Personal Agency
This page intentionally left blank
Personal Agency The Metaphysics of Mind and Action
E. J. Lowe
...
46 downloads
1120 Views
2MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Personal Agency
This page intentionally left blank
Personal Agency The Metaphysics of Mind and Action
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 reader. Furthermore, this method of proceeding means that the individual chapters, and certainly both Parts of the book, are still perfectly capable of being read independently of one another, which may suit the purposes of readers who are mainly interested only in some of the topics that the book covers.
Acknowledgements Various chapters of this book are based upon the following previously published papers of mine—all of them relatively recent—and I am grateful 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 Consciousness 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 Metaphysics 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: H¨older–Pichler–Tempsky, 2003). Chapter 9: ‘Rational Selves and Freedom 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 ‘NonCartesian 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 u¨ r 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 u¨ r 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 encouragement 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 List of Figures
xvii
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
Part II. Persons, Rational Action, and Free Will
119
6. Event Causation and Agent Causation
121
7. Personal Agency
141
8. Substance Causation, Persons, and Free Will
159
9. Rational Selves and Freedom of Action
179
10. Needs, Facts, Goodness, and Truth
199
Bibliography
213
Index
219
This page intentionally left blank
Contents List of Figures
xvii
Introduction
1
1. 2. 3. 4. 5. 6. 7.
Some questions and answers Event causation and agent causation Free action and causation Reasons and causes An interim summing up The challenge of—and to—physicalism A brief look ahead
Part I. Mental Causation, Causal Closure, and Emergent Dualism 1. Self, Agency, and Mental Causation 1.1. 1.2. 1.3. 1.4. 1.5. 1.6. 1.7. 1.8. 1.9. 1.10. 1.11. 1.12.
An apparently inconsistent triad The self is not its body Mental states are not physical states Selfhood requires agency Are the three claims inconsistent? Naturalistic dualism is possible On coincidental events A comparison between two possible worlds The significance of these findings Intentionality and mental causation An illustrative example An objection and a reply
2. Causal Closure Principles and Emergentism 2.1. Causal closure arguments for physicalism 2.2. The surprising variety of causal closure principles 2.3. Some causal closure principles that we may justifiably
ignore
2 3 6 8 11 12 14 17 19 19 20 22 23 25 26 27 29 33 34 37 39 41 41 43 45
xiv
Contents 2.4. A causal closure principle that is manifestly too weak 2.5. A stronger causal closure principle that avoids the
transitivity problem
46 48
2.6. Weak causal closure and non-coincidence through
mental causation
51
2.7. The consistency of strong causal closure with dualistic
interactionism
53
2.8. What can we reasonably demand of a model of mental
causation? 3. Physical Causal Closure and the Invisibility of Mental Causation
56 58
3.1. The early modern roots of the debate over interactive
dualism
58
3.2. The contemporary debate and the appeal to causal 3.3. 3.4. 3.5. 3.6.
closure Is the causal closure principle (CCP) true? Is the non-overdetermination principle (NOP) true? The invisibility of mental causation Objections, replies, and some philosophical lessons to be learnt
4. Could Volitions Be Epiphenomenal? Concerning epiphenomenalism Volitions and volitionism Automatisms and illusions of control Libet on the unconscious cerebral initiation of voluntary actions 4.5. Can the causal efficacy of the will coherently be doubted? 4.6. How is causal knowledge of the physical world possible? 4.7. The incoherence of full-blown epiphenomenalism 4.1. 4.2. 4.3. 4.4.
5. The Self as an Emergent Substance 5.1. Non-Cartesian substance dualism defined 5.2. The unity argument for NCSD 5.3. The causal closure argument against interactive dualism
62 65 68 74 75 79 79 81 82 84 85 87 89 92 93 95 99
Contents xv 5.4. Two different perspectives on the causal explanation of 5.5. 5.6. 5.7. 5.8. 5.9.
voluntary action A counterfactual-based argument against psychoneural identity theories Extending the argument to ‘realization’ accounts Intentional causation versus physical causation Context-dependency to the rescue? Reasons, causes, and freedom of action
Part II. Persons, Rational Action, and Free Will 6. Event Causation and Agent Causation 6.1. 6.2. 6.3. 6.4. 6.5. 6.6. 6.7. 6.8. 6.9. 6.10.
Agents and agent causation A putative analysis of agent causation Causative action verbs and basic actions The case for irreducible agent causation The problem of ‘free will’ Mental causation, rational choice, and freedom of action Basic actions and backward causation The conceptual priority of agent causation An analysis of event causation Implications for the notion of causality
7. Personal Agency 7.1. 7.2. 7.3. 7.4. 7.5. 7.6. 7.7. 7.8. 7.9.
Event causation and substance causation The primacy of substance causation The reduction of event causation to substance causation Agent causation and basic actions Personal agency and volitionism Causal powers and the nature of the will Classical agent causalism versus volitionism In defence of volitionism The will and its freedom Appendix
101 103 107 110 112 116 119 121 122 123 124 126 128 130 132 133 135 139 141 141 143 145 146 147 149 151 153 154 157
8. Substance Causation, Persons, and Free Will
159
8.1. The problem of the disappearing agent
159
xvi
Contents 8.2. 8.3. 8.4. 8.5.
More on the primacy of substance causation Human persons as psychological substances Volition as the executive element in intentional action The will as a spontaneous power
9. Rational Selves and Freedom of Action 9.1. 9.2. 9.3. 9.4. 9.5. 9.6.
Acting for a reason Objections and replies Choice, causation, and free agency Choice and chance Choice, agency, and control The pragmatic inconsistency of determinism
10. Needs, Facts, Goodness, and Truth 10.1. 10.2. 10.3. 10.4. 10.5. 10.6.
A summary of the story so far What kind of thing is a reason for action? Reasons for action versus reasons for belief Objective needs as reasons for action The logic of action versus the logic of belief Taking needs seriously
Bibliography Index
161 165 171 176 179 180 183 187 190 195 197 199 200 202 205 208 209 211 213 219
List of Figures 1.1. Two possible worlds
30
2.1. Simultaneous causation without overdetermination
48
2.2. Non-coincidence through mental causation
52
3.1. A dualistic counterexample to (NOP)
71
This page intentionally left blank
Introduction Most philosophy books benefit by having an introduction and this one is no exception. Most introductions, however, explain to the reader what the author intends to do in the order in which he or she intends to do it. In this respect, the present introduction is back-to-front, because I shall begin by focusing on what I intend to do in Part II of this book before turning, fairly briefly, to the themes of Part I. The chief reason for this is that I consider the ideas of Part II to be the more innovative, controversial, and potentially significant, not only for the philosophy of mind and action but also, more broadly, for our entire self-conception as rational and moral beings. It was in order to develop these ideas in detail that I decided to write the book. I do, however, regard the ideas and arguments of Part I as providing essential groundwork preparatory for those of Part II. Without that foundation, Part II might well seem unduly speculative, because it presupposes that the current physicalist consensus in the philosophy of mind is unjustified and indeed deeply mistaken. It is that consensus that I attack in Part I, as I shall explain more fully later in this introduction. This attack builds upon earlier work of mine in the philosophy of mind and to that extent will come as no surprise to those familiar with views on the subject that I have developed elsewhere.¹ Indeed, readers who are already sympathetic with those views and feel in need of no further persuasion concerning the general thrust of them could turn immediately after this introduction to Part II, all the chapters of which are perfectly intelligible independently of those in Part I. ¹ See, in particular, my Subjects of Experience (Cambridge: Cambridge University Press, 1996).
2
Introduction
1. Some Questions and Answers The most important questions that I raise and try to answer in Part II of this book are the following: (1) Are all causes events, or are at least some causes agents? (2) Are free actions uncaused, at least by antecedent events? and (3) Are an agent’s reasons for action causes of that agent’s actions? It will be observed that these questions presuppose certain distinctions: between events and agents, between caused and uncaused events, and between reasons for and causes of actions. I shall not seek to define the terms of these distinctions precisely here, but the following remarks may help to clarify what I have in mind in making them. I take events to be particular, datable occurrences, which consist in changes in the properties or relations of persisting, concrete objects.² I take agents to be persisting, concrete objects capable of surviving certain changes and hence capable of participating in certain events—and amongst agents I include individual human beings, in other words, ourselves. I take causation to be a relation (or family of relations) in which one or more items may stand to another, the former being causes of the latter, which is correlatively an effect of those causes—and I take effects to be events. Finally, I take reasons for action to be items which make specific actions rationally commendable, at least prima facie, and hence items whose existence may serve to justify, and in a certain manner explain, actions. The position to be defended in Part II will deliver the following answers to the foregoing questions: (1) In the most fundamental sense of ‘cause’, only agents are causes—although ‘agents’ understood in a very broad sense, to include inanimate objects as well as human beings. (2) Free actions are completely uncaused —but they need not on that account be deemed to be merely random or chance occurrences. (3) A rational agent’s reasons for action are never causes of his or her actions. ² See further my A Survey of Metaphysics (Oxford: Oxford University Press, 2002), ch. 13.
Introduction 3 In addition, I shall make the following claim: (4) All free actions either consist in, or are initiated by, an act of will —in other words, a volition—on the part of the agent. In the next three sections, I shall provide a quick overview of my reasons for favouring these answers, after which I shall say something about the themes of Part I of the book.
2. Event Causation and Agent Causation The distinction between event causation and agent causation is one that many philosophers of action have drawn, but they have not always done so entirely perspicuously. First of all, however, we may usefully distinguish between singular event-causal sentences or statements and singular agentcausal sentences or statements—these being sentences or statements of the grammatical forms, respectively, ‘c caused e’ and ‘A caused e’, where ‘c’ and ‘e’ are singular terms apt for denoting particular events and ‘A’ is a singular term apt for denoting a particular agent. I restrict consideration to past-tense sentences here simply for ease of illustration. An example of a singular event-causal sentence would be ‘The explosion of the bomb caused the collapse of the bridge’, while an example of a singular agent-causal sentence would be ‘The bomb caused the collapse of the bridge’ or ‘The saboteur caused the collapse of the bridge’. Having drawn this grammatical or syntactical distinction, we may next ask the semantic question of whether sentences of the one form are reducible to, or analysable in terms of, sentences of the other form. Typically, event causalists—as I shall call those philosophers who consider that all causation is fundamentally event causation—contend that the singular agent-causal sentence ‘A caused e’ is analysable in terms of an existentially quantified event-causal sentence, roughly along the lines of ‘Some event, c, involving A, caused e’. We may take it here that for A to be—in the relevant sense—involved in an event at least requires A to participate in that event. Thus, for instance, a bomb may be said to be ‘involved’ in its own explosion, inasmuch as the explosion is something that happens to the bomb—and this, I would say, is because the explosion just is a change of a certain kind in certain properties of the bomb. So, for example, the
4
Introduction
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 roughly along the lines of ‘Some agent, A, by acting in a c-type manner, caused e’—where ‘agent’ is understood in a broad sense to include, potentially, any individual substance possessing causal powers. Thus, for example, according to this way of thinking, the singular event-causal sentence ‘The explosion caused the collapse of the bridge’ is analysable as meaning something like ‘Some agent, by exploding, caused the collapse of the bridge’ and is thus entailed by the singular agent-causal sentence ‘The bomb, by exploding, caused the collapse of the bridge’. As for the singular agent-causal sentence ‘The bomb caused the collapse of the bridge’, I take this to be elliptical for ‘The bomb, by acting in some manner, caused the collapse of the bridge’, which is likewise entailed by ‘The bomb, by exploding, caused the collapse of the bridge’. All of this is, again, explained more fully in Chapter 6. Here, however, it may be asked how I propose to deal with my other earlier example of a singular agent-causal sentence, ‘The saboteur caused the collapse of the bridge’, in which the agent referred to is a human being rather than an inanimate object. But, in fact, my answer is just the same as before: I take this to be elliptical for ‘The saboteur, by acting in some manner, caused the collapse of the bridge’. And this, because I maintain it quite generally with regard to such examples, is why I reject the view of many contemporary self-styled agent causalists, as we shall shortly see. It should be clear now why I answer the first question posed at the outset in the way that I do: why, that is to say, I hold that, strictly speaking, only agents are causes. Events, in my view, may be said to be causes at best only in a loose and derivative sense, as a convenient fac¸on de parler. This, for what it is worth, was also George Berkeley’s view: although he notoriously held all agents—all individual substances—to be purely mental substances, or ‘spirits’.⁴ I certainly do not hold that, although I do hold that human persons are, in a perfectly good sense, ‘mental substances’, albeit ones possessing physical as well as psychological characteristics—a view that is very close, in some ways, to that of P. F. Strawson and one that I shall develop in more detail in Chapter 5.⁵ 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
3. Free Action and Causation So-called agent causalists in the philosophy of action typically say that human agents cause their own free actions and that those actions do not have sufficient causes in the form of antecedent events.⁶ They hope thereby to defend a form of libertarianism without falling prey to the charge that free actions, as they conceive of them, are merely random or chance events. The idea is that such actions are not causally determined by preceding events outside the control of the agents whose actions they are, but are nonetheless caused —and so are non-random—because they are caused by the agents themselves, who thereby intervene in the course of nature by initiating new causal chains of events, which would not have existed but for their intervention. This view relies on an acceptance that event causation is a genuine and irreducible form of causation which is manifested throughout the realm of inanimate nature, while contending that, in addition, agent causation exists as a sui generis phenomenon, in which the agent purely as such figures as the cause of certain events. I cannot accept the foregoing position, because I contend, first, that there is strictly speaking no such thing as event causation—only causation by agents—and, second, that whenever an agent causes an event, that agent always does so by acting in some manner —and hence is not a cause simpliciter of the event in question. The classical agent causalist, on the contrary, holds that when an agent causes an event, it is just the agent qua agent that causes the event, not the agent by acting in some manner. In other words, the classical agent causalist holds that in a genuine case of agent causation, a singular causal statement of the form ‘A caused e’ is not only strictly and literally true, but is not in any way elliptical —whereas I hold that such a statement is always elliptical either for something of the form ‘A, by acting in such-and-such a manner, caused e’, or for something of the form ‘For some manner of acting ϕ, A, by ϕing, caused e’. Classical agent causalism has, of course, been subjected to criticism on the point in question, notably, in the form of the following objection: what explains why the caused event occurred when it did, if its cause was just the agent as such?—for the agent as such is not a dated item and yet is, qua agent, ⁶ See further my article ‘Agent Causation’, in The Encyclopedia of Philosophy, 2nd edn, ed. D. M. Borchert (New York: Macmillan, 2006).
Introduction 7 supposedly a cause of many different events occurring at different times.⁷ This question does indeed seem difficult, even if not impossible, to answer on the classical account. My own position with regard to voluntary human action is basically this: that when a human agent, A, voluntarily causes some event, e, A causes e by willing to cause an e-type event.⁸ A’s so willing—A’s volition—is itself an event, but not merely an event: it is an action of A’s—indeed, it is a primitive action of A’s, because it is not further analysable in terms of more basic actions of A’s and the consequences of such actions. Willing, in my view, is the most basic kind of action that an intelligent, free agent can perform—it is a species of mental event and, more specifically, an executive act of mind with a determinate intentional content. A volition is always a volition to do such-and-such—typically, to bring about, or cause, an event of such-and-such a type. I further hold that volitions—which are our free actions, in the most basic sense—are completely uncaused, or entirely spontaneous. They are not, in my view, caused by prior events, any more than the spontaneous decay of a radium atom is. Nor, however, are they caused by the agents whose volitions they are.⁹ They could not be, given that they are our most basic actions and given, too, that an agent always causes an event by acting in some manner: for if they were so caused, then the actions by performing which agents caused their volitions would instead be our most basic actions. Rather than cause his or her own volitions, an agent simply performs, or executes, or enacts 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
them—and this kind of relationship between agent and volition is not a causal one, because it is an ‘internal’ one.¹⁰ In other words, ‘A performed volition v’ is really just a less perspicuous way of saying ‘Volition v was a willing of A’s’, because ‘volition v’ is an internal accusative of the verb ‘performed’ in ‘A performed volition v’, exactly as ‘sensation s’ is an internal accusative of the verb ‘felt’ in ‘A felt sensation s’. By analogy, thus, ‘A felt sensation s’ is really just a less perspicuous way of saying ‘Sensation s was a feeling of A’s’. An example may help to clarify the point, which is apt to be misunderstood. To say ‘A felt a pain in his toe’ is pleonastic, in a way that to say ‘A felt a rock on his toe’ is not, because a pain just is a kind of feeling and it is tautologous to speak of feeling a feeling. Analogously, then, to speak of ‘performing a volition’ amounts to speaking of doing a doing, which is similarly tautologous. This is why it is less misleading simply to say something like ‘A willed to ϕ’, rather than ‘A performed a volition to ϕ’. Now, as I have just indicated, volitions, as I conceive of them, are the exercises of a spontaneous, free power—the will —which is a two-way power to choose to act or refrain from acting in any one of what is normally a wide range of ways on any given occasion. Non-basic free actions consist in the agent’s causing some further event, e, by willing to do so. In such a case, the event e is, of course, caused —but the agent’s action is not e itself, but his or her causing e by a free act of will. So in this case, too, the action itself is completely uncaused. As to the question of why free actions, being uncaused, are not purely random or chance occurrences, my answer, which I shall develop a little further in a moment and much more fully in Chapter 9, is that they are not characterizable as such because they are performed for a reason—and the account of reasons for action that I am just about to sketch explains why that makes them relevantly different from random or chance occurrences.
4. Reasons and Causes Some causal theorists of action—notably Donald Davidson—hold that an agent’s reasons for acting are beliefs and desires of the agent which cause (or, 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 fac¸on 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 fac¸on 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
rational actions—namely, our beliefs and desires—are precisely not chosen by us, since we cannot adopt beliefs or desires at will. In my opinion, an agent’s reasons for action are not, typically, mental states of the agent at all, such as beliefs and desires, but states of affairs in the world, which are disclosed to the agent through perception or reflection and of which the agent may thus become aware. (I shall complicate this account somewhat in Chapter 10, in order to accommodate properly the distinction between practical and theoretical reason, but for present purposes the foregoing characterization of my position is not unduly misleading.) For instance, my reason for taking a longer step than normal as I walk along a path might be that there is a puddle in my way. It is not my belief that there is a puddle in my way, together with my desire to avoid it, that is my reason for taking a longer step. Indeed, having a belief that there is a puddle in one’s way would not constitute a reason for taking a longer step, if there were in fact no puddle in one’s way. It is the actual presence of the puddle, not the fact that one believes that it is there, that justifies—and so constitutes a reason for—the action of taking a longer step. Having a belief that there was a puddle in the absence of any puddle would not be a reason for taking a longer step, but at most a reason to have one’s eyesight checked. This is only a very rough-and-ready sketch of what I want to say about reasons for action and I shall say much more about them in Chapters 9 and 10, as well as elsewhere in Part II of this book. However, I should immediately acknowledge here an intellectual debt to Jonathan Dancy, whose work on reasons for action has persuaded me of the correctness of the general approach to this issue that I now adopt (although, of course, I can’t be sure that he would approve of my particular version of that approach).¹³ When an agent acts freely and rationally, he or she does so by exercising his or her will—by willing —in a manner that is responsive to the reasons for acting of which he or she is then aware. Acting freely and rationally is exercising the will in the light of the reasons for action that are disclosed to us through perception (and through any other means whereby reasons are disclosed to us—for example, through the exercise of our powers of deduction—which we may lump together under the general term ‘reflection’). Those reasons are, as remarked above, states of affairs in the 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 fac¸on 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.
6. The Challenge of—and to—Physicalism It will be immediately apparent that the views about rational human action defended in Part II of this book are very difficult, if not impossible, to square with the physicalist conception of the mind that has become, in effect, the default position of contemporary analytic philosophers. According to that conception, states of mind are just a special subset of bodily states and a human subject of such states is identifiable either with his or her physical body as a whole or with some distinguished proper part of it, such as the brain, or the central nervous system. A key argument—though, more accurately, it is a family of arguments—taken to support this conception of the mind is the notorious causal closure argument, which purports to show that the only way to reconcile the supposedly incontestable causal efficacy of mental states with the presumed fact that the physical world is causally closed is to identify those states with bodily states. For the purposes of this argument, any distinction that one might wish to draw between states and events is not really relevant, so that one may say, without undue distortion, that adherents of the argument assume—usually without examination—that all causation is fundamentally event causation. However, they do differ amongst themselves with regard to what exactly is or should be meant by saying that the physical world is ‘causally closed’, as we shall see most fully in Chapter 2: for many subtly different ‘causal closure
Introduction 13 principles’ have been invoked by different physicalist philosophers of mind. They also differ regarding the epistemological status of such principles, or the sorts of considerations that might or should be taken to support them. There is a general presumption, however, that current physical science (or perhaps only a future, ‘completed’ physics) somehow lends strong credence to some such principle and thereby to physicalism in the philosophy of mind. All of this I seek to challenge in Part I of this book. That is to say, I seek to undermine the challenge that physicalism presents to my theory of rational free action by challenging its own credentials and, in particular, the cogency of the causal closure argument in all of its versions. If my challenge to physicalism is successful, it shows at the very least that a dualistic philosophy of mind—one which holds that human mental states and their subjects cannot simply be identified with bodily states and human bodies respectively—is both metaphysically coherent and entirely consistent with known empirical facts concerning mental causation and causation in the physical domain. I should emphasize that it is a crucial part of this project to show that human subjects—we ourselves—are not at all plausibly identifiable with our bodies or parts of our bodies, as I argue especially in Chapter 5. That is to say—to use traditional terminology—the version of psychophysical dualism that I defend is a version of interactionist substance dualism. It is not, however, Cartesian dualism, as that doctrine is standardly understood, which is why I call my own version non-Cartesian substance dualism. As I indicated earlier, the crucial difference is that I am not at all committed to the doctrine that subjects of mental states—subjects of experience, as I like to call them—possess only mental properties and hence no physical ones. However, as I have already remarked, the overall position defended in Part I could be accepted by a philosopher who was not convinced by my account of rational free action in Part II even if—as I am strongly inclined to believe—an acceptance of the latter demands an acceptance also of something at least very close to the former. Of course, I consider that the most credible conception that we can have of ourselves as rational free agents is one which combines the general theses of Part I—in effect, some form of non-Cartesian substance dualism—with the specific claims of Part II regarding the nature of causation, freedom, and rationality. But it seems to me helpful to separate the overall package of views that I am
14
Introduction
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.
7. A Brief Look Ahead Part I of this book begins with Chapter 1, ‘Self, Agency, and Mental Causation’, in which I take my first look at the causal closure argument for physicalism in the philosophy of mind. I show how, consistently with accepting a fairly strong principle of physical causal closure, a dualistic theory of mental causation can be made plausible by emphasizing the explanatory role that can be accorded to mental states in accounting for what would, in their absence, appear to be mysterious coincidences in the ways in which apparently unconnected physiological events give rise to coordinated bodily movements. I argue that it is crucial to their occupying this kind of explanatory role that mental states are intentional states and that mental causation is distinctively intentional causation—the bringing about of intended effects. In Chapter 2, ‘Causal Closure Principles and Emergentism’, I look at a wide range of causal closure principles of differing strengths and argue that the stronger principles are less plausible than the weaker while the weaker are less capable of playing an adequate role in any causal closure argument for physicalism. The upshot is that there is plenty of scope for a psychophysical dualist to accept a moderately strong principle of physical causal closure while also advocating a genuine and autonomous causal role for mental states in the genesis of intended physical behaviour. Such a dualist may consistently espouse a form of emergentism in the philosophy of mind, according to which non-physical mental states are causally autonomous and yet are themselves ultimately the products of prior physical evolution. In Chapter 3, ‘Physical Causal Closure and the Invisibility of Mental Causation’, I develop in more detail models of mental causation of the kinds adumbrated in the first two chapters and explain why, if such models
Introduction 15 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
countenanced at all, is best understood as elliptical for talk in terms of agent causation, although only where the word ‘agent’ is very broadly construed, rather than being confined in its application to the much narrower class of rational agents, ourselves included. In Chapter 7, ‘Personal Agency’, I build upon the account of agent causation developed in the previous chapter to draw an initial sketch of my current view of rational human action, which falls midway between volitionism and classical agent causalism, while simultaneously endorsing libertarianism with regard to the question of free will and externalism concerning the nature of reasons for action. In Chapter 8, ‘Substance Causation, Persons, and Free Will’, I explain more fully why, in my opinion, all causation is fundamentally substance causation—that is, causation by ‘individual substances’, or persisting, concrete objects—rather than causation by events. I then explain how my versions of substance dualism in the philosophy of mind and libertarianism in the philosophy of action can be very naturally accommodated within a view of the physical world which takes seriously the fundamental status of substance causation. In Chapter 9, ‘Rational Selves and Freedom of Action’, I develop in more detail my externalist account of reasons for action in conjunction with my libertarian approach to the question of free will, in ways which I have already indicated in earlier sections of this introduction. Finally, in Chapter 10, ‘Needs, Facts, Goodness, and Truth’, I examine more closely than in previous chapters the important distinction between practical and theoretical reason and rationality. As a consequence, I modify the version of externalism regarding reasons for action advocated elsewhere in the book, contending that the sorts of items that properly qualify as reasons for actions, externalistically conceived, must differ fundamentally from those that properly qualify as reasons for belief. I also argue that this ontological difference between reasons for action and reasons for belief goes hand-in-hand with equally fundamental differences between the logic of action and the logic of belief.
PA RT I
Mental Causation, Causal Closure, and Emergent Dualism
This page intentionally left blank
1 Self, Agency, and Mental Causation 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.
1.1. An Apparently Inconsistent Triad All of the following claims may seem fairly compelling, upon reflection. And yet they may appear—for reasons that I shall set out in Section 1.5—not to form a consistent set: (1) The self, although physically embodied, is not to be identified with any physical body nor with any part of such a body. (2) The self is by its very nature an agent, something that is naturally capable of performing intentional actions, some of them with physical results.
20
Mental Causation, Closure, and Dualism (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.
1.2. The Self is not its Body I believe, first of all, that selves exist, not least because I believe that I exist and consider myself to be a ‘self’. I use the term ‘self’ interchangeably with the term ‘person’. I take it, however, that the term ‘self’ is a particularly appropriate synonym for ‘person’ because it reflects the fact that a necessary condition of personhood is a capacity for self-reference—a capacity which is manifested linguistically by use of the first-person pronoun, ‘I’. A person or self, in short, is a being that can have thoughts about itself, of the sort that are appropriately expressed (in English) by sentences containing the 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 first-person pronoun, ‘I’, as their grammatical subject—sentences such as ‘I feel hot’ and ‘I am six feet tall’. But I also believe that a person or self, even though physically embodied, is never to be identified with its physical body nor with any part of it, such as that body’s brain. This is claim (1) above. Our ordinary self-conception seems to involve a commitment to claim (1). For example, when I have a conscious first-person thought—such as the thought that I feel hot—I regard myself as being the subject of this thought, both in the sense of being the thing having the thought and in the sense of being the thing that the thought is about. But I am not at all inclined to regard my body or my brain as being its subject, in either sense. Since I am the subject of the thought but neither my body nor any part of it is, it follows that I am not identical with my body or any part of it. Of course, with the benefit of a little scientific knowledge, I may well be prepared to concede that, but for the existence and normal functioning of my brain, I could not so much as have this or any other thought: but that doesn’t (or shouldn’t) persuade me to believe that my brain is, after all, the subject of my thoughts. That would be like inferring that my feet run from the fact that I could not run without having feet. Anyway, quite apart from anything else, it seems clear that, even granted that I need a brain in order to be able to think, I don’t need to have the particular brain that I do have. I find nothing inconceivable in the thought that I might wake up one morning to be told (truly) that, overnight, I had undergone an operation in which my old organic brain was somehow replaced by a new inorganic one. Here it may be objected that, if I am not to be identified with my physical body nor with any part of it, then it only remains for me to be identified with something altogether non-physical, such as a spirit or soul or ‘Cartesian ego’—and this, it will be said, is a view wholly at odds with a naturalistically acceptable conception of persons. However, it is a simple mistake to suppose that if I am not to be identified with my physical body or any part of it, I must therefore be identified with something non-physical, that is, with something possessing no physical characteristics whatever. And, indeed, identifying oneself with something non-physical is quite as counter-intuitive as identifying oneself with one’s physical body or brain. It seems to me no less literally true that I have a certain height than it seems literally false that my brain has certain thoughts.
22
Mental Causation, Closure, and Dualism
The self can be a ‘physical’ thing—possess physical characteristics such as height—even though it has different identity conditions from those of the body or brain. Somewhat analogously, a statue can be a physical thing—possess physical characteristics such as shape—even though it has different identity conditions from those of the piece of matter which composes it. The analogy isn’t perfect, however, for I don’t want to say that the relation of embodiment is simply one of composition: I am not composed by my body, in the way that the statue is composed by bits of matter. Indeed, I don’t believe that the self is a composite entity at all: I don’t believe that it is literally made up of distinct and separable parts. The self, I want to say, possesses a strong kind of unity which is incompatible with its being a composite thing. I don’t want to digress from my main theme in order to argue for this view just here, but will return to the issue in Chapter 5.³ All I want to stress at present is that claim (1) above is not only plausible, but is perfectly consistent with the equally plausible claim that the self is a physical thing, in the sense of being a thing that possesses physical characteristics or states.
1.3. Mental States are not Physical States However, it doesn’t follow from the claim that the self is a physical thing that mental states of the self can intelligibly be thought of as being physical states of it, akin to such physical states as height and weight. Indeed, I very much want to deny that mental states are physical states, even though they are states of something physical—the self. This is because I can make no clearer sense of the idea that a conscious mental state might just be a physical state than I can of the idea that a physical object might just be a natural number.⁴ Consider a typical mental state, such as this one: consciously thinking of Paris. I know what it means to be in such a state, at least as clearly as I know what it means to be in the physical state of sitting in a chair. But I cannot at all understand what it would mean to say that the state of consciously thinking of Paris just is a ‘physical’ state. This is because—as I understand it—a physical state is, by its very nature, one whose possession ³ See also my Subjects of Experience, ch. 2. ⁴ Compare P. T. Geach, Truth, Love and Immortality: An Introduction to McTaggart’s Philosophy (London: Hutchinson, 1979), p. 134.
Self, Agency, and Mental Causation 23 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 unintelligible 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 dedicated 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 intelligible 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 physical states—all violate this principle by trying to identify items of quite different kinds.
1.4. Selfhood Requires Agency A word or two is now needed in defence of claim (2)—that the self is by its very nature an agent, something that is naturally capable of performing intentional actions, some of them with physical results. Since I have already characterized the self as something necessarily capable of self-reference, I have already implicitly characterized it as something necessarily possessing 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
agency, since self-reference is a species of intentional action. To refer to oneself as ‘I’, whether in speech or merely in thought, is to perform a kind of intentional act. If done merely in thought, this act may perhaps have no physical results, although if done in speech it clearly must. However, the idea that there might be a self which, throughout its life, was capable only of engaging in intentional actions of a purely mental kind—never, thus, in actions having physical results—is one that is hard to credit.⁷ Such a self would be constitutionally incapable of communicating with other selves. It is strongly arguable, however, that the development of self-awareness is necessarily linked to the development of other-awareness and that both are necessarily linked to the development of powers of communication, whether through language or merely through various kinds of non-verbal behaviour. If that is so, then there couldn’t be a self that was constitutionally incapable of communicating with other selves throughout its life—though there might, conceivably, be a self which lost this capability having once developed it, as is suggested by cases of so-called ‘locked-in syndrome’ (people who seem to remain self-aware even though they have lost all control over their bodies through complete paralysis of the non-autonomic nervous system). Another reason for thinking that a self must be capable—at least at some stage during its existence—of performing intentional actions which have physical results is that it is strongly arguable that only a being capable of such actions can develop a concept of causation and that possessing such a concept is a necessary condition of self-reference and thus of selfhood itself. (It is a necessary condition of self-reference because to self-refer is to perform an intentional action; to perform an intentional action is to act in a certain way knowing that one is so acting; the concept of intentional action is a causal concept; and knowledge is possible only for one who possesses the requisite concepts.) The thought here, then, is that a being that was condemned from birth to complete physical passivity, even though endowed with powers of sensation and perception, would be incapable of distinguishing between causal and non-causal sequences of events, because an ability to make this distinction depends upon an ability to intervene actively in the course of nature, with a view to discovering by means of experimental manipulation 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 evidence 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.
