Documente Academic
Documente Profesional
Documente Cultură
E. J. Lowe
1
1
Great Clarendon Street, Oxford ox2 6dp
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide in
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
Oxford is a registered trade mark of Oxford University Press
in the UK and in certain other countries
Published in the United States
by Oxford University Press Inc., New York
E. J. Lowe 2008
The moral rights of the author have been asserted
Database right Oxford University Press (maker)
First published 2008
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
without the prior permission in writing of Oxford University Press,
or as expressly permitted by law, or under terms agreed with the appropriate
reprographics rights organization. Enquiries concerning reproduction
outside the scope of the above should be sent to the Rights Department,
Oxford University Press, at the address above
You must not circulate this book in any other binding or cover
and you must impose the same condition on any acquirer
British Library Cataloguing in Publication Data
Data available
Library of Congress Cataloging in Publication Data
Data available
Typeset by Laserwords Private Limited, Chennai, India
Printed in Great Britain
on acid-free paper by
Biddles Ltd., King’s Lynn, Norfolk
ISBN 978–0–19–921714–4
10 9 8 7 6 5 4 3 2 1
Preface
I have been interested in the philosophy of action for more than thirty
years. In fact, my very first venture into print—‘Neither Intentional
nor Unintentional’, Analysis 38 (1978), pp. 117–18—was on the subject.
In retrospect, it seems to me that since then I have gradually been
putting together the parts of a complex jigsaw puzzle, which is at last
made complete by the comprehensive account of human action defended
in this book. Initially, I was primarily interested in understanding the
distinction between intentional and unintentional action, largely on account
of its moral significance. This then led me to examine the closely related
notion of voluntariness. Partly through a study of John Locke’s views on
the matter, I became convinced that a volitionist account of the nature
of voluntary action must be correct. Because I was at the same time
becoming increasingly attracted to a version of psychophysical dualism in
the metaphysics of mind—a version that I now call non-Cartesian substance
dualism—I also became interested in trying to overcome a popular line
of argument in favour of physicalism, the argument from causal closure.
Until this point, however, I felt relatively neutral on the question of
free will—neutral, that is, between compatibilist and libertarian responses
to this question—but my neutrality was finally overcome in favour of
libertarianism once I became persuaded that only a thoroughly externalist
account of reasons for action is defensible.
The last piece in the jigsaw puzzle concerns the distinction between
event causation and agent causation. For a long time, I considered that
all causation is fundamentally event causation and that I could happily
accommodate my volitionism within this broader view. More recently,
however, I have come to the conclusion that all causation is fundamentally
substance causation, with voluntary human action constituting a special case
of this. It struck me as being, in effect, a gross category mistake to talk of
events as literally being causes, when causal powers and liabilities manifestly
belong to substances—that is, to persisting, concrete objects—rather than
to events. According to my non-Cartesian substance dualism, human
vi Preface
persons—that is, human agents, for persons are necessarily also agents—are
‘individual substances’ of a distinct and irreducible kind. Following Locke,
I regard the will as a power possessed by such agents, which is exercised
by them whenever they engage in voluntary actions. However, whereas
Locke is at best ambiguous on the question of wherein our freedom of
action lies, I hold it to lie in the fact that our will is a spontaneous power,
which we are able to exercise freely—freely, that is, in the libertarian
sense—in the light of the reasons for action that our senses and intellects
reveal to us. I do not, then, espouse the kind of position in the philosophy
of action that normally goes under the description of ‘agent causalism’,
for this is typically associated with a rejection of volitionism and an
endorsement of the view that a human agent purely qua agent is a cause
of his or her intentional actions—whereas I want to say that an agent
always causes what he or she does voluntarily only by exercising his or her
power of will. Accordingly, my position with regard to voluntary action
steers a middle path between classical agent causalism on the one hand
and, on the other, those versions of volitionism that take all causation to
be, fundamentally, event causation. I explain all of this more fully in the
Introduction.
I have divided the chapters of this book into two Parts, to reflect the
difference between those aspects of my current position that are relatively
neutral concerning the distinction between event causation and agent
causation, the question of free will, and the nature of reasons for action and
those that commit me to the primacy of substance causation, libertarianism,
and a thoroughgoing externalism of reasons. A reader who was convinced
by the dualistic account of mental causation defended in Part I of the
book might not necessarily be persuaded to accept the main doctrines
advanced in Part II, even though I myself now regard the combination as a
natural and compelling one. Many of the chapters have appeared in print in
previous versions, but I have revised and rewritten them for the purposes of
this book in an effort to make it an integrated monograph rather than just
a collection of entirely independent papers. However, because the book’s
chapters were originally independent papers and I didn’t wish to disrupt
the internal thread of argument in each of them, there is inevitably some
repetition of themes and arguments across the book from time to time.
But even where this occurs, it is usually accompanied by some change of
perspective, focus or context, and so will not, I hope, unduly distract the
Preface vii
Various chapters of this book are based upon the following previously
published papers of mine—all of them relatively recent—and I am grate-
ful to the editors and publishers concerned for permission to reproduce
original material from them. I retain the titles of the original papers as
the corresponding chapter titles so as to avoid possible confusion, but in
many cases there are significant changes of content in the present versions.
Chapter 1: ‘Self, Agency, and Mental Causation’, Journal of Conscious-
ness Studies 6 (1999), pp. 225–39. Chapter 2: ‘Causal Closure Principles
and Emergentism’, Philosophy 75 (2000), pp. 571–85. Chapter 3: ‘Physical
Causal Closure and the Invisibility of Mental Causation’, in S. Walter
and H.-D. Heckmann (eds), Physicalism and Mental Causation: The Meta-
physics of Mind and Action (Exeter: Imprint Academic, 2003). Chapter 6:
‘Event Causation and Agent Causation’, Grazer Philosophische Studien 61
(2001), 1–20. Chapter 7: ‘Personal Agency’, in A. O’Hear (ed.), Minds
and Persons (Cambridge: Cambridge University Press, 2003). Chapter 8:
‘Substance Causation, Persons, and Free Will’, in C. Kanzian, J. Quitterer,
and E. Runggaldier (eds), Persons: An Interdisciplinary Approach (Vienna:
Hölder–Pichler–Tempsky, 2003). Chapter 9: ‘Rational Selves and Free-
dom of Action’, in A. Corradini, S. Galvan, and E. J. Lowe (eds), Analytic
Philosophy Without Naturalism (London and New York: Routledge, 2006).
Chapter 10: ‘Needs, Facts, Goodness, and Truth’, in S. Reader (ed.), The
Philosophy of Need (Cambridge: Cambridge University Press, 2006). In
addition, I have based parts of the Introduction on my ‘Between Agent
Causalism and Volitionism: A Middle Path’, in F. Castellani and J. Quitterer
(eds), Agency and Causation in the Human Sciences (Paderborn: Mentis, 2007).
I am grateful to the publisher and editors for permission to do this. I
have also based parts of Chapter 5 on material first published in my ‘Non-
Cartesian Dualism and the Problem of Mental Causation’, Erkenntnis 65
(2006), pp. 5–23. I am grateful to the publisher, Springer, for permission
to do this.
Acknowledgements ix
Most of the material in this book was first presented in the form of
invited conference papers or lectures, as detailed below, and I am indebted
to the organizers and audiences concerned for their encouragement and
help in the genesis and improvement of it. The Introduction draws upon
my paper ‘Between Agent Causalism and Volitionism: A Middle Path’,
which was delivered at a conference on Agency and Causation in the
Human Sciences, held in the University of Trento in June 2005. ‘Self,
Agency, and Mental Causation’ was first presented to an audience in
the University of Stirling in May 1997. ‘Causal Closure Principles and
Emergentism’ was delivered at a conference on The Completeness of
Physics, held in the University of London, School of Advanced Study,
in May 1999. ‘Could Volitions be Epiphenomenal?’ was presented to a
workshop on Epiphenomenalism at the conference of the Gesellschaft f ür
Analytische Philosophie, held in the University of Bielefeld in September
2003. ‘The Self as an Emergent Substance’ was delivered at a conference on
Emergence in Science and Philosophy held in the Catholic University of
Milan in June 2007. ‘Event Causation and Agent Causation’ was presented
to a workshop on Agents and their Actions at the conference of the
Gesellschaft f ür Analytische Philosophie, held in the University of Bielefeld
in September 2000. ‘Personal Agency’ was delivered as a Royal Institute
of Philosophy Lecture in London in February 2002. ‘Substance Causation,
Persons, and Free Will’ was an invited paper presented at the International
Wittgenstein Symposium, held in Kirchberg, Austria, in August 2002.
‘Rational Selves and Freedom of Action’ was presented at a conference on
Analytic Philosophy without Naturalism, held in the Catholic University
of Milan in June 2003. ‘Needs, Facts, Goodness, and Truth’ was presented
at the Royal Institute of Philosophy conference on The Philosophy of
Need, held in the University of Durham in September 2003.
Finally, I should especially like to express my gratitude to Peter
Momtchiloff of Oxford University Press for all his help and encour-
agement and to several anonymous readers for the Press for their careful
and constructive advice. I am also very grateful to Susan Lowe, Tim Lowe,
Maria Alvarez, and Meghan Griffith for commenting on all or part of
a late version of the manuscript. I have made various amendments as a
consequence.
This page intentionally left blank
Contents—Summary
Introduction 1
Part I. Mental Causation, Causal Closure, and Emergent
Dualism 17
1. Self, Agency, and Mental Causation 19
2. Causal Closure Principles and Emergentism 41
3. Physical Causal Closure and the Invisibility of Mental
Causation 58
4. Could Volitions Be Epiphenomenal? 79
5. The Self as an Emergent Substance 92
Bibliography 213
Index 219
This page intentionally left blank
Contents
Introduction 1
1. Some questions and answers 2
2. Event causation and agent causation 3
3. Free action and causation 6
4. Reasons and causes 8
5. An interim summing up 11
6. The challenge of—and to—physicalism 12
7. A brief look ahead 14
Bibliography 213
Index 219
List of Figures
² See further my A Survey of Metaphysics (Oxford: Oxford University Press, 2002), ch. 13.
Introduction 3
event causalist will say that the singular agent-causal sentence ‘The bomb
caused the collapse of the bridge’ is analysable as meaning (something like)
‘Some event involving the bomb caused the collapse of the bridge’—an
existentially quantified event-causal sentence that is entailed by, for instance,
the singular event-causal sentence ‘The explosion of the bomb caused the
collapse of the bridge’. I explain this sort of view much more fully in
Chapter 6.
Now, I wholly reject the foregoing approach, ultimately on the grounds
that events are causally impotent. In my view, only entities in the category
of substance—that is, persisting, concrete objects—possess causal powers
(and liabilities, for that matter).³ Strictly speaking, an event cannot do
anything and so cannot cause anything. For causings are a species of
doings—that is, in a very broad sense, actions—and doings are themselves
happenings. Thus, talk of an event doing something either involves a
gross category mistake—because, understood literally, it implies that one
happening is done by another —or else, taken less seriously, it may be
dismissed as being no more than a misleading manner of speaking. An
example of the latter possibility is provided, perhaps, by a throwaway
remark such as ‘That noise sounded loud’. Here it might be excessively
pedantic to complain that only things, such as bells and horns, can ‘sound
loud’—that is, make a loud sound, or even make a loud-seeming sound —and
that strictly speaking what we should say is that the noise was a loud
sound or, if this better captures the intent of the original remark, that the
noise seemed to be a loud sound. On the other hand, an example of the
former possibility—that of a blatant category mistake—is, I should say,
provided by the statement, expressed with serious and literal intent, ‘Your
collision with the car injured you’. Here I want to protest that it was the
car that injured me, by colliding with me. After all, only the car—not
the collision—had the right sorts of properties to cause me injury, being
massive, rigid, and fast.
Consequently, I want to contend, in direct opposition to the event
causalist’s approach, that the singular event-causal sentence ‘c caused e’
is analysable in terms of an existentially quantified agent-causal sentence,
³ For more on the notion of substance in play here, see my The Possibility of Metaphysics: Substance,
Identity, and Time (Oxford: Clarendon Press, 1998), ch. 6.
Introduction 5
⁴ See George Berkeley, A Treatise Concerning the Principles of Human Knowledge (1710), in his
Philosophical Works, ed. M. R. Ayers (London: Dent, 1975).
⁵ See P. F. Strawson, Individuals: An Essay in Descriptive Metaphysics (London: Methuen, 1959), ch. 3,
and also my Subjects of Experience, ch. 2.
6 Introduction
⁶ See further my article ‘Agent Causation’, in The Encyclopedia of Philosophy, 2nd edn, ed.
D. M. Borchert (New York: Macmillan, 2006).
Introduction 7
⁷ See, for example, C. D. Broad, ‘Determinism, Indeterminism, and Libertarianism’, in his Ethics
and the History of Philosophy: Selected Essays (London: Routledge and Kegan Paul, 1952).
⁸ See further my Subjects of Experience, ch. 5. I should stress that, to keep matters simple, I am focusing
now on cases in which the agent intends to cause what he or she voluntarily causes—thus setting aside
cases, for instance, in which an agent voluntarily causes a limb movement only in the course of carrying
out some more extended action plan, such as walking to the shops. Clearly, we must understand the
more basic cases first, in which an agent deliberately exercises voluntary control over the movements of
his or her limbs—for without an ability to do this, an agent could never acquire an ability to engage in
more extended action routines, which are typically learnt through step-by-step practice (as in learning
to play the piano or to type) and only later become ‘second nature’.
⁹ This claim sets me apart from those philosophers who are ‘classical’ agent causalists in the
sense explained earlier but who also invoke volitions (or something like them) in their theory of
action. Thomas Reid provides a well-known historical example, at least as his position is usually
interpreted. Until recently, modern agent causalists have tended not to favour this sort of position.
However, for a modern version of agent causalism in this mould, see Timothy O’Connor, Persons and
Causes: The Metaphysics of Free Will (New York: Oxford University Press, 2000). For more on Reid’s
theory, see Gideon Yaffe, Manifest Agency: Thomas Reid’s Theory of Action (Oxford: Clarendon Press,
2004)—especially pp. 150–8, where Yaffe in fact queries the now-dominant interpretation of Reid as
being an agent causalist in what I am calling the ‘classical’ sense.
8 Introduction
¹⁰ For more on the distinction between ‘internal’ and ‘external’ relations, see my The Four-Category
Ontology: A Metaphysical Foundation for Natural Science (Oxford: Clarendon Press, 2006), ch. 3.
Introduction 9
more accurately, the onsets of which cause) the agent’s actions.¹¹ I wholly
reject this view. I contend that behaviour that is caused by an agent’s beliefs
and desires is, on that very account, not rational, free action. Why? At bottom,
because causation of this sort is blind to reason—as I shall explain more fully
in Chapter 9. When beliefs and desires cause human behaviour—though
remember, here, that in my view this way of talking is just a façon de
parler —they do not do so because that behaviour is in any way rationally
commendable: rather, they do so simply in virtue of the existence of a general
psychophysical law that is applicable to the human subject in question. (It
may well be that, as Davidson has famously argued,¹² psychophysical laws
could not be ‘strict’—that is, perfectly exceptionless—laws, but I suspect
that at best only the most fundamental laws of physics could have this
status, so I do not regard psychophysical laws as being at all peculiar in
this regard.) The law in question doesn’t obtain for a reason—that is, to
fulfil some rational purpose—but simply as a matter of psychophysical fact.
Thus, I maintain, behaviour that is caused by an agent’s beliefs and desires
is typically either automatic or habitual in character, or else indicative of
a psychopathological condition. The so-called ‘problem of deviant causal
chains’, which besets causal theories of rational action of the foregoing sort,
is in my view just symptomatic of this more fundamental defect in their
very conception.
Query: if the fact that causation is ‘blind to reason’ poses a problem
for mainstream causal theories of rational action, why does it not also
for a volitionist theory which holds that volitions cause the bodily effects
involved in free, rational actions—even accepting that this is just a façon
de parler, meaning that agents cause those effects by willing? Answer: agents
choose which bodily effects to cause and so, although the causing of those
effects proceeds in accordance with psychophysical laws—which obtain
independently of our rational preferences or choices—the spontaneity of
our volitions means that we are able to exploit those laws in order to enact
our choices, rather than being mere slaves to those laws. Note here that
the mental items cited by mainstream causal theorists as the causes of our
¹¹ See Donald Davidson, ‘Actions, Reasons, and Causes’, in his Essays on Actions and Events (Oxford:
Clarendon Press, 1980) and, for further discussion, my An Introduction to the Philosophy of Mind
(Cambridge: Cambridge University Press, 2000), ch. 9.
¹² See Donald Davidson, ‘Mental Events’, in his Essays on Actions and Events. My own view of laws
is developed most fully in my The Four-Category Ontology, part III.
10 Introduction
¹³ See Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2000).
Introduction 11
world: but those states of affairs do not cause the agent to will in this or
that manner. (This is quite apart from the fact that—by my own account
of causation—states of affairs, like events, can in any case only be said as
a mere façon de parler to cause anything.) Being ‘responsive’ to a reason for
acting is not being caused to act in a certain way by that reason. And, most
importantly, it is because free, rational actions are responses to reasons that
they are not merely random events, like the spontaneous decay of a radium
atom. So, at least, I shall argue, especially in Chapter 9.
Here it may be asked: But is this account of free, rational action
compatible with what empirical science reveals to us about the nature of
the physical world, in which we are obliged to act as physically embodied
agents? Does physical science leave any scope for the existence of actions that
are at once uncaused and yet are not, like the spontaneous decay of radium
atoms, merely random or chance events? Well, physical science certainly
does allow for the existence of uncaused, and in that sense spontaneous,
events and it has nothing whatever—either positive or negative—to say
about the realm of reason and rational explanation, since that is a normative
rather than a factual domain. Hence, it is hard to see how the foregoing
account can be deemed to be incompatible with any of the deliverances
of physical science—as opposed to the deliverances of physicalist theories
of mind and action, which are philosophical rather than scientific theories.
Furthermore, if the foregoing account of rational action is a correct account
of what it is to act rationally—and that is a philosophical, rather than a
scientific, matter—then it is hard to see how one could rationally accept
the truth of any scientific theory that was incompatible with it. But as to
the philosophical theories of mind and action with which this account may
be deemed to be in conflict—which we may broadly refer to as being
‘physicalist’ ones—I shall deal with many of the challenges that they seem
to present to my sort of position in Part I of this book. I shall say more
about that in a moment.
5. An Interim Summing Up
The position defended in Part II of this book adopts a middle path between
classical agent causalism and volitionism. It accords to volitions the status
of basic actions, maintains that they are free and spontaneous exercises of
12 Introduction
the two-way power of the will, performed in the light of reason, and
contends that agents are the causal source of all change in the world—with
rational, free agents like ourselves having a special place in the causal order as
unmoved movers, or initiators of new causal chains. Rather than accepting the
notion of event causation as perfectly legitimate in the inanimate domain
and representing agent causation as a sui generis phenomenon restricted to
rational beings, in which an agent as such is a cause, it holds that all causation
is causation by agents, but that agents can only cause things to happen by
acting in suitable ways. This is not to say that their actings, rather than they,
are the causes of the effects thus brought about: it is just to say how, or in
what manner, they themselves bring about those effects. And it says that what
is special about rational agents is that they possess a distinctively rational
power—the power of will or choice.
offering into these distinct parcels. Both parts of the book contain much
that runs counter to prevailing thought in the philosophy of mind and
action, but this is even more true of Part II than Part I. However, I have
said enough by way of introduction to the themes of Part I and it is time
to get down to the business of developing them in detail. I shall detain the
start of that business only to provide, for those who find such things useful,
a brief chapter-by-chapter guide to the rest of the book.
are correct, it should nonetheless appear to be the case, from the perspective
of physical scientists, that all of the causation involved in human behaviour
is purely physical in character. In other words, I explain why mental
causation, thus conceived, should be invisible from such a perspective—the
implication being, of course, that its invisibility should not be regarded
as compelling evidence of its non-existence. In this manner, I hope to
take some of the wind out of the sails of those physicalists who presume
uncritically that ‘science is on their side’.
In Chapter 4, ‘Could Volitions be Epiphenomenal?’, I look at the
claims of some philosophers and philosophically minded psychologists that
volitions or acts of will are merely epiphenomenal, in the sense that they do
not actually play the causal role customarily assigned to them in the genesis
of our intentional physical behaviour. The claims are allegedly supported by
empirical studies supposedly showing that volitions are at best side-effects of
the neurological processes which, according to these theorists, really initiate
and sustain that behaviour. I argue that the empirical evidence in question
not only does not, but could not, support the interpretation favoured by
these theorists, because our very ability to conceive and investigate causal
hypotheses in the sciences is predicated upon the fact that we are beings
capable of actively intervening, at will, in the course of nature. In other
words, the viability of scientific method itself presupposes a conception of
human agency which is at odds with the claims of these theorists, which
they advance as supposed results of the application of that method.
In Chapter 5, ‘The Self as an Emergent Substance’, I explain why I
avoid traditional arguments in favour of substance dualism and offer in their
place a new argument which, however, supports only a non-Cartesian
version of this view, according to which beings like ourselves—persons or,
more generally, subjects of experience —are bearers of both psychological and
physical properties, while nonetheless being distinct from our biological
bodies and any parts of them. I then explain why this view is well-suited to
accommodating the dualistic psychophysical interactionism defended in the
earlier chapters and present a new—and, I think, compelling—argument
against physicalist theories of mind. This completes Part I of the book.
Part II begins with Chapter 6, ‘Event Causation and Agent Causation’,
in which I look at strategies for attempting to reduce one of these species
of causation to the other, coming down in the end in favour of the view
that talk in terms of event causation, to the extent that it should strictly be
16 Introduction
A person or self does not appear to be simply identifiable with his or her
organic body, nor with any part of it, such as the brain—and yet selves
seem to be agents, capable of bringing about physical events, such as bodily
movements, as causal consequences of certain of their conscious mental
states. How is this possible in a universe in which—or so it appears—every
physical event has a sufficient cause that is itself wholly physical? As we
shall see in the course of this chapter, the answer is that this is possible if a
certain kind of naturalistic dualism is true, according to which the conscious
mental states of selves, although not identifiable with physical states of their
brains, are emergent effects of prior physical causes. Moreover, as we shall
also see, mental causation on this model promises to explain certain aspects
of physical behaviour which may appear to be arbitrary and coincidental
from a purely physical point of view.
(3) Every physical event has a set of wholly physical causes which are
collectively causally sufficient for the occurrence of that event—and
rarely if ever is a physical event causally overdetermined.
The apparent inconsistency of this set of claims, or claims very like them,
seems to lie behind the fact that many philosophers feel obliged to reject
one or more of them. Some reject (1), either denying that there is any
such thing as the self, or else identifying it with something bodily, such as
an animal organism or brain. Some reject (2), holding that our experience
of volitional control over our bodies is merely illusory. And some reject
(3), maintaining that the self’s intentional states are non-physical causes of
certain physical events which lack sufficient wholly physical causes. This
appears to have been Descartes’s view. Instead, I shall argue that claims
(1), (2), and (3) are in fact perfectly consistent. Whether all of those claims
are true is another matter—although, clearly, if they are all not only fairly
compelling but also consistent, something is to be said in favour of their
all being true. Now, elsewhere I have argued in defence of claims (1) and
(2) and I continue to be convinced of their truth.¹ Consequently—in view
of the widespread acceptance of claim (3)—I have a vested interest in
establishing the consistency of the three claims.² So, before proceeding, let
me briefly explain why I think that claims (1) and (2) are true.
