231 106 1MB
English Pages 246 Year 2014
Libertarian Free Will
Libertarian Free Will Contemporary Debates
Edited by David Palmer
1
1 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford 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 certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016 © Oxford University Press 2014 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. Library of Congress Cataloging-in-Publication Data Libertarian free will : contemporary debates / [edited by] David Palmer. pages cm Includes bibliographical references and index. ISBN 978-0-19-986008-1 (bb : alk. paper) 1. Free will and determinism. 2. Libertarianism. I. Palmer, David (Professor) editor of compilation. BJ1461.L53 2014 123'.5—dc23 2014006492
1 3 5 7 9 8 6 4 2 Printed in the United States of America on acid-free paper
Con t e n t s
Acknowledgments vii Contributors ix 1. Free Will, Libertarianism, and Kane David Palmer
3
Part I: Libertarian Theories of Free Will 2. Can an Indeterministic Cause Leave a Choice Up to the Agent? 15 Carl Ginet 3. Free Will and Metaphysics Timothy O’Connor
27
Part II: The Luck Objection 4. Kane, Luck, and Control: Trying to Get by without Too Much Effort 37 Alfred R. Mele 5. Toward a Solution to the Luck Problem John Martin Fischer
52
Part III: Incompatibilism and Omissions 6. Compatibilist Ultimacy: Resisting the Threat of Kane’s U Condition 71 Michael McKenna 7. The Direct Argument for Incompatibilism David Widerker and Ira M. Schnall
88
8. Freedom, Responsibility, and Omitting to Act Randolph Clarke
107
( vi ) Contents
Part IV: The Significance of Free Will 9. Responsibility for Emotions, Alternative Possibilities, and Reasons 127 Ishtiyaque Haji 10. Moral Responsibility, the Reactive Attitudes, and the Significance of (Libertarian) Free Will 142 Dana Kay Nelkin 11. The Dialectic of Selfhood and the Significance of Free Will Derk Pereboom
161
Part V: Kane’s Reply 12. New Arguments in Debates on Libertarian Free Will: Responses to Contributors 179 Robert Kane References 215 Index 225
Ack n owl e d gm e n t s
Many people have helped me in putting together this book and I should like to thank them here. First, I thank the authors for their contributions and for their patience in responding to my requests for changes to their drafts. Second, I thank John Martin Fischer for helping me at the beginning of this project. Third, I would like to thank the staff at Oxford University Press, in particular Peter Ohlin, Lucy Randall, and Emily Sacharin, and also the anonymous reviewers who gave good suggestions regarding revisions. Fourth, I would like to thank Trevor Hedberg, a talented philosophy graduate student at the University of Tennessee, for helping me prepare the chapters, the reference section, and the index. Fifth, I thank my colleagues at the philosophy department at the University of Tennessee for giving me a supportive environment in which to work. Sixth, I would like to thank Robert Kane, a model philosopher and teacher. Any good work I have done has its roots in his fine mentorship. Finally, I would like to thank Jordan Palmer, my favorite critic.
Con t r ib u tor s
Randolph Clarke, Professor of Philosophy, Florida State University John Martin Fischer, Distinguished Professor of Philosophy and University of California President’s Chair (2006–2010), University of California, Riverside Carl Ginet, Professor of Philosophy Emeritus, Cornell University Ishtiyaque Haji, Professor of Philosophy, University of Calgary Robert Kane, University Distinguished Professor of Philosophy Emeritus and Professor of Law, University of Texas at Austin Michael McKenna, Professor and Keith Lehrer Chair of Philosophy, University of Arizona Alfred R. Mele, William H. and Lucyle T. Werkmeister Professor of Philosophy, Florida State University Dana Kay Nelkin, Professor of Philosophy, University of California, San Diego Timothy O’Connor, Professor of Philosophy, Indiana University David Palmer, Assistant Professor of Philosophy, University of Tennessee Derk Pereboom, Susan Linn Sage Professor of Philosophy, Cornell University Ira M. Schnall, Lecturer, Bar-Ilan University David Widerker, Professor of Philosophy, Bar-Ilan University
Libertarian Free Will
•
Chapter 1
Free Will, Libertarianism, and Kane David Palmer
O
ne of the central topics in philosophy, both historically and in the present, is free will. The basic puzzles of this topic are easily felt. For instance, it is easy to wonder whether factors beyond our control—our genetic constitution, the environment in which we were brought up, and so on—might be among the causes of our behavior. In the light of this, we might wonder whether it is really possible for us to act freely and be morally responsible for our actions or, instead, whether everything we do is ultimately shaped by these factors in such a way that undermines our free will and moral responsibility. In contemporary philosophical discussions, this concern is crystallized as a concern about the relationship between free will and moral responsibility on the one hand and causal determinism on the other hand.1 On this issue, the basic divide among philosophers is between compatibilism and incompatibilism. Compatibilists believe that free will and moral responsibility are compatible with the truth of causal determinism. By contrast, incompatibilists believe that free will and moral responsibility are incompatible with causal determinism. According to incompatibilists, if our actions are causally determined, then we cannot act freely and cannot be morally responsible for what we do.
1. Causal determinism is the view that, for any given time t, a complete statement of the facts at t, together with a complete statement of the laws of nature, entails every truth as to what happens after t (Fischer and Ravizza 1998, p. 14; van Inwagen 1983, pp. 2–8).
( 4 ) Libertarian Free Will
The central focus of the essays in this volume is the position known as libertarianism. Libertarianism about free will and moral responsibility is the view that people sometimes act freely and responsibly, but this freedom and responsibility are incompatible with causal determinism.2 Thus, libertarians are incompatibilists who believe that free will and moral responsibility exist. In short, if libertarianism is true, then people sometimes act without being causally determined to do so. Frequently maligned within the history of philosophy, this view has recently gained increasingly sympathetic attention among philosophers. But stark questions remain: How plausible is this view? If our actions are not causally determined, how can we have control over them? Why should we want our actions to be breaks in the deterministic causal chain? The recent resurgence of interest in libertarianism is due, most significantly, to Robert Kane, who is the leading contemporary defender of this view of free will and moral responsibility. In fact, Kane’s pioneering work, spanning a number of books and articles, has helped push libertarianism to the forefront of research in metaphysics, ethics, and philosophy of mind. His libertarian view initially emerged in Free Will and Values (Kane 1985), one of his earlier works, but it reached its maturity in his crowning work, The Significance of Free Will (Kane 1996). This book has variously been described as “one of the very finest books on free will and responsibility” (Pereboom 2000, p. 426), “quite simply, the most thoughtful and detailed defense of libertarianism currently available” (Mele 1998, p. 581), and “unrivalled” as a presentation of libertarianism (Nathan 1999, p. 180). Since the publication of The Significance of Free Will, Kane has continued to refine his view still further in a series of highly influential journal articles and book chapters (see, especially, Kane 1999b, 2000c, 2002a, 2003, 2011a), a contribution to a multiauthored book (Fischer, Kane, Pereboom, and Vargas, 2007) and a sole-authored introductory book (Kane 2005).3 Kane’s libertarian theory is nuanced. He allows that people can be morally responsible for actions that are causally determined by their character. But he insists that at least some of the prior actions that help shape our character must not be causally determined. He calls these causally undetermined actions “self-forming actions” or SFAs since it is by virtue of these 2. Libertarianism about free will and moral responsibility should, of course, be distinguished from libertarianism as a political doctrine. Libertarianism as a political doctrine is, roughly, the view that justice accords highest priority to respect for a wide range of basic liberties (see John Locke (1690) and Robert Nozick (1974) for classic historical and more contemporary statements of this sort of view). 3. In addition, Kane has also edited two editions of the Oxford Handbook of Free Will (Kane 2002c, 2011b) as well as a collection of readings on free will (Kane 2002a).
F ree W ill , L ibertarianism , and K ane
( 5 )
actions that we, in part, form the character or self that we have. By contrast, if all of the actions that shape our character were causally determined, then, Kane (1999b) says, “there would have been nothing we could have ever voluntarily done to make ourselves different than we are—a condition that I think is inconsistent with our having the kind of responsibility for being what we are which genuine free will requires” (p. 224). These SFAs occur at moments when people are conflicted about what to do in such a way that their presently formed characters do not causally determine for them which action they will perform. To borrow one of his examples, suppose that a businesswoman is on her way to an important business meeting when she comes across an assault victim in need of help (Kane 1996, p. 126). Suppose that she is torn between doing the moral thing (stopping and helping) and doing the self-interested thing (carrying on to her meeting), and her presently formed character leaves it causally open which choice she will make. She sees the reasons to do both and regards both sets of reasons as having roughly equal weight. Kane argues that, in the light of her conflict ing motivation, the businesswoman makes efforts to perform each action. If her subsequent action is to be a free and responsible action, then her effort, along with her reasons, nondeterministically cause her action. This and other proposals by Kane have been hugely influential in shaping contemporary discussions of libertarianism and of free will and moral responsibility in general. But his view and libertarianism more generally remain controversial. It is this controversy that forms the backdrop for this book. What follows are new essays on the libertarian position on free will and related issues that focus specifically on Kane’s views. Written by a distinguished group of philosophers, the essays cover various areas of philosophy including metaphysics, ethics, and philosophy of mind. Kane contributes a final essay, replying to the criticisms offered in the previous chapters and developing his view in new directions. The essays are divided into five sections: libertarian theories of free will, the luck problem, incompatibilism and omissions, the significance of free will, and Kane’s reply. In the rest of this chapter, I describe each of the essays under these headings and summarize their key claims.
1. Libertarian Theories of Free Will
In the contemporary free will debate, three prominent libertarian views have emerged: (i) the event-causal account, according to which a person’s free and responsible actions are caused solely by prior events or states of which
( 6 ) Libertarian Free Will
she is the subject, and there is some indeterminism in the causation of her actions by these antecedent events or states;4 (ii) the agent-causal account, according to which a person’s free and responsible actions are caused by her as a substance, where causation by a substance is not reducible to, nor composed of, causation by prior events or states; and (iii) the non-causal account, according to which a person’s free and responsible actions are not caused at all: they are neither caused by any antecedent events or states of which she is the subject nor are they caused by her as a substance. Kane’s is an event-causal account. According to Kane, if a person acts freely and responsibly (at least with respect to her SFAs), it is the event of her making an effort to perform it and the event of her having reasons to perform it that nondeterministically cause her action. But according to the first two essays in this volume, the event-causal account is not the most promising version of libertarianism. Carl Ginet (chapter two) argues for a non-causal libertarian account. Ginet is one of the main defenders of a noncausal libertarian view (see, for instance, Ginet 1990, 1997, 2003, 2007).5 In his essay, Ginet explains why he thinks that his non-causal view is to be preferred over Kane’s event-causal view. He also discusses how libertarians should respond to the so-called luck objection (to be introduced in the next section), and he considers the strength of our reasons for believing that uncaused actions might be found in the natural order. By contrast, Timothy O’Connor (chapter three) argues for an agentcausal libertarian account. O’Connor is one of the central proponents of the agent-causal position (see, for instance, O’Connor 2000, 2011). According to O’Connor, if we take seriously two metaphysical assumptions made, at least implicitly, by Kane—the view that persons are not immaterial minds and the view that causation is a real, nonreductionist relation—then both of these assumptions suggest, most naturally, an agent-causal libertarian account rather than an event-causal libertarian account favored by Kane. O’Connor then develops what he calls a “neo-Aristotelian” account of substance
4. Some event-causal libertarian views locate the indeterminism at an earlier stage than the causation of the agent’s action or decision by prior events or states. For instance, these views propose that while the action or decision might be causally determined by prior events or states, what is causally undetermined is what beliefs will come to mind when the agent is deliberating about what to do (e.g., Mele 1995). 5. Non-causal libertarian views are also defended by Stewart Goetz (1988, 2008) and Hugh McCann (1998b), among others.
F ree W ill , L ibertarianism , and K ane
( 7 )
causation, an account he thinks that Kane really ought to be committed to given the two metaphysical assumptions that Kane makes.
2. The Luck Objection
Perhaps the most significant objection to libertarian views of free will and moral responsibility is the “luck” objection, the objection that if people’s actions are not causally determined, then their actions or crucial facts about their actions become matters of luck or chance in a way that undermines their freedom and responsibility. In the first part of his essay, Alfred R. Mele (chapter four) introduces this objection in more detail and critically examines Kane’s response to it. Kane’s response rests, most centrally, on the claim that agents make “efforts of will” when acting freely and responsibly so that, whichever way they act, they voluntarily and rationally do something that they were trying to do. Mele responds to Kane’s proposal by arguing that people can act freely and responsibly only if these efforts themselves are freely made. In the second part of his chapter, Mele contrasts Kane’s event-causal libertarian view with his own event-causal libertarian view that does not appeal to efforts of will. Mele then defends both views against the objection that neither view provides people with any more control over their actions than they would have if their actions were causally determined. In the chapter five, John Martin Fischer develops his own solution to the luck objection. Unlike Kane, Fischer thinks that moral responsibility is compatible with the truth of causal determinism, but he also wants to show that moral responsibility is compatible with the falsity of causal determinism.6 In short, Fischer wants to show that people can be morally responsible for their actions whether or not causal determinism is true. With this as background, Fischer’s own solution to the luck objection turns on the claim that what is important for moral responsibility is that the prior states of the agent lead to her choice in the right kind of way. Fischer argues that the falsity of causal determinism, or the fact that indeterminism might make it a matter of luck that the person acts as she does rather than doing something else, is irrelevant to whether the person’s prior states lead 6. Fischer is officially agnostic on whether free will—understood as the freedom to do otherwise—is compatible with determinism, although he is inclined to think that it is not. Hence, Fischer calls his overall position “semicompatibilism,” the view that moral responsibility is compatible with causal determinism even if the freedom to do otherwise is not compatible with causal determinism. (For details, see Fischer 2006a, 2012.)
( 8 ) Libertarian Free Will
to her choice in the right kind of way. Hence, he argues that moral responsibility is compatible with the falsity of causal determinism as well as being compatible with the truth of causal determinism.
3. Incompatibilism and Omissions
Libertarianism is an incompatibilist view. But what are the most promising arguments for incompatibilism and how plausible are they? The traditional argument for incompatibilism about moral responsibility and determinism is the “alternative possibilities” or “leeway” argument: (1) People are morally responsible for their actions only if they could have done otherwise (often called the “principle of alternative possibilities”). (2) If determinism is true, then people cannot do otherwise. (3) Therefore, if determinism is true, people cannot be morally responsible for their actions. This is not, however, the argument for incompatibilism that Kane defends. Instead, Kane defends a version of the “source” or “ultimacy” argument for incompatibilism: (1) People are morally responsible for their actions only if they are the ultimate source of their actions. (2) If determinism is true, then people cannot be the ultimate source of their actions. (3) Therefore, if determinism is true, people cannot be morally responsible for their actions. Michael McKenna (chapter six) puts this source or ultimacy argument under the microscope and offers a nuanced assessment of it. McKenna argues that compatibilists should reject premise (2) of the argument. He argues against Kane that even if determinism is true, people can still be the ultimate source of their actions. He argues for this on the grounds that determining the ultimate source of something is sensitive to context. Moreover, if the relevant context is the discourse of ordinary people, then it is plausible to argue that determinism’s truth would not undermine our being the ultimate source of our actions. Hence, (2) can be rejected, and so Kane’s preferred argument for incompatibilism fails. David Widerker and Ira M. Schnall (chapter seven) discuss a further argument for incompatibilism, one they think has significant dialectical
F ree W ill , L ibertarianism , and K ane
( 9 )
advantages over the two arguments considered so far. Here is a simplified version of it: (1) If determinism is true, then the remote past (the past prior to the existence of any human beings) together with the laws of nature entail people’s actions. (2) No one is morally responsible for the remote past, no one is morally responsible for the laws of nature, and no one is morally responsible for the fact (if it is a fact) that the remote past together with the laws of nature entail people’s actions. (3) Therefore, if determinism is true, people cannot be morally responsible for their actions. This is known as the “direct” argument for incompatibilism. The argument is direct because, unlike the other two arguments, it does not rest on any claims about what moral responsibility requires: it does not rely on the claim that responsibility requires the freedom to do otherwise (as the alternative possibilities argument does) nor does it rely on the claim that responsibility requires that people are the ultimate source of their actions (as the source argument does). This directness is a dialectical advantage, Widerker and Schnall claim, since many compatibilists dispute these two claims about what moral responsibility requires. After distinguishing these three incompatibilist arguments, Widerker and Schnall critically evaluate some of the recent controversy surrounding the direct argument and offer a partial defense of it, contending that many of the alleged criticisms of it can be avoided. In the final essay in this section, Randolph Clarke (chapter eight) addresses the topic of omissions. We tend to think that as well as acting freely, people can also freely refrain from doing something and freely omit to do something. Moreover, we tend to think that people can be morally responsible for what they refrain from doing and what they omit to do. But what does it mean to freely omit to act or freely refrain from acting, and how might people be morally responsible for what they omit to do or refrain from doing? Clarke takes up these questions in his essay. As Clarke explains, freedom and responsibility for omissions is a significant topic for all camps in the free will debate, compatibilist and incompatibilist. With this in mind, Clarke aims to offer answers to these questions that are neutral with respect to the issue of whether free will and moral responsibility are compatible with determinism. However, Clarke pays special attention to some of Kane’s suggestions about the nature of omissions, comparing his own views to those suggested by Kane.
( 10 ) Libertarian Free Will 4. The Significance of Free Will
One of the central issues in Kane’s work is to explain why free will—especially a libertarian free will—is either significant or something worth wanting. Ishtiyaque Haji (chapter nine) takes up this issue and considers whether free will is required for us to have the attitudes associated with forgiveness and guilt. He argues that one cannot properly forgive or feel guilt unless one has pro tanto reasons of a certain sort. He then argues that having these pro tanto reasons entails that one could have done otherwise. So, our having the attitudes associated with forgiveness and guilt presupposes our having alternative possibilities. Haji then argues that if determinism rules out alternative possibilities, then determinism would similarly rule out our having the attitudes associated with forgiveness and guilt (though he is noncommittal on whether determinism does in fact rule out alternative possibilities). The remaining two essays by Dana Kay Nelkin (chapter ten) and Derk Pereboom (chapter eleven) investigate Kane’s arguments that a libertarian free will is significant and worth wanting. Kane begins by arguing that this sort of free will is valuable because it is required for a number of things that we value: moral responsibility, genuine creativity, autonomy, true desert for one’s achievements, being the suitable object of reactive attitudes, dignity or self-worth, a true sense of individuality or uniqueness as a person, genuine love and friendship, life-hopes requiring an open future, and the ability to say in the fullest sense that one acts of one’s own free will. But Kane concedes that compatibilists can argue that a compatibilist free will gives us versions of these things even if such a free will cannot underwrite these things in their exalted sense. In response to this concession, Kane gives two reasons why, despite the arguments of compatibilists, people nevertheless tend to value a libertarian free will: first, they desire to be independent selves, and second, they desire that their achievements have objective worth. In her essay, Nelkin argues that offering reasons why people tend to think that a libertarian free will is valuable is different from giving reasons to think that such a free will is in fact valuable. In the light of this, Nelkin suggests taking a different approach to uncovering free will’s significance. Rather than beginning by considering a libertarian free will, assuming it is valuable, and then asking what underwrites its value, she suggests that we instead begin by considering the things that all sides agree are valuable—moral responsibility, genuine creativity, autonomy, and so on. We should then explore whether these things really are valuable and then see whether they require a libertarian free will. Nelkin focuses on two valuable goods as case studies for her approach: moral responsibility and personal relationships.
F ree W ill , L ibertarianism , and K ane
( 11 )
In the first part of his essay, Pereboom evaluates Kane’s claim that we tend to think a libertarian free will is valuable because we desire to be independent selves and we desire that our achievements have objective worth. He argues that this picture is more complicated and suggests that a compatibilist free will might be able to satisfy both of these desires at least to some degree. Even if determinism were true, we might still possess a sort of freedom of action that would allow us to distinguish ourselves as being different from others and from the world itself, and our achievements might still have some degree of objective worth. Pereboom then turns to examine whether love and personal relationships might be undermined in the absence of a libertarian free will. He defends the view that these two things, at least to a significant degree, can still be had by us even if we were to lack a libertarian free will.
5. Kane’s reply
The final section of the book consists of Kane’s reply to the previous chapters. Kane discusses the criticisms of his view under three headings: the compatibility question (are free will and moral responsibility compatible with determinism?), the intelligibility question (can we make sense of a libertarian free will without reducing freely willed actions to mere matters of chance or luck?), and the significance question (why do we, or should we, want free will—especially a libertarian free will?). Along the way, Kane breaks new ground and further develops his views. The book thus closes in the spirit of philosophy that is keenly associated with Kane: a careful pursuit of truth shaped by a sympathetic consideration of the thoughts of critics. The points of disagreement with critics are never made simply to “score a point”—they are made while recognizing that philosophy is properly a cooperative enterprise. Kane’s central aim is to help us all make progress in answering, with greater clarity, the enduring philosophical puzzles of free will.
Pa r t I
••• Libertarian Theories of Free Will
•
Chapter 2
Can an Indeterministic Cause Leave a Choice Up to the Agent? C arl Ginet
L
ike Kane, I am an incompatibilist. We hold that there is a kind of freedom of choice (or action or decision) worth having that cannot obtain in a deterministic world. Indeed we agree on a more specific incompatibilist thesis: that a choice that was deterministically caused is one such that it cannot have been up to the subject at the time it occurred whether it would occur or not. But Kane’s incompatibilism does not, I believe, go as far as the incompatibilism I have come to believe in.1 I want to explain why that seems to me to be so and why the more extreme incompatibilism seems to me the better view. Let E be a specification of a particular event and let C be a description of all of the circumstances obtaining prior to this E that are causally relevant to its occurrence. Three cases are possible: (1) E is deterministically caused by (something in) C, (2) E is indeterministically caused by C, or (3) E is not caused at all (by anything). (I here ignore the idea of agent-causation of a decision (or other action) which some think is a real possibility distinct from all of these three, because neither I nor Kane believe it to be such a possibility. I think that talk 1. Which is “old time” incompatibilism: the freedom we want has to be “contracausal” freedom, to use C. A. Campbell’s term (Campbell 1951).
( 16 ) Libertarian Theories of Free Will
of an agent causing her decision is best understood as expressing a possibility that falls under (3), namely that, because the decision was not (event-) caused, it was entirely up to the agent at the time it was made: it was the agent determining what her decision would be simply by deciding.) I assume that causation is a real relation between particulars, that it is not reducible in Humean fashion to generalizations merely about what follows what. In causation one particular event really produces another.2 Causation is a relation, however, about which general laws of nature may hold, laws which say that circumstances of sort C always cause events of sort E, or that they have a certain propensity (expressible as a probability) to cause them. Let me explain how I understand the possibilities (1), (2), and (3). (1) E is deterministically caused by C just in case it is caused by C and the causal laws (that is, the laws that actually hold, not the ones that we currently know or believe to hold) entail that whenever circumstances of sort C occur they will cause an event of sort E in the same way. (2) There are, I believe, two different ways that E could be indeterministically caused by C, neither of which do we (yet) know to be impossible: (2a) C contains an indeterminate state that resolves to the determinate event E and this resolution is caused by something in C, but the causal laws relevant to the case do not entail that whenever C occurs the indeterminate state in it will causally produce E, or be caused to resolve to E: they entail only that whenever C occurs the indeterminate state in it will causally produce or be caused to resolve either to E or to one of one or more alternative determinate states. (An example at the subatomic level would be an electron’s being in a state of superposition where it has no determinate position but comes to have one when it encounters a device that measures position.) (2b) E is not a resolution of an indeterminate state; it is directly caused by a determinate part of C with no indeterminate state involved in the process; but the relevant causal laws do not entail 2. I haven’t found any passage in Kane’s work that explicitly addresses this issue, but I would be surprised if he does not share this view of causation. He emphasizes that causes produce their effects and, like any incompatibilist, he takes it that at least in deterministic causation the cause necessitates the effect. On a strict Humean view, causation does not necessitate anything and, indeed, is not a real phenomenon. It is rather an illusion produced by our experience of regularity. There is no genuine production of one thing by another. “Causal laws” are merely generalizations about regularities, about what follows what. This includes probabilistic or indeterministic “laws.” On the Humean view nothing is really caused; all events are uncaused. Thus, on this view, all of our actions and decisions meet a condition that is, on my view, necessary for them to be up to the agent at the time they occur (but of course I don’t regard that as a reason to accept the Humean view).
C an an I ndeterministic C ause L eave a C hoice ?
( 17 )
that whenever C occurs it will cause an E in that way: they entail only that whenever C occurs something in it will directly cause either E or one of one or more alternative determinate events. (3) A completely uncaused E might or might not be the resolution of an antecedent indeterminate state. If it is, then nothing caused this resolution; if it is not, then nothing caused E to occur. For Kane the paradigm of a choice that could be (and seems to the agent to be) up to the agent at the time she makes it is what he calls a “self-forming action” (an SFA). SFAs, he says, occur at times in life when we are torn between competing visions of what we should do or become. Perhaps we are torn between doing the moral thing or acting from self-interest, or between present desires and long-term goals, or we are faced with difficult tasks for which we have aversions. (Kane 1999b, p. 224)
SFAs are thus a species of “torn decisions,” ones in which the subject is torn between competing alternatives she is considering, that is, all of these alternatives are strongly appealing to her so that she has difficulty deciding which to go for.3 I think (and perhaps Kane would agree) that, not just an SFA, but any torn decision is one that will seem to the agent to be (and for all we know could be) up to the agent at the time she makes it. I think the same is true of many decisions that are not torn—decisions, for example, where the subject already has a preference for the chosen alternative over any of the others on offer, or regards the reasons favoring the chosen alternative as stronger than those favoring any of the others, or the agent has a policy favoring the chosen alternative among the sorts of options she’s considering. They too typically seem to the agents to be (and for all we know could be) up to them at the time they make them. Here it is not clear that Kane would agree. He says, indeterminism does not have to be involved in all free and responsible acts, even for incompatibilists or libertarians. Frequently we act from a will already formed; and it may well be that our actions are determined in such cases by our then 3. I borrow the term “torn decision” from Balaguer (2010), but I mean something less by it than he does. For him a torn decision must not only fit the description I’ve given but also be such that the subject “has no conscious belief as to which option is best, given her reasons, and . . . decides without resolving this conflict—that is, the person has the experience of ‘just choosing’” (Balaguer 2010, p. 71). On my use, a torn decision could be like this but could also be such that the subject does have (or in the course of deliberation arrives at) a belief as to which option is best.
( 18 ) Libertarian Theories of Free Will e xisting characters and motives . . . What [incompatibilists] should . . . say is that when we act from a will already formed (as we frequently do), it is “our own free will” by virtue of the fact that we formed it (at least in part) by earlier choices or actions which were not determined and for which we could have done otherwise voluntarily. . . . Undetermined SFAs are a subset of all of the actions done of our own free wills (many of which may be determined by our earlier formed character and motives). (Kane 1999b, pp. 223–224)
This seems to suggest that decisions that are not torn are determined— deterministically caused—by their antecedents, by the agent’s “already formed will.” This could be so but I don’t see that it must be so. I much prefer apricot jam to strawberry jam and when I am offered a choice between the two I do not feel torn but decide without difficulty to take the apricot. Nevertheless, my subjective impression is that up until I make the choice it is open to me to choose otherwise, that it is still up to me at the time which alternative I choose. This impression may be illusory but I don’t know that it is and I see no conceptual bar to its being veridical. But for my purposes here, we do not need to go into that issue or to take account of the features that distinguish SFAs from other torn decisions. Consider the event of a subject’s making a torn decision between two alternatives. It is the subject’s deciding to A rather than to B, where A and B are the alternatives between which the subject was torn. (The decision to A may coincide with the action of A-ing, but often it will not: the decision is to A at a later time. And often in such cases, after the decision is made more decisions are required in order to implement it.) I agree with Kane that if the decision was deterministically caused then it cannot have been up to the agent at the time the decision occurred whether it would occur or not. (Why we believe this is something else we need not go into here. I think that incompatibilists generally are persuaded by something along the lines of what has come to be called “the consequence argument,”4 though they may differ on details.) I also agree with Kane in rejecting the “luck” argument that if a decision was not deterministically caused then it must have been a “random” or “chance” event, in a sense that entails that it was not up to the agent at the time whether it would occur. Mele (1998) argues for the “luck” claim by noting that if an action was not determined by its past then there is an alternative possible world where the past and the causal laws are just the same but 4. Versions of this can be found in Ginet (1966; 1990, ch.5; 2003), van Inwagen (1983), and many other places.
C an an I ndeterministic C ause L eave a C hoice ?
( 19 )
the action does not occur and “there is nothing about the agent’s powers, capacities, states of mind, moral character and the like that explains this difference in outcome” (p. 583). If that is so, Mele claims, then “the difference is just a matter of luck” (in a sense that entails that it was not up to the agent at the time of the action which outcome would occur). It seems to Kane, as it does to me, that, if “there is nothing . . . that explains this difference” means there is nothing that causally determines (i.e., deterministically causes) it, then this inference is simply fallacious. It seems to me further that, if “explains” means just causes, whether deterministically or indeterministically, then the inference still just does not follow, that the non sequitur is just as plain. I don’t think Kane would agree with this further claim. Kane thinks that it is not so obvious that the “luck” inference does not follow if the case is one where, in not just the actual scenario, but also in the alternative scenario, the agent decides and acts voluntarily and intentionally. It is obvious the inference fails for a case where the agent’s succeeding in A-ing depends on some lack of deterministic causation downstream from her decision to A and her beginning to A, so that in the alternative scenario she still tries to A but fails to do so because the not–deterministically caused bit after the decision goes a different way. But, Kane thinks, it’s not so obvious if what lacks a deterministic cause is the agent’s deciding to A, if the alternative scenario has her deciding and trying to do an alternative action B. He thinks that in order to show how the inference fails in this latter case we need to show how either of the alternative decisions and intentional actions could be indeterministically caused by the same preceding events and circumstances. I confess that I don’t see this. To me, even if nothing in the preceding events and circumstances causes the agent to decide to A rather than to decide to B—nothing causally explains, deterministically or indeterministically, the difference in outcomes in the alternative scenarios—it’s just as clear that it does not follow that whether the agent decided to A or decided to B was not up to her at the time of deciding. Kane (1999b) gives the following account of an SFA where the choice is indeterministically caused: Consider a businesswoman who faces a conflict in her will . . . . She is on the way to a meeting important to her career when she observes an assault in an alley. An inner struggle ensues between her moral conscience, to stop and call for help, and her career ambitions, which tell her she cannot miss this meeting—a struggle she eventually resolves by turning back to help the victim . . . . Prior to choice, there was some indeterminacy in her neural processes stirred up by the conflict in her will. The indeterminism made it uncertain (and undetermined) whether she would go back to help or press onward.
( 20 ) Libertarian Theories of Free Will Suppose . . . that two recurrent and connected neural networks are involved . . . . Such networks circulate impulses and information in feedback loops and generally play a role in complex cognitive processing in the brain of the kind that one would expect to be involved in human deliberation. Moreover, recurrent networks are nonlinear, thus allowing (as some recent research suggests) for the possibility of chaotic activity, which would contribute to the plasticity and flexibility human brains display in creative problem solving (of which practical deliberation is an example). The input of one of these recurrent networks consists of the woman’s moral motives, and its output the choice to go back; the input of the other, her career ambitions, and its output, the choice to go on to her meeting. The two networks are connected, so that the indeterminism that made it uncertain that she would do the moral thing was coming from her desire to do the opposite, and vice versa—the indeterminism thus arising, as we said, from a conflict in the will. When her effort to overcome self-interested desires succeeded, this corresponded to one of the neural pathways reaching an activation threshold, overcoming the indeterminism generated by the other. (p. 225)
Where do the processes in the businesswoman’s brain leading up to her decision fit into the classification scheme I gave earlier? Clearly it is not a case of deterministic causation of the decision. But what sort of nondeterministic case is it? Are the two recurrent and connected and chaotic neural networks that realize the conflict in her will wholly determinate states and processes that directly but indeterministically cause the decision (so the case falls under 2b)? Or do they involve an indeterminate state or process and the outcome decision is a resolution of that indeterminacy? And if the latter, is the resolution of the indeterminacy caused (so the case falls under 2a) or uncaused (so the case falls under 3)? Early in the material quoted Kane speaks of an “indeterminacy in her neural processes stirred up by the conflict in her will.” This certainly suggests (though it does not entail) that Kane thinks of the failure of deterministic causation here as involving an indeterminate state in the process leading to the decision, which would make it a case falling under 2a or 3. The last sentence in the material just quoted speaks of “one of the neural pathways reaching an activation threshold, overcoming the indeterminism generated by the other” (emphasis added). This suggests that the reaching of an activation threshold by one of the neural pathways is the resolution of the indeterminacy and is the decision in favor of the output of that network, and it suggests that the indeterminate, chaotic state itself produces— indeterministically causes—its resolution.
C an an I ndeterministic C ause L eave a C hoice ?
( 21 )
Later in the same essay Kane (1999b) adds to the description of the case: Imagine that the businesswoman is trying or making an effort to solve two cognitive problems at once, or to complete two competing (deliberative) tasks at once—to make a moral choice and to make a choice for her ambitions (corresponding to the two competing neural networks involved in the earlier description). With respect to each task . . . she is being thwarted in her attempt to do what she is trying to do by indeterminism. But . . . the indeterminism . . . is coming from her own will, from her desire to do the opposite. Recall that the two crossing neural networks involved are connected, so that the indeterminism which is making it uncertain that she will do the moral thing is coming from her desire to do the opposite, and vice versa. She may therefore fail to do what she is trying to do. . . . But . . . if she nevertheless succeeds, then she can be held responsible because . . . she will have succeeded in doing what she was trying to do. And the interesting thing is that this will be true of her, whichever choice is made, because she was trying to make both choices and one is going to succeed. (p. 231)
Perhaps Kane’s picture is something like the following: the indeterministic causation of the woman’s decision is a matter of the resolution of an indeterminate state. That is, it is the resolution of a “superposition” of two or more states into one of them (on the analogy of the resolution to a determinate location for a subatomic particle from a superposition of two or more different locations).5 On this picture, the antecedent state from which the determinate event indeterministically results is a kind of disjunction of two or more (potential, as yet unrealized) determinate states, of which the resulting realized determinate state must be one. So if the decision, which entails the coming to be of a certain intention, is indeterministically caused in this way, the antecedent indeterminate disjunctive state from which it results must contain that intention as one of its disjuncts in a potential, not yet realized, form. This would explain why Kane wants to suppose that prior to her decision the woman is trying to do each of the competing alternatives, rather than just having a strong desire to do each of them. He thinks of each of this pair of predecision competing tryings as a kind of potential intention and the intention in the decision as the realization, the making actual, of one of them. Whether this is the right account of Kane’s picture, or instead there is no indeterminate state—no “superposition” of tryings involved and each 5. That this is Kane’s picture is further confirmed by his saying that free will requires us to “imagine an indeterministic world like the quantum world of modern physics (on standard interpretations of it) that allows for both indeterminateness of physical properties and the possibility of forks in history” (Kane 1996, p. 173, emphasis added).
( 22 ) Libertarian Theories of Free Will
competing trying is a fully determinate state and this compound of states indeterministically causes the decision—either way Kane’s account strikes me as an implausible account of what actually happens in torn decisions. It is not, I think, an impossible account: One can try to do each of two things one knows to be incompatible, but in such cases one is not intending to do each of the two things; one intends only to do one or the other.6 I’m not sure this would bother Kane. He might say that the trying he speaks of need not involve an actual, fully realized, intention but only a potential one, fitting for a disjunct of a disjunctive, indeterminate state. The account is not impossible. But it strikes me as implausible as an account of the actual phenomenology of torn decisions, at any rate of mine: they just don’t seem like that. A trying must—conceptually, it must—be an action of some sort. To try to do A it is not enough merely to intend or to desire to do A: one must make an attempt, one must act. But when I am torn between two strongly appealing alternatives, and eventually opt for one of them, I am aware of no action before the decision that could be described as my trying to do one of them. Nothing I’m aware of is an action that fits that description. There is just the strong desire to do the one and the strong desire to do the other. And isn’t it just as plausible as the account Kane gives to say that the winning desire directly causes the decision indeterministically, without any need to go through an indeterminate “superposition” of tryings, of potential actions or potential intentions? Kane (2000a) says that in a torn decision the agent will encounter resistance to her desire to do one alternative, resistance coming from the reasons for the other alternative and that “effort or trying is needed to overcome this resistance” (p. 343). But why must such effort or trying come before the decision rather than after? On the contrary, isn’t it more plausible to think that it is only after one has made up one’s mind that one needs, or is appropriately placed, to resist the pull of the rejected alternative(s)? Perhaps Kane thinks that talk merely of a desire as cause would not work as well as talk of a trying or an effort as cause in the following answer he proposes to a further pressing of the luck objection: 6. Here’s an example: I confront a set of double doors through which I wish to go. I know that the doors open only one way, either toward me or away from me, but I don’t remember which. I am in a hurry, so I simultaneously push away from me on one of the doors and pull toward me on the other. In pushing on the one door I am trying to push it open and in pulling on the other door I am trying to pull it open. But it is not the case that I intend to push open the one door and intend to pull open the other. It would be irrational of me to intend to do both of these two things when I know that doing both is impossible. But it is not irrational of me to try to do both things while knowing that at most one attempt will succeed, which entails my intending to make it the case that either I push open the one or pull open the other.
C an an I ndeterministic C ause L eave a C hoice ?
( 23 )
But one may still object that the businesswoman makes one choice rather than the other by chance, since it was undetermined right up to the last moment which choice she would make. If this is so, we may have the picture of her making an effort to over-come temptation (to go on to her meeting) and do the moral thing, and then at the last minute “chance takes over” and decides the issue for her. But this is the wrong picture. On the view just described, you cannot separate the indeterminism from the effort to overcome temptation in such a way that first the effort occurs followed by chance or luck (or vice versa). One must think of the effort and the indeterminism as fused; the effort is indeterminate and the indeterminism is a property of the effort, not something separate that occurs after or before the effort. The fact that the woman’s effort of will has this property of being indeterminate does not make it any less her effort. The complex recurrent neural network that realizes the effort in the brain is circulating impulses in feedback loops and there is some indeterminacy in these circulating impulses. But the whole process is her effort of will and it persists right up to the moment when the choice is made. There is no point at which the effort stops and chance “takes over.” (Kane 1999b, p. 232)
But does this give any more effective a response to the luck argument’s claim than it would if “effort” were replaced by “desire”? Can’t the luck objector respond by claiming that there need be no moment when the effort stops for it to be chance or luck that determines which of the competing efforts wins, that it is still chance or luck even if one of the efforts uninterruptedly becomes the decision? If the luck objection is plausible when it is just competing desires that precede the decision (with one of them indeterministically causing it), it is hard to see how we make it less plausible by insisting that it is competing efforts that precede the decision and that the winning effort becomes the decision without interruption. Or, to put it the other way around, if the luck objection is implausible when it is one of some competing efforts that uninterruptedly becomes the decision, why isn’t it just as implausible when it is one of some competing desires that does this? I suspect that, were Kane to concede this, he would still want to insist that showing how the torn decision is indeterministically caused by something in the agent’s “will” is the only way to show how it is up to the agent at the time it occurs. I question this too, and it is the important issue. I do not think that whether a decision was deterministically caused or only indeterministically caused (in the right sort of way)—whether the case falls under (1) or (2)—is what determines whether the decision was up to the agent at the time. I think, to the contrary, that what determines this is simply whether the decision was caused or not caused, whether the case falls under
( 24 ) Libertarian Theories of Free Will
(1) or (2) on the one hand or under (3) on the other. It is the difference between its being caused and its being uncaused that makes the difference as to whether the decision was up to the agent at the time. In the caused case, but not in the uncaused case, the decision is produced by the antecedent circumstances. What those circumstances produce—whether they produce that decision or one of the alternatives allowed by the relevant causal laws— has to be viewed as “decided” by those antecedent conditions and not by the agent. In contrast, in the uncaused case there is nothing but the agent to decide what the event will be, in simply making the decision that is that event: it is in the nature of a decision that it is up to the decider at the time of deciding what it will be if (and only if) this is not preemptively “decided” by an antecedent cause. If there is a genuinely appealing intuition behind the luck argument, it is this: if a decision is caused indeterministically, then what the agent decides is not determined by the agent at the time of decision but by nature’s “roll of the dice,” where the outcome probabilities are determined by the antecedent state of the world and the laws of nature. But the basic intuition here, it seems to me, applies to the case of causation generally, whether deterministic or indeterministic—it is nature that decides what its current state will produce next, whether by a deterministic or by an indeterministic “roll of the dice” (a deterministic roll being one where all the faces of the dice have the same number of spots on them). But the intuition does not apply to the case where the decision is not caused at all. There nothing decides what the choice is but the choice itself—that is to say, the chooser in making the choice.7 So I say that a choice that is up to the agent at the time it occurs cannot be caused, even indeterministically. But if this is so, how can such a choice have an explanation in terms of the agent’s reasons for making it? Surely most, if not all, of our choices that seem up to us at the time we make them are ones that we make for reasons. On the dessert menu both the chocolate cake and the bread pudding arouse in me strong desires; I agonize a few moments and finally order the bread pudding. Why did I order the bread pudding? In order to satisfy my desire to enjoy eating it (though I did not think that I would enjoy that more than eating the chocolate cake). We constantly believe many 7. We can, if we like, speak of uncaused actions or decisions as caused by the agent— meaning just that only the agent determines what they are. I find this a natural description. But I don’t read into it any commitment to the metaphysics of agent-causation. The naturalness of the description doesn’t change the fact that between uncaused acts or decisions and uncaused events of other sorts there is no real difference so far as their causation goes, and the only thing that makes that description natural in the case of uncaused acts or decisions (besides their being uncaused) is the fact that they are acts or decisions and their subjects are agents.
C an an I ndeterministic C ause L eave a C hoice ?