1.5. Are the Three Claims Inconsistent? Why should the three claims stated at the beginning of this chapter be thought to be inconsistent? For the following reasons, I imagine. First of all, claim (2) seems to imply—indeed, I agree that it does imply—that intentional states of the self can be causes of physical events. This is because the concept of an intentional action is a causal one: when an agent acts intentionally, an intentional state of that agent plays a causal role in the production of some event—an event which, in the case of an intentional action which has a physical result, will obviously be a physical one. Next, claim (1) seems to imply that intentional states of the self are states of something non-physical and are therefore themselves non-physical states. Now, of course, we have just seen that claim (1) does not, in fact, imply that the self is something non-physical. But we have also seen that there is, all the same, good reason to think that even though the self is physical, inasmuch as it possesses physical states, mental states of the self—including its intentional 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
states—are not physical states of it and so are indeed non-physical states. So, although claim (1) does not strictly have the implication it might seem to have—that intentional states of the self are non-physical states—I think that any adherent of claim (1) ought nonetheless to accept the thesis that intentional states of the self are non-physical states. Finally, claim (3) seems to imply that no physical event can have a non-physical state amongst its causes. (We shall examine this alleged implication in a moment.) Together, then, claims (1), (2), and (3)—or, more accurately, claims (2) and (3) together with the thesis, consistent with claim (1), that intentional states of the self are non-physical states—seem to imply that non-physical states both are and are not causes of physical events: a contradiction. However, even if we grant the alleged implications of claims (1) and (2), this reasoning is incorrect, because it ignores the transitivity of causation, as we shall now see. (The key point to appreciate here is the very simple one that if x is causally sufficient for y and y is causally sufficient for z, then, by transitivity, x is causally sufficient for z, but that this doesn’t imply that z is causally overdetermined by both x and y.)
1.6. Naturalistic Dualism is Possible It is possible for claim (3) to be true—that every physical event has a set of wholly physical causes which are collectively causally sufficient for the occurrence of that event—and yet for it also to be true that some physical event, P, has a non-physical event or state, M, amongst its causes (without envisaging this as involving the causal overdetermination of P). This is because M itself may have a set of wholly physical causes which are collectively causally sufficient for its occurrence. If M is a cause of P, then, by the transitivity of causation, all of those physical causes of M are also causes of P —and, clearly, they may form a subset of a set of wholly physical causes which are collectively causally sufficient for the occurrence of P. Hence, claims (1) and (2) are not inconsistent with claim (3), but only with something much stronger, such as (4) No physical event has a non-physical cause. Obviously, however, no one can pretend that claim (4) is strongly confirmed by empirical evidence, however much it may be an article
Self, Agency, and Mental Causation 27 of faith with some philosophers. Even claim (3), although significantly weaker than claim (4), is not exactly strongly confirmed empirically. A presumption in its favour, however, is that modern science encourages us to believe that the universe is a causally closed system whose origins were wholly physical. At the time of the ‘Big Bang’, we suppose, all events were wholly physical—and all subsequent physical events have been and will continue to be long-term effects of those initial events. (Here I am assuming a thoroughgoing causal determinism, but not much is affected by assuming instead that a good deal of causation is irreducibly probabilistic.) But this presumption in no way rules out the possibility that, at some stage during the evolution of the universe, non-physical events or states have come into existence, along with subjects of those events or states (that is, selves, conceived of in accordance with claim (1)). There is no reason to disparage this idea as ‘spooky’, since it need involve no element of supernaturalism—taking ‘supernaturalism’ to be the view that some events are brought about by agents (such as a divine being) which do not exist within the space–time universe. (Such an agent would, of course, be a non-physical thing, quite unlike human selves as I conceive of them.) Even if it is conceded that this is a genuine possibility and that claims (1), (2), and (3) are not logically inconsistent, it may nonetheless be thought that the suggestion that this is how things actually are is an extravagant one which somehow violates canons of parsimony or simplicity in matters metaphysical. On the contrary, I shall now attempt to show how the invocation of mental states, conceived of as non-physical causes of physical events, has the potential to strengthen our causal explanations of certain physical events. This is because such non-physical causes can be represented as rendering non-coincidental certain physical events which, from the perspective of purely physical causation, may appear to occur merely by coincidence.
1.7. On Coincidental Events An event occurs by coincidence, or coincidentally, in the sense that I now have in mind, when two or more events co-occur and jointly cause that event, but those causes are themselves causally independent, in the sense of having no common cause amongst their various causes. (Some philosophers describe
28
Mental Causation, Closure, and Dualism
the co-occurrence of two or more events which have no common cause as being a ‘coincidence’, and I have no quarrel with this usage: but my concern now is with the notion of a single event which occurs by coincidence, in the sense just explained. I am not concerned, then, with the question, which exercises some of those philosophers, of whether ‘coincidences’, in their sense, have causal explanations.⁹) Here is a simple example. A man walks past a house just as a gust of wind dislodges a slate from the roof, causing it to fall, with the result that he is hit by it and killed. The man’s walking there and the slate’s falling there co-occur and jointly cause his death, but, we assume, there was no common cause of the man’s walking there and the slate’s falling there. Consequently, his death occurred by coincidence. But if, say, the man’s approaching the house had set off a tripwire attached to the slate, causing it to fall just as he passed underneath it, then his walking there and its falling there would have had a common cause and so his death would not have been coincidental. An event which occurs by coincidence is not an uncaused event: it has causes, which themselves have causes, which likewise have causes, and so on—what makes it coincidental is the fact that its immediate causes have independent causal histories. An event which does not occur by coincidence is one whose immediate causes share a common cause, rendering the causal histories of its immediate causes non-independent. At least, this will do, to a first approximation, as an account of the distinction between an event which occurs by coincidence and one which does not. (We might need to refine this account in order to avoid having to describe as ‘non-coincidental’ certain events whose immediate causes do share some common cause, but only a relatively insignificant one lying in the remote past of their respective causal histories. After all, we have already conceded that all current physical events are ultimately effects of events which occurred at the time of the ‘Big Bang’, but we don’t want this to count as a reason for denying that certain current physical events are ‘coincidental’.) In our foregoing illustration of the distinction between coincidental and non-coincidental events, it is clear that the two different cases in which the man is killed by the falling slate differ not only in respect of some of the physical events which occur and are causally responsible for the man’s death 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.
1.8. A Comparison between Two Possible Worlds Suppose that two independent causal chains of physical events, P11 , P12 , P13 and P21 , P22 , P23 jointly give rise to a physical event P as the immediate effect of P13 and P23 . Here the occurrence of P is coincidental. But, I submit, it is metaphysically possible for P to have (in a sense explained below) exactly the same physical causal history and yet not to occur by coincidence, because it is metaphysically possible for the immediate physical causes of P —P13 and P23 —to share a common cause which links them by a non-physical causal chain, thereby rendering their causal histories non-independent. It might be the case, for instance, that in this alternative scenario P11 is a cause of a mental event M which is in turn a cause of P22 : see Figure 1.1. In the figure, each node, marked by a letter, represents a particular event and an arrow drawn between two nodes—whether or not it passes through other nodes—signifies that the event represented by the upper of those two nodes is a cause of the event represented by the lower of those two nodes. I should perhaps emphasize that to say that one event is a cause of another event is by no means to rule out the possibility that a third event, also, is a cause of that second event: that is to say, in the sense of ‘cause’ now in play, an event may have many different causes, without thereby being
30
Mental Causation, Closure, and Dualism P11
P21
P11
P21 M
P12
P22
P13
P23
P World 1
P12
P22
P13
P23
P World 2
Figure 1.1. Two possible worlds
causally overdetermined. I am taking it that to say that one event is a cause of another event is—barring the possibility of causal overdetermination—at least to imply that if that first event had not occurred, then that second event would not have occurred either. World 1 and world 2 are the same in the following respects: (i) the same physical events occur in both (in the space–time region with which we are concerned) and (ii) those events bear the same relations of purely physical causation to one another. By (ii) I mean that wherever two physical events in one of the worlds are linked by a certain chain of purely physical causation—causation not involving any non-physical event—they are linked in the same way in the other world. That is to say, wherever, in one of the worlds, a certain physical event is a cause of another physical event, either directly or via certain other intervening physical events, those events stand in that same relation in the other world as well: in the other world, too, the first physical event is a cause of the second physical event, again either directly or via the same intervening physical events.¹⁰ Of 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 course—assuming that causation is law-governed—the two worlds are not the same in respect of the causal laws which obtain in them, because in world 2 certain psychophysical laws obtain which do not obtain in world 1. I shall return to this point in a moment. But what is of special interest to us now is that in world 1 the occurrence of P is coincidental whereas in world 2 it is not coincidental. So we see that two worlds could contain the same physical events standing in the same relations of purely physical causation to one another and yet it be the case that in one of the worlds a certain physical event was coincidental whereas in the other it was not, because in the second world a certain non-physical (mental) event rendered the causal histories of that physical event’s immediate physical causes non-independent. Incidentally, the existence of this possibility does much, in my view, to undermine the popular—although rather obscure—thesis that mental events ‘supervene’ upon physical events: if the thesis is taken to be that worlds which are the same in respect of what physical events occur in them and what relations of purely physical causation obtain between those events are worlds which are the same in respect of what mental events occur in them and what causal relations those events stand in, then we see, first, that this could at best be true of a restricted range of worlds and, second, that there is no positive reason to suppose that our world is one of those worlds. Of course, not all true counterfactual conditionals concerning physical events in one of our two depicted worlds could also be truths in the other of those worlds. For instance, in world 2 of our diagram—where P11 is depicted as being a cause of P22 , albeit only via the mental event M and not, thus, via any chain of purely physical causation—it is true that if P11 had not occurred, then P22 would not have occurred (barring causal overdetermination, which is not at issue here), whereas in world 1 it is not true that if P11 had not occurred, then P22 would not have occurred. But that just reflects the fact that different causal laws are operative in our two worlds and that different events (though not different physical events) 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
occur in them. I should perhaps stress here, if it isn’t sufficiently obvious already, that I am by no means suggesting that the situations depicted in the two worlds of our diagram are compossible: thus, the actual world could not simultaneously be both as depicted in world 1 and as depicted in world 2. However, it is implicit in what I have said that if one were to know, concerning the physical events depicted in our diagram, only which events they were and what relations of purely physical causation they bore to one another, one would not be in a position to decide on that basis whether the actual world was world 1 or world 2, for the simple reason that world 1 and world 2 do not differ in these respects. The significance of this fact will emerge in a moment. Another important point to observe concerning our worlds 1 and 2 is that both of them—not just world 1—can be worlds in which claim (3) is true, that is, in which every physical event has a set of wholly physical causes which are collectively causally sufficient for the occurrence of that event. (Note here that the diagrams representing worlds 1 and 2 are not meant to be complete representations of those worlds, so that, for instance, it is not implied that P11 and P21 lack causes in those worlds.) For, as was pointed out earlier, it is perfectly possible that the non-physical event M of world 2 should itself have a set of wholly physical causes which are collectively causally sufficient for its occurrence. This can be so even if, as is plausible, M also has some other non-physical events amongst its causes (provided that each of these likewise has a set of wholly physical causes which are collectively causally sufficient for its occurrence). Incidentally, it is important to distinguish both world 1 and world 2 from yet another possible world—call it world 3—which is just like world 1 save that in it there is an additional relation of purely physical causation between event P11 and event P22 . In world 3, as in world 2, event P is not coincidental. But, of course, worlds 1 and 3 are not the same in both of the respects (i) and (ii) stated above—unlike worlds 1 and 2—since worlds 1 and 3 differ from each other in respect (ii): the physical events in them do not bear exactly the same relations of purely physical causation to one another. In consequence, P does not have exactly the same physical causal history in worlds 1 and 3, whereas it does in worlds 1 and 2. Naturally, I have no wish to deny that non-coincidental events may sometimes have a purely physical causal explanation, as in world 3: I am only concerned to show that, and how, they may sometimes have a causal explanation which
Self, Agency, and Mental Causation 33 is at least partly non-physical and which is nonetheless consistent with the truth of claim (3).
1.9. The Significance of these Findings What is the significance of these findings? Just this: they show that even if one has identified all the physical causes of a certain physical event within a certain space–time region—for instance, even if one has identified all the neural events causally responsible for a certain bodily movement—this doesn’t preclude the possibility that the existence of a non-physical (mental) event or state, such as a person’s belief or desire or intention, might serve to explain why that movement is non-coincidental, in a way in which the purely physical causal history of that movement does not. One might, thus, have truly discovered all the physical causes of the movement and this discovery might be consistent with the possibility that these were indeed all the causes of the movement—in which case the movement would be coincidental—and yet the discovery would also be consistent with the possibility that there were, in addition to these physical causes, certain nonphysical causes which served to render the movement non-coincidental. So, merely to have satisfied ourselves that we have discovered all the physical causes of such a movement and to have satisfied ourselves that such causes could provide a complete causal explanation for the occurrence of the movement, is not yet to have ruled out the possibility that they don’t in fact provide a complete causal explanation—because the question may still be left open as to whether or not the movement occurred by coincidence. Positing certain non-physical (mental) causes of a physical event, in addition to the physical causes which have already been discovered, may serve an explanatory purpose which cannot be served by appeal to the physical causes alone. The non-physical part of the explanation need not deny anything which has been discovered about the identity of the physical causes and their purely physical causal relations and is in this sense perfectly compatible with the purely physical part of the explanation—although, as has been noted, accepting the non-physical part of the explanation as well as the physical part, rather than just accepting the physical part alone, will require adopting a different view as to what causal laws and what counterfactual conditional truths obtain.
34
Mental Causation, Closure, and Dualism
Of course, if it should turn out, in a particular case, that a physical event can be discovered which renders a bodily movement (say) non-coincidental, then appeal to a non-physical event will be otiose in that particular case: the case will be like that of the non-coincidental death caused by the falling slate, where the triggering of a tripwire rendered the death noncoincidental. But what if it were to be discovered that this can be done in all cases in which we now see reason to invoke items such as beliefs in our causal explanations? That, I suppose, might be taken by some to be a reason for holding that beliefs and so forth just are physical items (neural states or events, perhaps). However, my own view, on the contrary, is that it would instead be (at best) a reason for holding either that beliefs and so forth do not really exist at all (eliminativism) or else that they are causally inefficacious (epiphenomenalism). I take this view of the matter because, as I explained earlier, I do not consider that the thesis that mental states ‘just are’ physical states is even an intelligible one (although, to be fair, I don’t regard either eliminativism or epiphenomenalism as being in much better shape, conceptually, so that I am pretty much committed to denying that we could, even in principle, make the discovery that has just been contemplated). What I am chiefly concerned to point out for the moment, however, is that it is perfectly conceivable that we should discover, in the case of some bodily movement, that all of the physical (say, neural) events which we can implicate in its occurrence are such that, as far as their purely physical causal relations to one another are concerned, they do nothing to show that that movement is anything other than coincidental. And in such a case, my claim is, we could quite consistently and plausibly invoke a non-physical (mental) event as rendering that movement non-coincidental, without denying anything that had hitherto been claimed about the identities of that movement’s physical causes and their purely physical causal relationships to one another and to the movement in question.
1.10. Intentionality and Mental Causation So far, I have said nothing about how mental events and states might cause physical events and states. For that matter, of course, neither have I said how physical events and states might cause physical events and
Self, Agency, and Mental Causation 35 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 intentional 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 ¹¹ See, further, my Subjects of Experience, ch. 5.
36
Mental Causation, Closure, and Dualism
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 noncoincidental, 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.
1.11. An Illustrative Example An assumption behind the following example will be that ‘fusions’ of events are, in general, themselves events: for instance, that if five battles occur over a certain period of time, then there is an event which occurs over that period of time and contains those five battles as parts (we might call this event a ‘campaign’, perhaps).¹³ Most of the macroscopic events we normally talk about are event-fusions in this way. For example, even so ‘simple’ an event as the rising of a person’s arm consists of many sub-events, such as the flexing of certain muscles and the movements of various parts of the arm. Suppose that, over a period of several minutes, the following series of purely physical events is observed to occur: one after another, all of the coloured balls remaining on a snooker table are struck by the cue ball and as a result fall into pockets. The fusion of all these events of a snooker ball falling into a pocket is itself an event, call it event E. Suppose, now, we ask why event E occurred. Clearly, in one sense, E occurred because each of the sub-events of which E is the fusion occurred—but this is not a causal explanation of E. However, each of the sub-events—each event of a snooker ball falling into a pocket—has a causal explanation and one might suppose that the causal explanation of E is simply the conjunction of all those causal explanations (though we shall see in a moment that there could be good reason to challenge this supposition). Moreover, one might suppose that each event of a snooker ball falling into a pocket has a wholly physical causal explanation, adverting solely to prior physical events, such as movements of the snooker cue, movements of the player’s hand (we are assuming here that there is just one player involved), neuronal events in the player’s efferent nerves and motor cortex, and so forth. However, I suggest, if that is all there is to the explanation of event E, then we shall have to regard event E as having happened merely ‘by coincidence’. 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
Moreover, this explanation will not serve to explain, in any interesting sense, why an event of this kind occurred: we shall only be able to say that an event of this kind occurred ‘because’ this particular event occurred and was an event of this kind (and such a ‘because’ is not causal in force). Now, of course, most of us would be extremely surprised if this were all there were to the explanation of event E. Most of us would surmise that E occurred because the snooker player had formed and acted upon a desire to pot all the coloured balls remaining on the table (and possessed the skill needed to achieve this). That desire would not be a desire specifically for event E to occur, but only a desire for the occurrence of an event of a certain kind, namely, an event consisting in the potting of all the remaining balls. We would surmise that, even if some of the balls had moved somewhat differently from the ways in which they actually moved—or, indeed, even if there had been more or fewer balls remaining on the table—the player would have adjusted his action so as to ensure the same kind of result, even though event E itself would not have occurred in those circumstances. Citing the player’s desire as an explanation of event E, then, explains not merely why E occurred, but, more interestingly, why an event of that kind occurred. And, it seems, no purely physical explanation of all of the sub-events of which E is the fusion can provide an interesting explanation of this sort. Such a purely physical explanation makes E appear to be a merely coincidental event and a ‘fluke’, in the sense that it provides us with no rational expectation that an event of this kind would still have occurred even if many of the individual movements of the balls had been rather different. Such a rational expectation can only be provided, it seems, by an explanation in terms of the player’s intentional state. And such an explanation requires us to assign a causal role to that intentional state, a role which no purely physical state seems apt to occupy. But what I have just said about the explanation of event E very often applies equally to the explanation of other event-fusions, such as the sort of event-fusion which constitutes an arm-rising. The advantage of focusing on event E, for our purposes, is simply that it is very much less likely, in its case, than it is in the case of an arm-rising, that such an event could be provided with a wholly physical causal explanation other than one which made it appear to be a mere ‘fluke’ which happened purely by coincidence.
Self, Agency, and Mental Causation 39
1.12. An Objection and a Reply At this point, I anticipate the following sort of objection. Surely, it may be said, someone could design a snooker-playing robot which could be pretty well relied upon to pot all of the snooker balls on a table—and when it did this there would occur an event of the same kind as E, which would be neither a ‘fluke’ nor ‘coincidental’: and yet, clearly, this event would have a purely physical causal explanation, since the robot would be a purely physical device possessing no mental states whatever, let alone a ‘desire’ to pot all of the snooker balls. So how can we be at all confident in denying that a wholly physical causal explanation of event E is available in the case of the human snooker player? My response is as follows. I agree that such a snooker-playing robot could, in principle, be designed and constructed. But note, first of all, that this is not to abandon appeal to intentional causation in our explanation of its feats, since we are quite explicitly appealing to the intentional states of the robot’s designer and maker. It is not remotely plausible to suppose that a device like this could come into being completely without any causal contribution from the intentional states of any thinking being. Second, note that, in describing the workings of the robot, we have conceded that it possesses no intentional states whatever, such as a desire to pot all of the snooker balls on a table. Indeed, the force of the objection that has been raised—that the robot provides an example of how an event of the same kind as event E could have a purely physical causal explanation—rests upon the presumption that the robot does not have any intentional states. But then it follows that if we are to see the example of the robot as providing an alternative paradigm for the explanation of event E in the case of the human snooker player, we have to regard that paradigm as an eliminativist one, in which appeal to the player’s ‘desire’ to pot all of the balls has no genuine explanatory force. Now, I concede that it is possible for every physical action that is performed by a human being to be caused in a wholly ‘robotic’ way, without any causal contribution from intentional states. But the point is that this possibility is an extremely remote one: we have no good reason whatever to suppose that it has been realized in this, the actual world. On the contrary, we have every reason to think that people’s beliefs and
40
Mental Causation, Closure, and Dualism
desires do contribute causally to their physical behaviour and help to explain it. What we cannot do, however, is try to combine this conviction with the thought that, somehow, human behaviour does, in principle, have a purely physical causal explanation along the lines of robotic behaviour, in the hope of reconciling intentional causation with a thoroughgoing physicalism. We can either take intentional causation seriously, in which case we must abandon physicalism, or else we can cleave to physicalism, in which case we must be eliminativists (or, perhaps, epiphenomenalists) about the mental. There is no middle ground which allows us to have it both ways. However, as I have tried to show, we can espouse a version of ‘dualism’ (for want of a better word) which preserves one central tenet of physicalism, namely, claim (3): that every physical event has a set of wholly physical causes which are collectively causally sufficient for the occurrence of that event. If it is only a concern that this claim is denied by dualism which persuades some philosophers to reject dualism in favour of physicalism, then I hope I have shown them why that concern is quite misplaced. If, on the other hand, their physicalism is motivated by a faith in claim (4)—that no physical event has a non-physical cause —then I can only say that it seems to me that their doctrine is an unwarranted dogma which commits them, whether they like it or not, to eliminativism or epiphenomenalism regarding the mental. The kind of ‘dualism’ that I am defending is fully deserving of the title ‘naturalistic’—provided that the term ‘naturalism’ is not hijacked as a mere synonym for ‘physicalism’, but is accorded its proper meaning as denoting a repudiation of the supernatural.
2 Causal Closure Principles and Emergentism I shall look, in this chapter, at a wide range of physical causal closure principles of differing strengths and argue that the stronger of these principles are significantly less plausible than the weaker while the weaker are considerably less capable of playing an adequate role in any causal closure argument for physicalism. The upshot will be that there is plenty of scope for a psychophysical dualist to accept a moderately strong principle of physical causal closure while also advocating a genuine and autonomous causal role for mental states in the genesis of intended physical behaviour. Such a dualist, we shall see, may consistently espouse a form of emergentism in the philosophy of mind, according to which non-physical mental states are causally autonomous and yet are themselves ultimately the products of prior physical evolution.
2.1. Causal Closure Arguments for Physicalism At one time, it was popular to argue against interactionist dualism by appeal to the conservation laws of physics, such as the laws of the conservation of energy and momentum. However, those laws are not sacrosanct and there need be nothing unscientific about questioning them—indeed they have been questioned from time to time by cosmologists and other physicists. Think, for instance, of Bondi and Gold’s steady-state model of the universe, which postulated the continuous creation of matter and hence of energy.¹ In any case, appeal to those laws can at best only be used to attack dualist ¹ See H. Bondi, Cosmology, 2nd edn (Cambridge: Cambridge University Press, 1961), ch. XII.
42
Mental Causation, Closure, and Dualism
models of psychophysical causation which attribute to the non-physical mind an ability to affect the energy or momentum of a physical system. I say ‘at best’ because some interactionist dualists, such as W. D. Hart, have postulated the existence of ‘psychic energy’, which is convertible into physical energy in accordance with the conservation laws.² It won’t do simply to object that energy is by definition a physical quantity, as this threatens to turn the dispute into a purely verbal one. In view of these and other limitations, it would serve physicalists well if they could frame a more general argument against interactionist dualism, which did not make specific appeal to the conservation laws.³ Many modern physicalists think that they have an argument of just this sort to hand, in the form of the causal closure argument.⁴ In fact, this is not so much a single argument as a family of arguments. A ‘causal closure argument’, as I shall be interpreting that phrase, has three premises: first, a physical causal closure principle; second, the claim—to which interactionist dualists are themselves committed—that at least some mental events are causes of physical events; and third, the claim that the physical effects of mental causes are not, in general, causally overdetermined.⁵ The conclusion of such an argument is that at least some mental events are identical with physical events. Incidentally, I am assuming for the time being that the relata of causal relations are events, broadly conceived to include both processes and states, but not too much turns on this assumption at the moment. Later on, however, I shall introduce a distinction between event causation and fact causation which may have a significant bearing on the issues under discussion.⁶ 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 As we shall see shortly, physicalists espousing causal closure arguments do not always agree on the way to formulate their causal closure principles. What is crucial to the success of such an argument is that the causal closure principle appealed to should be neither too strong nor too weak. If the principle is so strong, for instance, as to render redundant the third premise of the argument—the non-overdetermination claim—then the argument will become simply question-begging, because it will amount to an argument from two premises, one of which is accepted by the dualist, to a conclusion which is not accepted by the dualist: and in these circumstances the dualist may justly urge that his opponent is, in effect, merely asserting what he is denying. Moreover, to have any persuasive force, the causal closure principle must be one for which some measure of empirical support can plausibly be mustered, and this places limits on how strong it can be. On the other hand, the causal closure principle must, obviously, not be so weak that the argument in which it is deployed turns out to be invalid. What I shall try to show in this chapter is that it is in fact very difficult, if not impossible, for the physicalist to formulate a causal closure principle which is neither too strong nor too weak by these standards. In particular, I shall argue that there are various forms of naturalistic dualism, of an emergentist character, which are perfectly consistent with the strongest physical causal closure principles that can plausibly be advocated. If that is correct, then it means that causal closure arguments against interactionist dualism do not, after all, provide a superior alternative to the discredited arguments which appeal to the conservation laws of physics.
2.2. The Surprising Variety of Causal Closure Principles I mentioned a moment ago the variety of causal closure principles appealed to by physicalists. Let me briefly illustrate this variety by some examples. In a recent paper, David Papineau appeals to the following principle, which he calls the principle of ‘the completeness of physics’:⁷ (1A) All physical effects have sufficient physical causes. 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
But in an earlier book, he appeals instead to a somewhat different principle, which he calls by the same title, namely:⁸ (1B) All physical effects have complete physical causes. Papineau explains that by ‘complete’ here he means ‘ ‘‘complete’’ in the sense that those causes on their own suffice by physical law to fix the chances of those effects’. However, he does also advert to the latter formulation in a footnote to the more recent paper, remarking that ‘a stricter version ... would say that the chances of physical effects are always fixed by sufficient physical causes’.⁹ In another recent contribution to the literature on causal closure, Scott Sturgeon states what he too calls the principle of the completeness of physics as follows:¹⁰ (1C) Every physical effect has a fully revealing, purely physical history. However, Sturgeon himself glosses this in terms reminiscent of Papineau’s, as meaning that ‘physical effects have their chances fully determined by physical events alone’. In a paper commenting on Sturgeon’s, Paul Noordhof seizes upon this gloss, remarking that it ‘seems more perspicuous’ than Sturgeon’s official formulation.¹¹ Accordingly, Noordhof advances the following as his own preferred version of the causal closure principle: (1D) Every physical effect has its chance fully determined by physical events alone. It is worth mentioning, however, that in a further gloss on his version of the principle, Sturgeon asserts that ‘physics does not admit that physical effects have non-physical causes’,¹² and this suggests yet another formulation, namely: (1E) No physical effect has a non-physical cause. Pretty clearly, principles (1A) to (1E) are not all equivalent to one another, whatever one makes of the various glosses supplied by their advocates. ⁸ David Papineau, Philosophical Naturalism (Oxford: Blackwell, 1993), p. 22. ⁹ Papineau, ‘Mind the Gap’, p. 386, n. 4. ¹⁰ Scott Sturgeon, ‘Physicalism and Overdetermination’, Mind 107 (1998), pp. 411–32: see p. 413. ¹¹ Paul Noordhof, ‘The Overdetermination Argument versus the Cause-and-Essence Principle—No Contest’, Mind 108 (1999), pp. 367–75: see p. 367. ¹² Sturgeon, ‘Physicalism and Overdetermination’, p. 413.
Causal Closure Principles and Emergentism 45 And these by no means exhaust the variety of formulations to be found in the literature. One might have hoped for more exactitude and agreement amongst physicalists when it comes to the formulation of a principle so central to their position.
2.3. Some Causal Closure Principles that we may Justifiably Ignore One thing we should notice immediately is that the last version just cited, (1E), is too strong by the standards laid down earlier, because it is bound to render redundant the non-overdetermination premise of any causal closure argument in which it is deployed. To see this clearly, let me now state explicitly what the remaining premises and the conclusion of such an argument are supposed to be. Obviously, these premises and the conclusion are themselves susceptible to some variation in their formulation, but I shall utilize what I hope can be agreed to be uncontentious versions of them. The second premise, maintaining the existence of psychophysical causation, may be stated thus: (2) At least some mental events are causes of physical events. The third premise, excluding the possibility of systematic causal overdetermination, may be expressed as follows: (3) The physical effects of mental causes are not, in general, causally overdetermined. And the anti-dualist conclusion of the argument is supposed to be this: (4) At least some mental events are identical with physical events. However, it is quite evident that (1E) and (2) together entail (4), without the help of (3), so that (1E) can fairly be dismissed by the interactionist dualist as question-begging—as I am sure it would be. In what follows, I am not going to consider probabilistic versions of the causal closure principle, such as (1D), because they introduce complications which are not relevant to the main thrust of what I have to say in this chapter. Evidently, the chief reason why probabilistic versions have some favour amongst physicalists is that they do not want it to be objected
46
Mental Causation, Closure, and Dualism
against their argument that it presupposes a deterministic physics which is at odds with modern quantum theory. But since many dualists and physicalists would probably agree—rightly or wrongly—that quantumlevel phenomena are largely irrelevant to the problem of psychophysical causation, perhaps we can sidestep the complications raised by probabilistic causation without unduly oversimplifying the debate. It is true, of course, that some interactionist dualists, such as Popper and Eccles, have maintained that quantum-level phenomena are very much relevant to the problem of psychophysical causation,¹³ but it may perhaps be urged that they are an unrepresentative minority whom we can afford to ignore for present purposes. And, after all, if interactionist dualism can be rendered consistent with a deterministic version of the causal closure principle that is not excessively strong by the standards laid down earlier, it can hardly be more problematic to render it consistent with a probabilistic version. (That said, I shall myself, in later chapters, urge that quantum indeterminacy is indeed a very important consideration where matters to do with freedom of the will are concerned. So my setting aside of this consideration for present purposes is only provisional. In any case, I shall return to the issue in the next chapter, so further discussion of it here is unnecessary.)
2.4. A Causal Closure Principle that is Manifestly too Weak With this restriction in mind, let us look next at a version of the causal closure principle which is very close to Papineau’s version (1A) above, namely: (1F) Every physical event which has a cause has a sufficient physical cause. Here I should say that I understand a sufficient physical cause of a given event to be a non-empty set of physical events, each of which is a cause of the 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 nonoverdetermination 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 interactionist 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 physical 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 psychophysical 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
2.5. A Stronger Causal Closure Principle that avoids the Transitivity Problem The physicalist needs, then, a causal closure principle which is stronger than (1F). One likely candidate would be the following, which seems to avoid the problem created by the transitivity of causation:¹⁵ (1G) At every time at which any physical event has a cause, it has a sufficient physical cause. With (1G) in place, it might seem, we cannot postulate that a physical event P has a non-physical cause M existing at a time t1 in addition to the sufficient physical cause which, according to (1G), must exist at t1 , without violating the non-overdetermination premise (3). But in point of fact that is not so, because this neglects the possibility of simultaneous causation. Suppose that P occurs at t2 and that the postulated sufficient physical cause of P which exists at t1 causes P in part by means of causing M, which also exists at t1 . In that case, P is not causally overdetermined by M and the physical causes of P that exist at t1 . A situation of this sort is illustrated by Figure 2.1, in which an arrow between events represents the fact that the event at the tail of the arrow is an immediate cause of the event at the head of the arrow: t0
P01
P02
t1
P11
P12
t2
M
P
Figure 2.1. Simultaneous causation without overdetermination 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 In the situation depicted in Figure 2.1, P has a sufficient physical cause at time t1 , namely, the set of physical events {P11 , P12 }. But P also has as a cause at t1 the non-physical event M. P is not causally overdetermined by {P11 , P12 } and M, because it is not the case that in the absence of either one of them P would still have occurred. In particular, since P is caused by both P11 and P12 , if {P11 , P12 } had been absent, P would not have occurred. Equally, if M had been absent, P would not have occurred, for want of a sufficient cause: for we are supposing that {P11 , P12 } would not have been a sufficient cause of P if P12 had not caused P in part by means of M. Of course, the physicalist may well want to deny the possibility of simultaneous causation, but it should be acknowledged that this is a further substantive claim, without which the causal closure principle (1G) cannot be used to mount an attack on interactionist dualism.¹⁶ Here I pause to remark that, in his own version of the causal closure argument, Christopher Peacocke considers the possibility of a scenario very similar to the one depicted in Figure 2.1, but contends that it is already ruled out by the prohibition on causal overdetermination.¹⁷ This is what he says, replacing his labelling of events by my own: Even if P12 causes M, and so it is true ... that if P12 had not occurred, then M would not have occurred, the argument stands. For still there is a causal route from M to P, one that evidently cannot pass through P12 given the asymmetry of the causation relation; and so overdetermination of P remains the case. The fact that it is true that if P12 had not occurred, then M would not have occurred, does not show that there is not overdetermination. We can show this by considering another example. Your pressing a button to light a bonfire electrically may be a 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.