¹ See my Subjects of Experience (Cambridge: Cambridge University Press, 1996). I shall present further
arguments in their favour in later chapters of the present book.
² For a recent example of a philosopher who endorses claim (3)—and espouses a thoroughgoing
physicalism as a consequence—see David Papineau, Philosophical Naturalism (Oxford: Blackwell, 1993),
p. 22.
Self, Agency, and Mental Causation 21
by a thing makes some real difference to at least part of the space which
that thing occupies. Thus, my sitting qualifies as a physical state of me
because, in virtue of possessing it, I fill out a part of space in a certain way,
rendering that part of space relatively impenetrable by my presence. But
my consciously thinking of Paris has no spatial connotations of this sort
whatsoever, so far as I can see.⁵ In fact, the identity conditions of mental
states would appear to be thoroughly unlike those of physical states—as
unlike them as the identity conditions of physical objects are unlike those
of the natural numbers.⁶ And consequently the thesis that mental states ‘just
are’ (identical with) physical states is simply unintelligible.
A whole generation of philosophers has, alas, mistaken this unintelli-
gible thesis for something much more exciting, namely, a profound truth
which has only now begun to be revealed to us through the advance of
science. (I don’t expect to be able to shake their faith in its intelligibility,
however, any more than one could hope to shake the faith of a dedic-
ated Pythagorean—and so, partly for that reason, I shall try to show, in
Chapter 5, why the thesis cannot in fact be true, without presupposing its
unintelligibility.) Truths of identity simply cannot be exciting in the way
that such metaphysicians fondly imagine, because it can only be intelli-
gible to identify items of the same kind —that is, kinds importing the same
identity criteria for their instances: but the ‘exciting’ identifications—of
physical objects with mathematical objects, or of mental states with phys-
ical states—all violate this principle by trying to identify items of quite
different kinds.
⁵ Compare Colin McGinn, ‘Consciousness and Space’, Journal of Consciousness Studies 2 (1995),
pp. 220–30.
⁶ See, further, my Kinds of Being: A Study of Individuation, Identity and the Logic of Sortal Terms
(Oxford: Blackwell, 1989), pp. 131–3, and my Subjects of Experience, pp. 25–30.
24 Mental Causation, Closure, and Dualism
⁷ Thus I find Galen Strawson’s imaginary example of the ‘Weather Watchers’ highly implausible:
see his Mental Reality (Cambridge, MA: MIT Press, 1994), ch. 9.
Self, Agency, and Mental Causation 25
which events do or do not depend upon which other events.⁸ One’s own
inner mental life does not present a sufficiently independent arena in which
this capacity could be developed, it seems: one needs to be able, as it were,
to get to grips with things outside oneself in order to get any purchase on
the thought that some events stand in causal relations of dependence to one
another whereas others are only accidentally conjoined. (Against me here it
might be urged that a capacity to discriminate perceptually between at least
some causal and non-causal sequences of events could be innate, even in a
completely passive creature, and indeed that there is some empirical evid-
ence for such an innate capacity in human infants. However, it could still
be argued that such a capacity would inevitably be destined to lie dormant
or atrophy in any creature incapable of engaging in active exploration of its
perceptual environment—including here as ‘active exploration’ a creature’s
voluntary direction of its sense organs, such as its eyes, towards stimuli
selected by it for attention.) This general line of reasoning is, I confess,
only very sketchily presented here, but I shall return to it in later chapters.
⁸ Compare G. H. von Wright, Explanation and Understanding (London: Routledge and Kegan Paul,
1971), pp. 69–74. See also Chapter 4 and Chapter 6 below.
26 Mental Causation, Closure, and Dualism
⁹ See, for example, David Owens, Causes and Coincidences (Cambridge: Cambridge University Press,
1992), ch. 1, and Richard Sorabji, Necessity, Cause and Blame: Perspectives on Aristotle’s Theory (London:
Duckworth, 1980), ch. 1.
Self, Agency, and Mental Causation 29
in each case but also in respect of some of the relations of physical causation
which obtain between various physical events which occur in both cases.
Thus, in the non-coincidental case, but not in the coincidental case, the
physical event of the man’s setting off the tripwire occurs and is one of
the causes of his death. And in the non-coincidental case, but not in the
coincidental case, the physical event of the man’s approaching the house
is related by physical causation—via the movement of the tripwire—to
the physical event of the slate’s falling. This is because the common cause
which makes the difference between the coincidental and non-coincidental
cases in our illustration is not only a physical event itself, but also one which
links the causal histories of the immediate causes of the man’s death by
means of a chain of purely physical causation. But matters may be otherwise
if what links the causal histories of the immediate physical causes of some
non-coincidental physical event is a causal chain involving non-physical
events, as I shall now demonstrate.
P P
World 1 World 2
Figure 1.1. Two possible worlds
¹⁰ Since, manifestly, some events do not have exactly the same causes and effects in the two worlds
as I have represented them, I am assuming that it is not an implication of any acceptable principle of
Self, Agency, and Mental Causation 31
transworld identity for events that an event has the same causes and effects in any world in which it
occurs. But I take it that this is relatively uncontroversial, since to say that it is metaphysically impossible
for an event to have had causes and effects other than those which it actually has is to violate Hume’s
principle that there is no metaphysically necessary connection between cause and effect. I disagree,
then, with Sydney Shoemaker’s position as expressed in his ‘Causal and Metaphysical Necessity’, Pacific
Philosophical Quarterly 79 (1998), pp. 59–77.
32 Mental Causation, Closure, and Dualism
states. But there is good reason to suppose that mental causation has some
distinctive features which relate to the intrinsic natures of mental causes.
We have been taking mental causes to be items such as beliefs, desires,
and intentions—in short, intentional states of the self. (The onsets of such
states are events, but beliefs, desires, and intentions are states rather than
events—not that very much turns on the distinction between events and
states in what follows.) Of course, some mental states—such as ‘pure’
sensations (if such there be)—are not intentional states, since they lack any
intentional content (they are not ‘about’ anything, in the way that beliefs
and desires are always ‘about’ something). But I am not really concerned
with such non-intentional mental states at present. One distinctive feature
of mental causation by intentional states is that what is caused by such
states is intimately related to the intentional contents of those states. In
the case of normal voluntary action, movements of the agent’s body have
amongst their causes intentional states of that agent which are ‘about’ just
such movements. For instance, when I try to raise my arm and succeed in
doing so, my arm goes up—and amongst the causes of its going up are,
we may suppose, such items as a desire of mine that my arm should go up.
The intentional causes of physical events are always ‘directed’ upon the
occurrence of just such events, at least where normal voluntary action is
concerned.¹¹ Nothing like this seems to be the case when physical events
or states cause other physical events or states: such purely physical causation
always appears to be ‘undirected’ or ‘blind’.
Notice, however, that although, in normal voluntary action, an inten-
tional state of the agent is ‘directed’ upon an event of the kind which it
causes, it is not ‘directed’ upon the particular event which it causes. When
I try to raise my arm and succeed in doing so, my desire that my arm
should go up is, we may suppose, amongst the causes of the event of my
arm’s going up: but my desire is not that that particular event of arm-rising
should occur, but merely that an event of arm-rising of the appropriate
kind should occur at a certain time, or during a certain interval of time.
This has to be so because, even if I know that my attempt to raise my arm
will succeed, I cannot know in advance which particular event of arm-rising
will occur as a result of my success, since this will depend on factors outside
my knowledge and control, such as the speed with which my nervous
system reacts at the time of the attempt. Consequently, when my arm goes
up as a result of my successfully trying to raise it, what is causally explained
by my desire that my arm should go up is not specifically the occurrence
of this particular event of arm-rising, but rather the obtaining of the general
state of affairs of an event of that kind’s occurring during a certain interval
of time—a state of affairs which happens to be ‘realized’ on this occasion
by this particular event of arm-rising, but which could equally well have
been ‘realized’ by a different particular event of arm-rising, provided it had
been one of a suitable kind and had occurred at the right time.¹²
There is, I believe, a connection between this feature of intentional causal
explanation and the already proposed role of mental causes in rendering
certain of their physical effects non-coincidental. As we have seen, what
qualifies an event as being ‘non-coincidental’ is a fact about the causal
history of that event: the fact that its immediate causes have a common
cause, that is, the fact that its immediate causes do not have independent
causal histories. And I have suggested that when a mental state causes some
physical event, its causal role may be one of rendering that event non-
coincidental, which it can do by rendering non-independent the causal
histories of that event’s immediate physical causes. My further suggestion,
now, is that this feature of the causal role of mental states is intimately
related to the way in which they serve to provide causal explanations of
certain general physical states of affairs and not merely of particular physical
events. By causally connecting what would otherwise be independent
chains of physical causation, I suggest, a mental cause can render the
common effect of those chains non-coincidental and in so doing explain
why an event of that kind occurred, not merely why that particular event
occurred. For it seems that the way in which a mental cause interconnects
chains of physical causation is such as to ensure that the common effect of
such chains is, in the following sense, robust: in all relatively ‘close’ possible
worlds in which some of the physical events in those chains are different
from those of the actual world but the interconnecting mental cause is still
present, their common effect is nonetheless still of the same kind as that of
the actual world—namely, the kind specified by the intentional content of
¹² Not all philosophers like to include ‘states of affairs’ in their ontology—but see D. M. Armstrong,
A World of States of Affairs (Cambridge: Cambridge University Press, 1997)—and those who do
are not necessarily in favour of including general states of affairs. But that is a debate for another
occasion.
Self, Agency, and Mental Causation 37
that mental cause. This suggestion can perhaps best be elucidated by means
of an example.
¹³ On the notion of an ‘event-fusion’, see Judith Jarvis Thomson, Acts and Other Events (Ithaca, NY:
Cornell University Press, 1977), pp. 78–9.
38 Mental Causation, Closure, and Dualism
¹ See H. Bondi, Cosmology, 2nd edn (Cambridge: Cambridge University Press, 1961), ch. XII.
42 Mental Causation, Closure, and Dualism
² See W. D. Hart, The Engines of the Soul (Cambridge: Cambridge University Press, 1988), ch. 9.
³ For further discussion of the (ir)relevance of physical conservation laws to the problem of
psychophysical causation, see my Subjects of Experience (Cambridge: Cambridge University Press, 1996),
pp. 56–63.
⁴ A relatively early version of this form of argument may be found in Christopher Peacocke,
Holistic Explanation: Action, Space, Interpretation (Oxford: Clarendon Press, 1979), pp. 134–43, though
Peacocke’s version does not exactly conform to the pattern I specify below. I raise an objection to
Peacocke’s argument in my ‘Against an Argument for Token Identity’, Mind 90 (1981), pp. 120–1.
⁵ Why should systematic causal overdetermination be ruled out where psychophysical causation
is concerned? Perhaps it shouldn’t be, and this would be a quick way to dispose of causal closure
arguments: see further Eugene Mills, ‘Interactionism and Overdetermination’, American Philosophical
Quarterly 33 (1996), pp. 105–17. But I take it that most interactionist dualists would not wish to resort
to this strategy if possible, as it looks suspiciously ad hoc.
⁶ For more on the distinction between event causation and fact causation, see Jonathan Bennett,
Events and their Names (Oxford: Clarendon Press, 1988), pp. 21ff.
Causal Closure Principles and Emergentism 43
⁷ David Papineau, ‘Mind the Gap’, in James E. Tomberlin (ed.), Philosophical Perspectives, 12: Language,
Mind and Ontology (Oxford: Blackwell, 1998), pp. 373–88: see p. 375.
44 Mental Causation, Closure, and Dualism
¹³ See, for example, John C. Eccles, The Evolution of the Brain: Creation of the Self (London: Routledge,
1989), pp. 187–92.
Causal Closure Principles and Emergentism 47
given event and all of which jointly causally necessitate the occurrence
of the given event. Note that I have included the clause ‘which has a
cause’ in (1F) to allow for the possibility of uncaused physical events,
such as, perhaps, a putative ‘first’ physical event like the so-called ‘Big
Bang’.
Now, the trouble with (1F) is that it is too weak by the standards
laid down earlier, because it does not serve to entail, when conjoined
with premises (2) and (3), the anti-dualist conclusion (4). This, in brief,
is because (1F) fails to take into account the transitivity of causation.¹⁴
To see this, suppose that a certain non-physical mental event M is a
cause of a physical event P, such as a certain bodily movement. But
suppose also that M itself has a sufficient physical cause. Then, clearly,
P may have a sufficient physical cause, in accordance with principle
(1F), namely, a set of physical events which includes the postulated
sufficient physical cause of M. This is perfectly compatible with the non-
overdetermination premise, (3), because when a sufficient physical cause
of an event P causes P by means, inter alia, of causing an intervening
event M which is in turn a cause of P, that sufficient cause and M
do not thereby causally overdetermine P. More generally, an interac-
tionist dualist who espouses some version of emergentism can happily
endorse principle (1F). Such a dualist may consistently maintain that
the universe has evolved (without ‘outside’ influence by supernatural
powers) from a condition in which only physical events existed to one
in which both physical and mental events exist—the latter conceived
as non-physical—while also espousing causal determinism for all events.
On such a view, if we trace back the causal history of each mental
event, we eventually reach a time at which all of its causal antecedents
were wholly physical events, because on this view it is the physic-
al which has ultimately brought the non-physical realm of the mental
into being. A little later on, I shall look in more detail at how psy-
chophysical causation might operate in a dualistic world governed by
principle (1F).
¹⁴ Of course, some philosophers deny that causation is transitive—and I have even queried it myself:
see my ‘For Want of a Nail’, Analysis 40 (1980), pp. 50–2. But this is a minority view and, in any case,
it is strategically inadvisable for a philosopher who believes in physical causal closure to depend upon a
denial of transitivity, given the controversial nature of such a denial.
48 Mental Causation, Closure, and Dualism
t0 P01 P02
t1 P11 P12 M
t2 P
¹⁵ (1G) is very near to a version of the causal closure principle advanced by Jaegwon Kim, namely,
‘Any physical event which has a cause at time t has a physical cause at t’: see his Supervenience and Mind:
Selected Philosophical Essays (Cambridge: Cambridge University Press, 1993), p. 280 and, for a slightly
different wording, p. 360.
Causal Closure Principles and Emergentism 49
¹⁶ The physicalist may want to deny the possibility of simultaneous causation on the grounds
that it conflicts with the Special Theory of Relativity (STR). However, despite the almost universal
acceptance of STR by physical scientists, its credentials are certainly not beyond question: see, notably,
Michael Tooley, Time, Tense, and Causation (Oxford: Clarendon Press, 1997), ch. 11.
¹⁷ See Peacocke, Holistic Explanation, pp. 136–7. Peacocke’s scenario differs from that of Figure 2.1
in that, in his, P12 on its own is a sufficient cause of P, but this difference is irrelevant for present
purposes. It must be observed, however, that Peacocke also makes the following assumption about
P12 and P (again replacing his labelling by mine): ‘[W]e have a complete and wholly physical account
of ... the causal route from P12 to P in neurophysiological terms [and] this account completely explains
how the event P12 causes P’ (p. 134). Clearly, this assumption is much stronger than anything that
can be derived from the causal closure principle (1G). Indeed, on the most natural interpretation of
Peacocke’s assumption it simply rules out by definition the sort of situation depicted in Figure 2.1,
because in the latter there is a non-physical causal route from P12 to P in addition to a physical one.
But then one wonders why Peacocke bothered to discuss this sort of situation and object to it on the
grounds that it involves overdetermination.
50 Mental Causation, Closure, and Dualism
signal to me to put a lighted match to the bonfire at the same time; then it is true
that if you had not pressed the button, I would not have put the lighted match to
the bonfire, but this cannot show that the resulting fire was not overdetermined.
and that these physical events are the ultimate effects of two distinct physical
causal chains. From a purely physical perspective, these chains might appear
to be independent, with the consequence that P might appear to occur
by coincidence. But the truth might be that a mental event M served to
link those two physical causal chains, rendering P non-coincidental—as is
depicted in Figure 2.2 (compare Figure 1.1 of Chapter 1):
t0 P01 P02
t1 P11 M P12
t2 P21 P22
t3 P
Figure 2.2. Non-coincidence through mental causation
would be, in another sense, incomplete and would falsely represent the
occurrence of P as being coincidental. My suggestion is that the world of
Figure 2.2 may in fact be our world. If it is our world, physical science
can present us with the semblance of a complete explanation of our bodily
movements, and yet it will be an explanation which leaves something out,
giving our bodily movements the appearance of being coincidental events
arising from independent causal chains of events in our brains and nervous
systems. But isn’t that precisely what current physical science does appear
to do? As it traces back the physical causes of our bodily movements into
the maze of antecedent neural events, it seems to lose sight of any unifying
factor explaining why those apparently independent causal chains of neural
events should have converged upon the bodily movements in question.
In short, it leaves us with a kind of ‘binding’ problem, not unlike the
‘binding’ problem associated with conscious perceptual experience (the
problem of explaining how widely distributed and apparently unrelated
neural events in the cortex can support the existence of unified perceptual
experiences).¹⁸
¹⁸ For a description of the more familiar ‘binding’ problem, see Valerie Gray Hardcastle, ‘Psychology’s
Binding Problem and Possible Neurobiological Solutions’, Journal of Consciousness Studies 1 (1994),
pp. 66–90.
¹⁹ Something like (IH) is suggested by the following remark of Jaegwon Kim’s: ‘One way of stating
the principle of physical causal closure is this: If you pick any physical event and trace out its causal
ancestry or posterity, that will never take you outside the physical domain. That is, no causal chain
will ever cross the boundary between the physical and the non-physical.’ See Jaegwon Kim, Mind in a
54 Mental Causation, Closure, and Dualism
Physical World: An Essay on the Mind–Body Problem and Mental Causation (Cambridge, MA: MIT Press,
1998), p. 40. Of course, another reading of this remark might equate it with the excessively strong
principle (1E).
²⁰ For an earlier development of this conception of mental causation, see my Subjects of Experience,
pp. 64–71 and 82ff. There are some resemblances between my proposal and one made by Fred Dretske:
see his ‘Mental Events as Structuring Causes of Behaviour’, in J. Heil and A. R. Mele (eds), Mental
Causation (Oxford: Clarendon Press, 1993). I should stress that I don’t want to commit myself here to
the truth of this proposal, only to its coherence. Indeed, in Chapter 5 I shall endorse instead a rather
different way of accommodating a causal closure principle along the lines of (1H). It is still worth
examining the present proposal, however, since it reveals one more way in which physicalist dogma
can be challenged by dualists. It is also worth remarking that I am, of course, by no means committing
myself to the truth of principle (1H), only exploring ways to accommodate it within a dualistic theory
of mind.
Causal Closure Principles and Emergentism 55
principle why such a causal fact should not have a causal explanation and,
indeed, there may be a positive reason for seeking such an explanation,
as I shall explain later. Now, if a mental event M causes it to be the
case that certain physical events, P1 , P2 , ... Pn , have a certain physical
effect, P, then, it seems clear, M is itself a cause of P —but not an
immediate cause of P, nor an immediate cause of any of the physical events
P1 , P2 , ... Pn . In fact, it seems clear that M is not included in P’s transitive
causal closure and hence that M can be non-physical without violating
principle (1H).
There is nothing unintelligible in principle about the notion of a mental
event causing a physical causal fact, as the following theological example
shows. Suppose that principle (1H) is true in a world in which every
physical event has a sufficient physical cause at every time at which it has a
cause and no physical causation is either simultaneous or backward. Such a
world can have no beginning in time, because it can have no first physical
event. (Suppose it did have a first physical event. By hypothesis, that event
would have a sufficient physical cause, which would have to be an earlier
physical event or conjunction of such events, contradicting the supposition
just made.) And yet we could still ask of this world why it should exist or
be actual rather than any other. One intelligible answer would be to say
that this world was actual because God had chosen it to be actual. God’s
choice, then, would have caused it to be the case that a world containing
certain physical causal facts was actual—and this would be mental causation
of physical causal facts.
What is envisaged here, it must be emphasized, is not God’s causing
certain initial physical events, which then go on to cause other physical
events in accordance with causal laws chosen by God—for in the envisaged
world there are no ‘initial’ physical events, as the world has no beginning
in time. Now, of course, it may be asked when God’s choice was made.
If it was made at any time, it would seemingly have to have been made
earlier than the occurrence of any physical event. That is not impossible,
however, since a physical world which has no beginning in time—no
first physical event—need not have a past of infinite duration, any more
than a line of finite length need have a first or last point. On the other
hand, if God’s choice was not made at any time, because He is conceived
to be eternal or timeless, then it is true that we cannot say that His
choice was an event —but that in no way detracts from the fact that the
56 Mental Causation, Closure, and Dualism
²¹ See, further, G. E. M. Anscombe, ‘Causality and Determination’, in her Metaphysics and the
Philosophy of Mind: Collected Philosophical Papers, Volume II (Oxford: Blackwell, 1981) and Nancy
Cartwright, Nature’s Capacities and their Measurement (Oxford: Clarendon Press, 1989).
Causal Closure Principles and Emergentism 57
¹ For the historical background, see R. S. Woolhouse, ‘Leibniz’s Reaction to Cartesian Interaction-
ism’, Proceedings of the Aristotelian Society 86 (1985/86), pp. 69–82, and D. Garber, ‘Mind, Body, and
the Laws of Nature in Descartes and Leibniz’, Midwest Studies in Philosophy 8 (1983), pp. 105–33.
60 Mental Causation, Closure, and Dualism
to contend that the immaterial mind is a source of, and equally a sink
for, energy, which, he might allege, can be converted from material to
immaterial forms and vice versa, just as gravitational potential energy can be
converted to thermal energy or mass-energy to kinetic energy (the latter in
conformity with Einstein’s famous equation, E = mc 2 ).² The possibility of
defensive moves like these, even if they were not in fact exploited by any
of the early modern protagonists in the mind–body debate, demonstrates
that that debate, even in its early modern form, was not straightforwardly
resolvable by appeal to empirical considerations.
In its contemporary form, the mind–body debate involves relatively
little discussion of the role of the conservation laws, perhaps in implicit
recognition of the fact that the debate cannot be conclusively resolved
against interactive dualism by appeal to such laws. What has taken the
place of those laws in the contemporary debate is a higher-level principle,
or more strictly a family of such principles, invoking a notion of the
‘causal closure’ of the physical. And, just as it may once have seemed that
protagonists in the early modern debate could cite the conservation laws
as part of an effective empirical argument against interactive dualism, so it
seems to many present-day physicalists that they can argue convincingly
against interactive dualism, on empirical grounds, by appeal to a suitably
chosen principle of physical causal closure. In my view, the present-day
physicalists delude themselves in this quite as much as the early modern
opponents of Cartesian dualism did in appealing to the conservation laws.
Philosophical history is doomed to repeat itself, it seems.
Before we turn to the contemporary debate, however, I may still need
to convince some sceptics that Descartes need not, had he lived longer,
have felt at all disconcerted by the objections to his system that were
based on appeals to the conservation laws. Suppose that Descartes had
postulated, in order to explain the change in momentum undergone by the
‘animal spirits’, the existence of a distinct ‘mental force’. The postulation
of such a force would, no doubt, have sat uncomfortably alongside certain
other features of his philosophy of nature, but had Descartes lived to
see the success and general acceptance of Newton’s theory of gravitation,
perhaps he could have urged, with some justice, that such a ‘mental force’
² For a latter-day defence of this idea, see W. D. Hart, The Engines of the Soul (Cambridge: Cambridge
University Press, 1988), ch. 9. I discuss this and related issues more fully in my Subjects of Experience
(Cambridge: Cambridge University Press, 1996), ch. 3.
