ingenious counterexamples. Weinberg
complains about one counterexample in particular, the case of a “double-
lesioned testifier” (Weinberg’s description) that Lackey calls Consistent Liar.
Lackey’s presentation is somewhat long. Here is a truncated version:
Lackey offers a judgment about the case: Henry knows, on the basis of Ber-
tha’s testimony, that there was a deer on the trail. If she is right, then the
view that a hearer can come to know p via a speaker’s testimony only if the
speaker knows that p is false, since, in the case, Bertha does not believe, let
alone know, that there was a deer on the trail.
According to Weinberg, since her case is especially outlandish, Lackey’s
judgment about it is an example of an especially suspicious use of an intu-
ition as evidence. However, Lackey does not treat her intuition about the
case as evidence that Henry knows, via Bertha’s testimony, that there was
a deer on the trail. Instead, just as we have seen with respect to every other
case examined in this book, Lackey offers arguments—lots of them.
She points out, for example, that although Bertha is a highly unreliable
believer, her reporting is more reliable than that of many average testifiers.
Related to this is the fact that Bertha’s reportings (though not her believ-
ings) satisfy conditions that many epistemologists take to be prime indica-
tors of reliability; her report that there was a deer on the trail is sensitive in
Nozick’s sense, and safe in Sosa’s sense, for example. Furthermore, Lackey
argues on a variety of grounds that Bertha’s report is a genuine instance of
testimony, thereby blocking the objection that while Henry knows, it is not
via Bertha’s testimony. In addition, Lackey argues that it is no accident that
Bertha’s reports of her animal sightings are accurate; this was a deliberate
effect of the work the surgeon performed. Hence, says Lackey, the thought
that Henry might have a “Gettiered” justified true belief that there was a
deer on the path is a mistake. She adds to all of this that Henry has no rea-
son to doubt Bertha’s accuracy or sincerity. Indeed, he has plenty of good
inductive evidence that Bertha is an excellent source of information about
which wild animals have appeared where.
In short, Lackey concludes on a variety of grounds that we have plenty
of reason to accept, and no reason to deny, that Henry knows via Bertha’s
testimony that there was a deer on the path. That conclusion is the clear
product of inference and argument. Intuitions and intuitiveness are, once
again, irrelevant.
Its outlandishness perhaps does show that we should not treat intuitions
about the case as evidence for the truth of the judgment that Henry knows;
I want to be clear that I am not denying that. The point is an interpretive
one concerning Lackey’s method: Lackey very clearly does not intend that
we treat intuitions about her case as evidence for her judgment about it.
122 Chapter 5
Instead, she means for us to agree, on the basis of the arguments she gives,
that that judgment is true. A survey showing us that people’s intuitions
about the case vary along truth-irrelevant dimensions would not be cause
for alarm on Lackey’s part, since data of that kind would say nothing about
the quality of Lackey’s arguments for her judgment about the case.
6.1 Introduction
Many philosophers think that the negative xphi critique can and should
be resisted but do not think that xphiles have a mistaken account of philo-
sophical method. These philosophers think that intuitions do play a cru-
cial argumentative or evidential role in philosophical arguments, but they
think that this role can be defended against the intuitional diversity argu-
ments of negative xphiles. Even before xphi broke on to the scene, one
could find plenty of examples of philosophers claiming that intuitions play
a central role in their methods.
None of this, however, counts against my own claim, namely that one
very rarely finds appeals to intuitions as evidence in philosophy. This
claim, which I have been defending over the course of the last two chap-
ters, concerns first-order philosophy; that is, it concerns the ways in which
philosophizing about language and knowledge (and morality, free will, etc.)
is carried out. Second-order philosophy, or the philosophy of philosophy, is
another matter entirely. When philosophers don their methodologist hats
and start philosophizing about the nature of philosophy itself, they rou-
tinely mention intuitions and carve out a significant place for them in their
accounts of philosophical method. As an extreme example of the sort of
thing I have in mind, here is Stephen Neale on method in the philosophy
of language:
Our intuitive judgments about what A meant, said, and implied, and judgments
about whether what A said was true or false in specified situations constitute the
primary data for a theory of interpretation, the data it is the theory’s business to
explain. (Neale 2004, 79)
130 Chapter 6
This is extreme, since many who would say that consulting intuitions
about language is an important part of the methodology of the philosophy
of language would not go so far as to say that intuitions “constitute the
primary data” for a philosophical theory of language or that they are “the
data it is the theory’s business to explain.” These claims go far beyond typi-
cal second-order philosophical claims, which tend to limit intuitions to an
evidential role. Read literally, Neale’s account implies that philosophical
theories about language are theories of intuitions about what is meant, said,
and implied, instead of theories about, more simply, what is meant, said,
and implied. Perhaps Neale is taking his cue from linguistics, where, accord-
ing to some linguists, theories of grammar are theories of speakers’ “gram-
maticality judgments,” that is, intuitions concerning the grammaticality of
sentences of the speaker’s language. In any case, what one usually hears in
defense of the importance of intuitions in philosophy is, as I say, limited to
the claim that they are an important source of evidence.
There are plenty of examples of this more modest 2nd-order philosophi-
cal stance on intuitions. Joel Pust has written an entire book (Pust 2000) in
which he defends philosophy’s alleged use of intuitions as evidence. From
the perspective of the book you are now reading, Pust’s defense is unneces-
sary, since (first-order) philosophy does not treat intuitions as evidence.
However, to give a flavor for more moderate second-order pronouncements
about the evidential role of intuitions, here is a quote from Pust:
My aim in this first chapter is to demonstrate that much contemporary philosophi-
cal theorizing proceeds as though intuitions of various kinds constitute evidence
for and against the truth of a philosophical theory. Making clear just how widely
intuitions are treated as evidence in contemporary philosophy demonstrates the im-
portance of my discussion, in later chapters, of the normative question of whether
intuitions really ought to be so treated. (Pust 2000, 1)
Pust is not alone in describing the role of intuitions this way. The view that
intuitions play an evidential role in philosophy is very widespread. Here is
a quote from a recent paper in Mind by Sara-Ann Malmgren:
In brief outline, it [the “method of cases” in philosophy] has the following structure:
the hypothesis or theory that is under evaluation states or entails some modal claim
(typically a necessary bi-conditional or one-way implication) and in a thought ex-
periment we check that modal claim against our intuitive verdict on an imaginary
problem case. If the claim conflicts with our intuitive verdict, this is treated as strong
evidence against the theory—indeed the theory may be abandoned as a result. We
say that we found a counter-example to it. If not, that is treated as at least some evi-
Other Replies to Xphi 131
dence in support of the theory. We say that it accommodates our intuitions about the
case. (Malmgren 2011, 264)1
These quotations are just a small sample. One can find this second-order
view repeated again and again in self-consciously methodological work in
philosophy. The view is mistaken; in first-order philosophy, one simply
does not find intuitions playing the role assigned to them by second-order
pronouncements such as Neale’s, Pust’s, Malmgren’s, or Neta’s. But the mis-
take cries out for an explanation. Why have so many philosophers been
led astray?
This interesting question will not be pursued in much depth here. To
a large extent, I think methodologists have been misled by the rhetori-
cal style of first-order analytic philosophy, a style that includes many uses
of “intuitive” and its cognates. It cannot be denied that philosophers will
often say things such as “Intuitively p …” or “It is counterintuitive that q
…” Perhaps methodologists like Neale, Pust, Malmgren, and Neta, when
formulating their second-order views, think they are simply taking first-
order philosophy at its word. But it is a mistake to elevate “intuition talk”
into a substantial theory of philosophical method that takes that method
to treat intuitions as evidence. For one thing, as I have pointed out in sev-
eral places, there are plenty of examples of intuition talk that cannot be
plausibly read in an intuitions-as-evidence way.
Another part of the explanation, I believe, is that methodologists are not
sensitive to the state/content ambiguity of “intuition.” For, as I argued in
chapter 2, there is an uncontroversial sense in which intuitions are treated
as evidence in philosophy; their contents are treated as evidence. Indeed,
some of the claims in the quotations above would be merely mislead-
ing (instead of false) if occurrences of “intuition” and related terms were
132 Chapter 6
even to whether there is any evidence at all, for the truth of the proposi-
tion. To take an example, suppose that one thinks “Gödel” refers to Gödel
in Kripke’s Gödel Case but that the only evidence that this is so is that it is
intuitive that it does. Data showing that it is not intuitive after all, or that
we do not now know whether it is intuitive, will then force one to withdraw
the claim that “Gödel” refers to Gödel in the Gödel Case.
Even if one takes intuitions as evidence, one can wiggle a little here,
since one can object that the data collected so far do not show that we need
to withdraw any of the judgments about thought experiments we happen
to have made. The two major replies to the negative xphi critique that I
will examine in the next sections of this chapter wiggle in just this way, but
before getting to them I want to emphasize that there is a sense in which
they misgauge the depth of the challenge. One point on which positive and
negative xphiles agree is that both camps view the question of whether p is
intuitive as a straightforwardly empirical question. And what else could it
be, really? But if that is right, and one holds the methodological view that
intuitions are evidence in philosophy, then it seems that one is forced to
admit that the question of whether there is evidence for, for example, the
truth of the judgments philosophers make about thought experiments is
itself a straightforwardly empirical question, one to be answered by intu-
ition surveys of the type xphiles have already been running. How else does
one find out whether people intuitively judge that p except by asking them?
In the next two sections I examine the two major replies that have been
made against the intuitional diversity arguments of negative xphiles. These
replies are sometimes billed as defenses of “armchair philosophy” but, as I
have been arguing here, this is false advertising. The proponents of these
replies tend to agree with their negative xphile targets that intuitions are
evidence, and this, for the reasons laid out above, is in effect to agree that
philosophers must leave their armchairs and take data from intuition sur-
veys very seriously indeed. I think this ought to be sufficient indictment
of the replies; they do not accomplish what they set out to accomplish.
Furthermore, there is a whiff of desperation in them; both replies strike me
as very weak, and this, I suspect, is because their proponents are hamstrung
by their commitment to an intuitions-as-evidence methodology. The weak-
ness of the major replies to the negative xphile critique is another example
of the bad consequences of adopting that methodological view. Once one
accepts the methodology, the defenses against the critique are bound to be
134 Chapter 6
weak. However, the popularity of the major replies, the fact that they have
been vigorously made by famous philosophers, and the interesting side
issues they raise, all recommend more extensive discussion.
Most xphi intuition surveys have used nonphilosopher subject groups, uni-
versity undergraduates, mainly. A natural objection to some of the conclu-
sions that have been drawn from data about these groups’ intuitions is that
xphiles are surveying the wrong subjects. The intuitions of nonphilosophers
should not count for much; instead, it is the intuitions of philosophers—the
experts—that matter most. Most philosophers are trained to do philosophy
and spend a lot of their time engaged in the practice. Surely, this training
and immersion makes them better at the practice than your average uni-
versity undergraduate. Thought experimentation and the consideration of
hypothetical cases—including the making of intuitive judgments, on some
accounts—are part of this practice. Hence, we have prima facie good reason
to expect that the intuitive judgments of philosophers are more reliable
and trustworthy than those of nonphilosophers.
This reply, in at least roughly this form, is made by Michael Devitt (2011,
2012), Frank Jackson (2011), Steven Hales (2006), and Kirk Ludwig (2007).