( 25 )
such explanations of our and others’ actions. It would be absurd, quite out of the question, to doubt the truth of all of them, or even most of them. But can a choice or an action have such an explanation if it is not caused by the relevant desire or intention? Yes, it can, because it is sufficient for the truth of such a reasons explanation, roughly, that the agent intended concurrently with the choice or action that by taking the action chosen she would contribute to satisfying the relevant desire; and this sufficient condition does not entail that the choice or action is caused by that antecedent desire—only that it is explained by it.8 Kane (1996) disagrees: many libertarians . . . have held that incompatibilist free choices or actions cannot be given causal explanations of any kind, deterministic or probabilistic. . . . I think this familiar libertarian view is wrong . . . because I think it stands in the way of understanding the place of free will in the natural order in which causal explanations of deterministic or probabilistic kinds hold sway. . . . Causal explanations . . . must be part of the story if free will is to have a place in the natural order. And since deterministic causal explanations are ruled out for undetermined free willings, it is important to ask how explanations of them in terms of reasons are related to probabilistic or nondeterministic causal explanations. (p. 174)
But why could not entirely uncaused events be part of “the natural order” (even of a thoroughly physical natural order)? Nature has held many surprises. I see no a priori reason for thinking it impossible that one surprise that nature might have in store for us is that some events are not caused, that empirical investigation will persuade us that the best account of the laws of nature dictates for some kinds of events that their antecedents do not causally produce them but only limit what the ensuing event will be to one or another of a certain proper subset of the logical or metaphysical possibilities. Kane and I certainly agree that, on either his incompatibilism or on mine, the question whether we have free will—whether any of our decisions are up to us at the time we make them—is, in the phrase of Balaguer (2010), “an open scientific problem,” an as yet unsettled empirical question about a contingent matter of fact. And I must acknowledge that on Kane’s view the prospects for an affirmative answer—that they are up to us—are 8. For fuller exposition and defense of this account of reasons explanations, see Ginet (2002, 2008). For arguments against this account, see Clarke (2003, pp. 21–24; 2010a).
( 26 ) Libertarian Theories of Free Will
perhaps better than on mine. He conjectures a way in which our decisions might be indeterministically caused that is (as far as I know) compatible with what physics and physiology now know about causal processes at the micro level in our neural systems and that further investigation could find is actually realized for many of our decisions. We should not, I think, be quite sanguine about the prospects of our finding that most of our choices fit this bill: although we have quite strong evidence that determinism is false and that there is indeterministic causation in the world, we are not in a position to say that what current neuroscience knows about the workings of our brains makes it highly probable that most of our choices are indeterministically caused in the right way. But on my view the prospects are bleaker still: given what we now know about neural processes, it seems even less likely that we should find that the neural events that realize our decisions, in most of the instances where the decisions seem up to us at the time, are uncaused (though this is not, I think, incompatible with what is now known). And this is, it must be conceded, a reason for preferring Kane’s view—a nonevidential reason but nevertheless a rational one. By the same token, compatibilism is preferable to incompatibilism: on a suitably formulated compatibilist view we are assured of its being the case that our decisions that seem up to us at the time really are so. Many find this sort of motive hard to resist.9 But, however much weight an ideally rational person should give to this nonevidential sort of reason for believing something (namely, that it would be significantly more comfortable to believe it than not to do so), it is still the case (so it seems to me) that if only disinterested reasons—in the case of this philosophical question, only disinterested sound a priori arguments—are considered, they favor incompatibilism over compatibilism and they favor my incompatibilism over Kane’s.
9. John Fischer (2000), for example, has said, “One of my main motivations for being a compatibilist [about determinism and responsibility] is that I don’t want our personhood and our moral responsibility, as it were, to hang on a thread, or to be held hostage to the possible scientific discovery that determinism is in fact true” (p. 323).
•
Chapter 3
Free Will and Metaphysics Timothy O’Connor
F
or almost fifty years, Robert Kane has been a man on a mission. While he has addressed a range of questions in metaphysics and value theory throughout his distinguished career, he has sought with more tenacity than any other living philosopher to clarify the metaphysical underpinnings of human freedom and moral responsibility. Kane began thinking about the topic in the early 1960s, and I think it is worth noting that era’s climate of opinion regarding not just free will (something he frequently notes) but also metaphysics and the philosophy of mind more generally. While logical positivism was on the wane, empiricist suspicion of traditional metaphysical categories and arguments endured. Since that time, metaphysics has been re-born and is now flourishing. Most pertinent to the present essay, the metaphysics of causation and the ontology of mental states have returned as hotly debated issues. The central theme of the present essay is that an adequate account of free will must squarely engage these more fundamental metaphysical issues— they cannot be “bracketed off” in the way that even some contemporary theorists of free will tend to suppose. Doing so involves considering both empirical and philosophical issues raised by our fundamental physical theory, quantum mechanics, and how the processes it describes connect to the macro-level phenomena in the brain and mind. Interesting developments have occurred on this front, and Kane has tried to appropriate some of them in his thinking on free will. Some theorists have offered schematic models of how indeterministic quantum effects might be amplified in brain processes underlying human decision-making. More ambitiously, I think,
( 28 ) Libertarian Theories of Free Will
has been the general rethinking of reductionist metaphysics on which all macro-level processes are either identical to or wholly constituted by micro-level processes. Although scientific theorists’ ideas can be difficult to interpret in terms of philosophical categories, there has clearly been some form of anti-reductionism at work in much thought arising from complex systems theory, applied to physics, biology, and elsewhere. In philosophy of mind, there has been a reappreciation of the challenge to reductionism posed by consciousness and (more recently) the intentionality of mental states. I agree with (and here assume) Kane’s basic orientation to the problem of free will, on which metaphysical freedom consists in being the ultimate, reasons-guided causal source of an intention to act in the face of alternatives possibilities for action. Over the years, he has developed and refined a complex analysis of what the exercise of such ultimate causality consists in, an analysis that is intended to contrast with “agent-causal” accounts that take it as a conceptual and metaphysical primitive. I will argue, however, that when we draw out two metaphysical assumptions to which Kane’s account is plausibly committed, we can easily be led to an account that is just a particular version of something like the primitivist agent-causal theory itself. Put differently, when set within a plausible general metaphysical framework, Kane’s theory and the agent-causal theory are much closer than has so far been recognized.
1. Kanean Libertarianism
Central to Kane’s analysis of free will is the notion of “will-setting” or “selfforming” actions (SFAs). As many recent action theorists (and for that matter, cognitive psychologists) emphasize, much of our behavior is automatic, unfolding in accordance with entrenched action plans that are triggered, in some cases entirely unconsciously, by the appropriate stimuli, without any intervening choice or active intention-formation. For example, you are in your car at a stoplight thinking about the puzzle of free will. The light turns green and, without thinking about it, your foot moves from the brake to the gas pedal. However, the interesting cases are ones in which we do consider what to do and feel some pull in more than one direction. These might be overtly moral choices, where there is conflict between “duty” and “desire,” or choices between short- and long-term self-interest, or simply choices among a range of equally permissible and prudent actions that are all worthwhile from the agent’s point of view but not all of which can be undertaken. It is these cases, Kane believes, where it is most plausible to suppose that agents directly exercise their freedom of will. They are
F ree W ill and M etaphysics
( 29 )
cases when the agent’s own will is divided between incompatible courses of action, in the sense that she has a plurality of “volitional streams”— complexes of beliefs, desires, and intentions—that are aimed at different ends. The agent is trying to accomplish each of two or more incompatible goals, and hence is divided. He posits that there is an objective, nonnegligible chance that each of these efforts succeeds. Whatever the outcome of this internal struggle, it will have been the agent’s own effort that has brought it about, a choice that is both motivated and intended. Finally, as Aristotle taught, over time, individual choice outcomes affect the relative weight of subsequent volitional streams, and a character is formed that is partly a result of the agent’s previous choices (Kane 2011a, pp. 386–390). 2. Two Metaphysical Commitments of Kane’s Account
2.1 Ontological Irreducibility of Mental States Kane rejects the thesis that human persons are immaterial minds. I concur, but this negative thesis is consistent with a wide range of views concerning the nature of mental states. According to standard forms of both reductive and (putatively) “nonreductive” physicalism, token mental events supervene upon (and on most versions, are identical to) structured physical events in the brain. They differ in that nonreductive physicalism denies, while reductive physicalism affirms, that they are type identical. Jaegwon Kim has argued in numerous writings (see especially Kim 1998) that nonreductive physicalism is an untenable position, in that it is unable to secure the causal efficacy of mental events. While there are problems with Kim’s presentations of the argument, I believe that there is a sound version of it (O’Connor and Churchill 2010). And reductive physicalism is obviously an unsatisfactory position for one who affirms a libertarian position regarding free will. A more specific problem than Kim’s challenge for both types of physicalist view of mental states is that they appear to be subject to a Consequencestyle argument for their incompatibility with free will (Cover and Hawthorne 1996, pp. 58–60): for an arbitrary action A, let “P” be a proposition describing each of the constituent microphysical states and relations thereof that (according to physicalism) constitute my deciding to A at time t, and let “Q” be the proposition that I decide to A at time t. Plausibly, P and I have no choice whether P; and necessarily, if P then Q (by supervenience); hence Q, and I have no choice whether Q. The upshot, I suggest, is that one who affirms Kane’s position on the nature and reality of free will must reject all varieties of physicalism. Given an antecedent rejection of mind-body (substance) dualism, this result points
( 30 ) Libertarian Theories of Free Will
in the direction of a metaphysical form of emergentism. Kane (2011a) appears to concur (p. 396), but we need to be careful here. The term “emergence” is used frequently in relation to complex systems of all kinds. It is important that we distinguish metaphysical emergence theses from those that are, in one sense or another, merely epistemic. Conway’s simple cellular automaton, the Game of Life, vividly illustrates that the existence of strikingly novel patterns of behavior in complex macroscopic systems is consistent with the behavior of those systems being wholly determined by completely general low-level transition rules at the fundamental level. These high-level patterns are “emergent” only in the sense that one cannot—at least in any straightforward way—derive the patterns from the properties and patterns appropriate to the fundamental level alone; it remains the case that the high-level rules do not in any way modify or supplement the basic dynamics that drive Life’s evolution. In short, the transparent simplicity of Life worlds make plain to an observer that epistemic irreducibility (in some sense) of high-level patterns is consistent with metaphysical reductionism. It is true but misleading to say of Life worlds that cells caught up in stable macroscopic structures that follow different dynamical patterns are “constrained by” those high-level dynamics. Ultimately, what is happening is that individual cells constrain themselves, by “causing” there to be (in certain regions) such macroscopic structures and determining, moment by moment, what the precise state of those structures will be. That one can focus outward from those details and see large-scale patterns requiring a different form of description does not change the fundamental point that there is an asymmetrical dependency of macro-level patterns (where they occur here and there) upon the completely general micro-level patterns. In contrast to the epistemic emergence reflected in the Life game, what is needed for freedom of the will in stable systems such as ourselves who are wholly physically composed is a metaphysical form of emergence: our mental states and capacities must be ontologically basic, rather than token identical to complex physical states, making a nonredundant causal difference to the way we behave. On such an emergentist picture, certain of our conscious mental states (perceptual, cognitive, and conative) are ontologically basic states of enduring, though changing, biologically composed objects that causally contribute to our unfolding mental and physical behavior, and specifically to the states of intention or decision whereby we resolve deliberative uncertainty and embark on courses of action. It is sometimes suggested that there being metaphysically emergent capacities would be “spooky,” not amenable to empirical investigation. But this is simply not the case. While they are basic features of reality, emergent
F ree W ill and M etaphysics
( 31 )
capacities may nevertheless be fruitfully studied and eventually explained in detail in nonreductive fashion, by spelling out the basic inventory of emergent properties, detailing the precise conditions under which organized physical systems give rise to them, and isolating the precise behavioral impact their presence has on the system. Where we have reason to believe there are such metaphysically emergent capacities, it will be natural to suppose that they are caused to be by the object’s fundamental parts, which have latent dispositions awaiting only the right configurational context for manifestation. If, in human beings, the capacity to form choices that emerges operates indeterministically, its existence and causal nature could be studied in more fundamental physical terms, even though its outputs cannot be explained in purely physical terms. I said above that Kane appears to affirm emergentism regarding some or all conscious mental states, but the details of his suggestions concerning how things might go in the processes constituting freely willed choices make it unclear to me whether he has in mind Life-style epistemological emergence or a robust metaphysical emergence. On the one hand, he suggests that there might be mechanisms whereby local micro-indeterminacies in relevant aspects of the brain might get amplified, determining which of two large-scale competing neural networks that encode opposing motivational structures has its characteristic end realized in action (1996, pp. 130–142; 2011a, p. 387). This is perhaps most naturally understood as a special case of a Life scenario: micro-level processes naturally result in the formation of stable structures that “constrain” individual components, and the outcomes of these structures are determined non-linearly and in a way that is sensitive to small-scale and relatively localized indeterminacies. On the other hand, he speaks more generally of our exercising “macro-control of processes involving many neurons” (2011a, p. 395). While not wanting to discount the potential contributing role of processes described by the former suggestion, I contend that it is crucial to the viability of Kane’s claim that it is the agent herself that is controlling this outcome, that the central determinants of choice be metaphysically basic, agent-level powers.
2.2 Causal Nonreductionism A second plausible commitment of Kane’s account of free will (and one that was implicit in remarks above) is a realist, nonreductionist view of causation. Consider the neo-Humean reductionist view as developed by David Lewis. (I discuss Lewis’s picture only for the sake of concreteness;
( 32 ) Libertarian Theories of Free Will
the incongruity with Kane’s account of free will that I allege generalizes to any reductionist account.) According to it, causal facts and the laws of nature are reducible to facts concerning the global spatiotemporal arrangement of fundamental natural properties which we (allegedly) may conceive in nondispositional terms. Roughly, the laws are the best system of generalizations over such natural facts, where what is best is determined by the optimal balance of simplicity and explanatory “strength.” Causation in turn consists of a restricted kind of counterfactual dependence of one event on another, where the counterfactuals are grounded in cross-world similarities.1 Within this framework, intentional human agency is naturally understood in terms of the counterfactual dependence of behavior or behavior-guiding intentions on appropriate beliefs, desires, or intentions the agent had immediately before and as the behavior occurs. This reductionist metaphysics of causation yields an implausible understanding of the metaphysics of agential control. By taking the fact of A’s being a cause of B to be a reducible, massively extrinsic relation—grounded in what occurs elsewhere and elsewhen—we empty the fundamental idea that A “produces” or “brings about” B of any clear content. Since agency is a causal notion, the implausibility carries over: on a neo-Humean analysis, the sense in which my beliefs and desires here and now bring about my present action is at best very weak tea. A fortiori, extrinsic analyses, on which whether or not psychological factors cause behavior is largely determined (metaphysically) by what happens in the distant reaches of spacetime, provide a bizarre account of a free action’s being, as we commonly say, “directly controlled by” the agent, such that it was “up to her” what she would do in the particular circumstances. Our notion of agential control, and especially the ultimacy condition on freedom of the will, manifestly indicates something that supervenes on the local circumstances in which we act. Freedom of the will cannot survive a reductionist construal of causation, of which it is a particular form. The best alternative to a neo-Humean reductionist account of causation is a neo-Aristotelian causal powers account.2 On this account, the “natural” properties of objects are causal powers—power to bring about particular results in particular circumstances constitute their fundamental, intrinsic 1. The locus classicus is Lewis (1986). (I note that Lewis allows for temporally remote causation by defining causal chains in terms of stepwise counterfactual dependencies, but it is unnecessary to fuss about such details here.) 2. I here ignore the higher-order laws account proposed in different forms by Tooley (1987) and Armstrong (1984). There are well-known, quite fundamental problems with these accounts.
F ree W ill and M etaphysics
( 33 )
nature.3 Causation is the manifestation of such a power (or the collaborative manifestation of multiple powers in interacting objects). Indeterministic causation, no less than deterministic causation, is the manifestation of causal power, though here the power is associated with propensities that are “chancy” in the sense that the objective prior probability in a given circumstance that the power will be manifested as it in fact is is less than 1. These are propensities toward a plurality of possible effects, and thus propensities that are manifested in different ways on different occasions. Indeterministic causal powers are sufficient, relative to a context, for each of the possible outcomes in the sense that they are all that is needed, though not in the sense that they are a causally sufficient condition. Every indeterministic event is produced, though none is necessitated.
3. Neo-Aristotelian Freedom
In the previous paragraph, I have deliberately elided mention of the entity that is the cause—that which exercises the causal power. Within the broadly neo-Aristotelian framework, there are two different ways one might think about this issue that have significantly different implications for how we think about the metaphysics of freedom. According to the first, we should say that, in a given determinate situation type S, the having of a power P by object O1 at time t produces effect E in object O2. Or, perhaps more commonly, in situation type S, the having of power P1 by object O1 and the having of power P2 by object O2 jointly produce effect E.4 That is to say, causes are events. The second analysis maintains, instead, that in situation type S, the objects O1 and O2 jointly produce effect E, doing so in virtue of their having powers P1 and P2 at time t, respectively. They jointly exercise their respective powers P1 and P2 to contribute to bringing about E. That is to say, causes are objects/substances. Something like the first, event-causal understanding of causation is implicit in Kane’s discussion. That is unsurprising, since event-causal analyses of some form or other have been popular ever since Hume, and especially throughout the twentieth century. However, one may argue that the general identification of causes with events is a legacy of the Humean rejection of causal power and substance. Abandon these Humean deflationary projects and it becomes natural to understand causes as substances. On the 3. See, e.g., Shoemaker (1980), Heil (2003), Mumford (2004), Lowe (2008), Martin (2008), Bird (2010), and Jacobs (2011). Some authors say instead—misguidedly, I judge—that properties of necessity confer causal powers on their bearers. 4. The mutuality of causal interactions is much emphasized by Martin (2008).
( 34 ) Libertarian Theories of Free Will
neo-Aristotelian metaphysical framework, the world is fundamentally a world of things/substances, not events. Events are derivative from (constructed in part out of) objects. It is, in general, powerful particulars—objects— that exercise causal power, that do things in the world. To be sure, their acting in the ways that they do has an explanation: they reflect the causal powers that they have at the time, powers that are none other than (one or more) natural properties that they have, and also (typically) the presence of necessary manifestation conditions. (I say “typically” since the phenomena of radioactive particle decay seems to involve no manifestation conditions.) So, for example, two electrons, eddie and eleonore, mutually repel each other—that is, cause each other to accelerate along receding paths at a specific rate. They do so in virtue of their powers—that is, negative electric charge—and in the circumstance (necessary manifestation condition) of their being a certain distance apart. This description of the metaphysics of causation is natural within the neo-Aristotelian framework. If we accept it, we have a significant benefit for the problem of free will: the agent causalist’s “problem of the disappearing agent” worry concerning event causalist accounts of free will, such as Kane’s, melt away. Since all causation is substance causation, then (provided we have a nonreductive view of agents and their powers, per our first assumption) unreduced “agent causation” comes for free—it is not a fundamentally distinct variety of causation. If all this is right, then a Kanean “event causal” libertarian really is (or ought to be) a kind of agent causalist. I am, quite literally, the principal cause of my free choices—which choices involve significant macro-level indeterminism (of the right sort), dual rationality (weakly understood, requiring only that whichever choice I make, it would be motivated), and dual control (whichever choice I make, it be one that I bring about). Kane’s distinctive notion of opposing “efforts of will” is also compatible with this general metaphysical framework, and may be argued on its own merits.
Pa r t I I
••• The Luck Objection
•
Chapter 4
Kane, Luck, and Control: Trying to Get by without Too Much Effort Alfred R . Mele
I
started thinking seriously about free will about twenty years ago. I was not terribly interested in the familiar projects of arguing for incompatibilism about free will and determinism and rebutting such arguments. Such questions as the following two interested me much more: If compatibilism is true, what suffices for an agent’s freely A-ing? If incompatibilism is true, what suffices for an agent’s freely A-ing? My Autonomous Agents (Mele 1995) was guided partly by these two questions, and part of what led me to believe that an attractive answer to the second question might be forthcoming was Robert Kane’s early work on free will—especially Free Will and Values (Kane 1985) and “Two Kinds of Incompatibilism” (Kane 1989). Kane’s ideas about free will have evolved over the last twenty years, and so have mine. My focus here is on some relatively recent ideas of ours about how libertarians should deal with an apparent problem posed by something that might be called “luck.” But the history that I briefly set out in section 1 starts in the 1980s. In parts of this chapter, I draw on Mele (2006, 2013a, 2013b). This chapter was made possible through the support of a grant from the John Templeton Foundation. The opinions expressed here are my own and do not necessarily reflect the views of the John Templeton Foundation. I am grateful to David Palmer for advice about how to shorten a draft of this chapter.
( 38 ) The Luck Objection 1. Kane on Luck
Bruce Waller (1988, p. 151) raises a significant worry about luck in criticizing the view of free will and moral responsibility advocated in Kane (1985). Kane (1996) summarizes Waller’s objection as follows: “Suppose two persons had exactly the same pasts and made exactly the same efforts of will,” says Waller, but one does the moral or prudential thing while the other does not. Given that their pasts were exactly the same up to the moment of choice, as indeterminism requires, wouldn’t that mean that the outcome was a matter of luck? One of them got lucky and succeeded in overcoming temptation, the other failed. Would there then “be any grounds for distinguishing between [them], for saying that one deserves censure for a selfish decision and the other deserves praise for generosity? If they are really identical, and the difference in their acts results from chance, then it seems irrational to consider one more praiseworthy (or more blameworthy) than the other should be.” (p. 171)
Kane (1996) argues in response that efforts to resist temptation in free agents are indeterminate, and because that is so we cannot “imagine the same agent in two possible worlds with exactly the same pasts making exactly the same effort and getting lucky in one world and not the other. Exact sameness or difference of possible worlds is not defined if the possible world contains indeterminate efforts or indeterminate events of any kinds” (p. 172). In Mele (1999b), I argue that this appeal to indeterminacy is not particularly useful, even if it is granted that exact sameness and difference are not defined: Kane’s appeal to the indeterminacy of an effort makes it more difficult to formulate crisply the “objection from luck” to libertarianism. But the spirit of the objection survives. If John’s effort to resist temptation fails where John2’s effort succeeds, and there is nothing about the agents’ powers, capacities, states of mind, moral character, and the like that explains this difference in outcome, then the difference really is just a matter of luck. That their efforts are indeterminate explains why the outcomes of the efforts might not be the same, but this obviously does not explain (even nondeterministically or probabilistically) why John failed whereas John2 succeeded. (p. 280; also see Mele 1999a and Haji 1999)
Kane (1999b) reacts by modifying his view. Discussion of the modified view is the main business of this section.
K ane , L uck , and C ontrol
( 39 )
In some cases, Kane (1999b) contends, an agent simultaneously tries to make each of two competing choices or decisions.1 He claims that because the agent is trying to make each, she is morally responsible for whichever of the two decisions she makes and makes it freely (pp. 231–240), provided that “she endorse[s] the outcome as something she was trying and wanting to do all along” (p. 233). Someone who takes this position can consistently hold that even if the agent’s deciding to A, as she in fact did, rather than her instead deciding to B, as she did at the same time in another world with the same past and laws of nature—that is, that difference, D, between the two worlds—is just a matter of luck, the agent decides freely and is morally responsible for her decision. At least, that is so if, as it seems to me, this agent’s satisfying Kane’s alleged sufficient condition for free and morally responsible decision is consistent with D’s being just a matter of luck.2 What matters, in Kane’s view, is that the agent tries to make each decision (in both worlds) and endorses the outcome in the way just mentioned. If Kane is right, he has provided a successful answer to a certain challenge about luck (see Mele 2006, p. 50)—at least in scenarios of a certain kind. Part of the inspiration for Kane’s position is the observation that “indeterminism [sometimes] functions as an obstacle to success without precluding responsibility” and free action (1999b, p. 227). In one of his illustrations, “an assassin who is trying to kill the prime minister . . . might miss because” his indeterministic motor control system leaves open the possibility that he will fire a wild shot. Suppose the assassin succeeds. Then, Kane says, he “was responsible” for the killing “because he intentionally and voluntarily succeeded in doing what he was trying to do—kill the prime minister” (p. 277). It may be claimed, similarly, that the indeterminism in the scenario does not preclude the killing’s being a free action. If these claims are true, they are true even if the difference between the actual world at the time of the firing and any wild-shot world that does not diverge from the actual world before that time is just a matter of luck. Unfortunately, Kane’s observation does not get him very far. As Randolph Clarke observes, the presumption of those who judge that the assassin freely killed the prime minister is that he freely tried to kill him (2002, pp. 372– 373); if we are told that perhaps the assassination attempt was not free, all bets are off. In Kane (1999b), it is not claimed that in cases of dual efforts 1. Also see Kane (1999a, 2000b, 2002b, 2011a). Readers who balk at the thought that an agent may try to choose to A (Kane 1999b, pp. 231, 233–234) may prefer to think in terms of an agent’s trying to bring it about that he chooses to A. 2. Kane (1999b) writes: “The core meaning of ‘He got lucky’, which is implied by indeterminism, I suggest, is that ‘He succeeded despite the probability or chance of failure’; and this core meaning does not imply lack of responsibility, if he succeeds” (p. 233).
( 40 ) The Luck Objection
to choose, the choices made are products of freely made efforts. Nor did Kane put himself in a position to make this claim, for that article includes no account of what it is for an effort to choose to A to be freely made. Thus, there is a striking disanalogy between cases like that of Kane’s assassin and Kane’s dual trying cases: no grounds are offered for presuming that the dual efforts to choose are freely made. And if the agent’s efforts to choose in a dual trying scenario—unlike the assassin’s effort to kill the prime minister—are not freely made, it is hard to see why the choice in which such an effort culminates should be deemed free. Some readers may need assistance in appreciating this last point. They should imagine that a manipulator compels an agent, Eldon, simultaneously to try to choose to A and to try to choose to B, where A and B are competing courses of action that, in the absence of manipulation, Eldon would abhor performing. Imagine also that the manipulator does not allow Eldon to try to choose anything else at the time and that the manipulation is such that Eldon will endorse either relevant “outcome as something [he] was trying and wanting to do all along” (Kane 1999b, p. 233). The tryings are internally indeterministic, but Eldon does not freely try to make the choices he tries to make. Apparently, whatever he chooses, he does not freely choose it— especially when the sort of freedom at issue is the sort most closely associated with moral responsibility. To be sure, in this scenario the unfreedom of the efforts is tied to serious monkey business. But subtract the monkey business. Are Eldon’s efforts to choose now freely made? If so, in virtue of what is it true that he freely makes them? No answer is found in Kane (1999b), and one hopes (for a reason to be offered shortly) that any answer that is forthcoming will not appeal to competing efforts at a higher level—for example, an effort to make an effort to choose to A and an effort to refrain from making an effort to choose to A. If, on the other hand, the efforts to choose are not freely made even in the absence of monkey business, why should choices in which they may issue count as free? The combination of trying and endorsement that Kane describes does not suffice for freely making the decision one makes: that combination is present in Eldon’s case. It may be claimed that this combination would suffice for a free decision in the absence of monkey business. When I see an argument for that claim, I will try to assess it. The take-home moral here is that an argument by analogy from such cases as the assassin’s will not fly as long as the disanalogy I mentioned is in place. One way to try to eliminate the disanalogy is to produce an acceptable account of the freedom of an effort to choose to A and show that efforts at work in some dual trying scenarios satisfy the account. Kane makes the dual trying maneuver in response to an apparent problem about luck. The alleged fact that the agent succeeds in doing some-
K ane , L uck , and C ontrol
( 41 )
thing he is trying to do is supposed to carry a lot of weight. But it seems that this alleged fact will carry the weight it is supposed to carry only if the trying is free. So, in the case of an agent who allegedly simultaneously tries to make each of two competing choices, is he freely trying to make those choices, and, if so, in virtue of what are those tryings free? As I mentioned, Kane (1999b) does not answer this question. The absence of an answer is a significant shortcoming. Until we have been given a good reason for believing that the dual tryings Kane postulates are free, why should we believe that the choices in which they may issue are free? Kane’s recent distinction among “three freedoms” (2008, p. 142) is interesting in the present connection. He asserts that “Free acts may be”: (1) acts done voluntarily, on purpose and for reasons that are not coerced, compelled or otherwise constrained or subject to control by other agents. (2) acts [free in sense 1 that are also] done “of our own free will” in the sense of a will that we are ultimately responsible (UR) for forming. (3) “self-forming” acts (SFAs) or “will-setting” acts by which we form the will from which we act in sense 2. (p. 143)3 Kane observes that free actions of type 1, as he conceives of them, are compatible with determinism and that free actions of types 2 and 3 are not (p. 143). I asked in virtue of what dual efforts to choose might count as free actions. Now that Kane has given us three different “types” of free action to work with, we can look for an answer among these types. I start with type 3. All actions of this type, as Kane conceives of them, are indeterministically caused by their proximal causes. That they are so caused raises the problem about luck that Kane tries to solve with his dual trying maneuver. But applying that maneuver to efforts to choose themselves is definitely not to be recommended. In world 1 at time t, Ann makes an effort to choose to A. At t in another world with the same past and laws of nature, Ann does not make an effort to choose to A. Joan, who is worried about luck, asks Joe how Ann’s effort to choose to A can be free. His reply features the idea that Ann was making an effort to make an effort to choose to A while also making an effort not to make an effort to choose to A. Joe is appealing to Kane’s dual trying maneuver one level up, as it were; and Joe is in serious trouble. Not only is this idea farfetched, but it can also lead quickly to a vicious regress. (If Ann’s second-order efforts are free, in virtue of what is that so?) The dual trying maneuver that is supposed to solve a problem about 3. The brackets are present in the quoted text. On senses 2 and 3, also see Kane (1996, pp. 77–78).
( 42 ) The Luck Objection
luck for choices that are claimed to be free in sense 3 cannot reasonably be used to support the idea that efforts to choose to A are free in sense 3. And Kane has not provided an alternative way of supporting the idea that these efforts are free in that sense. I turn to type 2 free actions. Actions of this kind depend on earlier type 3 free actions by the same agent. So we cannot see how type 2 free action is possible until we see how type 3 free action is possible. I leave open the idea that some efforts to choose to A are type 2 free actions, but I remind the reader that we are not in a position to believe that there are any type 2 free actions until we are in a position to believe that there are some type 3 free actions. Type 1 free actions are next in line. Kane may propose that the following is sufficient for the truth of the assertion that an agent’s choosing to A was a type 3 free action: his choosing to A was indeterministically caused by its proximal causes, he was trying to choose to A, that attempt was successful, he endorses “the outcome as something [he] was trying and wanting to do all along” (1999b, p. 233), and his effort to choose to A was a type 1 free action. I will briefly discuss three questions or worries about this proposal— proposal P. First, why should it be thought that certain efforts to choose to A differ from actions in general in such a way that even though compatibilism is true of these efforts it is not true of actions in general? Elsewhere, I have floated a “soft libertarian” view according to which “free action and moral responsibility [may be] compatible with determinism but . . . the falsity of determinism is required for . . . more desirable species of” these things (Mele 2006, p. 95; see also Mele 1996). The brand of soft libertarianism that I present in Mele (1996, 2006) is relativistic: it maintains that at least some human agents are possessed of kinds of freedom and moral responsibility that are incompatible with determinism and are reasonably preferred by at least some of these agents to any kind of freedom or moral responsibility that is consistent with the truth of determinism. If Kane believes that type 1 free actions really are free actions and that agents are morally responsible for some such actions, he can beef up soft libertarianism to include a commitment to compatibilism. He can also defend an objective variant of my relativistic line. But the question with which I opened this paragraph remains. For a second worry, return to Kane’s assassin. Suppose the assassin’s effort to kill the prime minister was free in and only in sense 1. Even if his effort itself was indeterministically caused, it is difficult to see how the killing in which it issues can be free in sense 2 or sense 3, given that the effort is free only in sense 1. Arguably, the killing just is the successful effort to kill. How can the freedom of the killing outstrip the freedom of the
K ane , L uck , and C ontrol
( 43 )
effort? A parallel question arises about efforts to choose to A. How can the freedom of a choice to A outstrip the freedom of the effort to choose to A that culminated in that choice? I am not claiming that plausible answers to these questions are not available to Kane. Good answers to them would prove illuminating. A third worry is about the dual trying maneuver associated with proposal P. Kane is ordinarily pretty sensitive to the phenomenology of agency. But trying to choose to A while also trying to choose to perform some competing action seems remote from ordinary experience. We may occasionally have an experience of trying to bring it about that we choose to A. (For example, someone who knows that it would be best to quit smoking may try to vividly represent to himself the most important reasons for quitting, including the dangers of not quitting, with a view to bringing it about that he chooses to quit.) But how many of us have experienced simultaneously trying to bring it about that we choose to A and trying to bring it about that we choose a competing course of action? If such dual efforts never occur, they never underwrite free choices. Obviously, someone who grants that we never experience dual efforts to choose may wish to posit such efforts anyway in an attempt to solve a theoretical problem (see Kane 2011a, pp. 391–392). But alternative routes to a solution should also be explored. If dual efforts of the kind at issue never happen, is Kane unable to explain why some choices that are indeterministically caused by their proximal causes are free in sense 3? Well, if he is entitled to appeal to efforts to choose to A that are free in sense 1 in an attempt to solve a problem that luck apparently poses for the claim that these choices can be free in sense 3, he might consider appealing instead to an effort of another kind in this connection—an effort to settle on what to do (see Clarke 2011a, pp. 341–342). Sometimes we realize that we need to decide what to do about a particular issue pretty soon if we are to proceed reasonably. Ann was offered a job two weeks ago and was given two weeks to decide whether to accept it. She has not decided yet; she has until 5:00 p.m. today to make a decision, and it is already four o’clock. Ann tells herself that she had better set aside some time now to come to a decision. And she starts mulling the matter over again, in an attempt to bring it about that she decides the issue one way or the other before 5:00. Ann’s attempt is successful. She decides to reject the job offer and keep her present position, and that decision was indeterministically caused by, among other things, Ann’s effort to bring it about that she settled the matter. Suppose Ann’s effort to bring it about that she decides what to do is free in sense 1. If Kane regards an effort to decide or choose to A that is free only in sense 1 as adequate for the work assigned to efforts in his dual
( 44 ) The Luck Objection
efforts maneuver, might he deem Ann’s type-1 free effort to bring it about that she decides what to do about the job offer adequate for a comparable task? Because Ann’s effort is not an effort to bring it about that she decides to A, it is not a tight analogue of the assassin’s effort to kill the prime minister. In my story about Ann, the following claim is false: just as the indeterministic assassin who was trying to kill the prime minister freely killed him, Ann, who was trying to decide to reject the job offer, freely rejected it. Ann was not trying to decide to reject the job offer; nor was she trying to bring it about that she decided to reject the job offer. But she was trying to do something related. She was trying to bring it about that she decided what to do about the job offer. And she decided on one of the two options that she was entertaining. Readers who are attracted to an event-causal libertarian view along Kane’s lines but who doubt that Kane’s dual trying maneuver applies to actual human agents have another option to explore. Such readers may consider P* as an alternative to P: P*. The following is sufficient for the truth of the assertion that an agent’s choosing to A was a type 3 free action: His choosing to A was indeterministically caused by its proximal causes, he was trying to bring it about that he chose what to do, that attempt was successful, A was one of the options he was considering at the time, he endorses his choice to A, and his effort to bring it about that he chose what to do was a type 1 free action.
Of course, this proposal, like P, is subject to the second worry I discussed. One may try to meet that worry head on and show how the freedom of the efforts at issue can be outstripped by the freedom of the choices. Another option is to defend the idea that efforts to bring it about that one chooses or decides what to do may be free in sense 3. A philosopher who makes it that far may seek to defend—in a way that makes no special appeal to choice-related efforts—the idea that decisions or choices that are indeterministically caused by their proximal causes may be free in sense 3.4
2. Event-causal Libertarianism and Control
I myself am not a libertarian. (Nor am I a compatibilist. I am officially agnostic about compatibilism.) But I have floated a variety of libertarian views. The one that is closest to Kane’s view is my daring libertarian 4. For an approach of this kind, see Mele (2006, ch. 5).
K ane , L uck , and C ontrol
( 45 )
view (Mele 2006). I will discuss it shortly. Some background is in order first. Some philosophers distinguish between what may be termed late and early indeterministic processes in an action-producing stream (Dennett 1978, pp. 294–295; Mele 1995, p. 212; 2006, pp. 112–114). Late processes of this kind are still at work when the actions they issue in begin, and early processes are not. An indeterministic process that generates mental representations of options and stops some time before any option is selected is an example of an early process. An indeterministic decision-producing process that does not end before it indeterministically issues in a decision to A is an example of a late process. I hasten to add that, as I understand decisions to do things (as opposed to decisions about what is the case), they are very brief actions of intention formation, and in deciding to A one forms an intention to A (Mele 2003, ch. 9; see Kane 1996, p. 24; 1999b, p. 236). (Deliberating about what to do typically is not a very brief action, but deliberating must be distinguished from any act of deciding that is based on deliberation.) According to my daring libertarian view (DL), an agent never acts freely unless he sometimes successfully exercises “an initiatory power that, by definition, is at work only when the proximate causes of a decision do not deterministically cause it. This power gives agents causally open alternative futures; and . . . it gives them this openness at the moment at which some decisions are made” (Mele 2006, pp. 202–203).5 I call any free A-ings that occur at times at which the past (up to those times) and the laws of nature are consistent with the agent’s not A-ing then basically free actions. In principle, libertarians can hold that an agent’s basically free actions that are suitably related to his subsequent A-ing confer freedom on his A-ing even though he could not have done otherwise than A then. It is open to libertarians to accept or reject the thesis that the only free actions are what I am calling basically free actions (Mele 1995, pp. 207–209; 2006, p. 6). I turn from basically free actions in general to basically free decisions in particular. The following thesis about decisions is open to libertarians: BFD. An agent makes a basically free decision to A only if, at the pertinent time, he successfully exercises an initiatory power that, by definition, is at work only when the proximate causes of a decision do not deterministically cause it.
5. For complications introduced by Frankfurt-style cases, see Mele (2006 pp. 115–117, 203–205).
( 46 ) The Luck Objection
Daring libertarians accept BFD, and so does Kane. But Kane seems to go further; he seems to hold that what I am calling basically free decisions occur only in cases in which agents make dual efforts to decide or choose. My daring libertarian view shuns such efforts because I am skeptical about their existence.6 An objection that can be directed both at Kane’s view and at my daring libertarian view merits attention. Kane (2011a) calls it “the no-more-power objection” (p. 397). Like Kane’s view, DL is an event-causal libertarian view. Derk Pereboom (2001) claims that event-causal libertarianism fails because it “does not provide agents with any more control than compatibilism does” (p. 56). In the same vein, Randolph Clarke (2003) argues that “the active control that is exercised on [an event-causal libertarian] view is just the same as that exercised on an event-causal compatibilist account. [The] view fails to secure the agent’s exercise of any further positive powers to causally influence which of the alternative courses of events that are open will become actual” (p. 220). In Mele (2006), I suggest, in effect, that libertarians would do better to think in terms of kinds of control than in terms of amounts of control (as in “more control”; see Pereboom 2001, p. 56). Pereboom (2001) holds that agent causation, if it were to exist, would provide the “enhanced control” for which he calls (p. 55), and Clarke (2003) contends that “The requirement of agent causation . . . provides for the agent’s exercising when she acts, in addition to the active control secured by an event-causal view, a further power to causally influence which of the open alternatives will be made actual” (pp. 220–221). However, both Pereboom and Clarke are skeptical about agent causation—and rightly so, in my opinion (Mele 2006, ch. 3). Pereboom (2001) argues that although agent-causation is possible, it is extremely unlikely that anyone has agent-causal power (ch. 3). This is part of his argument for the thesis that no one has free will and no one is morally responsible for anything. In Clarke’s judgment, relevant arguments collectively “incline the balance against the possibility of substance causation in general and agent causation in particular” (2003, p. 209). Clarke also asserts that there is no evidence for the existence of agent causation (pp. 206–207). Libertarians contend that, in fact, some human 6. Kane tries to find a solution to a worry about luck in what is happening at and around the time of a decision. According to my daring libertarian view, it often is not the case that everything that is needed can be found there; we should also look back in time. A discussion of what a daring libertarian hopes to find in agents’ histories is beyond the scope of the present paper; interested readers may consult Mele (2006, pp. 111–134).
K ane , L uck , and C ontrol
( 47 )
beings sometimes act freely. And defending the claim that human beings sometimes engage in a kind of action that requires the existence of a species of causation for which there is no evidence is, to put it mildly, an unpromising project. What do I mean by kinds of control? We can distinguish, for example, between deterministic and indeterministic control. Control of the former kind is possible for agents in deterministic worlds and control of the latter kind is not. To forestall confusion, I point out that I do not place any special weight on the word “kinds” here. I need a term to contrast with “amounts” in the sphere of control, and “kinds” seems to be a reasonable choice. In discussions of comparative control in the free will literature, direct control is a prominent notion. Clarke (2003) writes: “Direct active control is exercised in acting, not before” (p. 166). Timothy O’Connor (2000) reports that “exerting active power is intrinsically a direct exercise of control over one’s own behavior” (p. 61). And Kane (1996) claims that agents exercise direct control over some of their choices (p. 144). In these cases, Kane says, the agent’s exercise of control is not “antecedent” to the choice; rather, it occurs “then and there,” when and where the choice is made. The following argument features direct control. I dub it the control argument against event-causal libertarianism (or CA, for short). (1) Necessarily, if an agent’s world is deterministic, then even if he has as much control as an agent can possibly have in a deterministic world, he lacks free will. (2) Necessarily, an agent with no agent-causal powers who has as much direct indeterministic control as can be had in the absence of agentcausal powers does not have a greater amount of control than an agent who has as much control as can be had in a deterministic world. (3) So, necessarily, even an agent with as much direct indeterministic control as can be had in the absence of agent-causal powers lacks free will if he has no agent-causal powers. If this argument is to be valid, it needs another premise. To see why, suppose that the following proposition is true: CAP. An agent with direct indeterministic control and no agent-causal powers can have free will even if he does not have a greater amount of control than an agent who has as much control as can be had in a deterministic world.
There is no explicit contradiction in the conjunction of (1), (2), and CAP; but CAP entails that (3) is false.
( 48 ) The Luck Objection
As far as I know, no one who argues that event-causal libertarianism is false on the basis of considerations of the sort featured in premises (1) and (2) has said how amounts of control are to be measured or how deterministic and indeterministic control are to be weighed on the same scale.7 Even so, I will grant premise (2) of CA for the sake of argument. If the premise is true, the agents at issue might have the same amount of control or it may be that direct indeterministic control and any kind of deterministic control are incommensurable. In any case, in the absence of an argument against CAP, CA is at best incomplete. Consider the claim that if two boats do not differ in horse power, the top speed of either cannot be greater than that of the other. This claim is false. Other features of boats are relevant to how fast they can move. Might amounts of control be like that in the sphere of free will? Someone might claim that if all relevant features of two agents that are not control features are equal, then if the agents do not differ in the amount of control they exercise at a time, either both act freely at that time or neither does. An argument for this claim may prove illuminating. For one thing, it may shed light on how to measure amounts of control and how to weigh deterministic and indeterministic control on the same scale. Premise (1) of CA—the incompatibilist premise—is relevant in this connection. A theorist who endorses it may say that agents in deterministic worlds do not have enough control over their actions to act freely, but the same theorist may add that these agents do not have enough control because they do not have the right kind of control. What is the right kind? According to some philosophers, having the right kind of control requires having agent-causal powers—powers that they themselves are inclined to regard as impossible (Clarke 2003, p. 209) and powers the existence of which they say we have no evidence for (Clarke 2003, pp. 206-207) or weighty evidence against (Pereboom 2001, ch. 3).8 According to others, the right kind of control is a species of direct indeterministic control that is not supplemented by agent-causal powers (Kane 1996, 1999b, 2011a). An event-causal libertarian may claim that an agent can exercise enough of this kind of control to act freely even if the amount of direct control he exercises does not surpass the greatest amount of control open to agents in deterministic worlds. It may be claimed that the amounts are incommensurable. Alternatively, it may be claimed that they might be equal. But anyone who makes the latter claim should tell us how to weigh deterministic and inde7. For a brief comparative discussion of amounts of indirect control in a particular connection, see Mele (2006, pp. 62–63). 8. Not all philosophers who claim that free will depends on agent causation are skeptics about agent causation. See O’Connor (2000).