62 Mental Causation, Closure, and Dualism
to have had. This doesn’t mean that I think that the mind–body problem
is necessarily insoluble, just that we are not going to make headway with it
by pursuing the sorts of considerations that have figured so prominently in
recent times.
The contemporary debate, as I have been calling it, centres on the
strategy, advanced by many physicalists, of defeating interactive dualism
by appealing to two sorts of principle: a principle of the causal closure
of the physical world, and a principle of the absence of systematic causal
overdetermination of physical by both physical and non-physical events.
Both of these sorts of principle have to be carefully formulated so as
not to beg the question against the dualist while at the same time not
relinquishing sufficient strength to constitute a valid argument against the
dualist’s position. Of the many closure principles that have been or might
be advanced, the following is perhaps as good as any by these standards:³
(CCP) For any physical event e, if e has a cause at time t, then e has a
wholly physical sufficient cause at t.
(One or more events constitute a sufficient cause of an event e just in case
the conjunction of those events causally necessitates the occurrence of e.)
As for a principle of non-overdetermination, the following might appear
to do equally well by those standards:
(NOP) Most physical events e are such that, if e has a mental cause at
time t, then e does not also have a wholly physical sufficient
cause at t which is wholly distinct from that mental cause.
(Note that ‘wholly distinct’ in (NOP) is being used in the standard
mereological sense, so that two events are ‘wholly distinct’ just in case
there is no event that is a common part of both of those events. Note
too that (NOP) is provably equivalent to the following principle, which
accordingly may be regarded as being merely a verbal variant of it: (NOP#)
‘Most physical events e are such that, if e has a wholly physical sufficient
cause at time t, then e does not also have a mental cause at t which is wholly
³ Compare Jaegwon Kim, Supervenience and Mind: Selected Philosophical Essays (Cambridge: Cambridge
University Press, 1993), p. 280 and p. 360. Apart from the explicit reference to time, (CCP) is very
close in meaning to a principle that David Papineau calls the principle of ‘the completeness of physics’,
namely, ‘All physical effects have sufficient physical causes’: see David Papineau, ‘Mind the Gap’, in
James E. Tomberlin (ed.), Philosophical Perspectives, 12: Language, Mind and Ontology (Oxford: Blackwell,
1998), pp. 373–88, at p. 375.
64 Mental Causation, Closure, and Dualism
distinct from that physical cause.’) It would seem that we can conclude
from (CCP) and (NOP) that many physical events are such that, if they
have mental causes at any time, those causes are identical with certain
physical causes—and this counts against any form of interactive dualism,
according to which, we may suppose, mental causes are never identical
with physical causes.
How does the argument from (CCP) and (NOP) proceed, exactly? Like
this. Suppose that e is a physical event which has a mental cause at a time
t and that e is like most such events according to (NOP), so that if e has
a wholly physical sufficient cause at t, then that wholly physical sufficient
cause is not wholly distinct from the mental cause in question and thus
includes that mental cause as a (proper or improper) part. (Here I am
assuming, because it is the simplest case, that the mental cause in question
is a mereologically simple event and so does not have any events as proper
parts: but the argument could easily be reformulated to accommodate
complex mental causes.) In that case, the mental cause, in being a part of
a wholly physical cause, must clearly be a physical cause itself. However,
given that e does have a mental cause at t, (CCP) implies that e does indeed
have a wholly physical sufficient cause at t. Hence, we can conclude that
the mental cause in question is indeed a physical cause. The interactive
dualist must allow, of course, that at least some physical events, such as e,
have mental causes, since that is a central plank of his theory. Consequently,
if (CCP) and (NOP) are correct, interactive dualism is untenable, because
it is forced to concede that at least some mental causes are physical causes,
contrary to its explicit denial of this. (An alternative way to construct the
argument, appealing to the variant of (NOP) mentioned earlier, (NOP#),
is as follows. Suppose again that e is a physical event which has a mental
cause at a time t —something that the interactive dualist should be happy
to assume. Since e is a physical event which has a cause at t, (CCP) implies
that e has a wholly physical sufficient cause at t. However, (NOP#) then
implies that e does not have a mental cause at t which is wholly distinct
from e’s wholly physical sufficient cause at t. And from this it follows once
more that the mental cause that e has at t is a physical cause, contrary to
interactive dualism.)
It would seem, on the face of it, that this argument meets exactly the
physicalist’s needs. Neither of the premises, (CCP) and (NOP), appear to
beg the question against the interactive dualist. That is to say, neither of the
Physical Closure and Mental Causation 65
premises is such that it, when conjoined with other assumptions that the
interactive dualist is disposed to accept (such as that at least some physical
events have mental causes), entails the falsity of interactive dualism. At
the same time, the premises, taken together, when conjoined with other
assumptions that the interactive dualist is disposed to accept, do entail the
falsity of interactive dualism, as we have just seen. If that were all there
was to be said, of course, then we could just conclude that interactive
dualism is false and close once and for all this particular debate. But now we
have to consider how premises (CCP) and (NOP) might be defended. It is
obviously not enough, for our acceptance of them as true, or as persuasive
in an argument against interactive dualism, that neither of these premises
begs the question against interactive dualism.
⁴ For a fuller description and discussion of this example, see, for instance, Michael Lockwood, Mind,
Brain and the Quantum: The Compound ‘I’ (Oxford: Blackwell, 1989), ch. 12.
66 Mental Causation, Closure, and Dualism
death has a wholly physical sufficient cause at t, then that cause will include
other physical events at t besides the decay of the radium atom—events
such as (if we can call it an ‘event’) the presence of the vial containing
cyanide. And at least some of these events will presumably themselves have
prior physical causes, even if the decay of the radium atom does not. So
the cat’s death will have a physical cause at some time earlier than t: and
yet, at that earlier time, it will not, apparently, have a sufficient physical
cause, because physical events occurring at that earlier time will have been
nomologically consistent both with the decay of the radium atom and with
its non-decay, and so consistent both with the cat’s death and with its
survival.
Some physicalist philosophers will say that none of this matters very
much, either because quantum physics is irrelevant to the issue of mental
causation, or because (CCP) can be reformulated in a probabilistic manner
which accommodates the indeterminacy of quantum physics. The first sort
of response is relatively feeble, however, since no one at present really
has much idea at all how far quantum physics might have a bearing on
questions of mental causation. All we can say with any confidence, indeed,
is that quantum physics aspires to offer a general explanatory framework
for all physical phenomena, not just physical phenomena which occur on
the very small scale. Any attempt to segregate physical phenomena in a
principled way into those that are ‘small-scale’ and submit to the principles
of quantum physics and those that are ‘large-scale’ and do not submit to
those principles is doomed to failure, as the very example of Schrödinger’s
cat demonstrates: for, by any standard, a radium atom is a ‘small-scale’
phenomenon and a cat is a ‘large-scale’ phenomenon, and yet in this case
we have a single physical system embracing them both and subject to the
principles of quantum physics.
The other sort of response canvassed a moment ago was to propose
a reformulation of (CCP) in probabilistic terms, according to which we
should speak of antecedent physical events ‘fixing the chances of’, rather
than ‘causing’, subsequent physical events.⁵ A difficulty with this line of
thought, however, is that (NOP) will obviously have to be modified
correspondingly in order to preserve the validity of the argument against
⁵ For this sort of formulation, see, for example, David Papineau, Philosophical Naturalism (Oxford:
Blackwell, 1993), p. 22.
Physical Closure and Mental Causation 67
at some earlier time t at which it has a cause. For instance, suppose that I
freely decide to move my finger and my decision is a cause of my finger’s
moving at a subsequent time. Because my decision was free, it did not
have a sufficient cause at any earlier time t.⁶ No doubt the finger’s moving
had some causes at t, but it cannot have had a sufficient cause at t, since if
it had then my decision could not both have been a cause of the finger’s
moving and have lacked a sufficient cause at t —unless, of course, the
finger’s moving was causally overdetermined, in which case, although my
decision was free, I none the less did not freely move my finger, because the
finger was caused to move independently of my free decision. Of course,
‘compatibilists’ will not be impressed by this sort of reason for rejecting
(CCP). But it should be emphasized that our current concern is with the
truth or falsity of interactive dualism, and it would surely be a weakness
in any attack on that doctrine that it should have to presuppose the truth
of compatibilism. After all, it is not as though adherents of a libertarian
conception of free will are automatically committed to interactive dualism,
even if most libertarians probably are dualists. (Certainly, my own support
of libertarianism is independent of my support of dualism.)
⁶ Compare John R. Searle, Rationality in Action (Cambridge, MA: MIT Press, 2001), ch. 3. I myself
shall be arguing extensively for this thesis in Part II of this book.
⁷ For more on the notion of ontological dependency in general, see my The Possibility of Metaphysics:
Substance, Identity, and Time (Oxford: Clarendon Press, 1998), ch. 6.
Physical Closure and Mental Causation 69
to share any event with any physical event as a common part, because any
such common part would have to be an event that was both mental and
physical, contrary to the claim that mental events are never identical with
physical events.)
This seems to me to be a perfectly fair objection on the part of non-
reductive physicalists against those reductivists who seek to appeal to a
principle like (NOP) to support their position. However, it may not be
immediately apparent how an interactive dualist could hope to exploit
the same sort of objection, because in the hands of the non-reductive
physicalist the objection requires one to explain the non-independent
character of the causal overdetermination involved in mental causation
in terms of the systematic obtaining of a ‘realization’ relation between
mental and physical events, and the dualist is committed to denying that
any such relation obtains. How can the dualist maintain that systematic,
non-coincidental causal overdetermination may be a widespread feature
of situations involving mental causation, without conceding that mental
events are ontologically dependent on physical events? The answer is
remarkably simple: he may do so by maintaining that mental events, while
not ontologically dependent on physical events, are causally dependent on
them in certain ways. This is hardly any sort of concession on the part of
the dualist, whose position is, after all, an interactive one, which allows that
mental events can be both causes and effects of physical events.
More specifically, what the interactive dualist may urge is the following.
Suppose, contrary to (NOP), that a certain physical event e has a mental
cause at time t and that e also has a wholly physical sufficient cause at t which
is wholly distinct from that mental cause. Does it follow that the mental
cause and the wholly physical sufficient cause must be entirely independent
of one another, as in the case of the two assassins’ shots? No—and not
just because, as the non-reductive physicalist would maintain, the mental
cause may be ‘realized by’ one or more physical events that are part of
the wholly physical sufficient cause. For it may be that the mental cause
is a cause of the physical event e while also being an effect of one or
more of the physical events that constitute the wholly physical sufficient
cause of e. Now, because the mental event in question occurs, we have
assumed, at t, which is the same time as the wholly physical cause of e has
been assumed to occur, it follows that in order for this proposal to work,
simultaneous causation between a physical event and a mental event must be
Physical Closure and Mental Causation 71
t0 P01 P02
t1 P11 P12 M
t2 P
Figure 3.1. A dualistic counterexample to (NOP)
72 Mental Causation, Closure, and Dualism
allow this sort of situation to occur commonly. And yet, of course, the
situation is quite unlike that of the independent assassins, because M is not
causally independent of the conjunction of P11 and P12 in the way in which
the two assassins’ shots are causally independent of one another. (Notice,
incidentally, that, in the situation depicted in Figure 3.1, P12 is a cause of
P in two different ways, both via M and via a purely physical causal route.
However, there is nothing particularly unusual about an event being a
cause of another event via two different routes. For example, by a single act
of touching a match to two different fuses which lead to the same bomb, I
may cause the bomb to explode via two different routes. Another example:
an avalanche causes the destruction of a house, partly by causing a lot of
snow to fall on the house and partly by causing a tree to fall on the house. I
should also perhaps remark that the situation depicted in Figure 3.1 is not,
of course, one that would be congenial to a libertarian dualist: however,
the truth or falsehood of libertarianism is not an issue with which I am now
concerned, even though I personally incline towards libertarianism.)
The situation depicted in Figure 3.1 is a case of ‘causal overdetermina-
tion’, at least in one sense of that phrase, because M is a cause of P at t1
even though there is another sufficient cause of P at t1 of which M is not a
part. However, this is not to imply that M is a ‘redundant’ cause of P, that
is, that, in the circumstances, even if M had not occurred, P would still
have occurred, being caused by the conjunction of P11 and P12 without
the mediation of M.⁸ For it is perfectly legitimate to postulate that, in
the situation depicted, P11 and P12 only succeed in causing P because P12
succeeds in causing M which in turn helps to cause P. That being the case,
if M had not occurred, then the conjunction of P11 and P12 , even if it had
occurred, would not have sufficed to cause P. In this respect, the situation
is again quite different from that of the two assassins, for in that case, we
suppose, each of the shots would have sufficed to bring about the death of
the victim in the absence of the other shot. (The situation is also different
in this respect from the one described a moment ago, in which I cause a
bomb to explode by touching a match to two different fuses, at least on
the assumption that each fuse functions independently of the other.)
⁸ That is why, in discussing the same situation in Chapter 2, where it is depicted by Figure 2.1, I
described it there as not involving ‘causal overdetermination’ (in another and, very arguably, more
intuitive sense of that phrase). So there is no conflict between my interpretations of the case in this and
the previous chapter.
Physical Closure and Mental Causation 73
been leaking’ and ‘If the floor had not been wet, then, even if the roof had
been leaking, the carpet would not have been ruined’. However, this is not
the place to enter into a long digression on the semantics of counterfactuals.
Suffice it to say that anyone who holds (2) to be true in the situation
depicted in Figure 3.1 should in any case agree with me that, if M had not
occurred, then P would not have occurred, so that M is not redundant,
which is the crucial point at issue.
It may be wondered whether this can really be true. After all, the scientist
may be able to interrogate the person whose brain he is investigating and
elicit from that person reliable testimony concerning the occurrence of
mental event M at time t1 . He may consequently come to believe, on
good grounds, that mental event M occurred simultaneously with physical
events P11 and P12 , which he knows constitute, together, a wholly physical
sufficient cause of P. Doesn’t this entitle him to believe that either
M played no causal role whatever in bringing about P —that M was
‘epiphenomenal’—or else that M was either identical with, or at least
‘realized by’, one or other of the physical events P11 and P12 or their
conjunction? The answer is clearly ‘No’, because the situation depicted in
Figure 3.1 rules out all of these options and yet is metaphysically perfectly
possible. At best the scientist has discovered a ‘correlation’ between the
mental event M and one or more of the brain events that he is investigating.
But such a correlation is consistent with various different relationships
between the mental event and the brain events concerned: identity and
‘realization’ are two such possible relationships, but causation is another.
I should perhaps emphasize that, of course, the situation depicted in
Figure 3.1 is a particularly simple one, much simpler than anything that
one would expect to find in a realistic case of mental causation. But the
crucial features of the situation depicted in Figure 3.1 could all be possessed
by more complicated situations: in fact, in more complicated situations,
it could only be more, not less, easy to confuse different possible sources
of ‘correlation’ between mental and physical events. And, of course, we
have no particular reason to suppose that this sort of situation is purely
hypothetical: it may, for all we presently know, be the sort of situation that
actually obtains with human minds and brains.
⁹ See his ‘Some Evidence for Physicalism’, in S. Walter and H.-D. Heckmann (eds), Physicalism and
Mental Causation: The Metaphysics of Mind and Action (Exeter: Imprint Academic, 2003).
Physical Closure and Mental Causation 77
is that we just have no right to suppose that reality operates along the most
‘economical’ lines—that every effect is always brought about in the simplest
possible way. (After all, the ‘simplest’ way to explain human behaviour
would no doubt be to eliminate reference to the mental altogether and
regard human beings as mere ‘zombies’.) It is true that it would be
extravagant to postulate ‘additional’ non-physical causes of physical events
quite generally, but, as I pointed out earlier, in the mind–body case we
start out with an initial intuition that mental events really do occur and
are completely different from physical events, so that, unless it can be
shown that dualistically conceived mental causation is either incoherent
or incompatible with the empirical evidence, it is perfectly reasonable
to suppose that interactive dualism may well in fact be true. Melnyk
also objects that my sort of dualist proposal is ‘very implausible’ and that
the mere metaphysical (or, as he calls it, logical) possibility of its truth
is insufficient to make it a hypothesis worthy of serious consideration.
However, plausibility is very much in the eye of the beholder and, as I
have just remarked, it is not as though interactive dualism is simply the
product of unmotivated speculation. On the contrary, if anything it has a
good claim to being the ‘default’ position in the philosophy of mind. As
for Melnyk’s charge that the kind of scheme that I propose involves some
sort of ‘inexplicable coincidence’, I have already explained in what sense
my proposed scheme precisely avoids the involvement of coincidence in
mind–body relations, and it is not at all clear to me what scope there is
for pressing this charge any further. If the suggestion is that coincidence
now emerges at the level of the laws governing causal interactions in the
mind–body system, I can only say that I don’t understand in what sense
there can be a ‘coincidence’ at this level. (Here it should also be borne in
mind that the scheme that Melnyk criticizes differs in important matters of
detail from the scheme that I am now proposing.)
Again I must emphasize that the situation depicted in Figure 3.1, and
other more complicated situations relevantly like it, are metaphysically possible
and are, for the reasons given earlier, of such a kind that, to investigators
equipped only with the empirical means to detect the occurrence of and
causal relations between purely physical events, the mental events involved
in them and their causal roles in helping to bring about physical events
would be perfectly ‘invisible’. One important lesson of this is that we cannot
necessarily expect any empirical findings relayed to us by neurophysiologists
78 Mental Causation, Closure, and Dualism
¹⁰ Compare Colin McGinn, The Problem of Consciousness (Oxford: Blackwell, 1991). I don’t mean to
imply that I favour such pessimism myself. Indeed, I shall try to show in Chapter 5 that, as far as the actual
world is concerned, there do in fact exist empirical considerations which, taken together with some
metaphysical ones, provide the materials for a compelling argument against physicalist theories of mind.
4
Could Volitions Be
Epiphenomenal?
The ‘argument from physical causal closure’ for the identity of mental with
physical events and states has as its premises (1) a principle of physical causal
closure, (2) a prohibition on the systematic causal overdetermination of
physical events and states, and (3) the presumption that mental events and
states have physical effects. By rejecting the third premise, epiphenomen-
alism can escape the conclusion of this argument. However, I have doubts
about the merits of the argument from physical causal closure in any case.
More specifically, I suspect that any version of the principle of physical
causal closure strong enough to secure the validity of the argument will
effectively beg the question against interactive dualism, or at least will make
a claim stronger than any that can realistically be alleged to receive empirical
support from current scientific knowledge. However, I do not want to
rehearse these doubts here, since I have already aired them extensively in the
preceding three chapters of this book. I simply shall put my cards on the table
by saying that I favour a version of interactive psychophysical dualism as the
most plausible and satisfactory theory of the mind–body relationship, taking
all relevant considerations, both metaphysical and empirical, into account.
Although I believe that epiphenomenalism lacks an adequate motivation,
because I think that the argument from physical causal closure can be
rebutted in more satisfactory ways than by rejecting the causal efficacy of
the mental, this is not yet to say that I think that epiphenomenalism could not
be true. Nor shall I attempt to argue against epiphenomenalism in the usual
fashion by arguing, for instance, that we could have no knowledge of mental
events and states if epiphenomenalism were true. I have no sympathy for
the sort of causal theory of knowledge that needs to be invoked for the latter
purpose. Rather, I begin by observing that if epiphenomenalism as a general
theory of the mind–body relationship were true, it would, of course, have
to be true with regard to all mental events and states. But, while it seems
perfectly coherent to suppose its truth with regard to sensory, cognitive,
and affective events and states—pains, beliefs, desires, and emotions, for
example—there is one class of mental events which appear to present a
much more serious obstacle. The mental events in question are volitions,
or ‘acts of will’—that is, what we might call conative events. The reason for
this should be obvious. While it is true that functionalism assumes that all
mental events and states have causal roles, taking those roles to be at least
partly constitutive of the nature of such events and states, functionalism is a
controversial theory, even if it is still the mainstream view in contemporary
Could Volitions be Epiphenomenal? 81
¹ For an excellent modern defence of volitionism, see Carl Ginet, On Action (Cambridge: Cambridge
University Press, 1990). I defend my own version of volitionism in my Subjects of Experience (Cambridge:
Cambridge University Press, 1996), ch. 5. I might remark that while a good many contemporary
philosophers of action are happy to talk about conative events in terms of trying, most of them would
repudiate any identification of trying with willing, as the latter has classically been conceived.
82 Mental Causation, Closure, and Dualism
² See, for example, Daniel M. Wegner, The Illusion of Conscious Will (Cambridge, MA: MIT Press,
2002).
Could Volitions be Epiphenomenal? 83
some brain-damaged subjects report that their hand or foot is acting ‘with a
will of its own’. And, on the other, sufferers of certain psychopathological
conditions imagine that they are exercising control over events which are
causally quite independent of them. Indeed, one does not have to be in
any way abnormal to be subject to occasional illusions of control. The
psychologist Daniel Wegner provides a nice example drawn from his own
experience: idly fiddling with a video game in a toy store one day, he
thought for a while that by jiggling the joystick he was making one of the
screen characters perform various antics—until he realized that the game
hadn’t even started and he was merely watching a ‘demo’ game.³
Such empirical data allegedly show that human actions and human
experiences of volition are ‘doubly dissociable’: someone can be acting in
a certain way even though he or she has no experience of exercising their
will to act in that way, and someone can have the experience of exercising
their will to act in a certain way even though they are not in fact acting
in that way. Some of the actual examples of such cases can be intriguing
and unsettling, but really it is not at all surprising that they can occur. They
go no way at all toward showing that, even in normal circumstances, our
volitions do not really cause the physical actions that we intend to cause
when we have the experience of exercising our will. If volitions are causes
of physical actions, then of course it must be possible to have a volition
without a corresponding physical action, or a physical action without a
corresponding volition—at least, on the standard assumption that causes
and effects are logically independent of one another. In fact, one of the
reasons why some philosophers are sceptical about the very existence of
volitions is that they think that volitions couldn’t be logically independent
of actions in the way they would be required to be if they were causes
of actions: this is the basis of the infamous ‘logical connection’ argument
against volitions.⁴ It is ironic, then, that these philosophers dismiss the
existence of volitions on the grounds that volitions and actions couldn’t
‘come apart’ as they would have to for the former to be causes of the latter,
while the psychologists appeal to the fact that volitions and actions can
‘come apart’ to support their view that volitions are not causes of actions.
The right thing to say, it seems to me, is that volitions and actions can
indeed ‘come apart’ and that is why the former are eligible as causes of
the latter.
⁵ See Benjamin Libet, ‘Unconscious Cerebral Initiative and the Role of Conscious Will in
Voluntary Action’, Behavioral and Brain Sciences 8 (1985), pp. 529–66, reprinted in his Neurophysiology of
Consciousness (Boston, MA: Birkhäuser, 1993).
