Deontic Morality and Control

This page intentionally left blank

Deontic Morality and Control

This book addresses a dilemma concerning freedom and moral obligation (obligation, right, and wrong). If determinism is true, then no one has control over one’s actions. If indeterminism is true, then no one has control over one’s actions. But it is morally obligatory, right or wrong, for one to perform some action only if one has control over it. Hence, no one ever performs an action that is morally obligatory, right or wrong. The author defends the view that this dilemma can be evaded but not in a way traditional compatibilists about freedom and moral responsibility will find congenial. For moral obligation is indeed incompatible with determinism but not with indeterminism. He concludes with an argument to the effect that, if determinism is true and no action is morally obligatory, right or wrong, then our world would be considerably morally impoverished as several sorts of moral appraisal would be unjustified. Ishtiyaque Haji is Associate Professor of Philosophy at the University of Minnesota, Morris.

cambridge studies in philosophy General editor

ernest sosa (Brown University)

Advisory editors: jonathan dancy (University of Reading) john haldane (University of St. Andrews) gilbert harman (Princeton University) frank jackson (Australian National University) william g. lycan (University of North Carolina at Chapel Hill) sydney shoemaker (Cornell University) judith j. thomson (Massachusetts Institute of Technology) recent titles: barry maund Colours michael devitt Coming to Our Senses michael zimmerman The Concept of Moral Obligation michael stocker with elizabeth hegeman Valuing Emotions sydney shoemaker The First-Person Perspective and Other Essays norton nelkin Consciousness and the Origins of Thought mark lance and john o’leary hawthorne The Grammar of Meaning d.m. armstrong A World of States of Affairs pierre jacob What Minds Can Do andre gallois The World Without the Mind Within fred feldman Utilitarianism, Hedonism, and Desert laurence bonjour In Defense of Pure Reason david lewis Papers in Philosophical Logic wayne davis Implicature david cockburn Other Times david lewis Papers on Metaphysics and Epistemology raymond martin Self-Concern annette barnes Seeing Through Self-Deception michael bratman Faces of Intention amie thomasson Fiction and Metaphysics david lewis Papers on Ethics and Social Philosophy fred dretske Perception, Knowledge and Belief lynne rudder baker Persons and Bodies john greco Putting Skeptics in Their Place derk pereboom Living Without Free Will brian ellis Scientific Essentialism julia driver Uneasy Virtue bennet m. helm Emotional Reason

Deontic Morality and Control

ISHTIYAQUE HAJI University of Minnesota, Morris

          The Pitt Building, Trumpington Street, Cambridge, United Kingdom    The Edinburgh Building, Cambridge CB2 2RU, UK 40 West 20th Street, New York, NY 10011-4211, USA 477 Williamstown Road, Port Melbourne, VIC 3207, Australia Ruiz de Alarcón 13, 28014 Madrid, Spain Dock House, The Waterfront, Cape Town 8001, South Africa http://www.cambridge.org © Ishtiyaque Haji 2004 First published in printed format 2002 ISBN 0-511-03026-6 eBook (Adobe Reader) ISBN 0-521-81387-5 hardback

for the love of my life, Shaheen

Contents

page xiii

Acknowledgments 1 Introduction Two Parallel Riddles Primary Goals Prospectus

1 1 5 8

part one: determinism and deontic morality 2 Obligation and Control Fundamentals of Moral Obligation ‘Can’ and Obligation 3 Frankfurt-Type Cases and Deontic Control Frankfurt-Type Examples A Requirement of Alternative Possibilities for Wrong Actions Principle CK and Wrongness The Plausibility of Principles CK and WC A Requirement of Alternative Possibilities for Deontic Morality An Alternative Argument 4 Control Requirements of Deontic Anchors: Some Objections Objections to K and Replies Frankfurt-Type Cases and K A Widerkerian Objection against K An Objection from Counterintuitiveness

ix

13 14 16 25 26 27 27 29 31 33 36 36 37 37 41

Fischer against K A Direct Threat against K from Frankfurt-Type Examples Self-Imposed Impossibility and K Pereboom on OW Genuine Moral Dilemmas and OW Appendix: Yaffe on K 5 Determinism and Deontic Anchors The Consequence Argument for the Incompatibility of Determinism and Alternative Possibilities Some Objections and Replies Why Determinism Undermines Deontic Anchors Objections to the New Incompatibility Thesis and Replies Appendix: Saka on ‘Ought’ Implies ‘Can’ and Determinism

43 46 47 52 53 54 59 60 62 65 70 77

part two: indeterminism and deontic morality 6 Transition: From Determinism to Indeterminism Introduction Synopsis R-Libertarianism Modest Meleian Libertarianism An Objection and a Reply Modest Meleian Libertarianism and Deontic Anchors 7 Robust Modest R-Libertarianism and Luck Robust Modest R-Libertarianism Robust R-Libertarianism and the Luck Objection Picking up the Gauntlet Refining PCEC Support for NCC Robust R-Libertarianism and Deontic Morality 8 Robust Modest R-Libertarianism and Deontic Anchors The Problem and the Outlines of a Solution A Case of CNC Manipulation: Psychohacker Robust Modest R-Libertarianism and CNC Manipulation Hierarchical Control and CNC Manipulation Normative Agency and CNC Manipulation Deontic Morality and CNC Manipulation

x

87 87 89 91 92 95 101 104 105 107 110 113 117 120 122 122 125 127 133 135 138

Antecedent Contrastive Control, Moral Responsibility, and Deontic Anchors Conclusion

142 147

part three: consequences of being deprived of deontic anchors 9 The Significance of the Possibility of Being without Deontic Anchors Transition and Synopsis Smilansky on the Ethical Advantages of Hard Determinism Pereboom on the Consequences of Determinism 10 Determinism, Deontic Anchors, and Appraisability Possible Justifications for the Objective View Supererogation, Suberogation, and the Objective View The Supererogatory and the Suberogatory Appraisability, the Supererogatory, and the Suberogatory Rejecting an Alleged Asymmetry between the Superand the Suberogatory Praiseless Supererogation and Blameless Suberogation The Objective View and the Significance of Alternative Possibilities The Argument from Fairness Two Faces of Responsibility Appendix: A Direct Challenge to the Objective View 11 Virtue Ethics without Metaphysical Freedom An Outline of a Version of Virtue Ethics Broad Contours of Another Virtue-Ethical Theory The Many Faces of Morality and the Autonomy of Deontic Appraisals Appendix: Kekes on K 12 On the Connection between Morality’s Dethronement and Deontic Anchors The Overridingness Thesis A Confusion Revealed Slote on Overridingness Kekes on Overridingness Shortcomings of the Overarching Standard – Reason On Reason’s Dethroning Morality

xi

151 151 153 155 162 164 167 167 170 176 177 182 184 191 194 197 198 202 205 211 221 221 225 226 230 234 235

Some Interim Conclusions Overridingness and Culpability A New Pair of Skeptical Arguments 13 Concluding Remarks On Not Being Sources of Deontic Morality Closing Comments

237 239 242 245 246 247

Notes Glossary and List of Principles References Index

252 269 272 283

xii

Acknowledgments

Work on this book began in the spring of 1998 while I was a visiting fellow at the Catholic University, Leuven. Very many thanks indeed to Stefaan Cuypers and Arnold Burms with whom I had a series of extremely illuminating discussions that influenced central themes of the project. I am also most grateful to many other philosophers for their critical comments, suggestions, or conversations that helped enormously. In particular, I thank Randolph Clarke, David Copp, Carl Ginet, Michael McKenna, Al Mele, and Bob Kane. John Davenport and Derk Pereboom read complete drafts and supplied very useful criticism and advice concerning sections of the manuscript. I am very thankful for their help. I suspect, with much delight, that my pertinent conversations with Derk will extend long into the future. It is my distinct pleasure to mention two individuals who deserve special acknowledgment for their extensive contributions. Michael J. Zimmerman read various renditions of the entire manuscript, many sections many times over, and provided very valuable feedback for revision. He pointed out several errors, raised a number of serious and penetrating objections (as well as a host of little ones), and offered judicious suggestions about how the book could be refined. He furnished much needed support throughout the trying period during which I endeavored to place the manuscript. He more than anyone forced me to rethink a number of my positions. As a result of his exceptional efforts, the manuscript is significantly improved. I am tremendously indebted to him, and I extend to him my deepest gratitude and respect. It should be obvious to any reader that I owe a very great deal to John Martin Fischer. He read two (quite) different versions of the entire

xiii

manuscript and firmly, but fairly, recommended a complete reorganization of the first version. He volunteered sage advice and gave me staunch encouragement during difficult times. His unwavering support and his extraordinary tutelage have been invaluable. I am deeply grateful to him. I give special thanks to production editor Louise Calabro, copyeditor Cynthia Benn, and the production staff at Cambridge University Press for their invaluable editorial assistance. I am thankful for permission to reproduce portions (sometimes modified) of previously published works. These works are listed in the References as: Haji 1998c – reproduced by permission of the editor of Philosophical Papers; 1999a – copyright 1999 University of Calgary Press, reproduced by permission of the editor of the Canadian Journal of Philosophy; 2000 – copyright 2000 Kluwer Academic Publishers, reprinted by permission of Kluwer Academic Publishers; and 2001 – copyright 2001 Blackwell Publishers, reproduced by permission of the editor of Pacific Philosophical Quarterly.