Williamson (2007, 2011), too, is on paper describing himself as a propo-
nent of the reply (though for reasons that will emerge, it is not entirely clear
that he really is). Here is a representative expression of the expertise reply
from Devitt (2011), who is discussing semantic intuitions in particular:
Still, are these referential intuitions likely to be right? I think we need to be cautious
in accepting them: semantics is notoriously hard and the folk are a long way from
being experts. Still it does seem to me that their intuitions about “simple” situa-
tions are likely to be right. This having been said, we should prefer the intuitions
of semanticists, usually philosophers, because they are much more expert (which is
not to say, very expert!). Just as the intuitions of paleontologists, physicists, and psy-
chologists in their respective domains are likely to be better than those of the folk,
so too the intuitions of the semanticists. (Devitt 2011, 426)
that none of its methods are any good at all—and a thoroughgoing skepti-
cism about philosophy as a whole is neither suggested by the xphile results,
nor recommended by negative xphiles themselves3—we can be reasonably
certain that at least some of its methods are fruitful and lead to philosophi-
cal knowledge.
However, the negative xphile critique targets what negative xphiles take
to be a specific philosophical method, namely that of appealing to intu-
itions to justify claims about what is true or false in philosophical cases and
thought experiments. (For brevity’s sake, I will henceforth refer to this class
of claims simply as judgments about cases. Examples include Gettier judg-
ments and Gödel-Case judgments, among many others.) I have argued in
previous chapters that this alleged method is merely alleged, and that, as a
matter of how first-order philosophizing actually transpires, there simply are
no appeals to intuitions to justify judgments about cases; argument plays
that role instead. But this is not something about which proponents of the
expertise reply would necessarily agree. Indeed, the most natural way to
understand the reply is to take it as agreeing that intuitions are used in the
way xphiles presume, namely as evidence for judgments about cases, but as
insisting that it is the intuitions of the experts that are the proper evidential
base, not those of the hoi polloi. Interpreted in this way, the expertise reply is
just as wrongheaded as the critique to which it replies: both wrongly assume
that intuitions are treated as evidence for certain sorts of philosophical judg-
ments. Their disagreement is simply over whose intuitions count.
As I understand it, the expertise reply not only agrees with negative
xphiles that intuitions are used as evidence for judgments about cases, they
also agree that the intuitive evidence is all the evidence there is for such judg-
ments. At the very least, the expertise reply, as I am understanding it here,
takes the intuitive evidence to be essential, where a kind of evidence is essen-
tial just in case the claims justified on its basis would not be justified in its
absence. Furthermore, as I understand them, proponents of the expertise
reply take the intuitive evidence for judgments about cases to be psycho-
logical facts concerning who intuits what; that is, they do not merely mean
to be asserting that, sometimes, the contents of expert intuitions qualify
as evidence for such judgments. They mean, instead, to be asserting that
psychological facts to the effect that the experts intuit p about a case—intu-
itions in the “state” sense (those of the experts)—are the only evidence (or
are, at least, essential evidence) that p is true in the case.
136 Chapter 6
Proponents are rarely explicit about these assumptions, but I think taking
them as implicit is the most natural way to understand the expertise reply.
After all, proponents of the reply do not complain that negative xphiles
have a completely inaccurate picture of the methodology of thought exper-
imentation; usually, they seem content to point out that xphiles have been
surveying the wrong subjects and leave the matter there. This suggests that
they think negative xphiles are mostly right, except about whom to survey.
Suppose, for now, that this is correct. A proponent of the expertise reply
implicitly makes these assumptions and hence agrees, to a large extent,
with the presuppositions of the negative xphile critique, but claims that the
wrong subjects have been surveyed. Can the matter be left there? Is this a
sufficient reply to the critique?
The xphi data, though it concerns the intuitions of nonphilosophers,
raises a question about philosophers’ intuitions: Is there reason to believe
that philosophers are less prone to making intuitive judgments about
philosophical cases that pattern along truth-irrelevant lines? Are philoso-
phers’ intuitions less prone to order-effects, for example, or to the effects of
culture? A defender of the expertise reply might presume that this is part
of what training in philosophy provides—an at least partial inoculation
against biases that affect the less-than-expert subjects surveyed thus far.
But where does the burden of proof lie? Who is responsible for showing
that the philosophical experts do—or do not—exhibit the same worrisome
patterns in their intuitive judgments about cases? I think the burden of proof,
when the expertise reply is understood as I have described it above, clearly
lies with the proponents of the reply; we need to be shown, via empirical
testing, that philosophers’ intuitive judgments will display less sensitivity to
the truth-irrelevant factors discovered to influence nonphilosophers’ intui-
tive judgments.4 It is irresponsible of the proponent of the expertise reply
simply to accuse xphiles of surveying the wrong subjects and leave the mat-
ter there. If the wrong subjects have been surveyed, then the proper thing
to do, if one thinks that what and how certain people intuit about a case is
essential evidence concerning what is true in the case, is to survey the right
ones. I think that those who have endorsed the expertise reply are obliged to
make good on it by empirically demonstrating that philosophers’ intuitions
are less sensitive to truth-irrelevant factors than the subjects surveyed thus
far. To date, none of the proponents of the expertise reply I cited above have
provided it with the experimental support it requires.5
Other Replies to Xphi 137
presumption that philosophers will be less prone to biases and effects into
knowledge that they are. And there is such a reason. After all, for the pro-
ponents of the reply, the claim that philosophers are less prone to biases
and effects is a central and important methodological claim. For they think
that, in doing philosophy, and, specifically, in determining what is true in
thought experiments and cases, we can, for the most part, rely on and trust
the intuitions of the experts—partly in virtue of the fact the experts are less
prone to intuiting along truth-irrelevant lines. But if we do not know that the
experts are less prone, then, to that extent, we do not know that we can
trust expert intuitions. And if we do not know that we can trust expert intu-
itions, then it seems clear that the methodology that recommends trust-
ing expert intuitions has not been properly defended. Hence, the expertise
reply has not been adequately made. Doing so requires demonstrating, and
thereby knowing, that the worrisome patterns of intuitions discovered in
nonphilosophical subject groups do not reappear in philosophical ones.
Furthermore, and fairly obviously, that demonstration is, or rather will be,
an empirical one: it will involve surveying the experts to discover their
intuitions.
However, suppose one thinks, as I do, that intuitions are methodologi-
cally unimportant in philosophy. Then there is no reason to worry about
whether expert philosophical intuitions are subject to effects of culture, for
example. One can admit that they might be, or even that they are in fact,
without that impugning philosophical method as one takes it to be. Strange
patterns in intuitions can be safely ignored, if one thinks intuitions play no
role in philosophical methods. But that option, of just ignoring the ques-
tion of what sorts of truth-irrelevant factors might influence even expert
intuitions, is simply not available to the proponents of the expertise reply.
As Weinberg et al. (2010) claim, the burden of proof—empirical proof—is
on the proponent of the expertise reply to show that truth-irrelevant fac-
tors have much less of an effect on the intuitions of philosophical experts
than they do on others.7
Proponents of the expertise reply are fond of analogies to other dis-
ciplines. We do not take the judgments of nonscientists (Hales 2006) or
nondoctors (Jackson 2011) or nonmathematicians (Ludwig 2007) or non-
lawyers (Williamson 2007) or nonpaleontologists, nonphysicists, and non-
psychologists (Devitt 2011) to count for much of anything in determining
the truth about science, medicine, mathematics, and so on. So why take
Other Replies to Xphi 139
in the “state” sense—against a challenge to that method. The view that phi-
losophers’ judgments about cases are better than those of nonphilosophers’
is not a methodological view; it implies nothing at all about the methods
that philosophers employ in arriving at these judgments, except, rather
trivially, that philosophical methods are best. We should trust the philo-
sophical experts because they are more likely to get things right in their
area of expertise, however it is that they go about trying to get them right.
But that is not a recommendation about how to do philosophy; it is just a
comment about the effect of knowing more philosophy. Knowing more
philosophy makes one a better judge of philosophical matters, including
what is true in philosophical cases. Williamson is surely right about that,
and right that none of the xphi data even come close to touching its truth
or our justification for believing it.
Alexander and Weinberg are never explicit about how the intuition-free,
nonpsychologistic version of the negative xphile critique is supposed to go,
but there is a hint in their comment that “at the present time philosophers
Other Replies to Xphi 143
may just not know what their evidence really is.” I take them to mean that,
somehow, the discovered diversity in judgments about cases shows that
philosophers are not justified in believing that their own judgments about
cases are correct. So, philosophers do not know that the contents of these
judgments are evidence.
But the question is: how is the discovered diversity in judgment sup-
posed to show this? If all there were to say in defense of a judgment about
a case is that it is intuitive, or intuitive to the experts, then diversity in
judgment about a case along truth-irrelevant lines would be a big prob-
lem. I have already admitted this in my discussion of the weakness of the
expertise reply. But if a judgment about a case is made for reasons, if the
judgment can be given argumentative support, then why should diversity
in judgments about the case, when those reasons and that support is not
clearly in play, or registered by the judges, matter in the slightest?
So far as I can tell, the only available answer is this: the existence of
diversity in judgments about a case, and along truth-irrelevant lines, will
always defeat whatever justification might be given for any particular judg-
ment about that case. This answer is clearly wrong, however. It is not true
in general that diversity in people’s judgments with respect to p defeats
one’s justification for judging that p. Whether it does depends, of course,
on the character and quality of one’s justification. Diversity in judgment
along truth-irrelevant lines is not an epistemic defeater no matter what. If
the intuition-free, nonpsychologistic version of the negative xphile critique
depends on insisting that it is, then the critique is badly misguided.
Many varieties of judgment display diversity along truth-irrelevant lines.
Think, for example, of the ways in which people’s political judgments vary
with respect to all sorts of demographics that do not matter to the truth
of the judgments. No one would seriously maintain that diversity of that
sort defeats every justification that has been or might be given for political
judgments. Sometimes people disagree politically, not for any good reason,
but because they are being pushed around, perhaps in subtle, difficult-to-
discern ways, by truth-irrelevant factors. That unfortunate fact of life sim-
ply has no bearing on whether there is any compelling justification for one
side of the disagreement over the other. In the United States, for example,
whether one supports stricter gun control laws depends to some disturb-
ingly large extent on one’s gender, one’s income, and one’s political party
(to pick three truth-irrelevant, judgment-affecting demographics among a
144 Chapter 6
large host of them). (See Pew Research Center 2011.) To think that this
implies that justifications for or against stricter gun laws are one and all
called into question is a downright bizarre epistemological view.
The gun-control example is just one of an enormous number of exam-
ples of judgments we know to vary along truth-irrelevant lines. But every
such example leaves open the question of whether there are good, truth-
relevant reasons to hold that the judgment is true (or false). Xphiles appear
to have shown that, in some circumstances, judgments about philosophical
cases vary along truth-irrelevant lines. But if such judgments are at least
sometimes made for reasons, if they are not just intuitions, then the mere
fact of that sort of variability need not pose an epistemic threat. Gettier’s
reasons for his judgment that his Smith character fails to know, in both of
the cases he describes in Gettier 1963, might be sufficient justification for
that judgment, regardless of the kind or amount of intuitional diversity we
find when presenting the cases to various subject groups, and even if we
find it in subject groups composed of philosophical experts. If Gettier’s rea-
sons are good enough, then philosophers can “know what their evidence
really is.” Their evidence really is that there are genuine counterexamples
to the JTB theory of knowledge, and some philosophers, those cognizant of
Gettier’s reasons, know it.