Could Volitions be Epiphenomenal? 85
position of a revolving point of light. Thus, they are required not to exercise
their will in the light of previous conscious deliberation and to focus their
attention only on what is consciously occurring in their mind and visual
field at the ‘moment of decision’. This is so far removed from the normal
circumstances of voluntary action that little confidence can be had that
it throws any real light upon the nature of such action. Knowing as
the subjects do that they must perform the specified bodily movement a
number of times without premeditation, they must somehow let an urge
to perform it creep up on them unawares and then consciously endorse it
as something they are willing now to do. It is presumably this conscious
endorsement that they report as the decision to act. But then it is hardly
surprising that this ‘decision’ did not initiate the action. Seen in this
light, the result of the experiment could almost have been predicted in
advance. Another example of psychological research restating the obvious
in misleading terms? However, we have been given little or no reason
to suppose that ‘decision’ operates like this in the normal circumstances
of deliberative voluntary action, when a choice is made in the light of
reasons reflected on by the agent in advance of choosing. In any case, the
apparent time of a volition, as judged by the subject, is the time at which
it represents itself as having occurred, and there is some reason to suppose
that this may be later than its actual time of occurrence—because the brain
has evolved mechanisms to disguise time lags between neural causes and
effects engendered by the relatively slow rate at which nerve signals are
propagated or response thresholds are reached.⁶ (Thus, for instance, if one
sees a sharp object strike one’s toe, one seems to feel the pain at the same
time, even though the neural pathway from toe to brain is much longer
than that from eye to brain.)
⁶ That the brain deploys such mechanisms is illustrated by other well-known experimental findings of
Libet and his colleagues, for which see again his Neurophysiology of Consciousness. For further discussion,
see Daniel C. Dennett, Consciousness Explained (London: Penguin, 1992), pp. 153ff.
86 Mental Causation, Closure, and Dualism
⁷ See William James, Principles of Psychology (New York: Henry Holt & Co., 1890), vol. 2, p. 490.
Could Volitions be Epiphenomenal? 87
satisfactorily. However, even in this sort of case we are dealing with subjects
whose will was, we suppose, causally efficacious at one time, before total
paralysis struck them. As for the malicious demon scenario, I am not really
interested in that for present purposes. When I ask whether the causal
efficacy of the will quite generally could be a cognitive illusion, I am not
thinking of an illusion of the Cartesian sceptical kind. In other words, I
am not considering the outermost reaches of the space of metaphysical
possibility, but whether it might really be the case, in the actual world, that
the will is causally entirely inefficacious. Our question concerns not what
is absolutely metaphysically possible—what is the case in some possible
world, however remote from actuality—but what is epistemically possible
with regard to the actual world.
⁸ See, further, G. H. von Wright, Explanation and Understanding (London: Routledge and Kegan
Paul, 1971), pp. 60ff. Compare also Ian Hacking, Representing and Intervening (Cambridge: Cambridge
University Press, 1983), Part B, and James Woodward, Making Things Happen: A Theory of Causal
Explanation (New York: Oxford University Press, 2003).
88 Mental Causation, Closure, and Dualism
⁹ See David Hume, An Enquiry Concerning Human Understanding, ed. L. A. Selby-Bigge and P. H.
Nidditch (Oxford: Clarendon Press, 1975), p. 66.
¹⁰ See, again, Wegner, The Illusion of Conscious Will, pp. 13–14, where he quotes Hume with
approval, in particular a passage from the Appendix to Hume’s Treatise in which he says ‘the actions
of the mind are, in this respect, the same with those of matter. We perceive only their constant
conjunction.’ See David Hume, A Treatise of Human Nature, ed. L. A. Selby-Bigge and P. H. Nidditch
(Oxford: Clarendon Press, 1978), p. 633.
Could Volitions be Epiphenomenal? 89
But a rational being must be able to direct its thoughts at will to this or
that subject matter. This is in the very nature of rational deliberation. If our
thoughts were in no way under our voluntary control, we would be mere
automata. We might be reasoning automata, like suitably programmed
computers, but we would not be rational beings in the proper sense of
the term.
We take ourselves to be rational beings in the proper sense of the term.
Might that itself be a cognitive illusion? If it is, then we are not in fact
rationally justified in the way we think we are to hold the beliefs that
we do, including all of our scientific beliefs. So we cannot rationally take
ourselves to be subject to such a cognitive illusion. If we are subject to such
an illusion, then we are not rational beings and a belief that we are subject
to the illusion will not be rational, simply because none of our beliefs will
be rational. If we are not subject to such an illusion, however, then again
a belief that we are subject to it will not be rational. So, either way, we
cannot rationally believe that we are subject to such an illusion. We are
rationally committed to the belief that we are rational beings in the proper
sense of the term, and this excludes the possibility that we might rationally
take ourselves to be reasoning automata. Pragmatically, at least, we can
conclude that we are not reasoning automata and hence that our thoughts
are subject to our voluntary control.
But does this require that our will can be causally efficacious with respect
to what we are thinking, or at least how we direct our thoughts? I think
it does. Recall that it is partly constitutive of the notion of a volition or
act of will that it be a mental event that is capable of causing its intended
effect. In willing to raise my arm, I will to bring about a rising of my arm
by means of that very act of will. The intentional contents of volitions
are causal in character, and necessarily so. If successful willing were merely
a matter of willing an event of a certain kind to occur, followed by the
occurrence of that kind of event, then our will would not need to be
causally efficacious in order for us to be able to engage successfully in
voluntary action. But that would make willing in no way significantly
different from mere desiring or wishing. We cannot will successfully unless
our volitions do in fact cause their intended effects. It is not sufficient for
such effects to follow our volitions but fail to be caused by them, for then
we would never in fact do what we intend to do when we exercise our
wills. We would all be, all of the time, victims of a massive ‘illusion of
Could Volitions be Epiphenomenal? 91
¹ See, for example, Jaegwon Kim, ‘Lonely Souls: Causality and Substance Dualism’, in K. J. Corcoran
(ed.), Soul, Body, and Survival: Essays on the Metaphysics of Human Persons (Ithaca, NY: Cornell University
Press, 2001).
The Self as an Emergent Substance 93
² See Jaegwon Kim, ‘Events as Property Exemplifications’, in M. Brand and D. Walton (eds), Action
Theory (Dordrecht: D. Reidel, 1980), pp. 159–77.
³ See Tim Crane and D. H. Mellor, ‘There is No Question of Physicalism’, Mind 99 (1990),
pp. 185–206.
94 Mental Causation, Closure, and Dualism
⁴ See, further, my Subjects of Experience (Cambridge: Cambridge University Press, 1996), ch. 1.
The Self as an Emergent Substance 95
body’ nor with any part of it,⁷ the one that I now consider to be the
strongest is what I call the unity argument—the unity in question being
the unity of the self or person as the unique subject of all and only its
own experiences. I shall present the argument in first-person terms for ease
of formulation. Thus formulated, the first premise of the unity argument
is this:
(1) I am the subject of all and only my own mental states,
which is surely a self-evident truth.⁸ The second premise is this:
(2) Neither my body as a whole nor any part of it could be the subject
of all and only my own mental states.
And its conclusion, which undoubtedly follows from the two premises, is
this:
(3) I am not identical with my body nor with any part of it.
Of course, (2) is the crucial premise, so let us see how it might be
defended. First, then, observe that my body as a whole does not need to
exist in order for me to have every one of the mental states that I do in fact
have. If, for instance, I were to lack the tip of one of my little fingers, I
might as a consequence lack some of the mental states that I do in fact have,
but surely not all of them. I might perhaps lack a certain mildly painful
sensation in the finger tip—a sensation that I do in fact have—but many
of my other mental states could surely be exactly the same as they actually
are, such as the thoughts that I am in fact having in composing this chapter.
Indeed, I could still even have that sensation ‘in my finger tip’, because the
phenomenon of ‘phantom’ pain is a well-attested one. However, I venture
to affirm that no entity can qualify as the subject of certain mental states
if those mental states could exist in the absence of that entity. After all, I
⁷ See, especially, my Subjects of Experience, ch. 2., but also my ‘Identity, Composition, and the
Simplicity of the Self ’, in K. J. Corcoran (ed.), Soul, Body, and Survival. In the latter, I show in detail
why the non-identity of the self with its body or any part of it implies that the self is a simple —that
is, a non-composite—substance. What I am calling the ‘unity argument’ is, however, new to the present
book.
⁸ It may perhaps be wondered whether the truth of premise (1) is threatened by the phenomena of
‘alien voices’ and ‘inserted thoughts’ that seem to characterize certain psychopathological conditions. I
discuss this issue and argue that there is really no such threat in my ‘Can the Self Disintegrate? Personal
Identity, Psychopathology, and Disunities of Consciousness’, in J. Hughes, S. Louw, and S. Sabat (eds),
Dementia: Mind, Meaning, and the Person (Oxford: Oxford University Press, 2006).
The Self as an Emergent Substance 97
certainly qualify as the subject of my mental states, as (1) asserts, but for
that very reason those mental states could not exist in my absence. Mental
states must always have a subject—some being whose mental states they
are—and the mental states that in fact belong to one subject could not
have belonged to another, let alone to no subject at all.⁹ But, as we have
just seen, very many and quite possibly all of my own mental states could
exist even if my body as a whole were not to exist—that is to say, even
if certain parts that my body actually possesses were not to exist. This, I
suggest, indicates that my body as a whole cannot qualify as the subject of
all and only my own mental states and so cannot be identified with me.
Now, many physicalists may agree with my reasoning so far, but draw
the conclusion that, rather than being identical with my body as a whole,
I am identical with some part of it, the most obvious candidate being my
brain. However, it is easy to see that the foregoing reasoning can now
just be repeated, replacing ‘my body as a whole’ by ‘my brain as a whole’
throughout. For it seems clear that, although I may well need to have a
brain in order to have mental states, neither my brain as a whole nor any
distinguished part of it is such that it in its entirety needs to exist in order
for me to have every one of the mental states that I do in fact have. Indeed,
even if every one of my mental states depends in this fashion upon some part
of my brain, it by no means follows, of course, that there is some part of my
brain upon which every one of my mental states thus depends. (To suppose
that this does follow would be to commit a so-called ‘quantifier-shift’
fallacy.) And yet I, being the subject of all and only my own mental states,
am such that every one of those mental states does depend upon me. Hence,
we may conclude, neither my brain as a whole nor any part of it can qualify
as the subject of all and only my mental states and so be identical with me.
Putting together the two stages of this train of reasoning, we may thus infer
that (2) is true and from that and (1) infer the truth of (3), the key claim
of NCSD.
I should perhaps stress that it is important to appreciate, when considering
the foregoing argument, that I am by no means denying that there may be
some part of my brain which is such that, were it to be completely destroyed,
all of my mental states would thereby cease to be. After all, I am happy to
¹⁰ See, for example, R. M. Chisholm, Person and Object: A Metaphysical Study (London: George
Allen and Unwin, 1976), ch. 3.
¹¹ See, for example, Peter van Inwagen, ‘The Doctrine of Arbitrary Undetached Parts’, Pacific
Philosophical Quarterly 62 (1981), pp. 123–37.
The Self as an Emergent Substance 99
B’s including the part—the finger tip—that O does not include? Isn’t the
material difference between B and O simply irrelevant to the case that can
be made in favour of either of them qualifying as the subject of T? But
in that case, we must either say that both B and O are subjects of T, or
else that neither of them is. We cannot say the former, however, because B
and O are numerically distinct objects, whereas the thoughts T have just
one subject—myself. We may conclude, hence, that neither B nor O is a
subject of T and thus that I, who am the subject of T, am identical with
neither of them. This sort of reasoning can then be repeated, as before,
with respect to any specific part of B, such as my brain.
However we exactly formulate the defence of premise (2), the basic
point of the unity argument, as I call it, is that my mental states do not all
depend on my body as a whole or on any part of it in the unified way
in which they all depend upon me as their subject. This point, it seems
to me, is a good one. Thus, I consider that the unity argument provides a
compelling reason for belief in the truth of NCSD’s central claim.
dualism, but since even the first and more credible of these positions has
relatively few modern advocates, I shall not consider them here.¹² In any
case, even those who do support them would presumably concede that
they would prefer to endorse interactive dualism if they thought that it
could meet the physicalist’s objections, so let us concentrate on seeing
how those objections can indeed be met, focusing on the causal closure
argument. Certainly, any dualist who holds, as I do, that a self or person
must have distinctive and independent causal powers must endorse a species of
interactive dualism.
The key premise of the causal closure argument against interactive
dualism is the principle of the causal closure of the physical domain.
This principle has received a number of different formulations—some of
which (as we saw in Chapter 2) are really too weak for the physicalist’s
purposes—but the relatively strong version of the principle that I shall
chiefly consider here is this:¹³
(4) No chain of event-causation can lead backwards from a purely
physical effect to antecedent causes some of which are non-physical
in character.
It may be objected on behalf of interactive dualism that (4) is simply
question-begging, because it rules out by fiat the possibility of there being
non-physical mental causes of some physical effects. However, as we shall
see, (4) does not in fact rule out this possibility. Dialectically, it is in
the dualist’s interests to concede to the physicalist a version of the causal
closure principle that is as strong as possible—provided that it still falls
short of entailing the falsehood of interactive dualism—because if the
causal closure argument in its strongest non-question-begging form can be
convincingly defeated, the physicalist will be left with no effective reply.
Weaker versions of the causal closure principle can, of course, be handled
by interactive dualists relatively easily, but too often tend to be dealt with
by them in implausible ways which leave the physicalist with a telling
response.
antecedent events in the agent’s nervous system and brain. Many of the
neural events concerned will be widely distributed across fairly large areas
of the motor cortex and have no single focus anywhere,¹⁴ with the causal
chains to which they belong possessing no distinct beginnings. And yet,
intuitively, the agent’s mental act of decision or choice to move the arm
would seem, from an introspective point of view, to be a singular and
unitary occurrence which somehow initiated his or her action of raising the
arm. The immediate question, then, is how, if at all, we can reconcile
these two apparent facts. It seems impossible to identify the agent’s act of
choice with any individual neural event, nor even with any combination
of individual neural events, because it and they seem to have such different
causal features or profiles. The act of choice seems to be unitary and to
have, all by itself, an ‘initiating’ role, whereas the neural events seem to be
thoroughly disunified and merely to contribute in different ways to a host
of different ongoing causal chains, many of which lead independently of
one another to the eventual arm-movement.
I believe that NCSD can enable us to see how both of these causal
perspectives on deliberative physical action can be correct, without one
being reducible to the other and without there existing any sort of rivalry
between the two. First of all, the act of choice is attributable to the person
whereas the neural events are attributable to parts of the person’s body: and
a person and his or her body are, according to this conception of ourselves,
distinct things, even if they are not separable things. Moreover, the act of
choice causally explains the bodily movement—the upward movement of
the arm—in a different way from the way in which the neural events
explain it. The neural events explain why the arm moved in the particular
way that it did—at such-and-such a speed and in such-and-such a direction
at a certain precise time. By contrast, the act of choice explains why
a movement of that general kind —in this case, a rising of the agent’s
arm—occurred around about the time that it did. It did so because shortly
beforehand the agent decided to raise that arm. The decision certainly
did not determine the precise speed, direction, and timing of the arm’s
movement, only that a movement of that general sort would occur around
about then. The difference between the two kinds of causal explanation
reveals itself clearly, I suggest, when one contemplates their respective
counterfactual implications. If the agent had not decided to raise his or her
arm, there wouldn’t have been an arm-movement of that kind at all —the
arm would either have remained at rest or, if the agent had decided to make
another movement instead, it would have moved in a quite different way.
It doesn’t seem, however, that one can isolate any neural event, nor even
any set of neural events, whose non-occurrence would have had exactly the
same consequences as the non-occurrence of the agent’s decision. Rather, the
most that one can say is that if this or that neural event, or set of neural
events, had not occurred, the arm-movement might have proceeded in a
somewhat different manner—more jerkily, perhaps, or more quickly—not
that the arm would have remained at rest, or would instead have moved in
a quite different kind of way.
¹⁶ See Benjamin Libet, ‘Unconscious Cerebral Initiative and the Role of Conscious Will in Voluntary
Action’, Behavioral and Brain Sciences 8 (1985), pp. 529–66. Libet’s experiments were not concerned
with premeditated actions, but only with ‘spontaneous’ ones. See the discussion in Chapter 4.
¹⁷ For discussion of this matter, see my A Survey of Metaphysics (Oxford: Oxford University Press,
2002), ch. 10. I should stress, then, that my present argument does not presume the truth of any form
of counterfactual theory of causation, where such a theory is one that seeks to analyse causal statements
in terms of counterfactual conditionals. Indeed, I am highly sceptical about the prospects for such a
theory.
The Self as an Emergent Substance 105
decision, then he or she would either have made a quite different decision
or else no decision at all. Either way—assuming that there is nothing
defective in the agent’s nervous system—the arm would not have risen
almost exactly as it did.
I suppose that a convinced physicalist might try to challenge the claim
that I have just made and contend that, indeed, if D had not occurred,
then another decision to raise the arm in virtually the same way would have
occurred instead, giving rise to a slightly different bodily event of the same
kind.¹⁸ But, on the face of it, this would appear to be a purely ad hoc
manoeuvre designed solely to save the envisaged physicalist’s position. One
serious problem with it is that contentful mental acts such as decisions are,
very plausibly, individuated at least partly by their contents—and yet their
contents surely cannot be as fine-grained as the physicalist’s conjectured
contention would appear to demand. How, exactly, would the content of
the decision that, supposedly, would have occurred if D had not occurred,
have differed from the content of D? If the putative difference in their
contents is to match the very slight difference between the bodily events
that are supposed to ensue from them, then a degree of fine-grainedness must
be attributed to those contents that, it seems to me, is utterly implausible
from a psychological point of view. For instance, we must suppose that
D is a decision to raise the agent’s arm along a quite specific trajectory
T, whereas if D had not occurred then the agent would instead have
decided to raise his or her arm along the very slightly different trajectory
T ∗ , where the spatiotemporal differences between T and T ∗ are of the
same order of magnitude as the very slight differences between the actual
arm-movement B and the arm-movement that would have occurred if
neural event N ∗ had occurred instead of neural event N. But the contents
of our decisions to act are surely never as fine-grained as this—not, at
least, if our conscious introspective awareness of those contents is to be
relied upon. And to propose that they always have much finer-grained
contents that are inaccessible to consciousness seems a desperate recourse on
the part of the physicalist. When, for instance, I decide to raise my arm
in a lecture in order to ask a question, I may indeed decide to raise it
quickly and vertically, but never—surely—along a quite specific trajectory
at a quite specific speed. Quite apart from anything else, I simply don’t
¹⁹ Thus, for example, Peter Menzies attributes the following causal inheritance principle to Jaegwon
Kim: ‘[I]f a second-order state S1 is realised by a first-order state S2 , then the causal powers of S1 are
108 Mental Causation, Closure, and Dualism
Then our argument may be extended in the following way to rebut such a
realization account of psychoneural relations.
Suppose that, in line with such an account, it is claimed that decision D
is not identical with the highly complex neural event N, but only realized
by N. And let us assume, as above, that ‘realization’ is supposed to be
a relation, distinct from identity, in virtue of which a ‘realized’ entity
inherits all of its causal powers in a given possible world from the entity
that ‘realizes’ it in that world—leaving open, thus, the possibility that the
‘realizing’ entity may possess other causal powers that are not inherited by
the ‘realized’ entity, as well as the possibility that one and the same entity
may be ‘realized’ by various numerically distinct entities, with somewhat
different causal powers, in different possible worlds. The implication surely
is that if D is actually realized by N, then a causal counterfactual cannot
be true of D in the actual world that is not true of N in the actual world
(even though the reverse may not be the case, so that there may be causal
counterfactuals true of N in the actual world that are not true of D in the
actual world). For if, on the contrary, some causal counterfactual were true
of D but not of N in the actual world, then that would surely indicate that
at least one of D’s actual causal powers was not inherited from N, in direct
conflict with the proposition that D is actually realized by N. However,
that is precisely what we do seem to have in the present case: a causal
counterfactual that is true of D but not true of N in the actual world. For
it is plausibly true of D in the actual world, but not of N, that if it had not
occurred, then the agent’s arm would not have risen at all.
Now, of course, extending our argument in exactly this way presupposes
a certain conception of the ‘realization’ relation that may not, perhaps,
be accepted by all self-styled realization theorists—namely, a conception
according to which, when one entity is realized by another, the former
inherits all of its causal powers from the latter.²⁰ So it is worth seeing that
identical with the causal powers of S2 (or at least a subset of them).’ See Peter Menzies, ‘The Causal
Efficacy of Mental States’, in S. Walter and H.-D. Heckmann (eds), Physicalism and Mental Causation:
The Metaphysics of Mind and Action (Exeter: Imprint Academic, 2003), p. 203.
²⁰ Peter Menzies, for instance, explicitly rejects Kim’s causal inheritance principle, but not the idea
that mental states are realized by neural states: see his ‘The Causal Efficacy of Mental States’, p. 221.
I do not have space to discuss Menzies’ own very interesting position, beyond remarking that one
fundamental point of disagreement between him and me is that he thinks that causation is not an
absolute but, rather, a contextually relativized relation: see again his ‘The Causal Efficacy of Mental
States’, p. 205.
The Self as an Emergent Substance 109
²¹ Further points made in Section 5.8 bear upon realization theories as well as identity theories. If
I seem to give rather short shrift in this book to ‘non-reductive’ physicalism, particularly in the shape
of realization theories, it is because I sympathize with Jaegwon Kim’s view that there is really no
110 Mental Causation, Closure, and Dualism
stable middle ground for non-reductive physicalism to occupy between genuine reductive physicalism
and full-blooded dualism (where by ‘reductive’ physicalism, however, I don’t just mean to include
so-called type–type identity theories but also logically weaker token–token identity theories which deny
the existence of type–type identities). For Kim’s view of the matter, see especially his Physicalism, or
Something Near Enough (Princeton, NJ: Princeton University Press, 2005).
²² Thus I am rejecting, in effect, what Menzies calls, following Tim Crane, the principle of the
homogeneity of mental and physical causation: see his ‘The Causal Efficacy of Mental States’, p. 200. See
further Tim Crane, ‘The Mental Causation Debate’, Proceedings of the Aristotelian Society (Supplementary
Volume) 69 (1995), pp. 211–36.
²³ For more on this distinction, see Jonathan Bennett, Events and their Names (Oxford: Clarendon
Press, 1988) and also my A Survey of Metaphysics, ch. 9.
The Self as an Emergent Substance 111
occurred is not necessarily yet to know why an event of that kind occurred,
as opposed to an event of some other kind. Intentional causation can
provide the latter type of explanation in cases in which bodily causation
cannot. More specifically: an event, such as a particular bodily movement,
which may appear to be merely coincidental from a purely physiological point
of view—inasmuch as it is the upshot of a host of independent neural
events preceding it—will by no means appear to be merely coincidental
from an intentional point of view, since it was an event of a kind that the
agent intended to produce. (See again Chapter 1 for more on this point.)
Notice, here, that the aforementioned fact—that a mental decision, D,
to perform a certain kind of bodily movement, cannot be said to cause
the particular bodily event, B, of that kind whose occurrence renders that
decision successful—is already implied by the argument that I developed
a little while ago in Section 5.5. For, given that D is not identical with the
actual neural cause, N, of B, the closest possible world in which N does
not occur is still a world in which D occurs—but in that world a slightly
different bodily movement, B∗ , ensues, being caused there by a slightly
different neural cause, N ∗ . (Clearly, if D is not identical with N, then there
is no reason to suppose that the closest world in which N does not occur
is also one in which D does not occur, for a world in which both of these
events do not occur evidently differs more from the actual world than a
world in which just one of them does not occur, other things being equal.)