xiv

1 Introduction

TWO PARALLEL RIDDLES

Very many of us are convinced that on numerous occasions in our lives we perform actions that are morally right, or wrong, or obligatory. But the seemingly innocuous view that our actions have such moral statuses may not be as secure as we initially believe. I want to generate a riddle about this view that, in many respects, bears striking resemblance to a much more widely known riddle. It will be helpful to start by saying something about this other riddle in order better to appreciate the new riddle. The venerable old riddle is the riddle about freedom and responsibility. Though fascinating and deeply puzzling, its essentials are easy to grasp. Almost all of us believe that people have been and will be morally responsible for at least some of their behavior. But suppose causal determinism – roughly, the view that all the facts of the past, in conjunction with all the laws of nature, entail one unique future – is true.1 Then it seems that, at each instant, we would lack genuinely open alternatives; contrary to popular belief, there would be no time at which we could do other than what we in fact did at that time. Causal determinism threatens our very natural picture of the future as a garden of forking paths. Perhaps a more apt figurative representation of what our lives, including our futures, would be like if causal determinism is true is captured by the image of trains chugging along the predestined grooves of a nonbranching trunk line (see Feinberg 1980: 36–7). In addition, if determinism is true, all our thoughts, choices, and actions are simply events deterministically caused by others in a long sequence of such events, the inaugural members of which, presumably, originated

1

with the birth of the universe. But if such is the case, then, arguably, we never ultimately initiate our actions; we are simply one transitional link in an extended deterministic chain that has its beginnings in the big bang.2 So it seems that causal determinism is incompatible both with the sort of control or freedom that involves genuinely open options – it expunges alternative possibilities – and persons being ultimate initiators of their actions – it undermines “ultimate” agency or responsibility. No wonder, then, that many have thought that determinism and responsibility are incompatible. For responsibility, it might plausibly be supposed, does require that we have the sort of control over our behavior that involves freedom to do otherwise and that we be the “ultimate sources” of our actions.3 This puzzle about freedom and responsibility swells with bewilderment because the falsity of determinism seems to imply nonresponsibility as well. The reasoning here is again fairly elementary. If determinism is false, then it looks as though at least some of our behavior must issue from choices, decisions, or other elements in the pathway of events culminating in that behavior that is undetermined in, roughly, this sense: Given exactly the same past and the laws of nature, one could have made a choice or decision other than the choice or decision that one actually made. So, for instance, in a world in which determinism is false and my choice is nondeterministically caused, it could happen that, torn between the delights of exploring Flanders on a fine sunny spring day and writing this chapter, although I decide to write, I could, right up to the moment just before this decision, have decided to explore, given exactly the same laws of nature and past circumstances that include all my prior reasons to write. But then my decision to write looks for all the world as though it is a matter of luck; “dumb luck” as Alfred Mele (1999a) says. If there is nothing about my powers, capacities, states of mind, character, and the like that explains why I decide to write rather than to explore, the undetermined decision that I do make really does smack of luck. Such luck seems incompatible with moral responsibility. So although indeterminacy in actional pathways culminating in behavior might free us from domination of the past, we might wonder how it could possibly help to ensure that the reigns of control are now securely in our hands. The denial of determinism (“indeterminism”), thus, seems just as irreconcilable as determinism is with responsibility. This celebrated old riddle is not of primary concern to me in this book. Rather, as I suggested earlier, it serves as a springboard to launch an analogous new riddle about moral statuses such as rightness, wrongness, and obligatoriness, on the one hand, and determinism and indeterminism,

2

on the other. The parallel between the two riddles can be brought out perspicuously in this fashion. Use the label “primary deontic properties” to refer to the moral properties of rightness, wrongness, and obligatoriness, and call any act that instantiates one or more of these properties a “deontic act” or a “(moral) deontic anchor.” The set of deontic acts comprises “deontic morality.” The dilemma concerning freedom and moral responsibility can be expressed in this way: (1MR) If determinism is true, then no one has control over one’s actions. Similarly, (2MR), if indeterminism is true, then no one has control over one’s actions. Now, (3MR), either determinism or indeterminism is true. Hence, (4MR), no one has control over one’s actions. The freedom-relevant component of responsibility says that, (5MR), one is morally responsible for one’s actions only if one has control over those actions. It follows that, (6MR), no one is morally responsible for one’s actions. This argument, I believe, can be modified so that it poses a dilemma concerning moral obligation, right, and wrong as follows. Premises (1MR)–(4MR) remain unchanged. The lemma (4MR) implies that whether determinism or its denial is true, no one has control over one’s actions. The revised fifth premise says that it is morally obligatory, right, or wrong for one to perform some action only if one has control over it. The troubling new conclusion, then, is that no action is such that it is morally obligatory, right, or wrong for one. Let’s put some flesh on this bare-bones sketch of the riddle about freedom and deontic morality, but only enough to kindle confidence that there really is a dilemma here worth exploring. Just as it is eminently reasonable to suppose that responsibility requires control – no one, for example, can be morally responsible for an action if one does not have “responsibility-grounding control” over the action – so it is reasonable to suppose that no one can perform an action that is morally right, or wrong, or obligatory unless one has appropriate “deontic-grounding control” over it. So, for instance, if Leno the lifeguard has been shackled to his seat against his will, and, as a result, he cannot save the child, we don’t think that he is under any moral obligation to save the child. He is free of any such obligation in his circumstances as, roughly, one is under an obligation to do something only if one is free to do that thing or one has control over doing that thing. ‘Ought,’ we believe, implies ‘can.’ Analogously, had Leno intended not to save the child but unbeknownst to him he could not have saved the child even if he had wanted to, then it seems that his failing to save the child is not wrong, again, because in order for an action

3

to be wrong for an agent, the agent must be able to control the action in an appropriate way. One promising strategy to establish the incompatibility of determinism and deontic morality proceeds by arguing that no actions can be right (or wrong, or obligatory) for a person unless that person was free to do otherwise. Exploiting this strategy, it is relatively straightforward to show that there is a requirement of alternative possibilities for moral wrongness; that is, no one can perform an action that is wrong unless one could have refrained from performing it. For suppose the principle that ‘ought’ implies ‘can’ is true. The principle says that if one ought to do something, then one can do it; and if one ought to refrain from doing something, then one can refrain from doing it. So we have: K: Agent S has a moral obligation to perform [to refrain from performing] action A (where A ranges over omissions as well) only if S can perform [refrain from performing] A. There is another stock principle of moral obligation that connects obligation with wrongness. It says that one has a moral obligation to do something if and only if it is wrong for one not to do that thing. The principle can be reformulated in this way: OW: Agent S has a moral obligation to perform [to refrain from performing] action A if and only if it is morally wrong for S to refrain from performing [to perform] A. Now, given OW, if it is wrong for one to do something, then it is obligatory for one to refrain from doing that thing. So, for instance, if it is wrong for Augustine to steal the pears, then it is obligatory for him to refrain from stealing them. Further, given K, if it is obligatory for one to refrain from doing something, then one can refrain from doing it. Hence, if it is obligatory for Augustine to refrain from stealing the pears, he can refrain from stealing them. It follows that if it is wrong for Augustine to steal the pears, then he can refrain from stealing them. Generalizing, if it is wrong for one to do something, then one can refrain from doing that thing. We can put the point crisply in this way. OW and K yield the further principle that there is a requirement of alternative possibilities for wrong actions: WAP: It is morally wrong for S to perform [to refrain from performing] A only if S can refrain from performing [perform] A.