It took skill, ingenuity, reflection, and careful reasoning to construct,
and argue for, counterexamples to a theory of knowledge that had reigned
in epistemology for thousands of years. So, of course, Gettier’s refutation
involved philosophical expertise. Gettier is in the business of arguing for
philosophical judgments, even those that qualify as intuitive on some
understandings of that troublesome term. Being in that line of work, he
is better placed than nonphilosophers to make justified judgments about
cases. There is no empirical evidence that shows or suggests otherwise.
What holds for Gettier holds, more or less, for the profession as a whole.
We philosophers are better placed than those who lack philosophical train-
ing to make justified judgments about philosophical cases. This does not
imply, however, that psychological facts about what intuitions philoso-
phers have are methodologically or evidentially significant. The problem
with the expertise reply is not that it appeals to expertise; philosophers
really are better at philosophy than nonphilosophers. The problem with
the reply is instead that, just like the critique to which it replies, it makes a
serious mistake about intuitions and their role in philosophy.
Other Replies to Xphi 145
Notice, however, that to view the xphi critique as failing in the way I just
described involves, in part, rejecting the picture of philosophy as treating
intuitions (in the “state” sense) as evidence. Unfortunately, many philoso-
phers are loath to give this up and are now scrambling desperately to save
the picture, even in the face of the ever-mounting empirical evidence that
people’s intuitions about philosophical cases vary along all sorts of truth-
irrelevant dimensions. The expertise defense is just one desperate attempt
among several. Another influential reply to the negative xphile critique—
one, again, that attempts to preserve the view that intuitions are treated as
evidence in philosophy—is to claim that the results of xphi surveys do not
necessarily indicate genuine disagreement. For example, perhaps people who
appear to disagree over Gettier cases presented in xphi surveys understand
the terms in which the questions are posed differently. And perhaps these
differences in interpretation, as opposed to genuine disagreement over one
and the same proposition, produce the appearance of disagreement in the
survey results.
Ernest Sosa (2007, 2010) is the main representative of this style of reply
to the negative xphile critique, but Frank Jackson (2011) and William Lycan
(2006) have also voiced support. To have a name, I will call it the mul-
tiple concepts reply, since a standard way of understanding the category of a
merely apparent disagreement is to see it as involving speakers who associate
different concepts with the same terms. One English speaker might appear
to disagree with another over the truth of “There is a bank in town,” for
example. But if one of them means ground near river while the other means
146 Chapter 6
This is all that the relevant xphi surveys have shown, according to propo-
nents of the multiple concepts reply—there exist apparent disagreements
over philosophical cases. They then quickly add that it is possible that these
apparent disagreements are merely apparent, owing to subjects associating
different concepts with the same terms. If so—if the disagreements are, in
this way, merely apparent—then the threat to armchair appeals to intu-
itions as evidence is defused. For example, for all that Weinberg et al.’s
(2001) cross-cultural study on Gettier intuitions shows, Westerners are right
about the Western concept of knowledge—it does not apply to agents in
Gettier cases—while East Asians are right about their different, East Asian
concept of knowledge—their concept does apply to agents in Gettier cases.
Despite the fact that the reply is very general (implausibly general, I will
soon argue), the cross-cultural differences in Gettier intuitions reported in
Weinberg et al. 2001 are the main target for proponents of the multiple
concepts reply. Here is Frank Jackson making the reply against these results
in particular:
We should be open to the possibility that different groups of people have different
concepts of knowledge, in the sense that the categorization the various groups effect
using the word “knowledge” may vary. There isn’t a law stating how people have to
use the term. Perhaps, therefore, at least some holdouts [that is, those who do not
share the Western Gettier intuition] have a different concept of knowledge; indeed,
it may be that their concept of knowledge is precisely true justified belief. (Jackson
2011, 469)
And here is Ernest Sosa making the same point while emphasizing that he
thinks that defenders of the use of intuitions as evidence need not assert
that there are any actual conceptual differences between survey respon-
dents; according to Sosa, the defense succeeds so long as the possibility that
there are such differences has not been ruled out:
In any case, here is the more important point: the response to the X-Phi challenge
does not need to commit to an actual divergence in meaning. Just remember: it is
the attackers who allege a real disagreement, one that is not merely verbal. But this
is an inference they are drawing from certain verbal reports. And these verbal reports
by rushers-by on a street corner are hard to take seriously as expressive of considered
views with full enough understanding of the issues under dispute, including those
concerning Gettier examples. Nevertheless opponents of the armchair infer from
those reports that there is real disagreement. Given the non-dismissible possibility
that the disagreement is just verbal, then, we need to be given reason to rule out that
possibility. (Sosa 2010, 422)10
148 Chapter 6
The question for the milder version of the multiple concepts reply
endorsed by Sosa and Jackson is this: just what, exactly, is required to “rule
out” the possibility that apparent disagreements are merely apparent? Must
there be conclusive reasons, ones that entail that the apparent disagreements
are also genuine? “Ruling out” suggests that this is the standard, but that
seems unreasonably strict. After all, we are not, in ordinary conversational
settings, always in a position to rule out, in this strong sense, the possibil-
ity that our interlocutors, when apparently disagreeing with us, are merely
apparently doing so. I assertively utter a sentence, S, and you seem to dis-
sent, saying, “No, not-S.” You have genuinely disagreed with me only if we
agree about what claim S expresses (or, better, about what claim I have used
S to express). But, typically, neither you nor I make any attempt at deter-
mining that there is this (metalinguistic) agreement between us. Besides,
any attempt at reaching the supposedly required agreement will involve
trading more sentences (perhaps, at this stage, ones about other sentences)
and the possibility that our agreement or disagreement at this next level is
merely apparent is just as much a possibility as is the possibility that the
disagreement with which we started is merely apparent.
In other words, the view that we know or justifiably believe that we
are genuinely disagreeing with our interlocutors only if we can rule out
the possibility that we are instead merely apparently disagreeing leads to
skepticism about our knowledge of, or justified belief in, the existence of
genuine disagreements. This skepticism is implausible—sometimes, surely,
we do know, or have good reason for supposing, that a disagreement is
genuine—and so we should reject the condition that requires that we, in
some conclusive fashion, rule out the possibility that the disagreements
are merely apparent. Sosa and Jackson are of course correct that it is pos-
sible that East Asians employ a different “concept of knowledge” and hence
say something different from Westerners when they apply, or withhold the
application of, knowledge ascriptions. But they are wrong to think that fail-
ing to conclusively rule out this possibility shows that xphiles do not know
or are unjustified in claiming that there are genuine disagreements between
East Asians and Westerners over Gettier cases.
Xphiles see this clearly. In essence, this is the response to the Sosa–Jack-
son variety of the multiple concepts reply that negative xphiles have made.
Alexander and Weinberg, for example, point out that the multiple concepts
reply, to the extent that it casts doubt on whether xphiles have uncovered
Other Replies to Xphi 151
Are there special reasons not to do so? I have already expressed my dis-
satisfaction with Lycan’s view, according to which the special reason is sim-
ply the existence of the apparent disagreement itself. That cannot be right,
since every instance of genuine disagreement is also an instance of appar-
ent disagreement. Something more is needed to substantiate the charge
that the apparent disagreement might, in some sense of “might” that goes
beyond bare logical possibility, be merely apparent. Lycan offers nothing
that will do the trick; however, neither do Sosa or Jackson.
Sosa is at pains to point out that his version of the multiple concepts
reply “does not need to commit to an actual divergence in meaning” (Sosa
2010, 422). He is adamant that he need not say that East Asians and West-
erners attach different meanings to “knowledge,” for example. But, while
that is true, his version of the multiple concepts reply does need to offer
some reason for thinking that the default justification we have for believing
that, in most cases, apparent disagreements are genuine is overridden in
this case. That can fall short of a reason for thinking that there is an actual
divergence in meaning. However, the mere possibility of a merely apparent
disagreement is insufficient in this regard, for even in those cases in which
we have very strong evidence that a disagreement is genuine, it is of course
still possible, even given the evidence, that subtle conceptual differences are
producing the relevant patterns of assent and dissent.
Similarly, Jackson, speaking of “knowledge,” is right when he says,
“there isn’t a law stating how people have to use the term” (Jackson 2011,
469). True, there is no such law, but there is, or so I have argued, a presump-
tion to the effect that speakers of a language communicate with each other
when speaking their shared language. That presumption implies that we
can reasonably take what looks like disagreement between two speakers
to be genuine. Jackson, like Sosa, offers nothing that might counter this
presumption beyond the fact that it is possible that apparent disagreements
between people over Gettier cases are merely apparent. But every appar-
ent disagreement is, possibly, merely apparent. Unless we accept that a
general skepticism about our knowledge of genuine disagreement follows
from this possibility, and hence that we can never know, about any putative
genuine disagreement that it is genuine, we will be unmoved by Jackson’s
point about the absence of laws governing how people must use the term
“knowledge.” The question is not whether it is possible that some people
use “knowledge” differently from other people. Instead, the question is
Other Replies to Xphi 153
whether our default justification for believing that this possibility does not
obtain is defeated or overridden in the specific case in which Jackson is
interested, namely the case of apparent cultural differences in judgments
about Gettier cases.
So it appears that, just as was true of the expertise reply, there is, in
the case of the multiple concepts reply, an issue to do with the burden of
proof. In the former case, I agreed with xphiles that the burden of proof was
on proponents of the expertise reply to show that philosophical experts
are less prone to the biases and effects that xphiles have found to affect
the intuitions of nonphilosopher folk. In the case of the multiple concepts
reply, I again agree with xphiles that, while it is possible that different sub-
ject groups are operating with different concepts and hence that what look
like genuine disagreements between them might be explicable in a way that
reveals this as mere appearance, this possibility, as things now stand, is idle.
Without some reason to believe otherwise, we—xphiles and the rest of us—
are justified in believing that there are genuine disagreements between, for
example, different cultural groups over philosophical cases. The burden of
proof is on proponents of the multiple concepts reply to show us that the
apparent disagreements are special in some way that reveals that it would
be too quick to take these disagreements at face value—to take them, that
is, as genuine.
negative xphile critique and reveals that the expertise and multiple con-
cepts replies are unnecessary. If intuitions are not treated as evidence, then
there is no need to fret about the sensitivity of intuition to truth-irrelevant
factors; nor is there any need to insist that philosophers intuit better or that
apparent disagreement in intuitions may be explicable in terms of concep-
tual differences. Both the critique and the major replies to it are predicated
on a mistake, the mistake of taking it to be part and parcel of philosophy to
rely on intuitions as evidence.
Conclusion: Armchairs versus Lab Coats?
It is now common to hear the dispute between xphiles and those who
defend analytic philosophy and its methods characterized as one over
whether philosophy can be pursued “from the armchair.” I myself, in the
last chapter, criticized the expertise and multiple concepts replies to the
negative xphile critique for failing to defend philosophy’s armchair meth-
ods. Nevertheless, the characterization of the dispute in terms of those for
or against armchair philosophy is misleading in at least two ways. First, the
characterization wrongly suggests that armchair philosophy is unscientific,
or unconcerned with empirical results related to its subject matter. Second,
the characterization unfairly casts xphi as a curative—a pro-science balm
designed to counteract the tendency to simply sit in an armchair and think.
Since typical survey-style xphi methods are clearly empirical, casting
xphiles as opposed to armchair philosophy suggests that armchair methods
are not empirical. But this is not true. By definition, a priori methods are not
empirical. But sitting in an armchair does not prevent one from appealing
to things one has learned a posteriori. Even judgments about purely hypo-
thetical cases are informed in various ways by one’s empirical knowledge.