However, this means that the occurrence of D is causally compatible with
the occurrence of two numerically different bodily movements of the same
kind, B and B∗ , and hence does not causally determine which of these
occurs, but only that some bodily movement of their kind occurs.
At this point, I anticipate the following possible objection on the part
of the physicalist. Couldn’t the physicalist simply concede that the complex
neural event N, in our example, is not identical with the mental decision
D —and thereby concede that D does not cause the particular bodily
movement, B, that is caused by N —while still insisting that D is identical
with some neural event, call it M, which has precisely the causal role that I
am attributing to D? According to this view, D is identical with a neural
event, M, which causally explains why some bodily movement of B’s kind
occurred, but not why B in particular occurred. No—such a position is
not tenable, for reasons which we have already encountered. Recall that
I argued that the following counterfactual conditional is true: ‘If D had
112 Mental Causation, Closure, and Dualism
not occurred, then no bodily movement of B’s kind would have occurred.’
That is to say, if the agent had not performed that decision to raise his
arm, his arm would not have risen in anything like the way that it did—it
would either have moved in some quite different way, or not at all, because
if he had not made that decision, he would either have decided to do
something quite different or else not have decided to do anything. Can
the same thing be said with regard to the putative neural event M? No, it
can’t. This is because, once again, plausibility demands that the physicalist
takes M to be an extremely complex neural event, composed of the firings
of very many individual neurons, so that the closest possible world in
which M itself does not occur will be one in which a neural event, M ∗ ,
occurs which differs from M only in respect of the firing of one or two
individual neurons. And it simply isn’t credible to suppose that this very
small difference between M and M ∗ should make all the difference between
the agent’s arm rising and some quite different kind of bodily movement
occurring. Consequently, the counterfactual conditional that is true of M
is this: ‘If M had not occurred, then a bodily movement of B’s kind would
still have occurred.’ So, once more, because different counterfactuals are
true of D and M, D and M cannot be identical. The physicalist’s new
proposal encounters exactly the same difficulty as did his original proposal.
The difficulty is that mental causes like D have a strong unity which fails to
characterize extremely complex neural events such as N and M. Because
of this lack of strong unity, the closest worlds in which events like N and
M do not occur are worlds in which the vast majority of their parts still
occur, with the consequence that similar bodily effects still ensue.
²⁴ See, further, my ‘The Truth About Counterfactuals’ and my A Survey of Metaphysics, pp. 148–52.
114 Mental Causation, Closure, and Dualism
psychological characteristics. But the point is that the objection that has
been raised seems to count strongly against precisely this stipulation. Hence,
the strategy is thoroughly question-begging. If this is the only resort of the
physicalist in the light of my argument, it is a desperate resort indeed and
tantamount, I should say, to a confession of defeat.
In any case, the foregoing strategy is a dangerous one for the physicalist
to pursue, for another reason. While it is plausible to maintain, quite gen-
erally, that the same counterfactual sentence may sometimes be interpreted
differently in different conversational contexts, what is much less plausible
to maintain is that such a sentence may receive in this way two different
causal interpretations so as to make it true on one such interpretation and
false on the other. For this would seem to imply that causation itself is a
context-relative rather than an absolute affair. And, while some philosoph-
ers have indeed expressed sympathy for this idea,²⁵ it seems difficult to
square it with the view, which physicalists generally espouse, that causal
relations are wholly objective and mind-independent in character. Thus, it
is going to be difficult for a physicalist to support the contention that the
counterfactual ‘If M had not occurred, then a bodily movement of B’s kind
would still have occurred’ is both true on one causal interpretation—as
it undoubtedly seems to be—and yet also false on another. But this is
what he needs to do if he both insists on the identity of M with D and
accepts, as I maintain he should, the truth of ‘If D had not occurred, then a
bodily movement’s of B’s kind would not have occurred’, on a very natural
causal interpretation of the latter. Moreover, a physicalist who did elect to
pursue the foregoing strategy would hardly be in a position to complain if
anti-physicalists were to do likewise. But it would be a relatively easy matter
for such anti-physicalists to circumvent by these means any form of causal
closure argument for the truth of physicalism. In short, then, the envisaged
strategy is a two-edged sword which threatens to serve the physicalist’s
ends only at the cost of playing directly into the hands of his opponents.
The convinced physicalist would be better advised, therefore, to try to
undermine my line of argument against him at a much earlier stage.
Even so, reflection on the context-dependency of counterfactuals might
lead a critic of my argument to suppose that there is, in any case, something
wrong with it that has to do with this feature of counterfactuals. For it
²⁵ See, for example, the reference to Peter Menzies’ position in note 20 above.
The Self as an Emergent Substance 115
might seem that parallel arguments deliver false conclusions in some cases
and are fallacious for precisely this sort of reason. Consider, for instance,
the following case.²⁶ Suppose that, on a certain occasion, I wave a red flag
at a bull and consequently cause it to charge. (Never mind if bulls don’t
really react in this way to red things!) Then it seems plausible to say that,
other things being equal, the following counterfactual is true: ‘If I hadn’t
waved my red flag, the bull wouldn’t have charged’. Why? Because, other
things being equal, the ‘closest’ world in which I don’t wave a red flag
is one in which I wave one of another colour, such as green, and only
red flags (we may suppose) enrage bulls. Suppose further, however, that
my red flag was a scarlet flag—rather than any other shade of red, such as
crimson—and that bulls are enraged by red flags no matter what shade of
red they may be. In that case, it might seem plausible to say that, other
things being equal, the following counterfactual is also true: ‘If I hadn’t
waved my scarlet flag, the bull would still have charged.’ Why? Because,
other things being equal, the ‘closest’ world in which I don’t wave a scarlet
flag is one in which I wave one of another shade of red, such as crimson,
which still enrages the bull. But now it seems that, if my argument against
the physicalist is valid, we ought to be able to conclude in this seemingly
parallel case that the event of my waving my red flag is neither identical
with nor even ‘realized’ by the event of my waving my scarlet flag. And
yet that conclusion seems absurd.
The trouble with this example, and others like it, is that the details of the
case are grossly underspecified and once we begin to fill them in, in any
remotely plausible way, we see that, after all, ‘other things’ are not ‘equal’
in the manner needed to make the example work. Why can we assume, for
instance, that the closest possible world in which I don’t wave my red flag
is one in which I wave a flag of another colour, such as green? Well, we
could assume this if we suppose that when I bought my flag, I went into a
shop selling flags, determined to buy a flag of some colour and found that
the shop sold only scarlet flags, green flags, and blue flags. However, on this
assumption, it simply isn’t the case that the closest possible world in which
I don’t wave my scarlet flag is one in which I wave one of another shade of
red, such as crimson. On the other hand, let us suppose instead that when I
bought my flag, I went into a shop selling flags, determined to buy a red flag
and found that the shop sold only scarlet flags, crimson flags, and green flags.
In that case, we can indeed assume that the closest possible world in which
I don’t wave my scarlet flag is one in which I wave one of another shade of
red, namely crimson. However, now we are no longer entitled to assume
that the closest possible world in which I don’t wave my red flag is one in
which I wave a flag of another colour, namely green: on the contrary, it
is one in which I wave a red flag of another shade of red, namely crimson.
It turns out, then, on either version of the story about how I obtained my
flag, only one of the relevant counterfactuals is true, not both of them. On
the first version, ‘If I hadn’t waved my red flag, the bull wouldn’t have
charged’ is true but ‘If I hadn’t waved my scarlet flag, the bull would still
have charged’ is false, and on the second version the reverse is the case. Thus
it transpires, on closer inspection, that this example doesn’t really provide
us with a genuine parallel to my argument against physicalism and thereby
cast doubt on the validity of that argument. For, in the case of my own
argument, there is no similar problem in evaluating both of the premises as
being true, without indulging in any sort of equivocation. Doubt could be
cast on the validity of my own argument only if the same could be said of
the premises of the ‘flag’ argument—but, as we have seen, it can’t.
²⁷ Compare Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2000).
The Self as an Emergent Substance 117
²⁸ Endorsed by, for example, Jaegwon Kim in his ‘The Non-Reductivist’s Troubles with Mental
Causation’.
118 Mental Causation, Closure, and Dualism
simply be false. It will be false, for instance, to say that if the agent had not
decided to raise his or her arm, then a rising of the agent’s arm would not
have occurred: rather, precisely the same bodily movement would still have
occurred, caused by precisely the same physical events that actually did cause
it—for if physical determinism is true, there was never any real possibility
that those physical events should not have occurred, nor that they should
have had different effects. Maybe so. But, in view of the developments in
quantum physics during the twentieth century, we now know that physical
causation is not in fact deterministic, so the objection is an idle one and can
safely be ignored. The model of intentional causation that I am proposing
may nonetheless still seem puzzling to many philosophers, but if so, then
I suggest that this will be because they are still in the grip of an unduly
simple conception of what causation involves—one which admits only of
the causation of one event by one or more antecedent events belonging to
one or more chains of causation which stretch back indefinitely far in time.
Since this seems to be the only sort of causation that is recognized by the
physical sciences, intentional causation on NCSD’s model is bound to be
invisible from the perspective of such a science (a point that I stressed earlier,
in Chapter 3). To a physicalist, this invisibility will seem like a reason to
dismiss NCSD’s conception of intentional causation as spurious, because
‘non-scientific’. Instead, I suggest, it is a reason to perceive no genuine
conflict between explanation in the physical and biological sciences on the
one hand and, on the other hand, another and irreducibly personal way of
explaining our intentional actions, by reference to our choices or decisions
and the reasons for which we make them. In sum, NCSD presents us
with a metaphysically coherent and empirically well-motivated conception
of ourselves as emergent substances, ontologically distinct from the purely
material substances of our ‘organized’ bodies and possessing causal powers
that complement those of our bodies, rather than being either reducible to
or in rivalry with them.
PA RT I I
¹ See, for example, Richard Taylor, Action and Purpose (Englewood Cliffs, NJ: Prentice-Hall, 1966),
ch. 8 and ch. 9, Roderick M. Chisholm, ‘The Agent as Cause’, in M. Brand and D. Walton (eds),
Action Theory (Dordrecht: D. Reidel, 1976) and Timothy O’Connor, Persons and Causes: The Metaphysics
of Free Will (New York: Oxford University Press, 2000), ch. 3 and ch. 4. For an opposing view, see
Donald Davidson, ‘Agency’, in R. Binkley, R. Bronaugh, and A. Marras (eds), Agent, Action, and
Reason ( Toronto: University of Toronto Press, 1971), reprinted in Donald Davidson, Essays on Actions
and Events (Oxford: Clarendon Press, 1980).
122 Persons, Rational Action, and Free Will
anomaly in the latter case being that the sense in which a person can be
‘in’ a sedan chair is quite different from the sense in which a person can
be ‘in’ a flood of tears.² The lesson seems to be, then, that event causation
and agent causation are distinct species of causation, because the sense in
which an event can be a ‘cause’ is quite different from the sense in which
an agent can be a ‘cause’.
² See Gilbert Ryle, The Concept of Mind (London: Hutchinson, 1949), ch. 1.
124 Persons, Rational Action, and Free Will
³ See Arthur C. Danto, ‘Basic Actions’, American Philosophical Quarterly 2 (1965), pp. 141–8 and also
his Analytical Philosophy of Action (Cambridge: Cambridge University Press, 1973), ch. 2.
⁴ I adopt a somewhat different definition of basic action in my Subjects of Experience (Cambridge:
Cambridge University Press, 1996), p. 151, for reasons which I shall not go into just here, but I shall
return to the matter in Chapter 7.
126 Persons, Rational Action, and Free Will
not all actions demand description in terms of causative action verbs, that is,
in terms which imply that the agent, in performing the action in question,
caused some effect. An inanimate object can push or pull another object
and in so doing it causes the latter object to move in certain ways. But
when we ask by what means an inanimate agent can push or pull another
object, our answer is likely to make reference to behaviour of the agent
which is not properly described in terms of causative action verbs. For
instance, an inanimate agent can push another object by rolling into it or
by falling on to it: but ‘roll’ and ‘fall’ are not causative action verbs, so that
there is no reason to suppose that the actions that they describe have a
means–end structure. Generally speaking, it is inappropriate to ask by what
means an inanimate object rolled into or fell on to another object. The first
object’s behaviour will, of course, be subject to causal explanation, but in
all probability we shall explain why that object rolled or fell by referring
to another object’s action upon it, rather than in terms of some further
behaviour of the object in question. Consequently, in the case of inanimate
agents, their non-causative behaviour provides an obvious terminus to the
threatened regress and a well-defined class of events ‘involving’ those agents
to which analysis (I) can appeal. For example, if a boulder pushes a tree by
rolling into it, providing us with a case in which an inanimate agent (the
boulder) causes motion of a certain kind in another object (the tree), we can
plausibly reduce this instance of agent causation to one of event causation,
in the way proposed by analysis (I), by saying that in this case it was the
rolling of the boulder —an event ‘involving’ the boulder—which caused the
motion of the tree. And since the boulder’s rolling is not itself a matter
of the boulder’s causing anything, because ‘roll’ is not a causative action
verb, no further question arises as to ‘by what means’ the boulder rolled.
(Though this, as I have just pointed out, is not to deny that the boulder’s
rolling can be causally explained by reference to some other object’s action
upon it.) Consequently, in such a case, no further fact of agent causation by
the boulder remains to jeopardize our claim to have analysed the boulder’s
agency wholly in terms of event causation.
But it will be evident that matters are not so simple when we are
concerned with animate agents, such as animals and human beings. When
a human agent spontaneously waves his hand, thus causing motion in his
hand, this seems not at all like the case of a boulder causing motion in
a tree by rolling into it. In describing the spontaneous waving of one’s
128 Persons, Rational Action, and Free Will
⁵ See Peter van Inwagen, An Essay on Free Will (Oxford: Clarendon Press, 1983), ch. 3. Although
van Inwagen’s argument has frequently been challenged, I think that it is basically sound.
Event Causation and Agent Causation 129
⁶ Compare Timothy O’Connor, Persons and Causes, pp. 52ff. and p. 87.
130 Persons, Rational Action, and Free Will
⁷ This, essentially, is the view of Donald Davidson: see his ‘Actions, Reasons, and Causes’, Journal of
Philosophy 60 (1963), pp. 685–700, reprinted in his Essays on Actions and Events.
Event Causation and Agent Causation 131
⁸ In this way, we may perhaps turn on its head Davidson’s well-known argument that reasons must
be causes: see further my An Introduction to the Philosophy of Mind (Cambridge: Cambridge University
Press, 2000), pp. 258ff.
⁹ I develop a volitionist theory of action along these lines in my Subjects of Experience, ch. 5.
¹⁰ But see, further, my An Introduction to the Philosophy of Mind, pp. 254ff., for a possible response to
this supposed dilemma from a volitionist standpoint.
132 Persons, Rational Action, and Free Will
that is apt to have an event as a cause. Not being the sort of thing that is apt
to have an event as a cause, its lack of a cause is only to be expected and can
carry no adverse implication that it defies explanation of any kind, since
there may be other kinds of explanation than causal explanation—notably,
rational explanation.
¹¹ See G. H. von Wright, Explanation and Understanding (London: Routledge and Kegan Paul, 1971),
pp. 76ff.
Event Causation and Agent Causation 133
¹² See George Berkeley, A Treatise Concerning the Principles of Human Knowledge, paras 25–6, in his
Philosophical Works, ed. M. R. Ayers (London: Dent, 1975), p. 84.
134 Persons, Rational Action, and Free Will
might advance for maintaining that the notion of agent causation is at least
conceptually prior to that of event causation, even if it is doubted whether
event causation is reducible to agent causation. For this consideration does
indeed focus on the special case of the intentional actions of rational agents.
Although, in a metaphysical inquiry such as the present one, our main
concern should be the nature of certain fundamental features of reality—in
this case, causation—we must inevitably be concerned also with certain
epistemological questions: for if we purport to have knowledge of a certain
feature of reality, our account of the nature of that feature should not be
such as to render our knowledge of it impossible or inexplicable. Consider,
then, our claim to have knowledge of at least some relations of causation
between events—and consider what mental capacities such a knowledge-
claim presupposes. It is strongly arguable that only a creature capable of
intentional action can acquire knowledge of causal relations between events
from experience. Such a creature must not only be an agent, but must also
be aware of being an agent—and so, it seems, must possess the concept of
agent causation. The argument for this conclusion is as follows.¹³
A purely passive being that was capable of observing its physical envir-
onment—something like, perhaps, an intelligent tree, if such a being is
indeed possible—would apparently be incapable of distinguishing between
causal sequences of events and purely coincidental sequences of events.
Such a being might be able to register the existence of certain regularities
or uniformities amongst types of events occurring in its observable envir-
onment, but it would be unable to discriminate between those regularities
that obtained purely by coincidence and those that obtained in virtue of
causal relations between the events concerned. The being might notice, for
instance, that night regularly followed day, that thunder regularly followed
lightning, and that the extinction of flames regularly followed their dousing
by water. All of these sequences, we know, are causal sequences—although
not in every case is the first member of such a sequence a cause of the
second member, since in some cases the two events are collateral effects
of a common cause. But, of course, it is also possible to experience
purely coincidental sequences of events on a regular basis. It might be,
for instance, that, within one’s experience, the fall of a certain kind of
¹³ Here I follow the lead of G. H. von Wright: see his Explanation and Understanding, pp. 60ff. I
sketched a version of this argument earlier, in Chapter 4.
Event Causation and Agent Causation 135
¹⁴ P. F. Strawson once famously pointed out that ‘There was a flash’ doesn’t appear to entail
‘Something flashed’: see his Individuals: An Essay in Descriptive Metaphysics (London: Methuen, 1959),
p. 54. So a flash, it might seem, could be an ‘agentless’ event—and yet it could still surely be a cause of
other events. I discuss this and related issues in my A Survey of Metaphysics (Oxford: Oxford University
Press, 2002), ch. 12 and ch. 13: see, especially, pp. 240–1, where I argue that Strawson’s example is not
Event Causation and Agent Causation 137
nearly as compelling as it might initially appear to be. Thus, for instance, although a flash of lightning is
unlike the flash of a lighthouse, in that the former does not consist in some one thing flashing, a lightning
flash may still plausibly be regarded as consisting in many agents’ acting together in a certain way—namely,
many electrically charged particles’ collectively discharging their electricity to the ground.
¹⁵ See, further, Jennifer Hornsby, Actions (London: Routledge and Kegan Paul, 1980), ch. 1.
138 Persons, Rational Action, and Free Will
the former sense and, clearly, in no way can the ‘hand-movement’ which
consists in A’s hand’s moving also be said to consist in A’s waving his hand.
So, it may be asked, how does analysis (II) enable us to account for
the causal relation between the muscular or neural event, e1 , and the
hand-movement, e2 ? The answer is quite straightforward. Provided that
we can say, as we plausibly can, that the muscular or neural event, e1 ,
consisted in some body-part’s acting in a certain way—for instance, it
might have consisted in a certain muscle fibre’s contracting or in a certain
neuron’s firing—and that that body part, by acting in that way, caused
(albeit indirectly) the hand-movement, e2 , then analysis (II) allows us to
say, as we want to, that e1 caused e2 . But notice that the ‘agent’ to which
we must now appeal, for the purpose of analysing this relationship of event
causation, is not the human agent, A, who waved his hand and, in so doing,
performed a ‘basic’ action. Rather, the ‘agent’ in question is merely an
organic part of A’s body and so not a rational agent nor even an ‘animate’
agent, in the sense in which an entire living organism is an ‘animate’ agent.
However, we need see no conflict here between the agency of the human
being, A, and the agency of A’s body-parts, such as particular muscle
fibres or neurons of A’s body. On the contrary, human agency, while not
reducible to the agency of human body-parts, clearly depends upon the
latter, in the sense that our possession of a body in working order is a
necessary condition of our ability to exercise physical agency on our own
account.
Whether there are any inescapable problems besetting analysis (II), or any
similar attempt to analyse event causation in terms of agent causation, I leave
for others to judge for the time being (though I shall return to the subject
in later chapters of this book). But in favour of the ontological primacy
of agent causation over event causation I would urge just this. It seems
proper to say that events of themselves possess no causal powers. Only persisting
objects—that is, individual ‘substances’—possess causal powers and, indeed,
causal liabilities. It is such objects that we describe as being magnetic,
corrosive, inflammable, soluble, and so forth. Objects manifest or display
their causal powers and liabilities by acting on things, or being acted upon,
in various appropriate ways—by attracting, corroding, burning, dissolving,
and so forth. In describing such activities we use, of course, the language
of agent causation, rather than the language of event causation. We resort
to the latter, I suggest, primarily when we are at least partially ignorant
Event Causation and Agent Causation 139
about the causal agents that are at work. Analysis (II) explains this, for
we could be sure that the right-hand side of (II) was satisfied in certain
circumstances, and hence that its left-hand side was true, even though
we could not identify a specific agent and manner of acting in virtue of
which the right-hand side of (II) was satisfied on that occasion. It might be
suggested, then, that those philosophers who accord ontological primacy
to event causation are tempted to do so by the fact that events seem to
be, very often, more ‘epistemically accessible’ than substances are. In short,
their approach may simply be the legacy of an empiricist epistemology
which distrusts all talk of causal machinery at work in the real world behind
the shifting scenes of appearance.
¹⁸ See Renford Bambrough, ‘Universals and Family Resemblances’, Proceedings of the Aristotelian
Society 60 (1960/61), pp. 207–22.
7
Personal Agency
Why does the problem of free will seem so intractable? I surmise that
in large measure it does so because the free will debate, at least in its
modern form, is conducted in terms of a mistaken approach to causality in
general. At the heart of this approach is the assumption that all causation
is fundamentally event causation. Of course, it is well known that some
philosophers of action want to invoke in addition an irreducible notion
of agent causation, applicable only in the sphere of intelligent agency.¹
But such a view is generally dismissed as being incompatible with the
naturalism that has now become orthodoxy amongst mainstream analytical
philosophers of mind. What I want to argue in this chapter—with a view
to reinforcing the conclusions of Chapter 6—is that substances, not events,
are the primary relata of causal relations and that agent causation should
properly be conceived of as a species of substance causation. I shall try
to show that by thus reconceiving the nature of causation and of agency,
the problem of free will can be made more tractable. I shall also argue
for a contention that may seem even less plausible at first sight—namely,
that such a view of agency is perfectly compatible with a volitionist theory
of action.
¹ See, for example, Richard Taylor, Action and Purpose (Englewood Cliffs, NJ: Prentice-Hall, 1966),
pp. 111–12.
142 Persons, Rational Action, and Free Will
others in everyday language, has both a transitive and causal sense and an
intransitive and non-causal sense, which are intimately related. For x to
dissolveT y is for x to cause y to dissolveI , just as for x to moveT y is for
x to cause y to moveI . DissolvingT and movingT are species of causation,
no less. And the entities that engage in these species of causation are
quite evidently individual substances, that is, particular persisting objects of
various kinds.