4

WAP implies that no one is able to perform an action that is wrong unless one is able to refrain from performing it. Suppose determinism is true and no one has freedom to do otherwise. Then, given WAP, it appears (bracketing, for now, a complication to be introduced and dealt with later), no actions of any person are wrong. Further, if some action A in a determined world is obligatory for some person, then failing to do A is wrong for that person in that world. But it is false that any action (or omission) in such a world is wrong for any person, and so it is false that failing to do A is wrong for that person. Hence (again, overlooking a complication to be handled later), determined worlds, it appears, will be devoid of acts that are obligatory or wrong. Consider, now, an indeterministic world in which some of our decisions and choices are nondeterministically caused. Assume that the decision to do one thing or another, like writing or exploring, is undetermined right up to the very moment before one or the other of these mental events actually occurs. In other words, assume that it is causally open, right up to the moment just prior to what one has decided, whether one will decide to write or to explore. Then it seems that the decision made and the subsequent behavior that occurs are infused with luck. But just as luck of this sort appears incompatible with moral responsibility, so one might think, luck of this sort is incompatible with deontic morality. How can the making of my decision be right, wrong, or obligatory if it is simply a matter of luck that I made one particular decision and not some other on a certain occasion? Hence, in an indeterministic world of this sort, deontic anchors appear to fall by the way as well. PRIMARY GOALS

My primary overall goals can now be recorded. First, as I have mentioned, my hope is to develop, with as much clarity as I can muster, a (somewhat) new riddle or dilemma about freedom and deontic morality that closely mirrors the grand old one about freedom and responsibility. A careful, detailed presentation of the new dilemma is desirable for a number of reasons. For one thing, a revealing formulation of this dilemma demands detailed inquiry into the sort of control required by deontic anchors, and this, I believe, is a worthy philosophical task. For another thing, by conscientiously exposing the presuppositions of the premises of the argument regarding the dilemma of freedom and deontic anchors (much as we have been doing and continue to do with the argument concerning the dilemma of freedom and responsibility), we will be in a better position

5

to assess these premises and to develop responses. Compatibilists about freedom and deontic anchors may want to try their hand at showing how such anchors are compatible with determinism; incompatibilists may want to argue to the contrary; and libertarians may undertake to convince us that the type of freedom required for deontic anchors is incompatible with determinism but that human beings sometimes perform actions that are right, or wrong, or obligatory. It would be premature to begin to respond to the new dilemma in any one of these or other ways unless its various nuances have been exposed and understood. Compatibilists about freedom and responsibility claim that causal determinism is compatible with moral responsibility. Some compatibilists, “semicompatibilists,” defend the interesting position that whereas causal determinism is incompatible with freedom to do otherwise, it is nevertheless compatible with responsibility.4 It strikes me that the “deontic” analogue to semicompatibilism, the view that determinism is incompatible with its being the case that one could have done otherwise and yet compatible with its being the case that one’s actions have at least one of the primary deontic properties of obligatoriness, rightness, and wrongness, is false. This brings me to my second overall goal. I aim to argue that promising compatibilist strategies, like that of the semicompatibilist, to show that determinism does not threaten moral responsibility, are ineffective in shielding deontic anchors from the clutches of determinism. Thus, for example, even if semicompatibilism affords some comfort to those who seek to defend moral responsibility against the threat of determinism, its analogue cannot be relied on to defend moral obligatoriness, rightness, or wrongness against this same threat. I do, then, take a stance on the whole issue of compatibility: I intend to argue that it is highly plausible that determinism is indeed incompatible with deontic morality. For all we yet know, our world is a deterministic one, and if it is, we are without deontic anchors. A third fundamental goal is to propose that there is some measure of hope that the new dilemma can be evaded but not in a fashion compatibilists would find congenial. To elaborate, one intriguing response to the dilemma concerning freedom and responsibility is the response of the modest libertarian. The basic aim of modest libertarianism is to specify the sort of control that suffices for moral responsibility without appealing to metaphysically exotic agents, such as Kantian noumenal selves, or forms of causation like agent-causation. I believe, and will defend the view, that the most sophisticated and robust versions of this type of libertarianism do not

6

succeed in achieving this aim. They fail, in part, because they run afoul of the problem of luck. However, I also believe, somewhat paradoxically, that these sorts of libertarian theories hold the key to accommodating deontic morality. The type of luck entailed by these theories may threaten responsibility but not deontic morality. My stance, then, will be that whereas determinism is incompatible with deontic morality, indeterminism (or more specifically, a version of an “indeterministic theory”) – even with its burden of luck – is compatible with such morality. It is reasonable to suppose that there are close conceptual connections between moral responsibility and deontic anchors. So, for instance, Susan Wolf has argued that the sort of control required for moral responsibility is the ability to do the morally right thing for the morally right reasons. The question of whether we have this ability, she says, “is not so much a metaphysical one as a metaethical, and perhaps also an ethical one” (1990: p. 71). So on Wolf ’s view, there is a close tie between the type of control needed for moral responsibility and deontic anchors. Others have contended that one cannot be morally blameworthy for performing an action unless that action is morally wrong,5 and analogously, one cannot be morally praiseworthy for performing an action unless that action is morally obligatory or at least morally right. Again, according to these theorists, there are intimate conceptual links between responsibility and deontic anchors. A fourth goal of mine is to examine some of these proposed connections and inquire into others, primarily with an eye toward exploring whether moral responsibility stands or falls with deontic morality. Can, for example, persons be genuinely morally responsible for at least some of their behavior in a world in which no behavior is morally right, wrong, or obligatory? Related to the second and third goals is a fifth: I introduce two “asymmetry theses,” one concerning an asymmetry in control requirements for responsibility and deontic anchors, and the other concerning an asymmetry in “agency presuppositions” of responsibility and deontic anchors. The first says that whereas there is a requirement of alternative possibilities for deontic anchors, there is no such requirement for responsibility. (If this thesis is correct, “semicompatibilism” with respect to deontic anchors is not a viable position.) So although there is, I believe, a close parallel between the dilemma of freedom (or control) and responsibility and that of freedom (or control) and deontic anchors, control requirements for deontic anchors differ from those for moral responsibility. This, if anything, renders the dilemma of freedom and deontic anchors more incisive than that of freedom and responsibility. Regarding this first thesis, I will defend

7

only the part of it that says that there is a requirement of alternative possibilities for deontic anchors; the other part concerning responsibility is something that I believe is true. But I will not undertake a full defense of it in this work.6 The part of the first thesis that I do defend bears directly on whether determinism is compatible with deontic morality. For if determinism effaces alternative possibilities, and there is a requirement of alternative possibilities for deontic acts, then it follows that determinism is not compatible with deontic morality. According to the second asymmetry thesis, the “agency presuppositions” of moral responsibility are stronger than those of deontic morality. The idea here, roughly, is that whereas one cannot be responsible for an action unless it is caused by things like beliefs, desires, and values that are “truly one’s own” – they are not, for example, the product of surreptitious conditioning – one can perform an action that is right, or wrong, or obligatory even though its causal precursors are not truly one’s own. This second thesis, as I will argue, has implications for whether versions of “indeterministic” theories can accommodate deontic anchors. Finally, in a world without deontic anchors nothing can be morally right, or wrong, or obligatory but other sorts of moral appraisal could, presumably, still be duly made. For instance, persons’ actions could still be intrinsically good or bad, or persons could still act “out of ” or “from” kindness, or generosity, or, more generally, they could still act from virtue. What, then, it might be queried is so important about deontic moral appraisals, and what loss would be suffered if, for example, deterministic worlds were indeed bereft of deontic morality? These orienting aims are intended to guide the reader by directing attention to the overall themes of this work. For further guidance, the book divides into three principal parts: The first (Chapters 2–5) establishes the incompatibility of determinism with deontic morality, the second (Chapters 6–8) substantiates the compatibility of a certain version of indeterminsitic theory with deontic morality, and the third (Chapters 9–13) addresses the significance of the possibility of being deprived of deontic anchors. PROSPECTUS

It will be helpful to start with an overview of each chapter. Deontic anchors require control; one cannot, for example, perform an action that is obligatory unless one has control over it. In Chapter 2, I begin with a fundamental principle of deontic control, principle K or the principle that

8

‘ought’ implies ‘can.’ The sense of ‘ought’ or ‘obligation’ in this principle is clarified as is the sense of ‘can.’ In addition, different versions of the principle are identified and examined. In Chapter 3 a complement to K, the principle that if one ought to do something, then one can refrain from doing it (CK), is defended. This principle, together with various other deontic ones like K, OW, and WAP, are enlisted to argue for a crucial asymmetry between the concepts of moral responsibility (praise- and blameworthiness) and those of deontic morality (obligation, right, and wrong), to the effect that even if moral responsibility is possible in the absence of freedom to do otherwise, deontic morality is not. Despite rebuttal in the literature of various objections against K, many people still regard K with a good deal of suspicion.7 But principle K, as we shall see, plays a central role in the dilemmatic argument for freedom and deontic anchors. The argument’s reliance on K may, consequently, appear to compromise its interest. Still, although K is controversial and rejected by many, it is widely accepted by others; indeed, some accord it the status of a deontic axiom. It may, then, prove instructive to those who find K plausible to see how K fuels the riddle about freedom and deontic anchors. I do, though, in Chapter 4, consider recent objections to K that I believe are especially relevant to the concerns of this work and I explain why they fail. In addition to defending K, OW is defended as well. In Chapter 5, I introduce the following argument that is one of the primary arguments of this book: No one can perform an action that is right, wrong, or obligatory unless one could have done otherwise. This is because deontic anchors require alternative possibilities. In addition, no one can do other than what one in fact does in a determined world. This is so because in a determined world, the laws of nature and the past preclude agents from doing other than what they in fact do. It follows from these two premises that deterministic worlds are without deontic anchors.8 Discussion in Chapter 5 focuses, in large measure, on the second premise, that determinism rules out alternative possibilities. Even if we conclude that this premise is controversial, it may still be true. And once again, taking a charitable approach, it would be useful to see what lessons can be discovered about deontic anchors on the assumption that this premise about determinism is true. In the first chapter of Part II (Chapter 6), I discuss and reject a modest form of libertarianism to which some might appeal to accommodate deontic anchors. This sort of libertarianism positions indeterminacy relatively early in the causal trajectory of an action at the point, roughly,