K ane , L uck , and C ontrol
( 49 )
terministic control on the same scale, something that proponents of the no-more-power objection have neglected to do. Conceptual sufficiency and contingent sufficiency are very different, and both are sometimes talked about in terms of what is enough for something. I have discovered that ten pizzas are not enough for a party attended by my graduate students and that fifteen pizzas are enough. This is a contingent matter. I also learned long ago that for something to be a line, it is enough that it be a curve. This is a matter of conceptual sufficiency. In light of the simple distinction just mentioned, it is easy to see an ambiguity in the claim that Joe did not exercise enough control to act freely. On one reading, the claim is that the control he exercised is not part of something conceptually sufficient for an action’s being free. If Joe satisfied all necessary conditions for having A-ed freely that are independent of control, he might have failed to act freely because he did not exercise a certain kind of control. On the reading currently under consideration, one may try to defend the claim that Joan, unlike Joe, did exercise enough control to have acted freely without saying anything at all about relative amounts of control. What matters, one may think, is that Joan exercised a species of direct indeterministic control whereas Joe exercised only deterministic control. On another reading of the claim about Joe, what is being asserted is that the amount of control he exercised falls short of some amount required for free action. Similarly, on a reading of this kind, the claim about Joan is that she exercised a greater amount of control than Joe did and an amount great enough for her to have acted freely. Someone who prefers these readings of the claims about Joe and Joan should tell us how to measure amounts of control and how to weigh deterministic and indeterministic control on the same scale.9 Being a curve is sufficient for being a line. This sufficiency is not a matter of an amount of anything; it is simply a matter of definition. Might it be, similarly, that a particular exercise of direct indeterministic control is part of something conceptually sufficient for a particular action’s being a free action although even a maximal exercise of deterministic control cannot play this role and the two exercises differ in kind but not in amount? Perhaps, in the category of amount, they are incommensurable. One might offer the following reply: other things being equal, if the former exercise of control was sufficient to play the role at issue but the latter was not, then the former must have involved more control than the latter. But this is simply to repeat a thought that I have been challenging. Would 9. For a suggestion about this, see Kane (2011a, pp. 397–398).
( 50 ) The Luck Objection
someone who has this thought feel compelled to explain the following comparative fact about sufficiency in terms of different amounts of something? Although being a curve is sufficient for being a line, being a banana is not. In any case, a philosopher who offers the reply at issue ought to tell us how to measure control and how to weigh deterministic and indeterministic control on the same scale. When a proposal about these matters is advanced, it can be evaluated and we can revisit premise (2) of the control argument (which I have been granting for the sake of argument). Why are there event-causal libertarians? Part of the answer may be that some believers in free will do not regard the argument(s) that persuade them that incompatibilism is true as forcing them all the way to agent causation and are skeptical about the possibility or existence of agent causation. These believers in free will may understandably seek a libertarian position with the following two features: its commitments do not include any metaphysical or conceptual impossibilities, and its positive side—the thesis that there are indeterministic agents who sometimes act freely—is supported by evidence. I close this section with a brief discussion of the issue of evidence. If one is seeking hard evidence of indeterministic brain processes in animals, it makes sense to start small. Alexander Maye and colleagues report what they regard as evidence of indeterministic brain processes in fruit flies that affect the flies’ behavior (Maye, Hseih, Sugihara, and Brembs 2007; also see Brembs 2011). They also offer two (mutually compatible) hypotheses about why indeterministic brain processes for behavior initiation might have evolved. One is that unpredictability is required for survival (Maye et al. 2007, p. 8), and indeterministic brain processes of the sort at issue would result in unpredictable behavior. A predictable pattern of response to pursuit or attack tends to make an animal relatively easy prey (Brembs 2011). The other is that an animal’s unpredictable behavior enables it to learn “which portions of the incoming sensory stream are under operant control by [its] behavior” (Maye et al. 2007, p. 8). It is possible for an animal to be unpredictable to potential predators and to itself even in a deterministic universe. Even so, if animal brains are indeterministic organs, as Maye and his coauthors argue, the evolution of adaptive unpredictability might have benefitted from that feature of brains. Perhaps indeterministic agent-internal processes that play a role in producing behavior are part of our evolutionary heritage, and perhaps some such processes evolved because of their contribution to survival-promoting unpredictability. Possibly, some of the indeterministic processes at work in us are late processes, including processes that indeterministically issue in decisions. Some people have speculated about the low-level mechanics of
K ane , L uck , and C ontrol
( 51 )
indeterministic agent-internal processes. Henry Stapp (2007) suggests that there are quantum probability clouds associated with calcium ions moving toward nerve terminals (pp. 30–32). Readers will have noticed the words “perhaps” and “possibly” in the preceding paragraph. I certainly am not claiming that we have convincing evidence of indeterministic processes in action-producing streams in the heads of human beings. And I do not see how I can be in a position to claim sincerely that libertarianism is true until I have seen good evidence that human brains work in ways they need to work if libertarianism (about human agents) is true.
3. Conclusion
In section 1, I raised some questions about Kane’s “dual efforts” response to a worry about luck, and I pointed in the direction of an alternative way for an event-causal libertarian to deal with luck at the time of choice or decision (developed in Mele 2006, ch. 5). In section 2, I replied to an argument against event-causal libertarianism that features agential control. If that argument were successful, it would undermine both Kane’s view and the daring libertarian view that I floated. If I am right, the argument leaves both views unscathed.
•
Chapter 5
Toward a Solution to the Luck Problem John Martin Fischer
1. Introduction
P
erhaps the central problem for developing an adequate libertarian account of freedom and moral responsibility is the “luck problem.”1 The problem is not just a challenge for the libertarian; it is also an issue of interest to a semicompatibilist who believes that our moral responsibility should not be conceptualized so that it “hangs on a thread.” I have indeed contended that our status as persons and morally responsible agents should not hinge on the falsity of causal determinism. To suppose that we would need to adjust such basic views of ourselves in light of the discovery of the truth of the doctrine of causal determinism is, in my view, implausible. And I think it is equally implausible to suppose that we would have to give up these central views of ourselves—as free and morally responsible agents—if we were to become convinced that causal indeterminism obtains. I have focused largely on what might be called the “deterministic side of the equation.” More specifically, I have argued that even if causal determinism were true, we could still be fully and robustly morally responsible for our behavior. But what if causal indeterminism obtains? After all, presumably it is possible, for all we know, that causal indeterminism obtains. Some of the material in this chapter, including a version of the Random Machine Example, appeared in Fischer (2010b). I have greatly benefited from insightful comments on a previous version of this paper by Christopher Franklin. 1. This is really a family of related problems, although I will not take pains to distinguish the various different versions of the underlying problem. For a helpful overview and analysis of the luck problem, see Franklin (2011).
T oward a S olution to the L uck P roblem
( 53 )
It would be desirable, therefore, to have a strategy for protecting our status as free and morally responsible agents if we were to wake up to a headline in The New York Times, “Causal Indeterminism Is True!” Robert Kane’s account of libertarian free will represents a significant intellectual achievement (Kane 1985, 1996, 2007a). And he has also provided important resources in the attempt to answer the objections stemming from the luck problem. I shall begin by sketching these resources. I then highlight some of the challenges that remain, despite the important contributions made by Kane, insofar as we seek a persuasive reply to the luck problem. Finally, I build on previous work of my own to suggest a way of meeting these challenges so as to provide a compelling reply to the luck problem.
2. The Luck Problem, Kane’s Distinction between Two Kinds of Control, and Three Examples
It is not straightforward to formulate the doctrine of causal determinism. Here I will simply take it that, whatever else would follow from the truth of causal determinism, it would be the case that, for any given time, a complete statement of facts about the universe at that time, together with a complete statement of the laws of nature, would entail the total sets of facts about the universe at all subsequent times. Now of course it could be the case that causal determinism does not obtain, but not in any way that affects the sequences that involve human choices and behavior. But in this chapter I shall focus on the possibility that causal determinism is false in such a way as to result in causally indeterministic sequences leading to human choices and behavior. Let us suppose that causal indeterminism (of this sort) obtains, and that I choose at time t2 to raise my hand at time t3. It follows that a statement of the total set of facts that obtained at t1, together with a statement of the laws of nature, fails to entail that I choose at t2 to raise my hand at t3. Thus, everything that obtained just prior to t2, including everything about me—all my preferences, beliefs, dispositions, traits of character, and so forth—is completely compatible with my not choosing at t2 to raise my hand at t3 (even on the supposition that the laws of nature are held fixed). Given this implication of causal indeterminism (of the sort under consideration), it can seem puzzling how my actual choice at t2 to raise my hand at t3 could really be mine—could be in my control in the sense required for moral responsibility. After all, everything about me could be the same and yet I not make this choice. How then could it be my distinctive contribution that makes the difference between my actual choice and an alternative
( 54 ) The Luck Objection
choice? How exactly was it really up to me whether or not I made the choice I made? It would seem that causal indeterminism threatens the notion that we have the sort of control required for moral responsibility. Robert Kane points out that causal indeterminism entails that agents lack the capacity to ensure or guarantee their behavior in advance. Kane (1996) calls this capacity, “antecedent determining control,” and he says that “the ability to be in, or bring about, conditions such that one can guarantee or determine which of a set of outcomes is going to occur before it occurs, whether the outcomes are one’s own actions, the actions of others, or events in the world generally is something the libertarian agent cannot have” (p. 144).2 Kane (1996) however goes on to say: No doubt, such “antecedent determining control” as we might call it, is valuable in many circumstances; and we cannot help but value it by virtue of an evolutionary imperative to seek security and get control of our surroundings. But it does not follow that because you cannot determine which of a set of outcomes occurs before it occurs, you lack control over which of them occurs, when it occurs. When the conditions of plural voluntary control are satisfied, agents exercise control over their future lives then and there in a manner that is not antecedently determined by their pasts. (p. 144)
Kane’s view, then, is that we do not need antecedent determining control— the capacity to ensure our choices and actions prior to them—in order to exercise a kind of control that would render us morally responsible. According to Kane, a distinct kind of control is all that is necessary for moral responsibility and is entirely compatible with causal indeterminism— “plural voluntary control”: To have such control over a set of options at a given time is to be able to bring about any of the options (to go more-than-one-way) at will or voluntarily at the time. That is to say, it is to be able to do whatever you will (or most want) to do among a set of options, whenever you will to do it, for the reasons you will to do it, and in such manner that neither your doing it nor willing to do it was coerced or compelled. (p. 111)
On Kane’s view, when an agent has plural voluntary control, then either way (among the options available to the agent) the choice will not be “inadvertent,” “accidental,” “capricious,” or “merely random.” According to Kane 2. See also Kane (1999b).
T oward a S olution to the L uck P roblem
( 55 )
(2007a), this is because the choice would be willed by the agent either way, and the choice would be made for reasons either way—“reasons [the agent] then and there endorses” (p. 29). An important part of Kane’s reply to worries stemming from the luck problem is the distinction between antecedent determining control and plural voluntary control and his claim that only the latter, and not also the former, is required for moral responsibility. Kane also offers three interesting examples in which he contends that indeterminism (in the relevant location) is compatible with the sort of control required for moral responsibility. Kane’s first example involves thinking through a mathematical problem where there is some indeterminacy in one’s neural processes complicating the task such that it is undetermined whether the person solves the problem. Kane (2007a) says, however, that if the person solves the problem, he is clearly responsible for doing so even though it was undetermined whether he would succeed (p. 27). His second example is that of an assassin trying to kill the prime minister, but owing to a nervous twitch in his arm, he misses his intended target and kills the minister’s aide instead (p. 17). Kane says that had the assassin hit his target and killed the prime minster, he could be responsible for doing so despite the indeterminism because he “intentionally and voluntarily succeeded in doing what he was trying to do—kill the prime minister” (p. 27). His third example is of a husband who breaks a table in a fit of rage. Kane argues that the husband can be responsible for the table’s breaking even if, owing to indeterminism in his outgoing neural pathways, it is undetermined whether the table will actually break right up to the moment it is struck (p. 27). I now want to present some problems with taking these examples to help establish that agents can be morally responsible for their behavior in indeterministic contexts. Return to Kane’s first example (of solving a difficult math problem). He points out that it is plausible that you did in fact solve it and are morally responsible for solving it, although the process was indeterministic and thus it was undetermined whether you would succeed. Intuitively, though, you’re responsible for solving the math problem only if it is the result of your own choice. Kane (2007a) seems to be aware of this and writes, “Yet, if you concentrate and solve the problem nonetheless, we have reason to say you did it and are responsible for it” (p. 27). But what if this choice itself is the result of an indeterministic process? If this possibility is kept firmly in mind, it is perhaps less evident that you are indeed praiseworthy for solving the math problem. Whatever worries are raised by the luck problem are present here, given that the choice to concentrate and thus the act of concentration are themselves the results of indeterministic processes. So (contra Kane) it is not
( 56 ) The Luck Objection
evident that you can indeed be deemed praiseworthy for solving the math problem. A similar problem afflicts Kane’s other two examples. Plausibly, the intuition on which Kane relies to the effect that the assassin can be held responsible because he did what he was trying to do depends on supposing that at least the assassin controlled his choice. If he did not control his choice, why would it matter that he was doing what he was trying to do? And on the assumption of causal indeterminism (of the sort under consideration here), the assassin’s choice itself would be the result of an indeterministic process. That is, I am assuming here that the choice itself is the result of a causally indeterministic process.3 The assumption that the assassin controls his choice is thus called into question. The same problem affects Kane’s third example of the angry husband breaking the table. Kane (2007a) writes, “We suppose that some indeterminism in his outgoing neural pathways makes the momentum of his arm indeterminate, so that it is undetermined whether the table will actually break right up to the moment when it is struck” (p. 27). But note again that, as with the other two examples, the indeterminism explicitly posited by Kane is downstream from the choice. This fact can distract us from the fact that the choice itself also must be assumed to be the result of an indeterministic process. If we were to assume that the husband controlled his choice, then I would agree with Kane that the downstream indeterminism would not vitiate the intuition that the husband is morally responsible for breaking the tabletop. But this assumption is illegitimate, and if we keep firmly in mind that the husband’s choice was also the result of a causally indeterministic process, it becomes unclear whether he can properly be deemed morally responsible for breaking the tabletop. As above, when we recognize that the choice itself is the result of an indeterministic process, it is not clear how adding the subsequent indeterminism really helps. More specifically, it does not seem that the example provides any new resources for addressing the luck problem in its “primary manifestation,” that is, as it relates to the process leading to the choice itself.4 3. Various libertarians place the gap in causal determination in different places along the sequence to the action. I am here presupposing that the lack of causal determination is placed just prior to the choice; I find this the most plausible location for the lack of causal determination (given a libertarian picture). 4. Kane (2007a) offers a further example of three assassins, two of whom had a less than 100 percent chance of succeeding in killing the prime minister because of indeterminism (p. 39). As with his other cases, I think that, in arguing that all would be equally guilty of they succeed, Kane is tacitly assuming that all three assassins control their choices to the same degree. But, as I have argued with respect to the previous three examples, this is an illicit assumption, given indeterminism.
T oward a S olution to the L uck P roblem
( 57 )
3. Residual Worries
Of course, the examples are not the only elements in Kane’s toolkit. He also offers helpful theoretical resources, notably the distinction between antecedent determining control and the analysis of plural voluntary control. But despite these resources, the fact of indeterminism leading to the agent’s choice (and subsequent behavior) has certain worrisome implications. I shall now lay out two such implications, as developed in important work by Peter Van Inwagen and Alfred Mele. The first anxiety is expressed by Peter van Inwagen’s “Rollback Argument” (van Inwagen 2000). The Rollback Argument is a way of highlighting the worry that, under the assumption of causal indeterminism, it is not the agent who makes the crucial difference as to what happens. In order to have the control required for moral responsibility, the choices and actions must flow from the agent in the right way—it must be genuinely the agent’s action. And in order for the choice and action to be genuinely the agent’s, they must be suitably related to the agent’s prior mental states. More specifically, the Rollback Argument seeks to explain precisely why it would follow from causal indeterminism (in the relevant places) that an agent’s choices and actions would not be an outflowing of the agent in the required sense; that is, it seeks to pinpoint the reason why, if indeterminism were to obtain, the required relationship between the agent’s prior mental states and his choice (and action) would not be present.5 For ease of discussion, let us call this relationship, the “responsibility-grounding relationship.” 6 Suppose, again, that I choose at t2 to raise my hand (at t3), and suppose, further, that causal indeterminism is true (in virtue of indeterminacy in the relevant location—between the agent’s prior states and his choice). Now imagine that God rolls back the universe to some prior time, say t1 (where we assume here that t1 is just one minute prior to t2) and allows things to go forward again. Indeed, we can imagine that God causes the universe to revert to its precise state at t1 and things to go forward a thousand times. Given the assumption of causal indeterminism, sometimes I will indeed choose at t2 to raise my hand but sometimes I will not choose 5. I am here operating within a framework of event-causation, which is congenial to Kane’s approach. I am thus putting aside, for the purposes of my discussion here, the possibility that invoking agent-causation can help with the problem of luck. I do not mean to rule this possibility out by fiat; I am simply focusing on the event-causal picture here. 6. More specifically, the responsibility-grounding relationship is a certain relationship between an agent’s prior mental states and his choice which is necessary, but not sufficient, for moral responsibility; thus, to posit that this relationship obtains is not thereby to claim that the agent is morally responsible.
( 58 ) The Luck Objection
at t2 to raise my hand. And recall that everything about me at t1 is held fixed; in all the scenarios in the thought-experiment, all of my mental states—my dispositions, traits of character, values, and so forth—are the same as they actually are. If everything about my mind is held fixed and yet sometimes I choose at t2 to raise my hand and sometimes I do not, then it seems that (in the relevant sense) I do not make the difference between choosing to raise my hand and not choosing to raise my hand. Put in other words, it would seem that my choice is not related to my prior states in the way required by the responsibility-grounding relationship.7 It is important to see that the Rollback Argument would apply even if Kane’s conditions for plural voluntary control obtain. So even if an agent has plural voluntary control in a causally indeterministic scenario, it nevertheless might seem that his choice is not an outflowing of the agent in the way required for moral responsibility. That is, the satisfaction of the conditions for plural voluntary control is compatible with the absence of the responsibility-grounding relationship between prior mental states and the choice in question. Alfred Mele (2006) has developed another way of highlighting a residual worry—a problem that would be present under the assumption of causal indeterminism, even given the satisfaction of the conditions for plural voluntary control. In Christopher Franklin’s terminology, this corresponds to the “Explanatory Formulation” of the luck problem (Franklin 2011). As many philosophers have pointed out, it appears that if causal indeterminism obtains, then we cannot give a contrastive explanation of an agent’s choices and behavior. (I do not include Mele himself in this group, although he gives a version of what might be called the Explanatory Formulation.) That is, it seems that we cannot give an explanation of why I choose as I actually choose at t1 rather than making a different choice. Given the putative lack of availability of contrastive explanations under indeterminism, many have concluded that agents in causally indeterministic worlds cannot have the sort of control required for moral responsibility; after all, if we cannot even in principle explain why the agent chooses X rather than Y, it does not seem that it is genuinely up to the agent whether to choose X rather than Y. Mele brings out this sort of worry by considering a goddess Diana who is creating agents who satisfy event-causal libertarianism. She worries, however, about the luck problem: 7. Note that it does not seem to matter for the purpose of the thought-experiment just how many times out of a thousand I choose at t2 to raise my hand and how many times I do not so choose at t2. All that appears to be required is that sometimes I choose at t2 to raise my hand and sometimes I do not so choose, holding fixed everything about my mind just prior to the actual choice.
T oward a S olution to the L uck P roblem
( 59 )
Her worry, more specifically, is that if the difference between the actual world, in which one of her agents judges it best to A straightaway and then, at t, decides accordingly, and any possible world with the same past up to t and the same laws of nature and he makes an alternative decision while the judgment persists is just a matter of luck, then he does not freely make that decision in that possible world, W. Diana suspects that his making that alternative decision rather than deciding in accordance with his best judgment—that is, the difference between W and the actual world—is just a matter of bad luck, or, more precisely, of worse luck in W for the agent than in the actual world. After all, because the worlds do not diverge before the agent decides, there is no difference in them to account for the difference in decisions. (Mele 2006, p. 8)
Mele’s contention is the cross-world difference between the actual world and W is just a matter of luck, and thus the existence of luck poses a significant challenge to the agent’s being free in either world. In his work on the luck problem, Mele emphasizes the challenge posed by cross-world differences in decisions and behavior that are not accompanied by differences in antecedent conditions. Precisely these sorts of cross-world differences will be present under the assumption of causal indeterminism, even given the further assumption that in a particular case Kane’s conditions for plural voluntary control are met. I will call the problem Mele highlights the “Problem of Bare Transworld Differences”: transworld differences in behavior unaccompanied by differences in antecedent conditions. Given these two residual worries, it is not clear that the mere satisfaction of the conditions for plural voluntary control is sufficient for the freedom required for moral responsibility. In the rest of the chapter, then, I shall supplement the resources provided by Kane in an attempt to address the worries presented by the Rollback Argument and the Problem of Bare Transworld Differences. I shall contend that these worries do not in fact show that an agent in an indeterministic scenario cannot possess the freedom (or control) required for moral responsibility.
4. The Random Machine Example
I now wish to call into question the efficacy of the Rollback Argument in establishing that causal indeterminism rules out the responsibility-grounding relationship. (After my discussion of the Rollback Argument, I shall go on to show how my critique can also be applied to the Problem of Bare Transworld Differences.) Imagine a causally deterministic world W1 in which everything
( 60 ) The Luck Objection
goes as it is supposed to in the sequence issuing in a given human choice and action. That is, suppose causal determinism obtains and I choose (for my own reasons, in the “ordinary way”) at t2 to raise my hand at t3, and I do in fact raise my hand at t3. Imagine, further, that whatever is required for the responsibility-grounding relationship between my prior states at t1 and my choice at t2 to raise my hand at t3 is present; that is, let us say that the requisite glue that connects my prior states with my choice at t2 is present. I am not sure exactly what this glue consists in; that is, I am not exactly sure what is required for the responsibility-grounding relationship. But we can suppose that in W1 it—whatever it is—obtains in the case of my choice at t2 to raise my hand at t3.8 (Of course, it is not enough for my freely choosing at t2 to raise my hand at t3 that this glue be present. But at least we cannot say in W1 that I do not freely choose at t2 to raise my hand at t3 because the glue is absent.) Now imagine another possible world W2 in which everything is the same as W1 in respect of the way the causal sequence that actually leads to my choice at t2 to raise my hand at t3 (everything, that is, apart from causal determination). In W2, as in W1, I choose for my own reasons, in the “normal way,” at t2 to raise my hand at t3 (and I do indeed raise my hand at t3). In general, whatever exactly it is that makes it the case that the responsibilitygrounding relationship is present in the actual sequence flowing through me to my choice at t2 (and action at t3)—everything, that is, apart from causal determination—is also present in the sequence that takes place in W2. But now we add that there is a genuinely random machine in W2, but not in W1. Let’s say that I begin my deliberations at t1 about whether to raise my hand; my last moment of deliberation is t1.9 and I make my choice at t2. The random machine “operates” in W2 between times t1 and t2. (By “operating” I simply mean that the machine goes through a series of internal states culminating in either M1 or some other state at t1.9.) For our purposes, we can focus on state M1. That is, if the machine is in state M1 at t1.9, there are two possibilities, each with a fifty-percent objective probability attached to it. The first possibility is that the machine does nothing— it “goes to sleep,” as it were, and does not trigger any causal interaction 8. Note that this case, as I’ve presented it in the text, will not move a committed skeptic about whether the glue can be present ever—even in a causally deterministic context. Rather, my primary target in this chapter is a theorist who thinks that indeterminism poses a special problem for moral responsibility in virtue of posing a special challenge to the possibility of the glue’s obtaining. That is, there are philosophers who do not worry about the choice and action’s being an “outflowing” of the agent under causal determinism, but do worry that indeterminism would call into question the contention that the relevant behavior is an outflowing of the agent. It is to such philosophers that I address my argument.
T oward a S olution to the L uck P roblem
( 61 )
with the world (including me). The second possibility is that it will initiate a causal sequence that would preempt my choice at t2 to raise my hand at t3. That is, on the second possibility, the machine would trigger a causal sequence that would terminate in (say) a direct electronic stimulation of my brain sufficient to ensure that I choose at t2 to refrain from raising my hand at t3. It is assumed that the process involving the machine in W2 is genuinely random (whatever is required for genuine randomness). Further, let us suppose that, as things actually go in W2, the machine’s state at t1.9 is indeed M1, and, further, the machine simply “goes to sleep” and never triggers any causal interference in the sequence flowing through me to my choice at t2 to raise my hand at t3. (That is, the machine is in M1 at t1.9 and the first possibility is realized—no intervention in the causal sequence flowing through me.) In both worlds W1 and W2, I choose and do exactly the same thing (typeidentical choices and actions), as a result of relevantly similar processes. More specifically, we have assumed that the causal process linking my prior states and my choice is the same in relevant respects in both worlds; thus, if the requisite glue connecting the prior states with the choice obtains in W1, it also obtains in W2. Presumably, the mere existence and operation of the machine in W2 should not in any way threaten these claims about the responsibility-grounding relationship. How could the mere existence of such a machine affect the responsibility-grounding relationship, given that the machine does not causally interact with the sequence flowing through me and issuing in the choice at t2?9 Indeed, it should be intuitively obvious that the mere existence and operation of the machine in W2 is irrelevant to whatever it is that makes it the case that the responsibility-grounding relationship obtains in the sequence flowing through me. Notice, however, that W2 is causally indeterministic during the relevant interval. Indeed, W2 is causally indeterministic in the relationship between my prior states at t1.9 and my choice at t2; after all, my actual deliberations could have been preempted by a causal sequence that was not in fact 9. As Randolph Clarke has reminded me, there are tricky issues here. Is the machine not entering state M2 a cause? If so, it might not be exactly right to say that the machine doesn’t causally interact with Jones. Still, I here rely on some intuitive notion— difficult to specify—according to which the machine does not actually causally interact with the sequence flowing through Jones. Note that similar worries come up in the context of evaluating the classical Frankfurt cases (and variations on them). For example, the untriggered machine in W2 is parallel to resting Black in the original Frankfurt case. In both cases my contention is that they (resting Black and the untriggered machine) are (in some sense) not part of the actual sequence flowing through Jones and thus are irrelevant to Jones’s moral responsibility. In the context of a defense of their version of a Frankfurt case, Mele and Robb (2003) discuss such issues.
( 62 ) The Luck Objection
triggered in W2. So we could obviously run the Rollback Argument with respect to W2: If we were to roll back the universe to t1.9 and allow it to go forward a thousand times, then in say 468 “replays” I will choose at t2 to raise my hand at t3, and in say 533 replays I will be caused to choose at t2 to refrain from raising my hand at t3. The Rollback Argument clearly “applies” to the conditions of W2, as W2 is explicitly an indeterministic world. But the key point is that intuitively it is obvious, as I claimed above, that the mere existence and operation of the machine in W2 cannot in itself show that the requisite glue is not present in W2—it cannot show that whatever underwrites the responsibility-grounding relationship is missing. (Recall, again, that the machine’s “operating” refers simply to its going through a sequence of internal states; it does not imply the triggering of its capacity to initiate a preempting sequence.) Perhaps my point could be put as follows. Whatever underlies the responsibility-grounding relationship— whatever constitutes the relevant glue that binds together the prior states of the agent with his choice—is a matter that is intrinsic (in some sense) to the relevant causal sequence. It is a matter of the way the prior states of the agent lead to the choice in question, and this cannot be affected by the mere presence of something (such as the random machine in W2) that plays no role in the causal sequence flowing through the agent.10 And if this is correct, then the mere fact that the Rollback Argument can successfully be run cannot in itself show that the responsibility-grounding relationship is not present. After all, in W2 the responsibility-grounding relationship is indeed present; it is present to the same extent that it is present in W1. Yet we can run the Rollback Argument relative to the conditions present in W2. In a nutshell, then, my argument is as follows. Let’s suppose, what is not implausible, that we have some sufficiently determinate intuitive notion of “the way a causal sequence goes,” where this notion abstracts away from whether causal determinism obtains or not. We now suppose that in W1 the actual causal sequence goes in the “normal” way typically thought to ground 10. Christopher Franklin has brought it to my attention that a libertarian might resist my claim here. He claims that libertarians must contend that extrinsic features of causal sequences are relevant. This is because whether the causation in question is deterministic or indeterministic might well be an extrinsic fact. But if this is so, I would claim that it brings out an implausible feature of libertarianism, rather than a problem with my argument. Note that in this chapter I do not seek to argue for libertarianism; rather, I wish to argue that indeterminism is compatible with the kind of control that grounds moral responsibility. Of course, libertarianism combines the contention that indeterminism is consistent with moral responsibility with the further claim (that I do not accept) to the effect that causal determinism is incompatible with moral responsibility. Thus, the distinction between causal determinism and causal indeterminism matters for a libertarian in a way in which it does not matter for me.
T oward a S olution to the L uck P roblem
( 63 )
attributions of moral responsibility, apart from considerations pertinent to causal determination. Now add to W1 that causal determinism does in fact obtain. It should be widely accepted that, whatever the requisite glue (the responsibility-grounding relationship) is, it obtains in W1. Given that the responsibility-grounding relationship is present in W1 and intuitively cannot be expunged simply because of the existence of the genuinely random machine in W2, it also is present in W2. But we can successfully run the Rollback Argument with respect to the conditions present in W2; indeed, the mere existence of the machine in W2 makes it the case that the relationship between my prior mental states and my choice at t2 is indeterministic (even though “the way the causal sequences go” in W1 and W2, as defined above, is the same). Thus, the mere fact of the application of the Rollback Argument does not show what it is intended to show, namely, that the responsibilitygrounding relationship is absent. So the application of the Rollback Argument cannot be the reason why causal indeterminism threatens the responsibility grounding relationship—threatens to make us come unglued, as it were. Now someone might object that I have, as it were, put the indeterminism in the wrong place. That is, I have invited us to imagine a genuinely random machine that is quite external to me. But, of course, the proponent of the luck objection worries about indeterminism that is internal to agents. So even if the existence of the genuinely random machine does not pose problems for the possession of the kind of control that is required for moral responsibility, this would not in itself show that casual indeterminism—of the sort under consideration here, in which there are gaps in causal determination along the sequence that flows through my mind—poses no problems for the possession of responsibility-grounding control. Note first, however, that the introduction of the random machine in a remote location does in fact imply that the relationship between the prior states of my mind and my choice is indeterministic. So indeterminism does in fact obtain (in the hypothetical scenarios of the example) in precisely the place in which it is supposed to obtain, if the example is to be relevant to the luck worries. But a critic might push the objection as follows. Although there is indeed indeterminism in the “right place” in the hypothetical scenarios of the Random Machine Example, this indeterminism is a mere implication of randomness that is entirely external to the relevant agent. What is required, in order for the example to show something about indeterminism of the sort that might well obtain in our minds, is indeterminism that is sourced internally; in the hypothetical scenarios of the example, the indeterminism is sourced externally. Now it is not entirely clear to me why it matters whether the source of the indeterminism in the mind is external or internal. It seems to me that
( 64 ) The Luck Objection
(arguably, at least) what matters is the presence of the indeterminism, rather than its source. But in any case we can adjust the example to reflect the concern that the indeterminism be sourced internally. So simply imagine that the Random Machine is micro-miniaturized and installed in my brain in such a manner that it is on the causal sequence that leads from my prior mental states to my choice at t2. Of course, we are now so conceptualizing the Random Machine that it functions like a frictionless module. Roughly speaking, and assuming some sort of materialist picture of the mind, we can imagine that the electrical impulse that is heading toward my choice at t2 to raise my hand at t3 simply “flows through” the module without being affected by it in any way; it is in this sense that we are here thinking of the Random Machine as functioning—in the actual causal sequence—like a completely frictionless module. I contend that the essential features of the Random Machine Example are still in place; that is, I claim that the mere existence of the Random Machine does not affect my moral responsibility.11 Again: if the responsibility-grounding relationship holds in W1 and W2 (as above, with the external Random Machine), it should also obtain in a scenario just like W2, except that the genuinely random machine has been micro-miniaturized and installed in the causal sequence leading from my mental states at t1 to my choice at t2 to raise my hand at t3. And here the source of the indeterminism is internal.12 I contend that the Random Machine Example establishes that the Rollback Argument does not show that we would lack the control required for moral responsibility in a causally indeterministic universe. Similarly, the Random Machine Example helps to show that the mere lack of the availability of a contrastive explanation (of the relevant kind) does not in itself show that an agent does not possess the sort of control required for moral responsibility. After all, in W1 but not in W2, it is presumably 11. As with food, perhaps it is better if the indeterminism is sourced locally! My argument in the text shows that the proponent of the Random Machine Argument can be a “philosophical locavore,” as it were. 12. Perhaps a critic might contend that the scenarios are still not sufficiently similar to the phenomena they purport to model because the worrisome phenomena in the brain might be described as “random processes,” whereas all we have in the imaginary scenarios are indeterministic events. But, again, I am not sure why this difference makes a difference. Also, we could mirror the processes in the brain more closely by simply adding to the imaginary scenarios in the Random Machine Example a series of steps in a process. At each step (or at relevant steps), we could envisage a random machine that functions in the actual sequence like a frictionless module (as above). Such a machine (or set of machines) ensures that we have an extended process that includes indeterministic elements; but insofar as the machines function in the actual sequence as frictionless modules, they should not affect the glue that holds us together as agents (and that grounds our moral responsibility).
T oward a S olution to the L uck P roblem
( 65 )
possible to provide a contrastive explanation of my behavior by reference to states “internal to me” in the relevant sense—my motivational states and whatever particular states realize or constitute them. Similarly, in W2 (but not W1) Mele could run his argument from cross-world differences. But, as we have seen above, the responsibility-underwriting glue is present in W2, just as much as in W1. I thus conclude that the mere fact that the relevant cross-world difference are purely a matter of luck does not in itself show that the agent’s behavior results from luck in a sense that would rule out his moral responsibility. Consider the pair of worlds, W2 and W2*. As we know, in W2 the state of the machine at t1.9 is M1 and the machine “goes to sleep.” Imagine that in W2* the state of the machine at t1.9 is M1, and it swings into action (the other possible result of being in M1). Nothing else is different about W2 and W2* (up to t1.9). Thus, Mele’s argument applies: if it is sound, one could conclude that the difference between W2 and W2* is just a matter of luck. But (as above) if the responsibility-underwriting glue is present is W1, it is present in W2. So, although Mele’s argument applies to W2, the glue is nevertheless present in W2. Mele’s argument thus cannot in itself show that there is an insuperable problem (pertaining to luck) with causal indeterminism. I conclude, then, that the Random Machine Example helps to show that, even if the Rollback Argument and the Bare Transworld Differences Problem applied in a given indeterministic context, it would not thereby follow that the responsibility-grounding relationship fails to hold. Thus it would not thereby follow that the relevant agent does not possess the kind of control required for moral responsibility.
5. Counterfactual Interveners: Untriggered Ensurers and Untriggered Preemptors
In order to “situate” my strategy for addressing the luck problem within a larger context, it might be helpful to recall the so-called Frankfurt-style counterexamples to the Principle of Alternative Possibilities (PAP). These much-discussed examples were originally introduced in contemporary philosophy by Harry Frankfurt (1969) in order to impugn PAP, according to which moral responsibility requires the kind of control that involves freedom to choose and do otherwise.13 Here is an updated version of a Frankfurt example: 13. For a selection of papers on Frankfurt-style examples, see Widerker and McKenna (2003).
( 66 ) The Luck Objection Suppose that Black, a Democrat, has secretly inserted a chip in Jones’s brain which enables Black to monitor and control Jones’s activities. Black can exercise this control through a sophisticated computer that he has programmed so that, among other things, it monitors Jones’s voting behavior. If Jones were to show any inclination to vote for McCain (or, let us say, anyone other than Obama), then the computer, through the chip in Jones’s brain, would intervene to assure that he actually decides to vote for Obama and does so vote. But if Jones decides on his own to vote for Obama (as Black, the old progressive would prefer), the computer does nothing but continue to monitor—without affecting—the goings-on in Jones’s head. Now suppose that Jones decides to vote for Obama on his own, just as he would have if Black had not inserted the chip in his head. It seems, upon first thinking about this case, that Jones can be held morally responsible for this choice and act of voting for Obama, although he could not have chosen otherwise and he could not have done otherwise.14
It seems to me that Black’s presence (as described in the example), perhaps together with other features, makes it the case that Jones cannot choose or do other than he actually does. Further, it seems to me that Black’s presence (in the context of those other features) is irrelevant to Jones’s moral responsibility. Frankfurt agrees, arguing that it would be “quite gratuitous” to assign any weight to Black’s presence in assessing Jones’s moral responsibility. After all: The circumstances that made it impossible for him [Jones] to do otherwise could have been subtracted from the situation without affecting what happened or why it happened in any way. (Frankfurt 1969, pp. 836–837)
In other words, Black did not play any role in the “actual sequence”—the actual causal pathway to Jones’s choice and action; Black’s device, although present, is untriggered. I think that the Frankfurt-style examples help to provide motivation for an “actual-sequence” approach to moral responsibility, according to which moral responsibility attributions depend on (possibly dispositional or modal) features of the actual sequence, rather than on the availability of genuinely open alternative possibilities. The mere presence of certain sorts of untriggered ensurers (such as Black’s device) rules out alternative possibilities without in any way affecting the actual sequence that issues in the relevant behavior. Now return to the Random Machine Example. The random machine in W2 is also a merely counterfactual intervener. It is an untriggered preemptor. 14. I originally presented such an example in Fischer (1982).
T oward a S olution to the L uck P roblem
( 67 )
But both untriggered ensurers (such as Black) and untriggered preemptors (like the random machine) are equally untriggered; that is, they are merely counterfactual interveners, and, as such, they are plausibly irrelevant to attributions of moral responsibility. The plausibility of the Frankfurt-style cases and the Random Machine Example rests on a basic intuition, first formulated by Harry Frankfurt, to the effect that mere counterfactual interventions are irrelevant to ascriptions of moral responsibility. In both cases it is crucial that we have a counterfactual intervener—an individual or device that is poised to intervene but remains dormant. As I would put it, moral responsibility is a matter of what happens in the actual sequence. Think of it this way. Assume (for simplicity’s sake) that the counterfactual intervener, Black, in the Frankfurt example makes it the case that Jones cannot do otherwise. (More carefully, it would be Black together with other factors that would have this implication, but put this aside for now.)15 Frankfurt’s point is that insofar as Black’s presence is irrelevant to Jones’s moral responsibility, it follows that the fact that Jones cannot do otherwise is similarly irrelevant to Jones’s moral responsibility. Now assume (again, for simplicity’s sake) that the genuinely random machine makes it the case that the Rollback Argument applies in my example above (and that Jones lacks antecedent determining control, the relevant bare cross-world differences exist, and so forth). Insofar as the mere presence of the machine is irrelevant to Jones’s moral responsibility in the example, it follows that the fact that the Rollback Argument applies (and that Jones lacks antecedent determining control, and so forth) is irrelevant to Jones’s moral responsibility.
6. Conclusion
Robert Kane has made significant contributions to the ongoing debates about free will and moral responsibility. He has honestly faced up to the worries stemming from the luck problem, and he has offered resources that make substantial progress in addressing these issues. Kane would surely agree with my contention, however, that more work needs to be done, in order fully and adequately to assuage the anxieties of the proponents of the luck problem. In this chapter I have offered the Random Machine Example in the hope that this will accomplish at least some of the remaining work. My strategy relies on precisely the same intuition that drives my “Frankfurt-style” approach to arguing that moral responsibility need not 15. For further discussion, see Fischer (2010a).
( 68 ) The Luck Objection
be conceptualized as requiring the sort of control that involves access to alternative possibilities. That is, the strategy relies on the intuitively plausible point that when some factor plays no role in the explanation of an event, it is thereby irrelevant to the issue of whether a given agent is morally responsible for the event. Unlike Kane, I do not seek to vindicate libertarianism as the true theory of freedom and moral responsibility. But like Kane, I believe that indeterminism is consistent with freedom and moral responsibility. Further, I welcome any argument that supports this consistency, because I hold that our status as free and morally responsible agents should not hang on a thread— should not depend on whether or not casual determinism obtains.
Pa r t I I I
••• Incompatibilism and Omissions
•
Chapter 6
Compatibilist Ultimacy: Resisting the Threat of Kane’s U Condition Michael M c Kenna
A
distinctive feature of Robert Kane’s libertarian theory of free will is the role played by the condition of ultimacy. According to Kane, ultimacy is necessary for free will: a person acts of her own free will only if her act originates from her as its ultimate source. Indeed, Kane makes ultimacy not only a necessary condition for free will, but its defining feature. “Free will,” Kane (1996) writes, is “the power of agents to be the ultimate creators (or originators) of their own ends and purposes” (p. 4, original emphasis). Compatibilists, Kane argues, cannot make room for ultimacy. While they can account for free action, because ultimacy is beyond their reach compatibilists cannot account for free will. Is this true? As a compatibilist, I shall argue that it is not.
I am delighted to contribute to this festschrift honoring the excellent work of Robert Kane. On top of making one of the most important contributions to work on free will in our time, reshaping the way many of us argue about this ancient problem, Bob has also distinguished himself as a kind, supportive man, willing to encourage young people, and always open to learning from anyone. Early in my career, when I was just one more anonymous youngster, he made time for me and offered his support and guidance. I am in his debt. For helpful conversations on the central ideas developed in this chapter, along with Bob himself, I would also like to thank Alfred Mele, Eddy Nahmias, David Palmer, and Carolina Sartorio. I am especially grateful to David for inviting me to contribute to this collection and for his hard work putting this volume together.