However, it is not enough for Peacocke to establish, by means of such an example, that ‘The fact that it is true that if P12 had not occurred, then M would not have occurred, does not show that there is not overdetermination’. Rather, it is incumbent upon him to establish that in the envisaged psychophysical scenario there is overdetermination. Now, the bonfire example obviously does involve overdetermination, because your pressing the button would have sufficed to light the bonfire even if it hadn’t caused me to light it with a match. But in the situation depicted in Figure 2.1, as I am interpreting it, {P11 , P12 } would not have sufficed to cause P if P12 had not caused M. Suppose one were to add to a causal closure argument invoking the causal closure principle (1G) a further premise ruling out simultaneous causation: would (1G) be strong enough to deliver the anti-dualist conclusion (4) in that case? The answer, as we shall see in due course, is that it would not. But before we examine that issue, let us consider whether (1G) is not already too strong by the standards laid down earlier. According to those standards, a causal closure principle should not be so strong that empirical support for it cannot plausibly be mustered. The weaker principle (1F) is certainly not too strong by these standards, since it is empirically plausible to suppose that the universe contained only physical events at an earlier epoch and that latter-day mental events are the ultimate effects of causal chains traceable back to events occurring at that earlier epoch. But when we consider that temporal ordering is dense, such that between any two times there is another time, (1G) appears unreasonably strong. Suppose that a physical event P, such as a bodily movement, is known on good empirical grounds to have a sufficient physical cause in a person’s brain at a time 100 milliseconds before the occurrence of P. And suppose it is further postulated that that sufficient physical cause causes P in part by causing an intermediate mental event M which occurs at a time 90 milliseconds before the occurrence of P. It cannot be illegitimate in principle to postulate something like this, given that temporal ordering is dense, and empirical evidence could quite conceivably support the claim that M occurred at the postulated time. However, principle (1G) then requires us to suppose that
Causal Closure Principles and Emergentism 51 there is a sufficient physical cause of P existing at a time 90 milliseconds before the occurrence of P, in advance of any empirical evidence in support of this supposition. That seems extravagant, especially when we consider that (1G) imposes the same requirement with regard to any intervening time between 100 and 0 milliseconds before the occurrence of P.
2.6. Weak Causal Closure and Non-Coincidence Through Mental Causation Let me now return to the weaker causal closure principle (1F) and consider how dualistic interaction might operate in a world governed by that principle. What I want to suggest is that, in such a world, mental events might serve to render non-coincidental certain physical events which, from a purely physical perspective, might appear to occur by coincidence—a suggestion that, it will be recalled, I began to develop towards the end of the previous chapter. For present purposes—and simplifying somewhat for ease of exposition—I take an event to be one which ‘occurs by coincidence’ if its immediate causes are the ultimate effects of independent causal chains. (One complication which I shall sidestep is the question of what is to count as an ‘immediate cause’ of a given event, to which different answers might be given according to more or less fine-grained ways of individuating events.) A simple example, discussed earlier in Chapter 1, is this. A man walks to the shops from his house one day, following his usual route, but as he passes by a neighbour’s house, a gust of wind dislodges a slate which falls on his head, killing him. In this case, the man’s death occurred by coincidence, because the causal chain leading to his being where he was when the slate fell was independent of the causal chain leading to the slate’s being where it was when the man passed by. We can easily imagine, however, another scenario in which the man’s death did not occur by coincidence: for example, a scenario in which a tripwire installed by a jealous neighbour is touched by the approaching man, dislodging the slate so that it falls just as he arrives beneath it. Changing now to an example involving psychophysical causation, suppose that a physical event P —such as a bodily movement, or a neural event which is a direct causal antecedent of such a bodily movement—has as its immediate causes two physical events, P21 and P22 , occurring at a time t2 ,
52
Mental Causation, Closure, and Dualism
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
t1
P11
t2
P21
t3
P02
M
P12
P22
P
Figure 2.2. Non-coincidence through mental causation
In the situation depicted in Figure 2.2, P has a sufficient physical cause at times t0 and t2 , namely, {P01 , P02 } and {P21 , P22 } respectively, but P does not have a sufficient physical cause at time t1 . This makes the situation consistent with principle (1F), which requires only that a physical event should have a sufficient physical cause at some time at which it has a cause, not at every time at which it has a cause—the latter requirement only being imposed by the stronger principle (1G). In a world containing the situation depicted in Figure 2.2, a physical scientist might be able to discover all the physical events depicted there, correctly identify each of them as a cause of P, and correctly claim that certain combinations of them are sufficient causes of P. In a sense, he would be in possession of a complete physical explanation of the occurrence of P, for he would have an explanation which correctly identifies all of P’s physical causes up to a certain time before the occurrence of P and a time (in fact, more than one time) at which a subset of those causes is wholly sufficient for P. And yet, because the mental event M would be, as it were, invisible to him as a purely physical scientist, his explanation
Causal Closure Principles and Emergentism 53 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
2.7. The Consistency of Strong Causal Closure with Dualistic Interactionism What I want to show next is that the relatively weak physical causal closure principle (1F) is far from being the strongest such principle that is compatible with dualistic interactionism. In fact, dualistic interactionism is even consistent with a causal closure principle that is stronger in some ways than principle (1G), the strongest principle that we have examined so far, apart from the question-begging principle (1E). The principle I have in mind is the following:¹⁹ (1H) Every physical event contains only other physical events in its transitive causal closure. 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
By the ‘transitive causal closure’ of an event P, I mean the set of events consisting of the immediate causes of P, the immediate causes of those causes, the immediate causes of those causes ... and so on: in short, the set which includes every event which stands in the ancestral of the ‘immediate cause’ relation to P. The implication of (1H), then, is that the immediate causes of all physical events are always and only other physical events. Principle (1H) clearly prohibits the situations depicted earlier in Figure 2.1 and Figure 2.2 (remembering that an arrow between events depicted there signifies immediate causation). Indeed, there is a clear sense in which principle (1H) is stronger than principle (1G)—even when the latter is supplemented by a prohibition on simultaneous causation—though it is true that (1H) is weaker in that it does not imply that any physical event has a sufficient physical cause. However, in this respect (1H) has an empirical advantage over (1G), inasmuch as it is compatible with the claim, which is plausible in the light of quantum mechanics, that at least some physical causation is irreducibly probabilistic. How could dualistic interaction be compatible with principle (1H)? For the following reason. It could conceivably be the case that, even though (1H) is true, sometimes a non-physical mental event M causes it to be the case that certain physical events, P1 , P2 , ... Pn , have a certain physical effect, P.²⁰ What this would involve is the causation by a mental event of a physical causal fact. So I am assuming now a distinction between event causation and fact causation. In a case of fact causation, what is brought about is not an event, but a fact or state of affairs. Some facts are causal facts, such as the fact that certain physical events, P1 , P2 , ... Pn , are causes of another physical event, P. There is no reason in 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
example demonstrates the intelligibility of mental causation of physical causal facts.
2.8. What can We Reasonably Demand of a Model of Mental Causation? But how could anything like the foregoing model of mental causation be envisaged as applying in cases of ordinary human behaviour? How, for instance, could a mental event, such as a choice of mine, cause it to be the case that certain physical events in my brain cause a certain bodily movement of mine? But what is being requested here? If what is being requested is a demonstration that causation of this kind—the causation of a physical causal fact by a mental event—is intelligible, then we have already provided it by means of the foregoing theological example. If, on the other hand, what is being requested is some account of the ‘mechanism’ that would be involved in such causation, in the human case, then it may be protested that the request is perhaps unreasonable—first of all because the term ‘mechanism’ is suggestive of purely physical causation, which is precisely not being envisaged here, and secondly because supplying more detail about the possible nature of such causation in the human case may not properly be a matter for armchair philosophical speculation. My aim has only been to show that dualistic interactionism is compatible with a surprisingly strong form of physical causal closure principle, not to help the would-be interactionist to construct a detailed model of psychophysical causation. This reply may seem unduly evasive. Is it not an unacceptable implication of the proposed form of mental causation that the mind would have to have the power to affect physical causal laws? Not necessarily, for we need not assume that all causal relations between physical events are subsumable under laws.²¹ There may be irreducibly singular causation between physical events, in which case a mental event could cause a physical causal fact without necessarily causing a change in physical laws. But if the real thought behind the objection is that mental causation of the kind envisaged 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 would somehow inevitably conflict with the conservation laws of physics, then I can only say, first, that the burden of proof in this matter lies with my opponent and, second, that if, after all, the physicalist must fall back on an appeal to the conservation laws in an attempt to defeat the dualist, then all the apparent advantages of appealing to causal closure principles have fallen away and we have returned to the tired old debate of earlier times. But why, it may be asked, should we feel any need to invoke mental causation of the kind just proposed? The answer, once again, might be that only by invoking such mental causation could we represent the physical effect of certain physical causes—such as a bodily movement produced by a large number of apparently independent neural events in the brain—as being non-coincidental, since there might be no identifiable physical event(s) which could be seen to link the chains of physical causation involved in an appropriate way. At the same time, there might be strong empirical evidence that a mental event was playing precisely such a role, and a strong intuition that the bodily movement in question was not an event which occurred ‘by coincidence’. All of this would be perfectly compatible with a naturalistic form of emergentism, moreover, since we could still suppose that every mental event ultimately has a sufficient physical cause. Of course, a physicalist might object that if empirical evidence did confirm that a mental event was playing the kind of causal role now being mooted, we would have to conclude that this event was identical with some physical event, almost certainly some neural event in the brain. But why would we have to conclude this, in the absence of any direct empirical evidence for the existence of such a physical event? What I have been trying to show in this chapter is that causal closure arguments do not provide the kind of justification for such a claim that physicalists may think they do.
3 Physical Causal Closure and the Invisibility of Mental Causation In the current debates about non-reductive physicalism and mental causation, there is a general presumption that the causation of physical events by non-physical mental causes would constitute an empirically detectable violation of any plausible principle of physical causal closure. This is supposed to spell trouble for interactive dualism. At the same time, this sort of consideration is thought by many to raise difficulties also for non-reductive physicalism and to tell ultimately in favour of either reductive physicalism or eliminative materialism. However, in this chapter, I shall argue that if interactive dualism were true, it would in fact be entirely reasonable to expect mental causation to be ‘invisible’, in the sense that no ‘gaps’ would be apparent in the causal relations between physical events and all physical events would seem to have wholly physical causal explanations. If correct, this renders the debate between interactive dualists, non-reductive physicalists, and reductionists or eliminativists one that is not amenable to resolution by appeal to empirical evidence of a certain kind. While this is not to say that there is no kind of empirical evidence that could have a bearing on this debate, the implication would seem to be that purely metaphysical arguments must have a larger role to play than is commonly supposed in determining which of these positions is superior.
3.1. The Early Modern Roots of the Debate over Interactive Dualism Why are the debates between physicalists and dualists in the philosophy of mind still no nearer to a resolution, more than three hundred years
Physical Closure and Mental Causation 59 after Descartes initiated those debates in their recognizably modern form? Descartes lived in the dawn of modern science and his contemporaries might understandably have expected that the mind–body problem would have been resolved through the advance of scientific knowledge in the course of the following three centuries. And yet it plainly hasn’t been, even if there is nowadays a general presumption in favour of some sort of physicalism in the philosophy of mind. Descartes himself, of course, was no empiricist in matters concerning the fundamental science of nature (although, to be sure, he appealed to empirical considerations where scientific issues of a less than fundamental character were concerned). His chief arguments in favour of interactive dualism were broadly a priori in character and so he, at least, would presumably not have been surprised that empirical evidence for or against dualism has been extremely elusive. But many other philosophers and scientists have been more sanguine about the relevance of empirical evidence to the mind–body problem. Indeed, some of the earliest criticisms of Descartes’s own interactionist scheme appealed to broadly empirical considerations, notably the apparent conflict between that scheme and various empirically confirmed conservation laws.¹ (Of course, such criticisms must be construed as objections not so much to Descartes’s arguments—given that the arguments were a priori while the objections turned on empirical considerations—but rather to the conclusions of those arguments.) Descartes himself believed, on a priori grounds, in a principle of the conservation of ‘motion’, the idea being that God, in creating the physical world, imparted a certain total ‘quantity of motion’ to matter, distributed amongst all material bodies, and that this total quantity, while it might become redistributed by the impact of one material body upon another, could never be altered except through His will. According to this conception of the physical world, quantity of motion could no more be created or destroyed by natural means than matter itself could be created ex nihilo or annihilated by natural means. Famously, Descartes postulated that the immaterial mind could alter the direction of flow of ‘animal spirits’ in the pineal gland of the human brain, without thereby affecting either the amount or the distribution of the ‘quantity of motion’ in the material ¹ For the historical background, see R. S. Woolhouse, ‘Leibniz’s Reaction to Cartesian Interactionism’, 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
world. Equally famously, Descartes’s near contemporaries, notably Leibniz, pointed out the empirical falsity of his principle of the conservation of ‘quantity of motion’ and the conflict between his interactionist scheme and the empirically well-confirmed principle of the conservation of momentum. Cartesian ‘quantity of motion’ differed from momentum, in the modern sense, in being a scalar rather than a vector quantity: the former was understood by Descartes as being, in effect, the product of a body’s mass and its speed, whereas the latter is the product of a body’s mass and its velocity. Because the velocity of the animal spirits would be altered by a change in their direction of flow, so too would their momentum be altered, without any compensating alteration in the momentum of other bodies so as to conserve momentum overall. Thus it appeared that there was conclusive empirical evidence against the feasibility of Descartes’s interactionist scheme, inducing dualists either to give up their dualism altogether or, like Leibniz, to espouse some non-interactionist version of the doctrine. In this early modern debate, we see prefigured some of the latter-day debates between physicalists and non-physicalists of recent years. But an outsider might wonder why the debate has been revived at all, given that it seemed long ago to be so firmly resolved to the detriment of Cartesian interactionism. Part of the answer, no doubt, is that it is an illusion to suppose that the early modern debate was in fact so conclusively resolved, because the Cartesian interactionist has many possible ways to counter the objection founded on appeal to the principle of the conservation of momentum. For one thing, it may be pointed out that the empirical evidence in support of that principle, such as it is, is all drawn from observations of or experiments upon purely physical systems in which material bodies interact with one another thing, whereas, ex hypothesi, mind–body interactions are not of that kind. For another, it may be pointed out that what the principle of the conservation of momentum actually asserts is that, in a system of mutually interacting material bodies upon which no external force is acting, the vector sum of the momenta of the bodies is constant. But what, then, is to prevent the interactive dualist from maintaining that the immaterial mind is able to exert an ‘external force’ upon such a system of bodies? To this it may be objected that now another conservation principle will come into play, namely, the principle of the conservation of energy, since energy is expended when work is done by the exertion of a force. However, it is now open to the interactive dualist
Physical Closure and Mental Causation 61 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
would be no more ‘occult’ than Newton’s mysterious ‘force of gravitation’, which supposedly operates instantaneously and at a distance, rather than by contact and impulse. (Since Einstein’s conception of gravitation is scarcely less mysterious, with its implication that the presence of matter somehow affects the metric of space–time, and we still lack any detailed alternative theory of quantum gravity, it can hardly be complained that we are now in a much more enlightened condition.) However, what would perhaps have disconcerted Descartes, and maybe rightly so, would have been the discovery—supposing it to have been made—that the change in momentum undergone by the ‘animal spirits’ and attributed by our hypothetical Descartes to the new ‘mental force’ was in fact entirely explicable in terms of the action of other, purely ‘physical’ forces. Suppose, for instance, in a further departure from the actual history of science, that electrostatic forces were already known of in those times and that the change of momentum of the ‘animal spirits’ could be attributed entirely to the action of such forces. Then, indeed, it might seem that our hypothetical Descartes would find himself in a quandary: not because his notion of a ‘mental force’ was lacking in scientific respectability, but simply because it would appear to be explanatorily redundant. But notice that what is now at issue is not the violation or non-violation of the conservation laws, but the question of whether a ‘non-physical’ causal explanation of certain physical phenomena is redundant or superfluous. In that respect, our hypothetical extension of the early modern debate has simply taken us directly into the territory of the contemporary debate, with its focus on physical causal closure principles and the rejection of any idea that physical phenomena might be independently causally overdetermined by both physical and non-physical causes.
3.2. The Contemporary Debate and the Appeal to Causal Closure So let us now leave behind the early modern debate, in both its actual and its counterfactual forms, and advance to the contemporary debate. As I have already indicated, it is part of my purpose here to argue that this contemporary debate really has no more prospect of achieving a straightforward empirical resolution than the early modern debate turns out
Physical Closure and Mental Causation 63 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.
3.3. Is the Causal Closure Principle (CCP) True? So let us look first at the credentials of (CCP). Plainly, (CCP) could not be advanced with any plausibility as being an a priori truth. If it is to be accepted at all, it must be on empirical grounds. But here already we are in difficult waters, because we have every reason to suppose that (CCP) is empirically false, its falsity being testified to by the empirical data which support the principles of modern quantum physics. Perhaps it will be replied that quantum physics only supports the contention that some physical events are uncaused, but does not challenge the claim that if a physical event has a cause at a time t, then it has a wholly physical sufficient cause at t, which is what (CCP) asserts. Consider, however, the notorious case of Schr¨odinger’s cat, which is alive or dead, as the case may be, depending on whether or not a radium atom has undergone spontaneous decay at an earlier time t.⁴ No doubt the event of the radium atom’s decay, supposing it to occur, is a physical event which is uncaused. But this doesn’t prevent us from saying that the cat’s death, in that case, has a physical cause at t and, indeed, a wholly physical sufficient cause at t. Part of that cause will be the event of the radium atom’s decay. However, that event alone will not be a sufficient cause of the cat’s death. If the cat’s 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 Schr¨odinger’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 interactive dualism, but it may not be so easy to accomplish this satisfactorily. Suppose that (CCP) is reformulated as: (CCP∗ ) For any physical event e, if e has its chances fixed by some event at time t, then e has its chances fixed by wholly physical events at t. Then one might think to reformulate (NOP) correspondingly in this way: (NOP∗ ) Most physical events e are such that, if e has its chances fixed by a mental event at time t, then e does not also have its chances fixed by wholly physical events at t which do not include that mental event. But (NOP∗ ) plainly cannot serve the physicalist’s purposes, because no one wants to say that any physical event ever has its chances fixed simply by a mental event at an earlier time: at most one would want to say only that a mental event could, along with other, physical events, help to fix the chances of some subsequent physical event. Thus, in the present context, it is no use simply replacing talk of causation by talk of the fixing of chances: we would need to talk in terms of some events ‘helping’ to fix the chances of other events. Maybe that can be done in a satisfactory way. But it is unlikely that the resulting principles to replace (CCP) and (NOP) will command immediate assent or even immediate comprehension, if only because our understanding of chance is far less firm than our understanding of causality—though even the latter can hardly be described as being particularly firm. We have seen that (CCP) must be regarded as false, if quantum physics (as it is standardly interpreted) is correct. But it is worth remarking that there is another reason for thinking that (CCP) must be false, because there is a reason for thinking that even the following weaker principle is false: (CCP#) For any physical event e, if e has a cause at time t, then e has a sufficient cause at t. The reason for thinking (CCP#) to be false is that its truth is incompatible with freedom of the will, and we cannot rationally believe that our will is not free. If the will is genuinely free, then sometimes a mental event (namely, an act of will or, in other words, a choice or decision) is a cause of a physical event, and yet that physical event does not have a sufficient cause
68
Mental Causation, Closure, and Dualism
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.)
3.4. Is the Non-Overdetermination Principle (NOP) True? Although I am firmly convinced that (CCP) is false, let us waive this point and assume, for the sake of argument, that (CCP) is true. After all, an interactive dualist should be prepared to contend that the dualist account of mind–body relations would still remain true even in a deterministic world in which neither quantum physics nor libertarian free will obtained. This is because dualism is an ontological doctrine, concerning the ontological status of mental events, holding that mental events do not stand in any relation of ontological dependency to physical events, so that they can be neither identical with, nor even ‘realized by’, physical events.⁷ This being so, our attention should now be focused on the second premise in the 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 argument against dualism, the non-overdetermination principle (NOP). This, recall, we stated as follows: (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. What is the motivation behind (NOP)? Apparently, it is the thought that if there were widespread causal overdetermination of the sort ruled out by (NOP), this would have to be seen as a massive and wildly improbable accident or coincidence. The model of causal overdetermination that may be in mind here is that of the classic ‘double assassination’ scenario, in which two assassins simultaneously and independently shoot at their victim and the impact of each bullet, even in the absence of the other, would have sufficed in the circumstances to bring about the victim’s death. Note, however, that (NOP) does not only serve to rule out causal overdetermination of this sort where mental causes are concerned: it also rules out causal overdetermination of another, non-independent sort. Some non-reductive physicalists have made a similar point in this connection. For instance, such physicalists who hold that mental events, while not identical with physical events, are none the less ‘realized by’ physical events, urge that when a mental event occurring at a time t causes a certain physical effect, that physical effect also has, at t, a wholly physical sufficient cause which is wholly distinct from the mental cause in question, because although that mental cause is realized by one or more physical events which are part of that sufficient physical cause, it is not identical with any part of that sufficient physical cause. They point out, however, that because the mental event stands in the ‘realization’ relation to the physical event or events in question, it and they are not independent in the way that the two assassins’ shots are, and consequently the sort of causal overdetermination that ensues is not, as in the assassination scenario, a mere coincidence. (Incidentally, if it sounds odd to say that a mental event, conceived as being ‘realized’ by one or more physical events, would be ‘wholly distinct’ from those physical events, it should be remembered that ‘wholly distinct’ is being understood here in a purely mereological sense, according to which events are wholly distinct just in case there is no event which they share as a common part. Clearly, if mental events are conceived as being ‘realized’ by physical events but never identical with physical events, no mental event can be understood
70
Mental Causation, Closure, and Dualism
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 nonreductive 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 presumed to be possible. But why should it not be possible? There may perhaps be objections to the possibility of simultaneous causation between a physical event and another physical event—for example, on the grounds that physical causation always involves some form of energy transfer, and that processes of energy transfer always take time; or, again, on the grounds that physical causes and effects cannot coincide spatiotemporally, and that there is (pace Newton) no instantaneous action at a distance. But considerations like these do not apply to causation between physical events and mental events, dualistically conceived. The dualist will typically deny that psychophysical causation involves any process of energy transfer (even if, as we saw earlier, dualism is not necessarily committed to such a denial) and will also typically deny that mental events have, in any straightforward sense, a spatial location. In order to make the foregoing proposal clear, it may be helpful to illustrate it with the aid of a diagram, Figure 3.1 (compare Figure 2.1 of Chapter 2). In Figure 3.1, P is a physical event occurring at a time t2 and M is a (nonphysical) mental cause of P occurring at the earlier time t1 . Consistently with (CCP), because P has a cause at t1 , we have represented P as having a wholly physical sufficient cause at t1 , namely, the conjunction of physical events P11 and P12 . However, we have also represented M as having one of those physical events, P12 , as a sufficient cause. Since P12 occurs at the same time as M, its causation of M is represented as being a case of simultaneous causation. In the situation depicted, principle (NOP) is violated, because (NOP) does not allow P to have, at t1 , both the mental cause M and the wholly distinct and wholly physical sufficient cause which consists of the conjunction of P11 and P12 —at least, (NOP) does not t0
P01
P02
t1
P11
P12
t2
P
Figure 3.1. A dualistic counterexample to (NOP)
M
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 overdetermination’, 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 Note that it would be wrong to object to the legitimacy of what has just been postulated by attempting to pose the following dilemma: either P11 and P12 need the help of M to bring about P, in which case they are not jointly sufficient for P, or else they do not need the help of M, in which case M is redundant. For there is no inconsistency in saying that P11 and P12 are jointly sufficient for P and yet that they also need the help of M to bring about P, provided that the sense in which they ‘need the help of M’ is just that, as is consistent with Figure 3.1, P11 and P12 need to bring about P at least in part via a causal route leading from P12 to M and thence to P. This is just to say that P11 and P12 cannot bring about P ‘immediately’, but only (in part) via an intermediate effect, M. It should be perfectly clear, though, that it is no threat to the sufficiency of a cause for one of its effects that it has to bring about that effect via some intermediate effect: after all, if physical determinism is true, certain physical events in the early history of the universe were causally sufficient for various physical events occurring today, despite the fact that those early events were only able to bring about their present-day effects via very long chains of intermediate effects. What may be leading our potential objector astray is the fact that, in Figure 3.1, M is represented as occurring at the same time as P11 and P12 , and we normally suppose that simultaneous events cannot be related as cause and effect. No such objection would, I suggest, seem even superficially plausible if M had been represented as occurring at a time intermediate between t1 and t2 , and thus later than P11 and P12 . But, clearly, such an alteration to M’s time of occurrence makes no difference to the causal structure of the situation depicted in Figure 3.1. Another objection that may be raised to something I said earlier is the following. I said that, as I was interpreting the situation depicted in Figure 3.1, the following is true: (1) if M had not occurred, then the conjunction of P11 and P12 , even if it had occurred, would not have sufficed to cause P. To this it may be objected that if, indeed, P12 is a cause of M, then, since causes are impossible without their effects, the following is in fact true: (2) if M had not occurred, then P12 , and hence also the conjunction of P11 and P12 , would not have occurred. However, (1) and (2) are not incompatible according to any plausible account of the semantics of counterfactual conditionals. We can affirm the joint truth of counterfactuals of these two forms, without contradicting ourselves. Compare: ‘If the floor had not been wet, then the roof would not have
74
Mental Causation, Closure, and Dualism
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.
3.5. The Invisibility of Mental Causation An interesting feature of the situation depicted in Figure 3.1 is this. Any scientist who was to examine that situation by empirical means, but who was restricted by his means of investigation to observing only purely physical events and causal relationships, would quite naturally come to the conclusion that the physical event P had a complete and wholly physical causal explanation, in terms of its immediate physical causes, P11 and P12 , and their antecedent physical causes, beginning with P01 and P02 . Such an investigator would notice no ‘gaps’ in the physical causation of P, and no lack of a sufficient and wholly physical cause of P at any time prior to P’s time of occurrence, t2 . For instance, if the investigator were a neuroscientist and all of the physical events concerned were brain events, such as neuron-firings, he might very reasonably form the impression that event P was entirely explicable in terms of such brain events. And yet, if the situation were really as Figure 3.1 depicts it, he would be mistaken in thinking this, because the complete causal explanation of P’s occurrence in fact involves the putatively non-physical mental event M. The existence and causal role of this mental event would be, as it were, ‘invisible’ to him, owing to the causal structure of the situation and his limited means of investigation—a point that we have touched on before, in Chapter 1. Not only would he not be able to detect M’s presence and its causal contribution directly, by somehow observing them; he would not even be able to detect M’s presence and causal contribution indirectly, by discovering the causal explanation of P in wholly physical terms to be incomplete. The causal explanation of P in wholly physical terms would in fact be incomplete, of course, but it would not appear to be incomplete, given the causal structure of the situation and his restricted means of empirical investigation.
Physical Closure and Mental Causation 75 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.
3.6. Objections, Replies, and Some Philosophical Lessons to be Learnt Here certain stock objections are likely to be raised by physicalists. It may be objected, for instance, that it is much more ‘economical’ to suppose that M is either identical with or realized by one or more brain events than to suppose that it is ontologically additional to all such events and plays a separate causal role in helping to bring about P. Well, certainly it is
76
Mental Causation, Closure, and Dualism
more economical to suppose this, if ‘economy’ is just a matter of how many ontologically independent events are invoked in the causal explanation of P. But, given that the situation depicted in Figure 3.1 is metaphysically possible, which I am convinced that it is, we can’t simply rule out that situation’s being actual on the grounds of such ‘economy’. To do so would render us vulnerable to a charge of being ‘economical’ with the truth, no less, by advocating an explanation that we know may well be less than completely true. I shall return to this issue in a moment. Again, it may be objected that the postulation of M as an ontologically independent non-physical event would be ad hoc and unprincipled. Well, I agree that it would be ad hoc and unprincipled to postulate ‘additional’ non-physical events to accompany physical events quite generally, even in the case of situations in which we have no testimony confirming the occurrence of mental events or anything like them. This may be a fair objection to raise against some forms of ‘panpsychism’. But given that we do, where human brains are concerned, have reliable testimony confirming the occurrence of mental events, which at least seem to be neither identical with nor ‘realized’ by brain events, there need be nothing ad hoc and unprincipled about postulating that these events are precisely what they seem to be, namely, ontologically ‘additional’ non-physical events. Yet another stock objection is, however, that postulating such events commits us to something deeply ‘mysterious’ and inexplicably at odds with what we have discovered about biological evolution, for it implies that at some stage in the course of evolution a wholly new kind of event suddenly sprang into existence in a way that is inconsistent with the gradualism characteristic of biological evolution in general. To this it may be replied, first, that it is widely disputed whether all biological evolution is in fact gradual in character and, second, that since, ex hypothesi, non-physical mental events are not biological events, there is no reason to expect their historical provenance to be governed by principles of evolutionary biology. Let me return, briefly, to the issue about ‘economy’. Contemplating a dualist proposal very similar to mine, Andrew Melnyk objects to it precisely as being ‘less economical’ than one which omits dualistically conceived mental events.⁹ My basic complaint against this sort of objection 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
and neuropsychologists to provide any grounds whatever for deeming interactive dualism to be false. The sort of empirical evidence that such scientists dispose of is simply incapable of discriminating between certain wholly physicalist accounts of mental causation and at least some interactive dualist accounts of it. On the other hand, this is not to say that there are no conceivable circumstances at all in which we could discover compelling empirical evidence in favour of some form of interactive dualism—which is just as well for dualism since, if that were the case, it would be a just ground for suspicion that the doctrine is vacuous. If, for instance, we had evidence that a human being could reliably recall veridical perceptual experiences which could only have occurred to him at a time at which his brain exhibited no electrical activity whatsoever (when, in short, he was ‘brain-dead’), this would surely constitute compelling evidence that at least some mental events are neither identical with nor ‘realized by’ brain events. But we have no guarantee that every world in which interactive dualism is true will be a world in which evidence like this for its truth would in principle be available. If interactive dualism is true in the actual world but mental causation always operates in the ‘invisible’ way examined earlier, then the actual world will turn out to be one in which, although interactive dualism is true, compelling empirical evidence of its truth is inevitably unavailable. This needn’t be taken to imply, however, that, in such a world, it would be altogether impossible to discover compelling reasons to believe in the truth of interactive dualism, just that those reasons could not be broadly empirical ones. The lesson of this is that we should be prepared to acknowledge that a priori metaphysical argumentation might in the end provide the best, or indeed the only, hope for a resolution of the mind–body problem. Failing that, we might simply have to learn to live with the possibility that we inhabit a world in which the resolution of that problem is impossible for beings with our cognitive capacities.¹⁰ 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? There are empirical neuropsychological data which suggest to some philosophers that volitions are epiphenomenal —that when we think that we are, by exercising our will, initiating a train of events which will issue in some voluntary bodily movement, in reality that movement is initiated by unconscious cerebral events occurring prior to our conscious experience of ‘willing’ and that the conscious experience is merely a causally inefficacious side effect of those prior cerebral events. This view seems to conflict deeply with our pre-theoretical conception of ourselves as free agents. In this chapter, I shall discuss whether we need to revise our conception of ourselves in the light of such evidence or whether, on the contrary, the evidence can and must be interpreted in a way which is consistent with our pre-theoretical self-conception. Perhaps unsurprisingly, I shall come down firmly in favour of the latter verdict.