9

between an agent’s reasoning concerning a prospective action of hers and the formation of a best judgment regarding that action. In Chapters 7 and 8, I consider a more robust form of libertarianism that places indeterminacy relatively late in the causal pathway of an action, at the juncture between the consideration of reasons and the making of a decision. I argue that such libertarianism probably cannot accommodate moral responsibility but can accommodate deontic morality, since (contrary to what is often maintained) it can accommodate the freedom to do otherwise but cannot accommodate the sort of self-expression required by moral responsibility (but not required by deontic morality). In Chapter 9 (the initial chapter of Part III), I begin scrutiny of the significance of the fact, if it is one, that no action is morally obligatory, right, or wrong. I propose that one cost of being deprived of deontic anchors is that, without these anchors, we would have no rational grounds for such reactive attitudes as indignation, resentment, and forgiveness. I argue in Chapter 10, for the independence of the concepts of moral responsibility from those of deontic morality, based on considerations concerning supererogation and suberogation. Among other things, the discussion in this chapter contributes to diffusing an argument for the view that a world without deontic morality is a world without moral responsibility. In Chapter 11 I discuss and reject the claim that the concepts of virtue and vice can both escape the new dilemma and replace the concepts of moral rightness, wrongness, and obligatoriness. The discussion highlights another cost of being deprived of deontic anchors: Without such anchors, deontic moral appraisals are not possible. I inquire in Chapter 12 into whether moral obligation overrides nonmoral obligation and whether, if so, this would enhance the significance of no action’s being morally obligatory, right, or wrong. Finally, in Chapter 13 I expose one further cost associated with no action’s being morally obligatory, right, or wrong: In a world without deontic anchors, no persons would be “sources” (in a fashion to be explained) of deontic morality. I conclude with some reflections on my view that certain robust versions of libertarianism can accommodate deontic morality but not moral responsibility.

10

Part One Determinism and Deontic Morality

2 Obligation and Control

One cannot perform an act that is morally right, wrong, or obligatory if one does not have appropriate control over that act; deontic acts presuppose control. This is, perhaps, most evident in the case of moral obligation. It is difficult to accept the verdict that Leno, though pinned to his seat, ought to have saved the child, or that the paraplegic ought to have walked across the lawn when ‘ought’ denotes moral obligation or requirement. Again, it is hard to see how Mitch did something that is morally wrong by failing to help the victim of the car crash when he could not have helped the victim. Or it is, minimally, puzzling to see how Zakir did moral wrong by stealing the loaf when he could not but have stolen it. One might, given apt circumstances, reasonably think that Zakir did something bad by stealing the loaf, but this judgment is consistent with Zakir’s not having done wrong. Or, again, it is paradoxical to suppose that the mother’s not spanking the child was morally right when she literally lacked the ability to spank her child. These enigmas are not merely pragmatic or do not simply have to do with substantive moral claims about what is fair but are conceptual; the verdict about the paraplegic, for instance, seems conceptually inconsistent. Such examples, and numerous others like them, motivate the view that one cannot do what is right, or wrong, or obligatory unless one has appropriate “deontic-grounding control” over what one does. But precisely what sort of control is at issue here? In this chapter, I lay the groundwork for responding to this leading question by focusing on moral obligation. I do this as there is a control principle with broad intuitive appeal associated with such obligation. This is the famous ‘ought’ implies ‘can’ principle and its corollary that ‘ought not’ implies ‘can avoid’ that says:

13

K: As of time t, an agent S ought morally to do something A at time t ∗ (where t may either be t ∗ or a time later than t ∗) only if S can, as of t, do A at t ∗; and, as of t, S ought not to do A at t ∗ only if S can, as of t, not to do A at t ∗. In other words, ignoring temporal indices for a time, K, as I explained in the opening chapter, says that one ought to do something only if one can do it, and that one ought to refrain from doing something only if one can refrain from doing it. We cannot, though, understand this principle unless two of its central terms, ‘ought’ and ‘can,’ are clarified. I start, in the next section, with ‘ought.’ In the subsequent section, I address ‘can.’

FUNDAMENTALS OF MORAL OBLIGATION

In this work, unless otherwise specified, I take ‘ought’ to express moral obligation or, equally, moral requirement or necessity. ‘Ought,’ is, of course, frequently used to express nonmoral obligation or requirement such as prudential, legal, etiquettical, or aesthetic requirement. “You ought not to have belched in the presence of the queen” probably expresses criticism based on etiquette rather than moral criticism. “You ought to diversify your financial portfolio” is likely to constitute a prudential rather than a moral recommendation. It seems that sometimes ‘ought’ is used to express an overarching or an all-encompassing sense of obligation as becomes fairly evident upon reflection on the fact that different sorts of obligation can conflict. For example, moral obligation can conflict with prudential or legal obligation. A person may, in a specific case, realize that she has a moral obligation to do one thing, a prudential obligation to do another, and a legal obligation to do yet something else, but still ask: “I know what my specific obligations are, and I realize that they conflict – they can’t be jointly executed – but what ought I to do period or what plain ought I to do?” When a person asks this question, surely she is not asking about what she morally, prudentially, or legally ought to do, for she confesses knowing what her “specific” obligations are. Rather, in asking this question, she is asking about which of her competing specific obligations, if any, takes “precedence” over or “overrides” others. It appears, then, that we cannot make sense of her question regarding what she plain ought to do unless we suppose that there is an overarching normative standard, one regarding an all-inclusive concept of obligation, that delivers verdicts on the relative normative importance or stringency of specific normative obligations such as moral,

14

legal, or prudential ones.1 Whether there is any such overarching standpoint is not in question – at least not here – though I shall have more to say about this later. The present concern is simply to record the fact that the concept of moral obligation is not to be identified with that of plain or overarching obligation, though if there is an overarching ought, that ought, some have professed, might turn out to be the moral one. Even within the moral domain, ‘ought’ doesn’t always express moral requirement or obligation. It expresses, instead, an ideal or a moral desideratum, as when we say, for instance, that “people ought not to suffer,” or “there ought to be plenty to eat for everyone.” We use ‘ought’ or ‘should’ to express such ideals even when we know that the ideal state or moral desideratum is something that may not be in anyone’s power to bring about. Presumably, none of us can bring it about that every child be happy, and hence, none of us has a moral obligation or ought morally to bring it about that every child be happy, yet, we can aspire toward achieving the moral ideal that every child be happy. The term ‘ought’ as it occurs in the principle that ‘ought’ implies ‘can’ expresses moral obligation or necessity and not a moral ideal or desideratum. Since W. D. Ross’s publication of The Right and the Good, we commonly distinguish between what Ross called “absolute obligation,” “actual obligation,” “duty proper,” or “duty sans phrase,” on the one hand; and “prima facie duty,” “conditional duty,” or “duty ceteris paribus” on the other (Ross 1930: 18–20, 28, 30). The rough idea between these two species of moral obligation is familiar enough, although the concept of prima facie obligation does not lend itself to easy analysis.2 In drawing the distinction, one leading idea of Ross’s seems to have been that one can frequently find oneself in circumstances in which different moral considerations or duties – the prima facie ones – bear on what one morally ought, “absolutely,” to do, all these morally relevant prima facie duties taken into account. As an illustration, on his way to meet his partner for a game of tennis, as he had promised, Doctor Kildare comes across an accident victim whom he could assist. Ross would say that Kildare has a prima facie duty (one of fidelity) to keep his promise, and a prima facie duty (one of beneficence) to render aid. These are “real” moral duties or considerations and not “on first glance” or “on the face of things” duties. Another guiding idea of Ross’s is that what Kildare morally ought absolutely to do is a function of the stringency of these prima facie duties or considerations. Presumably, in Kildare’s case, Kildare ought absolutely to stop and render aid. The concept of obligation operative in K is that of absolute or, as I prefer to say, “overall” moral obligation.