( 72 ) Incompatibilism and Omissions 1. Kane’s Account of Ultimacy
Kane’s libertarianism is unique insofar as he advances an ability to do otherwise (or alternative possibilities, or AP) condition for free will, but he does not rely upon arguments designed to show that this ability is incompatible with determinism. Typically, incompatibilists who endorse an AP condition argue that free will is incompatible with determinism because AP is incompatible with determinism.1 In Kane’s estimation, the dispute between compatibilists and incompatibilists cannot proceed past a stalemate by attending exclusively to AP (Kane 1996, p. 59). Why? Compatibilists can plausibly appeal to different modal intuitions and competitor semantics for terms like “can” and “able,” thereby giving them access to a compatibilist construal of AP. On Kane’s view, incompatibilism is best advanced by attending to a distinct condition of ultimate responsibility (UR). Kane’s UR concerns responsibility for the origin or, as he sometimes puts it, the arche of an agent’s action. It has to do with “the source or explanation of the action that is performed; that source must be ‘in us’ ” (Kane 1996, p. 34). When an agent acts of her own free will, the will from which that act issues is one that the agent constituted for herself; she made her will to be as it is. And, furthermore, there does not exist any further cause or explanation that is itself altogether distinct from her and is a source of her activity of making her will to be so. In developing UR, Kane (1996) distinguishes between an R condition and a U condition (p. 35). The R condition requires that for any act an agent performs of her own free will, she is personally responsible for that act in the sense that she voluntarily brought it about; her voluntarily doing so made a difference as to whether or not it occurred. The U condition then further requires that anything that is “an arche (or sufficient ground or cause or explanation)” of her act and her will—the will as source of the pertinent act—is also something for which the agent is personally responsible (p. 35). As Kane notes, the R condition can readily be accommodated from within the resources of a compatibilist theory (p. 73). Hence, if we are to account for free will in terms of AP+UR, given Kane’s openness to the prospects of compatibilist accounts of AP, it is the U condition alone that motivates an incompatibilist diagnosis (p. 73).2 1. For example, see Ginet (1990) or van Inwagen (1983). Both defend an AP condition, and both rely heavily on arguments for the incompatibility of determinism and the ability to do otherwise. For an example of incompatibilists who reject an AP condition, see Pereboom (2001) or Stump (1996). 2. To avoid any misunderstanding, I reject an AP condition. But that needn’t detain us here. For present purposes, we can grant Kane AP.
C ompatibilist U ltimacy
( 73 )
Clearly, given the way Kane formulates U, compatibilists cannot satisfy Kanean ultimacy. If determinism is true, then the facts of the remote past in conjunction with a complete statement of the laws of nature entail every truth about every later time. Suppose determinism is true. Then for any state of an agent’s will as it issues in a free act, there will exist a state of the world prior to that agent’s existence which is such that it, in conjunction with the laws of nature, entails that the agent’s will issue in that act—and this clearly violates Kane’s requirement U.3 Every act of an agent (a finite, nongodlike agent) will have a sufficient explanation that does not involve conditions for which that agent is personally responsible.4
2. Ultimacy Is Situated within Kane’s Theory of Free Will
How does U fit within Kane’s overall theory of free will? Three elements bring the role of U into clear view: (1) the nature of the will; (2) a distinction between three kinds of free action; and (3) the requirement of plurality and the nature of torn decisions. Will: For Kane, the will is a highly complex faculty enabling agents to be directed toward their ends (Kane 1996, pp. 21–31). Kane distinguishes between three different senses of “will” (p. 26) as the term has traditionally been used, each of which is complex in its own right, and each of which plays a role in his complete view of the will’s multifaceted nature. One sense concerns the wants, desires, and preferences that are, or give rise 3. I am assuming that an agent cannot be ultimately responsible for (temporally nonrelational) facts about the past prior to her birth, or for the laws of nature. Perhaps some compatibilists would wish to resist this. I don’t. David Palmer has offered an especially insightful comment about Kane’s assumption that U cannot be satisfied at deterministic worlds. Call Kane’s assumption UD. It appears that the basic inference is that if p entails q then p is a sufficient ground, cause, or explanation for q. Reflecting upon this, one can see that there are clear counterexamples. The truth of the proposition [2 + 2 = 4 or (p and not-p)] entails the truth of [2 + 2 = 4], but presumably the former proposition is not a sufficient ground, cause, or explanation of the latter. My suspicion is that Kane’s inference UD is more subtle. It has to do with the elements of the entailing facts as they bear on determinism. When the facts are about the past and the laws entailing future conduct, it is reasonable to think that this sort of entailment pertains to facts that bear upon being a sufficient ground, cause, or explanation. It’s an interesting issue, however, since it gives rise to the important question of just how we are to construe the notion of being what Kane calls an arche. I’ll set the issue aside in the remainder of this chapter and grant Kane UD. 4. This would not hold for deterministic worlds in which agents exist at every time in the past. For a compatibilist who exploits this prospect as it bears on a different set of incompatibilist arguments, see Joseph Campbell (2007).
( 74 ) Incompatibilism and Omissions
to, an agent’s reasons for action. A second concerns choices, decisions, and intentions. Choices and decisions are to be understood as mental acts whose function is to settle uncertainty, terminate deliberation, and issue in intentions (pp. 23–24). Intentions are, in turn, states of mind that persist through time and guide action by way of providing plans for organizing both immediate and future action (pp. 24–25). A third concerns tryings, or efforts of will, which have to do with efforts to complete overt acts as they issue from settled intentions, as well as follow through on our mental activities to do things such as persist in deliberation and settle on one’s intentions and her course of action (pp. 26–28). Kane labels them, respectively, appetitive will, rational will, and striving will (p. 27). Where is indeterminacy in the will required? Somewhere between appetitive will and rational will (Kane 1996, p. 27). This is crucial to understanding U’s role. On Kane’s view, it would be no threat to a libertarian conception of free will if the process giving rise to desires and preferences was a deterministic one. Nor would it threaten libertarian freedom if the relation between intentions once formed and overt actions was also deterministic. What would, however, undermine libertarian freedom is if the relation between an agent’s reasons once acquired and her subsequent choices and decisions was deterministic. Settling uncertainty about what to do, for Kane, given the reasons one has, is where the incompatibilists will locate distinctly libertarian freedom. So it is in this “space” in the etiology of action where Kane’s U will find a special place. It will give rise to an agent’s intentions exclusively by means which are such that the agent is also personally responsible for those means. Three types of free acts: Compatibilists, Kane (2008) maintains, rightly capture an important class of free acts by distinguishing acts that are done voluntarily, on purpose, and for reasons that are not coerced or compelled (pp. 142–143). For ease of reference, let us call these Plain Free Acts (PFAs). A more narrowly circumscribed class of acts is marked by those that are done of one’s own free will (p. 143). These acts issue from a will for which an agent is ultimately responsible. That is, not only did an agent perform the act voluntarily, on purpose, and for her own reasons (in keeping with PFAs), but the will that was the source of her voluntary act, the purpose, motive or character issuing in that act, was one for which she was ultimately responsible. For ease of reference let us call these Freely Willed Acts (FWAs). Yet a narrower class of free acts is marked by those acts by which one creates, shapes, settles, forms, or sustains her will. They are, as Kane (2008) puts it, “will-setting” acts, and so are ones by which one forms herself as a purposive being (p. 143). They are character-shaping. Following
C ompatibilist U ltimacy
( 75 )
Kane, let us call these Self-Forming Acts (SFAs). On Kane’s view, compatibilists cannot account for FWAs or SFAs. SFAs require direct exercises of undetermined free action in shaping one’s will from conditions that were not antecedently determined. FWAs require that the will issuing in one’s free acts be a will that was acquired by way of SFAs. Plurality and Torn Decisions: According to Kane, SFAs only occur in special contexts. If an agent faces a circumstance in which, given her reasons and motives as issuing from her appetitive will, only one course of action recommends itself, there is nothing to occasion her to do anything more to further shape her will. Hence, it would be natural to assume that her rational will would simply settle on an intention suited for the pertinent reasons and motives. But when in conditions of uncertainty an agent finds herself with compelling reasons for opposing courses of action, she is then faced with the prospect of settling what to do. In doing so, not only will she resolve any immediate questions about what intention to form and how to act, she will, in choosing or deciding what to do, also contribute to her own self and character, thereby shaping her will in the future. These conditions of uncertainty, however, require that, for a range of alternatives, an agent’s choosing amongst any of them in this range would count as rational, voluntary, and suitably controlled (Kane 1996, p. 114). This is Kane’s condition of plurality. An often-discussed example is Kane’s case of the businesswoman late for a meeting who witnesses an assault (Kane 1996, p. 126). She has competing reasons and motives. Self-interest counsels proceeding on to the meeting, but her moral compass counsels pausing to help. Both alternatives pull on her, and in different directions. Whichever way she chooses, her doing so would be voluntary, rational, and under her control; neither, given her psychic constitution, would be deviant in any of these ways. In exercising her rational will, she performs the mental act of choosing between competing values, not only settling on what to do with respect to the assault before her but also contributing to making her into the sort of person she is. Facing the situation, she is, as Kane (1996) puts it, “torn” about how to act (p. 126), and it is in facing decisions such as this one that the U condition has immediate application. Nothing independent of the agent can settle for her, just then, what to do, and in her settling on what to do at that time, in the context of a torn decision,5 she contributes to the making of herself.
5. Mark Balaguer (2004) deserves credit for the expression “torn decisions.”
( 76 ) Incompatibilism and Omissions 3. The Historical Significance of SFAs
In light of the preceding discussion, the following picture of free agency begins to emerge. Persons act freely in a straightforward manner captured by PFAs regardless of whether determinism is true. But if agents ever act of their own free will, those FWAs must issue from intentions or purposes that an agent came to acquire freely. What is required for an agent to act from a will that is free? In the cases of many FWAs, probably most, all that is required is that the pertinent psychological ingredients in the proximal etiology of her acts be ones for which she is personally responsible. These ingredients have to be hers, in the sense that she helped form them (Kane 2008, p. 144). These motives, purposes, and character traits might themselves be secured within her overall psychic constitution so that she “just is” the way she is at the time roughly coincident with the action. In this sense, she might, at that time, be unable to do otherwise, and so fail to satisfy an AP condition within the context of that time frame. Indeed, her performing of these FWAs thus might arise from locally deterministic causes that are internal to her own motivational psychology. But they will still count as FWAs so long as the proximal will-constituting psychic ingredients themselves are the products of other FWAs wherein, at earlier times in her history, the agent acted in such a way as to then directly shape her character, her motives, and purposes by choosing or deciding in conditions of uncertainty involving torn decisions. During these earlier periods of indeterminacy, the agent would then, at those times, be able to do otherwise in so fashioning herself, and there would be no antecedent sufficient conditions for her so forming of herself. In these moments, in performing these SFAs, she would directly satisfy the U condition in shaping her very nature. The relevant notion of will at play would be the rational will wherein agents freely decide or choose their ends and purposes. These SFAs would be the historical ancestral links that, at later times with subsequent FWAs, the agent’s freedom of will would “transfer through” her history. And so, for these later FWAs, it would also be true of them that U would be satisfied because the ultimate source of their motives at those later times could be traced by way of earlier SFAs to times at which the agent, in the context of torn decisions, made or created her will. Finally, on Kane’s historical view, the acquisition of character traits and will-setting by earlier SFAs is not typically the upshot of some single free choice. One shapes one’s will for subsequent FWAs by repeated activity, in piecemeal fashion, through a series of SFAs over the course of a life (Kane 2008, pp. 148–151). Hence, the freedom of our agency, when we act of our own free wills is, at least sometimes, an achievement that represents the long-term shaping of our dearest conceptions of our own characters.
C ompatibilist U ltimacy
( 77 )
4. Compatibilists’ Shortcomings?
Having now carefully presented Kane’s understanding of ultimacy, I wish to pose a challenge. Why, given the details presented thus far, can’t a compatibilist simply take on Kane’s penetrating picture? As I see it, there is but one particular point where compatibilists cannot agree with Kane. I’ll turn in the sections to follow to an examination of this point. But before doing so, I pause to consider where, other than with U itself, there are even any putative compatibilist shortcomings when measured against Kane’s thoughtful libertarian theory. Of course, Kane grants that compatibilists can explain PFAs. But what about FWAs in which, at the time of action, an agent’s freedom for them is to be accounted for by her earlier will-setting acts? Kane insists that only incompatibilists can account for such acts, since the requirements for willing freely from features of one’s own self at a later time must trace back to SFAs understood in a certain way—that is, as satisfying U. But setting the particular condition of U aside for the moment, it has to be granted that compatibilists can make some sense of the idea that in acting as she does, an agent might have previously shaped her character, her motives, and purposes through prior actions so that now (something like) her freedom in acting from the will she has is traced back to those earlier acts. So as not to beg any questions, let us call these FWA*s. But now, turn to Kane’s SFAs. It can also be granted that even if determinism is true, agents would sometimes face conditions in which they were torn and unsettled about what to do. And in those contexts, sometimes they would—in at least some sense of free—freely choose, and choose in ways that would after all shape their characters and their developing selves. Again, so as not to beg any questions, call these acts SFA*s. Kane should be able to grant that this is a kind of self-creation, even if it might remain in dispute that it is truly—really, genuinely—free in some deep way we all care about. There is, furthermore, nothing that should keep compatibilists from claiming that SFA*s typically shape an agent’s character and will over time by contributing incrementally to character shaping; it is typically a pattern of conduct over the evolving history of an agent’s life that settles her will and character.6 Now recall Kane’s tripartite distinction between the appetitive, the rational, and the striving will. The dispute between compatibilists and incompatibilists, when placed within the structure of Kane’s theory, reduces to 6. The compatibilist picture suggested here is similar to the sort of picture Mele (1995, 2006) has proposed (despite wishing to remain agnostic about the compatibility/incompatibility issue).
( 78 ) Incompatibilism and Omissions
one point of contention regarding the rational will—any freedom-enabling features of the appetitive and striving will both being friendly to a compatibilist treatment. Suppose, as I have just contended, that compatibilists can lay claim to FWA*s and SFA*s. And suppose that they too can tell a nuanced historical story about the incremental emergence of an agent who is able to act from a will that she participated in creating for herself. What remains in dispute is whether these compatibilist-friendly FWA*s and SFA*s actually are FWAs and SFAs, or if they are mere simulacra of the genuine article. This, in turn, apparently reduces to the question of whether, as regards SFAs, mental acts in the contexts of torn decisions can be acts for which an agent is indeed free even if Kane’s U is not satisfied. So what are compatibilists to make of the condition of ultimacy and Kane’s condition U?
5. A Misplaced Criticism of Kane’s Commitment to Ultimacy
A useful way to begin is by considering Galen Strawson’s formulation of a similar condition (Strawson 1986, 1994, 2002). According to Strawson, the concept of moral responsibility is incoherent since it requires conditions that are not metaphysically possible to satisfy. On Strawson’s view, in order to be morally responsible for what you do, you have to be morally responsible for the way you are mentally as it issues in what you do. But in order to be morally responsible for the way you are now mentally as it issues in what you do, you must be morally responsible for some prior way that you were mentally that then resulted in the way that you are now mentally (Strawson 2002, pp. 443–444). This regress only ends, Strawson argues, if you can be morally responsible for your very self, and this requires the incoherent condition that you are a causa sui—a cause of yourself. In short, moral responsibility requires ultimacy in the sense that it requires that you create yourself ex nihilo, which is an absurdity. In my estimation, Strawson’s impossibilist argument relies upon a conception of responsibility that invokes unnecessarily high standards. Strawson himself contends that its implausibility is beside the point; he’s talking about our ordinary concept of moral responsibility. Here, some might resist by arguing that he’s flatly wrong to find evidence for thinking of responsibility this way. But as I have remarked elsewhere (McKenna 2008c), it might well be that there is a strand in our ordinary thought that has this exceedingly high expectation built into it. We can grant Strawson this much. That, however, is no reason to conclude that the concept of
C ompatibilist U ltimacy
( 79 )
moral responsibility is constrained by this high standard. Our concept might be multifaceted. About this very high standard, John Martin Fischer (2006b) has astutely remarked that it reflects a kind of philosophical megalomania (p. 116). Why? Because it is exceedingly unreasonable to hold such a high standard for human agents, demanding of them that they have the power of Gods, only to turn around and conclude that none of us mere mortals can satisfy it. Why not settle for a more credible, attainable concept, one that actually informs our practices? Both Fischer (2006b) and I (McKenna 2008c) have made this a key point in our indictment of Strawson’s ultimacy argument. This criticism of Strawson cannot be leveled against Kane, and for two reasons. First, Kane is not committed to the absurdity of self-creation as a condition of free will and moral responsibility. All he is committed to is responsibility for creating pertinent features of one’s will as it issues in certain sorts of free actions. Kane can allow that there is much that is given and for which we are not ultimately responsible (e.g., character traits inherited from family and environment, primitive desires, and so forth). In brief, as in comparison with Strawson’s extreme view, Kane downgrades the standards for ultimacy and contends that they are in our reach. To this extent he does not deserve the indictment of philosophical megalomania. Second, Kane, is not committed to the key premise that leads Strawson to the extreme conclusion. Strawson contends that we are morally responsible for what we do only if we are morally responsible for the way we are mentally as it issues in what we do. With this in mind, focus just on an SFA. Consider the case of Kane’s businesswoman. Suppose she performs the mental act of choosing to help the victim. In this case, is it true that for her to be responsible for her SFA of choosing to help, she also has to be responsible for the way she was mentally as it issued in her choice to help? It seems not. Why? Because on Kane’s view, there is no settled way she was prior to her act of choosing that explains why she chose to stay and help rather than proceed on to her meeting.7 In so choosing, she was, just then, settling the way she was—she was just then 7. In correspondence, Kane has confirmed that I am correct to interpret him in this way. However, he points out a minor wrinkle that, as he notes, does not bear on the main point. For various SFAs, there may after all be a certain respect in which an agent is responsible for the way she was mentally as it played a role in an SFA. In particular, if a character trait had an influencing but not fully determinate role in how she chose, then some of her responsibility for her choosing in the context of that SFA might be linked to the developed character traits already in place. But in this sort of case, the agent would only be responsible for this trait if there were yet distinct, earlier SFAs on the basis of which she came to acquire that trait.
( 80 ) Incompatibilism and Omissions
forming herself (in a way that doesn’t involve Strawson’s megalomaniacal standards).8
6. The Ultimacy Argument and Kane’s Grounds for U
Grant that Kane’s condition of ultimacy avoids the indictment of philosophical megalomania. A different tactic is to charge that his commitment to U relies on nothing more than a brute incompatibilist intuition. Elsewhere, I have leveled such a charge against Saul Smilansky’s (2003) use of ultimacy, and I suggested that Kane too is open to this objection (McKenna 2008c, pp. 195–196). Is this a fair criticism of Kane? Upon reflection, I now do not believe it is. Explaining why will help point the way to a clearer assessment of how he advances U. Consider the following lean formulation of an argument that, so far as I can tell, Kane clearly endorses, The Ultimacy Argument (UA): (1) A person acts of her own free will only if she is the ultimate source of her act. (2) If determinism is true, no one is the ultimate source of her acts. (3) Therefore, if determinism is true, no one acts of her own free will.9 As I previously noted (McKenna 2008c), how a compatibilist responds to this argument depends crucially on how we are to understand ultimacy. If we simply grant that an agent’s being an ultimate source of her action requires something like Kane’s U, then the compatibilist must resist the first premise of the argument. U and other similar conditions are patently incompatible with determinism, and so under such a reading the second premise is unimpeachable. 8. In correspondence, Carolina Sartorio has suggested that Kane can be understood as accepting the following proposition: we are morally responsible for what we do if and only if, if there is a way we are mentally that issues in what we do, then we are morally responsible for the way we are mentally. This stops Strawson’s regress, as it is consistent with our being morally responsible for what we do in cases in which there is no settled way we are mentally that issues in what we do. What distinguishes Kane from Strawson, then, is that Kane rejects, and Strawson endorses, this proposition: we are responsible for what we do only if there is a way that we are mentally that issues in what we do, and we are responsible for that as well. 9. I have reworded the argument slightly to suit Kane’s commitment to the expression “free will.” Previously, I expressed it in terms of “acting freely in the sense required for true moral responsibility” (McKenna 2008c, p. 192).
C ompatibilist U ltimacy
( 81 )
The question immediately before us, however, is whether we should after all understand being the ultimate source of our actions in terms of a condition such as U. Above (sec. 1), I opted to paraphrase Kane’s U. Here it will be useful to quote him directly: (U) for every X and Y (where X and Y represent occurrences of events and/or states) if the agent is personally responsible for X, and if Y is an arche (or sufficient ground or cause or explanation) for X, then the agent must be personally responsible for Y. (Kane 1996, p. 35)
Note that, so formulated, U is not cast as a negative condition. But this is misleading. Kane’s U is after all a negative condition.10 It is equivalent to the claim that there does not exist an arche of anything for which an agent is responsible which is not also such that the agent is responsible for that arche. Indeed, just prior to introducing (U) in The Significance of Free Will, Kane approvingly quotes Martha Klein’s negative formulation: Agents should be ultimately responsible for their morally relevant decisions or choices—“ultimately” in the sense that nothing for which they are not also responsible should be the source [or cause] of their decisions or choices. (Klein 1990, p. 51; quoted in Kane 1996, p. 34)
Making perspicuous the negative claim is crucial. This shows that there is no way such a condition could be satisfied if determinism is true. Indeed, it looks nearly synonymous with the mere expression of the brute incompatibilist intuition that an agent acts of her own free will only if her act is not the product of a deterministic history. And if this is what U comes to, one wants to know if there is an argument for it. If not, those endorsing it in the context of issuing an argument for incompatibilism, such as UA, face the charge of begging the question; the second premise implicitly takes on what is tantamount to the conclusion of an argument for incompatibilism. This is my criticism of Smilansky (McKenna 2008c, pp. 194–197).
10. For the record, I think it is better to develop a principle like U in terms of a necessary condition rather than a sufficient one (McKenna 2008c, p. 192). Specifying exactly how to do so is, however, a delicate matter. Careful examination of Kane’s considered view of U as he develops it (in his 1996) suggests that he is alive to the difficulties. I’ll not pursue this here. For present purposes, U as Kane has formulated it will suffice.
( 82 ) Incompatibilism and Omissions
Interestingly, another incompatibilist who also endorses a condition similar to Kane’s U is Derk Pereboom. Indeed, Pereboom introduces his condition in the context of writing approvingly of Kane’s. Here is Pereboom’s: (O) If an agent is morally responsible for her deciding to perform an action, then the production of this decision must be something over which the agent has control, and an agent is not morally responsible for the decision if it is produced by a source over which she has no control. (Pereboom 2001, p. 4)11
Is Pereboom subject to the criticism I have leveled against Smilansky? No. As I have explained (McKenna 2008c, p. 197), Pereboom does not fall prey to the objection before us because he does not treat the Ultimacy Argument as a freestanding argument and implicitly make use of O as a premise in that argument. Rather, he treats O as the conclusion to a distinct argument for incompatibilism, a manipulation argument. Very roughly, without attending to various details, the general form of a manipulation argument proceeds as follows. Begin with the case of an agent who is massively manipulated by covert means into acting from a state that satisfies pertinent compatibilist minimally sufficient conditions for free will and moral responsibility. Then consider The Manipulation Argument (MA): (1) Any agent so manipulated does not act of her own free will and is not morally responsible for what she does. (2) Determinism is no different in any relevant respect from being so manipulated. (3) Therefore, if determinism is true, no one acts of her own free will and no one is morally responsible. Pereboom gets O by then inferring that what both determinism and such covert manipulation take away is (something like) ultimate sourcehood. In my estimation, Kane’s work is open to the very same interpretation. Although Kane is not as explicit about this dialectical maneuver, as I read him, it is clearly at the heart of his argument for U.12 Kane (1996, pp. 64–71) introduces a manipulation argument in an effort to refute Harry
11. Pereboom has more recently refined O, but his 2001 formulation will do for present purposes. 12. On this point in particular, I am especially indebted to Alfred Mele, who asked if we could read Kane in this way.
C ompatibilist U ltimacy
( 83 )
Frankfurt’s (1971) compatibilist account of free will.13 He then considers compatibilists’ efforts to resist the second premise of MA by showing that there is some relevant difference between mere determination of an agent’s will and covert nonconstraining control (CNC). Kane counters by pointing out that there remains a relevant similarity that eclipses any of the differences compatibilists might try to fix upon. He writes: [T]here is one power that persons essentially lose by being CNC controlled simply because the control is CNC; and this power is one that especially interests incompatibilists. It is the power to be the ultimate source or origin of one’s own ends or purposes rather than have that source be in something other than you. This is, of course, the governing image of free will that we tried to spell out in terms of UR: the origins or sources, or archai, of actions should be in the agents themselves and not in something outside the agents, if the agents are to be ultimately responsible for what they do and what they are. (Kane 1996, p. 70)
My contention is that, in this passage, Kane has all of the resources, just as Pereboom does, to contend that he is in no way appealing to a condition like U as a brute intuition. He is able to argue for it from the resources of a manipulation argument.
7. Compatibilists’ Options for Resisting the Ultimacy Argument
As I have explained, if ultimacy is understood in terms of U, compatibilists have no choice but to resist the first premise of UA. They must argue that acting of one’s own free will does not require being the ultimate source of one’s act. Does this matter? Perhaps compatibilists ought to avoid appeals to ultimacy. They could argue that the right kind of mediated sourcehood offers enough freedom worth wanting. Efforts to explain how a determined agent could be the ultimate source of her actions only invite unnecessary incompatibilist worries. Kane himself notes that for a class of free acts, PFAs, compatibilists have done well in explaining them and showing their 13. Due to space constraints, I’ll not set out the details of Kane’s way of discharging a manipulation argument here. I’ve also omitted presenting Pereboom’s as well. I’ve only offered the reader the general form of a manipulation argument. But this is enough, I hope, to make the general dialectical point at issue in this section and the next. Elsewhere (McKenna 2004), I’ve taken up Pereboom’s and Kane’s manipulation arguments directly.
( 84 ) Incompatibilism and Omissions
importance. So, even if compatibilists can resist Kane’s U, and so the second premise of UA, why bother? Why not just focus on the first premise of UA and deny that free will requires ultimate sourcehood? Just work on advancing super-nifty mediated sourcehood, and then show that it buys enough freedom to help justify our moral responsibility practices, show why life is meaningful, make room for human creativity, and so on.14 I think this would be a mistake. Compatibilists would be ill-advised to avoid a battle over ultimacy. An adequate theory of free agency should be able to make sense of a person’s shaping her life. There is, for instance, the familiar expression “a self-made man” (forgive the gendered formulation). And there is substance to the thought that sometimes, as one struggles with a hard choice, she settles not only on what to do just then, but she also shapes what sort of person she will become. She sets her will—or at least begins to. Kane’s lovely picture of free will, of the historical relation between FWAs and other SFAs in the shaping of a life, is not something that compatibilists should be prepared to give up. We want to capture the thought that for some features of our own agency we are, after all, their source. Thus did we freely shape ourselves. So, I reject the second premise of UA: if determinism is true, no one is the ultimate source of her acts. Incompatibilists appealing to some variant of U to defend this premise face a dilemma. Either U is the product of stipulation, and “ultimacy” as it appears in the premise functions as a term of art, one expressing a brute incompatibilist intuition; or U is something that can be shown by good argument to follow from reflection on ultimacy’s nature. But the former is off the table; it’s patently question begging. It is not how Kane himself proceeds. As for the latter, if the basis for accepting U rests on some instance of MA, then I would note, in opposition to Kane, that it is in dispute whether any formulation of it is sound. Space does not permit an examination of MA, but here it is worth considering that compatibilists have developed a variety of ways of resisting it.15 Assuming, then, that U rides on an adequate defense of MA, and that MA is open to dispute on compatibilist grounds, the compatibilist is dialectically positioned to contend that, barring the demonstrable success of MA, it is an open question how ultimacy is to be understood. What, then, beyond resisting MA, might I say in response to Kane to undermine his appeal to U while embracing ultimacy?
14. I suspect that those who find Frankfurt’s (1971) hierarchical theory of free will appealing are inclined to this sort of dialectical strategy. 15. See, for example, Frankfurt (1975), Fischer and Ravizza (1998), Haji (1998), and McKenna (2004, 2008a).
C ompatibilist U ltimacy
( 85 )
8. Claiming Ultimacy for Compatibilists
In Kane’s own incompatibilist formulation of ultimacy, and in a variety of the passages he cites from contemporary and historical incompatibilist figures (Kane 1996, pp. 33–35), there are two elements. One is that the source of an agent’s act be within her. Another, inferred from the first, is the negative claim that there is no further arche of what’s within her for which the agent is not also ultimately responsible. It is this inference that the compatibilist must resist. She needs to hold that an agent can be the ultimate source of her will and her acts even if there is a deterministic explanation for them that traces back to factors for which she is not ultimately responsible. When fixing just on the free will topic, it is easy to be drawn into thinking of ultimacy in such a way that it amounts to a contradiction to assert the first element and deny the second. But outside of this context, the natural way to think of X’s being the ultimate source of Y does not involve any such inference. Indeed, the very idea of it appears misplaced. We speak, for instance, of the original or ultimate source of Perrier drinking water as being a spring somewhere in the south of France, and we regard the refrigerator or the bottling plant as, by contrast with the spring, a mediated source for the sparkling water in our glass.16 In ordinary contexts, it would never dawn on us to think that whether the water in our glass really originated in France turned on whether determinism was true or not. Similarly, we ask about the origin of our species, and we seek the most immediate proximal link to homo sapiens, thinking that, when we discover it, we’ll have learnt the original or ultimate source of our biological kind. Was it Lucy and her kin, or is there some closer link? That’s an open question. No one need think, in such a context, that this explanation—this appeal to ultimate origins—is in competition with the Big Bang, and to be settled by discovering whether the laws of nature are after all deterministic. Think of how pervasive in the body of scientific knowledge accounts of origins are, and how natural it is to distinguish between mediated and ultimate sources. The difference between them, it appears, is a function of explanatory interest. This suggests that the notion of being an ultimate source of one’s will, in contrast with a mediated source, is a context-sensitive one, where that sensitivity tracks explanatory interest. It is therefore open to compatibilists to contend that, in the absence of stipulating that U is required for ultimate sourcehood, the proper context for settling whether a person ultimately formed herself, or pertinent features of her character 16. I develop this point elsewhere (2008c), and I credit Randolph Clarke with the example, as well as help in thinking through the central point of the criticism.
( 86 ) Incompatibilism and Omissions
or will, via an SFA, is located in the domain of ordinary folk psychological discourse. Was this choice or that decision a torn one? Did the agent struggle with it? Did her settling on what to do shape her? And so on. Here, it is open to the compatibilist to propose positive accounts of what is involved in making such decisions or choices freely. Was she choosing under duress, or in a condition of depression, or a state of ignorance? For my part, I would propose doing so in terms of reasons-responsiveness. The details do not matter here. What does matter is that, with the assumption that U is in dispute, it is open to compatibilists to offer positive accounts of free action in the contexts of SFAs and torn decisions. But, the incompatibilist might insist, when we are involved in debate about the metaphysics of free agency, ordinary folk psychological discourse is not the proper context. The proper context is, so to speak the ultimate one—something like the Nagelian view from nowhere. Fair enough. Of course, as a compatibilist, I am not inclined to agree. But here we can see where the lines in the sand are to be drawn. Now we can have a debate about what contexts should and should not fix claims about ultimate sourcehood, Nevertheless, it is an aid to the compatibilists to point out that there are numerous contexts in which ultimacy has perfectly intelligible applications that in no way are meant to turn on whether determinism is true. There are, on the other hand, few contexts, save for highly derived ones motivated by philosophical theorizing, where presuppositions of ultimacy are imposed from the view from nowhere. To the above remarks, the compatibilist might also add a warning. The incompatibilist should guard against a fallacy of reasoning similar to the sort at play when it is claimed that there are really no solid objects since there is so much space between subatomic particles. To make this mistake one must fail to appreciate that the sort of explanation underwriting the micro-level claim does not in any way compete with ordinary claims about which substances are solid and which are not. Likewise, a compatibilist might claim, locating the ultimate source of pertinent features of an agent’s character and will in her own torn choices or decisions in her own history is not—at least not on its face—in competition with an explanation of the causes of those same mental acts in terms of facts of the remote past and the laws of nature.17 17. Here, I am reminded of Eddy Nahmias’s (2011) recent work on bypassing. Nahmias contends that some intuitive motivation for incompatibilism, at least amongst the folk, is the product of a fallacy of reasoning according to which, if there is a microphysical explanation of the causes of our conduct in terms of things like states of the brain, then that bypasses mental states like desires, beliefs, intentions, choices, and so on. Once we come to see that mental states and processes are not eliminated by micro-level causation but constituted by it (identical with it), these explanations will no longer be regarded as competing. Perhaps a similar charge could be made here with
C ompatibilist U ltimacy
( 87 )
9. Concluding Remarks
Robert Kane has made the condition of ultimacy central to his arguments for incompatibilism and his libertarian theory of free will. He tells us that we need to attend to a condition of ultimacy, not an AP condition, to make progress. He has also made clear his strong conviction that the free will problem needs to be refashioned by a return to focusing upon the importance of the will itself and to questions of the will’s freedom. I hope that as a compatibilist I have honored him in this festschrift by taking him at his word and joining issue with him on his own terms. While I disagree with him about whether being the ultimate source of one’s freely willed actions requires the falsity of determinism, I’m quite certain that he is correct to insist that we attend to these matters more than we have, and I’ve profited immensely from studying his work.
respect to incompatibilist worries about ultimacy. There’s a bypassing worry: if there is some remote deterministic causal story whereby it is shown to be true that an agent at a later time performs the mental act of choosing or deciding in a condition of uncertainty, then that mental act can no longer be thought of as the agent’s ultimately settling for herself what to do and how to be as her life continues to unfold. The remote deterministic explanation “bypasses” the folk psychological one. Here, it might be countered, the former does not bypass the latter as ultimate source but gives rise to it.
•
Chapter 7
The Direct Argument for Incompatibilism David Widerker and Ira M. Schnall
1. Introduction
T
raditionally, incompatibilists have employed the following argument to show that determinism is incompatible with moral responsibility (henceforth the “Traditional Argument”):
(1) If determinism is true, then no agent could have avoided acting as he did. (The incompatibility of avoidability with determinism, IVD) (2) An agent is morally responsible for acting as he did only if he could have avoided acting as he did. (Principle of Alternative Possibilities, PAP) (3) Therefore, if determinism is true, then no agent is morally responsible for his acts.
In the early nineteen-eighties, Peter van Inwagen (1983) proposed another argument for the same conclusion, one that argues “directly” for it from some general and allegedly uncontroversial assumptions about moral responsibility and determinism (pp. 182–188). Let “PAST” stand for a true proposition describing the state of the world at some time before the existence of the human race, “LAWS” stand for the conjunction of the laws of nature, and “PRESENT” stand for any true propWe would like to thank Ori Beck, Carl Ginet, Dovid Gottlieb, and Michael Pauen for excellent comments and discussions on earlier versions of this chapter.
T he Direct A rgument for I ncompatibilism
( 89 )
osition about the state of the world now. Assume now the validity of the following two rules of inference: (A) ◻ p ⊢ NR (p) (B) NR (p), NR (p ⊃ q) ⊢ NR (q) where “◻” represents broadly logical necessity, “⊢” stands for “therefore,” “NR(p)” abbreviates “p and no one is (now), or ever has been even partly morally responsible for the fact that p,”1 and “⊃” stands for material implication. Let us call Rule (B) the “Transfer of Non-Responsibility Rule” or “Transfer NR” for short.2 Van Inwagen’s argument may then be formulated as follows: (1) Assume that the thesis of determinism is true. (2) ◻ [(PAST & LAWS) ⊃ PRESENT] From 1 (3) ◻ [PAST ⊃ (LAWS ⊃ PRESENT)] From 2, by logic (4) NR [PAST ⊃ (LAWS ⊃ PRESENT)] From 3, by (A) (5) NR (PAST) Assumption (6) NR (LAWS ⊃ PRESENT) From 4, 5, by Transfer NR (7) NR (LAWS) Assumption (8) NR (PRESENT) From 6, 7, by Transfer NR 1. An explanation of the “even partly” phrase is in order. Van Inwagen (1983) inserted it in order to forestall a certain type of counterexample to Rule B. For example, consider the following argument: NR(John throws a six) NR(John throws a six ⊃ John plays dice) Therefore, NR(John plays dice). If we omit the “even partly” phrase from the definition of “NR(p),” then it is easy to imagine a situation in which the premises are true but the conclusion false, and we have a counterexample to Rule B. However, if “even partly” is included, then, in the situation under consideration, the premise “NR(John throws a six)” is false, since John is partly responsible for throwing the six, by virtue of being responsible for playing dice in the first place, and so this argument is not a counterexample to Rule B (p. 243, n. 28). In that footnote, Van Inwagen also suggests that, instead of including the “not even partly” in the definition of “NR(p),” one could define “NR(p)” as “p, and no human being is morally responsible for p or for any logical consequence of p.” However, it is far from clear that this latter definition fully captures the intuitive meaning of “not being even partly responsible for p.” Hence, in what follows, we prefer to use van Inwagen’s original definition of “NR(p),” relying on an intuitive understanding of “being partly responsible” as bearing some responsibility for the fact that p. On this point, see Shabo’s perceptive remarks in his (2010, pp. 244–246). 2. Van Inwagen uses “N” where we use “NR.”
( 90 ) Incompatibilism and Omissions
Therefore, if determinism is true, then no one is, or ever has been, even partly morally responsible for PRESENT, that is, for any given fact that obtains now.3 Let us call this argument the “Direct Argument” or “DA” for short.4 DA has been welcomed by incompatibilists for a variety of reasons. Van Inwagen (1983) thought that he could, by means of it, establish the incompatibility of moral responsibility and determinism without employing the “could-have-done-otherwise” or avoidability notion of free will, and in this way sidestep disputes about the proper analysis of this notion (p. 183). Others have found DA attractive because of related worries compatibil ists have voiced about the incompatibilist defense, via the Consequence Argument (van Inwagen 1983, pp. 93–95), of premise (1) of the Traditional Argument. Since DA does not employ the notion of avoidability, it is not subject to those worries.5 Finally, some incompatibilists were dissatisfied with the Traditional Argument because they became convinced by Harry Frankfurt that premise (2) of that argument (PAP) is false.6 These incompatibilists, however, still hold that determinism is incompatible with moral responsibility. Naturally, they too have welcomed DA, precisely because it dispenses with the notion of avoidability, a notion that they think is irrelevant to moral responsibility.7 Although differing in their motivation, all of these parties saw in DA an important dialectical tool that could be used to convince compatibilists (or those who are undecided) of the incompatibility of moral responsibility and determinism. Robert Kane, too, has argued for the incompatibility of determinism with moral responsibility. He does so on the basis of his concept of ultimate responsibility, or UR. “The basic idea is this: to be ultimately responsible for an action, an agent must be responsible for anything that is a sufficient reason, cause, or motive for the action’s occurring” (Kane 2005, p. 121).8 Based on this concept, we may put forward the following incompatibilist argument:
3. From now on, for simplicity, we will use “responsible” to mean “at least partly responsible” (and “not responsible” to mean “not even partly responsible”), unless otherwise indicated. 4. We borrow the term “Direct Argument” from John Fischer and Mark Ravizza (1998, ch. 6). 5. For those compatibilist worries, see, for example, Lewis (1981), Horgan (1985), Flint (1987), and Kapitan (2002). 6. See Frankfurt (1969, pp. 829–839). 7. Here we have in mind, for example, Stump (1999), Zagzebski (2000), Hunt (2000), and Pereboom (2001, p. 33). These incompatibilists are sometimes referred to as “Causal History Incompatibilists” or as “Source Incompatibilists.” 8. For a more elaborate definition of this concept, see Kane (1996, p. 35).
T he Direct A rgument for I ncompatibilism
( 91 )
(1) If determinism were true, then (given the laws of nature), every human act would have a sufficient cause that occurred in the distant past, prior to the existence of human beings. (2) But no human being is responsible for something that occurred prior to the existence of human beings. (3) Therefore, if determinism were true, then no human being would be ultimately responsible for any action.9 This argument shares with DA the dialectical advantages of using neither IVD nor PAP as a premise and of not employing explicitly the controversial notion of avoidability. However, any argument based on UR also has an important dialectical disadvantage: it would not convince a compatibilist. That is, a compatibilist would claim that an agent can be fully morally responsible for an action with out being morally responsible for everything that is a sufficient reason, cause, or motive for the action’s occurring; and that therefore, despite the UR-based argument, one can be morally responsible for one’s actions even if determinism is true. In fact, applying contraposition to Kane’s characterization of UR quoted above, we obtain a version of the fundamental incompatibilist intuition that if an action is a causal result of factors for which the agent is not responsible, then the agent is not responsible for the action; and of course, the compatibilist, as such, denies the validity of this intuition. This disadvantage is not shared by DA, since it does not employ the notion of ultimate responsibility. Nevertheless, DA has recently come under severe attack from various quarters. In this essay, we offer a critical assessment of DA as an argument intended to refute compatibilism, or at least to shift the burden of proof onto the compatibilist. We examine a variety of objections to DA raised by John Fischer and Mark Ravizza (1998), David Widerker (2002), Ishtiyaque Haji (2009), Michael McKenna (2008b), and Seth Shabo (2010). We divide these into objections based primarily on dialectical considerations (section 2) and objections in the form of counterexamples to the Transfer NR inference rule involved in DA (section 3), although dialectical considerations will sometimes be relevant to the assessment of counterexamples. In the end, we contend that the proponent of DA can deal plausibly with most of the objections that have been raised to it. 9. Our UR-based argument differs somewhat from Kane’s own. See, for example, Kane (2005, pp. 122–123). We prefer to use ours simply because of its brevity. The remarks in the text on our argument apply also to Kane’s.
( 92 ) Incompatibilism and Omissions 2. Dialectical Objections
2.1 McKenna’s Dialectical Objection McKenna (2008b, p. 370) argues that it was dialectically improper for van Inwagen to use Transfer NR in the context of the debate between compatibilists and incompatibilists, and that therefore DA should never have gotten off the ground. To see what exactly is at issue, consider how van Inwagen justified the use of Transfer NR in DA. He did this by appealing to examples such as: NR (John was bitten by a cobra on his thirtieth birthday) NR (John was bitten by a cobra on his thirtieth birthday ⊃ John died on his thirtieth birthday) /∴ NR (John died on his thirtieth birthday) and NR (Plato died in antiquity) NR (Plato died in antiquity ⊃ Plato never met Hume) /∴ NR (Plato never met Hume)10 (Following McKenna, we shall call the two examples “Snakebite” and “Plato,” respectively.) Van Inwagen claims that the above arguments are intuitively valid and that their validity seems to be due to their exemplifying Transfer NR. Hence, these arguments show that prima facie Transfer NR is a valid rule of inference. Anyone who, at this point, wishes to deny the validity of Transfer NR bears the burden of proof. But McKenna disagrees. He argues that van Inwagen’s examples fail to establish Transfer NR as a rule of inference that can properly be employed in an argument against compatibilism. The reason for this failure, according to McKenna, is that neither of the examples involves a causal chain that passes through “a normally functioning agent who exercises unimpaired deliberative capacities” (McKenna 2008b, p. 376). Yet, DA was meant to apply precisely to cases involving such chains, since it is just such cases that are the focus of the debate between compatiblists and incompatibilists. Thus, the examples that van Inwagen adduces in support of Transfer NR are irrelevant to this debate. McKenna concludes that, with respect to the propriety of using Transfer NR in DA, the burden of proof is not on the 10. Both examples are from van Inwagen (1983, p. 187).