4.1. Concerning Epiphenomenalism Epiphenomenalism is normally understood to be the doctrine that mental events and states are real but non-physical and are effects but not causes of physical events and states, nor even causes of other mental events and states—in other words, that they are real but causally inert events and states that are distinct from, and not even ‘realized by’, physical events and states. The main motivation for epiphenomenalism seems to be the desire to respect the intuitive plausibility of psychophysical dualism without bearing the cost allegedly incurred by interactive psychophysical dualism, this cost being a supposed conflict with the causal closure of the physical domain.
80
Mental Causation, Closure, and Dualism
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, epiphenomenalism 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 philosophy of mind. It is at least disputable whether sensory, cognitive, and affective events and states are by their very nature items that participate in causal relationships with each other and with non-mental events and states. But it is built into the very notion of a volition or act of will that it is something apt to occupy a causal role of a very specific kind, one which requires volitions to be capable of having physical effects—at least, on the assumption that a physical world exists at all and that idealism is not true.
4.2. Volitions and Volitionism Now, regarding volitions much has been said, by both philosophers and psychologists, much of it of a rather sceptical character. Many contemporary philosophers of mind and action are sceptical about the very existence of volitions, for reasons which I do not think are particularly good.¹ Current philosophical opposition to volitionism is at least as much a product of prevailing fashion as it is of the cogency or persuasiveness of the arguments available against it. In the seventeenth and eighteenth centuries volitionism was as fashionable as it is now unfashionable. Surprisingly enough, however, there is much less scepticism about the existence of volitions amongst psychologists, a good many of whom exhibit no embarrassment whatever in announcing that they are engaged in empirical research on volitions. The confidence with which they assume that volitional events or states exist and are a suitable subject for empirical research will seem touchingly naive to most contemporary philosophers of mind and action, who have been brought up since their student days to dismiss the notion of ‘acts of will’ as an ill-founded philosophical fiction. However, while these psychologists assume unquestioningly the existence of volitions, they are apt to advance the view that ordinary folk are systematically deluded about the causal efficacy of their volitions. This seems to be part of a more general syndrome affecting some modern-day psychologists, which we also see exhibited in, for instance, psychological research on human reasoning. 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
Some psychologists seem to be peculiarly attracted to theories of mental functioning which imply that ordinary folk are systematically deluded about how their minds really work. Perhaps this is because such theories have shock value, which makes popularizing books about them more saleable to a gullible public. Or perhaps it is just because some psychologists are tired of the accusation that their research results only in complicated restatements of what is in any case obvious to common sense. But maybe I am being too cynical in attributing these motives. No matter: this is not really the place for me to speculate about the motivations of psychologists. It remains the case that a good many psychologists think that empirical research on volitions reveals that volitions do not play the sort of causal role that ordinary folk naively think they do. This is not to say that these psychologists consider that volitions are entirely epiphenomenal—that they have no effects whatever on physical or other mental phenomena. They could hardly believe that, given that most of them quite uncritically assume the truth of physicalism. Rather, what they often think that their research shows is that our volitions do not in fact play the causal role that is constitutively built into the very notion of a volition as a conative event. To put it simply, they do not think that our volitions cause our physical actions, such as our limb movements, when these are performed ‘voluntarily’. Roughly, they think that other physical events in the brain, of an unconscious character, cause both our volitions and our physical actions, so that rather than volitions being causes of physical actions they are just collateral effects of the real causes of physical actions. As such, volitions are fairly regular concomitants of physical actions, and this is supposed to account at least in part for our ‘illusion’ that they are actually causes of those actions—rather as children may be under the illusion that a flash of lightning causes a clap of thunder, when in fact both the seen flash and the heard bang are collateral effects of an electrical discharge.
4.3. Automatisms and Illusions of Control One sort of empirical evidence that such psychologists appeal to is the occurrence of automatisms and illusions of control.² Thus, on the one hand, 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 ³ See Wegner, The Illusion of Conscious Will, pp. 9–10. ⁴ See, for example, A. I. Melden, Free Action (London: Routledge & Kegan Paul, 1961), ch. 5. I criticize the argument in my Locke on Human Understanding (London: Routledge, 1995), pp. 124–6.
84
Mental Causation, Closure, and Dualism
indeed ‘come apart’ and that is why the former are eligible as causes of the latter.
4.4. Libet on the Unconscious Cerebral Initiation of Voluntary Actions Of course, it is not only automatisms and illusions of control that psychologists appeal to in order to shake our confidence in the causal efficacy of our volitions. Much has also been made of Benjamin Libet’s famous experiments which seemingly demonstrate the unconscious cerebral initiation of voluntary actions, because they appear to indicate that a ‘readiness potential’ starts to build up in the motor cortex a fraction of a second before the conscious experience of willing to perform a bodily movement, such as a flick of the wrist.⁵ Unsurprisingly, the experimental results and various interpretations of them have been hotly debated by both psychologists and philosophers. Libet himself holds that the results do not deny us ‘free will’, because he thinks that subjects retain a ‘veto power’ which enables them to prevent a willed action from going to completion, so that the action is not made inevitable by the initial unconscious cerebral process. However, the question of whether or not the will is ‘free’ in a libertarian sense is, of course, distinct from the question of whether volitions are causally efficacious. Even accepting Libet’s view about the ‘veto power’, it might still be urged that his experimental results indicate that our volitions at most disclose to us that an action of a certain kind is under way, rather than causing such an action. However, interesting though Libet’s results are, I doubt whether anything much can be made of them one way or another with regard to the question of the causal efficacy of the will. The experimental set-up inevitably makes the situation of the subjects concerned extremely artificial and abnormal. The subjects are instructed to perform a certain kind of bodily movement, such as a flick of the wrist, ‘spontaneously’ and without premeditation at arbitrary intervals within the course of the experiment, estimating when each of their conscious decisions to act occurred by noting the simultaneous 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: Birkh¨auser, 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.)
4.5. Can the Causal Efficacy of the Will Coherently Be Doubted? Rather than engage in further futile discussion about the relevance or irrelevance of experiments like Libet’s to the question of whether the will 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
is causally efficacious, I want now to engage in a more purely philosophical exercise. I want to inquire whether we can coherently doubt the causal efficacy of volition. Of course, as I have already indicated, I am perfectly happy to allow that volitions need not always give rise to their intended effects. This should be perfectly obvious in any case. One would have thought it evident that someone suddenly and unbeknownst to himself afflicted by a temporary paralysis of a limb could exercise his will to move the limb in a certain way and yet fail to move it. This is unlike cases of the illusion of control in which, typically, the subject is indeed performing some voluntary bodily action but is under the illusion that he is thereby bringing about some further event which in fact occurs independently of his action. In the paralysis case, all that the subject succeeds in doing is to exercise his will to perform a certain bodily movement and he is under no illusion that he has succeeded in doing so, because he can see or feel that the intended bodily movement has not occurred. There is another kind of case, famously reported by William James, in which a subject is not paralysed but is blindfolded and lacks any kinaesthetic sense of the movement of the limb concerned.⁷ This sort of subject may have the limb restrained, unbeknownst to himself, and exercise his will to move it, not realizing that he has failed to do so. In this sort of case too, then, we have the occurrence of a volition which fails to cause its intended effect. The question, then, is not whether we can coherently suppose any particular volition to be causally inefficacious, but whether we can coherently suppose all of our volitions to be causally inefficacious. Could the causal efficacy of the will quite generally be a cognitive illusion? Perhaps this is indeed possible, as an abstract metaphysical possibility. Perhaps a malicious demon, not unlike Descartes’s, could delude us into thinking that our will was causally efficacious, at least where our physical actions are concerned. Mental actions, such as deliberating and reasoning, may prove more problematic for reasons that we shall go into in due course. Certainly, it seems clear that an individual human being’s will may be completely causally inefficacious where physical actions are concerned, as tragic cases of so-called ‘locked-in syndrome’ demonstrate. Subjects suffering from this condition have lost all voluntary control over their bodily movements, although their autonomic nervous systems still function 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.
4.6. How is Causal knowledge of the Physical World Possible? Since our question is an epistemic one, it is pertinent to bring epistemic considerations to bear on our answer to it. One such consideration is this: we take ourselves to have some causal knowledge of the physical world. Indeed, it is a presupposition of natural science that such knowledge is available to us. Sometimes, at least, we can know or discover that something physical caused something else physical. But how can we ever know that? Causal knowledge of the physical world is presumably not a priori and presumably not innate. Causal knowledge of this kind, we think, arises from observation and experimentation. But, very arguably, it cannot arise always and only from mere observation. Experimentation is also necessary for the genesis of such causal knowledge.⁸ Observation can reveal to us that physical events of one kind are regularly followed by physical events of another kind but, contrary to Hume—or, at least, contrary to some interpretations of Hume—that events of one kind are regularly followed by events of another kind is not what causation between events consists in. In order to determine whether or not physical events of certain kinds are 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
causally or merely accidentally related, we need to conduct experiments. We need, for instance, to see whether we can prevent an event of a certain kind from occurring when an event of another kind, which normally precedes it, occurs. To do this, we need to exercise voluntary control over our bodies. If we were intelligent and sentient plants—trees with eyes and ears but no ability to control at will the disposition of our limbs—we could never engage in experimentation and never acquire causal knowledge of the physical world. Since I think that causal knowledge of the physical world is necessary for any physically embodied intelligent being, that means that I think that there could not in fact be such things as intelligent plants. What do these considerations serve to show? It may well be doubted whether they show that our volitions, or at least some of them, must actually be causally efficacious in bringing about their intended physical effects. But at least they seem to show that we must believe that they have this sort of causal efficacy if we are to engage in experimentation. If they show even that, then they seem to show by implication something else, namely, that we cannot acquire this belief in the causal efficacy of volitions in just the same way that we acquire beliefs about causal relationships between physical events, that is, by means of observation and experimentation. Thus they show that Hume’s account of how it is that we come to believe in the causal efficacy of volitions is mistaken. He considered that we acquire this causal belief in the same way that, according to him, we acquire any other causal belief: we observe a ‘constant conjunction’ between the occurrence of our volitions and the occurrence of their intended effects, and thence conclude that the former are causes of the latter. As he puts it: ‘We learn the influence of our will from experience alone. And experience only teaches us, how one event constantly follows another.’⁹ Even some modern psychologists think that something like this is the case.¹⁰ But it surely cannot be. It cannot be that I acquire the belief that a volition to raise my arm causes the rising of my arm because I repeatedly observe that whenever I have such a volition, it is followed by a rising of my arm. 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 When I have such a volition—when I will to raise my arm—I intend to cause my arm to rise by means of that very act of will.¹¹ It is part of the very intentional content of such a volition that this is my aim. But I cannot intend to do something unless I believe that it is possible for me to do that thing. (I cannot, for instance, intend to jump over a skyscraper, knowing that this is something naturally impossible for me to achieve.) So I cannot intend to cause my arm to rise by exercising my will unless I believe that a volition of mine to raise my arm can cause my arm to rise—can not just in the sense in which this expresses mere logical possibility, but rather in the sense in which it expresses real or natural possibility. Thus I must already believe in the causal efficacy of my will in order to exercise it, and so cannot acquire a belief in this efficacy by discovering a ‘constant conjunction’ between exercises of my will and certain physical events.
4.7. The Incoherence of Full-Blown Epiphenomenalism What we seem to be able to conclude so far is that a belief in the causal efficacy of the will is necessary for the very exercise of the will and hence that belief in the existence of causal relationships between physical events, which depends upon experimentation and hence on the exercise of our wills, is acquired in a very different way from belief in the causal efficacy of the will. The latter might even be innate in character. But as yet we still cannot rationally dismiss the possibility that this belief in the causal efficacy of the will, necessary though it may be for practical purposes and for the acquisition of scientific knowledge about the causal structure of the physical world, might turn out to be a cognitive illusion. This is because we have so far been focusing on physical actions, such as voluntary movements of parts of our bodies. But mental actions may also be voluntary, and at least some of them must be voluntary if we are to be rational beings. Thinking, reasoning, and imagining are things that we can and must be able to do voluntarily, if we are rational beings. This is not to deny that thoughts and images often come unbidden into our minds and that we sometimes engage in trains of unconscious reasoning quite independently of our wills. ¹¹ See, further, my Subjects of Experience, pp. 146ff.
90
Mental Causation, Closure, and Dualism
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 control’. Even if we could in principle be subject to such an illusion where our physical actions are concerned, however, we could not—or, at least, we cannot rationally believe that we could—be subject to it where our thoughts and reasonings are concerned. The implication seems to be that full-blown epiphenomenalism—the doctrine that mental events and states are never causally efficacious—is an incoherent doctrine, in the sense that it is one which we could not rationally believe to be true in our own case.
5 The Self as an Emergent Substance Substance dualism in the philosophy of mind is, naturally enough, commonly thought of on a Cartesian model, according to which it is a dualism of two radically different kinds of substance, one (the ‘body’) purely material and the other (the ‘mind’) wholly immaterial in nature. This view is subject to many familiar difficulties.¹ However, the almost universal rejection of Cartesian substance dualism has blinded many philosophers to the possibility of formulating other and more plausible versions of substance dualism. Non-Cartesian substance dualism (NCSD), as it may most perspicuously be called, is a dualism not of minds and bodies, but of persons—or, more generally, of subjects of experience —and their ‘organized’ bodies. This is an ontological distinction that is chiefly motivated not by some fanciful notion that there could be disembodied persons, but by much more solid considerations which require us to distinguish between the identity-conditions of persons and their bodies and also between the causal powers of persons and those of their bodies. Much of the intuitive appeal of Cartesian dualism is retained and explained by NCSD, without any of the former’s counterintuitive features and metaphysical difficulties. NCSD is, however, still a non-materialist position, because it is incompatible even with very weak forms of non-reductive physicalism. It may also be fairly described as an emergentist position, in that it regards the causal powers of persons as complementing and supplementing —rather than either being reducible to or existing entirely independently of—those of their bodies. In what follows, I shall begin by explaining and justifying NCSD’s distinctive ontology of persons, 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 before moving on to present and argue for its distinctively anti-physicalist account of the metaphysics of mental causation.
5.1. Non-Cartesian Substance Dualism Defined Dualism in the philosophy of mind is customarily divided into two chief kinds: substance dualism and property dualism, the former maintaining the distinctness of mental and physical substances and the latter maintaining the distinctness of mental and physical properties. But what are we supposed to understand by a mental or physical ‘substance’ in this context? I shall take it that by a substance, here, we should simply mean an individual object, or bearer of properties—where these properties include various distinctive and independent causal powers, since a ‘substance’ devoid of all independent causal powers would be a mere shadow of a substance rather than a genuine substance in its own right. I shall further take it that by a mental substance we should mean a bearer of mental or psychological properties, and that by a physical substance we should mean a bearer of physical properties. Thus, whereas the property dualist holds that mental and physical properties are distinct, the substance dualist additionally holds that certain bearers of those properties are distinct—the implication being that substance dualism entails property dualism but not vice versa. I assume, incidentally, that both kinds of dualism entail dualism with regard to mental and physical states and events, since I take these to consist in the exemplification of properties by objects at times.² All this being so, what is needed at this point is a defensible account of the two key concepts of a mental property and a physical property. These, it seems clear, are distinct concepts, although whether the properties of which they are concepts are themselves distinct is, of course, one of the main issues under dispute. However, it is one thing to say that these concepts are distinct and quite another to provide an account of that conceptual distinction that would satisfy everyone. In fact, it has proved remarkably difficult to produce an uncontentious characterization of either concept.³ 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
Fortunately, it is much easier to provide paradigm examples of mental and physical properties which almost all parties to the debate will be happy to accept as such. For instance, pain and desire are universally recognized as being mental properties, while mass and velocity are universally recognized as being physical properties. In what follows, therefore, I shall take it for granted that the conceptual distinction now at issue is a genuine one and that for practical purposes it can be captured by appeal to such paradigm examples. Now, substance dualists contend that certain bearers of mental properties, such as pain and desire, are distinct from—that is, are not to be identified with—certain bearers of physical properties, such as mass and velocity. What are these ‘bearers’, though? The bearers of mental properties may be called, quite generally, subjects of experience —understanding ‘experience’ here in a broad sense, to include not just sensory and perceptual experience, but also introspective and cognitive states or, in other words, ‘inner’ awareness and thoughts.⁴ Human persons—we ourselves—provide prime examples of subjects of experience, but no doubt we should also include examples drawn from the ‘higher’ reaches of the non-human animal domain. As for the bearers of physical properties, for the purposes of the present discussion I shall mostly be referring to bodies, or parts of bodies—on the understanding that what we are talking about here are not mere lumps or masses of matter, but organized bodies and their parts, the paradigm examples being the human body and its organic parts, such as the brain and the neurons and other kinds of cell making up the brain and central nervous system. In these terms, then, the substance dualist may be construed as holding that a person is not to be identified with his or her body, nor with any part of it, such as the brain. On this view, a person—not the person’s body or brain—feels pain and has desires, even if it is true to say that a person feels pain or has desires only because his or her body or brain is in a certain physical state. The physical state in question—a certain pattern of excitation in nerve cells, say—is not to be identified with the pain or desire consequently experienced by the person, according to the substance dualist. It is at this point that I want to introduce a key distinction between two different types of substance dualism. An implication of what I have said so ⁴ See, further, my Subjects of Experience (Cambridge: Cambridge University Press, 1996), ch. 1.
The Self as an Emergent Substance 95 far concerning substance dualism might seem to be that, according to it, a bearer of mental properties—a subject of experience—only bears mental properties, whereas as a bearer of physical properties, such as a human body or brain, only bears physical properties. This was indeed the view of the most famous substance dualist of all, Ren´e Descartes, for whom the human self or ego is an immaterial substance.⁵ However, even if I am distinct from—not to be identified with—my body or any part of it, as Descartes held, it does not automatically follow that I can have only mental, not physical properties. And, indeed, there is a modern form of substance dualism—which may be called, aptly enough, non-Cartesian substance dualism—which differs from Cartesian substance dualism precisely over this point. According to NCSD, it is I, and not my body nor any part of it, who am the bearer of mental properties, just as Descartes maintained. However, unlike Descartes, the advocate of NCSD does not make the further claim that I am not the bearer of any physical properties whatsoever. This sort of substance dualist may maintain that I possess certain physical properties in virtue of possessing a body that possesses those properties: that, for instance, I have a certain shape and size for this reason, and that for this reason I have a certain velocity when my body moves.⁶ It doesn’t follow that such a substance dualist should allow that every physical property possessed by my body is also possessed by me, however, for the possession of some of these properties may entail that the thing possessing them is a body—and the advocate of NCSD wants to deny, of course, that I am a body. One such property, for instance, would appear to be the property of being wholly composed of bodily parts, which is possessed by my body but presumably not by me.
5.2. The Unity Argument for NCSD Although I think that there are several good arguments in favour of NCSD’s key claim that a human person is not identical with his or her ‘organized ⁵ See Ren´e Descartes, Meditations on First Philosophy, in J. Cottingham, R. Stoothoff, and D. Murdoch (eds), The Philosophical Writings of Descartes (Cambridge: Cambridge University Press, 1984) and, for prominent modern sympathizers, Richard Swinburne, The Evolution of the Soul (Oxford: Clarendon Press, 1986) and John Foster, The Immaterial Self: A Defence of the Cartesian Dualist Conception of the Mind (London and New York: Routledge, 1991). ⁶ Compare my Subjects of Experience, ch. 2, and also Lynne Rudder Baker, Persons and Bodies: A Constitution View (Cambridge: Cambridge University Press, 2000).
96
Mental Causation, Closure, and Dualism
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 ⁹ See, further, P. F. Strawson, Individuals: An Essay in Descriptive Metaphysics (London: Methuen, 1959), ch. 3, and my Subjects of Experience, ch. 2.
98
Mental Causation, Closure, and Dualism
concede that this may very well be true of my brain as a whole—that if it were to be completely destroyed, all of my mental states would thereby cease to be. All that I am claiming is that there is no part of my brain which is such that, were any part of it —such as one particular neuron—to be destroyed, all of my mental states would thereby cease to be. That is to say, neither my brain as a whole, nor any distinguished part of it as a whole, is something with which I can be identified—any more than I can be identified with my body as a whole—because no such entity is such that all and only my mental states can be taken to depend on it, in the way that they clearly do depend on me. However, here it may be objected that the foregoing defence of premise (2) depends upon an illicit assumption, namely, that if my body as a whole were to lack a certain part, such as the tip of one of my little fingers, then it —my body as a whole—would not exist. This assumption, it may be said, is unwarranted because it presupposes, questionably, that every part of my body is an essential part of it, without which it could not exist. As it stands, this may be a fair objection—although it should be acknowledged there are some philosophers who do hold that every part of a composite object is essential to it.¹⁰ However, I think that the reasoning in favour of premise (2) can in fact be formulated slightly differently, so as to make it independent of the truth of this assumption. The initial insight still seems to be perfectly correct—that, as I put it, 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. Thus, to repeat, the thoughts that I am having in composing this chapter plausibly do not depend upon my body including as a part the tip of one of my little fingers. Call these thoughts T. Consider, then, that object which consists of my body as a whole minus that finger tip. Call this object O and call my body as a whole B. (It should be conceded here that there are some philosophers who would deny that any such object as O exists¹¹—but that is, to say the least, a controversial claim.) Suppose, now, that it is proposed that I am identical with B, and hence that B is the subject of the thoughts T. Then we can ask: on what grounds can B be regarded as the subject of T in preference to O, given that T do not depend upon 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.
5.3. The Causal Closure Argument against Interactive Dualism So far, I have only argued that NCSD is correct in claiming that a self or person is never to be identified with his or her organized body. But in order to qualify as a genuine substance in its own right, a self or person must, in my view, possess some distinctive and independent causal powers—otherwise, it would be a mere ‘shadow’ of a substance. So now I want to explore certain causal considerations that inevitably arise in the debate between dualism and its opponents. For dualism—whether we are talking about substance dualism or property dualism—is traditionally divisible into interactionist, epiphenomenalist, and parallelist varieties. Perhaps the most powerful argument against interactive dualism is the so-called causal closure argument. (We have, of course, looked at various versions of this argument in earlier chapters, so I can afford to be brief in my exposition of it here.) By interactive dualism, I mean the doctrine that mental events or states are not only distinct from physical events or states, but are also included amongst the causes and effects of physical events or states. Of course, the causal closure argument can have no force against either epiphenomenalist or non-interactive parallelist
100
Mental Causation, Closure, and Dualism
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. ¹² See William S. Robinson, Understanding Phenomenal Consciousness (Cambridge: Cambridge University Press, 2004) for an interesting recent defence of epiphenomenalism. However, I have already made it plain in Chapter 4 why I cannot regard epiphenomenalism—at least ‘full-blown’ epiphenomenalism—as tenable. ¹³ Compare Jaegwon Kim, ‘The Non-Reductivist’s Troubles with Mental Causation’, in J. Heil and A. Mele (eds), Mental Causation (Oxford: Clarendon Press, 1993).
The Self as an Emergent Substance 101 Let us now consider a version of the causal closure argument against interactive dualism that appeals to the very strong formulation of the causal closure principle embodied in premise (4)—that no chain of event-causation can lead backwards from a purely physical effect to antecedent causes some of which are non-physical in character. Two additional premises are needed. First, (5) Some purely physical effects have mental causes, which the interactive dualist accepts as true, of course. Second, (6) Any cause of a purely physical effect must belong to a chain of event-causation that leads backwards from that effect. These three premises entail the conclusion (7) All of the mental causes of purely physical effects are themselves physical in character, which contradicts the defining thesis of interactive dualism. My defence of interactive dualism will rest upon a challenge to premise (6). Moreover, it will endorse a version of interactive dualism which combines it with the sort of non-Cartesian substance dualism defended earlier.
5.4. Two Different Perspectives on the Causal Explanation of Voluntary Action In order to keep matters relatively simple and to confine my discussion to manageable proportions, I shall concentrate, in what follows, on issues concerning voluntary and deliberative human action, where it is most obviously pressing that some coherent story needs to be told as to how mental and neurophysiological causes interrelate with one another. So let us focus on a specific case of such an action, such as an agent’s deliberate—that is, premeditated and entirely voluntary—raising of an arm, for whatever reason: for instance, in order to catch a lecturer’s attention with a view to asking a question. Now, it seems relatively uncontroversial that if, in such a case, we were to trace the purely bodily causes of the upward movement of the agent’s arm backwards in time indefinitely far, we would find that those causes ramify, like the branches of a tree, into a complex maze of
102
Mental Causation, Closure, and Dualism
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 ¹⁴ See, for example, L. Deecke, P. Scheid and H. H. Kornhuber, ‘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, and K. R. Popper and J. C. Eccles, The Self and its Brain: An Argument for Interactionism (Berlin: Springer, 1977), pp. 282–94.
The Self as an Emergent Substance 103 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.
5.5. A Counterfactual-based Argument against Psychoneural Identity Theories This last point is an extremely important one and requires further elucidation. It is now standard practice amongst philosophers of logic and language to interpret counterfactual conditionals in terms of possible worlds, very roughly as follows.¹⁵ A counterfactual of the form ‘If it were the case that p, then it would be the case that q’ is said to be true if and only if, in the closest possible world in which p is the case, q is also the case—where the ‘closest’ possible world in question is the one in which p is the case but otherwise differs minimally from the actual world. Now, suppose that a physicalist in the philosophy of mind were to propose that the agent’s decision, D, to raise his or her arm on a given occasion—the agent’s mental act of choice—is identical with a certain neural event, N, which is correctly identifiable as being a cause of the subsequent bodily event, B, of the arm’s rising. Here I must stress that D, N, and B are, each of them, supposed to ¹⁵ See, especially, David K. Lewis, Counterfactuals (Oxford: Blackwell, 1973)—although I do not replicate every detail of his account, but only those that are germane to the issues now under discussion. Moreover, although I myself have doubts about his account—see my ‘The Truth About Counterfactuals’, Philosophical Quarterly 45 (1995), pp. 41–59—those doubts have no bearing on the issues being discussed here.
104
Mental Causation, Closure, and Dualism
be particular events, each occurring at a particular moment of time, with B occurring at least an appreciable fraction of a second later than D and N, since our decisions to act do not take effect immediately. The physicalist must furthermore suppose, of course, that D and N occur at the same time, since he holds them to be identical. (I should remark, incidentally, that I do not need to get embroiled here in the debates concerning Benjamin Libet’s celebrated but highly controversial experiments on the precise timing of volitions, as these are not relevant to the sort of case that we are now considering.¹⁶) Let us concede, consequently, that the following counterfactual is true: ‘If N had not occurred, then B would not have occurred.’ All that I am presupposing here is that if N was indeed a cause of B, then the foregoing counterfactual is true. The physicalist cannot, I think, have any quarrel with me on this account. I am not taking any advantage, then, of the various reasons that have been advanced for doubting, at least in some cases, whether causal statements entail the corresponding counterfactuals.¹⁷ What I am now interested in focusing on is the following question: what sort of event would have occurred, instead of B, if N had not occurred? In other words: in the closest possible world in which N does not occur, what sort of event occurs instead of B? My contention is that what occurs in this world is an event of the same sort as B, differing from B only very slightly. The reason for this is as follows. It seems evident, from what we know about the neural causes of an event such as B, that N must be an immensely complex neural event: it must be, in fact, the sum (or ‘fusion’) of a very large number of individual neural events, each of them consisting in some particular neuron’s firing in a particular way. Recall, here, that N must be supposed to occur an appreciable amount of time before B, at a time at which the neural antecedents of B are many and quite widely distributed across the agent’s cerebral cortex. It would be utterly implausible for the physicalist to maintain, for example, that the agent’s decision D is identical with the firing of just a single neuron, or even 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 of a small number of neurons. If D is identical with any neural event at all, it can surely only be identical with an extremely complex one, consisting in the firing of many neurons distributed over quite a large region of the agent’s cerebral cortex. However, it seems indisputable that if N is, thus, the sum of a very large number of individual neural events, then the closest world in which N itself does not occur is a world in which another highly complex neural event, N ∗ , occurs, differing only very slightly from N in respect of the individual neural events of which it is the sum. In other words, N ∗ will consist of almost exactly the same individual neural events as N, plus or minus one or two. Any possible world in which a neural event occurs that differs from N in more than this minimal way simply will not qualify as the closest possible world in which N does not occur. This is evidently what the standard semantics for counterfactuals requires us to say in this case. But, given what we know about the functioning of the brain and nervous system, it seems clear that, in the possible world in which N ∗ occurs, it causes a bodily event very similar to B, because such a small difference between N and N ∗ in respect of the individual neural events of which they are respectively the sums cannot be expected to make a very big difference between their bodily effects. There is, we know, a good deal of redundancy in the functioning of neural systems, so that the failure to fire of one or two motor neurons, or the abnormal firing of one or two others, will typically make at most only a minimal difference with regard to the peripheral bodily behaviour that ensues. Thus, the answer to the question posed earlier—what sort of bodily event would have occurred instead of B, if N had not occurred?—is this: a bodily event very similar to B. In other words, if N had not occurred, the agent’s arm would still have risen in almost exactly the same way as it actually did. Now, I hope, we can see the importance of this conclusion. For, if we ask what sort of bodily event would have occurred instead of B if the agent’s decision, D, to raise his or her arm had not occurred, then we plausibly get a very different answer. Very plausibly, if D had not occurred—if the agent had not made the very act of choice that he or she did to raise the arm—then the arm would not have risen at all. It is, I suggest, quite incredible to suppose that if the agent had not made that very decision, D, then he or she would have made another decision virtually indistinguishable from D—in other words, another decision to raise the arm in the same, or virtually the same, way. On the contrary, if the agent had not made that
106
Mental Causation, Closure, and Dualism
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 ¹⁸ I am grateful to Jos´e Berm´udez for pressing this line of response.
The Self as an Emergent Substance 107 possess sufficient voluntary control over my limb-movements to be able to decide to execute them with such precision. If all of this reasoning is correct, then it follows unavoidably that the decision D cannot be identical with the neural event N with which the physicalist proposes to identify it, for the counterfactual implications of the non-occurrence of these two events are quite different. If D had not occurred, the agent’s arm would not have risen at all, but if N had not occurred, it would have risen almost exactly as it did. One fundamental reason for this—according to the conception of human persons that I favour as an advocate of NCSD—is that a mental act of choice or decision is, in a strong sense, a singular and unitary event, unlike a highly complex sum or fusion of independent neural events, such as N. And N ∗ differs from N only in excluding one or two of the individual neural events composing N or including one or two others. That is why N and N ∗ can be so similar and thus have such similar effects. But D, I suggest, cannot intelligibly be thought of, in like manner, as being composed of myriads of little events—and that is at least partly why, in the closest possible world in which D itself does not occur, there does not occur another decision D∗ which differs from D as little as N ∗ differs from N. Note that this strong unity of our mental acts, whereby they resist decomposition into lesser parts, nicely parallels our own strong unity as ‘simple’ substances, revealed by the unity argument of Section 5.2.