To return, once again, to the example from Frankfurt with which this book
began, it is simply not true that we know, solely on a priori grounds, that
Frankfurt’s counterexample to the principle that one is morally responsible
only if one could have done otherwise is genuine. Understanding the coun-
terexample requires a great deal of empirical knowledge, for example about
people’s motives and the role these play in grounding responsibility, and
about the conditions under which people typically have or lack the ability
to do otherwise.1
So we need not accept the view that if philosophical knowledge is arm-
chair knowledge then it is entirely a priori. Thoroughgoing “naturalist”
158 Conclusion
philosophers, who are skeptical of the a priori, can and do make a place in
their epistemology of philosophy for judgments made from the armchair.
Williamson (2007) makes an excellent case for this, and even Hilary Korn-
blith (2006, 2007), who often appears hostile to traditional philosophizing,
is hostile mostly, and rightly so, to the account of philosophical practice
foisted on us by believers in the myth of the intuitive (though Kornblith,
unfortunately, is also a believer; he is hostile to what he takes to be an accu-
rate account of philosophical practice, and wants, therefore, that the prac-
tice be reformed). Kornblith thinks that philosophy needs a healthy dose of
input from the sciences, but, on his picture, this is just input; philosophy
itself, according to Kornblith, is an armchair pursuit, but one that needs
to be substantially informed and constrained by experimental/empirical
results. Kornblith does not think that philosophers need lab coats and the
labs to wear them in; he thinks that they need collaborators with these tools.
And of course Kornblith’s view is the correct one. Who in this day and
age could or would seriously deny that philosophy must pay close attention
to the results of those empirical sciences which, to take up another of Korn-
blith’s themes, deal with the very same not-purely-conceptual, real-world
phenomena (such as knowledge, reference, and moral responsibility) dealt
with by philosophers? Philosophy is not itself a science, but it cannot afford
to be positively antiscientific by ignoring the results of scientific investiga-
tions of its subject matter. Often, the best, most accurate picture we have
of a given phenomenon is the one we have from science. This is true, for
example, of our present-day understanding of the mind. A philosophy of
consciousness that simply ignores empirical results concerning “change
blindness” or “blindsight,” to name just two examples, is not going to be
complete or accurate.
But there is no real danger here. Kornblith has been misled by the myth
of the intuitive, with its emphasis on intuitions and a supposedly fully
a priori method for philosophy. As I have been stressing throughout the
book, the actual practice of philosophy does not look anything like the way
Kornblith and other believers in the myth take it to look. Philosophers do
not need advice to look closely and carefully at the sciences that bear on
their topics; most of us already heed this advice and have been doing so
at least since the rise of positivism and, arguably, for much, much longer.
Denying that armchair methods are entirely a priori is not to affirm that
such methods are entirely a posteriori. Discerning metaphysical possibilities
Conclusion 159
understanding ourselves and getting along with others. The work, if I may
take the liberty of summing up what is now a large body of results, is helping
to reveal just where, when, and how our considered moral judgments are
disconnected from the way we act in real-life situations that demand spon-
taneous moral assessments. In other words, it is the kind of experimental
philosophy that has clear philosophical relevance and is motivated by admi-
rable, broadly ethical aims. But none of it, as far as I can see, depends in any
way on commitment to the myth of the intuitive. One need not buy in to
the view that intuitions are evidence for moral theory, or are treated as such
in first-order theorizing in ethics, in order to map out the interesting connec-
tions (or lack thereof) between people’s spontaneous moral decision making
and their more considered moral opinions. None of the work experimentally
investigating this terrain is devalued, undermined, or called into question in
any way by the tenor or arguments of this book.
The distinction between spontaneous moral judgments and the consid-
ered moral judgments that figure in our moral theories is an instance of
the more general distinction between spontaneous and considered philo-
sophical judgments. And that more general distinction is ripe, it seems to
me, for exploration via experimental methods. I would welcome examples
of experimental philosophy that put aside specious claims about the link
between intuitions and philosophical evidence and focused simply on the
distinction between those two ways of cognizing a philosophical claim—
arriving at it seemingly without inference versus arriving at it through con-
scious inference.
This distinction, along with how the two modes of cognition play a role
in shaping not just our moral outlook but our philosophical outlook more
generally, is bound to have all sorts of broadly ethical implications. I encour-
age xphiles to unearth them—using surveys, but also whatever other empiri-
cal methods are required. An xphi unfettered by the myth of the intuitive
has an important place in philosophy and can make a significant contribu-
tion, but it is a mistake to think that this place and contribution will have
something to do with evidence for philosophical theory; xphi’s potential
contribution is different from what most of its practitioners take it to be.
It is also a potential that is yet to be fully realized. Hopefully, this book
will steer xphiles away from claims about how their studies and data bear
on the evidence for philosophical theory and toward the broadly ethical
goals that it has a genuine chance of serving.
Notes
1. From here on out, I will for the most part refer to the kind of philosophy that
some varieties of xphi seek to challenge as simply “philosophy,” though, as I hope is
clear by now, the kind in question is really that kind of analytic philosophy that
gives some sort of argumentative role to thought experiments and hypothetical
cases. Sometimes, I will refer to it as “traditional philosophy,” or “traditional ana-
lytic philosophy.” I am not especially happy with the usual definitions of “analytic
philosophy,” and “traditional philosophy” has connotations that are not quite
right. The general terms are stylistically useful, however. Really, all I mean by them
is: that kind of philosophy exemplified by Frankfurt’s argument about moral respon-
sibility and Gettier’s anti-JTB argument, along with several other examples we will
examine later.
2. “Those who do care about intuitions” are alleged by negative xphiles to include
not just traditional philosophers who (negative xphiles claim) appeal to intuitions,
but also xphiles in the other, more positive camp who also appeal to intuitions, but
do so in more scientifically respectable way. My favorite paper in the whole of the
xphi literature is Alexander et al. 2010. The paper, “Accentuate the Negative,” argues
that positive xphi is (nearly) as bad off as traditional “armchair” philosophy and
that the negative xphi critique works (nearly) as well against positive xphi as it does
against traditional philosophy. A main thesis of this book is that, whereas positive
xphi does carve out an important place for intuitions, traditional analytic philoso-
phy does not; hence the negative xphi critique works only against positive xphi.
Accentuate neither the positive nor the negative, I say, but the traditional (which is
not Mr. In-Between, but Mr. Entirely Different).
3. See Cappelen 2012 for an interesting argument to the effect that there is no uni-
vocal use of “intuition” in analytic philosophy. In fact, Cappelen argues that, not
only is there no univocal use, but the extreme unclarity in the intentions behind
these uses makes them literally meaningless. I disagree with Cappelen’s conclusion,
164 Notes
as I will make clear in section 1.5, but I do think that the premises of Cappelen’s
argument strongly support a different conclusion, which Cappelen also endorses,
namely that the ways in which “intuition” is used in philosophical texts do not
show that philosophers treat intuitions as evidence.
4. Several papers discuss divisions in xphi, including Knobe and Nichols 2007,
Appiah 2008, and Nadelhoffer and Nahmias 2007.
5. If xphi’s methods were irremediably methodologically flawed, then there would
be no actual grounds for the challenges to traditional methods I will discuss in this
chapter. At best, there would remain only hypothetical challenges of the form “If
groups of people were to intuit in such-and-such ways, then there would be thus-
and-so reasons to suspect traditional philosophical practice.” As a matter of fact, the
first xphi-ish challenge was hypothetical in just this way. When xphi was just a
twinkle in the eyes of some of his graduate students, high-muck-a-muck of xphiles
everywhere Stephen Stich hypothesized groups of people with significantly different
philosophical intuitions from “ours,” and concluded that this possibility seriously
undermines the methods of analytic philosophy, which Stich took, and still takes,
to be highly reliant on intuitions (Stich 1988). Xphiles, including Stich, now think
that this possibility has been shown to be actual. I agree, though I am not sure that
it matters very much that the possibility is actual, though that may be simply
because I am fond of hypotheticals and thought experiments. In any case, although
there are very likely some xphi studies that are methodologically unsound (this fol-
lows from the fact that there are experimental studies in all areas that are method-
ologically unsound), we can, given the sheer numbers of studies, and the care,
intelligence, and diligence of xphiles, be reasonably sure that this is not some sort of
in-principle problem with collecting people’s intuitive reactions in experimental set-
tings. All that said, I think one fairly general methodological problem is severe
enough to cast doubt on the methodological soundness of quite a few xphi studies.
This is the problem discussed in section 1.3.
6. It is difficult to make good sense of the current empirical data on Gettier intu-
itions. In addition to the studies I have mentioned in the main text, there is a study
carried out by Jennifer Nagel and her colleagues, the results of which have not yet
been published, but on which she reports in Nagel 2012. According to Nagel, her
studies reveal very few differences in Gettier intuitions due to ethnicity. Further-
more, unlike the findings in Starmans and Friedman 2012, Nagel’s results, she says,
reveal that people do tend to have the Gettier intuition. That is, significant majori-
ties of her subjects, in a range of Gettier cases, judge that the relevant agents do not
know the propositions in question. That is in rather direct conflict with Starmans
and Friedman’s findings. A forthcoming paper by Stephen Stich argues that Nagel
has not shown that we have reason to doubt that there are cross-cultural differences
in epistemic intuitions. Suffice it to say that the issue of diversity in Gettier intu-
itions is not yet fully settled. However, I will continue to assume in the main text
that there is still some (perhaps shrinking) evidence for cross-cultural diversity.
Notes 165
7. This is not to say that Knobe takes his work to have no bearing on philosophical
questions. Indeed, he has stressed in many places that questions about the workings
of the mind are central philosophical questions. He views his work as an empirically
informed investigation into such questions.
8. In English, most definite descriptions are familiar noun phrase constructions
beginning with the definite article “the,” e.g., “The man who won the 2012 US
presidential election.”
9. The “Tsu Ch’ung Chih Case” is a case very closely modeled on Machery et al.’s
Gödel Case, but modified to contain elements more immediately recognizable to
East Asians.
11. Or, if they do not quite think this, they at least think that their results show that
we are less entitled to the Kripkean judgment about the Gödel Case than we may
have initially supposed.
12. This is roughly Michael Devitt’s (2011, 2012) view of “folk intuitions” about
reference; they are evidence, but only a little bit of not especially good evidence. His
complaint about the “experimental semantics” exemplified by Machery et al. 2004
is that it assumes that semanticists treat such intuitions as having a very important
evidential role. They have an evidential role, Devitt thinks, but a far more minimal
one.
13. Negative xphiles rarely bother to spell out the form of the argument. Weinberg
2007 comes close, but Weinberg’s main concern in that paper is to explain how the
critique extends from conclusions about specific philosophical arguments (those rela-
tive to which truth-irrelevant variability in intuitions has been discovered) to the
entire practice that Weinberg describes as “philosophers’ appeals to intuitions.”
Joshua Alexander’s (2010) response to the reply to negative xphi he discerns in
Timothy Williamson’s metaphilosophical work makes it clear that at least some
negative xphiles (e.g., Alexander himself) do assume that analytic philosophers treat
intuitions about cases as essential evidence for what is true in those cases. In Alexan-
der’s reconstruction of Gettier’s anti-JTB argument, this is clear.