T he Direct A rgument for I ncompatibilism
( 93 )
critic to produce a counterexample to that rule; rather it is on the proponent of DA to provide supporting examples of the right sort for the rule. In the absence of support of the right kind, Transfer NR, and therefore DA, has no dialectical clout in the debate over compatibilism (McKenna 2008b, p. 377).
2.1.1 Response #1: Meeting McKenna’s challenge One way of responding to McKenna is to provide just the kind of example that he claims van Inwagen has failed to provide, that is, an intuitively valid instance of Transfer NR that involves a causal chain that passes through a normally functioning agent. Consider the following example, which we may call “Reprimand”: One morning, the commanding officer in the army’s anti-missile defense system receives a call from one of his subordinates, Jones. Jones says he is ill and cannot come in today. Unknown either to the officer or to Jones, there happens to be a man, Smith, in the area who looks exactly like Jones. Later in the day, the officer is on his way to another base when he sees Smith, whom he takes to be Jones, frolicking on the beach. The officer, after some deliberation, resolves to give Jones a severe reprimand the next day, which he does.
Jones, of course, does not deserve the reprimand. But is the officer morally responsible—that is, blameworthy—for reprimanding him? It seems not; for he is not blameworthy for believing that Jones was frolicking on the beach when he should have been defending the country, and he is not blameworthy for this belief’s leading him to reprimand Jones. That is, using “B” to stand for “the officer believes that Jones was frolicking on the beach” and “R” for “the officer reprimands Jones,” we can argue: NR (B), NR (B ⊃ R)/∴ NR (R). This argument obviously has the form of Transfer NR, and it seems to be intuitively valid. Furthermore, the officer is “a normally functioning agent who exercises unimpaired deliberative capacities.” So we seem to have here just the kind of example that McKenna requires.11 11. For a more elaborate discussion of this example, along with another one, see Schnall and Widerker (2012, pp. 30–31). In particular, we argue there that anyone who denies the conclusion of the Reprimand-argument is committed to denying one of the premises—a strong indication of its validity.
( 94 ) Incompatibilism and Omissions
2.1.2 Response #2: The Dialectical Situation Another way to respond to McKenna’s argument is to challenge his assessment of the dialectical situation—in particular, his claim that van Inwagen’s examples fail to support the validity of Transfer NR in a way that is relevant to the compatibilist/incompatibilist debate. We think that van Inwagen’s strategy is dialectically appropriate, and that by introducing Transfer NR into the debate, he has succeeded in shifting the burden of proof onto the compatibilist. In this sense, he has advanced the debate between compatibilists and incompatibilists. McKenna’s response indicates that he does not fully appreciate van Inwagen’s strategy. To see this, let’s describe van Inwagen’s strategy in some detail. First, let’s clarify the role played by the examples Snakebite and Plato in his argument. Recall Snakebite: NR (John was bitten by a cobra on his thirtieth birthday) NR (John was bitten by a cobra on his thirtieth birthday ⊃ John died on his thirtieth birthday) /∴ NR (John died on his thirtieth birthday) The first thing to note is that van Inwagen was not trying to inductively generalize from it (and similar examples) to the conclusion that most likely, all instances of Transfer NR are valid. (If he were, we might wonder how he could get away with generalizing from just two examples.) Rather, the role of the examples was to elicit a certain logical and conceptual intuition. We realize, through reflecting on an example like Snakebite, that if its premises are true, the conclusion must be true as well, and that therefore, Transfer NR, which encapsulates this intuition, must be valid. To see this more clearly, let’s apply what we call “the reverse argument” to Snakebite. Suppose Alice claims that someone is responsible for John’s dying on his thirtieth birthday. We ask Alice whether she thinks that some one set the snake on John, and therefore is responsible for the snake’s biting John. She says: No. We then ask her whether she thinks that someone was in a position to stop the snakebite from killing John—for instance, by sucking out the venom before it entered John’s bloodstream—but intentionally refrained from doing so, and therefore is (partially) responsible for the conditional fact that if the snake bites John on his birthday, then John dies on his birthday. She again answers: No. Alice’s claim is very puzzling; for since she admits that no one is responsible for the snake’s biting John or for the bite’s leading to John’s death, how can she say that someone is
T he Direct A rgument for I ncompatibilism
( 95 )
responsible for John’s death? Our puzzlement at Alice’s claims indicates that we find Snakebite, and thus Transfer NR, valid. Once the logical and conceptual intuition expressed by Transfer NR has been elicited by examples like Snakebite and Plato, Transfer NR acquires the status of being prima facie valid—always.12 For if a rule of inference, or an argument form, is valid, then it is always valid. With Transfer NR, van Inwagen now has at his disposal a new principle which is distinct from the principles to which incompatibilists standardly appeal. The new principle (which Transfer NR expresses in the form of a rule of inference) is that nonresponsibility transfers across a material conditional, or, more precisely, that nonresponsibility transfers from antecedent to consequent across a material conditional for (the truth of) which no one is responsible.13 At this point, van Inwagen is prima facie justified in applying this new principle, via DA, to all cases, including the controversial cases involving normal agency. The burden of proof is on the person who denies the unrestricted validity of Transfer NR. She must either (a) provide a clear counterexample to Transfer NR, or else (b) provide a plausible alternative to Transfer NR, which, like Transfer NR, would account for the validity of Snakebite but which, if substituted for Transfer NR in DA, would not yield an argument against compatibilism.14 McKenna does not attempt to satisfy (a). However, one might suggest that implicit in McKenna’s discussion is a proposal that satisfies (b); for he says: If the only uncontroversial cases one can cite to establish Transfer NR are such cases [i.e., cases like Snakebite and Plato], this strongly indicates that Transfer NR is best restricted to cases that do not “pass through” [normally functioning] . . . agents. (McKenna 2008b, p. 376)
In other words, McKenna proposes to account for the intuitive validity of the likes of Snakebite by appeal to a restricted version of Transfer NR, e xcluding 12. This status can be challenged—for instance, by presenting a counterexample to the validity of that rule. However, as long as such an example has not been presented, and as long as no alternative way to explain the intuition of validity is forthcoming, we are justified in regarding Transfer NR as a valid rule of inference. 13. Note that this is not the same as the principle that nonresponsibility is preserved through causal determination by factors for which the agent was not responsible. 14. For a beautiful example of how a philosophical principle may be disarmed by explaining away the examples that are used to support it, see William Rowe’s (1980) refutation of Richard Taylor’s argument for fatalism.
( 96 ) Incompatibilism and Omissions
from its range of application cases of normal agency. Note, however, that those excluded cases are precisely the controversial cases over which the incompatibilist and the compatibilist disagree. McKenna does not explain why Transfer NR should not apply to these cases; he simply asserts, without explanation, that the compatibilist is entitled to claim that it does not apply. So it seems that what McKenna recommends, in effect, is that Transfer NR is valid, except where it conflicts with compatibilist intuitions. To us, McKenna’s suggestion seems ad hoc and dialectically improper.15
2.2 Widerker’s Dialectical Criticisms of DA Widerker has raised two dialectical objections against DA. (W1) DA depends for its soundness on the assumption that determinism is incompatible with avoidability (IVD), an assumption which, since it involves the contested notion of avoidability, is controversial between compatibilists and incompatibilists. Hence, DA does not have the dialectical force envisaged for it by van Inwagen.16 (W2) The rationale for accepting the premise NR(LAWS) of the Direct Argument is that we do not have the power to render the laws of nature false, which means that the argument tacitly assumes the unwanted notion of avoidability that van Inwagen thought he could do without. (Widerker 2002, p. 324)
2.2.1 Assessment of (W1) To establish (W1), Widerker describes a deterministic scenario in which an agent, Jones, deliberately performs a heinous act A for a selfish reason, knowing very well that in doing so he is acting immorally, and believing that he could have done otherwise. Widerker then raises the question whether Jones is blameworthy for A. Incompatibilists would want to answer this question negatively, since the scenario is deterministic. But suppose, Widerker continues, someone claims that Jones could have avoided 15. For a more extended discussion of these dialectical issues, see Schnall and Widerker (2012, pp. 31–35). 16. Widerker (2002, pp. 322–323) originally aimed this argument at the validity of Transfer NR or any improved version thereof. However, we think that the argument is best construed as targeting DA in general. Of course, one who views Transfer NR as the most (or the only) vulnerable aspect of DA will interpret the argument as implicating Transfer NR specifically.
T he Direct A rgument for I ncompatibilism
( 97 )
acting as he did, and that therefore Jones is blameworthy; after all, he knew that he was acting wrongly and knew also that he could have acted otherwise. At this point, the proponent of DA seems to be faced with a dilemma. He cannot reply that, since the scenario under consideration is deterministic, Jones could not have refrained from acting immorally, for this would be, in effect, assuming (IVD)—something the proponent of DA would not want to do. On the other hand, he cannot afford to ignore the question of the avoidability of Jones’s act. Doing so would mean tacitly legitimizing the claim that Jones could have avoided acting immorally despite determinism, and therefore is blameworthy for what he did; and this, in turn, would mean that DA is unsound. Widerker also points out that the defender of DA cannot evade this objection by arguing that he is more sure of the soundness of DA than he is of an assumption that the objection presupposes, that is, (M) Someone who knowingly acted in a morally wrong way, and believed correctly that he could have avoided acting as he did, is blameworthy for his act.
For this assumption is beyond reproach, being (almost) self-evident.17 The conclusion Widerker (2002) draws from the above considerations is that one cannot maintain the soundness of DA without assuming the truth of (IVD), and therefore, DA cannot do the work van Inwagen wanted it to do (pp. 322–323). Some have claimed that Widerker’s objection (W1) does not succeed, since assumption (M), upon which it is based, is not beyond reproach, but is in fact false. McKenna (2008b), for instance, offers a counterexample to (M) in which an agent satisfies the conditions that (M) specifies, but is not morally responsible for her action because she is mentally deranged (p. 358). However, his counterexample is based on a misunderstanding of Widerker’s position. (M), as intended by Widerker, was meant to apply to a morally competent, normally functioning agent, whose mental health is not in question. In other words, (M) is to be understood as having an implicit clause to the effect that “all else is normal.”18 Though this critique of Widerker’s (W1) fails, we would now like to argue that (W1) is not nearly as damaging to DA as Widerker originally thought. 17. Note that (M) does not assume that avoidability is a necessary condition for moral responsibility (PAP). What it assumes is only that avoidability in conjunction with certain other assumptions is sufficient for moral responsibility. 18. We also think that the counterexample itself is problematic, as it is far from clear that the agent in the example, given that she is stipulated to be morally competent (despite being deranged), is not blameworthy for her action.
( 98 ) Incompatibilism and Omissions
Widerker wanted to show that the proponent of DA must assume (IVD), contrary to the main purpose of DA. However, all that follows from (W1) is that accepting DA commits one to holding (IVD). That is, it does not follow that DA presupposes (IVD); rather (IVD) is a logical consequence of DA plus (M). For given that the scenario that Widerker employs is a deterministic one, DA entails that in that scenario Jones is not blameworthy for what he did; and, given that he knowingly and deliberately acted in a morally wrong way, and also believed that he could have avoided acting as he did, it follows, via (M), that in that scenario he could not really have avoided acting wrongly. Widerker’s argument (W1) does entail that the proponent of DA must assume something about avoidability—namely (M). But whereas (IVD) is controversial, (M) is quite harmless and would be accepted by both compatibilists and incompatibilists. This point is important since it enables incompatibilists to restate the dialectical role of DA in a way that evades Widerker’s worry. In lieu of viewing it as an argument that does not employ the avoidability notion of free will at all, incompatibilists may view DA as an argument that is free of any controversial assumptions involving avoidability. The case for the incompatibility of determinism and moral responsibility based on DA would remain unimpaired.
2.2.2 Assessment of (W2) We turn now to (W2), Widerker’s second dialectical objection to DA. An essential premise of DA is that no one is responsible for the laws of nature— NR(L). But why, asks Widerker (2002) rhetorically, should one accept that premise, if not for the fact that no one could have prevented them from obtaining (p. 324)? That is, Widerker argues that (W-L) Unless one assumes that no one could have prevented the laws of nature from obtaining, one has no good reason to accept NR(L).
If Widerker is right, then DA depends for its plausibility on an assumption involving unavoidability, contrary to van Inwagen’s motivation for introducing DA. Both McKenna and Haji reject Widerker’s contention (W-L). They propose several ways of justifying acceptance of NR(L) that are not based on the laws’ unavoidability: (1) No one is responsible for the laws of nature because no human being brought them about;
T he Direct A rgument for I ncompatibilism
( 99 )
(2) No one is responsible for the laws of nature because no human being could bring them about; (3) One is not responsible for the laws of nature because one did not bring them about, and they obtain regardless of whether one exists or fails to exist; and, adopting a proposal made by Ted Warfield (1996, p. 218), (4) No human being is responsible for the laws of nature because their truth-makers operated before human beings came into existence. Hence, they claim that one can account for our nonresponsibility for the laws of nature, without assuming their unavoidability (McKenna 2008b, p. 360; Haji 2009, pp. 84–85). However, we do not think that (1)–(4) either singly or jointly account adequately for our nonresponsibility for the laws of nature. Consider the following example, which we may call “The Miracle Worker”: Suppose that the reasons for the nonresponsibility for the laws of nature stated in (1)–(4) all obtain; that is, no human being brought about, or could have brought about, the laws of nature; the laws hold regardless of whether any particular person exists; and the truth makers of the laws of nature obtained before any human being was born. But suppose that nevertheless, someone named Smith somehow has the power to bring it about that a given law of nature L1 does not obtain in a particular instance, and knows that he has it. Suppose further that Smith knows that by exercising this power, he could prevent a tragedy from occurring; but he intentionally refrains from suspending L1, for morally inadequate reasons, and the tragedy occurs.
Note that if (1)–(4) were adequate explanations of our nonresponsibility for the laws of nature, as McKenna and Haji maintain, then Smith would not be responsible for the obtaining of L1 in this scenario, and therefore he would not be responsible for the tragedy. However, it is clear that Smith is responsible (blameworthy) for the tragedy, and part of the reason why Smith is responsible for the tragedy is that he is responsible for L1’s obtaining in the circumstances; for Smith knew that L1’s obtaining would lead to the tragedy, and he knew that he could prevent L1’s obtaining (in the circumstances), and yet he did not prevent it. This example also shows that what renders (1)–(4) inadequate as explanations of our nonresponsibility for the laws of nature is precisely their failure to take into account the unavoidability of the laws.
( 100 ) Incompatibilism and Omissions
Note that one cannot parry this objection by arguing that no human being could possibly have it within his power to prevent a law of nature from obtaining, and that therefore the objection rests on an assumption that is necessarily false. If this is the critic’s only objection to our example, then he is ipso facto conceding Widerker’s claim (W-L) that to be able fully to account for our nonresponsibility for the laws of nature we have to assume their unavoidability. Thus Widerker’s claim (W2) seems to be correct, which means that strictly speaking, van Inwagen has not achieved his goal of constructing an argument for incompatibilism without appealing to assumptions involving avoidability. However, having said that, we must confess that this particular assumption—that is, the unavoidability of LAWS—strikes us as so intuitively plausible that it would be fair to say that DA’s dependence on it would not detract significantly from its philosophical importance.19 As emphasized earlier, DA’s dialectical role can be easily reformulated, that is, DA may be viewed as an argument for the incompatibility of moral responsibility and determinism that is free of any controversial assumptions about avoidability.
3. Alleged Counterexamples to Transfer NR
We turn now to a critical discussion of proposed counterexamples to Transfer NR.
3.1. Ravizza and Fischer’s Counterexample Mark Ravizza and John Fischer have presented the following a scenario, known as “Erosion”: Betty [a double agent who has been instructed to destroy the enemy’s camp] plants her explosives in the crevices of a glacier and detonates the charge at T1, causing an avalanche that crushes the enemy’s fortress at T3. Unbeknownst to Betty . . . the glacier is gradually melting, shifting, and eroding. Had Betty not placed the dynamite in the crevices, some ice and rocks would have broken free at T2, starting a natural avalanche that would have crushed the enemy’s camp at T3. (Ravizza 1994, pp. 72–73; Fischer and Ravizza 1998, p. 157)
19. That said, note that there are compatibilists who do deny this assumption. Here we have in mind Altered-Law Compatibilists such as Lewis (1981).
T he Direct A rgument for I ncompatibilism
( 101 )
They extract from the above scenario the following argument: NR (the glacier was eroding) NR (the glacier was eroding ⊃ the enemy’s camp was destroyed at T3) /∴ NR (the enemy’s camp was destroyed at T3) According to the scenario, the premises are true, but the conclusion is false. Since the argument has the form of Transfer NR, we see that Transfer NR is not valid.
3.1.1 McKenna’s Response Michael McKenna noted that this example is based on there being two causal paths to the same result—that is, either the glacier’s eroding or Betty’s explosives would result in the destruction of the enemy camp. So McKenna (2008b) suggested a version of Transfer NR that rules out such “two-track” cases by stipulating that there be just one causal path between p and q. His new principle is: Transfer NR∗∗∗: If a person is not responsible for a fact, and this fact is at some particular time the sole causal antecedent of the path to the bringing about of a further fact, and if the person is not responsible for any part of the path between the two facts, then he or she is not responsible for the further fact either. (p. 368)20
Fischer has responded that this “single-track” version of the transfer rule is question-begging. We think that what Fischer means is that McKenna’s version essentially involves reference to causality in a way that makes it merely a restatement of the standard incompatibilist intuition, which compatibilists reject, namely that no one is responsible for what is causally determined by factors for which no one is responsible. As we pointed out earlier, Transfer NR, unlike the standard incompatibilist intuition, is supposed to be acceptable to an open-minded compatibilist. So McKenna, by replacing Transfer NR with Transfer NR***, which is essentially a 20. We skip over an earlier version of McKenna’s (2001) single-track principle (p. 45). Fischer’s response, in the text below, was to the earlier version, but it applies equally to this one. Note also that although McKenna formulates Transfer NR*** as a conditional, we prefer to treat it as a rule of inference, since it is supposed to be a substitute for Transfer NR. The formal changes involved do not affect the discussion to follow.
( 102 ) Incompatibilism and Omissions
statement of the standard incompatibilist intuition, has taken a dialectical step backwards.21
3.1.2 Widerker’s Response David Widerker (2002, pp. 318–319) has argued that Fischer and Ravizza’s example can be sidestepped by drawing attention to the fact that its conclusion involves derivative responsibility. An agent S is derivatively responsible for p just in case S is responsible for p in virtue of being responsible for some other fact. S is directly responsible for p just in case S is responsible for p, but not in virtue of being responsible for some other fact. In Erosion, Betty is derivatively responsible for the destruction of the enemy’s camp by virtue of being responsible for detonating the charge that caused the destruction. The notion of derivative responsibility is parasitic upon the notion of direct or nonderivative responsibility, and is therefore only of secondary importance in the debate between compatibilists and incompatibilists.22 In other words, necessarily, if no one is directly responsible for anything, then no one is derivatively responsible for anything. Thus, if one were able to show that determinism is incompatible with direct responsibility, it would immediately follow that it is incompatible with derivative responsibility as well.23
3.2 Haji’s Example A very interesting counterexample to the validity of Transfer NR has been offered by Ishtiyaque Haji (2009, p. 97). He calls it “Hal’s Creation”: Hal is an amoral creator who creates a world W1, which is indeterministic. In W1, Michael, an agent with libertarian free will, kindly (and freely) decides at time T to help his friend Bill.
Given this scenario, Haji claims that the following Transfer NR argument has true premises and a false conclusion: 21. McKenna attempts to avoid the charge of begging the question by construing the causality involved in Transfer NR*** as probabilistic. However, this attempt is unsuccessful; for interpreting Transfer NR*** this way renders that principle false. The fact that one’s act is (merely) probabilistically caused by factors for which one is not responsible does not take one off the hook. 22. Similar remarks would apply to Widerker’s Brain Malfunction argument in Widerker (2008, p. 231), the conclusion of which also involves derivative responsibility. 23. Working out this response in greater detail would require, among other things, limiting the conclusion of Transfer NR to absence of direct responsibility rather than nonresponsibility sans phrase. We believe that the relevant details can be worked out.
T he Direct A rgument for I ncompatibilism
( 103 )
1H. NR (Hal creates W1) 2H. NR (Hal creates W1 ⊃ Michael decides at T to help Bill) /∴ 3H. NR (Michael decides at T to help Bill) 1H is true because Hal is the only candidate for moral responsibility here, and he is stipulated to be amoral. 3H is false because Michael is praiseworthy for deciding to help his friend. Haji’s justification of 2H is more subtle and complex. First, he puts forward two alternative epistemic constraints on moral responsibility: (E1) To be morally praiseworthy [blameworthy] for bringing about a state of affairs p, the agent must have brought it about in the belief that p is a morally good [morally bad] state of affairs. (E2) To be praiseworthy [blameworthy] for some state of affairs p, the agent ought to have known or believed that p is a morally good [morally bad] state of affairs. (Haji 2009, p. 99)24
Haji argues that neither Hal nor Michael can be said to satisfy either of these constraints with respect to the conditional state of affairs MC: (MC) Hal creates W1 ⊃ Michael decides to help Bill at T. This is obviously true for Hal, since Hal is amoral; but Haji thinks that it is also true for Michael, since there is no reason to think that Michael brought about MC in the belief that MC is a good state of affairs, nor is it true that he should have believed that MC is a morally good state of affairs. Finally, Haji points out (or stipulates) that if Hal and Michael are not responsible for MC, then no one is. Hence, he claims that 2H is true, and so Hal’s Creation is a counterexample to Transfer NR.25
3.2.1 Our Response Unlike Haji, we think that the argument based on Hal’s Creation is valid. We agree with Haji that the conclusion of the argument (3H) is false, that 24. Haji’s own formulation of these constraints differs from that in the text, but this difference can be safely ignored. 25. Haji (2009) also offers another counterexample to Transfer NR for the case that (W1) is a deterministic world, meant to appeal specifically to compatibilists (p. 98). However, Haji’s argument strikes us as overly dogmatic, as it simply has the compatibilist reject the Transfer NR-rule without addressing van Inwagen’s argument in support of its validity.
( 104 ) Incompatibilism and Omissions
is, Michael is responsible (praiseworthy) for deciding to help Bill. But we would argue that its second premise 2H. NR (Hal creates W1 ⊃ Michael decides at T to help Bill) is false as well. To see this, note that MC (the state of affairs which 2H says that no one is responsible for) is the same state of affairs as the disjunction (MC’) ~(Hal creates W1) ∨ (Michael decides to help Bill at T ), which, given the falsity of its first disjunct, obtains just in virtue of Michael’s deciding to help Bill. But then, since Michael is praiseworthy for deciding to help Bill, he can be said to be derivatively responsible for MC’ (and hence for MC), in virtue of being responsible for his decision to help Bill.26 Hence, the second premise of Haji’s argument is false. As for Haji’s epistemic constraints on responsibility to which he appeals in order to defend 2H, we would argue that these do not always apply to derivative responsibility, the kind of responsibility involved in (2H). For example, Jones might be praiseworthy for defusing a bomb planted in a mall, and hence be (derivatively) praiseworthy for the state of affairs consisting in little Jimmy’s life being saved. Yet, since he had no clue as to Jimmy’s existence or identity, he did not bring about that state of affairs in the belief that Jimmy’s life being saved is a good state of affairs, nor ought he to have believed that Jimmy’s life being saved is a good state of affairs (since, again, he knew nothing of Jimmy’s existence).
3.3 Shabo’s Counterexample Seth Shabo (2010) has also put forward a counterexample to the validity of Transfer NR. He bases it on a scenario which he calls “Bad Angle”:
26. Note that our claim that Michael is derivatively responsible for MC’, does not commit us to the acceptance of the principle that for any propositions p and q: If one is morally responsible for q, and p is false, then one is derivatively responsible for its being the case that ~p or q. This principle is obviously false. It would entail that one can be responsible for necessary truths such as (p or ~p). We do, however, subscribe to a weaker principle which is like the one above but constrained by the condition that (~p or q) is not an instance of a logical truth, nor a nomologically necessary truth, nor a truth that is nomologically necessitated by the circumstances.
T he Direct A rgument for I ncompatibilism
( 105 )
Jed murders Kenny by dislodging a boulder so that it falls onto him. A ranger could have prevented the murder by warning Kenny; but since he was viewing things from a bad angle, he misperceived the situation so that he mistakenly believed that Kenny was not in the boulder’s path, and so he did not warn him. (pp. 243–244)
The counterexample that Shabo extracts from this scenario is rather complex. Here is a simplified version which, we believe, does not detract from its main thrust. Let “M” stand for “The ranger misperceives the situation so that he mistakenly believes that the boulder will not hit Kenny,” and let “K” stand for “The boulder hits Kenny.” Consider now the following instance of Transfer NR: NR (M), NR (M ⊃ K)/∴ NR (K) This argument has true premises and a false conclusion, which implies that Transfer NR is invalid. The first premise is true, since no one is responsible for the ranger’s misperception. The second premise is true, since (M ⊃ K) is a necessary truth, and so no one is responsible for its truth. But the conclusion is false, since Jed is blameworthy for K. However, we think that the first premise, NR(M), is false as well. Note that M is true partially in virtue of the ranger’s mistakenly believing that the boulder will not hit Kenny.27 But mistakenly to believe that the boulder will not hit Kenny means to believe that the boulder will not hit Kenny, when in fact the boulder will hit Kenny. Thus, M is true partially in virtue of its being the case that the boulder will hit Kenny. But since Jed is responsible for the boulder hitting Kenny, it follows that Jed is partially responsible for M. Hence it is false that NR(M), and therefore we do not have here a counterexample to Transfer NR.
3.4 A New Example: Molecules Suppose that Jones fires a bullet at Smith at T1, a fact which, in the circumstances, provides a causally necessary and sufficient condition for the movement of some air molecules at T3 in his vicinity. The following argument shows that Transfer NR is invalid. 27. We may view M as equivalent to the following conjunction: (1) The ranger misperceives the situation, and (2) as a result, the ranger comes to believe that the boulder will not hit Kenny, but (3) the boulder will hit Kenny.
( 106 ) Incompatibilism and Omissions
1. NR (Some air molecules move at T3) (True) 2. NR (Some air molecules move at T3 ⊃ Jones fires a bullet at Smith at T1) (True) / ∴ 3. NR (Jones fires a bullet at Smith at T1) (False)28
Premise 1 is true, since intuitively the fact that some air molecules are moving at T3 is not something for which anyone is morally responsible, in these circumstances. And since we assumed that Jones’s firing a bullet at Smith at T1 is a causally necessary condition of the fact that some air molecules are moving at T3, premise 2 is true as well. However, the conclusion is false, since Jones is responsible for firing a bullet at Smith. This counterexample might be blocked by a further modification of Transfer NR. We might restrict Transfer NR to cases in which it is not the case that q temporally precedes p.29 Some might find this restriction ad hoc. Thus, dealing with Molecules requires a change in Transfer NR, which may detract from the rule’s broad intuitive appeal to some extent. We leave it to the reader to evaluate this worry.30
4. Conclusion
The picture that emerges on the basis of our assessment of DA is the following. Of the dialectical objections to DA, we have seen that McKenna’s is unsuccessful, whereas Widerker’s objections, though successful, nevertheless are not as damaging to DA as they initially seemed to be. As for the alleged counterexamples to Transfer NR, they are either unsuccessful or they can be adequately dealt with by relatively minor alterations in Transfer NR. Thus, DA seems to emerge from its battles somewhat weakened, but still a powerful argument that advances the compatibilist/incompatibilist debate by shifting the burden of proof onto the compatibilist.
28. This example is a variation on an example by Widerker (2002, p. 319). 29. For a similar restriction, see Ginet (2003, p. 607). 30. For another attempt to refute Transfer NR, which is based on a Frankfurt-type counterexample to PAP, see Campbell (2006, pp. 43–44). However, since Frankfurtcases are controversial, a full discussion of the issues involved is beyond the scope of this paper.
•
Chapter 8
Freedom, Responsibility, and Omitting to Act R andolph Clarke
L
et’s take it as given that we generally act freely and we’re often morally responsible for what we do when we so act. Is there such a thing as freely omitting to act or freely refraining from action, and can we be similarly responsible for omitting and refraining? One might think it obvious that the answers to these questions are affirmative; but it’s far from obvious how to understand our freedom and responsibility in such cases. In this chapter I’ll first examine what freely omitting or refraining might come to. Then I’ll take up the suggestion that sometimes one’s freedom in performing an action consists, in part, in being able not to act, rather than being able to act otherwise. I’ll then turn to the issue of responsibility for omitting or refraining. A question to be addressed is whether responsibility for such a thing can be basic, or whether, instead, it must derive from one’s responsibility for something else, such as a prior action. Drawing on the earlier discussion of freedom, I’ll argue in favor of the first alternative, and I’ll show that this position can be supported even given a view of responsiI want to express my appreciation to Bob Kane for the many significant contributions he’s made to philosophical discussion of free will, and for the support and encouragement he’s given me. For comments on earlier versions of this chapter, thanks to Keren Gorodeisky, Emese Mogyoródi, Dana Nelkin, David Palmer, Ian Proops, Carolina Sartorio, and Susan Wolf, and to audiences at North Carolina State University, Seoul National University, the University of Delaware, the 2012 Rocky Mountain Ethics Congress, and the January 2013 Big Questions in Free Will Conference. Work on the chapter was supported by a grant from the John Templeton Foundation. The opinions expressed here are my own and do not necessarily reflect the views of the John Templeton Foundation.
( 108 ) Incompatibilism and Omissions
bility that emphasizes control and the will. Finally, I’ll offer a sketch of what’s required for responsibility for omitting to act and discuss what’s needed to fill in the sketch.
1. Omitting Freely
There are distinctly different kinds of cases in which someone omits to do or refrains from doing a certain thing. And what an agent’s freedom might come to varies significantly from one kind of case to another. On some occasions, an agent’s not performing a certain action (e.g., not moving at all) might simply be that agent’s performing some other, incompatible action (e.g., holding very still); it might be an action designated negatively, in terms of something it isn’t. (We sometimes refer to objects in this way—“the largest non-planet directly orbiting the sun”—and we can refer in this way to actions as well.) When this is so, there’s no difficulty in seeing how one can freely omit or refrain. For example, when a child holds still while playing hide-and-seek, her not moving might be (identical with) her intentionally holding her body still (Clarke 2012).1 We’d then have one event—the child’s act—that can be designated with these two different expressions. In such a case, the child’s freely refraining from moving is just her freely holding still. Similar cases include some instances of speaking out (not keeping quiet), remaining standing (not sitting), and leaving (not staying). In another kind of case, an agent’s refraining from doing a certain thing consists in the agent’s performing some action by means of which he intentionally brings it about that he doesn’t do the thing in question. A policeman might keep his arm at his side—tensely holding it still—in order to prevent himself from shooting a fleeing suspect (Brand 1971, pp. 45–46). The policeman’s freely refraining from shooting would consist, in part, in his freely keeping his arm at his side. When one decides not to do a certain thing and intentionally omits to do it, one’s freedom in not doing that thing might in part stem from one’s freely making the decision. If I intentionally omit to save a drowning child, and I do so upon deciding not to help, my freedom in omitting to help might in part stem from my freely so deciding. 1. The example comes from Mele (1997, p. 232), who uses it for a different purpose. Lest one think that holding very still can’t be an intentional action, imagine the child crouching behind a sofa for several minutes to keep from being seen. Remaining in that position requires maintaining balance and making repeated adjustments of muscle tension in response to feedback, all resulting from the child’s intending, over the period in question, to remain very still.
F reedom , R esponsibility, and O mitting to A ct
( 109 )
But there are cases of intentionally not doing certain things in which the agents haven’t decided not to do the things in question. Granted, intentionally not doing something requires having an intention with relevant content, such as an intention not to do that thing. And we often come to have intentions by making decisions. But that’s not the only way to come to have an intention. We make decisions when we need to settle the question what to do; but sometimes one encounters no such practical uncertainty. When it’s perfectly clear to oneself what to do, one can come to have an intention in much the same way that one comes by inference to have a belief.2 When, upon seeing a sign saying, “Fresh Paint/Do Not Touch,” I refrain from touching a freshly painted object, I might have made no decision not to touch it. It might be that once I saw the sign, there was no question in my mind about whether to do so, no uncertainty that needed to be settled by making a decision. I simply came to intend not to touch it, and I refrained from touching it.3 Can it be that I freely refrained? In cases of intentionally not doing certain things, an intention with relevant content causes some of the individual’s subsequent thought and behavior (Clarke 2010b). It needn’t be the case that the intention directly causes any action at all; still, it plays a crucial causal role. Indeed, the intentionality of one’s not doing a certain thing can be undermined if the causation by the intention is abnormal in a certain way, such that the agent’s control is lost.4 One aspect of freedom in intentionally not doing certain things consists in the normal causal role played by intentions in such cases. (This is so in cases in which one has decided not to do the thing in question, as well as in cases in which one has nonactively come to have the relevant intention.) Suppose that my coming to intend not to touch the object came about in a normal way, free from various kinds of unfreedom. I read the sign and acquired an intention not to touch the object. My coming to so intend didn’t result from delusion, compulsion, brain manipulation by neuroscientists, or anything of this sort. We may count my freedom from such things as another aspect of my freedom in refraining from touching the object. Suppose, further, that when I refrained from touching the freshly painted object, I was able to touch it, and to do so intentionally. And suppose that it was up to me whether I refrained from touching it or, instead, touched it. These facts constitute further respects in which an agent might 2. For support of these claims, see Audi (1993, p. 64) and Mele (1992, p. 231). 3. I take it that, at least typically, when one refrains from doing a certain thing one intentionally doesn’t do that thing. The case at hand would be one of my intentionally not touching the object. 4. Imagine that one’s coming to intend not to get out of bed triggers—immediately and unexpectedly—a prolonged narcoleptic episode. Though one doesn’t get up, one doesn’t intentionally not get up.
( 110 ) Incompatibilism and Omissions
be free in intentionally not doing a certain thing, even when the agent’s not doing that thing isn’t an action and doesn’t result from a decision not to do the thing in question. There’s nevertheless a difference between the kind of control exercised in such a case and the kind exercised in performing an intentional action. The difference is simply that, in the former case, there need be no action that is performed. In intentionally not doing a certain thing, one need not have executed any intention to act.5 (One will almost always, while awake, be intentionally performing one action or another. The point is that no such intentional action is required in the kind of case we’re considering.) What of cases in which an omission isn’t intentional? Suppose that you intended to make up your mind well before the deadline whether to attend a certain conference. When the deadline rolls around, you realize that you haven’t made a decision. What might your freedom in omitting to decide come to? There will no doubt be numerous actions that you have performed during the relevant time. But none of these actions is your not deciding, and your freedom in doing these things isn’t your freedom in omitting to decide. Your omission is likely free from various kinds of unfreedom. We may suppose that you weren’t subject to insanity, phobia, hypnosis, or any such thing. This is of course an important fact about your omission. It might also be the case that you had a capacity to make such a decision well before the deadline, and during the relevant period you had a power to exercise that capacity. You were able to do so. This, too, seems an important respect in which you were free in omitting to decide. Still, the freedom of an intentional action isn’t merely the agent’s freedom from various kinds of unfreedom; and it isn’t merely a matter of one’s being able to do otherwise. It is the agent’s exercise of a certain kind of agential control in the performance of that action. (One hasn’t freely performed an intentional action unless one has, in the first place, performed an action.) In many cases of omission, there doesn’t seem to be any comparable thing. I’ve distinguished several kinds of case in which someone omits to do or refrains from doing a certain thing, and I’ve identified, in each kind of case, various aspects of freedom that one might have in omitting or refraining. Is there more? Robert Kane (1996) suggests that an agent can voluntarily omit to do a certain thing, and that one can be “ultimately responsible” for a voluntary omission (pp. 35, 156). An action or an omission is voluntary, on Kane’s 5. One will have had some intention, and that intention will have caused something. But the intention need not have been an intention to perform an action of any kind, and what’s required with respect to intention’s causal role is much less in a case of refraining than it is in intentional action. For more on this point, see Clarke (2010b).
F reedom , R esponsibility, and O mitting to A ct
( 111 )
view, just in case the agent does what she wills to do, for the reasons she wills to do it, and free from compulsion and coercion. It isn’t entirely clear what it is, on Kane’s view, for an action or an omission to be willed. Sometimes he characterizes it as the agent’s wanting to do that thing more than she wants to do otherwise (e.g., Kane 1996, p. 155); other times he says that it’s having reasons or motives for doing that thing that one wants to act on more than one wants to act on any others for doing otherwise (e.g., p. 30). On the first reading, the want in question is simply about one’s doing a certain thing; on the second reading, it is in part about one’s reasons or motives. I’ll turn later to the question of whether omissions can serve as grounds of moral responsibility. Here I want to consider whether Kane’s notion of voluntariness introduces some further respect in which an omission can be free. One factor that Kane brings up that I haven’t mentioned is that of omitting for reasons. Elsewhere (Clarke 2010b) I’ve offered a view concerning what omitting for a reason can be in cases of intentional omission. In brief, reason-states such as belief and desire cause the intentions that figure in such cases. However, in many cases of omissions that aren’t intentional, one doesn’t omit for a reason, in the normative sense of that word. When you omit to make up your mind well before the deadline, there need be no justifying or normative reason for which you so omit. If failures to act can be caused, there will likely be causes of this one; but there needn’t be any rationalizing causes of it. Hence, in a case of this sort, there’s no aspect of one’s freedom that consists in one’s omitting for reasons. What of the wanting more that is involved in Kane’s notion of voluntariness? He apparently has in mind motivational strength, rather than a judgment that doing a certain thing would be better than doing otherwise. But now, suppose that the policeman was strongly tempted to shoot, but he judged it better to refrain. He succeeded in doing what he judged better. If he continued to want more strongly to shoot than he wanted to do otherwise— or even to want more strongly to act on his reasons or motives for shooting than he wanted to respond to his reasons or motives for doing otherwise— this fact needn’t undermine his freedom in refraining from shooting. He might have freely overcome temptation. His refraining from shooting might have been voluntary in an ordinary sense: he omitted by design or intention, unconstrained by interference.6 Wanting more not to A, understood as Kane explains it, isn’t required for freely refraining from A-ing.7 6. “Voluntary.” Webster’s Ninth New Collegiate Dictionary. Springfield, MA: MerriamWebster, 1987. 7. Throughout the chapter I’ll use “A,” “A-s,” and “A-ing” as stand-ins for expressions indicating types of action.
( 112 ) Incompatibilism and Omissions 2. Ability Not to Act
Let’s say that one acts freely just in case one exercises free will in acting. And we might say that one exercises free will in A-ing on some occasion only if one is able to do otherwise then.8 What suffices for doing otherwise? Performing a significantly different action that is incompatible with one’s A-ing would suffice. I might do otherwise than choose to tell the truth if instead I choose to lie. But some writers appear to hold that it can suffice if one neither A-s nor performs such an incompatible action. Hugh McCann (1998a) says: If my action is autonomous . . . I must have had some alternative, if only the alternative of forbearance or inaction. Applied to deciding, this means that at the moment I make a decision, it must be possible for me to decide differently—or at least to commence to do so—or to forbear deciding, so that my inactivity, too, would count as a demonstration of my will, of my capacity to decide what I choose when I choose. (p. 174)
Similarly, Kane (1996) states: Failing to choose morally or prudentially may mean making an alternative choice. . . . But it may also mean simply not choosing or postponing choice . . . In this [latter] case, doing “otherwise” (than choosing morally) for that time period would be an omission or failure to choose rather than a choosing otherwise. (p. 156)
Sartre (1943, trans. 1966) was wrong, then; it isn’t always the case that “not to choose is, in fact, to choose not to choose” (p. 619). But how can inactivity “count as a demonstration of my will”? And what could an ability not to act be? I might refrain from touching the freshly painted object without having decided not to touch it, and there might be no action—indeed, nothing at all—that is my not touching it. Still, it seems that in this case I manifest an ability to intentionally not touch the object. We have here a manifested ability not to perform an action of a certain kind, and the manifestation isn’t the performance of any action of an incompatible kind. (Though I’m 8. Note that “freely” and “free will” are sometimes used to express a requirement for moral responsibility. It’s an open question whether that requirement includes an ability to do otherwise. Hence, when “free will” is used as I suggest in the text here, it’s an open question whether free will is a requirement for moral responsibility.
F reedom , R esponsibility, and O mitting to A ct
( 113 )
not entirely inactive on this occasion—I’m standing looking at the object— I’m inactive with respect to touching it or not.) The ability that’s manifested is thus not an ability to perform an action of some kind. It is an ability not to act in a certain way. Can we make sense of an unmanifested ability not to act, in a case of the sort that McCann and Kane have in mind—a case of decision making? I think we can. Our deliberations manifest a variety of related powers, some active and some passive. We can turn our attention to some practical question, try to think of alternatives or relevant considerations, attempt to remember various things, pursue some line of thought or switch to another, assess and weigh considerations against each other, make judgments about which reasons or alternatives are better or worse, and make decisions. Important here are certain receptive capacities: to have new considerations come to mind, to recollect, and to see implications of our thoughts. One’s success in deliberating and choosing well can turn on one’s manifesting at appropriate moments such passive powers. One can be inactive—albeit undergoing psychological change—in manifesting such powers. At moments during deliberation when one is actively doing something—perhaps at the moment when one makes a decision— one might have a power to instead have some new consideration come to mind. One would have been inactive in manifesting that power. Suppose that Carla is deliberating about whether to assist a friend in need. She has considered various factors and weighed their significance, but she’s unable to judge which alternative is better. Nevertheless, she intends to make up her mind, and at a certain moment, t, she decides to help. Suppose that there might have instead occurred to her at t some new relevant consideration, and that its occurring to her would have manifested one of her passive deliberative powers. Despite intending to make up her mind, she would not have done so at t, but the failure wouldn’t have been due to any neural glitch; it would have expressed a normal deliberative capacity. Imagine that Carla’s not making the decision then would have been free from various sorts of unfreedom: it wouldn’t have resulted from phobia, compulsion, direct brain manipulation by neuroscientists, and so on. Perhaps we have here something that merits being called an ability not to act at a certain moment when one does act. It isn’t the mere possibility of not acting but rather a passive psychological power that is a normal constituent of deliberative capacities. (I leave aside the question of whether a manifestation of such a power is properly called a demonstration of one’s will.) Carla was able to do otherwise than decide at t to help: she was able to be inactive with respect to making up her mind then. Nevertheless, if Carla
( 114 ) Incompatibilism and Omissions
wasn’t able to decide at t not to help, or to decide then not to settle the matter, her freedom in making her actual decision was narrowly circumscribed. An ability not to decide might be important, but so is an ability to decide otherwise. Consider another case. Sue decided earlier today to attend a colloquium this afternoon. In the time since she made the decision, she’s had a vague impression that she might have forgotten a conflicting commitment. She’s made a couple of attempts to think of what she might have forgotten, but she hasn’t thought of anything, though she’s been too busy to give it a great deal of thought. Sue still intends to go to the colloquium, and now, when she notices that it’s time to go, she gets up to go. Imagine that what might have occurred instead, when Sue noticed the time, is that she recalled the forgotten commitment. If she had, let’s suppose, the recollection would have manifested a normal psychological capacity, not some psychosis or neural glitch. And suppose that if the commitment had come to mind, Sue would have reconsidered whether to attend the colloquium. She would have been able to go, and able instead to change her mind and keep her prior commitment.9 A kind of ability not to act at some moment might be, as in this case, an important aspect of one’s freely performing an action. The manifestation of that ability might be necessary for one’s acting otherwise shortly afterward. (Sue’s recalling the forgotten commitment might have been necessary, not for her having been able to perform an action incompatible with getting up to go, but for her exercising that ability.) A passive power to do something other than act at a certain moment might thus be a valuable partner of an ability to act otherwise a moment later.