15

Finally, although this is not essential to the concerns of this work, I take what is overall obligatory among a set of alternatives to be that which is superior or best in some morally relevant way to its alternatives. I remain noncommittal on the axiological issue of what deontic bestness – deontic value – consists in. It may, for example, consist in maximizing intrinsic value, or in maximizing prima facie stringency, or in maximizing compliance with the Categorical Imperative, or in maximizing compliance with God’s commands, or in maximizing nonviolation of persons’ rights. As Michael Zimmerman has fruitfully suggested, the view that one ought (overall) to do the best one can, with emphasis on deontic value, is compatible with almost all substantive theories of obligation, whether consequentialist or nonconsequentialist (1996: 18. See, also, Nozick 1974: 29n.). A terminological concern is in order. Some theorists might reserve ‘ought’ for one concept and ‘duty’ for another (e.g., Feinberg 1968: 392), and with them I have no quarrel. But in this work I use ‘ought,’ ‘duty,’ or ‘obligation,’ to express the same concept, that of overall obligation. We can now address the second term of interest in control principle K, ‘can.’ ‘CAN’ AND OBLIGATION

The interpretation of ‘can’ in K is far more contentious than that of ‘ought’ in K, but there is at least general agreement that ‘can’ cannot plausibly be taken to express metaphysical possibility. Leno the lifeguard, tethered to his seat, can’t, in the sense of ‘can’ operative in K, save the drowning child, although there is some possible world in which he does. As Leno’s example suggests, ‘can’ in K is frequently associated with physical ability, the thought being that if one is “physically unable” to do something, one can’t be under an obligation to do it. Consider “Safe1,” a case in which the only way to disarm a bomb, that if detonated, will result in the death of millions, is by opening the safe and pressing the green button inside. Fox, aware of the bomb and the correct combination, cannot open the safe, despite being right next to it, because he is paralyzed. Even though he is “relevantly situated” (he is not, for example, on the other side of the moon but is within arm’s reach of the safe) and even though the environment cooperates (there is, for instance, no gale that prevents him from reaching the safe) his physical incapacity renders him unable to open the safe. In Safe1, as there is a transparent sense in which Fox lacks relevant control over opening the safe, he has no obligation to open it. ‘Can’ is sometimes affiliated with “know-how.” Safe2 is just like Safe1 except that Trot doesn’t know the combination and is not physically

16

disabled. Assume that even if Trot were to dial a million possible combinations, he would not hit on the right one, and the bomb would explode. One might sensibly claim that Trot cannot open the safe and thus has no obligation to do so. It’s still true, though, that as Trot is not physically disabled, he can dial any combination, and so can dial the correct one, and hence can open the safe even on the very first try. But if he tries, and tries and tries again, he won’t succeed, and the bomb will go off. Whereas Fox is physically disabled but appropriately informationally equipped, Trot is “informationally” or “know-how” impaired. Should we then agree that Trot just like Fox has no obligation to open the safe? Many would not accede to this. Witness, for example, Michael Zimmerman’s comments. Acknowledging that ‘ought’ implies ‘can,’ and having introduced the “know-how” sense of ‘can,’ he writes: [I]t is not the sense of “can” that I’m working with. . . . I can open the safe . . . in the present sense of “can,” and this is simply because I can (easily) perform those physical movements which would result in these things being done. Of course, there is a sense in which I’d be very lucky if I managed to do these things; but in this sense of “luck” it is not that their accomplishment is strictly out of my control, but rather that, among the very many things that I can apparently do, I have no reason to choose to make this particular set of movements (that would result in the safe’s being open . . .) over the many other sets of movements (that would be fruitless). (1996: 49, note omitted)3

Zimmerman’s position is reasonable, but it does generate a puzzle: Why exactly is it that if one lacks the physical ability to do something, one is under no moral obligation to do it, but if one lacks the know-how to do something, one can still be under an obligation to do it? One response is that physical disability is more normatively “basic” than “know-how” deficiency merely because, if one lacks relevant know-how but is not physically impaired, the deed can still be done, even if only (in some cases) by monumental luck. I find this rationale unpersuasive. For all rhyme and reason, the world in which Trot dials the right combination (imagine that the combination is complex consisting of a whole slew of numbers) is so distant that he is as incapacitated with respect to opening the safe as is Fox. To be sure, they are differently impaired, but their impairment prevents both of them from opening the safe. To make headway, let’s introduce some more cases involving different sorts of disability but in which we may be more confident that the agents still have an obligation to perform the germane acts. Then we can see whether Safe2 is sufficiently similar to these cases, and whether, in light of the similarity, we have better

17

grounds to conclude that lack of know-how is not, as I’ll say, “obligation subversive.” In Safe3 Tango, who is sitting by the safe and who knows the combination, is completely unaware of the bomb. She is deeply immersed in her novel, trying to anticipate Holmes’s next move. Her fingers, though, are in perfect working order and she can dial any combination. But she is, in her circumstances, not motivated to open the safe. Nor is she capable, let’s assume, in the time available, of acquiring motivation – acquiring an appropriate desire, for example – to open the safe. Even if her colleague were to yell, “Open the safe, Tang, and press the green button inside!” Tango would take too long to “disengage” from Sherlock’s world and would, consequently, not have sufficient time to acquire motivation to open the safe and respond. Barring certain complications, an agent who is incapable of being motivated to perform a certain action is not able to perform that action intentionally (Mele 1995: 150). Although Tango can open the safe, she cannot (in her circumstances and in the right space of time) do so intentionally. Here, one might credibly urge that though Tango doesn’t have any desire to open the safe, she is still under an obligation to do so (or that she can be under an obligation to do so). After all, not desiring to do something, it appears, does not provide legitimate escape from being obligated to do it. If we agree that Tango ought to open the safe even though she lacks motivation to do so, then ‘can’ in Principle K does not entail ‘intentionally can.’ One might object that Tango can (actively) move her fingers in the requisite “safe-opening way” only if she can intentionally do so. And if she currently lacks the necessary motivation to move her fingers intentionally and cannot (in the right space of time) acquire this motivation, then it would seem that she cannot open the safe, period. I’m unconvinced, though, by the first claim of this objection. For it seems that there are cases, much like Tango’s, in which Tango succeeds in unintentionally doing the relevant deed. So, for instance, it appears that Tango could unintentionally flick on the light switch, even though she currently lacks the required motivation to move her fingers intentionally in the right way, and can’t, in the right space of time, acquire this motivation. Admittedly, opening the safe does call for far more complicated physical movements, but I don’t see why, in principle, Tango couldn’t unintentionally move her fingers in the right way.4 Perhaps “motivational impairment” is sufficiently similar to “knowhow” impairment so that if the former is not obligation subversive, the latter shouldn’t be either. Well, are they relevantly alike? Both sorts of

18

impairment are consistent with the agent’s not being physically disabled. More revealingly, it seems that, barring the truth of some versions of internalism, one can believe that one ought to do something without being motivated to do it or without being able to muster motivation to do it.5 Suppose I have promised to help my neighbor to move heavy furniture, and in virtue of having promised, now have an obligation to help her move some pieces tomorrow. Suppose, just before the time of the move, the big fight airs on the television, and I subsequently find myself with no motivation – I have no desire at all, let’s assume – to help her move. Still, one might think, as I can help her move – I’m not physically disabled and I can acquire sufficient motivation if I try sufficiently hard – I do wrong by not helping her move despite being motivationally deprived. Analogously, it seems that I can believe that I ought, as of now, to do something later even though I will lack the germane know-how when the time comes to discharge my obligation. Let’s suppose, this time around, that I have promised the neighbor to clear her driveway of snow during her absence, should the need arise to do so, by using her snowblower. I don’t know how to use the blower, but I have assured her that I’ll be able to figure it out; I’ll just read the instruction booklet if the weather turns sour. Suppose it does snow. It’s imperative that the driveway be cleared quickly, as if it isn’t, the snowmelt will congeal into ice and create a dangerous mess. But suppose I have been lazy; despite ample warning of the impending storm, I simply fail to study the instructions. I end up not clearing the snow in time as I don’t know how to operate the blower. It seems that, as of the time I promise to clear the snow, I incur an obligation then even though I then lack the relevant know-how. Again, though, one might object that the case of the snowblower is unpersuasive as a counterexample to the claim that ‘can’ involves ‘knowhow.’ It may be that the reason why it is correct to say that I can on Monday blow snow on Wednesday is that I can on Monday study the instructions on Tuesday and thereby learn how to blow snow. But yet again, I’m not so sure about this objection. Suppose, aware of my vices, I believe correctly that there is an extremely remote possibility that I will read the instructions, and I believe this at the time of my promise. I’ve already decided to drink myself into a stupor on Monday, full well knowing that by so doing, I won’t be able to study the booklet before the expected storm on Wednesday. Nonetheless, I believe (truly) that if I don’t take to the rum, I can prepare myself, and I believe (truly) that there is an exceedingly remote chance that I won’t take to the rum. In this version of the story, it seems implausible to suppose that the reason why one