14. Note, however, that Swain et al. (2008) do not quite come out and say that
Lehrer’s Truetemp Case—based argument has been undermined. They say instead
that intuitions about Truetemp cases are not good evidence. That leaves it open that
there is different and perfectly good evidence for Lehrer’s judgment about the True-
temp Case. My view is that they intend the stronger claim; the argument that Lehrer
actually gives, they think, does rely essentially on intuitions. If that is what they
think, they are wrong. Lehrer argues for the judgment he makes about the Truetemp
Case. Cappelen (2012) argues convincingly for this.
166 Notes
15. Again, see Cappelen 2012 for the point that Lehrer himself argues for his judg-
ment about the Truetemp Case.
16. Why might it communicate this pragmatic meaning? Because people know (and
know that other people know) that, in a great many typical cases, when agents are
responsible/blameworthy for an outcome, this is partly in virtue of the fact that they
brought it about intentionally. That is the full explanation. Sometimes philosophers
demand special Gricean (see Grice 1989) explanations of how pragmatic meanings
arise. Though I do not have space to defend the view here, I think this is an unrea-
sonable demand. Grice’s theory cannot explain every case of pragmatic meaning. In
fact, there is a sense in which there can be no general theory of pragmatic meaning.
Whether an utterance gives rise to this or that pragmatic meaning is a highly contex-
tual matter. Also, Grice’s theory applies only to those pragmatic meanings that are
derived from speakers’ communicative intentions. But meanings that are generated
instead by hearers’ expectations and beliefs contribute to the phenomenon I am call-
ing pragmatic distortion.
17. Knobe, remember, does not intend his results to play a role in any negative xphi
argument, but, as I argued earlier, on the face of it, someone else might be inspired
to use them for that purpose. The point about pragmatic distortion is that, on a
second look, it is unclear whether the variability in answers to the survey question
concerning the Harm Case amounts to variability in intuitions about the Harm
Case.
18. For the record, I think there are cross-cultural differences in intuitions about
reference, but I do not think that Machery et al. 2004 should be credited with this
discovery. Later studies (as yet unpublished) that attempt to eliminate the pragmatic
distortion I describe in the main text appear to show that, yes, East Asians are more
attracted to descriptivism than Westerners. I should say here that, in the case of
Machery et al. 2004, the pragmatic meaning that I claim is conveyed by their survey
question and answer choices is conveyed in roughly the following way: it is common
knowledge between speakers and their audiences that, in typical cases, when one
uses a proper name to “talk about” a person, this is a person one intends to be talking
about. This, again, falls short of a Gricean explanation of how the pragmatic mean-
ing is generated, but that is not to say the explanation is faulty or incomplete. (See
note 15 above.)
19. Though see Bengson 2013 for grounds for saying exactly this. Bengson’s
grounds, all of which are based on the idea that one might answer a question with
“p” even when p does not “strike one” as true (and so is not, according to Bengson,
expressive of an intuition that p), are interesting, but negative xphiles are bound to
reply by legitimately demanding that Bengson provide evidence that the phenome-
non he points to really does affect their experimental designs, not just that it could.
20. There are reasons to suspect that the two studies on Gettier intuitions, and the
Knobe and Nichols study on intuitions about compatibilism, are affected by prag-
Notes 167
matic distortion as well. In the Gettier studies, subjects were asked whether an agent
described in the vignettes “really knows” something. That is different from asking
them whether an agent in the vignette knows something. That is, the intensifier,
“really,” potentially has an effect on how subjects understand the question. It is not
clear what this effect might be—perhaps some subjects understand it as asking
whether the agent described in the vignettes knows with certainty the relevant thing,
or perhaps some subjects understand it as asking whether they (the subjects) are
reasonably, or perhaps very, confident that the agent knows. In either case, a subject
might answer differently than they would if they took the question to be merely a
stylistic variant on the question of whether the agent (simply) knows. Perhaps this is
not pragmatic distortion, but rather semantic ambiguity; but in any case, the under-
lying problem is the same: it is impossible to tell from the results alone which ques-
tion subjects were answering. Regarding Knobe and Nichols 2007, a case can be
made that a “no” answer to the question asked about the concrete, affect-inducing
vignette, “Is Charles fully morally responsible for killing his wife and kids?” suggests
various things about how Charles ought to be treated—for example, that he ought
not be tried and punished for the act. If that is right, then, again, “no” answers do
not clearly conflict with “yes” answers, for one might think both that Charles is not
fully morally responsible but that, perhaps for nonmoral reasons, he should be tried
and punished.
One very general point about this objection-from-pragmatic-distortion deserves
airing. I do not claim simply that every study examined in this chapter might be
affected by pragmatic distortion. I claim instead that they are affected by it in fact.
So the objection is not the facile sort that one can make against any interpretation
of a given set of experimental data, namely the objection that has this form: “Well,
the results might be equally well explained in this other way.” In each case, I have
suggested specific pragmatic meanings and have gestured at how they might arise
and how they might influence a subject’s responses (these mights are wholly appro-
priate; there is no evidence that these are exactly the pragmatic meanings that arise,
that they arise in the very way I have indicated, or that any particular subject’s
response was distorted by them). Also, the objection does not take the form of a
complete alternative explanation of the data and is not intended to. The objection is
simply that we have reasons to suspect that, in some cases, subjects are not evaluat-
ing the semantic meanings of the survey questions. However, if, as is likely, other
subjects are evaluating the semantic meanings, then the existence of variable
answers to the survey questions is not clear evidence of variable intuitions.
21. It is well known that all sorts of judgments, even simple perceptual judgments,
can be affected by factors that are irrelevant to their truth. It would be extremely
surprising if this were not the case for some philosophical judgments.
22. Actually, these examples of intuitions from the traditional literature are general-
izations of more specific intuitive judgments about specific and detailed hypotheti-
cal examples. But you get the idea.
168 Notes
23. I suspect this is not the intended view. Rather, the view is that intuitions are
beliefs or judgments as opposed to a special mental kind such as an intellectual
seeming. That can be true even if many beliefs/judgments are not intuitions.
24. At any rate, Malmgren (2011) implies that there is widespread agreement in
philosophy that a, or perhaps the, method of philosophy is the “method of cases,” a
method she takes to depend heavily on intuitive judgments. It is not especially clear
that this means she thinks that “intuition talk” in philosophy is unified or coherent,
though my own view is that the “method of cases,” when characterized as Malmgren
characterizes it, is a myth. Cappellen would agree. I will say much more in later
chapters about the method of cases and why, under its usual descriptions, it is a
myth.
25. This is part of what is at issue, but only part. Those involved in the project also
disagree over the “logical forms” of what I am calling the “specific” Gettier judg-
ments. However, they agree at least about the “surface” forms. None would deny,
for example, that the surface form of the intuitive judgment relative to the Gettier’s
10 Coins Case is: Smith does not know that the man who will get the job has ten coins in
his pocket. Agreement over the surface forms suffices for identifying the paradigms, it
seems to me. I am not sure whether Malmgren would agree (her favored analysis of
the surface form of Gettier judgments is misleadingly described as the “real” Gettier
judgment), though I think she should.
1. Although there are important differences between us that will be spelled out later
in section 2.5 and chapter 3, Timothy Williamson (2007) takes a similar line against
negative xphi. There are also similarities between my view of philosophical method
and that of Jonathan Ichikawa (2012). My closest ally on such issues is Herman
Cappelen (2012).
3. Other diagnoses of the appeal of the myth of the intuitive miss this fairly obvious
source. For example, Cappelen’s diagnosis of the appeal of the myth (which he calls,
less derisively, Centrality) mentions several sources, the main one for Cappelen
being a relatively recent “verbal tic” afflicting analytic philosophers that causes
them to pepper their writing with plenty of (unmotivated, according to Cappelen)
uses of “intuitively” and cognates. The truth of (EC2) explains both the appeal of
the myth and the presence of the tic.
5. Ichikawa (2012) discusses the state/content ambiguity of “intuition” and uses it
to make some of the same points I make here.
6. There is no logical conflict between (EC1) and (EC2); one can consistently hold
both. Perhaps the xphi view of the analytic method is better described as the con-
junction of (EC1) and (EC2). Indeed, this conjunction of (EC1) and (EC2) seems to
be a popular metaphilosophical view even outside of xphi circles. Nevertheless, I
shall argue that, since (EC1) is false, the conjunctive view is too. Ichikawa (2012)
and Williamson (2007) acknowledge the state/content ambiguity of “intuition” and,
as I do, reject the view that intuitions in the “state” sense are treated as evidence in
philosophy.
7. Bealer (1998) is often taken by xphiles as a prime example of an analytic philoso-
pher who holds that intuitions are evidence in philosophy. Given the state/content
ambiguity of “intuition,” Bealer’s explicit endorsement of the idea that intuitions
are evidence might be merely an endorsement of (EC2). In fact, at one point, Bealer
says, “When I say that intuitions are evidence, I of course mean that the contents of
the intuitions count as evidence” (1998, 205). However, immediately following this,
Bealer adds, “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”
(1998, 205). The best reading of Bealer is thus one that takes his claim that intu-
itions are evidence as an endorsement of both (EC1) and (EC2).
9. An anonymous reviewer for MIT Press objected that I do not pay adequate atten-
tion to methods employed in nonoriginal presentations of philosophical thought
experiments and cases. David Chalmers (2014) raises the same objection against
Cappelen 2012 in a commentary on Cappelen’s book.
10. I leave aside, for now, the issue of how purely hypothetical cases could count
against the generalizations I have labeled the “JTB theory of knowledge” and the
“descriptivist theory of reference.” This issue is taken up in section 2.3.3.
11. At various places throughout the book, when discussing a philosopher’s presen-
tation of a hypothetical case, I will point out, as I do here with respect to Gettier’s
presentation of the 10 Coins Case, that the philosopher presenting the case does not
use “intuition” or cognates in the presentation. I should say, however, that I do not
take this to be conclusive evidence that intuitions are not being appealed to as evi-
dence in these presentations. What I do think is that the lack of intuition terminol-
ogy shifts the burden of proof to those inclined to see evidential appeals to intuitions
in these presentations: if there are implicit such appeals in the relevant presenta-
tions, then the burden, given the lack of explicit appeals, is on my opponent to
demonstrate this.
12. Kripke’s (1980) main stalking horse is a theory of meaning for proper names
unfortunately also known as the “descriptivist theory.” Descriptivism about the
meaning of a proper name is the view that the “semantic value” of a name is identi-
cal with the semantic value of the definite description, or “cluster” of such descrip-
tions, users of the name associate with it. Kripke’s arguments against descriptivism
as a theory of meaning for names will be left to one side here. What I have labeled
the “descriptivist theory of reference” is a theory of how names “have their refer-
ence fixed.” It is not a theory of a name’s “semantic value,” conceived as that contri-
bution a name makes to the propositional content of containing sentences. It is
important to recognize this distinction between the two varieties of descriptivism in
order to understand the aim of the Gödel Case, which I present below, in the main
text. The Gödel Case is designed to refute a certain theory of “reference fixing,”
namely the descriptivist theory of reference. It can succeed in this regardless of the
quality of Kripke’s arguments against descriptivism about meaning, since descriptiv-
ism about reference is independent of descriptivism about meaning, in sense that
one might be a reference descriptivist without being a meaning descriptivist. (How-
ever, if “meaning determines reference,” meaning descriptivism commits one to
reference descriptivism.)