5.6. Extending the Argument to ‘Realization’ Accounts It is important to observe that the foregoing line of argument sustains not only the conclusion that the mental and neural causes of voluntary bodily movements must be numerically distinct, but also the stronger conclusion that those mental causes cannot even be held to be realized by any of those neural causes—where ‘realization’ is taken to be a metaphysical relation, distinct from identity itself, in virtue of which ‘realized’ events or states supposedly ‘inherit’ all of their causal features (all of their ‘causal powers’, as it is commonly put) entirely from the events or states that ‘realize’ them.¹⁹ 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 our general line of argument may be extended to rebut realization accounts even without adopting this particular assumption. The crucial question that we need to ask in this context is the following. What, according to the realization theorist, is the character of the neural event N ∗ that occurs instead of N in the closest possible world in which no realizer of D occurs—in other words, in the closest possible world in which (according to such a theorist) D itself does not occur? Clearly, it must be an event that differs from N just enough not to qualify as a realizer of D—in other words, an event that differs minimally from N, consistently with its not being similar enough to N to qualify (according to the realization theorist) as an alternative possible realizer of D. However, this event N ∗ will surely still be similar enough to N to cause some kind of activity in the agent’s arm—even though not an arm-rising of the kind that the agent actually intended—as opposed to activity quite elsewhere in the agent’s body or no bodily activity at all. (Thus, N ∗ will presumably be like N at least in occurring in a region of the cortex associated with arm-movement, and indeed may well share with N many of the individual neuronal firings that are its parts.) For to suppose that N ∗ differs more than this much from N is surely not consistent with regarding N ∗ as the closest possible non-realizer of D. But then this implication of the realization approach conflicts, once more, with the very plausible claim that if D had not occurred, no arm activity at all would have occurred—that, rather, the agent would either have moved in a quite different way or else have remained still. However, it must be confessed that, although our original argument can plausibly be extended in this fashion to embrace realization accounts of psychoneural relations, the case against identity accounts is inevitably somewhat stronger for the very simple reason that realization theories tend to be considerably more obscure, as it is not always entirely clear what their adherents intend us to understand by the term ‘realization’. The most that I can claim, therefore, is that on any plausible construal of the significance of that term, the argument can be successfully extended along the foregoing lines.²¹ 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
5.7. Intentional Causation versus Physical Causation So far, I have tried to explain why the mental and neural causes of voluntary bodily movements must be distinct, consistently with allowing, as I do, that such movements have both mental and neural causes. Now I want to say a little more about the respects in which mental causation is distinctively different from bodily or physical causation. Most importantly, then, mental causation is intentional causation—it is the causation of an intended effect of a certain kind. Bodily causation is not like this. All physical causation is ‘blind’, in the sense that physical causes are not ‘directed towards’ their effects in the way that mental causes are. Both sorts of causation need to be invoked, I believe, in order to give a full explanation of human action and NCSD’s conception of human persons seems well-equipped to accommodate this fact. The very logic of intentional causation differs, I venture to say, from the logic of bodily causation.²² Intentional causation is fact causation, while bodily causation is event causation.²³ That is to say, a choice or decision to move one’s body in a certain way is causally responsible for the fact that a bodily movement of a certain kind occurs, whereas a neural event, or set of neural events, is causally responsible for a particular bodily movement, which is a particular event. The decision, unlike the neural event, doesn’t causally explain why that particular bodily movement occurs, not least because one cannot intend to bring about what one cannot voluntarily control —for, as I pointed out earlier, one cannot voluntarily control the precise bodily movement that occurs when one decides, say, to raise one’s arm. As I have just implied, the two species of causal explanation, mental and physical, are both required and are mutually complementary, for the following reason. Merely to know why a particular event of a certain kind
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.
5.8. Context-Dependency to the Rescue? Might it not be objected here, on the part of the physicalist, that I am illicitly assuming that there is only one way of determining what counts as the ‘closest’ possible world in which M does not occur, namely, by focusing purely on M’s neural characteristics—in particular, the set of individual neural firings that compose it? Perhaps, indeed, we must say that the complex neural event M ∗ will occur instead of M in that world when we compare worlds for closeness purely in neural terms. But why shouldn’t the physicalist say that, when we select the ‘closest’ possible world in which M
The Self as an Emergent Substance 113 does not occur by focusing, instead, on M’s psychological characteristics—in particular, the intentional content that it supposedly has in virtue of being identical with (or else ‘realizing’) the mental decision D—then we shall by no means be compelled to judge that it is M ∗ that will occur instead of M in that world. Well, I am happy to concede that, when it comes to evaluating counterfactual conditionals, different criteria of ‘closeness’ between possible worlds can sometimes come into play in different conversational contexts.²⁴ In principle, then, the context-dependency of counterfactuals is a consideration that the physicalist might be able to appeal to in order to protect his position against my argument. But it would be a mistake to suppose that no work needs to be done by the physicalist to earn the right to appeal to this consideration in the case in hand. After all, if such work never had to be done by a philosopher who wanted to use this consideration in order to evade the conclusion of a counterfactual-based argument, then the force of such arguments would entirely evaporate and philosophy would be deprived of a method of argumentation that is not only very common but generally regarded as worthy of respect. In short, for the physicalist merely to appeal, without more ado, to the context-dependency of counterfactuals as a reason for ignoring my argument would be a thoroughly ad hoc and unprincipled way of trying to protect his position. After all, if such a response was all that was needed, then it could be deployed by any kind of identity theorist, no matter how wild his or her theory might be. For example, someone who maintained that the mental decision D is identical with the emission E of an electron in the agent’s brain could employ this strategy to circumvent the seemingly compelling objection that the ‘closest’ world in which E does not occur is—unlike the ‘closest’ world in which D does not occur—a world in which the agent’s behaviour is almost exactly the same as in the actual world. Such a theorist could simply reply that when we select the ‘closest’ such world by focusing on the psychological characteristics that E allegedly has in virtue of its identity with D, instead of its sub-atomic characteristics, we find that in such a world the agent’s behaviour is relevantly different. Well, of course it is, because the theorist has stipulated that E is identical with D, and we know what the ‘closest’ world in which D doesn’t occur is like, when this is selected on the basis of D’s 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 generally, 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 philosophers 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 ²⁶ I am grateful to John Hawthorne for suggesting the example.
116
Mental Causation, Closure, and Dualism
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.
5.9. Reasons, Causes, and Freedom of Action Much more can and should be said on all of these matters, but since I have discussed a good many of them extensively in previous chapters, I shall rest content with the foregoing remarks for present purposes. Here, however, it may be asked: But what about the causes of an agent’s acts of decision or choice? Are these bodily, or mental, or both? My own opinion is that an act of decision or choice is free, in the ‘libertarian’ sense—that is to say, it is uncaused. (I shall mount an extensive defence of this view in Part II, and so shall say little more about it here.) This is not to say that decisions are simply inexplicable, only that they demand explanations of a non-causal sort. Decisions are explicable in terms of reasons, not causes. That is to say, if we want to know why an agent decided to act as he did, we need to inquire into the reasons in the light of which he chose so to act.²⁷ Since decisions are, ²⁷ Compare Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2000).
The Self as an Emergent Substance 117 according to NCSD’s conception of the mind, attributable to the person and not to the person’s body or any part of it, there is no implication here that any bodily event is uncaused. It’s not that I want to exclude altogether the idea of causal explanation in terms of mental states in favour of purely rational explanation in the psychological sphere—as my earlier arguments make manifest. However, I do want to help to reinstate the idea that reason-giving explanation is not a species of causal explanation and that it is one form of explanation that is distinctive of the psychological sphere. But now it may be wondered: How is it really possible for mental acts of decision to explain anything in the physical domain, if that domain is causally closed, as many contemporary philosophers of mind—and just about all physicalists—assume? This takes us back to the earlier concerns of Section 5.3. As we observed there, much turns on precisely how the putative causal closure of the physical domain is to be defined, for this is no simple matter (as indeed we saw in Chapter 2). According to one popular view,²⁸ the thesis of physical causal closure amounts to the claim that no chain of event-causation can lead backwards from a purely physical effect to antecedent causes some of which are non-physical in character. This was premise (4) of the version of the causal closure argument presented in Section 5.3. But intentional causation according to NCSD’s conception of human persons, as I have tried to characterize it earlier in this chapter, does not violate the thesis of physical causal closure just stated, since it does not postulate that mental acts of decision or choice are events mediating between bodily events in chains of causation leading to purely physical effects: it does not postulate that there are ‘gaps’ in chains of physical causation that are ‘filled’ by mental events. Thus, NCSD’s model of mental causation is consistent with premise (4) of the causal closure argument and avoids the conclusion of that argument by repudiating, instead, premise (6). As we have seen, according to NCSD’s conception of human persons, a decision can explain the fact that a bodily movement of a certain kind occurred on a given occasion, but not the particular movement that occurred. Even so, it may be protested that if physical causation is deterministic, then there is really no scope for intentional causation on the model that I am defending to explain anything physical, because the relevant counterfactuals will all 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
Persons, Rational Action, and Free Will
This page intentionally left blank
6 Event Causation and Agent Causation In Part I, I discussed issues concerning causation chiefly on the widely shared assumption that all causation is event causation—that is, causation by events—even though I was at pains to argue that sometimes, at least, events that are causes may have facts or ‘states of affairs’ as their effects. However, many philosophers—and particularly those concerned with the philosophy of action—consider that a further important species of causation is agent causation, in which the cause of some event or state of affairs is not (or not only) some other event or state of affairs, but is, rather, an agent of some kind.¹ My purpose in the present chapter is to try to understand the relationship between event causation and agent causation, with a view to seeing whether these two species of causation are equally fundamental or one of them is somehow prior to the other. In the end, I shall come down in favour of the view that talk in terms of event causation, to the extent that it should strictly be countenanced at all, is best understood as elliptical for talk in terms of agent causation, although only where the word ‘agent’ is very broadly construed, rather than being confined in its application to the much narrower class of rational agents, ourselves included. 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
6.1. Agents and Agent Causation An ‘agent’, in the sense I intend here, is a persisting object—or ‘individual substance’—possessing various properties, including, most importantly, certain causal powers and liabilities. A paradigm example of an agent would be a human being or other conscious creature, capable of performing intentional actions. Indeed, some philosophers of action would like to restrict the term ‘agent’ to entities such as these and, correspondingly, restrict the term ‘agent causation’ to cases of intentional action. For present purposes, however, I shall adopt a more liberal view, not least because we are not yet in a position to distinguish clearly between those things that are, and those that are not, capable of performing intentional actions, nor even to say what is distinctive about intentional, as opposed to unintentional, action. But later we shall see that there plausibly is something special about the causation of an event by an animate—and, more particularly, by a rational—agent. It cannot be denied that our ordinary ways of talking about action support, at least superficially, the idea that agent causation is a distinct species of causation. As well as making statements of event causation, such as ‘The explosion of the bomb caused the collapse of the bridge’, we say such things as ‘The bomber caused the collapse of the bridge’—where, by ‘the bomber’, we might be referring to an individual human being, or we might be referring to an aeroplane. In short, there is no doubt that the verb ‘to cause’ may take, as its grammatical subject, a noun-phrase referring to a persisting object, either human or inanimate, quite as well as a noun-phrase referring to a particular event. Moreover, since events and persisting objects are entities belonging to quite distinct ontological categories, it is strongly arguable that the verb ‘to cause’ must have a different sense when a term referring to a persisting object figures as its grammatical subject from the sense it has when a term referring to a particular event plays that role. This is brought out by the fact that, whereas it would not be incongruous to say, for example, ‘Smith and Jones together caused the collapse of the bridge’, it would indeed be incongruous to say ‘The explosion of the bomb and Jones together caused the collapse of the bridge’. The latter sentence appears to involve a category mistake, rather like the one that famously occurs in the sentence ‘She came home in a sedan chair and a flood of tears’—the
Event Causation and Agent Causation 123 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’.
6.2. A Putative Analysis of Agent Causation However, even if we accept this conclusion, it does not prevent us from proposing that one or other of these species of causation is ‘reducible’ to, or analysable in terms of, the other—that agent causation is reducible to event causation, or that event causation is reducible to agent causation. A reductionist thesis of this sort would maintain either that any statement of agent causation can be analysed in terms of a statement in which no notion of causation other than that of event causation is employed, or else, conversely, that any statement of event causation can be analysed in terms of a statement in which no notion of causation other than that of agent causation is employed. Opposing either form of reductionism would be the thesis that both species of causation exist but that neither is analysable in terms of the other. In order to see what is at issue here, let us first look at how one might attempt to reduce agent causation to event causation. And let us take, as the canonical form of a statement of agent causation, the following: ‘Agent A caused event e.’ (An alternative and perhaps more colloquial formulation would be ‘Agent A brought about event e’ or ‘Agent A made event e happen’, but I shall adhere to the preceding formulation because it explicitly employs the verb ‘to cause’.) Then the obvious suggestion would be that a statement of this form can be analysed along something like the following lines: (I) Agent A caused event e if and only if there was some event, x, such that x involved A and x caused e. Of course, analysis (I) is no clearer than the notion of ‘involvement’ which it exploits. How, precisely, must an event ‘involve’ an agent A if, in ² See Gilbert Ryle, The Concept of Mind (London: Hutchinson, 1949), ch. 1.
124
Persons, Rational Action, and Free Will
virtue of that event’s causing an event e, A may be said to have caused e? If we cannot spell out an appropriate notion of involvement, or can only do so in a way which appeals, either explicitly or implicitly, to the notion of agent causation, then analysis (I) will have failed. Perhaps, however, we can say something like the following by way of spelling out such a notion of involvement, at least to a first approximation: an event, x, ‘involves’ an agent, A, in the sense demanded by analysis (I), just in case x consists in some change in one or more of the properties of A. (The properties in question might be, but need not be, intrinsic properties of A: they might equally be relational properties of A.) For example, in the case of an animate agent A, such as a particular living creature, an event which is a movement of A is an event ‘involving’ A because a movement of A consists in A’s undergoing changes in the dispositions of its limbs and/or a change in its spatial relations to its environment. Perhaps this will suffice, if only for the time being, to clarify the notion of ‘involvement’ figuring in analysis (I).
6.3. Causative Action Verbs and Basic Actions In order to appreciate some of the attractions, as well as some of the potential limitations, of analysis (I), it is helpful to observe that many transitive verbs of action have quite specific causal implications. For instance, when we describe one person as having killed another, we imply that the first person—the agent of the action—caused the death of the second person. Even if ‘killing’ cannot be defined as simply meaning ‘causing death’, ‘killing’ does at least imply ‘causing death’. Similarly, ‘pushing’ implies ‘causing motion away from the agent’, ‘stopping’ implies ‘causing rest or cessation of motion’, ‘cutting’ implies ‘causing separation or division’, ‘crushing’ implies ‘causing compression or decrease in volume’, and so forth. Some of these causative action verbs, as we may call them, also imply the specific means by which a certain kind of effect is caused by the agent. For example, ‘pushing’ not merely implies ‘causing motion away from the agent’—it implies ‘causing motion away from the agent by the application of pressure’. Similarly, ‘cutting’ implies ‘causing separation or division by the transverse movement of a sharp edge’, whereas ‘ripping’ implies ‘causing separation or division by the application of tension in opposite directions’.
Event Causation and Agent Causation 125 Certainly, we can very often say, when we describe an agent’s action by using a causative action verb, by what means the agent caused the kind of effect which that verb implies, even if the verb that we use does not imply those specific means itself. For example, if we describe an agent as having killed someone, and thus as having caused that person’s death, we may well also be able to say how the killing was done—say, by shooting, or stabbing, or poisoning, each of which verbs does imply a certain kind of means of causing a certain kind of effect. However, there is a special class of actions which seem to defy description in these means–end terms. These are the so-called basic actions, such as a human agent’s spontaneous movement of one of his or her own limbs.³ For instance, when a human agent is described as having waved his hand, it is thereby implied that the agent caused a certain kind of motion in his hand: but nothing is implied as to the means, if any, by which the agent did this. Indeed, if we were to ask such an agent how —in the sense of ‘by what means’, rather than in the sense of ‘in what manner’—he waved his hand, it is very likely that his reply would be that he didn’t do this by any means at all: he would quite probably reply that he simply waved his hand. In contrast, a human agent will, in general, know precisely how he pushed, pulled, stopped, cut, crushed, or killed something or someone—that is to say, he will know by what means he caused effects of the kinds implied by these causative action verbs. It is tempting, then, to define a basic action as an action in which the agent causes an effect of a certain kind, but not ‘by’ any means whatever.⁴ (Of course, when a human agent waves his hand, certain muscular and neural events will precede the motion of his hand and be causally related to that motion. But it seems incorrect to say that the agent waves his hand ‘by means of’ causing these muscular and neural events, of which he is likely to know nothing. Indeed, even if one knows that such events invariably precede hand-motions of the kind occurring when one waves one’s hand, it seems more natural to say that one can cause the muscular and neural events ‘by means of’ waving one’s hand, rather than vice versa.) 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
6.4. The Case for Irreducible Agent Causation The relevance of these observations to the analysis of agent causation suggested above, analysis (I), is the following. On the one hand, the fact that many causative action verbs imply a means–end structure to the actions which they describe seems to favour analysis (I). For the use of such a verb to describe an action very often suggests an appropriate event, ‘involving’ the agent, which can be appealed to for the purpose of analysing the action in the manner proposed by analysis (I). For instance, suppose we describe a human agent, A, as having killed someone else, B, by poisoning. This is an instance of agent causation: A caused B’s death. But we also know by what means A caused B’s death, namely, by administering poison to B. But therefore, it seems, we can very easily identify an event, ‘involving’ A, which can be said to have caused B’s death, namely, A’s administration of the poison to B. So it seems that analysis (I) is vindicated in this instance. For here, it appears, is a case in which it is true to say that the agent, A, caused a certain effect in virtue of the fact that a certain event ‘involving’ A caused that effect—namely, the event of A’s administering the poison to B. On the other hand, the existence of the class of basic actions threatens to defeat analysis (I), for in the case of a basic action, as we have defined it, there appears to be no suitable event ‘involving’ the agent which can be called upon to provide the cause—in the event-causation sense of ‘cause’—of the effect which, in performing that action, the agent is said to cause. For example, when a human agent, A, spontaneously waves his hand, what event, ‘involving’ A, can we appeal to, for the purposes of analysis (I), as having been the cause of a certain kind of motion in A’s hand? Nothing analogous to the event appealed to in the poisoning case will be forthcoming, it seems, simply because, in a case of ‘basic’ action, the agent supposedly does not cause the given effect ‘by’ any means whatever: he simply causes it —that is to say, he causes it directly. Moreover, it would appear that there must be basic actions, if agents perform any actions at all: for if every action had a means–end structure, this would apparently generate a vicious infinite regress, whereby the means of each action is compelled to be the end of another. However, at least in the case of inanimate agents, there appears to be a way of avoiding such a regress and defending analysis (I). This is because
Event Causation and Agent Causation 127 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
hand as being a ‘basic’ action, we implied precisely that in such a case motion in the hand was not caused ‘by’ any means whatever: that is, that there was no behaviour or activity of the agent by means of which the agent caused motion in his hand, in the way that by rolling the boulder caused motion in the tree. Rather, what we seem to have in such a case is an instance of irreducible agent causation. Animate agents, we may feel tempted to say, are capable of spontaneous self-movement, which involves an agent’s causing motion in its own limbs or other body-parts directly. However, it is likely to be objected that this view of the matter is incompatible with a wholly naturalistic conception of animals, including human beings, and their causal powers. It may be urged that however much it may seem that animals engage in ‘spontaneous self-movement’, in the sense just proposed, in reality their agency must always be reducible to event causation, just as the agency of inanimate objects plausibly is.
6.5. The Problem of ‘Free Will’ But why ‘must’ this be so? There seems to be nothing unintelligible or incoherent about the notion of spontaneous self-movement. Moreover, the idea that there is this kind of irreducible agent causation holds out the prospect of a solution to the problem of ‘free will’ which dogs the reductionist approach to human action. If all human agency is ultimately just a matter of one event’s causing another, then, since the causal history of the events supposedly involved in any instance of human agency will plausibly be traceable back, through prior events, to times before the agent’s birth, we seem to lose all sense of the agent’s being genuinely responsible for—the author of—his or her own actions.⁵ A human agent must then be seen as no ‘freer’, in reality, than the boulder which rolls into the tree, its rolling being caused by the action of some other object upon it, which action is in turn caused by yet earlier events—and so on back to the dawn of time. Acknowledging that there is room for a certain amount of probabilistic causation between events provides no escape from this conclusion, since a boulder’s behaviour would be no ‘freer’ on that account than it would be in a perfectly deterministic universe. 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 But how, exactly, would the existence of irreducible agent causation, in the case of animate agents, provide a solution to the problem of ‘free will’? In the following way, perhaps. Suppose that, on a particular occasion, a human agent, A, caused an event e, such as motion in A’s hand, but that this instance of agent causation is not reducible to one of event causation. That is to say, suppose that it is not in virtue of some event’s causing e that A may be said to have caused e. Now, this still leaves open the question of whether or not some prior event or events were causes of e and, if so, whether or not that prior event or those prior events were causally sufficient for the occurrence of e. However, it is perfectly conceivable that no prior event or combination of events was causally sufficient for the occurrence of e, but that A’s agency on this occasion was causally necessary for the occurrence of e. That is to say, it might be the case that if A had not caused e, then e would not have occurred, since prior events alone were not causally sufficient for the occurrence of e. (This is perfectly compatible with the suggestion that certain prior events were, nonetheless, also causally necessary for the occurrence of e.) In that case, it seems, e occurred as a consequence of A’s agency and yet e was not causally determined by prior events (nor, we may suppose, did e have the probability of its occurrence fully determined by prior events). Hence, A was causally responsible for—was genuinely the author of—e and in causing e acted ‘freely’. But, it may be asked, isn’t it the case that A’s causing e was itself caused by some prior event or events? If so, then wasn’t the occurrence of e, after all, causally determined by prior events (or, at least, wasn’t the probability of e’s occurrence fully determined by prior events)? And if not, then wasn’t A’s causing e something quite inexplicable—a matter of pure chance? The proper response to these questions, I suggest, is that if A’s causing e was, as is now being supposed, an instance of irreducible agent causation, then we should not think of it as something which itself had a cause of any kind—but that, at the same time, we should not think of it as something which, in lacking a cause, would therefore have to have been a matter of pure chance.⁶ It is open to us to say, instead, that A’s causing e is subject solely to rational, not to causal, explanation. That is to say, it is open to us to say that A caused e for a reason, while denying that anything caused A’s causing e. I shall say more about this possibility in a moment. ⁶ Compare Timothy O’Connor, Persons and Causes, pp. 52ff. and p. 87.
130
Persons, Rational Action, and Free Will
6.6. Mental Causation, Rational Choice, and Freedom of Action In the preceding few paragraphs, I have been trying to make out as good a case as I can for the idea that agent causation is irreducible to event causation, on the grounds that there are plausible cases of ‘basic’ action by animate agents which involve irreducible agent causation—a ‘basic’ action being an agent’s causing a certain event and yet not causing that event ‘by’ any means whatever. However, an advocate of the thesis that agent causation is always reducible to event causation may urge that, even in what appears to be a case of ‘basic’ action as defined above, there always still is, in fact, an event (or conjunction of events) which causes the event, e—the event that the agent, A, is said to cause—and which, in the appropriate sense, ‘involves’ A. But, it may be added, the event (or conjunction of events) in question will be a mental event (or conjunction of events) of a certain kind—which should not surprise us, since apparently ‘basic’ action of the sort with which we are now concerned is exhibited precisely by animate agents, such as human beings. In particular, it may be suggested that the events in question will simply be the onsets of certain cognitive and/or affective states of the animate agents concerned—in short, the onsets of certain beliefs, desires, or emotions.⁷ Thus, it may be claimed, when a human agent ‘spontaneously’ waves his hand or otherwise ‘spontaneously’ moves some part of his body, the bodily event which the agent is thereby said to cause is one which is in fact caused by the onsets of such states as the agent’s desire to attract someone’s attention and his belief that by waving his hand he could achieve this. (If one adheres to a psychophysical identity theory of mind, one may also maintain that these mental events are in fact identical with certain physical events in the agent’s brain or nervous system, although it is not necessary to make this additional claim.) However, I imagine that the advocate of irreducible agent causation will object to this proposal on grounds which should now be familiar—namely, that the proposal leaves no room for genuine ‘free will’ in human affairs and that it distorts the relationship between an agent’s actions and his reasons for action. An agent’s beliefs and desires may give him a reason 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 to act in a certain way—for instance, to wave his hand—but if we treat those beliefs and desires as causing the motion of his hand, we cannot really say that the agent acts ‘for’ that reason, because to act ‘for’ a reason is to be guided by what one (reasonably) believes and desires but not to be caused to behave in a certain way by one’s beliefs and desires.⁸ In short, the proposal under consideration appears not to leave any room for rational choice on the part of the agent as to how to act in the light of his beliefs and desires. But, if this objection is deemed sound, why shouldn’t an opponent of the idea of irreducible agent causation simply say that, in what appears to be a case of ‘basic’ action as defined above, it is in fact the choice or decision of the agent—a mental event —that causes the event, e, which the agent, A, is said to cause? Another term for this kind of mental event would be a ‘volition’ or ‘act of will’, according to certain philosophers of action. My own view is that, if one wants to maintain that agent causation is reducible to event causation, then it is indeed preferable to appeal to mental events of this kind, rather than to the onsets of states of belief and desire, for the purpose of analysing apparently ‘spontaneous’ human action.⁹ But it may still be objected that to analyse spontaneous human action in terms of causation by mental events of any kind—even the ‘choices’ or ‘volitions’ of the agent—is to preclude any satisfactory solution to the problem of ‘free will’. This is because an event must either have or lack a cause, which on the view under scrutiny would have to be another event of some kind. And then we are faced with an apparent dilemma. For if our choices are caused, then we lack genuine freedom. And if our choices are uncaused, then they would seem to be mere chance happenings, so that we once again lack genuine freedom.¹⁰ In contrast, it seems that the advocate of irreducible agent causation can plausibly resist saying that A’s causing e, in a case of ‘basic’ human action, is itself something which has a cause, without thereby implying that A’s causing e is a mere chance happening: for A’s causing e is not a happening—an event—at all, and is therefore not the sort of thing 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.
6.7. Basic Actions and Backward Causation At this point I want to return, briefly, to an issue touched upon earlier, namely the relationship between a human agent’s ‘basic’ actions, such as his spontaneous hand-wavings, and the muscular and neural events that invariably precede and are causally related to the bodily motions involved in such actions, such as the motions of an agent’s hand. I remarked that, while most human agents know nothing of such muscular and neural events, even if one does know that such events invariably precede the sort of hand-motions that occur when one waves one’s hand, it seems more natural to say that one can cause the muscular and neural events ‘by means of’ waving one’s hand, rather than vice versa. Some philosophers find this view puzzling, however, because they think that a kind of ‘backward’ causation would have to be involved if an agent were to cause a neural event, e1 , ‘by means of’ causing a bodily motion, e2 —such as a hand-motion—where e1 occurs before e2 .¹¹ But in fact no ‘backward’ causation of any kind need be involved here. It is not being proposed that the later event, e2 , causes the earlier event, e1 , since we are not concerned here with event causation but rather with agent causation. Indeed, it can readily be accepted that, in such a case, the earlier event, e1 , is a contributory cause of the later event, e2 —for, as I have already made clear, an advocate of irreducible agent causation need not say that no events whatever are (contributory) causes of the event that is caused by an agent in a case of ‘basic’ action: all that is being claimed is that it is not in virtue of its causation by any events whatever that the agent may be said to cause that event. Nor can it be complained that ‘backward’ causation must be involved because the agent’s causing of e2 occurs later than e1 : for, quite apart from anything else, it simply isn’t obvious when, precisely, the agent’s ¹¹ See G. H. von Wright, Explanation and Understanding (London: Routledge and Kegan Paul, 1971), pp. 76ff.
Event Causation and Agent Causation 133 causing of e2 may be said to occur, if indeed it is proper to describe it as ‘occurring’ at all. (I say ‘if’ because it is events that are said to ‘occur’—and, according to the advocate of irreducible agent causation, an agent’s causing of an event is not itself an event of any kind.)
6.8. The Conceptual Priority of Agent Causation I hope I have done enough, now, to explain why some philosophers of action hold—I think with some justification—that agent causation is not universally reducible to event causation, in accordance with analysis (I) or any similar principle. But what about the suggestion that event causation is instead reducible to, or analysable in terms of, agent causation? This is a suggestion that has only rarely been explored—largely, I suppose, because it is commonly imagined that talk of ‘agent causation’ is appropriate only where the intentional actions of rational, or at least animate, agents are concerned. However, as I explained at the beginning of this chapter, we do not need to use the term ‘agent’ in a way which restricts its application to such beings. There is a perfectly good sense in which something as insensate as a boulder can be a causal agent. Of course, if agent causation could only be exercised by rational or animate agents, then the suggestion that all event causation is reducible to agent causation would amount, in effect, to the doctrine that nothing is caused to occur in the world save through the agency of rational or animate beings. It may be that certain so-called ‘primitive’ cultures have assumed such a view, attributing all physical events which lack a human or animal origin to the agency of gods or spirits. Equally, certain idealist philosophers, notably George Berkeley, have maintained that only thinking beings can cause anything (but that what they cause are certain purely mental effects in their own or others’ minds).¹² Clearly, however, if we are prepared to use the term ‘agent’ in a broader way, to apply to inanimate as well as animate beings, then there is scope to maintain that event causation is reducible to agent causation without embracing one or other of these more improbable doctrines. But before I explore this possibility, I want to discuss a reason that we 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 knowledgeclaim 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 environment—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 environment, 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 object has regularly been followed by a flash of light. However, if one is in doubt as to whether or not such a sequence is causal in nature, one can attempt to resolve that doubt by means of active intervention and experimentation. For instance, one can attempt to interrupt the fall of the next object of the kind in question, to see whether or not a flash of light will still occur even in the absence of the fall. Again, one can attempt to initiate the fall of such an object oneself, to see whether or not such a fall is still followed by a flash of light. Because we are causal agents ourselves and are aware of our ability to intervene in and manipulate the course of nature in such ways, we are able to test causal hypotheses experimentally and thereby endeavour to distinguish, albeit not infallibly, between causal and non-causal sequences of events. (Clearly, an important component of this ability to intervene in the course of nature is our capacity to move ourselves at will about our physical environment in accordance with our desires, at least within certain limits, rather than having our spatial location and orientation determined wholly by external factors.) But it seems that a purely passive creature, however acute its powers of observation, would be incapable of discriminating empirically between causal and non-causal sequences of events—from which it seems reasonable to conclude that it would have no concept of event causation, as it would have no empirical basis upon which to apply that concept. Certainly, it could not acquire the concept of event causation from experience, if the preceding argument is correct, but nor is it reasonable to suppose that it would possess the concept innately, since the possession of a concept that has no application can presumably confer no evolutionary advantage upon a creature.
6.9. An Analysis of Event Causation The lesson of the preceding argument seems to be that the concept of agent causation cannot be derived from that of event causation, because possession of the former concept is a prerequisite of possession of the latter concept. It cannot be that we first learn to apply the concept of event causation to observable events and only then learn to conceive of ourselves as agents by grasping the truth of some principle along the lines of analysis (I) above. It really does seem that agent causation must be, in this sense, conceptually prior to event causation. It does not follow from this, however,
136
Persons, Rational Action, and Free Will
that agent causation must be ontologically prior to event causation: it could still be true that analysis (I), or something like it, tells us correctly what agent causation consists in. On the other hand, provided that we don’t understand the notion of a causal ‘agent’ in too restricted a way, there is nothing to stop us from proposing that agent causation is ontologically, as well as conceptually, prior to event causation. Instead of explaining agent causation in terms of event causation, along the lines suggested by analysis (I), we could attempt to explain event causation in terms of agent causation, perhaps along something like the following lines: (II) Event c caused event e if and only if there was some agent, A, and some manner of acting, X, such that c consisted in A’s Xing and A, by Xing, caused e. Consider, for instance, one of our earlier examples of a statement of event causation: ‘The explosion of the bomb caused the collapse of the bridge.’ According to analysis (II), this is true because there was a certain agent (in the broad sense), namely, the bomb, and a certain manner of acting, namely, exploding, such that the event which was the explosion of the bomb consisted in that agent’s acting in that manner (that is, it consisted in the bomb’s exploding) and the agent, by so acting, caused the collapse of the bridge—that is to say, the bomb, by exploding, caused the collapse of the bridge. Plausible though analysis (II) may be in this sort of case, it may not be wholly free of difficulties. Some of these difficulties, however, may be relatively easy to deal with. For instance, it may be that where analysis (II) speaks of event c as consisting in a single agent’s acting in a certain way, a more refined version of (II) should allow c to consist in one or more agents’ acting in certain ways, either independently or in unison. But I shall ignore this sort of problem as one of mere detail. Much more serious would be the objection, if it is sound, that there are some events which cause other events and yet which do not consist in any agent’s or agents’ acting in any way or ways.¹⁴ However, I shall not pursue this sort of objection here 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 either, as it raises issues which are peripheral to my main concerns in this book. Instead, I shall focus on an apparent difficulty for analysis (II) which arises from my earlier proposal that one can cause the muscular and neural events which typically precede certain movements of one’s hand ‘by means of’ waving one’s hand. We want to say that such muscular and neural events are causes of the hand-movements, rather than their effects. Now, if my earlier proposal was correct, what we have in such circumstances is a human agent, A, and a manner of acting, namely, waving one’s hand, such that A, by acting in that manner, caused a certain muscular or neural event, e1 . But we also want to say that e1 caused, rather than was caused by, the hand-movement, e2 , which occurred on this occasion. Will analysis (II) allow us to say this? Not if we want to say that e2 consisted in A’s waving his hand. For if the hand-movement, e2 , consisted in A’s waving his hand and (as we have just proposed) A, by waving his hand, caused the muscular or neural event e1 , then analysis (II) implies that e2 caused e1 rather than vice versa—it implies that the hand-movement caused the muscular or neural event, whereas the truth is the reverse of this. Fortunately for an adherent of analysis (II), it seems clear that we should not say that the hand-movement, e2 , consisted in A’s waving his hand. A’s waving his hand was an action of A’s—it was A’s causing a certain kind of movement in his hand and so certainly cannot be identified with just such a movement of his hand. The hand-movement, e2 , rather than consisting in A’s waving his hand, in fact consisted in A’s hand’s moving in a certain way. We may be confused about this if we do not distinguish, as we clearly should, between transitive and intransitive senses of the verb ‘to move’.¹⁵ In talking of a ‘hand-movement’, we might either mean to talk of someone’s moving (in the transitive sense) his hand or alternatively mean to talk of someone’s hand’s moving (in the intransitive sense): but, in the present context, we should evidently mean the latter. Obviously, a ‘hand-movement’ in the latter sense can occur even if the person whose hand it is does not move it—that is, even if a ‘hand-movement’ in the former sense is not performed. A’s waving his hand is a ‘hand-movement’ in 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.