19

might claim that it is right to say that I ought, as of Monday, to blow snow on Wednesday is that I can on Monday study the instructions before Wednesday. For what’s true is that the situation is such that there is an exceptionally remote possibility of my studying the instructions before Wednesday, and I know that this is so. And this situation is very like the counterfactual one in which I can’t, as of Monday, study the instructions before Wednesday. Still, it strikes me as reasonable to suppose that as of the time of the promise, I incur an obligation then to clear the snow even though I then lack the relevant know-how. And if this is so, it seems that I can believe that, as of the time of my promise, I ought to clear the snow later (if it snows) even though as of the time of the promise, I don’t know how to operate the blower. So just as failing to have motivation to do something at a time is consistent with having an obligation then to do it, so failing to have appropriate know-how to do something at a time seems consistent with having an obligation then to do it. There is another feature of Safe3 that merits attention. Unlike Safe2, Safe3 involves ignorance of facts, lack of knowledge that something is the case, as opposed to not knowing how to do something. Tango has no reason to open the safe – she has no idea whatsoever that there is a bomb that can be disarmed only by pressing the green button within. Again, some might say that it is reasonable to suppose that Tango’s not believing that she ought to open the safe is consistent with her having an obligation to open it. But here, too, there is controversy. Ramon Lemos gives an example in which Matthew, who cannot swim, is on the verge of drowning in a secluded lake, his canoe having capsized. Mark, an excellent swimmer, is the only other person around, but he neither hears nor sees Matthew. Lemos’s verdict is that Mark has no obligation to render aid to Matthew because of his ignorance of the fact that Matthew needs his aid (1980: 301). Lemos adds, “If . . . we accept the principle that ought implies can, and if in the situation envisaged Mark cannot aid Matthew because of his ignorance of the fact that Matthew needs his aid, then he cannot have an obligation to aid Matthew and in not doing so does not act wrongly” (1980: 302). So it is clear that Lemos takes lack of “knowledge that” to be obligation subversive.6 Against Lemos, one might argue that just as there is a distinction between its being true that something is the case and belief about what is the case, so there is a distinction between one’s being obligated to do something and one’s believing that one is obligated to do that thing. And just as something can be true without one’s believing that it is so, so one can have an obligation to do something without believing or knowing

20

that one has such an obligation. Suppose I ought to return some keys by a certain date but completely forget about my obligation to do so. Suppose that, as a result of forgetting, I simply don’t believe that I ought to return the keys. Even though there is a sense in which I cannot return the keys – I don’t believe I have any reason to return them – it seems I can still have an obligation to do so. If what we overall ought to do is “objective” in one manner in which truth is – what’s true can obtain independently of what we believe is true – then, it appears, one will be drawn to the view that lack of “knowledge that” is not obligation subversive. I grant, though, that the sorts of considerations given against the view that motivational impairment, lack of know-how, and ignorance of facts are obligation subversive are not conclusive. One might, for instance, see informational impairment on a par with physical impairment and remain unpersuaded that the former is more normatively significant. One might balk at the suggestion that Tango has the right sort of control presupposed by obligation in virtue of the fact that she is physically and unintentionally able to open the safe. Lemos might simply insist that there is no incoherence and much sense in the idea that Mark, due to ignorance of facts, cannot save Matthew and, consequently, has no obligation to save him. Perhaps there is a problem with this piecemeal approach to understanding precisely what one is committed to when one claims that ‘ought’ implies ‘can.’ Maybe, figuring out the entailments of ‘can’ in principle K requires a more “holistic” approach that involves developing a theory of moral obligation, defending it against rivals, and if victorious, seeing what version, if any, of principle K it validates. This may, for example, be Zimmerman’s approach; his theory sanctions a version of K according to which one can do something even though one doesn’t know how to do it or doesn’t believe that one ought to do it (Zimmerman 1996). The proof of this approach is in the pudding, of course. Finally, ‘can’ in K is frequently affiliated with psychological ability; one factor that might make us say that someone cannot do something is that the person is psychologically unable to do that thing. Although Ken, for instance, is physically able to pick up the snake and believes that he won’t be harmed if he does pick it up, he just can’t bring himself to do it. In sum, conditional conclusions seem apt. There are stronger and weaker versions of Principle K. Among the requirements of a relatively weak version, “Moderate K,” are the requirements that one ought to do something only if (i) one is “relevantly situated” and one has the cooperation of the environment. These conditions are, roughly, associated with opportunity – loosely, one’s ability to do something – and are sensitive

21

to the fact that one can do something only if one does not lack the opportunity to do it. The concern is that not being relevantly situated or not having the cooperation of the environment would prevent one from exercising the pertinent abilities. If sharks would eat Leno before he could reach the drowning child, then even if he could save the child with no sharks around, in his circumstances, he lacks the opportunity to save the child; circumstantial factors prevent him from exercising relevant abilities. Similarly, internal factors, like psychological ones, can deprive one of opportunity. To save the child, Leno must be able to perceive that there is a child who is in need of help, and he must not suffer from a phobia that prevents him from entering the sea. We can adopt Kadri Vihvelin’s suggestion, understanding: opportunity broadly enough to include all circumstantial factors, both those inside and outside the skin, which would enable someone to exercise a relevant ability. On this view, someone lacks the opportunity to do x just in case there is something that prevents or would prevent a successful exercise of the relevant ability, regardless of whether it’s something external ( locked door, chairs, lack of piano) or something internal ( broken legs, cramps, state of unconsciousness).7

Moderate K entails that an agent S ought to perform an action A only if S has the opportunity to perform A; nothing prevents or would prevent S from successfully exercising the relevant abilities to perform A. Two other requirements of Moderate K when S ought to do A include (ii) S’s being physically and psychologically able to do A; and (iii) A’s accomplishment not being “strictly out of S’s control.” It should be stressed that the physical ability condition of Moderate K does not entail that S has the pertinent skill or know-how concerning what S is physically able to do. This is important as some people understand having an ability as entailing having the germane skills. Witness, for example, Vihvelin’s account of ability: I’ve been relying on our intuitive understanding of abilities as skills, capacities, or know-how – whatever it is that ensures that someone’s doing of something is not just a lucky fluke. . . . [S]omeone has an ability (as distinct from a mere capacity) to do x only if it’s in some sense under her volitional control whether or not she does x. We make judgments about ability on the basis of evidence of a reliable causal correlation between someone’s attempts to do x and the success of her attempts. There is a continuum of cases, ranging from the person with no ability to do x who keeps trying and sometimes gets lucky, to the person just learning to do x who succeeds more often, to the person highly skilled at x-ing, who typically succeeds most often. . . . What, then, is it to have the ability to do x? . . . I propose

22

the following [call it “volitional ability”]: someone has the [volitional] ability to do x just in case it’s true that there are some reasonably specifiable circumstances C . . . such that if she tried, in circumstances C, to do x, she would ( probably) succeed.8

Presumably, Vihvelin would say that Tango, in Safe2, does not have the volitional ability to open the safe – he just gets lucky when he opens it – although, she would say, he has the physical capacity to open it. The ability condition (ii) of Moderate K, we can now say, does not presuppose that the agent has the volitional ability to do whatever is of concern. To illustrate the luck requirement (iii) of Moderate K, if it were always the case that when an agent formed a here-and-now intention – that is, an intention to perform an action right away – it became highly improbable that she would act as she intended, then even if she did succeed in doing something, her accomplishment of it would be strictly out of her control. To formulate this requirement more revealingly, we can introduce rudiments of the notion of proximal control. Proximal control is one sort of control that an agent can exercise over an event he brings about or a state, such as a doxastic one, he might want to be in. This sort of control concerns the direct causal production of agent-involving events such as the agent’s having certain values, desires, and beliefs, his making a certain evaluative judgment, his forming a certain intention or arriving at a certain decision, his executing an intention, and his performing a certain action. In our example, the agent’s acting in accordance with her here-and-now intention is, pretty clearly, a matter of luck. I suggest that it is so as she lacks proximal control over the execution or translation of her intention into action. Similarly, it is easy to imagine circumstances under which the agent lacked proximal control over the very intentions that she forms; in these cases, the formation of her intentions would be strictly out of her control. Bearing in mind that according to Moderate K, motivational, “knowhow,” and “know-that” impairment are not obligation subversive, Moderate K can finally be summarized in this way: Moderate K: Agent S ought to do something A, only if S has the opportunity to do A, S is physically and psychologically able to do A, and A’s accomplishment is not “strictly out of S’s control.” The requirements of a strong version – “Strong K” – add to those of the moderate version by including motivational, “know-how,” and “know-that” capacities:

23

Strong K: Agent S ought to do something, A, only if S has the opportunity to do A, is physically, motivationally, and psychologically able to do A, has the relevant “know-how” to do A, is not ignorant of germane facts (this is meant to capture the “know-that” requirement), and A’s accomplishment is not strictly out of S’s control. Strong K implies that if S ought to do A, then S has both the opportunity and volitional ability to do A. Finally, note that both Moderate and Strong K, given the clause about “strict control,” imply that it can happen that one does A and yet “cannot” do A. This certainly has an odd ring to it. But it is acceptable if it is borne in mind that ‘can’ is meant to capture a notion of control. It may well be that, by sheer chance, I do succeed in firing up the boosters of the rocket even though I lack the relevant skill. Hence, if, for example, Strong K is true, although I do fire up the boosters, I don’t have “obligationgrounding control” over doing so. Fortunately, given the concerns of this work, it is not incumbent to decide between Moderate and Strong K. Indeed, I’ll argue that regardless of which version is true, although determinism is incompatible with deontic morality, some versions of indeterminism are not. But before we move on to these issues, we need to clarify the control requirements of deontic acts. In the next chapter, I develop the view that such acts require “dual” control – the powers both to perform and to refrain from performing the relevant actions.