13. It is true that Kripke first says, “It seems to me that we are not,” when consider-
ing the question of whether “Gödel” refers to Schmidt, immediately following this
with, “We simply are not” (1980, 84). However, it would be a stretch to interpret
this as an inference from “It is intuitive that we are not” to “We are not.” It seems
rather as though Kripke is replacing what strikes him as an overly cautious “seems”
claim with the more appropriate and bolder “is” claim.
Notes 171
14. I do not say that xphiles definitely do think that the best representation of the
arguments’ form requires a premise about intuitions in the “state” sense; xphiles are
unfortunately not very explicit about how best to represent the forms of the philo-
sophical arguments they seek to criticize. However, I do think that if they did assume
that the best representations of their target arguments include premises about intu-
itions in the “state” sense, then it would be understandable why they might then go
on to suppose that cross-cultural variability with respect to a philosophical intuition
poses a serious threat to an argument based on the content of that intuition. For, in
that case, the variability would be evidence that the premise is false.
15. Occasionally, philosophers do more than just strongly suggest that philosophi-
cal method involves appeals to intuitions in the “state” sense as evidence. Pust, after
describing a typical Gettier case, says that “most philosophers take the fact that they
have the intuition that S does not know p in this case to show that S does not know p”
(Pust 2000, 5; emphasis added). Here, it is clear that Pust thinks “most philosophers”
take the psychological state of intuiting that S does not know p (in the relevant case)
as essential evidence for the conclusion that S does not know p (in the case). Alexan-
der and Weinberg (2007), in describing a Gettier case, write, “According to Gettier,
despite now having a justified true belief that q, the person lacks knowledge that q.
Purportedly, when we consider this case, we will have the intuition that the person
whose epistemic position is detailed in the thought-experiment does not know that
q. Further, this is to count as sufficient evidence against the claim that a person
knows that p just in case that person’s true belief is justified” (2007, 57; emphasis
added). According to Alexander and Weinberg, it is our having the intuition, not
(just?) the truth of the content of the intuition, that serves, according to analytic
epistemologists, as evidence against the JTB theory of knowledge.
16. An anonymous reviewer for MIT Press says that it is strange to think that the
notion of a counterexample is a purely logical one and hence that there could be
counterexamples just, as he or she puts it, “out there, floating through the ether.” I
confess to not understanding the complaint. If there are pink cows, then there are
counterexamples to the claim that no cows are pink, regardless of whether anyone
has, does, or will believe that there are pink cows. To my ear, there is nothing the
least bit “weird,” as the same reviewer also puts it, about this purely logical, nonpsy-
chological conception of a counterexample.
17. Cf. Hilary Kornblith (1998, 134), who draws an apt analogy between counterex-
ampling in philosophy and a rock collector judging that a rock meeting certain
conditions does not count as a sample of a given kind.
ence by pointing out that the theory is inconsistent with the fact that “Madagascar”
refers to the island off the east coast of Africa, he was using the same method as other
philosophers who present counterexamples, despite the fact that his case was an
actual case, not a purely hypothetical one.
1. I do not mean to suggest that Williamson accepts that intuitions are noninferen-
tial judgments. He denies that there is any unifying account of “intuition” as it is
used in philosophy and even gives examples of judgments that are described as
“intuitive” by philosophers but are products of inference (2007, 217). Nor do I mean
to suggest, in claiming that he allows that some evidence is “basic,” that Williamson
is a foundationalist about evidence. It is rather that there are specific examples of
judgments made in philosophy that Williamson regards as noninferential and
standing in need of no inferential justification. His main example is “the Gettier
intuition.” Williamson takes this judgment to have “epistemic priority” (182) over
more general principles that might imply it, and argues at length (in chap. 7 of his
2007) that it may be regarded as evidence even if it does not receive any argumenta-
tive backing of its own.
2. Alexander (2010) expresses puzzlement about this, claiming at one point (see
Alexander 2010, 383) that Williamson perhaps thinks that our judgments that
agents in Gettier cases fail to know are self-justifying. This is not Williamson’s view.
Williamson’s view is that we can take it for granted that these judgments are true,
unless there is some legitimate challenge to them, one, for example, that does not
lead to judgment skepticism.
3. It somewhat misrepresents the view expressed in Weinberg et al. 2001 to describe
it as a view according to which the Gettier intuition is “mere cultural prejudice,”
however. Weinberg et al. 2001 never takes a stand on whether the Western Gettier
intuition is true or false. Yet representing its authors as holding that it is a mere cul-
tural prejudice, as Williamson does, suggests that they take it to be false, when in
fact they do not.
4. Alexander (2010) and Weatherson (forthcoming) argue that the dialectical con-
ception of evidence is correct.
6. Here again the state/content ambiguity of “intuition” comes into play. The idea
is that commitment to Evidence Neutrality forces one to trade regarding “the Gettier
intuition” in the “content” sense as evidence for regarding “the Gettier intuition” in
the “state” sense as evidence.
7. One might appeal to the general reliability of intuitive judgment to bridge the
gap: people intuit p and intuitions are reliable; hence p (or likely p). Brian Weather-
son (forthcoming) suggests this move as a way of taking the sting out of William-
son’s charge that Evidence Neutrality leads to a psychologization of the evidence in
philosophy. As I said earlier, I’m not opposed to the idea that the intuitiveness of a
proposition provides some highly defeasible evidence for its truth. What is troubling
about Evidence Neutrality is that it seems to lead to the view that the philosophical
evidence consists entirely of the psychological. That would be bad, since, in that
case, there would be no way to verify the supposed reliability of intuition.
2. I say a simple clash of belief because intuitive clashes are, usually, also cases of
clashing beliefs, since people tend to believe what they intuit. A simple clash of
beliefs is a case in which beliefs clash but it is not (also) a case of clashing
intuitions.
3. An anonymous reviewer for MIT Press claims that I am not careful enough, in
these passages in the main text, to note results from work in the epistemology of
disagreement. In my opinion, however, this work is not clearly applicable. Those
who work on the epistemology of disagreement agree that mere disagreement with
someone else over p is not sufficient grounds for withholding judgment regarding p.
That someone else must be an “epistemic peer,” for one thing, where someone is
disqualified from being one’s epistemic peer if they have more (or less) evidence for
p than one does.
4. I should stress that this is an empirical claim about extant first-order analytic
philosophy. It could be falsified tomorrow by the publication of a paper that uses a
thought experiment and in which it is clear that the author is appealing to the intu-
174 Notes
5. The view that philosophers treat intuitions about cases as (not the only, but nev-
ertheless) essential evidence for the truth about those cases is not quite inconsistent
with the actual practice of first-order philosophy, but that practice strongly suggests
that the view is false. If the view were true, we would at least sometimes see explicit
focus on the intuitiveness of a judgment about a case in the process of justifying the
judgment. But, I claim, we never do. (See Cappelen 2012 for more on this theme.)
That strongly suggests that view is false. In any case, the issue of whether philoso-
phers do treat intuitions about cases as essential evidence for their judgments about
them is an empirical issue that ought to be settled by canvassing the literature and
finding clear examples. To my knowledge, no xphile has ever done this.
7. Ironically, one reason for the derision is that Gettierology is supposed to lean
heavily on intuitions.
9. An anonymous reviewer for MIT Press objects that what I am describing here as an
argument for Gettier’s judgment about the 10 Coins Case is instead just a redescription
of certain features of the case. As he or she puts it: “Every premise of the argument is
nothing more than Gettier repeating some fact about the case he just presented.” He or
she points out, for example, that that (e) is true in virtue of the number of coins in Smith’s
pocket is simply a detail built into the earlier story that Gettier tells in presenting the
10 Coins Case. The objection appears to assume that the premises of the argument
can’t also be features or facts of the 10 Coins Case itself. But there is nothing to be said
for this assumption; indeed, the most likely place to find the premises of an argument
supporting a judgment about a thought experiment is right there in the specification
of the details of the thought experiment or case itself—where else? So, yes, the argu-
ment Gettier gives is a summary of some of the relevant facts of the 10 Coins Case;
Gettier is redescribing these facts, true. However, it is also an argument for the crucial
anti-JTB judgment that Smith does not know.
Notes 175
10. Gettier also gives an explicit reason for thinking that his Smith character fails to
know in both the 10 Coins and Brown in Barcelona cases. This explicit reason is not
the fact that Smith deduces his justified true belief from a justified false one in each
case, but instead that the fact that, in each case, the claims Smith justifiably and
truly believes are made true by facts disconnected from the facts that justify Smith
in believing them, and thus it is a lucky accident that the beliefs are true. I discuss
this explicit reason in a bit more fully in what follows in the main text.
11. Many authors say things that imply that Gettier does not argue for his judg-
ments about his cases, but it is rare to find someone explicitly asserting that there
are no arguments there to be found. As I say in the main text, Nagel 2012 is the only
exception of which I am aware.
12. An interesting example to consider in this connection is that of the very first
published attempt to patch up the JTB theory with an additional, fourth, necessary
condition, namely Clark 1963, in which it is proposed that, by disallowing infer-
ences from “false grounds,” we can formulate the correct theory of knowledge.
Roughly, according to Clark’s theory, knowledge is JTB in which the relevant belief
is not grounded by any falsehoods. The theory faces a number of clear counterex-
amples to the sufficiency of the proposed conditions, but, more interestingly from a
methodological perspective, to their necessity as well. This is interesting because it
seems to me that, despite the fact that the “no false grounds” condition is not
strictly necessary, if the condition goes unmet while the other conditions (J, T, and
B) are satisfied (as, for example, in Gettier’s original cases), that is strong evidence
that the relevant agent fails to know. One can argue, therefore, that agents in such
circumstances do not know because they infer based on false grounds. Indeed, this is
precisely what Clark 1963 does argue. In a sense, then, there is nothing wrong with
the argument except Clark’s further claim to the effect that the “no false grounds”
condition is strictly necessary (and that it, along with the original JTB conditions,
are sufficient) for knowledge. The presence of false grounds in an inferential chain is
a strong indicator, without being conclusive evidence, for the inferred belief failing
to qualify as knowledge.
in effect, been made earlier. More care was needed in showing that luckily true
beliefs can count as fully justified, less in just repeating what was known already,
namely that epistemic luck is knowledge-preventing. (Engel 1992 inspired this inter-
pretation of Gettier’s contribution.)
15. Clark’s proposal is actually a bit subtler than this description or my rough char-
acterization in note 12 suggests. Clark makes a distinction between those beliefs that
ground a given belief and the “second-order grounds” that ground the grounding
beliefs. Using this distinction, he presents a case in which all of an agent’s first-order
grounds are true and justify some true belief an agent has. But, in Clark’s case, one
of the second-order grounds of one of the agent’s grounding beliefs is false, and
Clark argues that this prevents knowledge in the case. Still, the proposal is that there
can be no falsity anywhere in an agent’s total set of grounding beliefs, where this
might include grounds for grounds, grounds for grounds for grounds, etc.
16. This is their “Tom Grabit” case. They argue that the Tom Grabit case is one
involving knowledge. (See Lehrer and Paxson 1969, 228–229.) They make no appeals
to brute intuition anywhere in the paper.
17. I hasten to add that there are things to be said in favor of the East Asian judg-
ment about Gettier cases too, though the examples I know of are due to Western
philosophers. Both Weatherson (2003) and Stephen Hetherington (1999; 2001)
argue that the JTB theory is true, and arguments for the JTB theory are ipso facto
arguments for the East Asian judgment about Gettier cases.