6.10. Implications for the Notion of Causality If we take agent causation to be both conceptually and ontologically prior to event causation, what bearing does this have on the question of whether the notion of causality is ultimately analysable or definable in non-causal terms? Elsewhere, I have surmised that the notion of causality might in fact be primitive and irreducible.¹⁶ But nothing that I have said in the present chapter encourages me to think otherwise. First of all, although—as I remarked earlier in this chapter—many transitive verbs of action are ‘causative’ verbs, I see no reason to suppose that the actions which such verbs describe are simply definable in terms of a generic notion of agent causation. For instance, although ‘killing’ clearly implies ‘causing death’, it is very doubtful that ‘killing’ just means ‘causing death’.¹⁷ Why? Because it seems proper to say that an agent may cause the death of a living creature without killing that creature—for example, by causing another agent to kill that creature. Thus, if a would-be assassin cannot bring himself to kill his intended victim himself, he may induce someone else to kill the victim for him—in which case, it seems, the would-be assassin is a cause of the victim’s death even though he does not kill the victim himself. (This isn’t to say, of course, that such a course of action need be any less reprehensible than ¹⁶ See my A Survey of Metaphysics, ch. 10. ¹⁷ See the editor’s introduction in Pieter M. Seuren (ed.), Semantic Syntax (Oxford: Oxford University Press, 1974), pp. 8ff. and Judith Jarvis Thomson, Acts and Other Events (Ithaca, NY: Cornell University Press, 1977), p. 128.
140
Persons, Rational Action, and Free Will
killing, nor that a court of law would be making a mistake in convicting such a person of murder.) Rather than saying that every causative action verb is synonymous with some verb-phrase in which the verb ‘to cause’ (in its agent-causation sense) appears, I am inclined to reverse the direction of semantic explanation and say that our grasp of the concept of agent causation is, as it were, a distillation of our grasp of a multiplicity of more specific action-concepts—concepts such as those of killing, pushing, stretching, attracting, corroding, burning, and dissolving. These and related action-concepts form a family, whose members bear ‘family resemblances’ to one another but possess, I suggest, no common and clearly definable ‘essential core’.¹⁸ Certainly we may say that the actions in question are all ways in which an agent may cause something to happen: but I think that this observation, rather than helping to elucidate the nature of such actions, better serves to elucidate what it means to speak of an agent’s causing something to happen—for we very arguably have a much firmer grasp on the nature of these and similar specific actions than we do on the generic notion of agent causation. However, clearly, we cannot hope to define agent causation by reference to any class of such specific actions, given that they form an open-ended family whose members bear only family resemblances to one another and share no common ‘essence’. So I conclude, albeit somewhat tentatively, that the concept of agent causation is primitive and indefinable, but that it is at the same time the product of our grasp of an open-ended multiplicity of specific ways in which individual objects can act upon one another. 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.
7.1. Event Causation and Substance Causation First, let us focus on the distinction between event causation and substance causation—the latter term being, of course, one that is little in use at present. What is at issue here is a question concerning the relata of causation, that is, 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
a question as to what ontological category or categories the items related by causal relations should be conceived to belong to. I speak of ‘causal relations’ in the plural here advisedly, because I think it is tendentious to assume that there is such a thing as ‘the’ causal relation, although this assumption is very widespread amongst contemporary analytical metaphysicians. The obvious and most appropriate way to begin to approach the question just raised is through a consideration of the syntax of causal statements in natural language—which is not to say, of course, that metaphysical questions have a purely linguistic resolution, only that syntax can be a guide to metaphysical presuppositions. We see, then, that some causal statements in everyday language undoubtedly have an event-causal formulation. These are statements such as ‘The explosion caused the collapse of the bridge’, where the transitive verb to cause has as its grammatical subject a nounphrase denoting a particular event and as its grammatical object a nounphrase denoting another particular event. However, we also have what might be called ‘mixed’ causal statements, such as ‘The bomb caused the collapse of the bridge’, in which the grammatical subject of the verb to cause is a noun-phrase denoting a particular persisting object or individual substance (to use the traditional vocabulary of metaphysics) while its grammatical object is a noun-phrase which once again denotes a particular event. Now, faced with such mixed causal statements, the devotee of event causation—that is to say, the metaphysician who holds that all causation is really or ultimately a matter of one event’s causing another—is likely to say something of the following sort. Such a mixed causal statement, he will say, is really just an abbreviation for some more complex statement of event causation which involves quantification over events. He will say, for instance, that what is really meant by the statement ‘The bomb caused the collapse of the bridge’ is something like this: ‘There was some event, e, such that e involved the bomb and e caused the collapse of the bridge.’ (See again my discussion of this example in Chapter 6.) To make this suggestion work, something must be said, of course, about what it means for a persisting object or individual substance, such as a bomb, to be ‘involved’ in an event—and this may not be altogether easy. That is to say, it may not be easy to spell out a suitable sense of the word ‘involve’ which does not in some manner compromise the proposal now under scrutiny—for instance, by rendering it implicitly circular. After all, the
Personal Agency 143 notion of ‘involvement’ that will need to be invoked cannot be something as thin as merely to imply, in the case in hand, a spatiotemporal overlap between the bomb and the event that is putatively the cause of the collapse of the bridge. That would allow the event in question to be something like a surge of flood water, which swept over the bomb and caused the collapse of the bridge, while the bomb sat inertly in its fixed position on the riverbed. Obviously, what the proposer has in mind is that it is some such event as the explosion of the bomb that caused the collapse of the bridge when we say truly, but by way of abbreviation, that the bomb caused the collapse of the bridge. So the relevant notion of ‘involvement’ must be one that would allow us to say that the bomb is involved in the explosion in a way in which it is not involved in the surge of flood water. But there, for the time being, I shall leave the issue of involvement and turn to other matters.
7.2. The Primacy of Substance Causation What we have just been looking at is a proposal to reduce all talk of substances as causes to talk of events as causes, in line with the following sort of schema: (1) Substance s caused event e2 just in case there was some event, e1 , such that e1 involved s and e1 caused e2 . However, one could readily acknowledge that (1) expresses a conceptual truth without agreeing that all causation is ‘really’ causation by events, rather than by substances. Indeed, it is perfectly possible to argue that (1) is a conceptual truth precisely because it is substances that are in the most fundamental sense ‘causes’. And in support of the latter claim one may urge that it is, after all, substances, not events, that possess causal powers and liabilities. Familiar examples of these causal powers and liabilities are the natural physical dispositions of material objects, such as magnetism and solubility. Substances manifest these powers when they interact causally—for example, when an electromagnet attracts a piece of iron or when the water in a glass dissolves a lump of sugar. What is it for the water to dissolve the sugar? It is, precisely, for the water to cause the sugar to dissolve. The verb to dissolve, like so many
144
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 wellknown 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
7.3. The Reduction of Event Causation to Substance Causation If we say, now, that it is substances that are in the most fundamental sense ‘causes’, what should we say about statements of event causation, such as ‘The explosion of the bomb caused the collapse of the bridge’? What we can say may be captured in the following schema, which is an elaboration of one first set out in Chapter 6. This schema need not, however, be seen as a rival to (1), inasmuch as both may be seen as capturing conceptual truths: (2) Event e1 caused event e2 just in case there was some substance, s1 , and some substance, s2 , and some manner of acting, F, and some manner of acting, G, such that e1 consisted in s1 ’s Fing and e2 consisted in s2 ’s Ging and s1 , by Fing, caused s2 to G. Thus, for example, it is correct, according to (2), to assert ‘The explosion of the bomb caused the collapse of the bridge’, because (we suppose) it was the bomb that, by exploding, caused the bridge to collapse. (I should perhaps remark that, in order to keep matters as simple as possible, I have stated (2) in a way which accommodates only causation by single substances, not causation by a plurality of substances acting in concert. But nothing of importance is compromised by this simplification.) Indeed, the truth of (2) can be used to explain the truth of (1), if we say, as we plausibly can, that for a substance to be ‘involved’ in an event, in the sense required by (1), is just for that event to consist in that substance’s acting in a certain manner.² (Again, I simplify, for ease of exposition, by considering only events that ‘involve’ a single substance.) This is to say, for example, that the explosion ‘involved’ the bomb precisely inasmuch as that event consisted in the bomb’s exploding—whereas, by contrast, the surge of flood water did not consist in the bomb’s doing anything and so did not in this sense ‘involve’ the bomb. For then (1) reduces to this: 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.
7.4. Agent Causation and Basic Actions However, (1∗∗ ), while plausible, is not incontrovertible and, indeed, will be rejected by some advocates of agent causation. The philosophers I have in mind are those who would contend that, for instance, when I raise my arm—cause my arm to rise—as a so-called ‘basic’ action, there is nothing I do by doing which I cause my arm to rise.³ For my own part, I am inclined to disagree with these philosophers, because I am sympathetic to a volitionist theory of action.⁴ I am inclined to believe, thus, that even when I raise my arm in the most ‘direct’ way possible—not, for instance, by pulling on a 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.
7.5. Personal Agency and Volitionism I hope it will now begin to be clear in what sense I can both be an advocate of agent causation and espouse volitionism. A person is an individual substance, in the broad metaphysical sense of that term, in which it denotes a persisting object of some kind. In my own opinion, ‘person’ denotes a basic kind of persisting object and is not what is sometimes called a ‘phase sortal’, denoting a status that an object of a more basic kind can take on during some part or phase of its existence.⁵ Those who think that all persons are animals, that is, biological organisms, take the latter view, for by their way of thinking no human organism comes into existence as a person: it is not a person during its early embryonic stages and very often not a person during the final stages of its existence, just before biological death ensues. As I say, I disagree with this view, because I consider that persons have persistence conditions which differ from those of any biological organism. However, this dispute is not immediately relevant to the present matter in hand—the nature of personal agency—since all that I need to insist on for the time being is that every person is a substance of some kind, even if persons do not constitute a basic kind in their own right. This then allows me to adopt a theory of personal agency which is a causal theory, while at the same time maintaining that all causation is fundamentally substance causation. On this view, persons are agents inasmuch as they are substance-causes of certain events. Which events, though? What we should not say, in my ⁵ See, further, my Subjects of Experience, ch. 2.
148
Persons, Rational Action, and Free Will
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 eventcausation 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 be exercised or manifested from time to time.⁷ Just as an electromagnet exercises or manifests its power to attract ferrous metals when it draws a piece of iron towards it, so a human agent exercises or manifests his or her power or capacity of will whenever he or she wills to do something.
7.6. Causal Powers and the Nature of the Will However, there are also important differences between the will, conceived thus as a species of power, and a natural causal power such as magnetism or a natural causal liability such as solubility. What they have in common is that they are all species of potentiality, in the sense in which the potential is to be contrasted with the actual.⁸ Something that has a power or potential to act in a manner F typically need not always be exercising that power by acting in that manner, but, typically, it will at least be naturally possible for it to act in that manner. I say ‘typically’ because we can easily think of relevant exceptions. For instance, every massive body has a gravitational power to attract others, but it is always exercising that power: gravity can’t be ‘switched off’. On the other hand, sometimes circumstances may render it naturally impossible for an object to manifest one of its powers or liabilities. Think, for example, of a fragile glass that is prevented from being shattered by a protective shield which, however, cannot be removed without vaporizing the glass: it is naturally impossible for this glass to be shattered and yet it can still be correct to say that it is fragile. Now, what is distinctive of causal powers and liabilities is that they are powers to affect or be affected by objects in certain ways. (In an older terminology, the distinction here between powers and liabilities is often registered as a distinction between ‘active’ and ‘passive’ powers, but I shall avoid this way of talking because I think that it can be misleading.) A causal power is a power to cause some object to act in a certain way—for instance, a power to cause sugar to dissolve. And a causal liability is a power to be caused by some object to act in a certain way—for instance, a power to be caused to dissolve by water. The will, however, is not a causal power, inasmuch as its exercise does not consist in the causing of 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.
7.7. Classical Agent Causalism versus Volitionism Before we pursue this issue, let us turn to another and strategically prior issue: the question of whether it is sensible even to attempt to combine an agent-causation approach to action with volitionism, given that our ultimate objective is to make headway with the problem of free will. For, classically, agent-causation theories are opposed to volitionism and exploit this opposition in attempting to solve the problem of free will.⁹ What the classical agent causalist may say about an episode of supposedly free action is something like this. Consider a case in which an agent, a, raises his arm ‘freely’ and as a so-called ‘basic’ action, or in the most ‘direct’ way possible (not, thus, by pulling on a rope attached to the arm, or some such thing). In such a case, the classical agent causalist will say, a causes his arm to rise but not ‘by’ doing anything else whatever—and so, in particular, not by ‘willing’. The classical agent causalist may well have positive reasons for doubting the existence of acts of will, but he also has the following strategic reason for denying their existence: it allows him to say that there really is nothing at all by doing which a causes his arm to rise in the sort of case under discussion. This in turn allows him to contend that in such a case a fact of agent causation does not imply any corresponding fact of event causation. As a consequence, he is in a position to block any attempt to reduce agent causation to event causation, along the lines suggested by schema (1). He 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
7.8. In Defence of Volitionism I concede that the classical agent causalist has a temporary strategic advantage over the volitionist for the foregoing reason, but consider that it is an advantage which is of little value in the longer run. This is partly because the classical agent causalist’s contention, in contravention of schema (1), that a substance s can cause an event e even though it is not by acting in any manner that s causes e, while perhaps being intelligible, appears both utterly improbable and unsupported by anything in our own experience of voluntary agency. It is true enough that when I raise my arm ‘directly’, or in the most ‘direct’ way possible, there is no other bodily movement that I perform by performing which I cause my arm to rise—or, at least, no such bodily movement that I perform intentionally, by performing which I cause my arm to rise (for, after all, it may be said that I move certain muscles in my shoulder, albeit not intentionally, and by so doing cause my arm to rise). That, I suggest, is all that we are entitled to understand by performing a bodily movement ‘directly’, or in the most ‘direct’ way possible—namely, performing it without doing so by intentionally performing any other bodily movement.¹² It is altogether tendentious of the classical agent causalist simply to define a ‘basic’ or ‘direct’ bodily action as one which is performed without doing so by doing anything else whatever. I would, moreover, contend—I know somewhat controversially—that the phenomenology of voluntary action in fact supports volitionism, in that we are at least sometimes consciously aware of our willings as such. The most obvious examples of this arise in cases of sudden and unexpected paralysis, when we knowingly exercise our will but completely fail to effect any bodily movement. In such cases, we are aware of acting, but know also that our action cannot consist in our causing any bodily event. This is not to say that our action might not be a bodily event: it will be one if any of various forms of physicalism is true. However, the truth or falsity of 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
physicalism is not at present my concern, although I think that there is in fact good reason to suppose that volitions cannot be identical with neural or cerebral events of any kind. The classical agent causalist may be well-equipped, then, to duck the question of prior causes which is at the heart of the problem of free will, but can do so only at the expense of espousing a doctrine which is at once metaphysically and psychologically implausible. The volitionist, it seems, must acknowledge that volitions, even if they are actions, are also events and as such entities of the right category to be assigned causes and to be subject to causal explanation. So it may seem that all the work that I have been doing to accommodate volitionism within a substance-causal view is not going to pay dividends as far as the problem of free will is concerned. I disagree. Although I am prepared to acknowledge that volitions are events, it is more important that they are actions and still more important that all causation, if I am right, is fundamentally substance causation, that is, causation by substances exercising their powers upon other substances. According to this way of thinking, all talk of event causation is really mere shadow talk—legitimate enough for purposes of abbreviation, but not revelatory of the real causal machinery that operates at the level of substances and their powers. Furthermore, as we saw earlier, the ontology of powers allows scope for various fine-grained distinctions between kinds of powers, which we may hope to exploit in attempting to make headway with the problem of free will.
7.9. The Will and its Freedom An important point to emphasize here is that one can endorse both schema (1) and schema (2), while also making the following claims. First, that schema (2) is genuinely reductive, in the metaphysical sense, in that it tells us what event causation consists in, whereas schema (1) is simply a conceptual truth which is explicable in terms of the truth of schema (2) together with the principle (rejected by the classical agent causalist) that substances always cause their effects by acting in some manner. Second, that not every event—that is, not every case of a substance’s acting in some specific manner—is caused. By ‘caused’ here I mean, of course, caused by a substance acting in some manner, which I hold to be the fundamental sense in which
Personal Agency 155 anything is ever caused. Given my substance causalism, to say that some event may be uncaused is just to say that, sometimes, a substance may act in a certain manner even though it is not the case that any substance, itself included, causes it to act in that manner (which, of course, it could only do by acting in some manner, if I am right). Such an event will be an exercise of a ‘spontaneous’ power of the substance concerned, to use the terminology introduced earlier. And we have seen that there plausibly are such spontaneous powers in the natural world, such as the spontaneous radioactivity of radium atoms. Consequently, there is nothing unintelligible in the idea that the will itself may be a spontaneous power in this sense. Indeed, it is very natural to suppose that the will is such a power, because we have already seen that it appears to be neither a causal power nor a causal liability. Certainly, the will seems to be quite unlike such natural causal powers and liabilities as magnetism and solubility. It is a characteristic of such natural causal powers and liabilities that the objects possessing them have, as we say, no choice as to whether or not to exercise or manifest them: when a piece of iron is placed near to a magnet, the magnet cannot refrain from attracting it, nor can a lump of sugar refrain from being dissolved when it is placed in water. It is equally true, however, of the spontaneous power of radium to undergo radioactive decay that a radium atom has no choice as to whether or not to manifest that power on any given occasion. But what do we mean when we talk thus of ‘choice’? We mean, I would say, precisely an exercise of the power of will. Willing precisely is choosing a course of action, normally in preference to some other course of action: choosing, say, to raise one’s arm rather than to leave it down. So our very conception of the will is as a power the conditions of whose exercise are quite different from those governing the exercise of other natural powers, whether causal or spontaneous. We conceive of it neither as a power whose exercise is characteristically determined by the causal influence of other objects, nor as one whose manifestations are merely the outcome of chance. Rather, we conceive of it as being a power that is characteristically exercised in the light of reason. We conceive of it, that is to say, as being a rational power.¹³ A power’s being a spontaneous power is clearly not a sufficient condition of its being a rational power, as the example of radium’s radioactivity ¹³ The notion of a ‘rational’ power is, once again, Aristotelian in origin: see Metaphysics, Book θ.
156
Persons, Rational Action, and Free Will
shows. But it is a necessary condition, I would urge, because a power whose nature is such that its exercise or manifestation is characteristically determined, or even merely made more probable, by the causal influence of objects acting upon the possessor of the power cannot, for that very reason, qualify as a rational power. For a substance to act in a certain way because it was caused to act in that way, whether by another substance or even by itself acting upon itself, is not for that substance to act rationally, or genuinely ‘in the light’ of any reasons it may have for so acting. By so acting, a substance may act in a way that is in accordance with certain of its reasons for acting, if it has any, but it cannot be acting for any of those reasons, given that it was caused to act in that way.¹⁴ To act for a reason is to act in a way that is responsive to the cogency of certain considerations in favour of one’s so acting—and this is incompatible with one’s being caused to act in that way, because causal processes bring about their effects with complete indifference to the question of whether those effects have cogent considerations in their favour. To act in the light of one’s reasons for acting in this or that way, one must, then, be able to choose so to act, where choice is understood as being the exercise of a rational power.¹⁵ However, this doesn’t mean that when one so acts there must always be an alternative course of action open to one, other than in the minimal sense that one must be able to refrain from choosing to act in a certain way. It may well be that, in certain circumstances, whatever I choose to do, a certain action-result will eventuate. For example, my hand may be strapped into a device which allows me either to move my finger down, thereby depressing a button and exploding a bomb, or else to leave my finger where it is, in which case I shall prevent a lever from falling, as a consequence of which again the bomb will explode. In this case, whatever I do and however I exercise my power of choice, I shall cause the bomb to explode. Of course, I am presented with alternative courses of action as far as the movement of my finger is concerned, but even this feature of the example may be eliminated by envisaging some device which is directly 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 activated by my mental act of choice, rather than via an intervening bodily movement. In that case, it is beyond my control to make any difference to the course of events affected by my will. Whether I choose to act or choose to refrain from acting, my choice—my volition or act of will—will have the same effect. But in respect of the act of choice itself, I suggest, I am and must be presented with a genuine alternative: to choose to act in a certain way or to refrain from choosing so to act. Choice being the exercise of a rational power, it cannot coherently be supposed to be determined by prior causes: and nothing that we know about physical nature or the human mind, I suggest, demonstrates that one could not be in possession of such a rational power. Indeed, I would go further and say that one cannot rationally believe that one is not in possession of such a power, for this would be to undermine one’s own claim to be a rational agent and thus a rational being—and to deny oneself the title of a rational being is to deprive oneself of any right to appeal to reason in support of any of one’s beliefs. Consequently, it is incoherent to suppose that one might encounter either evidence of an empirical or scientific nature, or arguments of a metaphysical character, which would rationally compel one to judge that one is not in possession of a rational power of choice, such as I conceive the human will to be. I shall return to this theme in later chapters of this book.
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
Now let us make the following further assumption: (3) (∀t)(∀G)(t’s Ging = s’s Fing → t = s & G = F) and consider the right-hand side of (1∗ #), namely, (4) (∃e1 )(∃F)(e1 = s’s Fing & (∃t)(∃G)(e1 = t’s Ging & t by Ging caused e2 )) (4) may be rewritten as (5) (∃e1 )(∃F)(∃t)(∃G)(e1 = s’s Fing & e1 = t’s Ging & t by Ging caused e2 ) Now, ‘e1 = s’s Fing & e1 = t’s Ging’ in (5) entails, by the laws of identity, ‘t’s Ging = s’s Fing’, which, given assumption (3), entails ‘t = s & G = F’. It follows that, by Leibniz’s law, (5) entails (6) (∃e1 )(∃F)(e1 = s’s Fing & s by Fing caused e2 ), which in turn entails, and is therefore equivalent to, (5). Hence, (6) is equivalent to and so may be substituted for the right-hand side of (1∗ #), to yield as a statement equivalent to (1∗ #) and hence equivalent to (1∗ ) (1∗∗ ) s caused e2 ↔ (∃e1 )(∃F) (e1 = s’s Fing & s by Fing caused e2 ) Q.E.D. 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.
8.1. The Problem of the Disappearing Agent Part of our intuitive sense of freedom is our common-sense conviction that we, as free agents, can intervene in the natural world and make a difference
160
Persons, Rational Action, and Free Will
to the historical course of events. The supposedly ‘scientific’ picture of the natural world painted by some physicalist philosophers appears to threaten this conviction in various ways. Often, the threat is diagnosed simply in terms of a conflict between the ‘libertarian’ conception of the will and the supposed implication of the scientific world-view that, in the natural world, everything happens as a result of some mixture of causality and chance. Such a diagnosis makes libertarianism seem an incoherent position, impossibly seeking some middle ground between lawful causal determination and mere randomness. The problem may be posed in the form of a dilemma. All of our actions, including our supposedly free ones, are (it may be said) just events, and all events are either causally determined by prior events or else are chance occurrences (though their chances may be fixed by prior events). Either way, there is no room for the notion that we are the authors of our actions in any sense that would suit the libertarian. For the libertarian wants to say that we are the authors of our actions in a way which renders those actions neither mere chance occurrences nor events that are wholly causally determined by prior events. The libertarian wants to say that we sometimes make a difference to the historical course of events—that is, make the actual course of events different from what it would have been without our interventions. But this requires our ‘interventions’ not to be seen merely as certain events amongst others in the natural course of events. Once we see our own ‘interventions’ in the latter way, they cease to qualify as ‘interventions’ in the appropriate sense, but become instead just part of the natural order which, as free agents, we conceive ourselves as being able to affect through our free decisions. This is a phenomenon that is sometimes described, quite aptly, as the ‘disappearance of the agent’.¹ We begin to see ourselves not as independent authors of our own actions who by acting freely can intervene to affect the course of nature, but just as part of the natural stream of events ourselves, swept up by and even constituted by that stream, whose flow is consequential upon factors independent of us, in the shape of causal laws and chance. Physicalists may say that the conception of ourselves that we thereby give up is one that is well lost, because it is a pre-scientific and ultimately incoherent one. There is simply no room between chance and causal necessity, they 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 reconceptualization 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.
8.2. More on the Primacy of Substance Causation However, matters are not so simple as this stark picture represents them as being. Our conceptions of reason and rationality are not so easily negotiable as the picture suggests. In fact, the image that comes to mind when contemplating the naturalistic philosopher attempting to construct that picture is that of the hapless home-decorator who ‘paints himself into a corner’ or the inept forester who saws off the branch that he is sitting upon. Or, again, that of the artist who, when asked to paint a portrait of the world in his vicinity (‘The World As I See It’), inevitably omits
162
Persons, Rational Action, and Free Will
something of himself from the scene. I am not going to attempt to prove right now that the naturalistic world-view, as I have been calling it, is rationally unacceptable, although I believe that it quite literally is so: that is, that one could never be confronted with empirical evidence which made it rational for one to accept that that picture is correct. (I shall return to that theme in Chapter 9.) Rather, I want to challenge the idea that the libertarian conception of ourselves is incapable of being accommodated within a world-view which incorporates a fundamental distinction between free agency and natural causality. In other words, I want to argue that we are not in fact faced with a stark choice between a return to ‘animism’ and a complete acceptance of the ‘scientific’ world picture as painted by most physicalists, in which libertarian free will has no place. Other philosophers have also attempted to do this, of course, and have generally failed to convince the physicalists. One reason for the failure of some previous attempts, I believe, is that they already accepted too much of the ‘scientific’ world-view before trying to rein it in where human agency is concerned. The key, I think, lies in the very notion of causality itself and the presumption that all causation is fundamentally event causation. Once it is assumed that all causation is a matter of one or more events causing another, it is difficult to resist the thought that human decisions and actions are, after all, just events and as such woven into the universal web of event causation, the only alternative being that they are random outcomes of chance. I should make it clear that my object now is not to defend the distinction between ‘agent causation’ and ‘event causation’, as this would classically be understood. Classical agent causalists hold that an irreducibly distinct species of causation comes into play uniquely in the sphere of intentional agency, while event causation reigns elsewhere.² According to the classical agent causalist, when an intelligent agent, A, intentionally performs a so-called ‘basic’ action, such as raising his arm, A is literally the cause of a certain occurrence—in this case, a rising of his arm—in a distinctive sense in which no event is or could be its cause and in a way which never happens in the world of inanimate objects.³ In particular, the classical agent causalist 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
of intentional agency. Rather, my point is that substances, not events, are the engines of all change in the world. It is substances, not events, that have causal powers and liabilities—the powers to act and be acted upon in various ways—and causation is fundamentally a matter of substances exercising their causal powers to act upon other substances possessing suitable causal liabilities. So, substances are causes, in the fundamental sense of ‘cause’: they are the entities that do all the causing in the world. Causing is acting, and it is substances that act. More particularly, causing is acting upon some substance so as to make that substance in turn act in some way. This happens all the time in the inanimate world. A magnetized piece of iron acts upon some nearby iron filings to make them move towards it: it ‘attracts’ them. Some sulphuric acid acts upon some bits of copper to make them dissolve and transform their constituent atoms into copper ions: it ‘oxidizes’ them. In all such cases, a substance brings about its effects by acting in some distinctive way. The same substance may act in different ways in different circumstances to bring about different effects. For instance, the very same piece of magnetized iron which acts upon the iron filings to make them move towards it may also act in a different way to extend a steel spring from which it is suspended. In the one case, the piece of iron exerts its power of magnetic attraction, while in the other it exerts its weight —this being a power that it has in virtue of the force of gravitational attraction that it and the earth exert upon each other, owing to their respective masses and the distance between their centres of mass. To say, as I do, that all causation is fundamentally substance causation is not necessarily to rule out all talk of event causation as entirely erroneous or misconceived: it is just to say that such talk is never ontologically fundamental. I am not, after all, denying the existence of events. Events occur when substances act or are acted upon in certain ways: indeed, events just consist in the doings of substances.⁴ A death is the dying of a living substance, such as an animal or plant. A marriage is a legal transaction between two human beings. A collision is the coming into contact of two material substances. We can make true statements of event causation, as when we say that a certain collision caused a certain death—for instance, that the collision of the car and the pedestrian caused the pedestrian’s death. What this means, 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.
8.3. Human Persons as Psychological Substances I have spoken of substance causation as being fundamental, and this requires me to say something about the notion of a substance. By a ‘substance’—or, to be more precise, an ‘individual substance’—I mean only what philosophers have traditionally meant by this term: a substance is an ontologically independent entity that bears properties, stands in relations to other substances, persists through time, and undergoes qualitative change over time. Most importantly, for our purposes, substances possess causal powers and liabilities. These are species of disposition. I shall not offer here a general ⁵ For an examination of some of these attempts, see my A Survey of Metaphysics, ch. 10.
166
Persons, Rational Action, and Free Will
theory of dispositions.⁶ It does not much matter to me for present purposes whether we categorize dispositions and thus causal powers as being ‘properties’ of substances, or whether we say, as some philosophers do, that properties ‘confer’ causal powers upon substances which possess those properties. It suffices that we all instantly recognize many clear examples of causal powers and liabilities, such as magnetism and solubility. The causal powers and liabilities of substances are evidently intimately connected with the natures of those substances, that is, with what kinds of substances they are and how they are constituted. The magnetic power of a piece of iron arises from the way in which the atoms making it up are aligned. The oxidizing power of a quantity of sulphuric acid arises from the orbital electron structure of its constituent ions. But some powers may well be fundamental, such as the power of an electron to repel another electron or to attract a proton. This is because an electron may well be a simple substance, with no constituent structure. But, of course, almost all of the macroscopic substances with which we are empirically acquainted are certainly not simple substances, but highly complex ones. Living organisms provide an excellent example. These have characteristic causal powers and liabilities which arise from their inner natures and constitutions. For example, many plants have the power to turn their leaves towards the sun, to convert sunlight into stored energy, and to draw water from the soil. I make these obvious points just to remind us that talk of causal powers and liabilities is ubiquitous in scientific descriptions of the workings of the natural world. It is at best a philosopher’s myth to suppose that the causal language of science is the language of event causation—a myth born, perhaps, of positivism, with its presumption that science is only concerned with recording observable events and noticing their patterns of recurrence, in order to facilitate reliable prediction. Real science is concerned, rather, with revealing the causal mechanisms underlying and explaining recurrent patterns of events. As I remarked earlier, I see human agency as involving substance causation in the foregoing sense. This is because I see human agents—that is, human persons—as being, themselves, substances of a distinctive kind. One view that I am emphatically opposing here is the ‘Humean’ view of 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
persons or selves as being mere ‘bundles of perceptions’, which is to regard persons as being constituted by mental events. If a person were no more than a collection or series of mental events—thoughts, feelings, urges, and so forth—then, because events are, as I have remarked, utterly powerless, so too would persons be. But we certainly conceive of ourselves as being ‘substances’, in the sense explained earlier: we regard ourselves as being bearers of properties, as persisting through change, and, most importantly, as possessing distinctive causal powers and liabilities. Different kinds of substance are distinguished not least by their characteristic existence and identity conditions (under which we may subsume their so-called persistence conditions)—and persons constitute a distinctive kind of substance for this reason, or so I maintain. There are some philosophers, who characterize their position as ‘animalism’, according to whom human persons just are human animals.⁷ (This position, of course, is quite distinct from the ‘animism’ discussed earlier.) More precisely, they take ‘person’ to be a term which characterizes a human animal only during certain phases of its existence—namely, those during which the animal is capable of conscious thought, reason and self-reflection. According to this view, I existed before I was a person and may well continue to exist after I cease to be one. I reject this view as being incompatible with the characteristic existence and identity conditions of persons. Very plausibly, persons can, in principle, persist through changes which cannot be persisted through by animals—and vice versa. Animals are essentially biological beings, but persons are not: persons are essentially psychological beings. In this respect, at least, I agree with the ‘Humean’ view and, indeed, with Locke’s conception of persons. But against Locke and Hume I insist, with Descartes, that the person who I am is a psychological substance, not a mere collection of psychological events happening in or to an animal body. On the other hand, I do not agree with Descartes that I am an essentially immaterial substance—that I occupy no space and have no spatial dimensions and no mass. Rather, I hold that I am a psychological substance coinciding with, but numerically distinct from, my animal body. (See again Chapter 5.) A position quite close to mine holds that I am constituted by my body, rather as a bronze statue is constituted by, but not identical with, the lump 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’.)