Note, however, that nothing that I have just said implies that we
can straightforwardly analyse a species of substance causation, such as
dissolvingT , in terms of a generic notion of substance causation together
with a specific non-causal notion, such as dissolvingI . Indeed, it is well-
known that such a suggestion faces difficulties at least in some cases:
that, for instance, it is hard to maintain that to kill just means to cause
to die. (See, again, our discussion of this case in Chapter 6.) In this case,
there are plausible counterexamples. For example, if Jones forces Smith at
gunpoint to kill Brown, then it would seem that Jones may be held causally
responsible for Brown’s death, even though it was Smith and not Jones
who killed Brown. However, I have no desire to defend the analytical
claim in question. Indeed, I would prefer to represent the direction of
semantic explanation as running, if anything, the opposite way: from
a multiplicity of quite specific notions of substance causation, bearing
various overlapping family resemblances to one another, to an abstract
generic notion of substance causation, conceived of as being the common
residue of these more specific notions (as indeed I indicated at the end
of Chapter 6). This is one reason why I expressed a reluctance to speak
of ‘the’ causal relation, as though there is really just one such relation, in
which anything describable as a ‘cause’ may be said to stand to anything
describable as an ‘effect’. As I see it, dissolvingT , movingT , attractingT , and
the rest are simply all distinct, but in some ways similar and in other ways
dissimilar, species of causal relation: they are not all analysable in terms of
some single and independently intelligible notion of causation of a generic
character. That there is no such independently intelligible generic notion
of causation would explain, of course, why it is that philosophers have
so signally failed to provide a satisfactory analysis of what it means to say
that one thing (whether it be a substance or an event) was a ‘cause’ of, or
‘caused’, another.
Personal Agency 145
² For a defence of a view of the nature of events consistent with the present proposal, see my A
Survey of Metaphysics (Oxford: Oxford University Press, 2002), ch. 14.
146 Persons, Rational Action, and Free Will
(1∗ ) Substance s caused event e2 just in case there was some event, e1 ,
and some manner of acting, F, such that e1 consisted in s’s Fing and
e1 caused e2 .
But if (2) is correct, then it obviously goes a long way towards explaining
the truth of (1∗ ), because ‘e1 caused e2 ’, according to (2), is true just in case
the following is true: ‘There was some substance, s1 , and some manner of
acting, F, such that e1 consisted in s1 ’s Fing and s1 , by Fing, caused e2 .’ We
see, thus, that, on the assumption that (2) is true, (1∗ ) is equivalent to the
following (see the Appendix to this chapter for a formal proof):
(1∗∗ ) Substance s caused event e2 just in case there was some event, e1 ,
and some manner of acting, F, such that e1 consisted in s’s Fing
and s, by Fing, caused e2 .
Now, (1∗∗ ) is plausibly true, because it is plausible to say that whenever
a substance causes an event, it does so by acting in a certain manner and
that its acting in such a manner constitutes an event. So, the implication
is that (1) is true simply because it is equivalent to a plausible principle
of substance causation, (1∗∗ ), and hence that, rather than its truth being
indicative of the reducibility of substance causation to event causation, it is
indicative of the very reverse of this.
³ For the notion of a ‘basic’ action, see Arthur C. Danto, Analytical Philosophy of Action (Cambridge:
Cambridge University Press, 1973), p. 28.
⁴ I defend such a theory in my Subjects of Experience (Cambridge: Cambridge University Press, 1996),
ch. 5.
Personal Agency 147
rope attached to it—I still cause my arm to rise by doing something, namely,
by willing to raise my arm. So, in short, I am prepared to allow that schema
(1) is true, being prepared to explain this by appealing to the truth of (1∗∗ )
and, ultimately, to the truth of schema (2). At the same time, however, I
am prepared to allow that it is at least intelligible to suppose that (2) is true
but (1) is not. That is to say, it is conceivable that all event causation may
be explicable in terms of substance causation and yet that some statements
of substance causation do not imply corresponding statements of event
causation.
opinion, is that persons are causes of their actions. Rather, we should say
that a person’s actions almost exclusively consist in that person’s causing
certain events—events that we may call ‘action-results’. The distinction
between action and action-result may be illustrated by the distinction between
the action of raising one’s arm and the event of arm-rising which occurs
as a result of that action. In English, we do not have a transitive verb ‘to
rise’, but instead use the distinct verb ‘to raise’. Raising one’s arm is a
matter of causing one’s arm to rise, however, just as dissolvingT some sugar
is a matter of causing the sugar to dissolveI . Notice, however, that I said
only that a person’s actions almost exclusively consist in that person’s causing
certain events—namely, action-results. The most important exception is
the action of willing itself, for willing does not consist in causing an event
of an appropriate kind, in the way in which arm-raising, for example,
consists in causing a rising of an arm.⁶ According to volitionism (or, at
least, my version of it), acts of will, or ‘volitions’, are indeed causes, in
the event-causation sense, of all the action-results of our voluntary actions.
But this doesn’t imply that to will is to cause something to happen. (In
the terminology of Chapter 6, ‘to will’ is not a causative action verb.) And,
indeed, of course, it is perfectly possible to will to no effect: possible, for
example, to will to raise one’s arm and yet for one’s arm not to rise as
a result.
I have just said that acts of will, or volitions, are causes, in the event-
causation sense, of action-results, where voluntary actions are concerned.
There is no conflict here with my advocacy of the primacy of substance
causation. An act of will consists in some agent’s—some person’s—willing
something. And, in line with schema (2), whenever such an act causes an
event, such as the rising of an arm, it does so simply because the act consists
in some person’s willing something—for instance, to raise an arm—and
that person, by so willing, causes the event in question to occur. Thus, one
can consistently be a substance causalist and espouse a volitionist theory
of action. Volitions are causes—not of actions, but of action-results—but
only insofar as they consist in persons’ exercising their power of will and
by so doing causing certain events. Talk here of the ‘power of will’ is to
be taken literally, not in the popular sense of ‘will-power’, but to mean
quite simply that the will is a power or natural capacity of agents, which can
⁶ Compare Carl Ginet, On Action (Cambridge: Cambridge University Press, 1990), p. 11.
Personal Agency 149
⁷ This, of course, was Locke’s view: for discussion, see my Locke on Human Understanding (London:
Routledge, 1995), ch. 6.
⁸ The locus classicus for this contrast is, of course, Aristotle, Metaphysics, Book θ.
150 Persons, Rational Action, and Free Will
some relevant kind of effect, even though its exercise does normally have
an effect of a predictable kind, namely, an action-result appropriate to the
kind of action that the agent wills to perform—for instance, an arm-rising,
in a case in which the agent wills to raise an arm. (If the will were a causal
power, it could not be exercised without a suitable effect actually occurring,
any more than water’s power to cause sugar to dissolve can be exercised
without some sugar actually dissolving. But I can, for example, exercise my
will by willing to raise my arm even if my arm does not actually rise as a
result of my so willing. Thus, an exercise of my will consists, in itself, merely
in my willing to do something, not in my actually doing that thing as a
consequence of my so willing.) Nor, clearly, is the will a causal liability, like
solubility. The exercise or manifestation of an agent’s will does not consist
in the agent’s being caused to will, in the way that the manifestation of
a lump of sugar’s solubility consists in the sugar’s being caused to dissolve
by some suitable solvent. This is not yet to deny that acts of will might in
principle be caused, only to deny that the power that is the will just is a
liability to be caused to act in a certain way (although I do, in fact, want to
deny that acts of will may be caused, even in principle, for reasons that I
shall go into in Chapters 8 and 9).
Because the will is a power, but is neither a causal power nor a causal
liability, I shall call it a spontaneous power. There are other such powers in
nature, such as radium’s spontaneous power to undergo radioactive decay,
its atoms splitting into various fission products. Of course, it is possible to
cause a radium atom to split up in such a way, by bombarding it with
high-energy particles. But current scientific orthodoxy has it that radium
can also decay ‘spontaneously’ and that when it does so there is simply no
prior event which can properly be said to be the cause of the event of
splitting: the latter, it is maintained, is a genuinely uncaused event. It is not
even the case that such a splitting event, in such a case, has a non-deterministic
or probabilistic cause: that is to say, it is not even the case that, whenever
such a splitting event occurs ‘spontaneously’, there is always some prior
event whose occurrence raises the probability (or ‘chance’) of the splitting
event’s occurrence, albeit only to a probability that still falls short of unity.
Now, of course, it may seem tendentious in the extreme to describe the
will as a ‘spontaneous’ power, given that the spontaneous radioactivity of
radium is taken to be a typical example of such a power. For it may seem
that we do not in like manner have any compelling reason to suppose that
Personal Agency 151
acts of the will are generally, or indeed ever, uncaused by prior events.
However, we shall see in due course that there may indeed be good reason
to suppose precisely this.
⁹ I don’t mean to deny that there have been, and still are, agent causalists holding that what agents
most directly cause are volitions or something like them, such as states of intention. Thomas Reid in
the eighteenth century and Timothy O’Connor today immediately come to mind in this connection:
see, for example, Timothy O’Connor, Persons and Causes: The Metaphysics of Free Will (New York:
Oxford University Press, 2000), p. 72. But, in speaking of ‘classical’ agent causalism, I am now thinking
primarily of philosophers of action writing in the middle years of the twentieth century, such as Richard
Taylor: see his Action and Purpose, pp. 64–8, where he expressly repudiates volitions. These philosophers
wrote at a time when volitionism was under attack from all sides and ‘basic’ actions were taken to
be bodily rather than mental. For further discussion, see my ‘Agent Causation’, in The Encyclopedia
of Philosophy, 2nd edn, ed. D. M. Borchert (New York: Macmillan, 2006). See also Section 3 of the
Introduction to the present book.
152 Persons, Rational Action, and Free Will
may contend that a’s causing the event of arm-rising, in such a case, is a
primitive causal fact which obtains at least to some extent independently
of any facts of event causation that may also obtain. He may also urge
that, inasmuch as a’s action of arm-raising consists in his thus irreducibly
causing the event of arm-rising, his action itself is not properly categorized
as being an event: rather, it is an irreducible causing of an event by an agent.
But then, if the action is not an event, it is not something which is apt to
be assigned a cause, on pain of committing a category mistake.¹⁰ Hence,
if we seek an explanation for the action, it cannot be right to seek a causal
explanation for it, but at most a rational explanation in terms of the agent’s
reasons for so acting. Moreover, now there is no room to agree with
the Davidsonian contention that reasons are causes insofar as they explain
actions.¹¹
By contrast, it may seem that volitionism faces a much more difficult
task in explaining how supposedly free action could escape the causal net
of events, for the volitionist is in no position to dispute the correctness
of schema (1). The volitionist believes that it is always by willing that we
cause action-results, such as arm-risings, to occur when we act freely and
that, indeed, our willings or volitions are causes of those events, in the
event-causation sense of ‘cause’. Precisely because, as we have seen, willings
are not causings, they may qualify as events and thus as causes and effects of
other events. So, to acknowledge that whenever we act freely we exercise
our will and that our willings or volitions are events is to invite the question
of whether our volitions are the effects, deterministic or probabilistic, of
prior causes. The volitionist cannot, like the classical agent causalist, hope
to sidestep the question of prior causes simply by urging that actions are
in the wrong ontological category to qualify as entities apt to be assigned
and therefore explained in terms of causes. For volitions themselves seem
to be at once actions and events—and to be events that are constitutively
involved in all episodes of supposedly free action.
¹⁰ Compare my earlier remarks to this effect in Chapter 6. But even if it is allowed that agent a’s
causing event e does itself qualify as an event, there are reasons for denying that such an event may be
supposed to be caused: see O’Connor, Persons and Causes, pp. 52–5.
¹¹ See Donald Davidson, ‘Actions, Reasons, and Causes’, in his Essays on Actions and Events (Oxford:
Clarendon Press, 1980).
Personal Agency 153
¹² See, further, my Subjects of Experience, pp. 150–2. Thus, I am now distancing myself from the way
of characterizing ‘basic’ action that was deployed in Chapter 6.
154 Persons, Rational Action, and Free Will
¹³ The notion of a ‘rational’ power is, once again, Aristotelian in origin: see Metaphysics, Book θ.
156 Persons, Rational Action, and Free Will
¹⁴ See, further, my An Introduction to the Philosophy of Mind (Cambridge: Cambridge University Press,
2000), pp. 252–62. For a similar view, see John R. Searle, Rationality in Action (Cambridge, MA: MIT
Press, 2001), pp. 12–17 and ch. 3.
¹⁵ There is no implication here that one must be able to choose to choose, or will to will, much less that
the conception of choice now being defended falls foul of a vicious infinite regress. For discussion, see
my Locke on Human Understanding, pp. 133–4.
Personal Agency 157
Appendix
Here is a formal proof that (1∗ ) is equivalent to (1∗∗ ), given certain further
assumptions that will be made explicit. Let us write (1∗ ) more formally as
follows:
(1∗ ) s caused e2 ↔ (∃e1 )(∃F)(e1 = s’s Fing & e1 caused e2 )
Then let it be assumed that schema (2) embraces the following schema as a
special case:
(2∗ ) e1 caused e2 ↔ (∃t)(∃G)(e1 = t’s Ging & t by Ging caused e2 )
Substituting the right-hand side of (2∗ ) for ‘e1 caused e2 ’ in (1∗ ), we get
(1∗ #) s caused e2 ↔ (∃e1 )(∃F)(e1 = s’s Fing & (∃t)(∃G)(e1 = t’s Ging
& t by Ging caused e2 ))
158 Persons, Rational Action, and Free Will
The key assumption in the proof is principle (3) which, however, seems
relatively uncontroversial. Informally expressed, this is the principle that if
a substance t’s acting in a manner G is identical with a substance s’s acting
in a manner F, then t is identical with s and G is identical with F.
8
Substance Causation, Persons,
and Free Will
In this chapter, I shall try to strengthen the case made in Chapter 7
for the view that the most fundamental species of causation is not event
causation—the causation of one event by another event—but substance
causation, the causation of an event by an individual substance. I maintain
that substances, rather than events, are the possessors of causal powers
and liabilities and that a substance causes an event by exercising its causal
powers. In my view, event causation may and should be analysed in terms
of this prior notion of substance causation. Personal agency, then, is a special
case of substance causation—persons being substances of a distinctive kind,
with distinctive existence and identity conditions and a characteristic range
of causal powers and liabilities. Persons are, however, unlike inanimate and
non-conscious agents in possessing rational powers, the exercise of which
is involved in episodes of free voluntary action. The events that are the
effects of the exercise of such powers do not have sufficient causes, in the
event-causation sense, at times prior to the exercise of those powers, nor
do they even have their chances of occurrence entirely fixed by events at
such prior times. And yet these effects are not arbitrary or incompletely
explicable, for they are at least in part explicable in terms of the agent’s
reasons for action. An implication of this picture of free agency is that
the domain of physical events is not completely closed under the relation
of event causation but, properly understood, this should not be seen as
incompatible with any legitimate claim of the physical sciences.
¹ Compare what Gideon Yaffe calls the ‘Where’s the Agent Problem’, in his Liberty Worth the Name:
Locke on Free Agency (Princeton, NJ: Princeton University Press, 2000), pp. 121ff.
Substance Causation, Persons, and Free Will 161
may say, for such a conception of ourselves to find a foothold. The rejected
conception belongs to an animistic world-view which sees all change in
the natural world as brought about by spirits, demons, or gods. According
to that world-view, nothing ever happens either by chance or by mindless
causal necessity. What the libertarian hankers after is an inconsistent mixture
of this pre-scientific conception and the modern naturalistic world-view,
so that animism is preserved in the exclusive domain of human action while
chance and causal laws reign everywhere else. But, so the charge goes, there
is no consistent way of combining these views because they clash at the
interface between human agency and the natural world. Either the natural
world must be reabsorbed into the world of free agents, as in animism, or
else human agency must be absorbed into the realm of chance and causal
necessity. The latter path is the only one that is realistically now open
to us and it means abandoning our pre-scientific self-conception. If we
want to retain anything that was valuable in that self-conception, we must
reinterpret ourselves in new ways by, for example, developing a convincing
compatibilist conception of freedom and a naturalistic theory of reasons
and rationality. We must accept, for instance, the dictum that ‘reasons
are causes’, abandoning the idea that there is a fundamental distinction
to be drawn between causal and rational explanation. Likewise, we must
abandon ‘final’ causes and teleological explanation, except in some suitably
naturalized form. The agents that will ‘disappear’ under this reconceptual-
ization are merely mythical in any case. Human kind will finally come of
age when it recognizes itself as being merely part of the natural order, not
set over and against it as a species of demigods.
² See, for example, Richard Taylor, Action and Purpose (Englewood Cliffs, NJ: Prentice-Hall, 1966),
pp. 111–12. For a more recent defence of agent causalism, see Timothy O’Connor, Persons and Causes:
The Metaphysics of Free Will (New York: Oxford University Press, 2000).
³ For the notion of a ‘basic’ action in play here, see Arthur C. Danto, Analytical Philosophy of Action
(Cambridge: Cambridge University Press, 1973), p. 28.
Substance Causation, Persons, and Free Will 163
will deny that A’s causing the arm-rising consists in, or even in any way
involves, the causation of the arm-rising by some action or event in which
A participates or of which A is the subject. So, for example, it will be
denied that the arm-rising is caused by a volition or act of will of A’s, or by
the onset of some combination of belief and desire attributable to A. This is
supposed to rule out as fundamentally misconceived any question of the
sort ‘How, or by doing what, did A cause the arm-rising?’. We are supposed
to accept that the agent, A, was the (or at least a) cause of the arm-rising in
a sense which precludes there being anything about A in virtue of which
A caused that occurrence. Moreover, we are to suppose that this sort of
thing is uniquely distinctive of intelligent agency, so that whenever, in the
realm of inanimate objects, we speak of such an object being the ‘cause’
of an event, we can take this as being elliptical for some statement to the
effect that some event ‘involving’ that object caused the effect in question,
or that the object caused that effect by behaving in a certain way.
Classical agent causation is regarded by its naturalistic opponents as being
utterly mysterious—and I sympathize with this attitude to a large extent.
How can a person simply as such—the person holus-bolus, as it were—be
a ‘cause’ of some event? How, for instance, would this explain why the
event occurred when it did—for the same person could cause different
events at different times and yet in each case we are supposed to believe
that the event just has the same cause, namely, that person. There are ways
in which the classical agent causalist can attempt to answer this sort of
objection—for example, by adverting to a time-reference in the agent’s
reasons for acting, or by appealing to other events as contributory causes, or
causal preconditions, of the given effect. But such strategies are not, I think,
particularly convincing. My own view is that classical agent causalism is an
unstable halfway house. It rightly sees that the notion of the causation of an
event by something in the category of substance —in this case, a person—is
an important key to understanding human agency, but mistakenly treats this
as something special and distinct from the sort of causation that goes on in
the inanimate world, where event causation supposedly reigns supreme. My
challenge to the event causalist presumptions of modern physicalism go far
deeper. I think, as I argued in Chapter 7, that all causation is fundamentally
substance causation—the causation of events by substances—even in the
world of inanimate objects. This is not a revival of ‘animism’, of course. I
am not proposing that everything that happens does so as a consequence
164 Persons, Rational Action, and Free Will
⁴ For a defence of such a view of events, see my A Survey of Metaphysics (Oxford: Oxford University
Press, 2002) ch. 14.
Substance Causation, Persons, and Free Will 165
however, is just that the car, by coming forcefully into contact with the
pedestrian, caused the pedestrian to suffer lethal injury. The source of all
change in the world lies in substances, in virtue of their causal powers
and liabilities, which they characteristically exercise or manifest when they
enter into suitable relationships with one another. But philosophers may be
misled by event-causation talk into supposing that causation is fundamentally
a relation between events—that, ultimately, it is events that bring about
changes (that is, other events). This is to invest events themselves with causal
powers. But events as such are utterly powerless. They are mere changes
in things and not the source of those changes. This is why a pure event
ontology would be an ontology without causation in any serious sense. It
is unsurprising, indeed, that idealists and phenomenalists often explicitly
acknowledge this—that, for instance, Berkeley thinks that ideas really cause
nothing, but merely succeed and are succeeded by other ideas in certain
regular patterns of occurrence (although, of course, he also thinks that
spiritual substances—ourselves and God—are truly causes of ideas). This
also helps to explain why it is that attempts to ‘analyse’ event causation—for
example, in terms of counterfactual dependence amongst events—have so
signally failed, generating epicycles upon epicycles in order to overcome
counterexamples to proposed analyses.⁵ Precisely because causation is not,
fundamentally, a relation amongst events, but a family of transactions
between substances, attempts to analyse it as such unsurprisingly fail to
accommodate all of our pre-theoretical intuitions concerning causality.
⁵ For an examination of some of these attempts, see my A Survey of Metaphysics, ch. 10.
166 Persons, Rational Action, and Free Will
⁶ For the sort of theory I favour, see my The Four-Category Ontology: A Metaphysical Foundation for
Natural Science (Oxford: Clarendon Press, 2006), Part III.
Substance Causation, Persons, and Free Will 167
⁷ See, for example, Eric T. Olson, The Human Animal: Personal Identity without Psychology (New
York: Oxford University Press, 1997).
168 Persons, Rational Action, and Free Will
of bronze whose particles compose both the statue and the lump at the
time of their coincidence.⁸ This is not my view, because I agree with
Descartes that I am a simple substance, composed of no proper parts. In fact,
I think that persons and perhaps some other species of conscious, embodied
beings, are the only macroscopic simple substances. The relationship of
embodiment in which a person stands to his or her body I regard as sui
generis, being neither identity nor constitution and yet implying, as those
relations also do, spatial coincidence. I don’t expect many philosophers
to be immediately convinced by all of these claims, although I think that
strong arguments can be marshalled directly in their favour.⁹ I also think
that this conception of the ontology of persons enables us to make more
sense of human agency—and to demonstrate this will be part of my object
in what follows.
If persons are psychological substances, distinct if not separable from
their biological bodies, then one should expect them to have causal powers
and liabilities that are also distinct from those of their bodies. And indeed
I think that this is so (as I argued in Chapter 5). Central amongst our
distinctive psychological powers are our powers of perception, thought,
reason, and will. It is I who perceive, think, reason, and will, not my body
nor any distinguished part of it, such as my brain or central nervous system.
Of course, animalists will vigorously deny this. They will charge my sort
of substance dualism—for that is what it is—with harbouring absurdities
and contradictions. For instance, they will urge that all it takes to create a
thinking being such as myself is the creation of a biological body organized
in the right sort of way, with a suitably functioning central nervous system.
Hence, they will say, all of my psychological features must supervene upon
physical features of my biological body, whence it is absurd to deny that my
body itself possesses those very psychological features: it perceives, thinks,
reasons, and wills, whenever I do. But then, if we deny that I am identical
with my body, we must say that two distinct beings are simultaneously the
subjects of the same or exactly similar perceptions, thoughts, reasonings,
⁸ See, for example, Lynne Rudder Baker, Persons and Bodies: A Constitution View (Cambridge:
Cambridge University Press, 2000).
⁹ For more details, see—in addition to Chapter 5—my Subjects of Experience (Cambridge: Cambridge
University Press, 1996) and also my ‘Identity, Composition and the Simplicity of the Self ’, in
K. J. Corcoran (ed.), Soul, Body, and Survival: Essays on the Metaphysics of Human Persons (Ithaca, NY:
Cornell University Press, 2001).