5 More Evidence for the Evidence and the Relocation Problem Redux
1. It is really a very large reach to take the quotation from Kripke I have been dis-
cussing as expressing an overarching methodological view about the relationship
between intuitions and evidence in philosophy. Any remaining doubts on this score
ought to be settled by reminding ourselves of the sentence that comes immediately
after the passage I have reproduced, concluding the passage and ending Kripke’s
commentary on theme of the meaningfulness of the distinction between essential
and accidental properties. The concluding sentence is, “But, in any event, people
who think the notion of accidental property unintuitive have intuition reversed, I
think” (Kripke 1980, 42). This makes it very clear, I think, that Kripke is not suggest-
Notes 177
ing some very general methodological view about intuitions and evidence in phi-
losophy. He has a much more limited principle in mind, one to do with the
connection between ordinary people drawing certain distinctions and the meaning-
fulness of the distinctions thus drawn.
2. These other counterexamples and their supporting arguments, though they are,
strictly, counterexamples to a stronger form of descriptivism than that tested by the
Gödel Case, can nevertheless be viewed as supporting Kripke’s judgment about the
Gödel Case. These earlier counterexamples raise the suspicion that associating
descriptive material with a name will not explain the referential facts about names.
3. Strictly speaking, speakers can go wrong in uttering “Peano discovered the axioms
of number theory,” even if descriptivism about reference is true. It could turn out
that no one discovered the axioms of number theory. That possibility also shows that
Kripke’s example does not express a trivial truth, even if descriptivism about mean-
ing is true. This means only that the wrong example sentences have been chosen;
the underlying point still holds. Consider: “Peano, if he exists, discovered the axioms
of number theory.” On meaning descriptivism, this sentence is trivially true, while,
in fact, it is false. Similarly, the sentence, given reference descriptivism, cannot be
uttered falsely, when, in fact, any actual utterance of it is false.
4. An anonymous reviewer for MIT Press complains, as he or she did about my ren-
dition of Gettier’s argument for his judgment about the 10 Coins Case (see note 9 of
chapter 4), that the premise of this “argument” that I claim that Lehrer gives is
simply a feature or fact of the case as Lehrer stipulates it. My reply is the same: Yes,
the premise is a feature or fact of the case, but not simply that. It is also a premise in
Lehrer’s argument for his judgment about Mr. Truetemp.
5. One important facet of Lehrer’s argument that is skipped over by both Swain et
al. and Cappelen is that the argument is embedded in a larger argument that has a
much larger target, namely every extant “externalist” theory of knowledge, not just
Goldman-esque reliability theories. The Truetemp Case is presented after what
might as well be called Lehrer’s Thermometer Case. In the Thermometer Case,
Lehrer imagines an ordinary thermometer used to measure the temperature of an oil
of some kind. Lehrer supposes it to be highly reliable and to be reading accurately
on some occasion that the temperature of some sample of the relevant sort of oil is
104 degrees. Lehrer takes it for granted that the thermometer does not possess
knowledge that the temperature of the oil is 104 degrees. Ordinary thermometers do
not know things; this is treated by Lehrer as a datum (as it surely ought to be). And
yet, the thermometer satisfies various externalist conditions on knowing. Its reading
is produced by a reliable temperature-reading process, its reading “tracks the truth”
about the oil’s temperature, etc. So the explanation of why the thermometer does
not know cannot be failing to satisfy these externalist conditions. Lehrer then con-
siders whether what is missing in the Thermometer Case is simply the capacity for
thought. Suppose, he says, that we add this, producing the Thoughtful Thermometer
178 Notes
Case. The Thoughtful Thermometer literally believes that the temperature of the oil
is 104 degrees. But, still, Lehrer says, it does not know. And here he offers a reason:
though it believes that the temperature of the oil is 104 degrees, it “might have no
idea that it is an accurate temperature-reading device” (Lehrer 1990, 162). This
reason is then used to justify the judgment in the later Truetemp Case, but it is clear,
I think, that Lehrer takes it to be no more plausible that Mr. Truetemp knows than
that the Thoughtful Thermometer knows, and no more plausible that the Thought-
ful Thermometer knows than that the Thermometer knows. So there is not just an
argument that appeals to a general principle to the effect that knowledge requires
the satisfaction of some “internalist” condition, such as justified belief about the
reliability of one’s beliefs; there is also an argument by analogy to the Thermometer
and Thoughtful Thermometer cases for Lehrer’s judgment in the Truetemp Case.
The reasoning is highly reminiscent of Kripke’s reasoning about the Gödel Case in
which we find that a comparison between a somewhat controversial case (Truetemp
in Lehrer; Gödel in Kripke) and much less controversial ones (Thermometer and
Thoughtful Thermometer in Lehrer; Einstein, Peano, and Columbus in Kripke) is
used to bolster the judgment about the controversial case.
6. Actually, this judgment about the Mary case, framed as it is in terms of “knowing
what it is like to see red,” does not appear in Jackson 1982. There, the judgment
about Mary is instead that, when she starts receiving color television transmissions,
“she will learn something about the world and our visual experience of it” (Jackson
1982, 130). The “what-it-is-like” formulation occurs in Jackson 1986. The phrase
also appears in the discussion of extra-color Fred in Jackson 1982.
7. In his 1982 and 1986, at any rate—Notoriously, Jackson, has changed his mind
and now rejects the knowledge argument and the antiphysicalism he once claimed
it established. (See Jackson 1998, 2003, and 2006.)
8. I will admit, however, that Jackson’s case of Mary is one of the better candidates;
Jackson’s defense of his judgment about the Mary case is fairly minimal in compari-
son to the kinds of backing arguments for judgments about cases that one typically
finds in analytic philosophy.
9. This explains why many people have mistakenly taken the knowledge argument
to rest on a brute intuition about the Mary case.
10. I think there is an all but explicit suggestion, in Jackson’s presentations of the
knowledge argument, that facts about the qualitative characters of experiences are
knowable only via having experiences with those characters. This is why the various
physical facts we might learn about Fred will not add up to knowledge of what it is
like to see the extra color, and why the various physical facts Mary might learn
about color and color vision do not add up to knowledge of what it is like to see red.
The suggestion is implicit in Jackson’s repeated assertions to the effect that the vari-
ous things learnable by Mary while in the room, or by us before we have the opera-
tion that makes our visual systems Fred-like, do not come anywhere close to giving
Notes 179
us the crucial bit of knowledge concerning what it is like. I take this suggestion as an
additional argument for Jackson’s judgments about the cases: Mary learns what it is
like to see red only after leaving the room and having visual experiences with “red
qualia” because that is the only way one could come to know what it is like to see
red. Experience is the only teacher when it comes to knowledge of qualia, as the
general principle might be put. The impression that this general principle concern-
ing knowledge of qualia is part of what justifies Jackson’s judgment that Mary learns
what it is like to see red only after leaving the room and actually seeing red is fairly
common. Martine Nida-Rümelin’s (2010) presentation of the knowledge argument
in the Stanford Encyclopedia of Philosophy, for example, suggests that the judgment
about the case is partly justified by appeal to such a principle. (See especially the last
paragraph of the second section of Nida-Rümelin’s entry.)
11. In his “Author’s Response” in the original (1980) article, Searle explicitly denies
that the Chinese room argument evidentially depends on intuitions. (See the sec-
tion titled “Intuitions” in the Author’s Response section of Searle 1980.) In a passage
presaging the general methodological stance taken by me in this chapter and in the
book as a whole, really, Searle writes:
In sum, though in some sense intuition figures in any argument, you will mistake the nature of
the present dispute entirely if you think it is a matter of my intuitions against someone else’s or
that some set of contrary intuitions has equal validity. The claim that I don’t speak Chinese or
that my thermostat lacks beliefs aren’t just things I find myself mysteriously inclined to say.
(Searle 1980, 451)
Consider Putnam’s 1975 argument to show that meanings, as he put it, “just ain’t in the head.”
Putnam argues persuasively that what words mean depends on more than “what is in the head.”
He tells a number of stories the moral of which is that aspects of the natural history of how
someone learned the use of a word necessarily make a difference to what the word means. It
seems to follow that two people might be in physically identical states, and yet mean different
things by the same words. (Davidson 1987, 443)
This is not the language of someone who thinks the case for externalism (either his
own or Putnam’s) rests on intuitions about thought experiments.
14. By saying that it is largely a choice which propositions philosophers take as their
starting points, I do not mean to be saying that they start with propositions that
they do not take to be true. I assume that, in most cases, philosophers take their
starting points to be true (though of course sometimes an initial premise is treated as
an assumption or held true only “for the sake of argument”). The choice is among
propositions they take to be true, but it is not constrained such that the starting
point propositions must have some “mark of the intuitive.” I thank an anonymous
reviewer for MIT Press for forcing me to be clearer on this point.
180 Notes
15. An anonymous reviewer for MIT Press objects that my stance here risks possible
future xphi challenges. Perhaps current xphi results do nothing to challenge any
extant philosophical argument (because the challenges are not directed at rock
bottom evidence), but we can imagine data that do challenge such evidence. My
reply is that, yes, of course, we can imagine this, but that leaves everything as it is;
there is no reason right now to think that Gettier is wrong about his cases, or suspend
judgment on his verdicts about them, for example. Furthermore, the issue of what
sorts of propositions could qualify as rock bottom evidence is an extremely vexed
one in epistemology. Whether there will ever be any agreement on this issue is an
open question. So it is unclear whether we will ever be able to even recognize
whether some bit of xphi data challenges rock bottom philosophical evidence or
not. And, it should be borne in mind that I am, at this point, granting for the sake of
argument that it is not “arguments all the way down” and that there is founda-
tional, rock bottom evidence. Officially, I am agnostic. If there is always another
argument to which to appeal, then there is no possibility of a “crucial” xphi experi-
ment. Finally, it is pure speculation that there will be truth-irrelevant-variability
with respect to judgments about propositions that qualify as rock bottom evidence.
Given what we now know, this is no more and no less likely than that there won’t
be such variability with respect to such judgments.
16. This objection was raised by an anonymous reviewer for MIT Press.
2. On the other hand, the fact that philosophy has been practiced for millennia and
until very recently without any respectable, properly scientific attempts to determine
whether people share their intuitive judgments about thought experiments is yet
more evidence that first-order philosophical practice does not rely on intuitions as
evidence.
3. In a paper in which they object to the expertise reply, Weinberg, Gonnerman,
Buckner, and Alexander (2010) write: “It borders on the trivial to claim that philoso-
Notes 181
phers’ training makes them at least somewhat better than the folk, at least at some
philosophically relevant tasks” (Weinberg et al. 2010).
4. If expert intuitions were not taken by proponents of the expertise reply to be
essential evidence for judgments about cases, then there might be some other way
to defend the experts’ judgments other than by appeal to what and how they intuit.
(For example, one might appeal to whatever reasons the experts give for judging as
they do!) But the expertise reply, as I am understanding it here, agrees that intu-
itions are essential evidence. This makes its proponents vulnerable to the charge
that the burden of empirical proof is on them.
5. Xphiles have already begun to do this work for proponents of the expertise reply,
and the early findings seem to suggest that philosophers’ intuitions are no less
immune to certain kinds of biases than are the intuitions of the “folk.” See, e.g.,
Shulz and Cokely 2012 and Schwitzgebel and Cushman 2012.