8.4. Volition as the Executive Element in Intentional Action Suppose we accept, then, for the foregoing reasons, the claim that psychological substances—that is, conscious, thinking beings, including human persons—are distinct if not separable from their physical bodies and possessed of distinctive and irreducible psychological powers, including the central powers of perception, thought, reason, and will. Some philosophers will persist in purporting to find this view mysterious, but that is because the only sort of account of persons and their psychological powers that they would not find ‘mysterious’ is one that is couched in reductive biological terms—and such an account, it seems clear, is simply not to be had. We just have to accept that there are aspects of nature which defy ‘explanation’ in the reductive terms that alone seem to make such philosophers feel comfortable. A datum that is not to be denied is that we ourselves exist and have the psychological powers that are definitive of ourselves as persons. This is simply an immovable framework assumption of rational debate in this or any other area of philosophy or science, for to deny one’s own existence or rationality is automatically to exclude oneself from any such debate. What remains is to see how, precisely, the conclusions that we have
172
Persons, Rational Action, and Free Will
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 unfashionable.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 causation, 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
if my arm were strapped down, or if I were suddenly to be struck with paralysis in my arm. What is this thing that I do? It is a mental act—an exercise of a distinctive mental power of an ‘executive’ character. The power in question is what philosophers have traditionally called ‘the will’ and our exercises of that power are our ‘volitions’ or ‘acts of will’.¹³ Acts of will are events, for they are the doings of substances, namely, of persons. But they are not mere happenings, in the sense that the falling of a stone is a mere happening. We can describe what the stone is ‘doing’ by saying that it is falling, but a ‘doing’ in this sense is not distinctively ‘active’ in the way that exercises of the will are. What is it, then, about exercises of the will that requires us to say that they are ‘active’ in a way that the mere falling of a stone is not? When it falls, a stone is manifesting—and in this sense exercising—one of its causal powers: the power in question is its weight, which, as we noted earlier, enables such an object to extend a steel spring from which it is suspended. Such a power is ‘passive’, however, because the stone has no option but to manifest the power in some appropriate way whenever suitable circumstances arise. When the stone is suspended from a steel spring, it inevitably manifests its weight by extending the spring, unless some other object prevents it from doing so by exercising a contrary power—as, for example, a strong magnet might do by restraining the spring from extending and so counteracting the weight of the stone. The will, however, is not like this. The way in which it is exercised is not determined by external objects and circumstances. Of course, some philosophers will try to persuade us otherwise. They will say, perhaps, that volitions, or exercises of the will, assuming them to occur, must all be caused to occur by prior events, such as the very onsets of belief and desire of which we have already spoken. That, indeed, is one reason why such philosophers may urge that it is unnecessary to invoke volitions in the analysis of intentional action, because (so they will say) volitions are mere idle wheels between the real causal antecedents of intentional action—the onsets of belief and desire—and the ensuing actions. But to suppose that exercises of the will are causally determined, even by the onsets of belief and desire, is precisely to deny their essentially ‘active’ 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
character. Volitions so conceived would indeed be mere idle wheels in the aetiology of action. What is distinctive of the will as a power is that its exercises are precisely not exogenously determined by exterior causes. In willing we initiate trains of action, whereas the ‘passive’ characterization of the will now under discussion represents willing as being at best a mere passing on of causal influence from prior causes, rather as one billiard ball inherits its momentum from another that collides with it. Why, then, do so many philosophers find the belief–desire analysis convincing? Partly, no doubt, because they recognize that it fits in so well with the presumed sovereignty of event causation. But another important factor is that it may seem that unless we accord beliefs and desires the sort of causal role just described, volition and action will appear to be blind and unmotivated. When we act intentionally, our beliefs and desires must certainly play some role in explaining why we act, for to deny this would be to represent all of our actions as being done merely arbitrarily, or on a whim. Sometimes, indeed, I do act merely on a whim: I kick a stone, perhaps, as I am walking along, and do this ‘for no reason at all’, although not merely unintentionally or accidentally. Explaining my action as one that I did ‘just because I felt like doing it’ scarcely seems to be much of an explanation at all, and certainly not one that we should wish to extend to all of our intentional actions. Generally, the contents of one’s beliefs and desires constitute one’s reasons for acting intentionally in the ways one does. When one acts rationally, one acts ‘in the light of’ one’s beliefs and desires. From this it may seem but a small step to say that those beliefs and desires—or their ‘onsets’—are causes of one’s actions and explain one’s actions causally, while at the same time ‘rationalizing’ one’s actions. But this seemingly small step is, I believe, an erroneous one of enormous magnitude. I was careful to say, a moment ago, only that the contents of one’s beliefs and desires constitute one’s reasons for acting intentionally in the ways one does: I emphatically did not say that one’s beliefs and desires themselves constitute those reasons. Suppose that I decide to raise my arm so as to catch the speaker’s eye at a lecture, with the aim of asking a question, and exercise my will accordingly by endeavouring, successfully, to raise my arm. Asked to explain why I raised my arm I might respond as follows: ‘My reason for so acting was that I wanted to ask a question and believed that by raising my arm I would catch the speaker’s eye.’ However,
Substance Causation, Persons, and Free Will
175
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
one is aware, by exercising one’s will in an appropriate way. (I shall say more in explication and defence of this view shortly, in Chapter 9.)
8.5. The Will as a Spontaneous Power I have been making these points in the course of trying to explain in what sense it is that the will is an ‘active’ power. Part of the upshot is that we have to conceive of the will as a power whose exercises are not characteristically brought about by the operation of prior causes. To this extent we may call the will a ‘spontaneous power’. We have every reason to believe that such powers exist, even in the inanimate world. The power of a radium atom to undergo radioactive decay is, apparently, just such a power, as I remarked in Chapter 7. Where it differs from the will is in that its manifestations are not responsive to the demands of reason. When a radium atom decays, its decay is not caused by prior events, that is, by the causal influence of other objects acting upon it. And yet, of course, the atom does not decay ‘for a reason’, or ‘in the light of’ its beliefs and desires, for it has none. Indeed, there is no ‘reason’ for it to decay or not to decay. What is special about the will is that it is a rational power, that is, a power whose exercises are not only spontaneous but also responsive to the demands of reason. This, at least, is how I think we implicitly do, and plausibly must, regard the will if we are to regard ourselves as rational beings—as indeed I argued in Chapter 7. But it may be wondered whether the ontology of powers, substance causation, and free will being proposed here is compatible with what physical science reveals about the causal structure of the natural world. I believe that it is. Certainly, we must abandon any very strong principle of the closure of the domain of physical events under the relation of event causation: it cannot be the case that every physical event has a sufficient cause consisting of prior physical events. But we know from the example of the radium atom that such a principle is untenable in any case. It is not even the case that the decay of such an atom has its chances of occurrence fixed by prior physical events, for whether or not it decays at a given time is completely insensitive to such prior conditions. Certainly, radium has a distinctive ‘half-life’, meaning that, given a large number of radium atoms, there is a certain objective probability that half of them will have
Substance Causation, Persons, and Free Will
177
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: Birkh¨auser, 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 Is a naturalistic account of rational human action possible? Obviously, we can’t answer this question without being told what the questioner means by a ‘naturalistic account’ of some phenomenon. For most philosophers, however, ‘naturalistic’ just means ‘physicalistic’.¹ For many of these philosophers, a naturalistic account of rational human action would be one which represented human actions as being wholly physical events with wholly physical causes and the rationality of an action could only have something to do with how and by what it was caused. For example, it might be held that every human action is simply a bodily movement of some kind and that such an action qualifies as rational just in case it was caused by the onsets of certain psychological states of the agent whose contents represented such an action as serving the agent’s interests in the circumstances in which the agent found himself. For such an account to qualify as wholly naturalistic in the sense now under consideration, it would have to incorporate a physicalistic account of mental representation and of an agent’s interests. Perhaps such an account of mental representation could be provided in causal or teleofunctional terms and perhaps such an account of an agent’s interests could be provided in terms of evolutionary adaptation. I shall not, however, be pursuing questions about mental representation any further in the present chapter, because I think that a physicalist account of rational human action along the lines just proposed inevitably falls at an earlier hurdle, in virtue of endeavouring to explain the occurrence of such actions in wholly causal terms. I shall give my reasons for saying this shortly. Before coming to these reasons, though—which are, of course, 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.
9.1. Acting for a Reason What is an action? What is a reason to act? And what is it to act for a certain reason? An action is something done by an agent, such as opening a door, raising an arm, uttering a sentence, or imagining a situation. Some actions, such as imagining, are mental actions. Others, such as raising an arm, are physical actions. An ‘agent’, in the sense now intended, is a psychological subject —something that can have thoughts, feelings, beliefs, desires, intentions, urges, and so forth. Persons or selves are psychological subjects with a capacity for reason and self-reflection. And it is with such agents—that is, with beings like ourselves—that I shall be exclusively concerned in what follows. On the issue of the nature of reasons for action, I espouse what would commonly be called an ‘externalist’ view, at least in one clear sense of this often confusingly used epithet. A reason for an agent S to perform some action a consists, typically, in some fact concerning the agent’s circumstances. For example, if S is standing near to a dilapidated building, the fact that a loose slate is sliding from the roof and is about to fall on S’s head constitutes a reason for S to jump out of the way. Of course, S will be in no position to act upon that reason and jump out of the way unless S is aware of the impending danger. It seems safe to say, then, that only if S believes that the slate is about to fall on his head will S jump out of the way for that reason. But this is very far from saying either that S’s belief that the slate is falling is his reason, or part of his reason, for jumping or that this belief is a cause of his jumping.² That an agent has a certain belief can be, but typically isn’t, a reason for the agent to act in a certain way—but a belief as such cannot ever be a reason for action, because it is not even something 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 in the right ontological category to constitute such a reason. And, although a belief can sometimes cause or help to cause some of an agent’s behaviour, a rational action is precisely one that isn’t caused by an agent’s beliefs and desires. Rather, it is one that the agent chooses to perform ‘in the light of’ his or her beliefs and desires, which is a very different matter. Let me now expand a little upon these claims. I am happy to acknowledge that sometimes the fact that an agent has a certain belief constitutes a reason for that agent to act in a certain way. Here I can do no better than to adapt an example of Jonathan Dancy’s.³ The fact that someone has the paranoid belief that he is being pursued by alien beings from Mars is a good reason for that person to visit a psychiatrist—although, of course, it is unlikely that such a person will be able to act upon this reason, because it is unlikely that he will be aware that his belief is paranoid. By contrast, in our earlier example of the falling roof-slate, it is not the fact that S believes that the slate is falling that constitutes a reason for S to jump out of the way: rather, it is the fact that the slate is falling—although, once more, S must be aware of this fact and so believe, or at least suspect, that the slate is falling if he is to be in a position to act upon that reason. The mistake that is characteristically made by those who favour an ‘internalist’ view of reasons for action is to regard such psychological states themselves, rather than the contents of such states, as constituting reasons for the agent to act—where those ‘contents’ are the facts or states of affairs whose existence is revealed to the agent in virtue of his being in such psychological states. The mistake is twofold in character. First, it represents a reason for action as being a state of the agent, when in fact it is a state of affairs. Second, it represents a reason for action as always being psychological in nature, when in fact this is only the case in exceptional circumstances, such as that of the paranoid believer. Even in the latter sort of case, the relevant reason for action is a psychological state of affairs—the fact that the agent has the paranoid belief—rather than the psychological state of the agent, his paranoid belief. I am also happy to acknowledge that there are many cases in which an agent’s beliefs do cause, or help to cause, some behaviour of the agent’s. The example of the paranoid believer in persecuting Martians will again serve us well. An agent with such a paranoid belief may well be caused by that belief to act in various seemingly bizarre ways, such as hiding behind ³ See Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2000), p. 125.
182
Persons, Rational Action, and Free Will
bushes when approached by strangers in the park. Now, of course, the fact—were it to obtain—that one is being pursued by Martians might well constitute a good reason to hide behind bushes when approached by strangers, and so we are in a better position to understand this agent’s seemingly bizarre behaviour once we know that he has the paranoid belief. But we would be wrong to judge that the agent is acting rationally, or for a reason, just because we know that he has this belief and that its content would, if the belief were true, provide him with a reason for acting in the way that he does. If his paranoid belief is causing him so to act, then he is not choosing to act in this way in the light of that belief: he is simply being impelled by the belief and as such is not acting rationally. It is already well known that there is a problem of ‘deviant causal chains’ for those theorists who maintain that to act rationally is to be caused to act in a certain way by beliefs and desires which represent such an action as being in the agent’s interests in the circumstances in which he finds himself. To use Donald Davidson’s own famous example, a climber may be caused to let go of the rope supporting his companion by his strong desire to save himself and his belief that by letting go he would increase his own chances of survival—and yet the climber’s behaviour may be entirely unintentional, as it was only because the onset of this belief and desire so unnerved him that his grip slackened.⁴ It has proved notoriously difficult to say, quite generally, how beliefs and desires should be supposed to cause behaviour if the behaviour is to qualify as rational action performed in the light of those beliefs and desires. What I am now contending, however, is that there can be no correct account of rational action in such causal terms, because an action’s being caused by the agent’s beliefs and desires is incompatible with its being an action that the agent chose to perform in the light of those beliefs and desires—and no action is rational if the agent does not freely choose to perform it. An action can be in accordance with reason, or reasonable, without being rational. To be rational, it must be done for a reason which the agent freely chooses to act upon—or so I claim. For example, jumping out of the way of a falling slate is a reasonable thing to do, but it is not done for a reason and so rational unless the agent chooses to jump out of the way in the light of his belief that the slate is falling. 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 This may seem to turn on its head a famous argument of Davidson’s in defence of the doctrine that ‘reasons are causes’.⁵ And so it does. The argument appeals to the fact that it is not enough, to explain why an agent acted in the way he did, just to point out that the agent was aware of certain reasons for so acting. For the agent might well have been aware of various different reasons for so acting and, indeed, of various reasons for acting in other ways instead. To explain why the agent acted as he did, we need to know which of these reasons was the reason for which he actually did so act. We are supposed to conclude that the reason for which he actually acted was the one that actually moved him so to act, that is, the one that caused him so to act—which would have to be a psychological state of the agent, such as a certain combination of belief and desire. However, I have just maintained that being caused to act by certain of one’s beliefs and desires is in fact incompatible with one’s acting rationally, even though it may be compatible with one’s acting ‘reasonably’, or ‘in accordance with reason’. And the problem of deviant causal chains is grist to my mill here. Admittedly, this still leaves me needing to answer the question that Davidson’s argument poses: which of those reasons for acting of which the agent may have been aware at the time of acting was the reason for which he actually acted, if there was one? But my answer to this is straightforward: the reason for which the agent acted is the reason which the agent chose to act upon. Normally, the agent himself will be able to tell us which reason this was. In situations in which an agent is aware of a variety of reasons for acting in a certain way, and of other reasons for acting in other ways, the agent is confronted with a choice as to how to act and on which reason to act. By choosing to act in a certain way in the light of a given reason, the agent makes that reason the reason for which he acted on that occasion.⁶ This is something that the agent himself brings about through his mental act of free choice.
9.2. Objections and Replies I am aware, of course, that this talk of a ‘mental act of free choice’ will ring alarm bells with all right-thinking ‘naturalists’. I shall say more about 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 ⁹ Compare Jonathan Dancy, Practical Reality, pp. 131ff.
186
Persons, Rational Action, and Free Will
irrationally. For it cannot be a requirement of rational agency that rational agents be infallible in their exercise of such powers. By that standard, only God could be a rational agent. Thus, in the absence of an external reason for an agent’s acting as he did, we are not obliged to provide an internal one—in the shape of a false belief of the agent’s—in order to defend the plausible claim that the agent still acted rationally. So let us turn to the second question raised a moment ago, namely, how a mere state of affairs, such as the falling of a slate, could constitute a reason for acting in a certain way save in conjunction with some appropriate desire on the part of the agent, such as a desire not to be hit by the slate. What I am inclined to say about this is similar to what I said about the agent’s beliefs earlier: first, that a desire as such is the wrong sort of thing to constitute a reason for action and, second, that the fact that the agent has a certain desire, while it could constitute a reason, or part of a reason, for the agent to act in a certain way, does not normally, much less necessarily, do so. Here I am again simply following Dancy’s lead. However, I am happy—as Dancy may or may not be—to say that some reasons for action manifest themselves to the agent’s consciousness as the contents of desires while others do so as the contents of beliefs. In the case of the falling slate, the agent’s awareness of this reason for jumping manifests itself as his belief, or at least suspicion, that the slate is falling. But jumping in these circumstances is a reasonable thing to do only for an agent in whose interest it is not to be hit by such a slate. Human beings can be seriously injured or killed by falling slates and it is consequently in their interests not to be hit by them. They will, correspondingly, tend to desire not to be hit by falling slates. The full reason for a human being to jump out of the way of a falling slate is that the slate will very probably injure him and it is in his interest not to be injured.¹⁰ However, although his awareness of this interest will manifest itself, at least in part, as a desire not to be injured, his action of jumping out of the way of the slate will only qualify as a free and so rational action if he chooses to jump out of the way in the light of this desire. If the desire merely causes him to jump out of the way and the power of choice is not exercised by the agent on this occasion, his behaviour is undoubtedly reasonable, in that it furthers his interests, but it is not an instance of rational action. Or so I want to claim. 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 Another complication that I should at least touch upon here concerns the distinction between moral and practical reason. I have been talking so far only about reasons for action which consist in the agent’s having certain ‘interests’ and his being in a certain kind of situation—for instance, in his being in danger of being hit by a falling slate and its being in his interest not to be injured by it. But sometimes we consider it rational for an agent to act contrary to his own interests, in the furtherance of some moral consideration, such as the requirement to alleviate the suffering of others. I am perfectly happy, at least for present purposes, to endorse this view and to accommodate it by allowing that some reasons for action consist in moral facts or states of affairs (although I shall return to this issue in the next chapter). Of course, many naturalists think that the very notion of a ‘moral fact’ is at best obscure and at worst unintelligible—witness J. L. Mackie’s famous argument from ‘queerness’ against such facts.¹¹ Not being a naturalist myself, I have no such problem. This is not, of course, to say that nothing more needs to be said about how and why moral facts can exist. However, morality is not my present concern and I shall say no more about it here, leaving further discussion of the issue to the next chapter. Many rational actions are purely prudential and have nothing of a moral character about them—and it is with such actions that I am now primarily concerned. What I now want to discuss is the fundamentally metaphysical question of how it is that ‘free choice’—which I take to be an essential ingredient in all rational action—is possible and how, in particular, we can avoid the conclusion that mental acts of ‘choice’, assuming them to occur at all, would have to be either causally determined by prior events or else merely the random products of chance. For neither of these alternatives is compatible with the view, which I wish to support, that in acting rationally we freely choose between genuine alternatives in the light of the reasons for action of which we are aware.
9.3. Choice, Causation, and Free Agency Here is the picture of rational action that I wish to endorse. In certain situations, a rational agent is confronted with two or more alternative ¹¹ See J. L. Mackie, Ethics: Inventing Right and Wrong (Harmondsworth: Penguin Books, 1977), pp. 38ff.
188
Persons, Rational Action, and Free Will
courses of action, which are genuine alternatives in the sense that, at the time at which the agent must choose between them, no sufficient cause already exists of one of the prospective outcomes. The agent then deliberates, by weighing the reasons for and against the various different alternatives. Finally, the agent makes a decision and chooses to act in one way rather than another. His mental act of choice may then, and normally will, contribute causally to a chain of subsequent events issuing in the performance of the chosen action. So, for example, the agent is attending a lecture and is confronted with a choice between raising his arm to ask the speaker a question or refraining from raising his arm and instead listening to other people’s questions. The agent is aware of certain considerations favouring the first course of action and other considerations favouring the second. For instance, in favour of the first may be the fact that his question is an important one and no one else is likely to raise it. In favour of the second may be the fact that, being of a shy and nervous disposition, he is not as good as many other people at articulating questions clearly in public. Having deliberated for a while, the agent then decides to ask his question, chooses to raise his arm and, by this act of choice, initiates a train of events which results in the rising of his arm. We are to suppose that, prior to the agent’s act of choice, no causally sufficient condition of the arm’s going up when it did already existed. The first moment at which there existed a causally sufficient condition of this event was the moment at which the agent chose to raise his arm. The agent’s act of choice was perfectly free, in the sense that it had no antecedent cause whatsoever. Instead of that act of choice occurring, it was equally possible, given the state of the universe prior to that moment, that a different choice should have been made by the agent—a choice to refrain from raising his arm—in which case his arm would not have risen when it later did. I suspect that many people will agree that this is how they think of themselves as proceeding when confronted with what appears to be a choice between two or more alternative courses of action. But does this picture really make sense? Self-styled ‘naturalistic’ philosophers will almost certainly say no—and even many ‘antinaturalistic’ philosophers may be inclined to agree with them. Some will argue as follows. Suppose the agent chooses to raise his arm and his arm duly goes up. We need to ask why the agent chose to raise his arm. There must surely be some explanation
Rational Selves and Freedom of Action 189 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.’
9.4. Choice and Chance At this point, our philosophical interrogator, seeing that he is not going to get the agent to admit that anything caused him to choose as he did, may take another tack and try to persuade the agent that if his choice was genuinely uncaused, then it must have been somehow arbitrary—a mere chance event—and as such the very antithesis of anything rational. Moreover, if acts of choice are mere chance events, then how can any agent be said to have an ability to choose to act in one way rather than another? For to possess such an ability the agent must surely have some sort of control over how he chooses and no one can, almost by definition, have any sort of control over what happens merely by chance. One way of pursuing this line of argument is through a thought experiment suggested by Peter van Inwagen.¹² We are to suppose that Alice is faced with a choice between lying and telling the truth on a given occasion and, after much deliberation, chooses to tell the truth. The 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
really involves and if choice really is causally undetermined in the way the libertarian maintains, then the libertarian should not meekly accept the terms in which the thought experiment involving Alice is described. For the libertarian should reject the idea that one could, even in principle, set up circumstances at any time prior to Alice’s act of choice in such a fashion that she was bound to go through the same process of deliberation as she actually did. This is not to say that the libertarian should redescribe Alice’s episode of rational deliberation and action simply as a punctuated sequence of discrete indeterministic choices separated by deterministically evolving mental processes—first, choices about how to direct her thoughts in gathering and evaluating reasons for and against truth-telling and finally a choice, in the light of those reasons, whether to tell the truth or a lie. Rather, the whole train of deliberation which culminates in the final choice should be seen as a continuous indeterministic process, capable of evolving differently at every moment at which it is going on.¹³ Seen in this light there is, I think, no intelligible way of representing what is going on with Alice as having any numerical degree of objective probability or chance associated with it, even setting aside my earlier objections to this notion. The final thing that I want to say about the Alice thought experiment is this. The choice that Alice actually made was made in the light of reasons of which she was aware, favouring the course of action that she chose: but if she had chosen differently, whether or not after a different process of deliberation, she would still have chosen as she did in the light of reasons of which she was aware. Choice, by its very nature, is never exercised ‘blindly’, but is always informed by or responsive to reasons for action. In this respect, it is utterly unlike any mere chance event, such as the fall of a die or the decay of a radium atom. It doesn’t, I believe, make sense to think of an agent as choosing between two or more possible courses of action without being aware of any consideration whatever in favour of carrying out any of them. This is not to deny that an agent may find himself faced with a choice between two courses of action for which equally good reasons seem to present themselves—a ‘Buridan’s ass’ situation—so that he can do nothing but make an arbitrary choice in favour of one of them. Nor ¹³ See, further, Storrs McCall and E. J. Lowe, ‘Indeterminist Free Will’, Philosophy and Phenomenological Research 70 (2005), pp. 681–90.
Rational Selves and Freedom of Action 195 is it to deny that we sometimes choose to do things ‘just because we felt like doing them’—as, for example, when we choose to kick a small stone lying in our path as we stroll along. Even in this sort of case, our choice is made in the light of the consideration that no harm is likely to come from kicking the stone and that we shall get some small degree of pleasure or satisfaction from kicking it.
9.5. Choice, Agency, and Control But what about the notion that rational action requires that the agent be able to exercise control over what he does? Is the libertarian notion of choice at odds with this idea? That it is at odds with the idea of control is, of course, one thing that was meant to be shown by the thought experiment involving Alice. Alice’s case was supposed to show this by showing that a free choice, on the libertarian conception of what such a choice involves, can only be a mere chance event—and, almost by definition, one cannot have control over a mere chance event. But even if we have succeeded in defending libertarianism against the charge that free choices as it conceives of them must be mere chance events, this doesn’t of itself serve to explain how, or in what sense, according to libertarianism, an agent has control over what he chooses to do. However, the idea that what is needed is an account of how the agent has control over how he exercises his power of free choice is, I believe, a deeply confused one. According to libertarianism, it is precisely because we have a power of choice which we can exercise freely—that is, a power of choice whose exercises are not determined by prior events—that we have control over our actions. If our choices were causally determined for us by prior events, then indeed we would not have genuine control over our actions. In fact, our ‘choices’ would not really be choices at all, in any serious sense. Our choices, I have said, are necessarily informed by or responsive to reasons for action, but are nonetheless necessarily free, in the sense of being undetermined. The notion of free choice exercised in the light of reason provides us with the very paradigm of what it is to be ‘in control’ of our actions. To demand that we somehow be ‘in control’ of our choices seems either superfluous or incoherent. On the one hand, freely exercising our power of choice precisely is being ‘in control’ and as such needs no further exercise of ‘control’. On the other hand, what is in control
196
Persons, Rational Action, and Free Will
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.
9.6. The Pragmatic Inconsistency of Determinism The final issue that I want to address in this chapter is the following. Having argued that our intuitive self-conception is perfectly coherent in representing ourselves as beings whose rationality in action consists in a power of free choice exercised in the light of reason, we may still seem to be faced with the question as to whether this picture of ourselves is true or false. We may have established that it could be true, but could it nonetheless be the case that it is in fact false and that in painting this picture we are deluding ourselves, perhaps in a way which we find psychologically unavoidable? To those who pose such a question I wish to pose the following question in return. Can we make sense of the thought that we might be confronted either with evidence of an empirical nature, or arguments of a philosophical or logical character, which would be rationally compelling and speak in favour of the falsehood of the picture that has just been painted? I do not believe that we can, because I do not believe that we can separate rationality of belief from rationality in action. To form a rational belief that we are not free to act in the light of reason, we should have to exercise rationality in action at the very least in directing our thoughts towards putative sources of evidence for that belief and in evaluating the deliverances of those sources of evidence, or in considering and evaluating arguments in favour of that belief. If we are not free to act in the light of reason, then we are not free to deploy our judgement in the light of reason in seeking out and assessing evidence and arguments for or against this or that belief. If, lacking freedom of rational action, we were to acquire the belief that we lack that freedom, not through the free direction of our thoughts and the free use of our power of judgement, but rather as a consequence of prior causes determining the contents of our beliefs, then we would not have acquired that belief rationally and would not be rationally justified in holding it. There is a perfectly good sense, then, in which we simply cannot rationally believe that we lack freedom of rational action. If we came to believe that we lack that freedom, it would either be a false belief that we had acquired
198
Persons, Rational Action, and Free Will
through a faulty exercise of our rationality, or else it would be a true belief which we did not hold rationally because, in virtue of its truth, we would not be rational beings. What cannot be the case is that we should hold the belief rationally and the belief be true. Nothing, therefore, can rationally commend the belief, and reason demands that we dismiss it as false.
10 Needs, Facts, Goodness, and Truth In this final chapter, I want to explore certain parallels between the logic of action and the logic of belief or, as it might otherwise be put, between practical and theoretical reasoning and rationality. The parallels will be seen to involve an ontological dimension as well as psychological and linguistic dimensions. It may help to begin by mentioning how I was drawn into an examination of these parallels. This was through becoming convinced of the correctness of an externalist account of reasons for action, having been persuaded of this by, amongst other things, arguments to be found in Jonathan Dancy’s important book on the subject, Practical Reality.¹ Externalism about reasons for action appeared to me to be, on reflection, the only view that one could plausibly adopt in conjunction with a libertarian account of free will—the latter being a position which I am now convinced is not only coherent but entirely defensible and indeed correct (as indeed I argued in Chapter 9). Oddly enough, however, recent debates concerning internalist versus externalist accounts of reasons for action tend to have been dominated by moral philosophers, whereas those concerning compatibilist versus libertarian accounts of free will tend to have been dominated by philosophers of action. As a consequence of this, the two debates have been carried on relatively independently of each other—in my view, to the detriment of both. The present chapter is, then, part of a larger exercise of trying to bring them together. ¹ See Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2000).
200
Persons, Rational Action, and Free Will
10.1. A Summary of the Story so Far Very roughly, what I want to say about a voluntary action, freely undertaken after deliberation, is this (here I summarize the view developed in Chapter 9). The agent makes a choice in the light of reasons disclosed by the deliberative process. This choice is a mental event—what early modern philosophers like Locke would have called a volition or exercise of the will. But it is an event which does not have a sufficient cause in the form of antecedent events of either a mental or a physical nature. This does not mean, however, that it is a mere ‘chance’ event, like the spontaneous decay of a radium atom. For choice is normally made in the light of reason. Reasons for action are considerations which speak in favour of or against one’s acting in a certain way. When one acts freely in the light of such reasons, by choosing to act in one way rather than another, one is not caused so to choose by certain of those reasons, for then one’s action would not be free. The famous arguments of Donald Davidson notwithstanding, reasons are not causes.² On the Davidsonian account, reasons for action are beliefs and desires, and the reasons for which an agent acted on a given occasion are beliefs and desires which caused him so to act, while also ‘rationalizing’ the action by representing it, under an appropriate description, as something which would serve to satisfy those desires given the truth of those beliefs. However, to be caused to act in a certain way by certain of one’s beliefs and desires is not consistent with one’s acting rationally. It may be that by so acting one is doing something reasonable, but because one is not freely choosing to do it in the light of objective considerations favouring that choice, one is not acting rationally. The basic point here is that causes bring about their effects with complete indifference to the question of whether or not there are objective considerations in favour of those effects being brought about. But free and rational agents cannot act freely and rationally with indifference to such considerations and so cannot, in acting freely and rationally, be caused to act by their beliefs and desires. This is not to deny that we are sometimes caused to act as we do by our beliefs and desires, only to deny that on those occasions we are acting 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 rationally. Nor is it to deny that when we act rationally we possess beliefs and desires whose contents are relevant to the way in which we act. For it is the contents of such beliefs and desires that constitute our reasons for acting as we do, but not the beliefs and desires themselves, as an internalist account of reasons for action would maintain. (I shall modify this claim later in the chapter, but not in a way that undercuts my opposition to an internalist account of reasons for action.) A simple example—discussed earlier in Chapter 9—may help to illustrate what is at issue here. Suppose that I am standing under the eaves of a dilapidated house and that a loose slate is sliding down the roof in such a way that it is likely to land on my head. This conjunction of circumstances gives me a reason to jump away from the spot on which I am standing. But, obviously, I cannot act on that reason unless I am aware of the circumstances. I must, for instance, believe, or at least suspect, that a slate is sliding towards me and that if it hits me I shall be seriously injured. Now, it may be that my acquisition of this belief, together with other factors, causes me to jump out of the way without any process of deliberation and choice taking place. In that case, my action is instinctive and automatic. It is certainly a reasonable way to act in the circumstances, but it is not a rational action because it is not freely chosen in the light of the reason for so acting which my belief discloses to me. Saying that the action is in this case not rational is by no means intended as a criticism of it or of myself as its agent. On the contrary, we would surely not survive in a hazardous world if all of our actions had to be freely chosen in the light of reasons disclosed to us. At the same time, it is important to recognize that what makes us different from other animals is that we are at least sometimes able to act rationally. If we adapt the example of the falling slate, allowing me plenty of time to be aware of its trajectory and the danger that its fall presents to me, we can envisage my subsequent act of jumping to be a genuinely rational one. It will be so, if I jump as a result of choosing to jump, in the light of the reason for jumping that is disclosed to me by my awareness of the falling slate and the danger that it presents. But, whether I jump rationally or merely instinctively, in the latter case being caused to jump by my belief that the slate is falling and presents a danger, my reason for jumping is not that belief —a psychological state of mind—but its content. Or, more precisely, it is the state of affairs represented as obtaining by the propositional content of that belief, namely, the fact that the slate is falling and presents a danger to me. It is an insistence upon this
202
Persons, Rational Action, and Free Will
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.