3. Basic Responsibility
In section 1 I examined what an agent’s freedom in omitting or refraining might come to. I move now to the topic of moral responsibility. I’ll begin with a familiar distinction. Sometimes someone is morally responsible for something because that thing results from something else for which that individual is responsible. A parent might be blameworthy for the death of a child because the child’s death resulted from abuse by the parent and the parent is blameworthy for abusing the child. Responsibility for such things, I’ll say, is derived or nonbasic, and the parent is indirectly responsible for the death. Our responsibility 9. I discuss this example in Clarke (2000).
F reedom , R esponsibility, and O mitting to A ct
( 115 )
for outcomes of our actions, when those outcomes are not themselves our actions, is commonly derived. Some of one’s actions are consequences of one’s earlier actions. And responsibility for such a resulting action can be derived. A heroin addict might be responsible for injecting herself with the drug on some occasion even if she can’t reasonably be expected to resist her urge to do so. Her responsibility would at least in part stem from her responsibility for prior actions of hers that resulted in her addiction and thus in her behavior on this occasion. But not all of our moral responsibility is derived, or not solely so. Sometimes one is responsible for something and one’s responsibility for that thing doesn’t entirely stem from one’s responsibility for something else. Responsibility for that thing is then, at least in part, basic. I’ll say that in such a case the agent is, at least in part, directly responsible for that thing. Responsibility for an action can be in part derived and in part basic. One might be responsible for that action partly because it results—by way of some of one’s traits, habits, beliefs, or desires that also result—from one’s prior actions, and partly independently of the fact that it so results, in virtue of one’s freely performing that action. We can be indirectly responsible for various kinds of things: actions; omissions; habits and character traits; psychological states such as belief, desire, and intention; episodes of emotion such as anger or hatred; and outcomes not involving ourselves, such as the suffering or death of another person. For what kinds of thing can we be, at least in part, directly responsible? It’s commonly recognized that free actions, or some kinds of them, are included. And it’s sometimes said that it’s only for free actions, or only for some kinds of them, that agents can be directly responsible. Kane appears to hold a view of this kind. As do many theorists, he takes responsibility to require freedom. And with respect to the latter, he endorses what he calls Bramhall’s Thesis: “The freedom of the agent is from the freedom of the will” (Kane 1996, p. 124).10 We’re free in a nonderivative way, he holds, only in the case of “self-forming willings,” which “include such things as choices, decisions, judgments, formations of intention, and efforts or trying, all of which are mental ‘acts’ or actions of one sort or another” (p. 125).11 As these acts of will are the source of our freedom, so they ground our moral responsibility. 10. For Bramhall’s statement of this view, see his response to Hobbes in Chappell (1999, p. 44). 11. Kane (1996) subsequently adds to this list certain “voluntary omissions” (p. 156), though, as I’ve noted, in many cases an omission isn’t an action of any kind at all.
( 116 ) Incompatibilism and Omissions
At least part of what motivates such views, I suspect, is adherence to a Kantian principle according to which moral worth depends solely on the good will. Thomas Nagel (1979) finds a reflection of this principle in our ordinary moral thinking, in the idea that “people cannot be morally assessed for . . . what is due to factors beyond their control” (p. 25). We might be thought to directly control, in the required manner, only our free acts of will. Whether such acts result in anything else depends, at least to some extent, on luck. On such a view, our basic, underived responsibility is only for these willings. Our responsibility for anything else traces back to our responsibility for these acts of will.
4. Responsibility for Omissions
What should we say, then, about moral responsibility for omissions? If the child playing hide-and-seek is responsible, by way of responsibility for some act of will, for holding very still, then she might in the same way be responsible for not moving. If there is some act of will involved when the policeman keeps his arm at his side, his responsibility for refraining from shooting might derive from his responsibility for this act of will. When one decides not to do a certain thing and then intentionally omits to do it, one’s responsibility for the omission might, at least in part, stem from one’s responsibility for freely deciding not to do that thing.12 We saw that when I refrain from touching the newly painted object, I need not have decided not to touch it, and there need be no action at all that is my not touching it. Nevertheless I might freely refrain from touching it. And it certainly seems that I can be responsible—perhaps praiseworthy—for not touching the object and messing up the paint job. In this case, it seems, we might have something—an intentional omission or a refraining—that isn’t an action of any kind and for which one can be directly responsible. A restriction of basic responsibility to acts of will—or even to actions of some broader kind—is thus mistaken. What about cases in which an omission isn’t intentional? Suppose that I planned to stop and buy milk on the way home, but I forgot to do so. I didn’t intentionally omit to stop and buy milk. Perhaps earlier I had decided to make the stop, but I’m blameless for that act of decision making, 12. If basic responsibility is limited to acts of will, then in none of these cases will responsibility for an omission or for refraining count as basic. But then, on this view of basic responsibility, the same goes for most actions—raising one’s arm, walking to the library, driving a car, etc. On a more liberal view that allows for basic responsibility for more kinds of action, the child’s responsibility for not moving might count as basic.
F reedom , R esponsibility, and O mitting to A ct
( 117 )
and surely any blame I’m due for failing to get the milk doesn’t derive from my responsibility for that earlier decision. While driving home, I started to think about omissions. If I hadn’t, my earlier decision to stop and get milk would have caused my doing so. Perhaps my responsibility for failing to get the milk derives from responsibility for this mental act. The trouble with this suggestion is that, we may imagine, I didn’t foresee that the failure would result from my thinking about omissions. And the case is easily imagined such that it isn’t plausible that I should have foreseen this. I routinely think about my work while driving home, typically without untoward consequences. It’s not that I should have refrained from thinking about omissions; what I should have done is get the milk. Similarly, we may suppose that my responsibility for omitting to get the milk can’t be traced back to my responsibility for earlier actions that resulted in my having a poor memory. I don’t have a poor memory, just an average one. And we may imagine that in performing earlier actions I didn’t foresee, and it can’t plausibly be said that when I performed them I should have foreseen, that they would result in my becoming the sort of person who would forget what I planned to do on occasions like this. Perhaps I should have made myself a note or created a reminder on my phone. But my omitting to do these things, too, we may imagine, wasn’t intentional. And if responsibility for such an omission can’t be basic, we face the same difficulty in finding something my responsibility for which grounds my responsibility for these omissions. The crux of the matter is that some epistemic (or better, cognitive) condition must be met when responsibility for something derives from responsibility for prior actions. One must have foreseen, or it must be that one should have foreseen, that the outcome would or was likely to ensue. In many cases of omissions that aren’t intentional, it seems, this cognitive requirement isn’t satisfied. It thus appears that if responsibility for such omissions can’t be basic, then there’s considerably less responsibility for omissions than we generally think. We might take it as a moral datum—one that a theory of responsibility should fit, and perhaps ought to explain—that we’re commonly responsible for omissions, and often for omissions that aren’t intentional. A view on which this can’t be so convicts very many of our everyday moral judgments of falsehood. A view that saves more of the phenomena is in an important respect preferable. But a more conservative view might be thought to conflict with the Kantian principle mentioned earlier. It might appear that where there’s no willing—and, indeed, no intention—there’s no bad will. Is there no way of
( 118 ) Incompatibilism and Omissions
accepting our everyday judgments without rejecting what seems right about the Kantian principle?
5. Moral Assessability
Several writers on moral responsibility distinguish a kind or aspect of it that they call attributability. An individual bears such responsibility for all that can appropriately be taken as a basis of moral assessment of her. The things in question are those that reflect one’s evaluative judgments or disclose one’s practical commitments. Our answerability for such things is a matter of its making sense that we be asked to defend them, to provide our justifications of them.13 Writers taking this view sometimes argue that there’s no special role for action in grounding moral responsibility. One can be appropriately morally assessed for feeling envious, for holding an uncharitable belief, or for forgetting a friend’s birthday.14 Such things reflect one’s evaluative judgments or practical commitments, and one can be asked to justify them, and to offer excuse or apology if one can’t. There’s no problem fitting basic responsibility for omitting to act—even when the omission isn’t intentional—into such a view. For a failure to act, too, can reflect a lack of concern. One can be asked to justify such a thing, and one can be expected to apologize if one lacks justification or excuse. However, it seems to me that there’s a kind of moral assessability that’s more closely tied than is attributability (of the sort just described) to certain other aspects of our practice of finding and holding people responsible. And it appears that omitting to act, along with action, might play a special role with regard to assessability of this second kind, in a way that respects the Kantian principle. Sometimes when someone is negatively morally assessable for something, it’s appropriate to have toward that person one or another of a certain class of reactive attitudes that are closely tied to finding someone responsible, attitudes including resentment, indignation, and (when the offender is oneself) guilt. You might appropriately judge a colleague blameworthy and feel resentment when you discover that she has deceived you about an important matter. In other cases of appropriate negative assessment, such attitudes are out of place. Imagine “a squash player who, while suffering an ignominious 13. See, for example, Scanlon (1998, ch. 6), Smith (2005), and Watson (1996). 14. Smith (2005) emphasizes this point.
F reedom , R esponsibility, and O mitting to A ct
( 119 )
defeat, desires to smash his opponent in the face with the racquet” (Watson 1975, p. 210).15 Even if the angry player gets a grip on himself and dismisses his desire without acting on it, some judgment of moral failing, and some not-entirely-cognitive attitude toward him, will be fitting. But something like distaste or distrust—or perhaps amusement—is what would fit. Indignation toward him is out of place. Resentment, indignation, and guilt reflect the sort of demand that can appropriately be made with regard to action, one that can be grounded in moral obligation. We’re morally obligated to act or not act in certain ways, and because we are, we can appropriately demand of one another (and ourselves) that we act in ways that manifest goodwill, or at least don’t manifest ill will, and that we not act in ways that manifest the latter. The attitudes in question are often apt responses to failures to meet such a demand.16 In contrast, although coming to desire to do something that one is obligated not to do can manifest a moral flaw, one doesn’t violate a moral obligation simply in coming to have the desire. Cases of derived responsibility aside, attitudes that can appropriately accompany assessments for having such nonvoluntary attitudes don’t reflect the kind of demand that’s reflected in resentment, indignation, and guilt. Similarly, one’s assessability for actions is tied to the justification of sanctions in a way that assessability for having nonvoluntary attitudes is not. One can merit an overt response meant as a sanction when one has freely acted wrongly. Merely desiring to do something that one ought not do can merit some kind of overt response, even one that costs the offender, but not, I think, one that is meant as administering a sanction. Again, the difference seems due to a difference in the applicability of moral obligation. Some sanctioning behavior, it seems, can be merited only by violations of obligation. It thus appears that there’s a kind of moral assessability with respect to which actions play a role that isn’t played by nonvoluntary states of 15. Watson employs the case for a different purpose. 16. “The personal reactive attitudes [such as resentment] rest on, and reflect, an expectation of, and demand for, the manifestation of a certain degree of goodwill or regard on the part of other human beings towards ourselves; or at least on the expectation of, and demand for, an absence of the manifestation of active ill will or indifferent disregard. . . . The generalized or vicarious analogues of the personal reactive attitudes [such as indignation] rest on, and reflect, exactly the same expectation or demand in a generalized form; they rest on, or reflect, that is, the demand for the manifestation of a reasonable degree of goodwill or regard, on the part of others, not simply towards oneself, but towards all those on whose behalf moral indignation may be felt, i.e. as we now think, towards all men” (Strawson 1962, p. 200).
( 120 ) Incompatibilism and Omissions
mind. This is a kind of assessability that concerns whether one has acted appropriately in response to moral obligations. Might omissions play a similar role with respect to this kind of moral assessability? In many cases of omission, there’s no action that is the omission, and there’s no decision to omit. Still, in some such cases, one complies with an obligation simply by omitting to do a certain thing. Moral obligation is often simply a requirement that one not do anything of some kind; it’s commonly indifferent to what, when not doing that thing, one does do. I might comply with an obligation—an obligation not to ruin the paint job—simply by not touching the freshly painted object. I might merit a favorable moral assessment for doing so. The demand that I not manifest active ill will is satisfied simply by my refraining from touching the object. No action is required. Further, often when in omitting to do a certain thing one fails to fulfill an obligation, one’s failure is one that was avoidable simply by performing the action in question. If I had promised to stop and get milk and was thereby obligated to do so, all I had to do to fulfill the obligation—and to satisfy the demand that I manifest proper regard for the person to whom I’d made the promise—was get the milk. In contrast, any obligation one might have to avoid having bad desires would be an obligation to act so as to prevent one’s having them (or not act so as to bring them about). Obligation applies in a more direct way to omitting to act than it does to non-voluntary states. Earlier I indicated several respects in which omitting to act can be free. When one intentionally omits to do something, even when one hasn’t decided not to do that thing, an aspect of one’s freedom consists in the normal causal role played by one’s intention in producing some of one’s subsequent thought and behavior. Further, one’s coming to have that intention will commonly be free from various kinds of unfreedom. When one refrains after deciding not to do a certain thing, one’s decision will typically have been free. And ordinarily one will have been able to perform the action that in fact one omitted. In cases of omissions that aren’t intentional, commonly one’s thinking or doing what one actually did will be free from various sorts of unfreedom, and one will have been able to do what one omitted. In many of these cases, one doesn’t exercise the agential control that is exercised in action, but one has a kind of control that shares several features with such agential control. Of course, omissions that aren’t intentional are typically unwitting: one isn’t aware that one isn’t doing the thing in question. Indeed, commonly in such cases one wasn’t aware at any earlier time that one wasn’t going to do the thing. The Kantian principle is sometimes understood as requiring
F reedom , R esponsibility, and O mitting to A ct
( 121 )
some such awareness.17 But the quality of one’s will can be reflected not just in what one knowingly wills; it can be reflected in a failure to will. Even unwitting failures to will, then, might play a role in grounding responsibility that isn’t played by nonvoluntary states such as desire. We have a kind of control over the former that we lack over the latter. Note that it would be mistaken to think that, when I omit to get the milk, I’m unable to do so because I’ve forgotten my plan. Recalling my intention might be required for the exercise of my ability, but it isn’t required for being able. Forgetting to do something isn’t disabling. Indeed, it would be fair to say that it remains up to me whether I get the milk.
6. A Proposal
It’s commonly thought that, in addition to having free will, being morally responsible requires having a capacity to recognize and respond to specifically moral reasons. Let’s confine our attention to agents who have this capacity. What’s needed for them to be directly responsible for omitting to do or refraining from doing certain things? Consider the following proposal: such an agent is directly morally responsible for omitting to A on some occasion just in case she freely omits to A then. The proposal is at best the skeleton of a correct view. I’ve suggested several factors that might constitute one’s freedom in omitting or refraining, including: freely deciding not to do the thing in question, or being free from various kinds of unfreedom in nonactively coming to have one’s intention; the normal causal role played by an intention one has; and being able to perform the omitted action. The suggestions are adaptations of off-the-shelf components of familiar accounts of free action. An informative proposal would have to spell out these suggestions more fully than I’ve done here. Moreover, as it stands, the proposal doesn’t say whether all of the suggested factors, or only some, are required for basic responsibility for not doing a certain thing. We might wonder, in particular, whether an ability to do that thing is required. Note that an inability to do a certain thing can preclude one’s not doing that thing on some occasion from counting as an omission or an instance of refraining.18 I haven’t omitted to buy milk at the store if, whether I know it 17. Sher (2009) construes it this way and argues against it. 18. There are different kinds of ability to act, and hence inabilities are of several kinds. Not every kind of inability precludes one’s not doing a certain thing from counting as omitting or refraining. I develop this point in Clarke (2014, ch. 4).
( 122 ) Incompatibilism and Omissions
or not, the store is out of milk. (I might have omitted to stop at the store or to try to buy milk, but that’s another matter.) Nevertheless, even when, due to lack of an ability, an agent’s not doing a certain thing doesn’t count as an omission or an instance of refraining, there can still be a question of whether the agent is responsible for not doing that thing, and indeed she can be. If I’m unable today to go to the store and get milk because my car won’t start, and my car won’t start today because I left its lights on yesterday evening, I might be responsible for not getting milk. In this case, it appears, my responsibility for not getting the milk derives from my responsibility for leaving the car lights on. Hence there’s no reason here to think that ability to do a certain thing isn’t required for basic responsibility for not doing it. But we’d need to consider whether there’s any such reason elsewhere.19 The proposal also says nothing explicitly about cognitive requirements for responsibility. Though I’ve suggested there might be cases of basic responsibility for unwitting omissions, I mean to leave this question open. Hence as it stands the proposal doesn’t say whether awareness that one isn’t doing a certain thing (or earlier awareness that one isn’t going to do it) is part of the required freedom.20 There might be an epistemic or cognitive requirement for responsibility that is additional to the requirement of freedom, as some authors maintain.21 If there is, and if the requirement applies even to basic responsibility, then the proposal may need to be revised accordingly. I won’t attempt to work out exactly what this further requirement, if there is one, comes to. Finally, the proposal is silent on whether the requisite freedom in omitting is compatible with determinism. In cases in which one decides not to do a certain thing, must that decision be undetermined? When an intentional omission doesn’t result from a decision, must one’s coming to have 19. Frankfurt omission cases might be thought to show that an ability to do the omitted thing isn’t required for responsibility. For discussion see Clarke (2011b), Fischer and Ravizza (1998, ch. 5), Frankfurt (1994), McIntyre (1994), and Sartorio (2005). 20. Intentionally omitting to do a certain thing might itself require something with respect to awareness. However, intentionally omitting to A at t doesn’t require that one be aware at t that one isn’t A-ing. I might intentionally omit to attend a conference in May if I decide in January not to go and in May I’m no longer thinking about the matter. Indeed, I might intentionally omit to meet you at the airport at midnight even if I’m asleep at midnight. But generally in either case I’ll have been aware earlier that I’m not going to do the thing in question. If intentionally omitting requires awareness, intentionally omitting freely does. 21. See, for example, Fischer and Ravizza (1998, pp. 12–13), Ginet (2000), and (for the most thorough treatment of this issue) Sher (2009).
F reedom , R esponsibility, and O mitting to A ct
( 123 )
the relevant intention be undetermined? Or in these cases must what one’s intention causes be undetermined? In cases in which an omission isn’t intentional, must what one is actually thinking or doing during some relevant time period be undetermined? I mean to leave these questions open, leaving a proposal that can be fleshed out in accord with compatibilism or, alternatively, one way or another in accord with incompatibilism. Whichever view is correct for moral responsibility generally should be workable here.
Pa r t I V
••• The Significance of Free Will
•
Chapter 9
Responsibility for Emotions, Alternative Possibilities, and Reasons Ishtiyaque Haji
I
n The Significance of Free Will, a book that has left and will leave an enduring and highly impressionable mark in the free will literature, Robert Kane (1996) argues that determinism is incompatible with free action and moral responsibility. He proposes that determinism may well jeopardize many other things we value, such as genuine creativity, autonomy, interpersonal relationships, and objective worth, but this tantalizing suggestion is left fairly undeveloped. In this chapter, I explore a facet of this suggestion, confining my attention to the moral sentiments because many of them are intimately connected with interpersonal relationships of fundamental importance in our lives. I begin with a summary of Angela Smith’s rational relations view of moral responsibility as it purportedly is especially congenial to accommodate responsibility for our attitudes, emotions, or other mental states that seemingly do not involve voluntary choice. Discussion of key elements of this view reveals that some of our attitudes are essentially associated with “objective” (or “external”) reasons.1 I argue, furthermore, that one cannot have such reasons unless one has access to alternatives. Provided these alternatives are of the sort that are incompatible with determinism—“strong Many thanks, indeed, to David Palmer for his insightful comments and suggestions. This essay was completed during my tenure of a Social Sciences and Humanities Research Council of Canada (SSHRC) grant. I am most grateful to this granting agency for its support. 1. Henceforth, “attitudes” will be used as an umbrella term for the items of interest, such as various emotions and the reactive attitudes.
( 128 ) The Significance of Free Will
alternatives”—Kane’s claim that determinism undermines sundry things of value in our lives is vindicated. Should, however, these alternatives be “weak” alternatives that can coexist with determinism, several versions of semicompatibilism—the thesis that determinism is compatible with the truth of various normative assessments even though determinism is incompatible with our being able to do otherwise—are not sustainable.
1. The Rational Relations View
Smith (2005) develops the rational relations view as a competitor to volitional accounts of responsibility according to which “choice or voluntary control is a precondition of legitimate moral assessment” (p. 236). To formulate a generic version of the volitional account, distinguish between directly and indirectly free actions. An indirectly free action derives its freedom from the freedom of other actions to which it is suitably related. A directly free action does not inherit its freedom from the freedom of other actions. The sense of “free” here is the personal sense operative in propositions such as she acted with the freedom required for her to be morally blameworthy for having broken trust. Similarly, one is indirectly responsible for something just in case one is responsible for it by way of being responsible for something else; one is directly responsible for something just in case one is responsible for it but not indirectly so. Responsibility tracks control in that we can be directly responsible for something only if it is in our direct control. The generic volitional view states that one is directly responsible for only one’s directly free (mental or otherwise) actions. This view entails, for instance, that if we are responsible for having a certain emotion, or making a certain evaluative judgment, we are only indirectly so. Addressing the rival of interest, the rational relations view, Smith (2007) proposes that to say that “a person is morally responsible for some thing is to say that it can be attributed to her in the way that is required in order for it to be a basis for moral appraisal,” where nothing is implied about what that appraisal, if any, should be (pp. 467–468). Smith (2008) ventures that being responsible is primarily about actions, choices, or attitudes being properly attributable to, and so reflective of, one’s “rational agency or activity in a way that would make them an appropriate basis for moral” appraisal (p. 381). When negative, the appraisal expresses a demand to its target, calling upon her to be answerable for her rational activity. To be answerable for such activity, at least in principle she must be able to explain or justify this activity in terms of her reasons, and “to acknowledge fault if such a justification cannot be provided” (p. 381). This justificatory demand
R esponsibility for E motions
( 129 )
implies that appraisals of being responsible (or answerable) are “appropriately directed only at features of a person that can be said to reflect her practical agency” (p. 381). These features include any that express “her judgments or evaluative assessments, regardless of whether . . . those features reflect or have resulted from a voluntary choice on her part” (p. 382). According to Smith (2005), the evaluative judgments for which an agent is morally answerable are not necessarily held conscious beliefs, but rather tendencies to regard certain things as having evaluative significance. These judgments, taken together, make up the basic evaluative framework through which we view the world. They comprise the things we care about or regard as important or significant. “Judgments” in this sense do not always arise from conscious choices or decisions, and they need not be consciously recognized by the person who holds them. (pp. 251–252)
Such “judgments” include various values, cares, or emotional commitments. One may discover that one values the intellectual freedom and autonomy associated with a career in academia more than the economic benefits associated with a career in law, or one cares more about being liked by others than standing up for one’s moral principles (Smith 2005, p. 252). An evaluative judgment is sensitive to rational activity—it is “reasonssensitive”—only if it is open to revision or modification through one’s own process of rational reflection where such reflection need involve no choice. Suppose you consider the reasons for (or against) a judgment, such as you care more about being liked by others than standing up for your moral principles, and you realize that you have ample reason to renounce this judgment in favor of another. Assuming that you can give up the judgment on the basis of rational reflection, and you do give it up on this basis, this judgment is reasons-sensitive. Smith (2005) endorses a “constitutive” thesis that various attitudes “such as contempt, jealousy, and regret seem to be partially constituted by certain kinds of evaluative judgments or appraisals” (p. 250). For instance, to feel contempt toward some person “involves the judgment that she has some feature or has behaved in some way which makes her unworthy of one’s respect, and to feel regret involves the judgment that something of value has been lost” (p. 250). Smith proposes that there “seems to be a conceptual connection between having these attitudes and making, or being disposed to make, certain kinds of judgments” (p. 250). The rational relations view denies the generic volitional account’s tenet that the sole objects of direct responsibility are free actions, and holds,
( 130 ) The Significance of Free Will
instead, that if we are directly responsible, we are so only for the reasonssensitive evaluative judgments that various actions, choices, emotions or other mental states express, responsibility for these latter items being indirect (Smith 2008, pp. 383, 388–390). The following passages speak to this significant point: [I]f a paranoid schizophrenic attacks me in a moment of severe mental delusion, I may see that action or his condition as “bad” but I will not consider him blameworthy for it. . . . Why? Because such an act cannot be said to reflect his judgments, and therefore it would make no sense to demand that he justify it. But if I assess an (otherwise normal) adult human being as “cruel” I am not merely attributing a trait to him for which he may or may not be responsible. . . . I am (in part) making a demand of him, a demand that he justify the objectionable judgments his actions and attitudes express concerning the moral status of others. This demand by its very nature implies responsibility, for it is directed at his judgmental activity, activity for which we must regard him as responsible if we are to regard him as a moral agent in any sense. (p. 388) [W]e cannot help but regard ourselves as responsible and answerable for the particular judgments expressed in our actions and attitudes, regardless of what circumstances may have shaped these assessments. A cruel person, whether she became so through her own efforts or through unfortunate circumstances, is someone who judges that the fact that something will cause pain or suffering to another is no reason to avoid it (and is perhaps even a reason to pursue it). It is that judgment, as reflected in her actions and attitudes, for which we consider her answerable and expect her to defend or give up. (pp. 389–390) We need not say that a person is not responsible for her judgments in order to take into account the relevance of her formative circumstances when it comes to our moral assessment of her in light of those judgments. We can acknowledge that a person who has not been exposed to morally appropriate values may have a more difficult time appreciating the importance of these values than someone who has had the benefit of a decent moral upbringing. Such people may be open to less serious moral criticism for their vicious attitudes than others, but they are just as fully responsible for them. (p. 390)
The crucial difference between volitional and nonvolitional accounts centers on how they both answer the following question: “What are the things for which people can be directly responsible?” For volitionists, people can
R esponsibility for E motions
( 131 )
be directly responsible only for actions; for nonvolitionists, people can be directly responsible only for reasons-sensitive evaluative judgments. With this difference in hand, we see how the “nonvolitional” aspect of the nonvolitional view then falls right out of nonvolitionists’ answer to this question: since reasons-sensitive evaluative judgments are not (typically) things that are under our voluntary control, the nonvolitional view is committed to the claim that we can be directly responsible for things that are not under our voluntary control. In sum, the rational relations view streamlines to the following: Attributability: An agent is morally responsible (in the sense of being responsible) for doing some action, or making a choice, or being in a feeling (or emotional) state if and only if the action, choice, or feeling state expresses (or is) a reasons-sensitive evaluative judgment.
Regarding blameworthiness for a choice (action, etc.), Smith (2007) proposes the following: To judge that a person is morally culpable for an action or attitude is to judge that she is responsible for it (it is attributable to her in the relevant sense) and that it is morally wrong or objectionable in some way. . . . We will judge a person morally culpable or at fault if her actions or attitudes violate the moral norms or requirements we take to apply to her. (p. 476)
In other words: Culpability: An agent is morally blameworthy for performing some action, or making a choice, or being in a feeling (or emotional) state if and only if the action, choice, or feeling state expresses (or is) a reasons-sensitive evaluative judgment, and it is wrong (or morally objectionable) for the agent to do this action (or make the choice or be in the feeling state).
Turning to a brief assessment, first, in a noteworthy passage, Smith (2005) proposes that an agent is not blameworthy for an attitude that, unbeknown to her, has been acquired as a result of manipulation or “implantation” (p. 262). Smith explains that what differentiates such attitudes from those one is responsible for is that these implanted “attitudes are not based upon the agent’s own evaluative appraisal of her situation and surroundings but are induced in a way that bypasses her rational capacities altogether” (p. 262). However, critical assessment itself can be tainted in a
( 132 ) The Significance of Free Will
manner that renders it responsibility-subversive. This is so because such assessment cannot be divorced from possible items of induction such as one’s deliberative principles, pertinent values, and germane desires and beliefs. These items are, partly, constitutive of reasoning. So, to simplify, there is nothing such as “input-independent” reasoning. Imagine a case in which germane “inputs” have been surreptitiously induced. On the basis of such contrived inputs, the agent reasons about a particular evaluative judgment that is the basis for an attitude of hers. Assuming all other requirements of responsibility have been met, there is little reason to believe that the agent is directly morally responsible for this judgment and, thus, indirectly responsible for the attitude that expresses this judgment. The germane lesson is that the agent’s reasoning that underlies a pertinent evaluative judgment must not itself be relevantly tainted if she is to be morally responsible for that judgment. Such reasoning won’t be so tainted if its inputs, partly constitutive of this very reasoning, are, as Smith (2005) says, “the agent’s own” (p. 262). Among such inputs are, presumably, desires or other motivationally encompassing attitudes. If these turn out not to be the agent’s own, then, again, reasoning that implicates such desires will be undesirably tainted. But in Smith’s estimation, these desires will qualify as one’s own only if they express reasons-sensitive evaluative judgments that have the sanction of untainted reasoning. But such reasoning, again, will itself have inputs, such as desires, partly constitutive of it. So, it appears that Smith’s account is committed to a troubling regress. Second, Smith is at pains to stress that the nonvolitional account of responsibility she defends is a genuine alternative to volitional accounts. The rational relations account, however, can deliver verdicts fundamentally different from those various volitional accounts deliver even where one would reasonably expect convergence. Suppose a significant set of Tess’ actions nondeviantly and causally issue from reasons-sensitive evaluative judgments they express, in conjunction with other antecedents of action, such as desires. Suppose, furthermore, that unbeknown to her, whenever she performs such an action, the proximal desire (or the relevant motivational encompassing attitude) in its etiology is made irresistible by an evervigilant actual, as opposed to a Frankfurt-style counterfactual, intervener. Assuming other conditions of responsibility are not in question, the rational relations view yields the result that Tess is morally responsible for such an action: it nondeviantly issues from a reasons-sensitive evaluative judgment for which she is supposedly directly responsible. However, because the action’s proximal desire is irresistible, many volitionists would resist the claim that Tess is responsible for such an action. It’s not just that the rational relations view delivers a “responsible” verdict whereas the
R esponsibility for E motions
( 133 )
volitional view delivers a “not responsible” verdict. It is also that intuitively Tess is not responsible for such an action. So, the rational relations view clashes not only with the verdict of the volitional view but also with our intuitions about the Tess case. Third, a vital component of any account of responsibility is an epistemic one: broadly, the component says that one must not be relevantly ignorant of what one is doing or expressing if one is to be morally responsible for this thing. Limiting attention to blameworthiness, an implausibly strong epistemic condition says that one is morally blameworthy for an action (choice, judgment, attitude, etc.) only if one performs (makes, forms, etc.) it on the basis of knowing that it is morally impermissible for one to perform (make, form, etc.) it. A better candidate (Blame-1) is that one is to blame for an action (choice, judgment, etc.) only if one does (makes, forms, etc.) it on the basis of the nonculpable belief that it is impermissible for one to perform (make, form, etc.) it: despite taking it to be impermissible one still does (makes, forms, etc.) it.2 This condition may be supplemented with the further condition (Blame-2) that one is to blame for an action (choice, judgment, etc.) only if it is impermissible for one to perform (make, form, etc.) it. If one is blameworthy for an evaluative judgment, presumably, the epistemic condition for blameworthiness must be satisfied. Now ponder this case. You reflect on reasons for and against cheating on your tax returns, some of which appeal to relevant moral principles. On the basis of your reasoning, you form the true judgment (“IMC”) that it is impermissible for you to cheat. Still, you intentionally cheat. Here, we appear to have a paradigm instance of blameworthiness. What verdict does Smith’s account deliver in a case of this sort? In her nonvolitional account, if one is indeed indirectly responsible for cheating, responsibility for cheating derives from direct responsibility for the evaluative judgment, IMC (or some similar judgment), that cheating expresses. If Blame-1 is an essential component of the epistemic condition for blameworthiness, then to be responsible for judgment, IMC—it is impermissible for you to cheat—you must make IMC despite your nonculpable belief that it is impermissible for you to make it. But it would seem that unless circumstances are truly out of the ordinary, why would you nonculpably believe that it is impermissible for you to make the (true) judgment that it is impermissible for you to cheat? Under ordinary conditions, this further condition concerning judgment IMC itself would not be satisfied, and so, assuming Blame-1, you would not be to blame for cheating, a disturbing deflationary result. 2. “Wrong” and “impermissible” are used interchangeably, as are “right” and “permissible.”
( 134 ) The Significance of Free Will
Furthermore, suppose that to be blameworthy for IMC, condition Blame-2 must also be satisfied: it must also be the case that it is impermissible for you to make the (true) judgment that it is impermissible for you to cheat. Then Smith’s nonvolitional account condemns as morally impermissible the very activity—making a (correct) evaluative judgment—that is central to moral blameworthiness. In other words, suppose you freely cheat despite your true nonculpable belief (or judgment) that it is impermissible for you to cheat. Why should it be supposed that in all such cases it is also true that it is impermissible for you to make the judgment that it is impermissible for you to cheat (if the condition Blame-2 specifies is necessary for blameworthiness)? Finally, as Alfred Mele (1995) remarks, the determinants of a token feeling state can extend well beyond any evaluative assessment of that state, including the formation of any decisive best judgment regarding it (p. 109). Mele explains that in many cases feeling states are the “products of desires, habits of interpretation, learned patterns of emotional response, other emotions, our physiological condition, and so on” (p. 108). This is also true of many of our emotional commitments or cares (Shoemaker 2011, pp. 610–611). Imagine that Jane is in a type of state in which she finds herself in various stressful situations, a feeling state of inadequacy. Furthermore, this is a state that is avoidable for her on the occasion of interest, it has not been formed on the basis of any evaluative reasoning— manifestations of this sort of state trace to various trying events in her childhood—and it is the primary motivational component of a particular action of hers. Suppose she performs this action despite her nonculpable belief that she is doing moral wrong in performing it. Intuitively, Jane is morally blameworthy for this action as many volitionists would agree. Principle Culpability, in contrast, appears to sustain the result that she is not blameworthy for it: she is not directly answerable for the token feeling state because it does not reflect her rational activity; hence, she is not indirectly culpable for the action that owes its generation to this state. So, again, not only does the verdict of the rational relations view conflict with that of the volitional view, it also conflicts with our (pretheoretic) intuitions about the case. Some may object that Jane’s belief that she is doing wrong in performing the action is the evaluative judgment that grounds blameworthiness for her action. Again, though, there is a problem with Jane’s alleged direct blameworthiness for this judgment. Jane may not form this (let us assume) true judgment on the basis of the nonculpable belief that she is doing wrong in forming it. Nor may it be wrong for her to form the judgment that it is wrong for her to perform this action.
R esponsibility for E motions
( 135 )
2. Attitudes and Reasons
Let’s transition from these concerns to Smith’s insightful constitutive thesis that says that pertinent evaluative judgments are partly constitutive of various attitudes. This thesis implies that reasons are integral to many attitudes because evaluative judgments, as Smith conceives of them, are formed on the basis of considering reasons for or against them. I advance what may simply be a variation of the constitutive thesis: sundry attitudes are essentially associated with reasons because, roughly, one cannot be in the particular attitudinal state unless one has reasons of a certain sort. I illustrate with examples. Forgiveness is associated with reasons in a number of ways. For instance, forgiving someone gives one reasons to do things. Your forgiving may give you reason to strengthen the relation with the person you have forgiven. Or your forgiving may give you occasion to reflect on what you may have done that might have triggered the other party to hurt you. But forgiveness is also tied to reasons in another way: one forgives for reasons. There may be many different reasons why we forgive when we do. One may forgive because one thinks it is morally obligatory for one to do so or believes that doing so would please someone else whom one wants to please. Or one may forgive because one is willing to cease to regard the wrong done to one as a reason to weaken or dissolve the relationship (see, e.g., Pereboom 2001). It seems that one cannot fail to have this reason when one forgives because being willing to cease to regard some wrong as a reason to weaken or terminate a relationship is essential to forgiveness. Similarly, when we feel guilt, we feel guilt for reasons. Legitimate guilt contrasts with, for instance, feeling “guilt” even when one believes correctly that one has done no wrong at all. One reason for which one feels legitimate guilt is that one takes oneself to have intentionally done wrong (Zimmerman 1988, 2008; Haji 1998, 2012). It appears that one cannot fail to have this reason when one feels guilt. If one fails to have this reason, then the state one is allegedly in when one allegedly feels guilt is not a state of legitimate guilt, largely, because one’s taking oneself to have intentionally done wrong is essential to guilt. What sorts of reasons are at issue when one forgives or feels guilt? A brief cataloguing of different sorts of reasons will be helpful. Objective reasons are, roughly, facts dissociated from the agent’s desires or attitudes with “built-in” motivation. More carefully, use “motivating desire” to refer to attitudinal states of mind (or their neural realizers) that are, constitute, or include motivation. If R is an objective reason for an agent, S, to do something, R is not a motivating desire of S, does not have a motivating
( 136 ) The Significance of Free Will
desire of S as a constituent, and is not even in part a fact, truth, proposition, or the like about any actual motivating desire of S (Mele 2003, p. 77). Something’s being intrinsically good or someone’s recognizing that something is intrinsically good, for instance, is an objective reason for one to desire or pursue it. Pro tanto objective reasons can be outweighed by other reasons, as opposed to all-things-considered reasons, which cannot be outweighed. If you have several alternatives, and your pro tanto reasons to act in some way are stronger than those to act in any other way, then you reasons-wise ought to act in this way. If you have sufficient pro tanto reason to act in two or more ways and no better reason to act in any other way, then it is reasons-wise permissible for you to act in either of these ways. Finally, if you have most pro tanto reason not to act in a certain way, then acting in this way is reasons-wise impermissible for you. Objective pro tanto reasons contrast with subjective pro tanto reasons: an agent has a subjective pro tanto reason to do something if and only if the agent believes that she has an objective pro tanto reason to do it. Davidsonian reasons are, typically, complexes of desires and beliefs. One may think of such reasons as “internal” because they are constituted by “internal” mental states (or, perhaps, their physical realizers) of the agent whose Davidsonian reasons they are. Some Davidsonian reasons are normative, prescribing what we reasons-wise ought to do—as opposed to being merely explanatory, providing a rational or reasons explanation of the behavior in question. For example, suppose you nonculpably believe that you have decisive objective pro tanto reasons to administer medicine A to a patient because you nonculpably believe, on excellent authority, that giving A will save this patient. You believe, also, that you morally ought to administer A. However, you have decisive objective pro tanto reasons against giving A because doing so will kill the patient. Suppose, furthermore, that you desire to cure the patient, and you give A partly on the basis of the nonculpable beliefs that you can cure by giving A, and that you morally ought to cure. This desire, together with these beliefs, constitute a normative Davidsonian reason to give A. The reasons one must essentially have when one forgives or feels guilt are, presumably, objective pro tanto. To motivate this view, attention is restricted to forgiveness. Preliminarily, we can formulate the principle that ties forgiveness to reasons in this way: Forgiveness-pre: Necessarily, if S forgives T for something, then S has a pro tanto reason (ceasing to regard the wrong done as a reason to weaken or end the relationship—reason “FR”) to forgive T.
R esponsibility for E motions
( 137 )
When one forgives, we should be careful to distinguish between having reason FR as a reason for forgiving and forgiving for this reason. My claim has to do with the former and not the later. An analogy will help to explain the distinction. The following principle seems true: Obligations are tied to reasons (OR): If an agent has a moral obligation to do something, A, then the agent has an objective pro tanto reason to do A.
Suppose you morally ought to visit your friend in hospital. Then you have an objective pro tanto reason to visit: your being obligated to visit. You may, however, have other reasons to visit. For example, you may believe that if you fail to visit, you will feel guilty, and you want to avoid feeling guilty. Suppose it is some such Davidsonian reason for which you visit. It does not follow that you do not have the relevant objective pro tanto reason: it is obligatory for you to visit. Similarly, you may forgive because you think it is morally obligatory for you to do so. In this case, nothing precludes you from having reason FR to forgive; it is simply that you do not forgive for FR. I have proposed that reason FR is an objective pro tanto one. Why might one conjecture otherwise? A motivating desire is an attitudinal state of mind (or the neural realizer of this state) that is, constitutes, or includes motivation. One might think that you cannot be willing to cease to regard something as such and such, or you cannot cease to regard something as such and such, without having a pertinent motivating desire. But this is false. One’s being willing to regard something—a brusque remark, for instance—as out of character does not imply that one has a motivating desire concerning being willing to regard the remark as out of character. As another example, you may be willing to regard some person you know to be culpable as innocent without having any motivating desire to regard this person as innocent. Alternatively, one might argue that to regard something as such and such is to engage in mental activity. Perhaps, when you regard something as such and such you perform a mental action or you engage in mental activity that is somewhat similar to performing a mental action in that this mental activity may well have causal antecedents such as beliefs and desires. But then, it may be claimed, regarding something as such and such is indirectly linked to some motivating desire. Suppose we agree that being willing to regard something as such and such, or regarding something as such and such, qualifies as a mental action, and, furthermore, we concur that as a mental action, there is some motivating desire that this action has in its etiology. It would, however, not follow that reason FR itself is a motivating desire. Nor would it follow that FR has a motivating desire as a proper constituent as, for instance, would be the case with the desire
( 138 ) The Significance of Free Will
component of a Davidsonian reason of this sort: Fred opened the window because he wanted to let in some fresh air and believed he could do so by opening the window. I can think of no other reasons to disqualify reason FR from being an objective reason. Principle Forgiveness-pre may now be regimented in this way: Forgiveness: Necessarily, if S forgives T for something, then S has an objective pro tanto reason—being willing to cease to regard the wrong done to S as a reason to weaken or dissolve the relationship—to forgive T.