6. That the demonstrated biases and effects are rather surprising is important and
seems to me to tip the scale slightly in favor of the view that philosophers would not
be less prone. Consider the effect of culture on Gettier judgments, for example. This
is a surprising effect, and part of what makes it surprising is that we do not have any
reason to believe it would diminish among expert subject groups. So it is unlike
something such as lack of attention to thought-experimental detail, which one might
reasonably expect to diminish with training of the sort philosophers receive.
7. Weinberg et al. (2010) claim that there are three ways to reply to the negative
xphile critique. One can object to the experimental design of the studies that appear
to show biases and effects in the intuitions of the folk. One can claim that the biases
and effects are not what they appear to be; perhaps, for example, East Asians are
simply operating with a different concept of knowledge, and the different pattern of
answers to Gettier cases among East Asian subject groups does not indicate genuine
disagreement between Western and East Asian populations. (This second way of
replying, by appealing to differing concepts, is the topic of the next section of this
chapter.) Or, one can say that biases and effects found in the intuitions of the folk
will not reappear in the intuitions of the experts—the expertise reply. But there is a
fourth way to reply, which is the way I have been recommending in this book: deny
that intuitions, whether of the folk or the experts, play any sort of evidential role in
philosophy.
8. This is not to deny that there are cases in which one determines that that some
proposition, of, say, mathematics, is true by consulting mathematicians. There are
plenty of such cases, of course. For example, I know that Fermat’s Last Theorem is
true because expert mathematicians have told me so. But consulting mathemati-
cians is not part of the methodology of mathematics, and that mathematicians
judge that Fermat’s Last Theorem is true is not part of the evidential basis on which
mathematicians ground their belief that the theorem is true.
182 Notes
9. Here I don’t mean to be denying that there is, so far, some empirical evidence
that, as a group, philosophers are no less immune to biases in intuitive judgments.
(See the references in note 5.) My point is that rather that, on the whole, philoso-
phers’ judgments about philosophical cases, which by my lights are usually not
intuitive judgments, are bound to be better and more accurate than nonphiloso-
phers’ judgments about cases.
10. In this passage, Sosa explicitly mentions Gettier intuitions and clearly has the
results of Weinberg et al. 2001 in mind. However, more than other proponents of
the multiple concepts reply, Sosa has made it clear that he takes the reply to be very
general, and has deployed the reply in connection with other xphi results (see Sosa
2007). He argues, for example, that an ambiguity in “moral responsibility” makes it
possible, or perhaps even likely, that some of the apparent disagreement between
the subjects surveyed in Nichols and Knobe 2007 over whether agents in determinis-
tic universes are morally responsible is merely apparent, merely verbal, disagreement.
And he approvingly cites Nichols and Ulatowski 2007, an xphi paper that proposes
to explain some of the apparent disagreement between its subject groups over when
actions are intentionally produced by appeal to an alleged ambiguity in action
ascriptions formulated with the term “intentionally.”
11. The irony is that the multiple concepts reply, though intended as a defense of
an intuitions-as-evidence methodology, engenders a kind of skepticism that is even
more damaging to that sort of methodology than the negative xphi critique.
12. I here ignore the possibility of combining the expertise and multiple concepts
replies, which, to some extent, is what both Sosa and Jackson do. It should be clear
from what I have said thus far that such a combination is, I think, ineffective twice
over.
1. Putting it this way, in terms of knowing the conditions under which so-and-so, or
knowing what grounds what, will suggest conceptual analysis to some ears. But that
is not what I intend. We sometimes know such things simply by living in and
observing the world—in a fully a posteriori manner, in other words. Some of the
facts about the conditions under which people are morally responsible for their
actions are surely knowable a posteriori, simply by observing people, the things they
do, and what they are held accountable for.
2. Jonathan Haidt’s work on “moral dumbfounding” is a good example of the kind I
have in mind. Haidt 2012 is a good overview.
References
Adams, Fred, and Annie Steadman. 2004. Intentional action in ordinary language:
Core concept or pragmatic understanding? Analysis 64 (2): 173–181.
Alexander, Joshua, and Jonathan Weinberg. 2007. Analytic epistemology and exper-
imental philosophy. Philosophy Compass 2 (1): 56–80.
Alexander, Joshua, and Jonathan Weinberg. 2014. The challenge of sticking with
intuitions through thick and thin. In Intuitions, ed. Anthony Booth and Darrell Row-
bottom, 187–212. Oxford: Oxford University Press.
Bealer, George. 1996. A priori knowledge and the scope of philosophy. Philosophical
Studies 81 (2–3): 121–142.
Bealer, George. 1998. Intuition and the autonomy of philosophy. In Rethinking Intu-
ition: The Psychology of Intuition and Its Role in Philosophical Inquiry, ed. Michael
DePaul and William Ramsey, 201–240. New York: Rowman & Littlefield.
Bonjour, Laurence. 1985. The Structure of Empirical Knowledge. Cambridge, MA: Har-
vard University Press.
Clark, Michael. 1963. Knowledge and grounds: A comment on Mr. Gettier’s paper,
“Is Justified True Belief Knowledge?” Analysis 24:46–48.
Davidson, Donald. 1987. Knowing one’s own mind. Proceedings and Addresses of the
American Philosophical Association 60 (3): 441–458.
DePaul, Michael R., and William Ramsey. 1998. Rethinking Intuition: The Psychology
of Intuition and Its Role in Philosophical Inquiry. New York: Rowman & Littlefield.
Deutsch, Max. 2009. Experimental philosophy and the theory of reference. Mind and
Language 24 (4): 445–466.
Engel, Mylan, Jr. 1992. Is epistemic luck compatible with knowledge? Southern Jour-
nal of Philosophy 30 (2): 59–75.
Evans, Gareth. 1973. The causal theory of names: Part I. Aristotelian Society suppl.
vol. 47:187–208.
Goldman, Alvin I. 1986. Epistemology and Cognition. Cambridge, MA: Harvard Uni-
versity Press.
Goldman, Alvin I. 2007. Philosophical intuitions: Their target, their source, and
their epistemic status. Grazer Philosophische Studien 74:1–26.
Grice, H. Paul. 1989. Studies in the Way of Words. Cambridge, MA: Harvard Univer-
sity Press.
References 185
Haidt, Jonathan. 2012. The Righteous Mind: Why Good People Are Divided by Religion
and Politics. New York: Pantheon Books.
Hales, Steven D. 2006. Relativism and the Foundations of Philosophy. Cambridge, MA:
MIT Press.
Jackson, Frank. 1986. What Mary didn’t know. Journal of Philosophy 83:291–295.
Jackson, Frank. 1998. Poscript on qualia. In Mind, Method, and Conditionals, ed. Frank
Jackson. London: Routledge.
Jackson, Frank. 2003. Mind and Illusion. Cambridge: Cambridge University Press.
Jackson, Frank. 2006. On ensuring that physicalism is not a dual attribute theory in
sheep’s clothing. Philosophical Studies: An International Journal for Philosophy in the
Analytic Tradition 131 (1): 227–249.
Jackson, Frank. 2011. On Gettier holdouts. Mind and Language 26 (4): 468–481.
Knobe, Joshua. 2003. Intentional action and side effects in ordinary language. Anal-
ysis 63 (3): 190–194.
Kripke, Saul. 1980. Naming and Necessity. Cambridge, MA: Harvard University Press.
Lehrer, Keith, and Thomas Paxson. 1969. Knowledge: Undefeated justified true
belief. Journal of Philosophy 66: 225–237.
Lewis, David. 1983. Philosophical Papers, vol. 1. Oxford: Oxford University Press.
Ludwig, Kirk. 2007. The epistemology of thought experiments: First person versus
third person approaches. Midwest Studies in Philosophy 31:128–159.
Lycan, William G. 2006. On the gettier problem problem. In Epistemology Futures, ed.
Stephen Herrington, 148–168. Oxford: Oxford University Press.
Machery, Edouard, Ron Mallon, Shaun Nichols, and Stephen P. Stich. 2004. Seman-
tics, cross-cultural style. Cognition 92 (3): B1–B12.
Machery, Edouard, Ron Mallon, Shaun Nichols, and Stephen P. Stich. 2013. If folk
intuitions vary, then what? Philosophy and Phenomenological Research 86: 618–635.
Mallon, Ron, Edouard Machery, Shaun Nichols, and Stephen P. Stich. 2009. Against
arguments from reference. Philosophy and Phenomenological Research 79 (2): 332–356.
Nadelhoffer, Thomas, and Eddy Nahmias. 2007. The past and future of experimental
philosophy. Philosophical Explorations 10 (2): 123–149.
Nagel, Jennifer. 2012. Intuitions and experiments: A defense of the case method in
epistemology. Philosophy and Phenomenological Research 85 (3): 495–527.
Neale, Stephen. 2004. This, that, and the other. In Descriptions and Beyond, ed. Anne
Bezuidenhout and Marga Reimer, 68–82. Oxford: Oxford University Press.
Neta, Ram. 2012. Knowing from the armchair that our intuitions are reliable. Monist
95 (2): 329–351.
References 187
Nichols, Shaun. 2004. The folk psychology of free will: Fits and starts. Mind and
Language 19 (5): 473–502.
Nichols, Shaun, and Joshua Knobe. 2007. Moral responsibility and determinism:
The cognitive science of folk intuitions. Noûs 41 (4): 663–685.
Nichols, Shaun, and Joseph Ulatowski. 2007. Intuitions and individual differences:
The Knobe effect revisited. Mind and Language 22 (4): 346–365.
Pew Research Center. 2011. Views of gun control—A detailed demographic break-
down. Pew Research Center, Washington, DC, January 13. http://www.pewresearch
.org/2011/01/13/views-of-gun-control-a-detailed-demographic-breakdown/.
Putnam, H. 1975. Mind, language, and reality. In his Philosophical Papers, vol. 2.
Cambridge: Cambridge University Press.
Saunders, John Turk, and Narayan Champawat. 1964. Mr. Clark’s definition of
“knowledge.” Analysis 25:8–9.
Schwitzgebel, E., and F. Cushman. 2012. Expertise in moral reasoning? Order effects
on moral judgment in professional philosophers and non-philosophers. Mind and
Language 12:135–153.
Searle, John R. 1980. Minds, brains, and programs. Behavioral and Brain Sciences 3 (3):
417–457.
Shulz, E., E. T. Cokely, and A. Feltz. 2011. Persistent bias in expert judgments about
free will and moral responsibility: A test of the expertise defense. Consciousness and
Cognition 20 (4): 1722–1731.
Starmans, Christina, and Ori Friedman. 2012. The folk conception of knowledge.
Cognition 124 (3): 272–283.
Stich, Stephen. 1988. Reflective equilibrium, analytic epistemology, and the prob-
lem of cognitive diversity. Synthese 74:391–413.
Swain, Stacey, Joshua Alexander, and Jonathan M. Weinberg. 2008. The instability
of philosophical intuitions: Running hot and cold on Truetemp. Philosophy and Phe-
nomenological Research 76 (1): 138–155.
Vahid, Hamid. 2001. Knowledge and varieties of epistemic luck. Dialectica 55 (4):
351–362.
Weinberg, Jonathan M., Chad Gonnerman, Cameron Buckner, and Joshua Alexan-
der. 2010. Are philosophers expert intuiters? Philosophical Psychology 23 (3):
331–355.
Weinberg, Jonathan M., Shaun Nichols, and Stephen Stich. 2001. Normativity and
epistemic intuitions. Philosophical Topics 29 (1–2): 429–460.
Williamson, Timothy. 2007. The Philosophy of Philosophy. New York: John Wiley &
Sons.