24

3 Frankfurt-Type Cases and Deontic Control

In a seminal paper, Harry Frankfurt advanced the now famous “Frankfurttype example” primarily to argue against the Principle of Alternative Possibilities (PAP). The principle, which purports to specify the sort of freedom or control required for moral responsibility, is formulated by Frankfurt as follows: PAP: A person is morally responsible for what he has done only if he could have done otherwise (Frankfurt 1969: 829). In addition to revealing the lessons that they do about responsibility, Frankfurt-type examples are especially useful in probing the control requirements for deontic anchors. I appeal to them to defend a thesis about deontic control that says that deontic anchors require freedom to do otherwise: No one can perform an action that is right, wrong, or obligatory unless it is an action that one could have avoided performing. I am convinced that Frankfurt-type examples exert commanding pressure against PAP, though I shall not argue for this view in this book. If PAP is indeed called into question by Frankfurt-type cases, and if deontic morality requires “two-way” control, then moral responsibility and deontic morality are asymmetric with respect to the requirement of alternative possibilities: Whereas one can be morally responsible for an act that one could not have avoided performing, one cannot perform an act that has primary deontic properties unless one could have avoided performing it. In this chapter, after introducing Frankfurt-type examples, I propose two different lines of reasoning for the view that deontic morality requires two-way control.

25

FRANKFURT-TYPE EXAMPLES

A standard Frankfurt-type example is one in which an agent seemingly has various alternatives from which to choose. He makes his choice and performs the action in question not knowing that, had he revealed even the slightest inclination to act differently, something, a “counterfactual intervener,” would have forced him to act as he in fact did. Since he acted on his own, in that the intervener did not play any role in his action – it’s just that the intervener would have forced him to act as he did had he shown any signs of acting differently – it seems that nothing stands in the way of his being morally responsible for his deed (Frankfurt 1969: 834–7). In one variation of this type of case, there is no counterfactual intervener. Rather, there are two operating sequences that can culminate in the agent’s performance of the relevant action, one deterministic and the other indeterministic. The latter is indeterministic in virtue of its being the case that, given exactly the same past and the laws of nature, the agent could have decided one way or another. The indeterministic sequence actually leads to the end result, but the situation is such that the deterministic sequence would have resulted in the same sort of outcome if the indeterministic sequence had not. Here is one incarnation of such a case recently advanced by Mele and David Robb: At t1, Black initiates a certain deterministic process P in Bob’s brain with the intention of thereby causing Bob to decide at t2 (an hour later, say) to steal Ann’s car. The process, which is screened off from Bob’s consciousness, will deterministically culminate in Bob’s deciding at t2 to steal Ann’s car unless he decides on his own at t2 to steal it or is incapable at t2 of making a decision (because, for example, he is dead by t2). (Black is unaware that it is open to Bob to decide on his own at t2 to steal the car; he is confident that P will cause Bob to decide as he wants Bob to decide.) The process is in no way sensitive to any “sign” of what Bob will decide. As it happens, at t2 Bob decides on his own to steal the car, on the basis of his own indeterministic deliberation about whether to steal it, and his decision has no deterministic cause. But if he had not just then decided on his own to steal it, P would have deterministically issued, at t2, in his deciding to steal it. Rest assured that P in no way influences the indeterministic decision-making process that actually issues in Bob’s decision. (1998: 101–2, notes omitted)

In this case, the verdict that Bob is morally responsible for his choice and subsequent action even though he could neither have decided, nor, having so decided, then done otherwise, seems reasonable.1

26

A REQUIREMENT OF ALTERNATIVE POSSIBILITIES FOR WRONG ACTIONS

What do these examples suggest, if anything, about the sort of control required for deontic acts? Bob cannot, in his situation, refrain from stealing Ann’s car. With Black having initiated deterministic process P, although Bob has the ability to refrain from stealing the car, he lacks the opportunity to refrain from doing so. Now, the deontic principle: OW: Agent S ought to do [not to do] A if and only if it is morally wrong for S not to do [to do] A; a fundamental, groundwork principle of moral obligation, in conjunction with K, our control principle for moral obligation, K: S ought to do [not to do] A only if S can do [refrain from doing] A (temporal indices have been omitted), entails that there is a requirement of alternative possibilities for overall wrong actions: WAP: It is morally wrong for S to do [not to do] A only if S can refrain from doing [do] A. According to WAP, no one is able to perform an action that is (overall) wrong unless one is able to refrain from performing it. Assume (as I have done) that the sense in which, according to K, one must ‘be able’ to do something to have a moral obligation to do it is the same sense in which, according to WAP, one must ‘be able’ to do otherwise for what one does to be wrong. Assume, further, that this same sense of ‘being able’ to do something is operative in the Frankfurt example in which Bob is unable to do something other than what he in fact did. I see no compelling reason to deny these assumptions. As Bob cannot refrain from stealing Ann’s car, WAP and the facts of the case entail that Bob’s stealing the car is not (overall) wrong. This is perfectly consistent with other moral appraisals such as Bob brings about something bad by stealing, or as Frankfurt would presumably urge, Bob is morally responsible for stealing the car, or as I believe, Bob is blameworthy for stealing Ann’s car. PRINCIPLE CK AND WRONGNESS

So far, I have shown that there is a requirement of alternative possibilities for deontic wrongness; no action can be wrong for a person unless

27

she could have refrained from performing it. But this does not suffice to establish the stronger view that under conditions in which one lacks alternatives, as in Frankfurt-type cases, nothing can be wrong for one. For consider a world, w, in which agent S has no alternative to doing action A, S can do A, and S does A.2 In w, S cannot refrain from doing A. We might wonder about the very normative status of the state of affairs, S’s refraining from doing A. Is it wrong (in w)? WAP entails that it is wrong for S to refrain from doing A only if S can do A. But as S can do A in w, it might be objected that nothing stands in the way of S’s refraining from doing A being wrong for S in w. A more detailed development of this concern might be illuminating. OW and K won’t allow us to infer that S’s refraining from doing A is not wrong for S in w. For, from these two principles, we can infer that it is wrong for one to refrain from doing A only if one can do A, but this condition is satisfied by S in w. We cannot infer from OW and K what would be needed to establish that S’s refraining from doing A is not wrong for S in w: WC: It is wrong for one to do A only if one can do A; and it is wrong for one to refrain from doing A only if one can refrain from doing A. To establish WC we need a “complement” to K, CK, that says: CK: If one ought to do A, then one can refrain from doing A (and if one ought not to do A, then one can do A). OW and CK entail WC. As WC implies that it is wrong for one to refrain from doing something only if one can refrain from doing that thing, and as S cannot refrain from doing A in w, if WC is true, then there would be an easy route to the conclusion that S’s refraining from doing A is not wrong for S in w. If WC is false, though, it can be wrong for S to do A even if S can’t do A, and it can be wrong for S to refrain from doing A even if S can’t refrain from doing A. In other words, even if a person lacks alternative possibilities in some world, some things, it might be claimed, can still be wrong for that person in that world (provided WC is false).3 As I have not established the truth of CK, I have not shown that WC is true. But then WC could be false, and if it is, nothing that I have yet said establishes the stronger view that, in a world in which a person lacks alternative possibilities, no state of affairs can be wrong for that person.

28

THE PLAUSIBILITY OF PRINCIPLES CK AND WC

Well, are CK and WC true? I believe that these two principles are just as plausible as K. I’ll advance various considerations in their support apart from the following preliminary comment. When we ponder what we should do from the moral point of view, we ask: Which of our alternatives is, say, obligatory, permissible, or wrong? If something is not a genuine option for one, then, it seems, we standardly believe that it cannot be obligatory, permissible, or wrong for one though, of course, it might be obligatory, permissible, or wrong for others. If this is roughly correct, then it appears that if A is wrong for S, then A is an alternative for S; and A is an alternative for S only if S can do A.4 Addressing WC first, suppose the diabetic, strapped to his chair, cannot eat the piece of cake topped with sugar-rich frosting. Then it’s false that he ought to eat it, given K. Suppose one insists that it is or can still be wrong for him to eat the piece even though he can’t eat it. It would then seem that though having an obligation to do something, A, presupposes having control over A-ing, its being wrong for one to do A does not. But why the asymmetry regarding metaphysical presuppositions of control? After all, wrongness and obligatoriness are normative appraisals of the same family; they are deontic normative statuses and thus it would seem that, barring cogent explanation to the contrary, the control-relevant presuppositions of the one should also be those of the other. An analogy might help. Praise- and blameworthiness are different normative appraisals of the same ilk. One augments one’s moral worth by doing something if one is praiseworthy for doing it, and one diminishes one’s moral worth by doing something if one is blameworthy for doing it. But it would be farfetched, to say the least, if praiseworthiness for an action required one’s having a kind of control over that action that is fundamentally different than the kind of control required for blameworthiness for an action.5 Turning, now, directly to CK, the first consideration in its favor, surfaces upon closer scrutiny of its second conjunct that says that if one ought not to do A, then one can do A. If one ought not to do something, then it is wrong for one to do that thing. CK’s second conjunct, consequently, is equivalent to the view that if it is wrong for one to do A, then one can do A. As claimed in support of this view, if we accept the principle that ‘ought’ implies ‘can,’ there is little reason not to accept the principle that ‘wrong,’ too, implies ‘can.’ Again, it is difficult to see why control requirements of deontic obligatoriness would differ, in this respect, from control requirements of deontic wrongness. The principle that ‘wrong’ implies