3. A Requirement of Alternative Possibilities for Objective Pro Tanto Reasons
Elsewhere, I have argued that there is an alternative-possibilities requirement for our having objective pro tanto reasons. No one can have such a reason to do something, unless one could have done otherwise. Here, I give a very brief sketch of the argument. Just as the moral “ought” implies “can” (Kant’s Law), so the objective reasons-wise “ought” implies “can” (see, e.g., Streumer 2007; Haji 2012). Indeed, Kant’s Law just is a special instance of this principle: If one reasons- wise ought to do A, then one can do A; and if one reasons-wise ought not to do A, then one can refrain from doing A. If reasons-wise “ought” implies “can,” then reasons-wise “permissibility” (and “impermissibility”) imply “can” too. The most fundamental consideration for this symmetry is straightforward. Starting with a comparison, moral praiseworthiness and blameworthiness require freedom (or control). An essential element of the freedom requirement of these appraisals is captured by the principle that one is morally praiseworthy or blameworthy for doing something only if one could have done it. It would be implausible to suppose that while praiseworthiness requires control, blameworthiness does not (or vice versa). This principle highlights a link that holds of conceptual necessity between such appraisals and freedom. Again, the link is simply that praiseworthiness and blameworthiness both require control. Moreover, if we think that praiseworthiness requires a certain variety of control, then without good reason to believe the contrary, blameworthiness, too, requires this variety of control. Similarly, obligation, permissibility, and impermissibility require control. As with the former sorts of appraisals, it is implausible to suppose that, for instance, while obligation requires control, impermissibility or permissibility do not. Furthermore, it would seem that the control requirements of moral obligation, unless we have strong reason to think otherwise, should also be the very ones of
R esponsibility for E motions
( 139 )
moral impermissibility and permissibility. Kant’s Law can be conceptualized as a control principle for moral obligation: if you ought to perform an action, you have obligation-relevant control in performing it. If Kant’s Law expresses just one more instantiation of the association between morality and freedom, then, again, in the absence of special reason to believe otherwise, it should also be the case that the principles that “impermissible” implies “can” and “permissible” implies “can” express two other instances of this association. In sum, if obligation—whether it is reasons-wise obligation or moral obligation—requires control, so do permissibility and impermissibility. And, if obligation requires a certain sort of basic control—that we can do what we are obligated to do—permissibility and impermissibility require this sort of basic control as well. We may now argue as follows. (In what is to come, unless otherwise specified, “a reason” refers to an objective pro tanto reason.) If it is reasonswise impermissible for one to do something, one reasons-wise ought not to do it. If one reasons-wise ought not to do something, one can refrain from doing it. Hence, if it is reasons-wise impermissible for one to do something, one can refrain from doing it. But it is also true that if it is reasonswise impermissible for one to do something, one can do it. So, there is an alternative possibilities requirement for reasons-wise impermissibility. Regarding reasons-wise obligation, if one reasons-wise ought to refrain from doing something, it is reasons-wise impermissible for one to do it. If it is reasons-wise impermissible for one to do something, one can do it (from reasons-wise “impermissible” implies “can”). Therefore, if one reasonswise ought to refrain from doing something, one can do it. But it is also true that if one reasons-wise ought to refrain from doing something, one can refrain from doing it. In other words, just as there is a requirement of alternative possibilities for reasons-wise impermissibility, so there is such a requirement for reasons-wise obligation. If reasons-wise impermissibility and reasons-wise obligation require alternative possibilities, there is little reason to deny that reasons-wise permissibility, too, requires alternatives. We may conclude that there is an alternative possibilities requirement for the truth of judgments of objective pro tanto reasons.3
4. Attitudes and Alternative Possibilities
Consider the following template: various things we value are essentially associated with objective pro tanto reasons. So, there is an alternative possibilities requirement for these things because there is such a requirement 3. See Haji (2012, ch. 2) for further discussion.
( 140 ) The Significance of Free Will
for having such reasons. The instantiation of this template so far proposed involves various attitudes such as forgiveness or guilt. With forgiveness, for instance, necessarily, if you forgive someone for something, you have an objective pro tanto reason—being willing to cease to regard the wrong done to you as a reason to weaken or dissolve the relationship—to forgive this person. This fact, together with the principle that there is an alternative possibilities requirement for having objective reasons, yields the result that certain attitudes, such as forgiveness, presupposes our having alternatives.
5. Determinism and the Reactive Attitudes
Does determinism imperil our having attitudes that are essentially associated with objective pro tanto reasons, and so with our having alternatives, in the fashion previously explained? That depends on whether the alternatives are weak compatibilist ones or strong incompatibilist ones.4 It is customarily acknowledged that determinism effaces strong alternatives (e.g., Wiggins 1973, van Inwagen 1975, Fischer 1994, Ginet 2003). If, on the one hand, the alternatives many of our attitudes presuppose our having are strong, then determinism undermines the rationality of these attitudes. Similarly, if, for example, moral obligation (see principle OR above), or certain axiological assessments concerning intrinsic goodness, or a number of aretaic assessments, are essentially tied to our having objective pro tanto reasons, and these reasons require that we have strong alternatives, then Kane’s claim that determinism undercuts many things of value to us is on firm footing. If, on the other hand, the pertinent alternatives are weak, then sundry forms of semicompatibilism are imperiled. Assuming that obligation is essentially tied to our having objective pro tanto reasons, for instance, the position that determinism is compatible with obligation independently of whether it is compatible with freedom to do otherwise is untenable. Let’s briefly revisit Smith’s rational relations view. Smith proposes that one is morally responsible for an attitude, choice, or overt action only if one is responsible for the evaluative judgment it expresses. One is responsible 4. Strong alternatives may be more precisely characterized in this way: assuming agent S does action A at time t in world w, S has a strong alternative at t if the combination of w’s past and w’s laws of nature is consistent with S’s not A-ing at t. In one conception of weak alternatives, for instance, although an agent does one thing—she donates to UNICEF, say—she would have refrained from donating had she wanted, tried, intended, or chosen not to donate (e.g., Berofsky 2003, 2011; Vihvelin 2004).
R esponsibility for E motions
( 141 )
for such a judgment if it is formed on the basis of one’s rational activity; it is responsive to reasons one has. It would be implausible for a theorist of this sort to suppose that whenever an agent is morally responsible for some judgment, the agent has formed this judgment on the basis of reasons none of which can be an objective one. There is no reason whatsoever for a theorist who accepts the rational relations view, or more broadly, some reasons-responsiveness account of responsibility, to think that, necessarily, or coincidentally, or nomically, responsibility is wedded to our being aptly responsive only to “nonobjective” reasons. For example, such a theorist should presumably grant that there are occasions when an agent who is blameworthy, and so responsible, for doing something, recognizes that that thing is impermissible but still does it despite nonculpably believing that it is impermissible for her to do it. But the fact that something is impermissible for one is a paradigmatic example of some fact’s being an objective reason for one. It appears, then, that if one is inclined to accept a Smith-like reasons-responsiveness account, such an account will have to appeal to agents’ having alternatives on at least some occasions when they are responsible. And this is because they will have formed their evaluative judgments on the basis of, or will have been suitably responsive to, reasons, at least some of which are objective, and there is an alternative possibilities requirement for having such reasons. In conclusion, one might take issue with the considerations advanced for the view that several attitudes or emotions are essentially associated with objective pro tanto reasons. Still, it may be well worth keeping in mind the following template. Various things we value are essentially connected with objective pro tanto reasons. So, there is an alternative possibilities requirement for these things because there is such a requirement for having such reasons. As a general strategy to support Kane’s view that determinism poses a threat to many things we value, one is invited to plug into this template things one believes are suitable candidates for what we value.
•
Chapter 10
Moral Responsibility, the Reactive Attitudes, and the Significance of (Libertarian) Free Will Dana Kay Nelkin
1. Introduction
I
t is often tempting to take it as a given that the topic of free will is an important and central one in human life, and then go on to engage in the rigorous debate about what having free will requires and whether we have it. Robert Kane’s The Significance of Free Will takes nothing for granted. Though the book is perhaps best known for its wonderfully detailed and subtle development of a libertarian theory of free will, the task of the central chapter of the book, as the title also makes clear, is to ask and answer “the Significance Question.” The question turns out to be a closely related pair of questions: The Significance Question: Why do we, or should we, want to possess a free will that is incompatible with determinism? Is it a kind of freedom “worth wanting” (to use Dennett’s useful phrase) and, if so, why? (Kane 1996, p. 13) I am very grateful to David Palmer, Derk Pereboom, and Sam Rickless for most helpful comments and discussions on previous drafts of this chapter. Many thanks, too, to David Palmer for his wonderful work in conceiving, editing, and shepherding this volume through and, finally, to Robert Kane for providing the rich and thought-provoking material that made it possible.
M oral R esponsibility, R eactive Attitudes
( 143 )
Or, to put it in other words, why have so many people—both traditional philosophers and others—wanted a kind of agency or authorship of their actions that requires indeterminism, and why should we? (p. 80). Now I take it that one question is really psychological or etiological (what is it about people that makes them want this form of agency?), and the other normative (what is valuable about libertarian freedom?). Kane’s answers to these questions have clearly contributed both to framing and further driving the debate about the significance of free will. In what follows, I will set out those answers, which raise a number of independently interesting issues, and which include an analysis of the free will debate itself, commitments to meta-ethical principles in value theory, and substantive psychological explanations for why we care about what we do. As Kane (1996) notes at the outset, “the usual answer” is that libertarian free will is necessary for a number of other things that people want and that are worth wanting. Candidates include: “(1) genuine creativity; (2) autonomy (self-legislation) or self-creation; (3) true desert for one’s achievements; (4) moral responsibility in an ultimate sense; (5) being suitable objects of reactive attitudes such as admiration, gratitude, resentment, and indignation; (6) dignity or self-worth; (7) a true sense of individuality or uniqueness as a person; (8) life-hopes requiring an open future; (9) genuine (freely given) love and friendship between persons (or in religious contexts, freely given love toward God); and (10) the ability to say in the fullest sense that one acts of one’s own free will” (p. 80). But this is just the first step, and Kane makes clear that ultimately the answer lies in focusing directly on the value of the indeterministic origination, within one’s self, of one’s actions. In section 2, I set out and assess Kane’s approach to the Question of Significance. In section 3, I offer a different approach, and explore the application of the competing approaches to moral responsibility and the reactive attitudes thought to presuppose it.
2. Answering the Significance Question from the Value of Origination
According to Kane, identifying the things worth wanting that are thought to presuppose libertarian free will is the first step in the debate over the Significance Question. But, as Kane is quick to point out, this answer has not been accepted by compatibilists in the debate. Compatibilists tend to agree that these are all things worth wanting, but they disagree that libertarian free will is required to achieve them. Rather, compatibilists offer “interpretations” of each of the ten items that do not require libertarian
( 144 ) The Significance of Free Will
free will in particular. To that, incompatibilists have replied in turn that the compatibilist interpretations are not “what we really want.” According to Kane, this is the first stage in the dialectic about libertarian free will. For example, when it comes to (1) on the list—creativity—Kane quotes Karl Popper, writing that determinism “destroys” the idea of creativity. It reveals an illusion in the idea that one has created a lecture, since instead “there was no more in it, according to physical determinism, than that certain parts of my body put down black marks on white paper” (Kane 1996, p. 81).1 Or consider (3) and (4), desert and moral responsibility. Kane has us imagine a medical researcher whose discovery of a potent new antibiotic drug marks a great achievement. When colleagues say that it is all a matter of luck, since she was just in the right place at the right time, she is indignant, claiming that she is responsible for some of those very circumstances that led her to discover the drug. “What this shows,” writes Kane, “is that the woman’s conception of true desert for her accomplishments involves the idea that she is, to some degree, ultimately responsible for them,” where ultimate responsibility entails indeterminism (p. 82).2 Now substitute for the scientific discovery a morally praiseworthy act like a man’s saving a drowning child. He feels demeaned by the charge that he is undeserving because he just responded out of instinct, and would insist that the act “was not entirely caused by circumstances that were not up to him and not ultimately his doing” (p. 83). Similar points apply, argues Kane, to blameworthy acts, particularly when punishment is involved. Compatibilists respond to each of these points by claiming that in perfectly ordinary senses of the words, determinism does not preclude creativity or novelty, desert for achievements, or blameworthiness and praiseworthiness. For example, Beethoven’s Fifth Symphony would still be a remarkable new contribution to human history, even if determinism were true, and Beethoven would be admirably creative for having produced it. Of course, there is another notion of creativity that would be precluded by determinism, but that requires defining such a notion precisely in opposition to “determinism” and is not the ordinary notion. Similarly, when it comes 1. Kane’s quote comes from Popper (1972, p. 222). 2. Ultimate responsibility, as Kane (1996) sets it out, is as follows: an agent is ultimately responsible for some (event or state) E’s occurring only if (R) the agent is personally responsible for E’s occurring in a sense which entails that something the agent voluntarily (or willingly) did or omitted, and for which the agent could have voluntarily done otherwise, either was, or causally contributed to, E’s occurrence, and made a difference to whether or not E occurred; and (U) for every X and Y (where X and Y represent occurrences of events and/or states) if the agent is personally responsible for X, and if Y is an arche (or sufficient ground or cause or explanation) for X, then the agent must also be personally responsible for Y (p. 72).
M oral R esponsibility, R eactive Attitudes
( 145 )
to moral responsibility, in ordinary life, the default is to hold people responsible unless there are some excusing or mitigating conditions, such as coercion, ignorance, delusion, or manipulation.3 No mention is made of determinism. And, writes Kane, if moral responsibility can be understood in these terms, so can desert for achievements and more (p. 90). At this point, Kane’s incompatibilist will reply that the compatibilist is correct that there are notions that are compatible with determinism, but that there are more “exalted” counterparts that require indeterminism and that we also want (p. 91). Yet Kane takes it that the compatibilist will reply that this last claim is question-begging. If the more exalted notions are just defined as incompatibilist versions of the ordinary notions, then the claim appears to build in the conclusion. Hence, we have what looks like a kind of stalemate. The resolution, as Kane (1996) sees it, is to “look more closely at the notions of ultimate responsibility and underived origination themselves. Why are they so important? What lies behind the widespread conviction . . . that ultimate responsibility and underived origination confer greater value on each of the goods (1) to (10)?” (p. 91). Compatibilists are often entitled to their notions for everyday life and the law courts, but when it comes to our place in the world, we must become metaphysicians, and ask these questions about the importance of concepts that build in indeterminism. (In what follows, I will use “indeterministic origination” to refer to ultimate responsibility and underived origination, for simplicity and to highlight the required indeterminism.) Importantly, Kane here reverses the direction with which the debate, as described earlier, began. Rather than asking about the nature of what is claimed to be at stake and then enquiring into whether each phenomenon, precisely identified and evaluated, really requires indeterminism, we are to instead begin with the value of indeterministic origination itself. Kane’s approach is also distinctive in the particular way in which it answers the Significance Question. For, as we will see, Kane offers an explanation for why we want to be undetermined originators in developmental psychological terms, and at the same time he explicitly omits to endorse the claim that we are all rationally required to value such undetermined origination.4 3. A compatibilist might also add a counterpart for excuses when it comes to praiseworthy action. If the saving of the child Kane describes were a result only of instinct, so that it wasn’t done even with a view to preventing a harm to the child, then compatibilists could also have a reason for withholding moral praise. But if the person saved the child because she saw immediately the point of doing so without having to think about it, then the intuition that this description of her action would be demeaning in the least, or detract from her praiseworthiness in a sense worth wanting, is far less powerful. 4. Another perhaps surprising element is the concession that compatibilist notions are often all that is needed in ordinary life and in the courts. This provides some mutual
( 146 ) The Significance of Free Will
Kane offers two lines of reasoning to explain why we value indeterministic origination, the first with roots in a developmental story about human experience of the self, the second relating it to the notion of objective worth. Reflecting on the experience of human infants who take great pleasure and a keen interest in what they cause by acts of their wills, Kane observes that this experience is what helps give infants a sense of self, distinct from the rest of the world. As our understanding grows, we become aware that we are influenced by the world in myriad ways and that we are physical objects just like those in the rest of the world. This provokes a “spiritual crisis,” according to Kane (1996, p. 94). And our worries about determinism can be seen as an extension of this sort of doubt. It is “a stage of the dialectic of selfhood” (p. 95), a stage in a process in which we learn about ourselves and our relation to the world, all the while trying to hold on to the idea that we are independent sources of motion in the world distinct from the forces of the external world. It is in this way, according to Kane, that the dialectic of selfhood leads us to the notion of indeterministic origination itself. Thus, at least one reason we want indeterministic origination is the same reason that children and adults revel in their accomplishments, even their very earliest, namely, that they are thereby enabled to distinguish themselves from the world (p. 98). A second explanation for why we value indeterministic origination is its connection to the notion of objective worth. Because we value our own objective worth, we value such origination. Kane offers this example in illustration: Alan is a despondent artist. A rich friend who wants to lift his spirits arranges to have his paintings bought under an assumed name for a large sum of money. Alan believes that his paintings are being bought because of their artistic value and cheers up. He dies happy, but under an illusion. Kane then has us compare this world to one in which Alan really is a great artist.5 Suppose that Alan could stand outside both worlds, would it make a difference to Alan which world he inhabits (not that he would know)? We can easily imagine that Alan would think his life more valuable if he inhabited the second, despite the fact that he would be “subjectively” happy in both. If we value objective worth, then we value indeterministic origination because the latter requires us to view the universe from an support for the general approach. If one is willing to concede that ordinary notions of moral responsibility, say, don’t implicate indeterminism, then the direct question about the value of indeterministic origination takes on a kind of priority. However, it seems that it is precisely our ordinary concepts and practices that are central in the debate. As Kane recognizes, too, for example, our very justification for punishment often seems to hinge on resolving the metaphysical question at issue one way or the other. 5. As Kane notes, this example is similar to Nozick’s “experience machine” (Nozick 1974, pp. 42–45).
M oral R esponsibility, R eactive Attitudes
( 147 )
bjective perspective. If we care about our worthiness above and beyond o our being pleased, then we want to be undetermined originators. Thus far, we have two explanations for why we want undetermined origination. Do these explanations also justify the claim that such origination is objectively valuable? Do we have an answer to the second part of the Significance Question? Kane does not go so far. He claims that there cannot be an argument that we are rationally compelled to value such origination; rather, at this point, argument gives out and we can simply use these explanations to commend this kind of freedom (p. 99). While the explanations are intriguing, I believe the particular explanatory claims are stronger than the evidence warrants in the end. Thus, questions remain about just why, and even whether, we value indeterministic origination.6 At the same time, as I will argue, the general approach to claims about objective value does not go far enough. While the restraint is admirable in itself, it is in tension with what I take to be intuitive judgments about other values—including many items on the list of what is supposed to be at stake in the debate. To see why, consider the explanatory arguments first. The early developmental story about the will and our sense of self is both plausible and important. But the crucial question is whether we should see our concern with determinism as an extension of the desire to be willing actors in the world, a desire for something that underlies our very sense of ourselves as selves. This is the move that is not supported here. We can distinguish the things we do from the things that happen to us even if determinism is true. And we can establish our sense of a self, distinct from the rest of the world, by simply understanding that we control some things in the world. We do not in turn need to have control over the source of that control to distinguish ourselves and feel our effects in the world. It is a further philosophical question whether we need a sort of ultimate control for other things we might want—moral responsibility, the justification of punishment, the justification of the reactive attitudes, genuine friendship, and so on. But the account of the development of one’s sense of self by itself does not support this strong conclusion.7
6. It is important to note that Kane (1996) recognizes that not everyone does so value indeterministic origination, referring to certain Buddhists as one example (pp. 99–100). 7. See Norton Nelkin (1995, pp. 299–315) for a detailed defense of this point. It includes a developmental story that appeals to key findings in the developmental literature, shares key aspects of Kane’s story of the fundamental importance of the will to the sense of self (and the world), and yet explicitly rejects the idea that this story is continuous with our concern about determinism.
( 148 ) The Significance of Free Will
The second line of reasoning takes the desire for indeterministic origination to be of a piece with the desire for objective worth. The particular argument contains a powerful intuition pump and does pump (at least for me) the intuition that the value of a life depends on more than just a subject’s feelings and attitudes. We should conclude both that there is a strong case for the idea that being under an illusion adds disvalue to a life and that it is better to be a great artist than not to be.8 But neither of these points adds up to indeterministic origination—unless either being undeceived or being a great artist is understood in a way that presupposes it. The first idea seems a nonstarter, since it seems clearly possible to be undeceived with respect to one’s (lack of) talent in a deterministic world. If we were to try to extend the second idea, it seems that the two worlds we should compare are these: a world in which Alan is a talented artist in a deterministic world and knows it and a world in which Alan is a talented artist in an indeterministic world and knows it. If he (and we) would choose the latter, this would show that we value indeterministic origination. But at least for me, this intuition pump does not have the force of the first, and, interestingly, it does not even appeal to objective worth per se.9 We could instead compare two worlds that differ only according to Alan’s deservingness, and perhaps this is what Kane has in mind. But here I think we could only reach the conclusion about our valuing of indeterministic origination if we already assume desert for artistic achievement requires indeterminism (as opposed to requiring only talent, hard work, dedication, and so on). This would be to put us back at what Kane sees as the first stage 8. At the same time, a different pair of worlds is needed to provide full support for the conclusion that living without illusion is valuable in itself. For the worlds described here differ not only in there being deception in one and not the other, but in the fact that Alan is a great artist in one and not in the other. To isolate the deception, consider a world like that of the film, The Truman Show, in which although inhabiting a world in which one has the love of his family and friends is better for Truman than one in which he doesn’t, it is still better overall to not be deceived. Interestingly, Kane (1996) suggests that a third world in which Alan finds out the sad truth would be worse than the other two worlds (p. 97), but I do not have a strong intuition about this case, at least not with respect to the case as described thus far. 9. See Fischer (2002), who also rejects Kane’s thought experiment but for somewhat different reasons. He attributes to Kane the premise that compatibilist views are committed only to subjective value because he assumes compatibilism does not have a “historical” component and is only concerned with the structural arrangements of mental states. Because, as Fischer points out, his own account includes a requirement that the responsible agent have taken responsibility for the mechanism that generates his action and because it requires that the agent have the right connections to the world, his view is not targeted by this argument (p. 15). I agree and simply emphasize that even a nonhistorical compatibilist account (e.g., a reasons-responsive account without an additional historical condition) can accommodate the intuition that is pumped by the thought experiment.
M oral R esponsibility, R eactive Attitudes
( 149 )
of the debate, rather than showing independently something more fundamental about the value of indeterministic origination.10 Thus, I do not think that either explanation is adequate to account for the desire for indeterministic origination. Importantly, though, attention to human development does contribute a great deal to our understanding of why the will—and control—are important to us. Nevertheless, once we leave behind the extended developmental account that includes attitudes toward indeterminism, it is less clear just how widespread the desire for indeterministic origination is. At the same time, Kane may give up too soon on the question of whether we can make more progress on the question of whether such origination is worth wanting. The question is important because the significance of the entire debate about free will depends on at least something being worth wanting—indeterministic origination or any of the ten (or more) things thought to depend on it. At this point, it should be noted that whether it is rationally required for someone to actually value something is not the same question as whether it is actually worth wanting. (Thus, Kane’s pessimism about the possibility of rational persuasion need not be the same as pessimism about objective value.) But the two questions are connected—if we are to show that something is worth wanting, we need to give reasons of some sort for it. If indeterministic origination turned out to be like a case in which someone enjoys collecting small bits of paper and if we cannot say why this activity is valuable beyond his liking it, then there would be a serious gap in our account of the significance of such origination, and it seems we would have left the question of whether it is worth wanting behind altogether.11 But there seems to be something in between being a mere matter of taste (like collecting paper, perhaps) and something that everyone is rationally required to value for himself or herself, and that might be captured by the idea that it is reasonable for some people to value it.12 For example, I may not value 10. See Kane (1994) for the case of Alan as an intuition pump for our view of objective worth but with no conclusion drawn about indeterministic origination. 11. It is true that desire satisfaction itself is a candidate for making the desired object worth wanting. And it seems right, as Frankfurt (1971) points out, that a desideratum in any account of free will should explain why we would want it. But if this were all we could say, then the significance of indeterministic origination, or free will, or anything on the list of what is at stake, would seem to be of a shallow sort. 12. See Mele (2006), which presents a view that seems to seek a middle ground and also seems in the spirit of Kane’s position. In particular, Mele considers what he calls, “Daring Soft Libertarianism,” or DSL, according to which libertarian freedom is of a kind that some find more valuable than a compatibilist kind. This view shares with Kane’s the rejection of the claim that everyone ought to value being undetermined sources of their actions. As Mele writes, “The soft libertarian need not claim that everyone—or every reasonable person—values being an indeterministic initiator” (p. 98).
( 150 ) The Significance of Free Will
my being able to sing well in front of large numbers of people, but I understand that doing so is valuable and contributes significantly to lives worth living. What can it mean to be reasonable that goes beyond a matter of mere taste? One possibility is that it is one way among many to achieve something that is worth wanting for all—for example, a life in which one employs one’s talents to their fullest.13 This seems to explain well our judgments about a number of lives that are intuitively flourishing ones but quite different in their details. But it is less clear how to apply this idea to indeterministic origination. What would be the larger category of good for which this is one among other ways to achieve it? Without an answer to this question, we seem unable to go beyond the claim that our desire for indeterministic origination is a matter of taste. It is a notoriously difficult question how we can show anything to have objective value or to be rational, or even reasonable (in the sense just described), to value. But Kane’s discussion shows us in a vivid way that we must confront such questions if we are even to interpret correctly the second part of the Significance Question or to see it as a well-formed one. While it may be very hard to elicit direct intuitions about the value of being indeterministic originators in the absence of other things which may depend on our so being, this very fact seems to me to point to the fruitfulness of the first general direction of debate—eliciting such intuitions about the value of other things and then figuring out whether they require indeterministic origination after all.14 Ideally, we will be able to do more than elicit But, just as a gambler might show a risk-averse person what the former finds appealing in gambling, a soft libertarian might do something similar for one who does not herself value libertarian freedom. The particular analogy used suggests that DSL takes libertarian freedom to occupy a place closer on the spectrum to being a matter of taste than to being something of objective value. Like Kane’s idea of being able to “commend” indeterministic freedom, Mele offers us the idea that a gambler might be able to explain to others who don’t share his desire why it is that he likes it. 13. Kane (1994) seems to adopt something like this view, where he takes it that a number of different lives are valuable, but no one life is called for by all. The view can be described as a pluralist conception of an “objective list” view of well-being (e.g., Hurka 1993). 14. It is worth noting two additional reasons why one might continue with the original direction of enquiry in the debate. At the outset, it is possible to be more optimistic than Kane seems to be about the possibility of progress if we begin with the value of the items thought to be at stake. For example, for at least some items on the list—such as the reactive attitudes—there is arguably a univocal notion shared by compatibilists and incompatibilists alike, and then it seems natural to focus on the questions of whether that agreed-upon phenomenon does or does not require indeterminism, and whether the items really are worth wanting, once we have a better idea of their nature. A second reason to resist reversing direction targets the idea that the presence of indeterministic origination is what is supposed to explain the exalted status of the indeterministic version of each item on the list. But one might ask why,
M oral R esponsibility, R eactive Attitudes
( 151 )
intuitions. While it may be that at some point, argument runs out in the sense that there is no more fundamental value on which the value of, say, friendship or love rests, that need not mean that argument runs out in the sense that we can’t make further appeal to the value’s explanatory power, connection to other commitments we have, and intuitive appeal. In the second part of the chapter, I compare the approach just assessed to the one that proceeds by first asking about the nature and value of those things thought to be at stake in the debate by applying both approaches to moral responsibility and the reactive attitudes that are thought in turn to depend on it. Call this the “what is at stake” approach. I do not claim that the latter approach has (yet) given us a definitive answer, but I hope to show both some of the fruitful ways in which this alternative approach has already been pursued, and a potential direction for further progress. Finally, I return to an insight of Kane’s, showing how it might be combined with the “what is at stake” approach in a promising way.
3. A Contrasting Approach and a Case Study
My approach contrasts with Kane’s by beginning with what is at stake, asking about its value, and asking whether what is valuable really requires libertarian freedom (or even freedom at all). To illustrate the application of the two approaches, turn first to a central item on the list, possibly the one that has been most discussed in recent decades: moral responsibility. According to Kane’s approach, there are two notions of moral responsibility, one that requires indeterminism and one that does not. The compatibilist version works for many everyday purposes, but it is reasonable to want the “more exalted” variety. We cannot claim that everyone should value this, on pain of irrationality, but there is a deep developmental explanation for our desire, and it is connected with something that seems of obvious value—our sense of selves and understanding of the very distinction between the self and world. But if this explanation fails for the reasons described earlier, then not only do we not have an argument for why we should value it, we lack even a motivation for locating this phenomenon in the world. even if we could show why indeterministic origination is desirable, we should thereby favor an indeterministic version of friendship, say, rather than simply a compatibilist version of friendship, together with the additional, but distinct, fact of our being undetermined originators. Still, neither of these points detracts from the importance of asking directly about the value of indeterministic origination, for even if there were no obvious connection to any items on the list, it would still be a question worth asking.
( 152 ) The Significance of Free Will
On the “what is at stake” approach, we should begin with the nature and value of moral responsibility and, separately, investigate whether it requires or presupposes indeterministic origination. Even here, there are a variety of ways we might proceed. For moral responsibility might be valuable in itself, or it might be valuable because of what it is in turn required for, or, as I believe, both. For present purposes, I will follow Kane—and many others—in accepting that moral responsibility is presupposed by the reactive attitudes, those attitudes of people that, in the words of Peter Strawson (1962), “are directly involved in transactions with each other . . . the attitudes and reactions of offended parties and beneficiaries . . . such things as gratitude, resentment, forgiveness, love and hurt feelings” (p. 190).15 On some views, responsibility is simply constituted by the appropriateness of the attitudes,16 but I will make a more minimal assumption here. The appropriateness of the reactive attitudes depends on our being responsible agents. This assumption is consistent with responsible agency being constituted by wholly response-independent conditions. What is essential here is that the attitudes are appropriate only if their targets are morally responsible agents. If this is correct, then it follows that if the reactive attitudes and our susceptibility to them are themselves valuable, then the value of our being morally responsible is (at least in part) derived from the value of appropriate reactive attitudes. Thus, in what follows, I will approach the question of the value of moral responsibility itself through the question of the value of the reactive attitudes and our disposition to them.17 15. The connection between reactive attitudes and moral responsibility is often taken to mark the fact that the notion of responsibility in question is that of “accountability,” a notion distinguished from “attributability” (see Watson (1996) for the introduction of these terms). One might blame someone in the sense of attributing a moral fault to an agent. One might call a piece of behavior (e.g., a colleague’s failure to return a book) “shoddy,” for example. So far, this is only to blame in the “attributability” sense. But this is not yet to blame in a sense that presupposes that the agent deserves to be held responsible, that demands can appropriately be made, and so on. The reactive attitudes are thought to embody the “making of demands” (Strawson 1962, Watson 1975). Interestingly for our purposes here, these two notions are sometimes taken to correspond to compatibilist and incompatibilist notions, with accountability requiring indeterminism. But there are compatibilist and incompatibilist theories of accountability, so for now, I will treat the notion in a neutral way. 16. See, for example, Wallace (1994). This is also a natural interpretation of Strawson (e.g., Watson (1987)). 17. In fact, I believe that moral responsibility is also valuable in virtue of what moral responsibility consists in. It is, of course, a matter of great controversy under what conditions we are responsible. I have elsewhere defended the view that we are responsible insofar as we act with the ability to do the right thing (or a right thing) for good reasons (Nelkin 2011). Although it is controversial whether this captures the conditions of responsibility, it seems to me much less controversial that acting with this ability is valuable. A third way in which responsibility (whether moral or not) may be thought to be valuable is its being required for our living a life without illusion. If,
M oral R esponsibility, R eactive Attitudes
( 153 )
Even having narrowed our focus in this way, there remain multiple routes to travel. For example, the reactive attitudes might be valuable because in turn they contribute to still other things that are valuable, or they might be valuable in themselves. If the former, then they might contribute to other goods necessarily or only contingently. The current debate so far seems to be centering on the idea that reactive attitudes are valuable for their contribution to further goods. And here there is great variety in the further goods identified; they include good social structures and features like decreases in violence and disharmony, as well as the very existence of personal relationships. There is seemingly vast consensus that these are things worth wanting, so the debate has largely focused on whether the reactive attitudes do in fact contribute to these goods. More precisely, the debate has largely centered on whether the reactive attitudes are the best—or only—way to achieve them. Here I enter the debate in one of its aspects, focusing on the claim that there is an unavoidable (or at least for most of us, unavoidable) connection between the reactive attitudes and personal, caring, relationships. For if it can be shown that the reactive attitudes are valuable because they (at least for most) make personal and caring relationships possible, then we would be able to show not only the value of the reactive attitudes but also—in one way—of morally responsible agency itself. At the same time that Strawson introduced the term “reactive attitudes,” he argued for its importance to the debate about free will precisely by claim ing their essential connection to personal relationships (pp. 190–192). Without these attitudes that are reactions to the good or ill will of others and through which we demand good will from those with whom we stand in a variety of personal relationships, we are left, on Strawson’s view, to adopt the “objective” attitude, to see a person “perhaps as an object of social policy; as a subject for what, in a wide range of sense, might be called treatment; as something certainly to be taken account, perhaps precautionary account, of; to be managed or handled or cured or trained; perhaps simply to be avoided” (p. 194). If these are our only alternatives, then the high value of the reactive attitudes does not seem to require much in the way of argument. But that the objective attitude and the reactive attitudes that presuppose responsibility are exhaustive alternatives is an assumption that has for example, we must take ourselves to be free and responsible agents, then our failure to be such agents would condemn us to a life of illusion. I discuss this proposal in detail elsewhere (Nelkin 2011). See Wolf (1981) for discussion about how this point is connected to the value of the attitudes themselves, and Milam (2013) for a valuable response.
( 154 ) The Significance of Free Will
not held up well. In fact, Derk Pereboom has offered a rich and subtle account of a whole set of other options, or what he calls “analogues” of the reactive attitudes.18 Even if we were to set aside the reactive attitudes that presuppose responsibility, many other personal attitudes remain, including moral sadness, disappointment, resolve, and analogues to forgiveness and gratitude that retain many positive features of each while eliminating any presupposition of responsibility. But this is not the end of the debate. For at this point, there remains a question of whether we can eliminate our susceptibility to, or even regularly disavow, the reactive attitudes and still maintain our personal relationships. If we cannot—even if only because of our human psychological limitations—this suggests that such susceptibility has significant value, contingent though it may be. And, even if we settle this (at least partly) empirical question, there is a further question: Are the admittedly personal relationships that would remain without the specifically reactive attitudes as valuable as relationships in which we are susceptible to such attitudes? I address each of these questions in turn. Taking up Strawson’s cause, Seth Shabo (2012) has recently addressed the first question with a qualified “no.” While there may be exceptional people who can disavow the reactive attitudes and still maintain caring personal relationships, most of us cannot do so. Shabo argues that genuine loving relationships, such as romantic ones, are ones in which the participants care not only for the other person but also care about the other’s attitudes and feelings about them. That is, if one is not susceptible to hurt feelings if, say, one’s partner is unfaithful, or shows one disrespect or unkindness, then this is a sure sign that one does not have the kind of caring relationship that is characteristic of romantic love. One has to care about the other’s feelings and attitudes displayed in his or her behavior, and to do this one must be at least susceptible to hurt feelings. As Shabo puts it, one must be susceptible to “taking things personally,” and having a certain degree of “emotional vulnerability” (p. 112). In turn, hurt feelings “often beget resentment” (p. 114). While the first connection—between a caring relationship and taking things personally—is claimed to be conceptual, the second—between taking things personally and the reactive attitudes—is a psychological generalization. An example is meant to support this: Learning from a trusted friend that her husband openly scoffs at her artistic ambitions behind her back, Caitlin’s feelings are badly hurt. . . . As her initial embarrassment subsides, the first stirrings of resentment set in. He has been so 18. See, for example, Pereboom (2001, 2009, 2014).
M oral R esponsibility, R eactive Attitudes
( 155 )
supportive of their friends’ creative endeavors, why not hers? When she thinks that this is her husband, her resentment shades into moral indignation. (p. 115)
We are then asked to imagine that Caitlin wants to disavow such feelings of resentment and indignation because she believes them to be morally objectionable. But it seems that she cannot do so without “emotionally divesting” in a way that not only sets aside the feelings of resentment and indignation, but also the vulnerability to the hurt feelings (p. 115). As part of his response, Pereboom (chapter 11, this volume) offers a kind of real case in which many are able to both take things personally and disavow and even eliminate reactive attitudes: that of parents of teenagers who in a normal developmental stage show disrespect. As he writes, Very commonly, teenagers go through a period when they have attitudes of disregard and disrespect for parents, expression of which can result in deeply hurt feelings. But often such expressions of disregard and disrespect do not occasion the parents’ resentment, but rather their disappointment and sadness. Although these emotions are not reactive attitudes, they are nevertheless manifestations of vulnerability on the part of the parent. Most significantly, they are also personal on Shabo’s characterization, since the attitudes the teenager has toward his parents matter to them in their own right, independently of the impact these attitudes might have on their interests. It’s not unusual for parents in such situations to be resentful as well, but often they are not. (p. 173)
Both kinds of cases—that of Caitlin and that of at least some parents of teenagers—are powerful, but each has limitations. As mentioned, even Shabo does not claim that eliminating or disavowing resentment in a personal relationship is impossible, only that it is difficult or impossible for most of us. But there is another limitation as well. The very case that Shabo uses to make the psychological generalization plausible has a particular feature that is not present in a number of cases of hurt feelings, a feature that may be quite relevant to determining how far our reaction to it should generalize. The case is one in which Caitlin is wronged by her husband and one in which we naturally think resentment warranted. It is one in which we ourselves may be indignant on her behalf—and this has nothing to do with our susceptibility to hurt feelings in the case; we don’t even know Caitlin’s husband. If we are trying to show that the special susceptibility to hurt feelings that characterizes personal relationships gives rise to resentment (for most of us), then it is worth seeking cases in which third parties
( 156 ) The Significance of Free Will
do not have such feelings of indignation. Take a case in which one’s friend postpones a dinner in order to help another friend. A third party might feel sorry for one but not indignant. Still, one might be hurt, even as one recognizes that one’s friend did nothing wrong and fails to feel resentment. Or take cases of jealousy, in which the object of a person’s love does nothing wrong. Or cases in which a child does not want to spend time with another child. These seem to be common sorts of cases in which one can have genuinely hurt feelings and yet not have the reactive attitudes. Still, it might be argued that one must be able to take all sorts of things personally, notably including cases in which one is wronged, to participate in a significant personal relationship, and in these cases in which one is wronged, it is impossible for most people to detach resentment from taking things personally. Note that this would be a substantially modified argument, but let us accept for the moment the premise that one must be susceptible to hurt feelings in all areas, including wrongs. At this point, we might return to the case of parents of teenagers, which would seem to be a kind of counterexample to the psychological generalization, now restricted to the cases in which we take wrongs personally, that we cannot (or almost never can) separate out resentment. Perhaps this kind of case shows that in a large number of real cases of genuine wrongs, people are perfectly well able to separate out resentment from taking things personally. But even if we are as optimistic as Pereboom about how commonly parents of teenagers are able to react without resentment and instead with disappointment and sadness, we can ask—as Pereboom does—whether this special ability is unique to the parental relationship. It also may be special to the relationship between parents and not fully mature adult children, in particular, insofar as there seems to be growing recognition of at least partial excuse of the behavior of adolescents.19 There may even be some reason to view these actions as not wrong, or not as violations of obligations for the very reason that teenage behavior is seen as a normal and healthy developmental stage. Thus, though the case provides some evidence that we are psychologically capable of setting aside resentment in personal relationships, how far it generalizes is also an open question.20 19. See Brink (2004), for example. 20. It is an interesting empirical question just how the process works for many parents of teenagers who behave in disrespectful ways. If (as seems sometimes to be the case) the recommended thought processes include thinking of the behavior as normal and healthy and ultimately in the service of establishing adolescents’ own identity, and so on, the fact that it is disrespectful becomes less salient. These sorts of thought processes would seem to have some effect not only in reducing resentment but also in taking the particular behavior personally (since the behavior is, in a sense, not about
M oral R esponsibility, R eactive Attitudes
( 157 )
As Pereboom (chapter 11, this volume) points out, there is reason to think that our psychological flexibility is greater than Strawson and Shabo believe because in many cases we have been able to disavow and even eliminate our reactive attitudes to criminals, the insane, and children, among others (p. 174). In addition, it simply seems to remain an open empirical question to what extent we could either eliminate or disavow the reactive attitudes, particularly in nonparental personal relationships. But this does not mean progress has not been made. To the contrary, we have been able to recognize the importance of a specific empirical question and a substantial start to the process of answering it. A more settled answer will contribute directly to the question of the value of free agency. Further, suppose, for the moment, that we had settled this question so that we found that we could make substantial changes in the way we conduct our personal relationships so as to set aside the reactive attitudes while maintaining the personal aspects of them. At this point, we would be left without an affirmative answer to the question of whether the reactive attitudes are valuable. We would know that they are not valuable insofar as they are required for maintaining personal relationships. But this does not show them not to be valuable. Of course, they might be valuable for some of the other reasons mentioned earlier—for example, they might be valuable because they contribute to social harmony and a reduction in violence.21 But we might be able to make further progress on the question of value even while continuing to consider personal relationships. To do this, turn to the second question set out earlier—assuming that we could maintain personal relationships in which we care and are disposed to take things personally, would we thereby lose something else of value? In particular, would we lose something of value in itself, namely a particular kind of personal relationship? The question is whether relationships in which we are susceptible to the reactive attitudes have a special value that isn’t equaled by the value of other sorts of personal ones.
the parent). Still, this is just one route to the reduction of resentment, and there may be others that would leave one’s “taking it personally” fully intact. Further, even if parents of teens with reduced resentment also take the relevant behavior “less personally,” it is a plausible hypothesis that this need entail nothing about a lower quality of the personal relationship. To the contrary, it is often recommended to parents that they take this approach. See, for example, Alfie Kohn (2005). 21. See, for example, Shaun Nichols’ (2007) argument for this empirical claim and responses by Pereboom (2009, 2014). This can be seen as a kind of parallel to the debate described in the text that centers on the connection between the reactive attitudes and personal relationships. And it, too, already contains appeals on both sides to empirical evidence of a great variety.