Substance Causation, Persons, and Free Will 169
and volitions. And then how do I know which of these beings I am? This is
the notorious ‘multiple thinkers’ problem.¹⁰
However, this sort of objection is entirely spurious, in my view. What
it ignores is the quite general point that one can only attribute dispositions,
including causal powers and liabilities, to substances which possess suitable
identity conditions. And biological substances, I would claim, do not possess
identity conditions suitable for the attribution to them of psychological
powers. The general point can be brought out in the following way. When
we attribute a disposition to a substance, we imply something about how it
will behave in certain hypothetical or future circumstances, and this requires
that the attribution be sensitive to the substance’s identity conditions,
including its persistence conditions—for it is these that determine which
substance, if any, can be identified with the subject of the attribution in
the hypothetical or future circumstances in question. For instance, when
I say of a rubber ball that it is ‘elastic’, I imply, amongst other things,
that it will bounce if dropped from a height on to a hard floor—and this
presupposes an account of the identity conditions of such a ball, licensing
the assumption that this very entity could persist through a change in
which it first falls through the earth’s gravitational field and then undergoes
a change of shape and motion upon impact with a solid object. Similarly,
then, when I attribute a power of thought or reason to a substance, I must
attribute it to a substance whose identity conditions are such that, of its
very nature, it can persist identically through the sorts of changes that are
constitutively involved in processes of thinking and reasoning. Thinking
and reasoning take time, and it makes no sense to suppose that a single train
of thought or reasoning might have different substances as its subject at
different times—as though one subject could ‘begin’ a thought and a quite
different subject ‘end’ it. Equally, it makes no sense to suppose that different
‘parts’ of a single thought might have different substances as their subject
at one and the same time. For this sort of reason, biological substances are
ill-suited by their identity conditions to qualify as psychological subjects.
Largely, this is because of their highly composite and mutable nature, as I
shall now try to make clear. (Recall that, on my own view, persons are
simple substances.)
¹⁰ The problem has several different versions: see, for example, Trenton Merricks, Objects and Persons
(Oxford: Clarendon Press, 2001), pp. 47–53.
170 Persons, Rational Action, and Free Will
No one doubts, of course, that the brain and central nervous system of
an embodied human person are the site of numerous biological processes
which are somehow intimately connected with the person’s psychological
processes of thought and reasoning. But the biological processes in question
cannot simply be, or in any way constitute, those psychological processes
of thought and reasoning. Why not? Because neither the organism as a
whole nor any biologically distinguishable part of it, such as the brain or
any specific region of the brain, can be singled out at any given time, or
over time, as being uniquely related in an appropriate way to any particular
psychological state or process, much less to a multiplicity of such states
and processes all of which are assignable to a single subject. Suppose, for
instance, that I am now wondering about the state of the weather in
Vienna. I could engage in that very same train of thought whether or not
my body lacked some minor organic parts here or there—for instance,
whether or not it lacked my right ear or my left little finger. So my body
as a whole is no better qualified than my body minus my right ear or
my body minus my left little finger as being, in virtue of the biological
processes that it is undergoing, the subject of my train of thought about
the state of the weather in Vienna. But we can’t say that all three of
these biological entities (plus indefinitely many more) are having those
same thoughts, on pain of falling into the very ‘multiple thinkers’ problem
of which my sort of substance dualism itself stands falsely accused. We
should conclude, then, that none of these biological entities qualifies as the
subject of my train of thought in virtue of the biological processes that it
is undergoing. And hence we should further conclude that the psychological
process in question—my train of thought about the state of the weather in
Vienna—is neither identical with nor constituted by any of the biological
processes being undergone by those biological entities: for if it were, then,
contrary to what we have already just concluded, one or more of those
biological entities would, after all, qualify as being the subject of my train
of thought about the state of the weather in Vienna.
Exactly the same form of argument can be run with my brain, or any
biologically distinguishable region of it, being substituted for my whole
body in the foregoing version of the argument. In place of minor parts of
my body, such as my right ear or left little finger, we need merely substitute
minor parts of my brain or of the selected region of it, such as a few cells
here or there. No doubt, of course, I couldn’t have my train of thought
Substance Causation, Persons, and Free Will 171
about the state of the weather in Vienna if my brain was lacking some
quite large part. But that is quite irrelevant and obviously doesn’t imply that
precisely that part of my brain qualifies just as well as I do as being the subject
of the train of thought in question. Indeed, various different parts of my
brain will doubtless be necessary for my capacity to think various different
thoughts, whether at the same or at different times, but these thoughts
will nonetheless all be thoughts of a single subject—me. Clearly, there can
be no identifiable subset of organic parts of my brain of which it can be
said that that subset is uniquely necessary for the having of all and only
my thoughts. The lesson of all this is, as I suggested earlier, that the unity
of thought and reasoning requires a unity of their psychological subject
which is inconsistent with the identity conditions of biological substances,
given the enormous compositional complexity and mutability of the latter.
(Readers who are already familiar with Chapter 5 will recognize here a
version of what I there called the ‘unity argument’.)
so far reached bear upon the issues in the philosophy of action with which
we began. For this purpose, the psychological power that we must focus
on is the power of will, or volition.
Talk of volition in the philosophy of action is still somewhat unfash-
ionable.¹¹ This is partly because there is a broad consensus that the mental
causes of human action are combinations of belief and desire or, more
precisely—since such items are taken to be mental states rather than mental
events—the ‘onsets’ of combinations of belief and desire.¹² Adherents of this
view are, of course, event causalists and hostile to classical agent causalism.
Volitionism has remained a minority view in the philosophy of action
because it is opposed both by adherents of the belief–desire approach and
by classical agent causalists. One thing that is wrong with the belief–desire
approach, quite apart from its assumption of the sovereignty of event caus-
ation, is that it makes no room at all for the executive element in intentional
action. It effectively treats all intentional actions as mere happenings with
distinctive mental causes. On this approach, the only difference between
my arm’s rising as a result of a muscular spasm and its rising because I raised it
intentionally is that, in the latter case, the arm-rising has amongst its causes
certain mental events—the onsets of certain beliefs and desires—where
the beliefs and desires in question also constitute my ‘reasons’ for raising
my arm on that occasion. But this analysis precisely leaves out what is
distinctively active about my raising my arm and ‘passive’ about my arm’s
rising as a result of a muscular spasm. In the former case, I did something
to cause my arm to rise, but not so in the latter. What did I do? The
classical agent causalist, while agreeing that there is something distinctively
‘active’ about my raising my arm, will decline to answer this question: such
a theorist will contend that, because the action is a so-called ‘basic’ one,
there is nothing more to be said than that I simply caused my arm to rise,
as a primitive case of agent causation. However, not only does this make
little sense metaphysically, it is false to the phenomenology of intentional
action. When I raise my arm, I feel that I am doing something to cause it
to rise—something that I could characterize as trying to raise it. I could do
this same thing on another occasion and fail to raise my arm—for instance,
¹¹ But for an important defence, see Carl Ginet, On Action (Cambridge: Cambridge University Press,
1990). I defend volitionism myself in my Subjects of Experience, ch. 5.
¹² The locus classicus for this view is, of course, the work of Donald Davidson: see his Essays on Actions
and Events (Oxford: Clarendon Press, 1980).
Substance Causation, Persons, and Free Will 173
¹³ Locke famously held precisely this view: for discussion, see my Locke on Human Understanding
(London: Routledge, 1995), ch. 6.
174 Persons, Rational Action, and Free Will
there are two ways of interpreting this response, only one of which is really
intelligible. On the correct interpretation, it was the contents of my belief
and desire—what it was that I believed and desired—that constituted my
reason for acting as I did. On the incorrect interpretation, it was my belief
and desire themselves—those very mental states of mine—that constituted
my reason for acting as I did. This, I believe, makes no sense. A reason
for acting is a ground which would justify the performance of a certain
action. The contents of belief and desire can constitute such grounds, but
a belief or desire as such cannot, because these are not even items in the
right ontological category to constitute grounds.
Now, of course, a ground may exist which constitutes a reason for me
to act in a certain way in certain circumstances, even though I am unaware
of that ground. For instance, if a stone is about to fall on my head, this fact
constitutes a good ground for me to take avoiding action by jumping out
of the way. Unless I am aware of the fact, however, I will be unlikely to
take the appropriate action. I need to be aware of how the world is and
of how it had better be to serve my interests, if I am to act rationally in
the furtherance of those interests. But it is not my believing that the world
is thus-and-so and my desiring to further those interests that constitute a
reason for me to act in a certain way, just its being thus-and-so and my
having those interests. (In saying this I am advocating a so-called ‘externalist’
account of reasons for action,¹⁴ about which I shall say much more in the
next two chapters.) I must obviously be aware of that reason if I am to act
‘in the light of’ it, but we should not confuse the reason itself with the
mental states involved in my awareness of that reason. Most importantly,
moreover, to act ‘in the light of’ a certain reason is not to be caused to act
in a certain way by the mental states involved in one’s awareness of the
reason. In fact, to be caused to act by certain of one’s beliefs and desires is
precisely not to act freely and rationally, even if the ensuing behaviour is
not contrary to reason. This is because an action which is so caused lacks
the crucial element of choice, which is precisely the contribution of the
will.¹⁵ To act rationally one must freely choose to act on a reason of which
¹⁴ See, for example, Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2000),
pp. 15–17.
¹⁵ See, further, my An Introduction to the Philosophy of Mind (Cambridge: Cambridge University Press,
2000), pp. 252–62, and compare John R. Searle, Rationality in Action (Cambridge, MA: MIT Press,
2001), ch. 3.
176 Persons, Rational Action, and Free Will
decayed in a certain period of time. But this doesn’t mean that the chances
of any particular atom’s decaying at any given time, or during any given
period of time, are fixed in any way by prior events. When such an atom
decays, it does so without any cause and, of course, for no reason at all.
The difference with the will is just that, although its exercise has no cause,
it is characteristically exercised for a reason, which the person in question is
aware of and normally able to articulate. And the will, recall, is a power of
the person and not of the person’s body nor of any distinctive part of it, so
we are not required to suppose that purely physical substances have rational
powers.
Even so, it may be wondered how exercises of the will can be causally
efficacious in bringing about bodily effects: that is to say, how a person,
by exercising his or her will, can cause part of his or her body to move.
There would be a problem here if we had to suppose that all bodily
effects, because they are physical events, must have sufficient causes in
the form of prior physical events. But we have already rejected any strong
principle of physical causal closure that would have this requirement. Again,
it may be objected that our proposal requires some mysterious form of
‘downward causation’ of mental acts upon the behaviour of physical objects,
in violation of the causal autonomy of the physical world.¹⁶ But what can
justify a belief in such autonomy, other than a discredited form of causal
closure principle of the sort that we have just rejected? I am not suggesting
that the will somehow exploits ‘gaps’ in physical causation left open by
the indeterminacy of the quantum world, as though my will becomes
causally efficacious by bringing about such events as the radioactive decay
of atoms. For events such as the latter really are uncaused and so, a fortiori,
not even caused by exercises of the will. Moreover, if they were so caused
at some times and not at others, quantum physical statistical data such as
those relating to atomic half-lives would be inexplicable. My appeal to
the case of radioactive decay was just intended to establish the fact that
spontaneous powers exist in nature, so that not all physical events have
sufficient physical causes, contrary to the strong causal closure principle.
Physical events brought about by the exercise of the will are, of course,
caused, at least in part by those exercises, that is, by our volitions. Indeed,
¹⁶ See, for example, Jaegwon Kim, ‘ ‘‘Downward Causation’’ in Emergentism and Nonreductive
Physicalism’, in A. Beckermann, H. Flohr, and J. Kim (eds), Emergence or Reduction? Essays on the
Prospects of Nonreductive Physicalism (Berlin: de Gruyter, 1992).
178 Persons, Rational Action, and Free Will
we may suppose those physical events to have sufficient causes (albeit not
wholly physical ones), at least at the times at which the volitions in question
occur. However, the volitions themselves, I have said, are uncaused—and
they, by my account, are not physical events, since they are exercises of a
non-physical mental power. The only significant implication of all this as
far as physical events are concerned is just that some of them are such that,
at certain times prior to their occurrence, they do not possess sufficient
physical causes—for instance, a certain willed movement of my arm does
not possess a sufficient physical cause at any time prior to my volition to
move my arm in that way. But this is consistent with what is currently
empirically known about the neurophysiology of voluntary action.¹⁷ And,
as we have just seen, no acceptable principle of physical causal closure
stands in the way of its truth. So, while I would not pretend to have
explained how a person can, by willing, effect physical changes in his or
her body—and maybe this is something that is destined forever to remain
a mystery to us—I see no philosophical or scientific ground for denying
that when we do so we exercise a non-physical power which is at once
spontaneous and rational, in the sense that I have proposed.
¹⁷ See Benjamin Libet, Neurophysiology of Consciousness: Selected Papers and New Essays (Boston:
Birkhäuser, 1993) and B. Libet, A. Freeman, and K. Sutherland (eds), The Volitional Brain: Towards a
Neuroscience of Free Will ( Thorverton: Imprint Academic, 1999).
9
Rational Selves and Freedom
of Action
¹ It should be evident, of course, that when I spoke favourably, in Chapter 1, of ‘naturalistic dualism’,
I was not using the term ‘naturalistic’ in this commonly received sense.
180 Persons, Rational Action, and Free Will
reasons for belief —I need to say something about the notion of a reason
for action. Only at the very end of the chapter shall I say something about
the connection between reasons for belief and reasons for action.
² Unfortunately, the English word ‘belief ’ is ambiguous, being used sometimes to denote a kind of
psychological state—which is how I myself always use the word in this book, when I am deploying it
as a technical philosophical term—and sometimes the content of a state of that kind. ( The latter use is
illustrated by a remark such as ‘One of my beliefs is that extraterrestrial life exists somewhere in the
universe’.) We shall shortly see that this distinction is immensely important for present purposes, so that
it is vital not to confuse the two uses.
Rational Selves and Freedom of Action 181
³ See Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2000), p. 125.
182 Persons, Rational Action, and Free Will
⁴ See Donald Davidson, ‘Freedom to Act’, in his Essays on Actions and Events (Oxford: Clarendon
Press, 1980). The example appears on p. 79.
Rational Selves and Freedom of Action 183
⁵ See Donald Davidson, ‘Actions, Reasons, and Causes’, in his Essays on Actions and Events.
⁶ Compare John R. Searle, Rationality in Action (Cambridge, MA: MIT Press, 2001), p. 16.
184 Persons, Rational Action, and Free Will
it in due course. Let me say at once, though, that talk of such mental acts
raises no spectre of a vicious infinite regress. There is no danger of our
being compelled to say that each mental act of choice must itself be the
consequence of a prior act of choice. If choices are to play the role in
rational action that I have assigned to them, they must be uncaused and
so ‘free’ in the libertarian sense of ‘freedom’. A fortiori, they cannot be
caused by prior choices. Choices can be causes of subsequent events, but
they cannot be effects of prior ones. I shall return to this matter shortly.
First, there are some other matters to clear up.
A reason for action, I have claimed, is some fact or state of affairs.⁷ More
particularly, it is a fact or state of affairs which justifies a particular course
of action—as, for example, the fact of the roof-slate’s falling justifies the
action of jumping out of its way. But two problems are likely to be raised
for me here. First, what do I say about an agent who jumps in the mistaken
belief that a slate is about to fall on his head? Second, how can we say that
it is rational, tout court, to jump out of the way of a falling slate: mustn’t we
say, rather, that it is rational to do this if one desires not to be hit by it?
In which case, mustn’t we admit, after all, that psychological states of the
agent are at least partly constitutive of an agent’s reasons for action?
In answer to the first question, I could say the following. Whether or
not an agent is mistaken in believing that a slate is about to fall on his head,
that a slate is about to fall on one’s head is, by any standard, a good reason to
jump out of the way. If an agent is to act rationally in a certain manner, he
must be aware of a reason for so acting. But ‘being aware of a reason’ in this
sense doesn’t necessarily require that the reason should actually obtain. Here
we might usefully deploy Alvin Plantinga’s style of talking about facts and
states of affairs.⁸ A state of affairs may or may not actually obtain, but if it
does then it is a fact and may consequently be expressed by a true sentence.
Reasons for acting are states of affairs and one must be aware of them if one
is to act rationally. If the world is as the agent believes it to be, a reason for
acting of which he is aware will be a state of affairs that obtains, and so a fact.
If the relevant beliefs of the agent are false, however, he may still be aware
of the same reason for acting and act rationally in the light of that reason,
but the reason for which he acts will turn out to be a state of affairs that does
⁷ I shall modify this claim in Chapter 10, but for present purposes I regard it as being close enough
to the truth.
⁸ See Alvin Plantinga, The Nature of Necessity (Oxford: Clarendon Press, 1974), pp. 44ff.
Rational Selves and Freedom of Action 185
not actually obtain.⁹ In that case, the agent might well have done better to
have acted differently, but he did not act irrationally, since he still acted for
a reason that he chose to act upon. Since reasons are not causes, it does not
matter that an agent’s reason for acting in a certain way may on occasion be a
state of affairs which does not obtain and hence which is not a fact. Reasons
for action, even in the case of agents with mistaken beliefs, remain perfectly
objective and, typically, non-psychological in character. (Here I should
concede that I am glossing over the further question of what distinguishes
‘good’ from ‘bad’ reasons for action, if indeed such a distinction is a proper
one to make. No doubt, for an action to be rational, the reason in the light
of which the agent performed that action must at least be one that the agent
judged to be ‘good’, in the sense that he judged the state of affairs in question
to justify acting as he did. But this is a further complication which I do not
want to go into at present, as it would distract me from my main theme.)
Perhaps the foregoing answer to the first question will not satisfy
everyone. For those who are dissatisfied with it, I have a rather different
answer which they may prefer. (For my own part, I am presently undecided
as to which is the better answer, although I am somewhat more inclined
towards the one that I am about to articulate.) We could say that if an
agent jumps out of the way in the light of a mistaken belief that a slate is
about to fall on his head, then he does not really have a reason to act in
that way on that occasion. At the same time, however, we can and should
insist that such an agent is not acting irrationally merely insofar as his belief
is false—for if it had been true, he really would have had a reason to act in
precisely the way that he does. On this view of the matter, only facts—that
is, states of affairs which obtain—can actually be reasons for action, but an
agent can, nonetheless, act for a would-be reason which does not actually
obtain and in doing so act rationally in precisely the same sense in which
an unmistaken agent can. Most importantly, we do not have to regard
the mistaken agent’s false belief —a psychological state of the agent—as
being his reason for action, and thus treat mistaken and unmistaken agents
asymmetrically. Strictly and literally speaking—the externalist may say—an
agent who acts in the light of a false belief has no reason for acting as he does,
but this by no means implies that such an agent has not properly exercised
his powers of reason in deciding how to act and so is in that sense acting
¹⁰ Again I should remark that I shall modify this account somewhat in Chapter 10, although in a
way that need not detain us here since it makes no concession to the ‘internalist’ approach to reasons
for action.
Rational Selves and Freedom of Action 187
¹¹ See J. L. Mackie, Ethics: Inventing Right and Wrong (Harmondsworth: Penguin Books, 1977),
pp. 38ff.
188 Persons, Rational Action, and Free Will
as to why the agent chose as he did and the most plausible explanation is
that the agent’s mental state immediately prior to the act of choice caused
it to occur. So, it may be said, at some stage in the agent’s deliberations
he formed the belief that, all things considered, asking a question was
the sensible thing to do and this belief, together with his desire to do
what he considered sensible, caused him to choose to act in the way he
did, by raising his arm. However, now we are back to the picture of
rational action according to which a rational action is one that is not freely
chosen in the light of the agent’s beliefs and desires but caused by the
agent’s beliefs and desires. And this is precisely the picture that I was at
pains to reject earlier. I think it is a picture that is radically false to the
phenomenology of rational action and, worse still, impossible to square
with any conception of rationality that we can really make sense of. That
is to say, when we take ourselves to be acting rationally it never seems
to us that we are being caused to act in the ways we do by our beliefs
and desires, and once we suppose that, on a given occasion, we were in
fact caused to act in a certain way by our beliefs and desires, we find
ourselves obliged to withdraw any claim to have acted rationally on that
occasion.
Let us go back to the question that precipitated these difficulties. The
question was why the agent chose to raise his arm. I suggested that what
prompts this question is the thought that there must surely be some
explanation as to why the agent chose as he did—and I then suggested
that, at least for many philosophers, the most plausible explanation is that
the agent’s mental state immediately prior to the act of choice caused it
to occur. But suppose that the question is addressed to the agent and he
is asked ‘Why did you chose to raise your arm?’ The agent will no doubt
answer that he chose to raise his arm in order to ask a question. If pressed a
little further, he will say, perhaps, that he chose to raise his arm because he
had an important question to ask which no one else was likely to raise. In
other words, he will cite certain reasons that he was aware of for raising his
arm. Pressed still further, he may admit that he was also aware of certain
reasons for refraining from raising his arm. ‘So,’ we may ask him, ‘why did
you chose to act on the first set of reasons and not on the second?’ He may
answer, ‘Well, after some deliberation I came to the conclusion that the
first set of reasons was better than the second, so I chose to act on the first.’
For most ordinary folk, this is where the questioning would come to an
190 Persons, Rational Action, and Free Will
end. But philosophers, like little children, sometimes don’t know when to
stop asking ‘why?’ They may now ask, ‘But why did you choose to act on
the reasons that you thought were the better ones?’ Here our agent, if he
is not a philosopher, is likely to return a blank stare of bewilderment. If he
says anything, it is likely to be sarcastic or rhetorical, such as, ‘Well, don’t
you usually choose to act on what you consider to be the best reasons?’
Our philosophical interrogator may now pounce and say, ‘So, you really
couldn’t help choosing as you did, once you came to the conclusion that
one set of reasons was better than the other—in other words, your coming
to that conclusion caused you to choose as you did.’ But our agent is likely
to resist this suggestion very forcefully and reply, ‘No, of course I could
have chosen to act otherwise and nothing made me choose to act as I
did—although if I had chosen to act otherwise, it would have been against
my better judgement, and sometimes I have done precisely that and later
regretted it.’
¹² See Peter van Inwagen, ‘Free Will Remains a Mystery’, in Robert Kane (ed.), The Oxford
Handbook of Free Will (Oxford: Oxford University Press, 2002).
Rational Selves and Freedom of Action 191
libertarian will want to represent this choice as being genuinely free, in the
sense of being uncaused. This means that, according to the libertarian, the
state of the universe prior to Alice’s act of choice contained no causally
sufficient condition of that act and, up until the moment of choice, there
were possible futures in which Alice tells the truth and also possible futures
in which she lies. Suppose, then, that all of the circumstances leading up to
the moment of Alice’s choice were to be replicated or ‘replayed’ a number
of times—say, 1,000 times. Since we are supposing that Alice’s choice in
the original ‘play’ of these circumstances was causally undetermined by
preceding states of affairs, we have to suppose that Alice’s choice is likewise
causally undetermined by preceding states of affairs in each of the replays.
So, it seems, we must suppose that in some of the replays she chooses as she
did originally and tells the truth and in the others she chooses differently
and tells a lie. It seems that there will be some specific number of replays
in which she chooses to tell the truth—say, 513—and in the remaining
487 replays she chooses to tell a lie. But it was arbitrary to pick 1,000
as the number of replays. We can generalize and say that there will be
some particular proportion of all the possible replays of Alice’s situation
in which she chooses to tell the truth. This proportion would have to
be less than unity, for to suggest that Alice would have chosen to tell
the truth in every possible replay of the situation is surely to imply that
her choice was, after all, determined by preceding states of affairs. But
if the proportion is less than unity—say, it is one half—then what this
appears to signify is that Alice’s choice in the original situation was simply
a chance event whose objective probability of occurrence was 0.5, or 50
per cent.