10.2. What Kind of Thing is a Reason for Action? So far, I have been trying to sketch and defend a picture of rational action which is at once libertarian and externalist, in strong contrast with the dominant contemporary model, which is compatibilist and internalist. Now, however, I have to explain what all of this has to do with the subject of needs, which figures prominently in the title of this chapter. For all that ³ See Donald Davidson, ‘Freedom to Act’, in his Essays on Actions and Events.
Needs, Facts, Goodness, and Truth 203 I have said so far—and, indeed, as I assumed throughout Chapter 9—it might appear that reasons for action, construed in the externalist fashion that I have recommended, are facts or states of affairs, disclosed to the agent as the contents of certain of his or her propositional attitudes, most obviously his or her beliefs. This seems to fit the case of the falling slate well enough, at first sight. My reason for jumping out of the way, it may be said, is the fact that the slate is falling towards me and thus presents a danger to me. This fact is something of which I may become aware through perception, thereby becoming the content of a belief of mine, and in the light of that belief I may choose to jump out of the way—assuming that I have time to choose, and do not simply act instinctively on account of that belief. But now it may be complained, invoking the venerable authority of David Hume, that I cannot be expected to jump out of the way in virtue of holding that belief unless I also possess an appropriate desire —in this case the desire not to be injured by the slate. Belief alone, it is said, is insufficient to move any agent to action. My first response to this is to say that this may perhaps be so where non-rational actions are concerned. In the case in which I instinctively jump out of the way of the slate, without exercising a free choice, perhaps it must indeed be the case that I am caused so to move at least in part by a desire not to be injured, not just by the belief that the slate is likely to injure me. However, whether or not this is so seems to me to be an empirical psychological question, which cannot be settled simply by a priori philosophical reflection on the nature of such actions. Be that as it may, I am not primarily concerned, in any case, with non-rational actions in which I am caused to act as I do by my antecedent psychological states, whether these be beliefs or desires or combinations of them both. I am primarily concerned with rational actions, taking these to be understood along the lines sketched previously. So the issue for me now is whether any comparable complaint can be raised against this view of rational action on account of its taking reasons for action to be constituted solely by facts or states of affairs disclosed to the agent through his or her beliefs. It would seem that an analogous complaint can indeed be raised. This is that facts can only constitute reasons for belief, not reasons for action. If it sometimes seems as though we can present reasons for action as being facts of a certain kind, this is really only a subterfuge, or so it may be said. For example, the fact that the slate is falling and about to hit me certainly
204
Persons, Rational Action, and Free Will
constitutes a reason for me to believe that I am likely to be injured unless I take avoiding action by jumping out of its way. But it doesn’t seem that this fact, or this fact alone, suffices to constitute a reason for me to jump in these circumstances. Something is still missing from our account. It might be suggested that what is missing is the fact, or supposed fact, that being injured in this way is not in my interests. After all, it may be said, we can surely envisage cases in which it would be in an agent’s interests to be injured by a falling slate—for example, so that he or she could collect a large amount of money from an insurance company. We were just taking it for granted in the present case that it was not in my interests to be injured, so this presumed fact must be added to give a proper and full account of what constituted a reason for me to jump on this occasion. But what kind of fact is it that something or other is ‘not in my interests’, or that something else is ‘in my interests’? This sort of locution is hardly perspicuous, even if it is commonplace. The danger at this point is that we shall fall back into the clutches of internalism and be persuaded to say that what something’s being ‘in my interests’ ultimately amounts to—all that it can ultimately amount to—is the fact that I desire it. The very word ‘interest’ encourages this identification with something of a psychological character, and desire naturally presents itself as the appropriate psychological category to invoke in order to classify it. The only way to avoid this sort of psychologism about reasons for action and to represent reasons for action as something objective and external, while at the same time locating them within the realm of facts or states of affairs, is to invoke some sort of distinction between those facts or states of affairs which are apt only to constitute reasons for belief and those that are apt instead to constitute reasons for action. The former may perhaps be distinguished as being purely ‘descriptive’ facts, the latter as being ‘normative’ facts—in a broad sense of ‘normative’, no doubt, which is suited to embrace both properly ‘moral’ facts and also merely ‘prudential’ facts. However, if we proceed down this route we shall obviously find ourselves mired in the seemingly endless disputes about the so-called fact/value distinction. Many philosophers find the notion of a ‘normative fact’ virtually a contradiction in terms. Facts as such, they say, are value-free. Values are projected by us and perhaps by other intelligent beings upon a world that is essentially or intrinsically valueless. The idea that there might be moral or even prudential facts is dismissed as ‘queer’ and incompatible
Needs, Facts, Goodness, and Truth 205 with a naturalistic view of the world as disclosed to us by science.⁴ Such facts, it seems, would not be empirically detectable in the way that scientific facts supposedly always are, unless we impute to ourselves a special faculty of moral sense which is apparently inconsistent with a naturalistic account of the physical and psychological make-up of human beings and other animals. In the light of such seeming metaphysical difficulties, the advocate of normative facts may be pressured into according them a subordinate status, constructing them as ‘social’ facts founded on culturally accepted conventions or as ‘subjective’ facts resulting from the ‘projection’ of our desires and emotions upon aspects of our social and physical environments. Such subterfuges should be resolutely resisted by advocates of the objectivity of reasons for action. A wrong turning was taken in trying to force reasons for action into the mould of reasons for belief, for this is almost bound to result in the former being construed as facts or states of affairs of a peculiar and ill-defined kind. A better strategy by far for the objectivist is to find an altogether different ontological category to house reasons for action. And, indeed, the very language of action and practical reasoning suggests that precisely this is the right strategy to adopt.
10.3. Reasons for Action versus Reasons for Belief Belief is rightly classified as a propositional attitude whose aim is truth—and truth consists in correspondence to fact. I take this to be a truism, rather than a rash commitment to a specific theory of truth. That is to say, I am not suggesting that the concept of truth may be defined in terms of independently intelligible concepts of ‘correspondence’ and ‘fact’, though nor do I wish to rule out the possibility of such a definition. My point at present is just that the verb ‘believe’ may be complemented by a ‘that’clause and that when it is the case that S believes truly that p, it follows that it is a fact that p, or that what S believes is a fact. This is a purely semantic point which can hardly be denied by any theory of truth that aspires to respect the meanings of the words ‘believe’, ‘true’, and ‘fact’ 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 to this kind of content as ‘actional content’, in deliberate contrast with the ‘propositional content’ of beliefs and other so-called propositional attitudes. The virtue of this kind of analysis lies not only in the comparisons that it sustains between belief and action but also in the means it provides to distinguish perspicuously between actions, conceived as particular events, and the abstract items that form the ‘contents’ not only of actions themselves but also of those psychological states of an agent that are specifically ‘directed’ upon action—states such as intending and trying, which we could collectively call ‘conative’ states and which constitute the present-day equivalents of the volitions or acts of will favoured by early modern philosophers of mind and action such as Thomas Hobbes and John Locke. What we intend or try is to do such-and-such, where ‘such-and-such’ is a placeholder for what I am calling an ‘act’. (Of course, it is also sometimes what we desire, for we must distinguish between desiring that and desiring to.) I intend or try, for instance, to raise my arm. If I am successful, I perform an action which consists in my doing just what I intended or tried to do, namely, raise my arm. It is important here to appreciate that what I intend or try to do is not the action which, if I am successful, I will go on to perform. That action is a particular event, which has not yet occurred when I begin my attempt to act. To be successful in my attempt, I do not have to intend to bring about the very event which, as it turns out, will constitute my successful action. That would make intentional action virtually impossible for beings with our cognitive limitations (as indeed I pointed out in Chapter 5). For my attempt to be successful, it suffices that some action of mine should ensue which satisfies the content of my prior intention, which was to raise my arm. Any one of many—perhaps even infinitely many—numerically distinct possible actions could meet this requirement. This sort of consideration, then, appears to justify the kind of articulation of action-statements that I am proposing, with its distinctions between agent, action, and act. Acts, thus conceived, are abstract items in the same way—whatever exactly that is—that propositions are and must be distinguished clearly from the concrete events that are particular actions, just as propositions must be distinguished from the concrete states that are particular beliefs.⁵ ⁵ On the abstract/concrete distinction in general, see further my ‘The Metaphysics of Abstract Objects’, Journal of Philosophy 92 (1995), pp. 509–24.
208
Persons, Rational Action, and Free Will
However, there the similarity between acts and propositions ends. Propositions are expressed by that-clauses, or by whole sentences, whereas acts are expressed by infinitival verb-phrases, such as ‘[to] raise my arm’. This verb-phrase is apt, as we have seen, to express both what I may intend or try to do on a given occasion, and what it is that I succeed in doing when such an intention or attempt is successful. In both cases, it expresses a certain act. Because acts are not propositions, they cannot, of course, have truth-values. Propositions are true or false, and beliefs are correspondingly true or false in virtue of the truth or falsehood of their propositional contents. Because acts cannot have truth-values, however, reasons for action, unlike reasons for belief, cannot be facts. They must be items belonging to some other and suitable ontological category. What could that category be? The items in question must be ones which in some appropriate sense support actions, in a manner analogous to that in which facts support beliefs.
10.4. Objective Needs as Reasons for Action We do not have to look far for a suitable category of items. The category in question is surely the category of needs. These items have, as it were, precisely the right sort of ‘logical shape’ to constitute reasons for action. Again, a simple example will help. Just as the fact that distant ships disappear over the horizon supports the belief that the earth is round so, it might well be said, the need to eat supports the action of sowing corn. Or again, to return to our example of the falling slate, the need to avoid life-threatening injury supports the action of jumping out of the way of a falling slate. But what exactly is a need, then? This is a bit like asking: what exactly is a fact? By pointing this out, I don’t mean to suggest that the question is an illegitimate one, only that it is no less difficult than the corresponding question concerning facts and also that this difficulty should no more deter us from referring to needs and citing them as reasons than we should be deterred from doing the same in the case of facts. I hope it will not seem disappointing if I say that I shall not attempt to answer this difficult question in the present chapter. For my present purposes it suffices to identify needs as items belonging to a distinct ontological category and as having a ‘logical shape’ which suits them, unlike facts, to constitute reasons for action of an objective kind.
Needs, Facts, Goodness, and Truth 209 Some may press me at this point concerning my assumption that needs may be ‘objective’—by which I mean ‘mind-independent’, in the sense that a need is an item that can exist whether or not anyone is aware of its existence and so does not exist simply because one or more people believe that it exists. However, there are philosophers who would equally want to press my assumption that facts are ‘objective’ in this sense—anti-realists of various stripes. This is much too big an issue to take on here. And the same applies with regard to the corresponding question concerning needs. The important strategic advantage of identifying reasons for actions as needs rather than as a putative special class of ‘normative’ facts is that it puts needs and facts on the same footing with regard to such big questions of realism versus anti-realism.
10.5. The Logic of Action versus the Logic of Belief There is an obvious question which arises at this point, given that the parallels between reasons for action and reasons for belief so far advanced are accepted. What is it that stands to needs as truth stands to facts? We have noted that acts—the contents of actions—cannot have truth-values in the way that the contents of beliefs—propositions—do. Correspondingly, actions cannot be true or false in virtue of their contents, in the way that beliefs can. The obvious, and I think correct, thing to say is that the equivalent of truth in the domain of action is goodness, understood in a suitably broad sense which is not exclusively moral in import, but which includes moral goodness as a special case. Just as a true belief is one which corresponds to fact, so a good action is one which corresponds to need. In another idiom, just as facts are the truth-makers of true beliefs, so needs are the goodness-makers of good actions. We can extend the parallel between goodness and truth to capture what is similar and what is different between practical and theoretical reasoning, or the logic of action and the logic of belief. A valid argument in the logic of belief is one that is truth-preserving, while an inductively strong argument is one which preserves probability of truth. Analogously, then, we may say that a valid argument in the logic of action is one that is goodness-preserving, while an inductively strong argument is one which preserves probability of goodness. A simple example will serve to illustrate
210
Persons, Rational Action, and Free Will
the notion of goodness-preservation. Given that it is good to do both x and y, it follows that it is good to do x. For instance, given that it is good for me to both eat and drink, it follows that it is good for me to eat. However, we should not expect the logic of action to resemble in every respect the logic of belief, simply because goodness-makers are very different from truth-makers. Needs are very different from facts. One very important difference is that facts cannot conflict with one another, but needs can—both the needs of different people and different needs of the same person. Because facts cannot conflict with one another, truth is unitary and indivisible.⁶ But goodness, it seems, is not like that. Suppose, for example, that I am lost in the desert with a very small supply of water in my water-bottle. I need to drink, but I also need to conserve water. These needs conflict, however. The action of drinking would be good, as would be the action of conserving water, but I cannot perform both actions—at least, I cannot both drink all of the water in the bottle now and also conserve some of it for later use, or both conserve all of it for later use and drink some of it now. The good thing to do is to compromise, by drinking some of the water now and conserving the rest for later use. By contrast, compromise is never required in the domain of facts, because facts cannot conflict. Formally, this difference is reflected in the contrast between the following two inference-patterns, the first of which is valid and the second of which is not: ‘It is true that p and it is true that q, therefore it is true that p and q’, as opposed to ‘It is good to x and it is good to y, therefore it is good to x and y’. We have already seen one putative counterexample to the validity of the second inference-pattern. Here is another. Suppose it is agreed that it is good to be self-sufficient and that it is also good to give all of one’s possessions to the poor. It can hardly be concluded that it is good to both be self-sufficient and give all of one’s possessions to the poor, because it is impossible to do both. We have to choose between two incompatible courses of action, each of them good in its own way—or else we have to compromise. While we likewise have to choose between incompatible beliefs, however, we cannot regard each such belief as being true in its own way, nor can we compromise about truth. There is only one way in which beliefs can be true. Truth is single and indivisible, but goodness is not. 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.
10.6. Taking Needs Seriously What I have been proposing in this chapter is that we should take needs seriously, from an ontological point of view. Only then, I have suggested, shall we be able to understand clearly how the logic of action both
212
Persons, Rational Action, and Free Will
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 Anscombe, G. E. M., ‘Causality and Determination’, in her Metaphysics and the Philosophy of Mind: Collected Philosophical Papers, Volume II (Oxford: Blackwell, 1981). Armstrong, D. M., A World of States of Affairs (Cambridge: Cambridge University Press, 1997). Baker, L. R., Persons and Bodies: A Constitution View (Cambridge: Cambridge University Press, 2000). Bambrough, R., ‘Universals and Family Resemblances’, Proceedings of the Aristotelian Society 60 (1960/61), pp. 207–22. Beckermann, A., Flohr, H., and Kim, J. (eds), Emergence or Reduction? Essays on the Prospects of Nonreductive Physicalism (Berlin: de Gruyter, 1992). Bennett, J., Events and their Names (Oxford: Clarendon Press, 1988). Berkeley, G., A Treatise Concerning the Principles of Human Knowledge, in his Philosophical Works, ed. M. R. Ayers (London: Dent, 1975). Bondi, H., Cosmology, 2nd edn (Cambridge: Cambridge University Press, 1961). Brand, M. and Walton, D. (eds), Action Theory (Dordrecht: D. Reidel, 1976). Broad, C. D., ‘Determinism, Indeterminism, and Libertarianism’, in his Ethics and the History of Philosophy: Selected Essays (London: Routledge and Kegan Paul, 1952). Cartwright, N., Nature’s Capacities and their Measurement (Oxford: Clarendon Press, 1989). Chisholm, R. M., ‘The Agent as Cause’, in M. Brand and D. Walton (eds), Action Theory (Dordrecht: D. Reidel, 1976). Person and Object: A Metaphysical Study (London: George Allen and Unwin, 1976). Corcoran, K. J. (ed.), Soul, Body, and Survival: Essays on the Metaphysics of Human Persons (Ithaca, NY: Cornell University Press, 2001). Crane, T., ‘The Mental Causation Debate’, Proceedings of the Aristotelian Society (Supplementary Volume) 69 (1995), pp. 211–36. and Mellor, D. H., ‘There is No Question of Physicalism’, Mind 99 (1990), pp. 185–206. Dancy, J., Practical Reality (Oxford: Oxford University Press, 2000). Danto, A. C., Analytical Philosophy of Action (Cambridge: Cambridge University Press, 1973).
214 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: Routledge, 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 Hume, D., An Enquiry Concerning Human Understanding, ed. L. A. Selby-Bigge and P. H. Nidditch (Oxford: Clarendon Press, 1975). A Treatise of Human Nature, ed. L. A. Selby-Bigge and P. H. Nidditch (Oxford: Clarendon Press, 1978). James, W., Principles of Psychology (New York: Henry Holt & Co., 1890). Kane, R. (ed.), The Oxford Handbook of Free Will (Oxford: Oxford University Press, 2002). Kim, J., ‘ ‘‘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). ‘Events as Property Exemplifications’, in M. Brand and D. Walton (eds), Action Theory (Dordrecht: D. Reidel, 1980). ‘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). Mind in a Physical World: An Essay on the Mind–Body Problem and Mental Causation (Cambridge, MA: MIT Press, 1998). ‘The Non-Reductivist’s Troubles with Mental Causation’, in J. Heil and A. Mele (eds), Mental Causation (Oxford: Clarendon Press, 1993). Physicalism, or Something Near Enough (Princeton, NJ: Princeton University Press, 2005). Supervenience and Mind: Selected Philosophical Essays (Cambridge: Cambridge University Press, 1993). Lewis, D. K., Counterfactuals (Oxford: Blackwell, 1973). Libet, B., Neurophysiology of Consciousness: Selected Papers and New Essays (Boston, MA: Birkh¨auser, 1993). ‘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. Freeman, A., and Sutherland, K. (eds), The Volitional Brain: Towards a Neuroscience of Free Will (Thorverton: Imprint Academic, 1999). Lockwood, M., Mind, Brain and the Quantum: The Compound ‘I’ (Oxford: Blackwell, 1989). Lowe, E. J., ‘Against an Argument for Token Identity’, Mind 90 (1981), pp. 120–1. ‘Agent Causation’, in The Encyclopedia of Philosophy, 2nd edn, ed. D. M. Borchert (New York: Macmillan, 2006). ‘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).
216 Bibliography Lowe, E. J., ‘For Want of a Nail’, Analysis 40 (1980), pp. 50–2. The Four-Category Ontology: A Metaphysical Foundation for Natural Science (Oxford: Clarendon Press, 2006). ‘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). An Introduction to the Philosophy of Mind (Cambridge: Cambridge University Press, 2000). Kinds of Being: A Study of Individuation, Identity and the Logic of Sortal Terms (Oxford: Blackwell, 1989). Locke on Human Understanding (London: Routledge, 1995). ‘The Metaphysics of Abstract Objects’, Journal of Philosophy 92 (1995), pp. 509–24. The Possibility of Metaphysics: Substance, Identity, and Time (Oxford: Clarendon Press, 1998). Subjects of Experience (Cambridge: Cambridge University Press, 1996). A Survey of Metaphysics (Oxford: Oxford University Press, 2002). ‘The Truth About Counterfactuals’, Philosophical Quarterly 45 (1995), pp. 41–59. McCall, S. and Lowe, E. J., ‘Indeterminist Free Will’, Philosophy and Phenomenological Research 70 (2005), pp. 681–90. McGinn, C., ‘Consciousness and Space’, Journal of Consciousness Studies 2 (1995), pp. 220–30. The Problem of Consciousness (Oxford: Blackwell, 1991). Mackie, J. L., Ethics: Inventing Right and Wrong (Harmondsworth: Penguin Books, 1977). Melden, A. I., Free Action (London: Routledge and Kegan Paul, 1961). Melnyk, A., ‘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). Menzies, P., ‘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). Merricks, T., Objects and Persons (Oxford: Clarendon Press, 2001). Mills, E., ‘Interactionism and Overdetermination’, American Philosophical Quarterly 33 (1996), pp. 105–17. Noordhof, P., ‘The Overdetermination Argument versus the Cause-and-Essence Principle—No Contest’, Mind 108 (1999), pp. 367–75. O’Connor, T., Persons and Causes: The Metaphysics of Free Will (New York: Oxford University Press, 2000).
Bibliography 217 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
Wittgenstein, L., Tractatus Logico-Philosophicus, trans. C. K. Ogden (London: Routledge and Kegan Paul, 1922). Woodward, J., Making Things Happen: A Theory of Causal Explanation (New York: Oxford University Press, 2003). Woolhouse, R. S., ‘Leibniz’s Reaction to Cartesian Interactionism’, Proceedings of the Aristotelian Society 86 (1985/86), pp. 69–82. Yaffe, G., Liberty Worth the Name: Locke on Free Agency (Princeton, NJ: Princeton University Press, 2000). Manifest Agency: Thomas Reid’s Theory of Action (Oxford: Clarendon Press, 2004).
Index abstract/concrete distinction 207 act/action distinction 206–8 action-results 148, 150, 156 actions 180, 206–7 akratic 193 basic 7, 124–33, 146, 151, 153, 162, 172 free 6, 116, 184, 186, 188, 191 instinctive 201, 203 intentional 24, 122, 172–5 rational 175–6, 181, 182, 186, 189, 200–1, 203 reasonable 182, 200–1 voluntary 101 activeness (versus passiveness) 173, 176 agency 23 agent causalism (classical) 6–7, 151–4, 162–3, 172 agents 2, 6, 19, 122, 125, 127, 133, 138, 159–61, 166, 180, 186, 206 Alice thought experiment 190–4, 195 alternatives (for action) 188 animalism 167–8 animism 161 Anscombe, G. E. M. 56n Aristotle 149n, 155n Armstrong, D. M. 36n, 212 automatisms 82 Baker, L. R. 95n, 168n Bambrough, R. 140n beliefs 9, 10, 34, 35, 130–1, 163, 172, 173–5, 180–3, 189, 200–2, 206 mistaken 184–5 paranoid 181–2 rational 197–8 Bennett, J. 42n, 110n Berkeley, G. 5, 133, 165 Berm´udez, J. L. 106n Big Bang 27, 28, 47 binding problem 53 bodies 21, 94–8
Bondi, H. 41 brains 22, 94, 97–8, 170–1 Broad, C. D. 7n Buridan’s ass situation 194 causal closure argument 12, 13, 14, 42, 80, 99–101 closure principles 42–56, 61–5, 117, 176–8 overdetermination 26, 30, 42–5, 48–9, 63–72, 80 powers: see powers relations 142, 144 causation 2, 6, 9, 24, 114, 134 agent 3, 15, 16, 121–40, 141, 146–7 backward 132–3 downward 177 event 3, 12, 15, 42, 54, 121–40, 141–7, 154, 159, 162, 166, 174 fact 42, 54, 110, 121 intentional 14, 39–40, 110–11, 117–18 mental 14–15, 35 physical 35, 110 probabilistic 45–6, 54, 66, 128–9, 150 simultaneous 48–50, 54, 70–1, substance 16, 141–9, 154–5, 159, 161–5 transitivity of 26, 47–8 causative action verbs 124–5, 127, 139–40, 148 chance 6, 8, 11, 44, 66–7, 129, 131, 150, 155, 160–1, 176–7, 187, 190–5 Chisholm, R. M. 98n, 121n choice 9, 56, 67, 102, 116, 131, 155–7, 175–6, 181–4, 186, 187–97 coincidence 14, 27–32, 36–9, 51–3, 57, 69, 77, 111, 134 compatibilism 68 composition 22 compromise 210 concepts 93
220 index conceptual priority (of agent causation) 133–5 conservation laws 41–3, 57, 59–62 content (intentional) 36, 89–90, 106, 113, 174–5, 181, 206–207 control 6, 35, 82, 107, 110, 190, 195–7 illusions of: see illusion counterfactual conditionals 31, 73–4, 103–17 context-dependency of 112–14 Crane, T. 93n, 110n Dancy, J. 10, 116, 175n, 181n, 185n, 186, 199 Danto, A. C. 125n, 146n, 162n Davidson, D. 8, 9, 121n, 130n, 131n, 152, 172n, 182–3, 200, 202 decisions 67–8, 102–11, 116, 188 Deecke, L. 102n deliberation 85, 90, 101, 188, 192–4 Dennett, D. C. 85n Descartes, R. 20, 59–62, 86, 95, 167, 168 desires 9, 10, 35, 38–9, 130–1, 163, 172, 173–5, 184, 186, 189, 200–2, 204 determinism 27, 46–7, 117–18, 128 pragmatic inconsistency of 197–8 deviant causal chains 9, 182–3, 202 disappearance of the agent 159–61 dispositions (see also powers) 165–6, 169 Dretske, F. 54n dualism 13, 19, 26, 40, 41–65, 70–1, 77–9, 92–3, 99, 168 non-Cartesian substance (NCSD) 13, 92–110, 117–18 Eccles, J. C. 46, 102n economy (ontological) 75–7 Einstein, A. 61, 62 eliminativism 34, 39–40 emergentism 14, 41, 47, 57, 92 energy 41, 42, 60, 61, 71 epiphenomenalism 15, 34, 79–82, 91, 99, 100n events 2, 4, 37, 93, 122, 131–2, 133, 138–9, 152, 160, 164–5 evolution 76 experimentation 24, 87–8, 135 explanation causal 37–9, 117, 132, 189–90
rational 132, 152, 189–90 externalism (regarding reasons for action) 16, 175, 180, 184–6, 199, 202, 203 fact/value distinction 204–5 facts 184–5, 203, 205–6, 208–10 moral 187, 204–5 family resemblances 140 first-person thoughts 21 forces 60–2 Foster, J. 95n Frankfurt, H. G. 196n Frankfurt-style cases 196 free will 67, 84, 128–9, 130–1, 141, 151, 154–7 functionalism 80 fusions (of events) 37–8, 104 Garber, D. 59n Geach, P. T. 22n Ginet, C. 81n, 148n, 172n God 55, 59, 165, 186, 192 Gold, T. 41 goodness 209–11 gravitation 61, 62, 149, 164, 169 Hacking, I. 87n Hardcastle, V. G. 53n Hart, W. D. 42, 61n Hawthorne, J. 115n Hobbes, T. 207 Hornsby, J. 137n Hume, D. 31n, 87–8, 167, 203 identity 23 criteria and conditions of 23, 92, 169 illusion cognitive 89–90 of control 82–3, 86, 90 individuation 106, 211 innateness 25 intentionality (see also content) 34–5 interests (of an agent) 186, 187, 204 internalism (regarding reasons for action) 181, 202 intervention 24, 160 invisibility (of mental causation) 15, 52, 58, 74, 77–8, 118
index 221 involvement (of agents in events) 123–4, 126, 127, 130, 142–3, 145, 163
Olson, E. T. 167n Owens, D. 28n
James, W. 86
panpsychism 76 Papineau, D. 20n, 43–4, 46, 63n, 66n paralysis 86–7 Peacocke, C. A. B. 42n, 49–50 persons (see also selves) 5, 15, 19–20, 92–6, 147, 159, 165–71, 180 phantom pains 96 physicalism 11, 12, 13, 40–1, 59, 69, 153 Plantinga, A. 184 Popper, K. 46, 102n possible worlds 29, 103, 192 powers (see also dispositions) 4, 92–3, 99–100, 107–8, 118, 122, 138, 143, 148–51, 154–7, 159, 164–6, 168–9, 173 rational 155–7, 176–8 practical reasoning 16, 187 probability (see also chance) 209 properties (mental and physical) 93–5 propositions 206, 208 Pythagoreanism 23
Kim, J. 48n, 53n, 63n, 92n, 93n, 100n, 107n, 108n, 109n, 117n, 177n knowledge (and causality) 80, 87–8, 134–5 Kornhuber, H. H. 102n laws (causal or natural) 9, 31, 44, 55–6, 160–1 Leibniz, G. W. 60 Lewis, D. K. 103n libertarianism 191–4, 202 Libet, B. 84, 85, 104, 178n Locke, J. 149n, 167, 173n, 200, 207 locked-in syndrome 24, 86 Lockwood, M. 65n logic (of action and of belief) 209–11 Mackie, J. L. 187, 205n McCall, S. 194n McGinn, C. 23n, 78n means (causal) 124–5, 126 Melden, A. I. 83n Mellor, D. H. 93n Melnyk, A. 76–7 mental causation: see causation representation 179 states: see states Menzies, P. 107n, 108n, 110n, 114n mereology 63, 64, 69 Merricks, T. 169n Mills, E. 42n momentum 42, 60, 61 multiple thinkers problem 169–71
quantum physics 46, 54, 62, 65–7, 118, 177 queerness, argument from 204–5
naturalism 161–3, 179, 187, 205, 212 needs 202, 208–9, 211–12 neural events 102–12, 132–3, 137–8, 154 Newton, I. 61, 62, 71 Noordhof, P. 44 numbers 22, 23
radioactive decay 11, 65–6, 150, 155, 176–7, 192, 194 realization 69, 70, 75, 107, 108–9 reasons for action 2, 8–10, 85, 116, 129, 130–1, 152, 155–6, 174–5, 177, 180–7, 189–90, 201–9 for belief 203–6 moral 187 prudential 187 redundancy (of causes) 72–4 regress, infinite 184 Reid, T. 7n, 151n replay argument 191–4 responsiveness (to reasons) 176, 194, 195 Robinson, W. S. 100n robots 39
O’Connor, T. 7n, 121n, 129n, 151n, 152n, 162n
Scheid, P. 102n Schr¨odinger’s cat 65–6
222 index Searle, J. R. 68n, 156n, 175n, 183n self-reference 20, 23–4 selves (see also persons and subjects) 19–24, 96 sensations 8, 35 Seuren, P. M. 139n Shoemaker, S. 31n simplicity (of the self) 96n, 168 Sorabji, R. 28n souls 21 special theory of relativity 49n spontaneity (of action or power) 7, 9, 11, 65, 84, 126–8, 130, 150, 155, 176–8, 192 states (mental and physical) 22–6, 38, 93, 96–9 states of affairs 10, 36, 184–5, 201, 203 Strawson, G. 24n Strawson, P. F. 5, 97n, 136n Sturgeon, S. 44 subjects (of thought and experience) 13, 15, 21, 92–4, 97–8, 170–1, 180 substances (individual) 4, 93, 99, 122, 142, 143–4, 159, 165–71 supernatural 27, 40, 47 Swinburne, R. 95n
Taylor, R. 121n, 141n, 151n, 162n Thomson, J. J. 37n, 139n thoughts 21, 89, 98, 169–71 Tooley, M. 49n transitive verbs 137, 144 truth 205–6, 209–11 trying 81n, 172, 207 unity argument 95–9, 107, 171 values 204–5 van Inwagen, P. 98n, 128n, 190 veto power 84 volitions 3, 7–9, 15, 79–90, 104, 131, 148, 152, 154, 157, 163, 171–4, 177–8, 207 von Wright, G. H. 25n, 87n, 132n, 134n Wegner, D. M. 82n, 83, 88n will 10, 12, 148–51, 154–7, 173–8 willing (see also volitions) 147, 153 Wittgenstein, L. 212 Woodward, J. 87n Woolhouse, R. S. 59n Yaffe, G. 7n, 160n