( 158 ) The Significance of Free Will
Here is one way to argue for an affirmative answer. The reactive attitudes are our way of making moral (and perhaps other) demands on ourselves and each other. Through the reactive attitudes we hold each other to our obligations to treat each other in certain ways. Some of these obligations seem to be partly defining of particular relationships—for instance, friends have special obligations to friends, parents have special obligations to their children. (This is, of course, consistent with our all having obligations to all fellow humans and even to all sentient creatures.) The reactive attitudes embody the demand that others (as well as ourselves) fulfill these obligations, or the holding of others to the standards defining of the relationship.22 Relationships in which we cannot appropriately hold others to standards (but can only encourage or express disappointment about their failures to meet them) and in which we do not live up to or fall short of meeting demands might be thought to be missing something of importance. It is a truism that it is a gift and a sign of respect to hold others to standards, and it is thought to be an admirable trait when we do so for ourselves. Not all truisms are true, of course, but there is something appealing in the idea that there is special value in relationships in which we do so. To elaborate, it is helpful to note that the idea of “mutual demand” is here to be contrasted with the idea of “expectations” in a sense of “prediction.” Two friends might in fact let each other down a fair portion of the time, but it might be argued that part of what makes them friends rather than two people who happen to care about each other is that they have expectations of each other in a sense distinct from that of prediction. If this is correct, then friendship itself would seem to require the holding to standards or the making of demands. But even if we are more expansive in our understanding of friendship, so that even two people who care about each other, without holding each other to the standards of friendship, still count as friends, the very recognition of and holding to standards seems to mark a real distinction between two kinds of relationships. Thus, a weaker conclusion is not that friendship itself requires the holding to mutual standards, but that there is a kind of friendship that requires it, and that kind of friendship has a special value.23 22. See, for example, Strawson (1962). 23. A closely related argument, that in one way is less and in another more difficult to defend than the one in the text, is one that begins with a distinct premise about the nature of friendship: (1) Friendship necessarily generates special obligations or duties; (2) special obligations, being obligations, require the freedom to meet them. Therefore, (3) friendship requires a certain kind of free agency. This special-obligations-based argument contrasts with the one in the text: (1') friendship (or a special kind of friendship) entails the mutual recognition and appropriate holding to standards, (2') appropriately holding to standards entails free agency, (3') friendship (or a special kind of
M oral R esponsibility, R eactive Attitudes
( 159 )
It is important to emphasize that the value of being disposed to the reactive attitudes is consistent with it being the case that relationships that contain less actual resentment and indignation are better than those with more. And it is consistent with it being the case that relationships that contain more positive and less negative ones are better. The idea is simply that the disposition and recognition of a disposition to such attitudes marks a particularly valuable kind of relationship. A full evaluation of the approach sketched here would require examining more closely each of the two main premises in the larger reasoning: that the reactive attitudes are the only, or best, way to hold others to demands and that holding others or oneself to demands is of any special inherent value. Either might be questioned. Interestingly, though, in the context of the larger project of exploring the value of free agency, it would not be enough for an opponent to simply undermine the first premise. For if the holding to demands itself presupposes free agency, then even if we could show the particular reactive attitudes unnecessary as a vehicle for doing so, the larger point that something of significant value presupposes free agency, of which the reactive attitudes might be one vehicle, remains. Here I believe we can see at work a key insight of Kane’s—there are different kinds of relationships, and some may have a kind of value that at least in one aspect goes beyond that of others. Interestingly, the approach here diverges on two dimensions from Kane’s. First, Kane (1996) is primarily concerned to argue that freely chosen love, say, is superior, or valued more by many (p. 88).24 In contrast, the suggestion here is that relationships presuppose a kind of free agency via the disposition to the reactive attitudes that partly define the relationships. This view is perfectly consistent with the idea that the feelings and commitments of friendship and loving relationships are not themselves undertaken freely.25 Second, where Kane describes the increase in value in personal relationships as depending on indeterministic origination, this approach is neutral as to what the friendship) requires free agency. Interestingly, while (1) may be more widely discussed as expressing a truth about friendship than (1'), I believe that (2') is more widely assumed than (2). A fuller comparison and evaluation of both arguments awaits another occasion. 24. See Pereboom (2009, pp. 184–187) for a direct response to this claim of Kane’s. 25. There is, however, an appealing view that the special obligations of particular relationships only exist where there is voluntary assumption of them (e.g., Jeske 2008). The relationship between voluntarism about special obligations and free agency is one I set aside here, however, in order to focus on what engaging in special relationships presupposes. It may be that there are multiple grounds for thinking that certain kinds of special relationships require free agency (e.g., in the free assumption of obligations necessary to relationships and in the appropriate holding to standards within relationships), but it is also possible to consider these grounds separately.
( 160 ) The Significance of Free Will
relevant free agency requires in the way of determinism or indeterminism. But the insight that there may be special forms of some of the items on the list that depend on a substantial notion of free agency is shared between these approaches. In sum, we have seen two ways of arguing that the reactive attitudes, and our disposition to them, are valuable, as well as questions that have been (and might be) raised about both. If either argument is ultimately successful, we will have shown not only the value of the reactive attitudes and moral responsibility but also, through them, the dependence of certain kinds of valuable relationships on a sort of free agency. At the same time, although we won’t have shown that the free agency in question must be libertarian, we will have shown that several interrelated and important things are indeed at stake in the question. Thus, we will have shown the significance of a robust notion of free agency and the significance of the question of whether that agency requires indeterminism.
4. Conclusion
We have seen two quite different approaches. Kane’s approach in one way goes further: by concluding that there is special significance to libertarian free will. But that significance is limited in another way: the significance depends on the reasons why people want it but not on the reasons it is worth wanting. The alternative is to start with things that seem to be “worth wanting,” even of the “exalted” kind that require free agency, explore whether they really are worth wanting (or worth wanting more than possible alternatives) and then see whether they require libertarian free agency in particular. I have sketched only one part of the second, “what is at stake” approach here, namely the exploration of the value of the reactive attitudes that presuppose moral responsibility and their connection to certain kinds of personal relationships. But I believe that it is enough to reveal that, despite a lack of consensus, we are far from a stalemate about questions of value. And as Kane’s own work on the nature of free agency perfectly exemplifies, despite a lack of consensus on this point, as well, we are engaged in a lively and fruitful debate.
•
Chapter 11
The Dialectic of Selfhood and the Significance of Free Will Derk Pereboom
W
hat is the significance of free will? Robert Kane (1996) specifies a number of goods for which free will could be necessary: “(1) genuine creativity; (2) autonomy (self-legislation) or self-creation; (3) true desert for one’s achievements; (4) moral responsibility in an ultimate sense; (5) being suitable objects of reactive attitudes such as admiration, gratitude, resentment, and indignation; (6) dignity or self-worth; (7) a true sense of individuality or uniqueness as a person; (8) life-hopes requiring an open future; (9) genuine (freely given) love and friendship between persons (or in religious contexts, freely given love toward God); and (10) the ability to say in the fullest sense that one acts of one’s own free will” (p. 80). The method he pursues has us first focus on the value of the indeterministic origination of one’s actions. The idea is that by ascertaining why indeterministic origination might be of value, we stand to discover the significance of free will. In one line of thought Kane pursues, we value indeterministic origination because it is a key component of our sense of self as distinct from the rest of the world. He makes this idea plausible by providing an account of how this sense of self develops. Infants take pleasure and interest in what they cause by an act of their will, which produces a sense of themselves as distinct from the external world. But they also discern that they are influenced by the objects they find in their environment that makes them realize that they are like things in the external world. This opposition between Thanks to Dana Nelkin and David Palmer for valuable comments and discussion.
( 162 ) The Significance of Free Will
distinctness from the external world and resembling and being immersed in it provokes a “spiritual crisis” (Kane 1996, p. 94), of which our concerns about causal determination are a component. This crisis is “a stage of the dialectic of selfhood” (p. 95), a process in which we aim to combine and balance our sense of ourselves as an independent source of activity with our realization that we are subject to causal influence from the world external to us. On Kane’s account, indeterministic origination is valuable insofar as it is a stable component of the resulting self-conception (p. 98). One can accept Kane’s dialectic of selfhood, and the specific story he tells, while arguing that the dialectical process doesn’t end where he leaves off. Or more modestly—and this is what I believe is most plausible—it might be that there are several more or less equally stable endpoints of this dialectical process, each of which is nonetheless unstable to a certain degree, and Kane’s position is just one of them. A case can be made that the synthesis at which he thinks the dialectic stops is at least somewhat unstable and, in particular, gives rise to a new conflict. A first salient problem is that when we affirm indeterministic origination as a stance that contrasts with absorption in the world, we are prone to exaggerate our independence from the world. One evident way we do this, no matter what one’s view about free will, is by discounting the extent to which our accomplishments are dependent on the help of others, auspicious external circumstances, and fortunate genetic endowment. When we laud successful entrepreneurs, we often disregard the contribution of the infrastructure of our market system which is dependent on government funding and management. Closer to home, we’re apt to exaggerate our own contribution to our academic achievements, where success is indebted to genetic endowment and one’s teachers. Moreover, our will and effort may be involved in studying existing contributions in the relevant field, but once that’s done, what we value most— coming up with an original idea—has a passive phenomenology. As in the Taoist element in Kane’s account of free will, here our contribution amounts to relaxing the mind so that it’s open to the reception of new ideas (Kane 1996, pp. 164–167). But even if the relaxation of mind is indeterministically and freely produced, the extent to which we tend to believe we deserve credit for these ideas far outstrips what can be accounted for by the effort involved in mastering existing contributions and relaxing one’s mind. Alternatively, then, we might think of the indeterministic-origins stage as a step along the way in the dialectical process of self-understanding. A subsequent stage should not amount to giving in to antithesis according to which our agency is submerged by the world’s contributions. Rather, it might represent a synthesis that involves seeing oneself as distinguished from external stimuli first of all by one’s capacity for reasons-responsive
Dialectic of S elfhood, S ignificance of F ree W ill
( 163 )
agency (Fischer and Ravizza 1998). As an agent, one has the capacity to be sensitive to and to act on the basis of one’s evaluation of reasons, and this sets one apart from the causal mechanisms in nature that are not mediated by reasons-responsiveness. It also sets one apart from others, whose agency is distinct from one’s own. To this we may add one’s capacity to relax one’s mind to wait for inspiration, and this is a further ability that sets one apart from common natural processes and from others who have this capacity but whose agency is distinct. We can also point to the emotional sensitivities that allow an agent to enjoy the kinds of personal relationships specific to human beings, sensitivities that we don’t find in external natural processes. Here too, one comes to understand that one’s emotional sensitivities are distinct from those of other agents. This proposal for a further movement in the dialectic of selfhood preserves the sense that we are separate from natural processes in a subtler and less dramatic way than we find in the affirmation of indeterministic origination. There may be several more or less stable syntheses that fit this description, and among them are various other compatibilist compromises. In the synthesis I propose, indeterminism may still have a role, but not in a way that generates the supposition that we deserve, in a fundamental sense, praise or credit for our achievements (Pereboom 2001, 2014). For example, when we relax our minds and allow ideas to occur to us as a result of an indeterministic process, we may be tempted to claim that we deserve, in the fundamental sense, credit for the result. But in actual fact it may be that something has happened in us that we can enjoy and appreciate, without the added thought that we basically deserve credit for it. Kane is right to say that this sense of basically deserved credit is hard to give up, but here too there is a stance intermediate between basic desert and mere mechanical process that can be embraced. At this point we can consider a challenge to this proposal in the form of a connection Kane proposes between indeterministic origination and our sense of objective worth. He presents this connection by way of an example (Kane 1996, p. 97). Alan, a despondent artist, has a wealthy friend who wants to lift his spirits. The friend buys Alan’s paintings under an assumed name for large sums of money. Alan believes that his paintings are being bought because of their artistic value and his despondency evaporates. He develops a sense of self-worth but, crucially, this is due to an illusion. Kane then has us compare this world to one in which Alan really is a great and successful artist. Supposing that Alan could stand outside both worlds, would he be indifferent about which world he inhabited? It’s plausible that he would think his life more valuable if he inhabited the second, despite the fact that he would experience the same degree of subjective satisfaction in each. Kane then claims that if we value objective worth, then we value indeterministic
( 164 ) The Significance of Free Will
origination because the latter requires us to view the universe from an objective perspective. If we care about our worthiness independently of subjective satisfaction, we would want to be undetermined originators. But how, more specifically, might indeterministic origination and objective worth be linked? In this type of context, Ted Honderich (1988) asks whether without indeterministic origination—and, we’re assuming, therefore without basic desert praise- and credit-worthiness—we could legitimately retain a sense of achievement for successes that make our lives fulfilled, happy, or worthwhile. Let’s keep in mind that it’s a sense of achievement connected to objective worth that is the quarry, not a sense of achievement that deluded Alan might still have. It might be argued that without indeterministic origination there would be no accomplishment that is objectively valuable for the agent, for it couldn’t be objectively valuable for an agent if his achievement was inevitable due to being caused by factors beyond his control. Otherwise, any worth the accomplishment might have passes backwards along the deterministic causal chain, eventually to some factor external to the agent. However, objectively valuable achievement would not seem to be as closely tied to indeterministic origination as this objection would have it. If an agent hopes to achieve success in a project she undertakes, and if she accomplishes what she hoped for by a reasons-responsive process, intuitively this outcome would be an objectively valuable achievement of hers even if she is not its indeterministic originator. At the same time, the degree to which it is objectively valuable might be diminished relative to a world in which she is its indeterministic originator. My sense is that the absence of indeterministic origination reduces our objective value relative to a world in which we have this capacity, but that we would retain enough objective worth for our lives to be meaningful and fulfilling. By contrast, Saul Smilansky (1997) contends that although the absence of indeterministic origination allows for a limited foundation of the sense of self-worth that derives from achievement and virtue, the free will skeptic’s perspective can nevertheless be “extremely damaging to our view of ourselves, to our sense of achievement, worth, and self-respect” (p. 94), especially when it comes to achievement in the formation of one’s own moral character. In response to this concern, he argues that it would be best for us to foster the illusion that we have free will (Smilansky 1997, 2000). Smilansky’s solution highlights my main worry about Kane’s stopping point in the dialectic of self-formation. While supposing indeterministic origination may give us a strong sense of objective self-worth, this supposition itself is threatened by the prospect of illusion, since any specific account of such origination may not be sufficiently empirically plausible or might not
Dialectic of S elfhood, S ignificance of F ree W ill
( 165 )
deliver the requisite control in action (Pereboom 2001, pp. 38–68; 2014). This threat of illusion stands to undermine the conviction of objective value it is apt to generate. By contrast, the alternative conception of achievement without indeterministic origination places its objective worth, albeit diminished, on an epistemically more secure foundation. To address Smilansky’s specific concern, in the typical case an agent’s moral character was formed to some significant degree by her upbringing. Parents typically regard themselves as having failed in raising their children if they turn out with immoral dispositions, and parents often take great care to bring their children up to prevent such a result. Accordingly, people often come to believe that they have the good moral character they do largely because they were raised with love and skill. But those who come to believe this about themselves seldom experience a dispiriting sense of diminished objective worth because of it. By contrast, they often come to feel more fortunate and thankful. This shows, to my mind, that for some sorts of achievements or apparent achievements, giving up indeterministic origination does not incur a loss.
1. Love, Personal Relationships, and Reactive Attitudes
Kane (1996) suggests that denying undetermined origination would subvert love, at least love between mature persons (pp. 86–90). Let’s assess the suggested challenge, and consider how the alternative synthesis in the dialectic of selfhood might accommodate it (Pereboom 1995, 2001, 2009, 2014). A first thought might be that love that is freely willed is clearly valuable, and that love is not worth nearly as much if it is not freely willed, and that the free will in question involves indeterministic origination. However, against this, note that parents’ love for their children—a paradigmatic sort of love—is often produced independently of the parents’ will. Kane endorses this last claim and a similar view about romantic love, but he nevertheless argues that a certain valuable type of love we want would be endangered if someone who loved in this way knew that there were factors beyond her control that causally determined her to have it: There is a kind of love we desire from others—parents, children (when they are old enough), spouses, lovers and friends—whose significance is diminished . . . by the thought that they are determined to love us entirely by instinct or circumstances beyond their control or not entirely up to them. . . . To be loved by others in this desired sense requires that the ultimate source of others’ love lies in their own wills. (Kane 1996, p. 88)
( 166 ) The Significance of Free Will
But setting aside free will for a moment, in what sorts of cases does the will plausibly play any role in generating love for another? When the intensity of a personal relationship is waning, the partners will at times decide to try to make it succeed and to attempt to restore the intimacy they once enjoyed. When a marriage is arranged by parents, the partners may choose to take steps to come to love each other. But while in such situations we might desire that the other make a decision to love, we don’t seem to have reason to want the decision to be freely willed. A decision to love on the part of another might greatly enhance one’s personal life, but it is unclear what value the decision’s being freely willed would add. Moreover, while in circumstances of these sorts we might desire that the other make a decision to love, we would typically prefer the situation in which the love did not involve a decision. This is so not only for romantically intimate relationships but also for friendships and for relationships between parents and children. Perhaps the will can have a key role in maintaining love over an extended period. Søren Kierkegaard (1843) suggests that a marital relationship ideally involves a commitment that is continuously voluntarily renewed. A relationship with this sort of voluntary aspect might be highly desirable. Nevertheless, it is difficult to see what value might be added by these continuously repeated decisions being freely willed, by contrast with being a voluntary expression of what an agent truly cares about. Thus although one might initially find it intuitive that love that is freely willed is valuable, it’s difficult to see precisely how free will might have a desirable role in producing, maintaining, or enhancing love. But Kane’s core concern is that we would think that love that is causally determined by factors beyond one’s control has significantly less value that love that is indeterministically originated. For a striking case, one that John Milton addresses in Paradise Lost, we might ask whether one’s love for God would be valuable if it were causally determined by God. Milton’s God delivers a decisively negative answer: Of true allegiance, constant Faith or Love / Where only what they needs must do, appeared / Not what they would? What praise could they receive?/ What pleasure I from such obedience paid / When Will and Reason (Reason also is choice) / Useless and vain, of freedom both despoiled / Made passive both, had served necessity / Not me. (Book III, lines 104–111)1 1. John Milton 1667. Paradise Lost, ed. Gordon Teskey. New York: W. W. Norton & Company, 2005.
Dialectic of S elfhood, S ignificance of F ree W ill
( 167 )
In response, even in the extreme case in which one is causally determined to love by the person one loves, whether the causal determination would undermine the value of the love would depend on particular features of the case. Imagine that Maddy causally determines you to love her by manipulating your brain so that you are oblivious to her flaws of character and by adding a love potion into your drink. This would clearly jeopardize the value of the resulting love. Now imagine instead that you have a self-destructive tendency to love people who are apt to hurt you and not to love those who would be good to you. This is so partly because you have a tendency to overlook people’s valuable characteristics, such as kindness and concern for the well-being of others. Now suppose that Violet slips a drug into your drink that eliminates this disposition, as a result of which you are now able to appreciate her good qualities and you are causally determined to love her. How bad would this be? Perhaps what would undermine the value of love in these sorts of cases is not being causally determined to love by the other party per se, but rather how one is causally determined. And if love can be valuable even when it is causally determined by the other, it can plausibly be valuable when it is causally determined by impersonal factors. My sense is that when it comes to love in personal relationships, so far we haven’t seen that rejecting indeterministic origination results in giving up significant objective value. Dana Nelkin (chapter 10, this volume) proposes a different way of connecting the value of personal relationships with indeterministic origination. In her account, various demands one makes of others and of oneself are essential to such relationships. Crucially, on her proposal the way in which we human beings are sensitive to such demands is by reactive attitudes such as moral resentment, indignation, guilt and gratitude: Through the reactive attitudes we hold each other to our obligations to treat each other in certain ways. Some of these obligations seem to be partly defining of particular relationships—for instance, friends have special obligations to friends, parents have special obligations to their children. (This is, of course, consistent with our all having obligations to all fellow humans and even to all sentient creatures.) The reactive attitudes embody the demand that others (as well as ourselves) fulfill these obligations, or the holding of others to the standards defining of the relationship. Relationships in which we cannot appropriately hold others to standards (but can only encourage or express disappointment about their failures to meet them) and in which we do not live up to or fall short of meeting demands might be thought to be missing something of importance. It is a truism that it is a gift and a sign of respect to hold others to standards, and it is thought to be an admirable trait when we do so for ourselves. (p. 158)
( 168 ) The Significance of Free Will
Thus certain kinds of relationships confer objective value on our lives. Having such relationships requires making and holding each other to certain kinds of obligation-involving demands. Holding each other to such demands in turn requires susceptibility to the reactive attitudes, which presupposes that the potential target of the reactive attitude is free. Kane would at this point claim that this freedom requires indeterministic origination, although Nelkin would disagree. In Nelkin’s view there is another, more direct, route to free will. The central sorts of demands, by her characterization, communicate moral obligations, and for an agent to be subject to such an obligation, it must be the case that she can act so as to satisfy it: If S ought to do A, then S can do A. If S ought not do A, then S can refrain from doing A. So suppose Bill does not meet a demand that he not be abusive. For the demand to apply, he must be so morally obligated, which entails that he could have acted so as to meet the obligation. Thus for this demand to be in place, he must have had the free will to refrain from acting as he does, which on an incompatibilist conception such as Kane’s, requires that he was the indeterministic originator of his abusive behavior. Nelkin (2011) invokes a compatibilist notion of free will at this point, but here I agree with Kane that the sort of free will in play must involve indeterministic origination (Pereboom 2013, 2014). But then if the dialectic of selfhood unfolds as I am envisioning, and we come to reject indeterministic origination in favor of reasons-responsiveness and other determinismfriendly capacities, we will not see ourselves as the sorts of beings for whom demands of moral obligation are in place. But still, it would seem that Nelkin is right to claim that certain especially significant kinds of relationships essentially involve demands. What I would need is a notion of demand that does not require the ability to do otherwise and sensitivity to which does not require the reactive attitudes. I take these requirements in turn. The kinds of demands at issue, we can agree, can be stated as “ought” claims. It would initially appear that all claims to the effect that an agent ought to behave in a certain morally recommended way would count as judgments of moral obligation. But I want to deny this supposition and instead to defend, following Kate Manne (2011), the legitimacy of a kind of moral “ought” claim that does not invoke moral obligation but yet qualifies as a kind of demand (Pereboom 2013, 2014). “Ought” has a variety of correct uses, as C. D. Broad (1952) contended, and some are not linked to the sort of “ought” implies “can” requirement that is plausibly a condition of
Dialectic of S elfhood, S ignificance of F ree W ill
( 169 )
judgments of moral obligation. For a case in point, Ruth Marcus (1966), Lloyd Humberstone (1971), and Gilbert Harman (1977) distinguish an “ought” that applies to action from one that applies to states. An “ought to do,” Harman specifies, “implies that an agent has a reason to perform an action, while an “ought to be” evaluates a state of affairs and does not by itself imply that any particular agent has a reason to contribute to bringing about that state of affairs” (1977, p. 87; cf. Hobbs 2012). Nelkin (2011) argues that that “ought” judgments that specify what an agent ought to do are essentially action directed, so that if “S ought not do A” is true, then as a matter of the meaning of “ought” propositions, or of the essential nature of obligation, S is thereby directed to refraining from A, and this entails that S can refrain from A (p. 111). We could say, then, that while an “ought to be” is an “ought” of axiological evaluation, or sometimes of axiological ideality, which does not (at least directly) entail a “can” claim, an “ought to do” expresses a demand that an agent in a particular circumstance perform some action, which does entail that the agent can comply. One might call this second type an “ought” of specific action demand. It’s my sense that if causal determinism precludes alternative possibilities for motivation and for action, a central sense of “ought to do” will be compromised. If you know that it was not possible for Bill to have been motivated to perform some morally exemplary action at some particular time, it would be unfair and, I think, mistaken for you to claim that he was morally obligated to perform that action at that time. However, it might well not be unfair for you to recommend to him that he perform an action of that type at some future time, supposing it is epistemically open that he will develop the required motivation by then and in particular if it’s reasonable to believe in advance that so recommending would contribute to producing that motivation. You might make this recommendation by telling Bill he ought to perform such an action at some point in the future, and you would do so legitimately and without making any kind of error. The sense of “ought” invoked here would thus be distinct from the “ought” of specific action demand. I propose that given determinism and that determinism precludes alternatives, when one tells an agent that he ought to refrain from performing an action of some kind in the future, the “ought” of axiological evaluation, by contrast with the “ought” of specific action demand, is legitimately invoked. Judgments that feature this kind of “ought” recommend to an agent as morally valuable a state of affairs in which he so refrains. Accordingly, we might call this the “ought” of axiological recommendation. Unlike the moral “ought” of specific action demand, it is not an “ought” of moral obligation. Like the “ought” of specific action demand, the “ought” of axiological recommendation
( 170 ) The Significance of Free Will
essentially concerns agents and actions they might consider in deliberation. But as for all claims about what ought to be, this use of “ought” should not be understood as presupposing a route accessible to an agent, by way of reasons for action, to acting or refraining from acting in the relevant way (Hobbs 2012). One might be unsure about whether such a route is in fact accessible, while one’s making an “ought” judgment of this sort is nevertheless legitimate. Thus this type of demand, by contrast with the demand of moral obligation, does not presuppose the freedom to act and to refrain, and thus it does not invoke, even prima facie, indeterministic origination. But is it strong enough a notion of demand to do the work required in personal relationships? Some demands reflect only what we’d like the other to do, and in the typical case they are plausibly classified as axiological recommendations. But others are demands that someone not perform actions that are intuitively morally wrong. When Bill is repeatedly abusive, one would want to protest his behavior by demanding that it cease and that he extinguish the disposition to behave this way (Hieronymi 2001, p. 546). Is such a demand compatible with the absence of indeterministic origination? One would want to say that the demand is appropriate because the behavior is morally wrong. But Ishtiyaque Haji (1998, 2012) contends that because of the connection between moral obligation and moral wrongness, the threat posed by causal determination to judgments of obligation extends to judgments of moral wrongness. His argument essentially involves the endorsement of the following principle: S has a moral obligation to perform [not to perform] A if and only if it is morally wrong for S not to perform [to perform] A. If this principle is true, then if judgments of moral obligation are ruled out by determinism, judgments of moral wrongness will be ruled out as well. While this biconditional principle may be attractive considered in the abstract, my sense is that it captures only one sense or aspect of the complex concept of moral wrongness. This can be made plausible by reflecting on the fact that while one half of this biconditional If S has a moral obligation to perform [not to perform] A then it is morally wrong for S not to perform [to perform] A, is clearly true, the other half, that is, If it is morally wrong for S not to perform [to perform] A, then S has a moral obligation to perform [not to perform] A,
Dialectic of S elfhood, S ignificance of F ree W ill
( 171 )
seems less secure (Pereboom 2001). I can’t think of a case in which, intuitively, an agent has a moral obligation not to perform an action and it is not morally wrong for the agent to perform it. But there are situations in which it seems quite credible that it is morally wrong for the agent to perform the action, but not that the agent has a moral obligation not to perform it. For example, suppose that a serial killer could not have done otherwise due to his psychological disorder. Even if we’d deny on the basis of the “ought” implies “can” principle that he was obligated to refrain from killing, it remains plausible that what he did was morally wrong. An alternative nondeontological notion of moral wrongness, one that isn’t biconditionally linked to obligation, accommodates this intuition. Alastair Norcross (2006) suggests a purely axiological ethics, the core of which involves specifying for each action-relevant situation the possible options for acting ranked in order of value realized. An option for acting can be classified as morally wrong when its value is low enough in this ranking, in its context, for it to be prima facie morally justified (on whichever normative ethical theory is endorsed) for a relevantly situated interlocutor, such as a partner in a personal relationship, to protest the action. True, this proposal does not characterize wrongness independently of when it is morally appropriate to protest the action, and thus it cannot ground the appropriateness of protest in wrongness. For this reason it does not meet one significant intuition we have about the relationship between wrongness and the appropriateness of protest. But this proposal does have the result that it is appropriate to protest only when acting for some reason is morally wrong, and in this respect it satisfies another intuition we have about this relationship. Moreover, it won’t be that it’s appropriate to protest just in case the action is morally wrong, since the prima facie appropriateness may be overridden by, for example, the difficulty the other may have in complying with the protest. So the link between wrongness and appropriateness of protest won’t be trivial.2 One might think that on this proposal no demands will have the edge that demands of moral obligation have. This seems right and arguably counts as a loss. However, in certain kinds of morally charged situations one might invoke demands of a hypothetical form that have considerable force and clearly don’t presuppose indeterministic origination. For example, one might say to Bill: “If you don’t stop abusing me, I’m leaving you.” The legitimacy of this demand can be grounded in the right to self-protection, as well as the good of potential moral reform. But suppose Bill can’t stop being abusive. The appropriateness of the hypothetical demand is not thereby contravened. 2. Thanks to Dana Nelkin for discussion of this notion of wrongness.
( 172 ) The Significance of Free Will 2. The Reactive Attitudes and the Significance of Free Will
Nelkin argues that the reactive attitudes are part of the human way of being sensitive to whether demands in relationships are being met, and that an agent’s being an appropriate target of a reactive attitude presupposes that he has free will. Here again she regards a compatibilist notion of free will as sufficient. On the view I defend (Pereboom 1995, 2001, 2014), the legitimacy of the expression of reactive attitudes such as indignation and resentment requires that their targets have a kind control in action, indeterministic origination, which is not unlikely to be illusory. So here as well I would want to explore the accommodation that the further step in the dialectic of selfhood might make. One might object that moral resentment and indignation are crucial to effective communication that demands are not being met, and if we dispelled or modified these attitudes, an important component of human relationships would be lost. Are there emotions other than the free will–presupposing reactive attitudes that can do the requisite work in relationships? When we are wronged in our relationships there are other such emotions present or available whose expression can also convey the relevant information. They include feeling hurt or shocked or disappointed about what the offending agent has done and moral sadness or sorrow and concern for him. These emotions, unlike the reactive attitudes, don’t presuppose free will. It is reasonable to be disappointed by or be saddened by events or states even if they are not caused by agents at all, and thus even if no agent has a role in producing them. So it could be reasonable to be disappointed or saddened by actions even if they are causally determined by factors beyond the agent’s control. An important objection to this proposal is advanced by Seth Shabo (2012) who also argues, in a way that complements Nelkin’s line of thought, that the reactive attitudes are required for good personal relationships. In his view, our susceptibility to resentment and indignation is a consequence of the sort of vulnerability that is a feature of the best kinds of close personal relationships. On Shabo’s account, the specifically personal aspect of such a relationship requires vulnerability to moral resentment. Conditioning ourselves to be immune to resentment in personal relationships threatens to eradicate this personal aspect. As he sees it, to care about someone’s attitudes personally is to care about those attitudes in their own right or else as expressions of her deeper attitudes. In particular, it is to care about the morally significant attitudes she has toward oneself. If one has this sort of care in a close relationship, one will naturally be resentful when one is made the target of an expression of a disrespectful or demeaning attitude
Dialectic of S elfhood, S ignificance of F ree W ill
( 173 )
(Shabo 2012, p. 112). Shabo remarks: “[I]t is in feelings of resentment that our susceptibility to take disregard or ill will personally is characteristically manifested” (p. 113). His crucial claim is that in an intimate personal relationship that features reciprocal love, when the other expresses a disrespectful or demeaning attitude, the resentment that one naturally feels as a result cannot reliably be extinguished or disavowed while retaining the personal nature of the caring involved in the relationship (pp. 113–114). First of all, I agree that we very generally do not forgo or disavow resentment in such contexts. But this does not all by itself indicate that we cannot take measures that would result in the substitution of disappointment and sadness for resentment and in these contexts express such attitudes instead. I think it’s mistaken to claim that we are not now capable of disavowing resentment in a way that can have an effect on which expressions we regard as justified and even on which responses we actually have. However, the question Shabo is asking is whether we are capable of such substitution and disavowal while preserving the specifically personal nature of our caring about the other’s attitudes. It’s instructive to examine a kind of personal relationship in which it is not unusual for us to respond to expressions of disregard and disrespect with attitudes that are personal but not reactive, even though it is not the kind of relationship on which Shabo focuses (Pereboom 2014). Very commonly, teenagers go through a period when they have attitudes of disregard and disrespect for parents, expression of which can result in deeply hurt feelings. But often such expressions of disregard and disrespect do not occasion the parents’ resentment, but rather their disappointment and sadness. Although these emotions are not reactive attitudes, they are nevertheless manifestations of vulnerability on the part of the parent. Most significantly, they are also personal on Shabo’s characterization, since the attitudes the teenager has toward his parents matter to them in their own right, independently of the impact these attitudes might have on their interests. It’s not unusual for parents in such situations to be resentful as well, but often they are not. Thus certain relationships that are very important to us feature care for the other’s attitudes that is personal but without susceptibility to resentment. To these claims one might object that the absence of susceptibility to resentment is an artifact of the specific nature of the parental relationship, which does not transfer on a significant scale to personal relationships between adult peers. However, it’s my sense that many such adult mutual relationships are free from resentment and feature disappointment and sadness instead. An evolution to this alternative kind of emotional profile may be an option for those of us who currently are susceptible to resentful
( 174 ) The Significance of Free Will
reactions. More generally, the last two hundred years have witnessed very significant changes in attitudes toward criminals, the insane, and children, and for this reason it cannot plausibly be argued that large-scale attitudinal change is impossible for us. Supposing such change is unlikely, disavowal remains as an option, and this can be achieved without compromising the personal nature of the care one has for the quality of the other’s will. When one is hurt by an expression of the attitudes of the other, one will not only feel resentment but also disappointment and sadness. Suppose that for an agent these three attitudes are not, as a matter of psychological fact, separable. Each of these attitudes is personal in Shabo’s sense. But only resentment is plausibly linked with the presupposition that its target has free will. Thus disavowing resentment because one rejects the free will at issue need not involve also disavowing the personal nature of the set of attitudes one has, since by setting aside resentment in this way one need not at the same time renounce disappointment and sadness. Shabo argues that we often can’t forgo or disavow resentment without forgoing or disavowing the hurt feelings that occasion it. But all of these attitudes are occasioned by these hurt feelings, and thus disavowing resentment but not disappointment and sadness need not involve renouncing the hurt feelings. We might acknowledge the legitimacy of the hurt feelings and of the disappointment and sadness that result while distancing ourselves from resentment. Disavowal and distancing of this kind are significant not only in their own right but also because they issue in rejection of reprisal grounded in resentment. With this added specification, the proposed further synthesis in the dialectic of selfhood has the drawback of counting as illegitimate reactive attitudes that come naturally to us, which communicate obligations that are not satisfied. But such reactive attitudes and moral obligations presuppose free will, and Kane and I agree that the relevant kind of free will involves indeterministic origination. The serious possibility that indeterministic origination is illusory occasions the motivation for moving to the new synthesis. Instead of demands that involve moral obligation, we substitute demands of axiological recommendation or, in extreme cases, hypothetical demands that contain threats. Disavowing or forgoing resentment and indignation, we express our personal care for the attitudes of the other in sadness and disappointment. This picture requires modification of our natural emotional life, and this may be difficult to do, but this aspect of the new synthesis arguably does not result in a very significant loss.
Dialectic of S elfhood, S ignificance of F ree W ill
( 175 )
3. Final Words
In the view I’ve developed, the dialectic of selfhood that Kane invokes evolves so as to reduce the threat of illusion posed by the supposition of indeterministic origination. Among the costs are a somewhat diminished sense of objective worth and the loss of the legitimacy of certain natural emotional predispositions in personal relationships. Perhaps, as Spinoza (1677) argued, in addition to the freedom from illusion, there is also the benefit of enhanced equanimity in the face of the hardships of this world (Pereboom 2001, 2014). The synthesis that results, I think, is fairly stable. Given our nature, it’s not perfectly stable, for the reason that in certain respects it will prove difficult to realize in our lives. But I believe there is no synthesis that is free from illusion and perfectly stable, and I contend that the one I propose does as well as any in achieving a balance of practical workability and facing the truth.
Pa r t V
••• Kane’s Reply
•
Chapter 12
New Arguments in Debates on Libertarian Free Will: Responses to Contributors Robert Kane
D
iscussions over the past several decades with many of the contributors to this volume have been invaluable in refining my views. To explain the organization of my responses, some historical remarks will help. When I began thinking about free will issues in the 1960s, the landscape of free will debates was simpler. The assumption was that if you have scientific leanings, you would be a compatibilist about free will, unless you denied free will altogether, as did hard determinists. If, by contrast, you were a libertarian, believing in a free will that was incompatible with determinism, it was assumed you must appeal to some kinds of obscure or mysterious forms of agency to make sense of it—uncaused causes, noumenal selves, nonevent agent causes, prime movers unmoved, or other examples of what P. F. Strawson (1962) called the “panicky metaphysics” of libertarianism. My project was to see if one could give an intelligible account of libertarian free will without appealing to any such unusual forms of agency and to consider in the process how such a view might be reconciled with modern science. It was a more complex task than I imagined. For it would require rethinking every facet of the free will problem. I am grateful to the contributors to this volume and to David Palmer for putting it all together.
( 180 ) Kane’s Reply
That especially meant rethinking the three central questions discussed in preceding essays and which I address here: the Compatibility Question (is free will compatible or incompatible with determinism?), the Intelligibility Question (can one make sense of an incompatibilist free will requiring indeterminism without reducing it to mere chance or mystery?), and the Significance Question (why do we want free will and what sort of free will is “worth wanting”?).
1. The Compatibility Question: McKenna, Haji, Widerker and Schnall, and Clarke
1.1 McKenna and UR As Michael McKenna explains in his characteristically clear and insightful essay, one of several distinctive features of my approach to the Compatibility Question is the role played in it by the condition of ultimate responsibility (UR). Focusing on this condition was a departure from much historical (and recent) debate about compatibility, which has turned on whether determinism was compatible with “the condition of alternative possibilities” (which I call AP)—the requirement that the free agent “could have done otherwise.” As is well known, most arguments for incompatibilism, such as the Con sequence Argument, which appealed to this condition AP, frequently tended to stalemate over differing interpretations of “can,” “power,” “ability,” and “could have done otherwise.” They have done so, I argued, for good reasons having to do with different possible meanings of freedom. The result I came to believe, as McKenna notes, was that, while alternative possibilities were important, a narrow focus on them alone would not allow one to resolve issues surrounding the Compatibility Question. A closer look at the history of the free will problem suggested to me that there was another criterion fueling incompatibilist intuitions even more important than AP, a condition I called UR (Kane 1985, 1989, 1996, 2002b, 2005, 2008). McKenna insightfully describes the technicalities involved in defining this condition; and I refer you to his essay for these details. But the basic idea is this: to be ultimately responsible for an action, an agent must be responsible for anything that is a sufficient reason (condition, cause or motive) for the action’s occurring. If, for example, a choice issues from, and can be sufficiently explained by, an agent’s character and motives (together with background conditions), then to be ultimately responsible for the choice, the agent must be responsible to some degree for the c haracter
L ibertarian F ree W ill
( 181 )
or motives he or she now has.1 Compare Aristotle’s claim that if a man is responsible for wicked acts that flow from his character, he must at some time in the past have been responsible for forming the wicked character from which these acts flow. As McKenna points out, this condition UR does not require that we could have done otherwise (AP) for every act done of our own free wills. But it does require that we could have done otherwise with respect to some acts in our life histories by which we formed our present wills. I call these “selfforming actions,” or SFAs, and they are also informatively discussed in McKenna’s essay (sections 2–4). The idea behind them is that we often act from a will (character, motives and purposes) already formed, but it is “our own free will” by virtue of the fact that we formed it by other choices or actions in the past (SFAs) for which we could have done otherwise. If this were not so, I argue, there is nothing we could have ever done differently in our entire lifetimes to make ourselves different than we are. Free will is thus not merely about free action, but about self-formation. Were we ultimately responsible to some degree for having the wills we do have, or can the sources of our wills be completely traced backward to something over which we had no control, such as Fate, God, heredity, environment, social conditioning, and so on? Therein lies the traditional problem of free will. In the light of these implications, McKenna argues that compatibilists, like him, must reject UR. For if the case for incompatibility cannot be made on AP alone, he notes, it could be made if UR were added. If agents must be responsible to some degree for anything that is a sufficient cause or motive for their actions, an impossible infinite regress of past actions would be required unless some actions in the agent’s life history (SFAs) did not have sufficient causes or motives (and hence were undetermined). But while McKenna thinks compatibilists must reject UR, he also interestingly argues that it “would be a mistake” for compatibilists to “deny that free will requires” some notion of “ultimate sourcehood” altogether: Compatibilists would be ill advised to avoid a battle over ultimacy. An adequate theory of free agency should be able to make sense of a person’s shaping her life. . . . Kane’s lovely picture of free will, of the historical relation between FWAs
1. Other incompatibilists who later also came to stress the importance of a condition like my UR for free will came to be called “source incompatibilists.” Some of them argued that APs were not required at all, though I never shared that view. Timpe (2013) helpfully distinguishes between “narrow” and “wide” source incompatibilists. The former believe that an ultimacy or source condition like UR was all that was required for free will, whereas wide source incompatibilists hold that both UR and AP are required. I am of this wide variety, for reasons we’ll see.
( 182 ) Kane’s Reply [acts done “of our own free will”] and . . . SFAs [self-forming actions] in the shaping of a life, is not something that compatibilists should be prepared to give up. We want to capture the thought that for some features of our own agency we are, after all, their source. Thus did we freely shape ourselves.” (p. 84)2
What compatibilists must do then, he argues, is resist the move from ultimacy to the denial of determinism. They must argue that “an agent can be the ultimate source of her will and acts even if there is a deterministic explanation for them that traces back to factors for which she is not ultimately responsible” (p. 85). To do this, McKenna suggests, compatibilists must appeal to the fact that accounts of origins or ultimate sources in science and everyday life are a “contextual” matter, dependent on one’s explanatory interests. Thus, we speak of the ultimate source of Perrier drinking water as a spring in the south of France. If we were geologists, we might trace the origins of the spring still further back to the formation of the earth’s crust. But “[i]n ordinary contexts,” he says, “it would never dawn on us to think that whether the water in our glass really originated in France turned on whether determinism was true or not” (p. 85). True enough. But in such contexts, we are not concerned with the moral responsibility of the spring or other source in producing what it did produce; and that is crucial. For the specific context of ultimate responsibility (and free will) is that of a person or agent with a will (character, motives, and purposes) who can in principle be held praiseworthy or blameworthy for being the sort of person she is, with the will she has, which led to her doing what she did. And in that context, the context of responsible agency, it does matter whether the agent is personally responsible to some degree by virtue of some past choices or actions for becoming the sort of person she now is with the will she has, or whether the formation of that will is entirely traceable to factors the origins of which she had no role in producing or bringing about. UR expresses this idea. It entails that for a person (or agent) to be personally responsible for being the sort of person she is at any time t' with the specific will she has, something the person voluntarily and intentionally did or omitted must have played a role in producing or bringing it about that it was settled that she would be the sort of person she now is. (Where, formally stated, it is “settled” after a time t that something p will be case at
2. All page numbers in this essay without specified dates refer to pages of this volume.
L ibertarian F ree W ill
( 183 )
a time t' (t