The first thing that I want to say about this intriguing argument is that,
even if we can make complete sense of the notion of a ‘replay’ of the
circumstances preceding Alice’s act of choice and even if we can agree
that, given that her choice was causally undetermined, it follows that in
some replays she would choose to tell the truth while in others she would
choose to tell a lie, it doesn’t follow that there must be some determinate
proportion of all possible replays in which she chooses to tell the truth. In
other words—and avoiding the perhaps questionable notion of a ‘replay’
of Alice’s situation—the fact that Alice might have chosen otherwise than
she actually did does not imply that there was a certain numerically precise
objective probability of her choosing as she did in the one ‘play’ that
192 Persons, Rational Action, and Free Will
actually did happen. Consequently, it is far from clear that we are entitled
to regard her choice as a ‘chance event’. Compare Alice’s situation with
that which occurs in a genuine game of chance, such as a dice game. Let
us suppose that on a certain occasion a perfectly fair die is thrown and it
lands with the six uppermost. And let us suppose that its so landing is a
genuinely indeterministic event, like the spontaneous decay of a radium
atom—even though there is reason to doubt that this is really so. If we now
contemplate all possible ‘replays’ of this situation, we do indeed have some
reason to suppose that a definite proportion of them—one sixth—will
result in the die landing with the six uppermost, and hence that there
was an objective probability or chance of one in six of this happening
when the die was actually thrown. Our reason to think so is obviously
based on symmetry considerations concerning the structure of the die.
However, nothing remotely like this reason is able to support a similar
conclusion in Alice’s case. It may be objected that we likewise lack any
such reason to assign a numerical degree of chance to an indeterministic
atomic event, such as the spontaneous decay of a radium atom, and yet
numerical probabilities are associated with such events. However, this is
because large amounts of statistical data enable us to assign ‘half-lives’ to
radioactive isotopes. Nothing similar even to this is actually available in the
case of Alice. By being invited to contemplate a large number of ‘replays’ of
Alice’s situation, we are being invited to suppose that statistical data of this
sort is in principle available concerning Alice’s situation, albeit data that is
not all drawn from the actual world, as in the case of the data which enables
us to assign half-lives to radioactive isotopes: rather, the ‘data’ in Alice’s case
is, as it were, distributed across all the possible worlds compatible with the
history of actual events up until the moment of Alice’s choice. I suppose
that there are some conceptions of possible worlds that would support the
idea that such ‘data’ really exists, but for my own part I see no reason to
think that it does. I see no reason to think that even God could assign an
objective numerical probability or degree of chance to Alice’s choosing to
tell the truth.
The next thing I want to say about the Alice thought experiment is this.
Alice, we have said, deliberated at some length before deciding how to
act and choosing accordingly. The way in which the thought experiment
is set up requires us to suppose that in all ‘replays’ of her situation her
deliberations proceed in exactly the same way that they actually did, but
Rational Selves and Freedom of Action 193
that at the end of some of these replays she still chooses to tell the truth
while at the end of others she chooses instead to tell a lie. However, if
Alice deliberates and chooses rationally, at the end of her deliberations
she will have formed a judgement as to whether the reasons in favour
of telling the truth are better than those in favour of telling a lie. Now,
certainly, if Alice’s choice was genuinely undetermined, it was possible
for her to have chosen to tell a lie in spite of having formed such a
judgement. On the other hand, there is a clear sense in which, in the
light of that judgement, Alice would have been acting irrationally in telling
a lie. A libertarian should happily accept that our freedom to choose is
a freedom even to choose irrationally—and we all recognize, I think,
that sometimes we do make irrational choices. So-called ‘akratic’ action
involves precisely this, it would seem, for such an action is one that the
agent chooses to perform ‘against his better judgement’, as we say. At
the same time, little sense can be made of the notion of a rational agent
who frequently, or as often as not, chooses to act irrationally rather than
rationally.
We have to distinguish clearly between two different conceptions of how
Alice might have chosen differently. She might have chosen differently even
if the result of her deliberations had been the same, that is, even if she
had formed the same judgement about which action had better reasons in
favour of it. A libertarian must certainly say this. But another thing that
can and should be said by the libertarian is that Alice might have deliberated
differently and formed a different judgement as to which action had better
reasons in favour of it, in which case, too, she might have chosen to act
differently—only this time she would not have chosen ‘akratically’. The
reason why the libertarian should say this is that deliberation itself should be
seen as a process which involves choice. Deliberation isn’t simply a matter
of the agent’s being confronted with ready-made reasons in favour of doing
this or that action and then having to weigh those reasons against one
another. Deliberation involves the active seeking of reasons for or against
carrying out one of a range of alternative possible actions. Consequently,
it involves the agent in making choices about what sources of evidence to
consult or think about with a view to revealing reasons for or against acting
in a certain way. We can choose not only how to move our bodies but
also how to direct our thoughts, and the latter kind of choice is essentially
involved in all processes of deliberation. But if this is what deliberation
194 Persons, Rational Action, and Free Will
¹³ See, further, Storrs McCall and E. J. Lowe, ‘Indeterminist Free Will’, Philosophy and Phenomeno-
logical Research 70 (2005), pp. 681–90.
Rational Selves and Freedom of Action 195
cannot, for that very reason, be controlled in its exercise of control, if this
is understood as involving some further agency. If I say to somebody to
whom I assign some special responsibility, ‘OK, you are in control now,’
I cannot consistently then go on to say, ‘But remember that I shall be
controlling your every move.’
To this it may be objected that so-called ‘Frankfurt-style cases’ suggest
otherwise.¹⁴ Couldn’t it, in principle, be the case that some mad scientist
is monitoring my brain to see what choices I make, allowing only those
choices to take effect which are compatible with what he decides should
happen—but that it just so happens that everything I choose to do coincides
with his plans? In that case, isn’t there a sense in which I am controlling
what I do, by exercising my power of choice, even though he has control
over me because he would prevent me from doing anything incompatible
with his plans? Notice, however, that even if this thought experiment
makes sense—which may be questionable—there is no suggestion that
the mad scientist has control over my power of choice, only control over its
effects. He can make sure that if I choose to do something incompatible
with his plans, my choice will be ineffective. But nothing has been said to
suggest that he can make sure how I exercise my power of choice—what
choices I make. Nor does it make sense that he could do this. For him
to have, per impossibile, control over how I exercise my power of choice
would be for him to deprive me of that power, so that this is, logically,
not a sort of control that he can have over me. If he had such control
over me that I no longer had any power of choice, he would now be
the only one of us capable of possessing such a power: there would not
be two distinct powers of choice, mine and his, with his the dominant
power. He would be in control of me, but not in control of my power
of choice, for I would have none. By the same token then, it would
make no sense for me to be in control of my own power of choice,
for then I would have to have two such powers, one dominating the
other—and yet the supposedly dominated power of choice would simply
be extinguished, not controlled, by the supposedly dominating power. This,
I think, demonstrates that it is incoherent to demand of the libertarian that
he provide an account of how we have ‘control’ over our power of choice.
¹⁴ See Harry G. Frankfurt, ‘Alternate Possibilities and Moral Responsibility’, in his The Importance of
What We Care About: Philosophical Essays (Cambridge: Cambridge University Press, 1988).
Rational Selves and Freedom of Action 197
Having a power of choice gives us all the control we could ever have
or need.
¹ See Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2000).
200 Persons, Rational Action, and Free Will
² See Donald Davidson, ‘Actions, Reasons, and Causes’, in his Essays on Actions and Events (Oxford:
Clarendon Press, 1980).
Needs, Facts, Goodness, and Truth 201
point that distinguishes the externalist from the internalist view of reasons
for action.
Davidson famously argued that the only way of identifying the reason for
which an agent acted on a given occasion—as opposed to reasons which
the agent may have possessed for so acting but did not in fact act upon—is
to pick it out as the one that caused the action. He, of course, was assuming
the truth of internalism and took the reason in question to be a belief of
the agent, not just the content of any such belief. (More precisely, he took it
to be a conjunction of certain beliefs and desires of the agent.) But it may
seem that internalism is supported by the fact that Davidson has this way of
answering the question as to which reason was, as things turned out, the
agent’s actual reason for acting as he did. Not so. First of all, Davidson’s
account famously falls foul of the problem of wayward or deviant causal
chains, which has never been convincingly solved. The problem involves
cases, like Davidson’s own well-known example of the nervous climber,
in which belief and desire both cause and ‘rationalize’ an action which is,
nonetheless, unintentional on the part of the agent.³ Second, and more
importantly in the present context, the libertarian account of freedom that
I am endorsing has a satisfactory answer to Davidson’s question on its own
terms. When an agent acts freely and rationally, the reason for which he acts
is simply the reason on which he chose to act. The agent himself determines,
by exercising his power of choice, which of various reasons for or against
acting in a certain way he acts upon in any given case—selecting, of course,
only from those reasons of which he is aware as a result of his deliberations,
for one evidently cannot choose to act in the light of a reason of which
one is not aware.
³ See Donald Davidson, ‘Freedom to Act’, in his Essays on Actions and Events.
Needs, Facts, Goodness, and Truth 203
⁴ I have in mind here J. L. Mackie’s famous ‘argument from queerness’: see his Ethics: Inventing Right
and Wrong (Harmondsworth: Penguin Books, 1977), pp. 38–42.
206 Persons, Rational Action, and Free Will
that belong to the English language and have their equivalents in all other
human languages known to us. Because it is the case that what we believe,
when we believe truly, is a fact, it is also the case that facts are those items
that are apt to constitute reasons for belief. A fact can be a reason for belief,
because it can entail or confirm—more generally, it can support —the truth
of a belief. For example, the belief that the earth is round is supported by
the fact, amongst others, that ships disappear over the horizon as they get
more distant from the observer.
In order to see what sort of item can constitute a reason for action,
it will help to consider how the language of action both resembles
and differs from the language of belief. Beliefs, we have noted, are
propositional attitudes: they are psychological states whose ‘contents’ are
propositions. We may say this much uncontentiously, without delving
into the metaphysics of propositions. For present purposes, we need not
concern ourselves overmuch with what propositions themselves are. Now,
actions are certainly not propositional attitudes. For one thing, they are
events rather than states. Still, an action has an agent, just as a belief has a
subject. And in both cases this is a person, or at least an animal of some kind.
Indeed, it would seem to be the case that those things that are subjects of
belief are necessarily also agents of actions, and vice versa: only a believer
can be an agent, and only an agent can be a believer.
However, while subjects ‘hold’ beliefs, they ‘perform’ actions. We can
regiment belief-statements in the general form ‘S believes that p’. Here
the subject of the belief is S, the belief held by the subject is the belief
that p, and the content of the subject’s belief is the proposition p. How
can we analogously regiment action-statements? In the following way,
which brings out both the similarities and the differences between belief
and action. The general form of an action-statement, we may say, is ‘A
does x’. Here A is the agent of the action, the action performed by the
agent is the action of doing x, and finally, we might say, the content
of the agent’s action is the act x. An example will help. Suppose that
A raises his arm. We may regiment this in the form—which admittedly
sounds either slightly strained or somewhat archaic—‘A does raise his arm’.
The point to get across, however, is that what the agent does—raise his
arm—is distinguishable from his action of raising his arm in a manner that is
analogous to the distinction between a belief and its propositional content.
I propose to reserve the term ‘act’ for the ‘content’ of an action, and refer
Needs, Facts, Goodness, and Truth 207
⁶ See, further, my The Four-Category Ontology: A Metaphysical Foundation for Natural Science (Oxford:
Clarendon Press, 2006), ch. 11.
Needs, Facts, Goodness, and Truth 211
What I have said about this matter is extremely sketchy and may fail
to convince for that reason. In a fuller account, I would need to be
much more specific about the individuation of acts, actions, and needs. I
have spoken, for instance, of a need to eat, an action which satisfies that
need—someone’s eating something—and the act that is the ‘content’ of
such an action, the act of eating something. But, of course, we could and
perhaps should be much more specific about all items of these kinds. In like
manner, we may loosely speak of the belief that it is raining, appreciating
that we should more precisely talk about a particular person’s belief on
a particular occasion that it is raining in a specified place at a specified
time. Similar precisifications no doubt can and in principle should be
included in our talk about acts, actions, and needs. We could talk, thus, of
a particular person’s action on a particular occasion of eating some specific
object in a specified place at a specified time. However, I see no reason
to suppose that the main thrust of what I have said so far about the logic
of action and the relationship between goodness and needs would be at
all affected by such precisification. Sometimes being precise is not a virtue
in philosophy, because it hinders our ability to see the main features of a
position clearly.
Another issue that I have skirted around is the following apparent
difference between goodness and truth: goodness seems to be a matter of
degree, while truth does not. Relatedly, it may be said that the goodness
of some actions outweighs that of others, but that all truths have the same
weight. However, these apparent differences between goodness and truth
are not incontestable. Some logicians maintain that there are degrees of
truth and try to explain various semantic phenomena of vagueness in these
terms. Again, it is often urged by philosophers of pragmatist leanings that
some truths are more important than others. In view of such disagreements,
I think it better to leave these questions aside for present purposes, important
though they are.
resembles and differs from the logic of belief and how practical rationality
both resembles and differs from theoretical rationality. Needs constitute
objective reasons for actions. They are what make actions good, analogously
to the way in which facts make beliefs true. Very arguably, just as we are not
obliged to invoke falsehood-makers to account for false beliefs, neither are
we obliged to invoke badness-makers to account for bad actions. Plausibly,
just as a false belief is one that fails to correspond to any fact, so a bad
action is one that fails to correspond to any need. Be that as it may, the
important point, ontologically speaking, is that needs exist and are items
that belong to a distinctive ontological category, different from that of facts.
To articulate clearly the nature of needs, we would have to provide a full
account of the existence and identity conditions of needs, and that is not
something that I have attempted to do here. It is a very big task which goes
far beyond the limited ambitions of the present chapter. The central lesson,
however, of the present chapter is that the world is not just a world of states
of affairs—to echo the title of a well-known and excellent book by David
Armstrong.⁷ To put it another way, the world is not just, in the famous
words of Ludwig Wittgenstein, ‘everything that is the case’.⁸ Rather, it is
a world that is also permeated by needs—to the extent, at least, that it
is a world inhabited by agents who possess reasons for action. Whether
those needs can be accounted for in wholly ‘naturalistic’ terms—whatever
precisely that is taken to mean—is another question that is too big for me to
address on this occasion. I would only emphasize that if, by a ‘naturalistic’
account of needs, what is meant is an account which somehow reduces the
existence of needs to the existence of certain scientifically describable facts,
then I hold out no hope for such an account.
⁷ See D. M. Armstrong, A World of States of Affairs (Cambridge: Cambridge University Press, 1997).
⁸ See Ludwig Wittgenstein, Tractatus Logico-Philosophicus, trans. C. K. Ogden (London: Routledge
and Kegan Paul, 1922), 1.
Bibliography
Danto, A. C., ‘Basic Actions’, American Philosophical Quarterly 2 (1965), pp. 141–8.
Davidson, D., ‘Actions, Reasons, and Causes’, Journal of Philosophy 60 (1963),
pp. 685–700, reprinted in his Essays on Actions and Events.
‘Agency’, in R. Binkley, R. Bronaugh, and A. Marras (eds), Agent, Action,
and Reason (Toronto: University of Toronto Press, 1971), reprinted in his Essays
on Actions and Events.
Essays on Actions and Events (Oxford: Clarendon Press, 1980).
‘Freedom to Act’, in his Essays on Actions and Events.
‘Mental Events’, in his Essays on Actions and Events.
Deecke, L., Scheid, P., and Kornhuber, H. H., ‘Distribution of Readiness Potential,
Pre-Motion Positivity and Motor Potential of the Human Cerebral Cortex
Preceding Voluntary Finger Movements’, Experimental Brain Research 7 (1969),
pp. 158–68.
Dennett, D. C., Consciousness Explained (London: Penguin, 1992).
Descartes, R., Meditations on First Philosophy, in J. Cottingham, R. Stoothoff, and
D. Murdoch (eds), The Philosophical Writings of Descartes (Cambridge: Cambridge
University Press, 1984).
Dretske, F., ‘Mental Events as Structuring Causes of Behaviour’, in J. Heil and
A. R. Mele (eds), Mental Causation (Oxford: Clarendon Press, 1993).
Eccles, J. C., The Evolution of the Brain: Creation of the Self (London: Rout-
ledge, 1989).
Foster, J., The Immaterial Self: A Defence of the Cartesian Dualist Conception of the
Mind (London and New York: Routledge, 1991).
Frankfurt, H. G., ‘Alternate Possibilities and Moral Responsibility’, in his The
Importance of What We Care About: Philosophical Essays (Cambridge: Cambridge
University Press, 1988).
Garber, D., ‘Mind, Body, and the Laws of Nature in Descartes and Leibniz’,
Midwest Studies in Philosophy 8 (1983), pp. 105–33.
Geach, P. T., Truth, Love and Immortality: An Introduction to McTaggart’s Philosophy
(London: Hutchinson, 1979).
Ginet, C., On Action (Cambridge: Cambridge University Press, 1990).
Hacking, I., Representing and Intervening (Cambridge: Cambridge University Press,
1983).
Hardcastle, V. G., ‘Psychology’s Binding Problem and Possible Neurobiological
Solutions’, Journal of Consciousness Studies 1 (1994), pp. 66–90.
Hart, W. D., The Engines of the Soul (Cambridge: Cambridge University Press,
1988).
Heil, J. and Mele, A. R. (eds), Mental Causation (Oxford: Clarendon Press, 1993).
Hornsby, J., Actions (London: Routledge and Kegan Paul, 1980).
Hughes, J., Louw, S., and Sabat, S. (eds), Dementia: Mind, Meaning, and the Person
(Oxford: Oxford University Press, 2006).
Bibliography 215
Olson, E. T., The Human Animal: Personal Identity without Psychology (New York:
Oxford University Press, 1997).
Owens, D., Causes and Coincidences (Cambridge: Cambridge University Press,
1992).
Papineau, D., ‘Mind the Gap’, in J. E. Tomberlin (ed.), Philosophical Perspectives,
12: Language, Mind and Ontology (Oxford: Blackwell, 1998).
Philosophical Naturalism (Oxford: Blackwell, 1993).
Peacocke, C. A. B., Holistic Explanation: Action, Space, Interpretation (Oxford:
Clarendon Press, 1979).
Plantinga, A., The Nature of Necessity (Oxford: Clarendon Press, 1974).
Popper, K. R. and Eccles, J. C., The Self and its Brain: An Argument for Interactionism
(Berlin: Springer, 1977).
Robinson, W. S., Understanding Phenomenal Consciousness (Cambridge: Cambridge
University Press, 2004).
Ryle, G., The Concept of Mind (London: Hutchinson, 1949).
Searle, J. R., Rationality in Action (Cambridge, MA: MIT Press, 2001).
Seuren, P. M. (ed.), Semantic Syntax (Oxford: Oxford University Press, 1974).
Shoemaker, S., ‘Causal and Metaphysical Necessity’, Pacific Philosophical Quarterly
79 (1998), pp. 59–77.
Sorabji, R., Necessity, Cause and Blame: Perspectives on Aristotle’s Theory (London:
Duckworth, 1980).
Strawson, G., Mental Reality (Cambridge, MA: MIT Press, 1994).
Strawson, P. F., Individuals: An Essay in Descriptive Metaphysics (London:
Methuen, 1959).
Sturgeon, S., ‘Physicalism and Overdetermination’, Mind 107 (1998), pp. 411–32.
Swinburne, R., The Evolution of the Soul (Oxford: Clarendon Press, 1986).
Taylor, R., Action and Purpose (Englewood Cliffs, NJ: Prentice-Hall, 1966).
Thomson, J. J., Acts and Other Events (Ithaca, NY: Cornell University Press, 1977).
Tooley, M., Time, Tense, and Causation (Oxford: Clarendon Press, 1997).
van Inwagen, P., ‘The Doctrine of Arbitrary Undetached Parts’, Pacific Philosophical
Quarterly 62 (1981), pp. 123–37.
An Essay on Free Will (Oxford: Clarendon Press, 1983).
‘Free Will Remains a Mystery’, in R. Kane (ed.), The Oxford Handbook of
Free Will (Oxford: Oxford University Press, 2002).
von Wright, G. H., Explanation and Understanding (London: Routledge and Kegan
Paul, 1971).
Walter, S. and Heckmann, H.-D. (eds), Physicalism and Mental Causation: The
Metaphysics of Mind and Action (Exeter: Imprint Academic, 2003).
Wegner, D. M., The Illusion of Conscious Will (Cambridge, MA: MIT Press, 2002).
218 Bibliography
James, W. 86 panpsychism 76
Papineau, D. 20n, 43–4, 46, 63n, 66n
Kim, J. 48n, 53n, 63n, 92n, 93n, 100n, paralysis 86–7
107n, 108n, 109n, 117n, 177n Peacocke, C. A. B. 42n, 49–50
knowledge (and causality) 80, 87–8, 134–5 persons (see also selves) 5, 15, 19–20, 92–6,
Kornhuber, H. H. 102n 147, 159, 165–71, 180
phantom pains 96
physicalism 11, 12, 13, 40–1, 59, 69, 153
laws (causal or natural) 9, 31, 44, 55–6, Plantinga, A. 184
160–1 Popper, K. 46, 102n
Leibniz, G. W. 60 possible worlds 29, 103, 192
Lewis, D. K. 103n powers (see also dispositions) 4, 92–3,
libertarianism 191–4, 202 99–100, 107–8, 118, 122, 138, 143,
Libet, B. 84, 85, 104, 178n 148–51, 154–7, 159, 164–6, 168–9,
Locke, J. 149n, 167, 173n, 200, 207 173
locked-in syndrome 24, 86 rational 155–7, 176–8
Lockwood, M. 65n practical reasoning 16, 187
logic (of action and of belief) 209–11 probability (see also chance) 209
properties (mental and physical) 93–5
Mackie, J. L. 187, 205n propositions 206, 208
McCall, S. 194n Pythagoreanism 23
McGinn, C. 23n, 78n
means (causal) 124–5, 126 quantum physics 46, 54, 62, 65–7, 118,
Melden, A. I. 83n 177
Mellor, D. H. 93n queerness, argument from 204–5
Melnyk, A. 76–7
mental
causation: see causation radioactive decay 11, 65–6, 150, 155,
representation 179 176–7, 192, 194
states: see states realization 69, 70, 75, 107, 108–9
Menzies, P. 107n, 108n, 110n, 114n reasons
mereology 63, 64, 69 for action 2, 8–10, 85, 116, 129, 130–1,
Merricks, T. 169n 152, 155–6, 174–5, 177, 180–7,
Mills, E. 42n 189–90, 201–9
momentum 42, 60, 61 for belief 203–6
multiple thinkers problem 169–71 moral 187
prudential 187
redundancy (of causes) 72–4
naturalism 161–3, 179, 187, 205, 212 regress, infinite 184
needs 202, 208–9, 211–12 Reid, T. 7n, 151n
neural events 102–12, 132–3, 137–8, 154 replay argument 191–4
Newton, I. 61, 62, 71 responsiveness (to reasons) 176, 194, 195
Noordhof, P. 44 Robinson, W. S. 100n
numbers 22, 23 robots 39