29

‘can,’ in turn, in conjunction with the principle (derived from OW) that if it is obligatory for one to refrain from doing A, then it is wrong for one to do A, yields yet another principle: To wit, if it is obligatory for one to refrain from doing A, then one can do A. In other words, the ‘wrong’ implies ‘can’ analogue of K, together with OW, generates the principle that there is a requirement of alternative possibilities for obligatory acts, and this is precisely what the first conjunct of CK implies. A second defense for CK, a “theory-fueled” one, also supports CK’s complement K. A highly promising theory of what it is for an act to be obligatory is Fred Feldman’s.6 The theory builds on the idea that at each time of moral choice, there are several possible worlds accessible to a person: There are, at the time, various ways in which a person might live out her life; for each of these complete “life histories,” there is a possible world – the one that would exist if she were to live out her life in that way. Roughly, a possible world is accessible to a person at a time if and only if it is still possible, at that time, for the person to see to the occurrence of that world. A world may be accessible to a person at a time, but once the person behaves in some way other than the way in which she behaves in that world, it is no longer accessible; it has been “bypassed.” Once bypassed, a world never again becomes accessible. As a person moves through life, she inexorably pares down the stock of worlds accessible to her. Making use of the notion of accessibility, one can say that a state of affairs is possible for a person as of a time if and only if it occurs in some world still accessible to the person at that time. Let ‘Ks,t,p’ abbreviate ‘there is a world accessible to s as of t in which state of affairs p occurs.’ ‘Ks,t,p’ is equivalent to ‘as of t,s can still see to the occurrence of p.’ Corresponding to the ‘K’ operator, which is a sort of possibility operator, is a ‘U’ operator, which is a sort of necessity (or “unalterability”) operator. ‘Us,t,p’ abbreviates ‘p occurs in every world accessible to s at t.’ Thus if p is a “hard fact” about the past, or if p will occur no matter what I do, or if p is a metaphysical or physical necessity, p is unalterable for me now. On Feldman’s theory, actions are morally judged not by appeal to the value of their outcomes, but by appeal to the values of the accessible possible worlds in which they are performed. Worlds may be ranked in accordance to a value relation; each world is as good as, or better than, or worse than, each other world. A world is best if no world is better than it is. Again, for purposes of ranking worlds by value, one can supply one’s favorite axiology. The theory can now be stated in this way:

30

MO: A person s ought, as of t, to see to the occurrence of a state of affairs p if and only if p occurs in all the intrinsically best worlds accessible to s as of t.7 Let ‘O,s,t,p’ abbreviate ‘p occurs in all the intrinsically best worlds accessible to s at t.’ MO validates a version of the Kantian principle that ‘ought’ implies ‘can,’ since Os,t,p entails Ks,t,p. Strikingly, it also turns out that unalterability implies obligation. That is, this is also true if MO is true: UO: Us,t,p entails O,s,t,p (Feldman 1986: 42–3). UO says, roughly, that if something is unavoidable for an agent as of a time, then it is obligatory for that agent as of that time. UO kindles a problem. In the Frankfurt-type case discussed above, it is unavoidable for Bob that he steals Ann’s car. Suppose the action that is unavoidable for Bob in his Frankfurt-type situation is that he kills Ann. Given this supposition, UO entails that it is obligatory for Bob that he kills Ann. Understandably, many will balk at this result. Feldman indicates that should one find UO unacceptable, one could introduce a new obligation operator, ‘O ∗ ,’ defined in this way: ‘O ∗ s,t,p’ is true if and only if p does not occur in all the worlds accessible to s as of t, but p does occur in all the best of these worlds (Feldman 1986: 43). ‘O ∗ s,t,p,’ then, means the same as ‘O,s,t,p and Ks,t,not-p.’ This type of moral obligation entails ‘can avoid.’ As Zimmerman has argued (1996: 27–31), a sort of theory structurally similar to MO but with emphasis on “deontic value” – whatever it is that such value consists in – rather than on intrinsic value is compatible with almost all substantive theories of obligation, whether consequentialist or nonconsequentialist. And if the theory does capture the nature of moral obligation, then there is a powerful theory-fueled incentive to endorse both K and its complement CK. A REQUIREMENT OF ALTERNATIVE POSSIBILITIES FOR DEONTIC MORALITY

We can now show that under conditions in which an agent lacks alternative possibilities, the agent’s actions, besides not being overall wrong, are neither overall obligatory nor overall right. For, if some action A is obligatory for some person, then failing to do A – an omission – is wrong for that person. But it is false that any action (or omission) in a world in which a person lacks alternative possibilities is wrong for any person, given WAP,

31

and so it is false that failing to do A is wrong for that person. Hence, in such a world (given CK), no action is morally wrong or obligatory for that person. Could some actions in this sort of world be morally permissible for our agent? From the fact that an action is neither wrong nor obligatory for a person, it does not follow that it is morally permissible for her. An alternative is that the action lacks a primary deontic property altogether, not being right, wrong, or obligatory; it is, we can say, “amoral” for the person. This alternative is more plausible than the supposition that, in a world in which an agent lacks alternative possibilities, actions are morally permissible for that agent. It’s more plausible for the simple reason that in such a world no action is wrong or obligatory for our agent, and so in such a world, if actions performed by our agent did have deontic moral status, they would be either amoral or morally permissible for our agent. Consequently, killing someone in cold blood, or any other heinous deed, if not amoral would be morally permissible for our agent. This is a result that resists acceptance. It seems, then, that if our agent lacks alternative possibilities, nothing can be right, wrong, or obligatory for her. Let’s collect results. The argument that there is a requirement of alternative possibilities for deontic morality depends on three principles: OW: It is obligatory for one to do something if and only if it is wrong for one to refrain from doing it. K∗ : If it is obligatory for one to do something, then one can do it. (From K.) CK∗ : If it is obligatory for one to do something, then one can refrain from doing it. (From CK.) These principles yield two further principles: WAP: If it is wrong for one to refrain from doing something, then one can refrain from doing it. WC: If it is wrong for one to do something, then one can do it. The arguments for WAP and WC, respectively, are simply these: Argument 1 (1) If it is wrong for one to do A, then it is obligatory for one to refrain from doing A. (From OW.) (2) If it is obligatory for one to refrain from doing A, then one can refrain from doing A. (From K.) (3) Therefore, if it is wrong for one to do A, then one can refrain from doing A.

32

Argument 2 (1) If it is wrong for one to do A, then it is obligatory for one to refrain from doing A. (From OW.) (2) If it is obligatory for one to refrain from doing A, then one can do A. (From CK.) (3) Therefore, if it is wrong for one to do A, then one can do A.

From the proposition, then, that obligatoriness requires alternative possibilities, conjoined with OW, we may derive the proposition that wrongness also requires alternative possibilities. There is no similar way to derive the proposition that rightness likewise requires alternative possibilities; for even if it is agreed that ‘may’ implies ‘can,’ there is no principle like OW that will allow us to infer that ‘may’ does imply ‘can refrain.’ Nevertheless, it is very plausible that ‘may’ does imply ‘can refrain.’ Otherwise, as we noted, inasmuch as obligatoriness and wrongness do require alternative possibilities, we are in danger of being burdened with the intolerable view that it is morally permissible for one to do whatever heinous acts one cannot avoid doing.

AN ALTERNATIVE ARGUMENT

Some might object to the strategy utilized above to show that obligatoriness, rightness, and wrongness require alternative possibilities. For one might claim that, when trying to establish that there is a requirement of alternative possibilities for all three primary deontic properties, it is nonproprietary to assume not only that ‘ought’ implies ‘can’ (see line 2 of Argument 1) but also that it implies ‘can refrain’ (see line 2 of Argument 2 ). I should, though, indicate, that line 2 of Argument 2 derives from CK and there are good reasons to accept CK. Indeed, when I defended CK, I hinted at another sort of argument for the view that deontic morality requires alternative possibilities. This second line of reasoning requires only that we assume that, just like ‘ought,’ ‘right’ and ‘wrong’ also imply ‘can.’ This is, as I suggested, a very natural assumption to grant once it has been granted with respect to ‘ought.’ Alternatively, barring cogent reasons to believe otherwise, if we assume that ‘ought’ implies ‘can,’ there seems to be little reason to deny that ‘right’ and ‘wrong’ also imply ‘can.’ Then one can argue from this assumption to the conclusion that all three primary deontic properties require alternative possibilities. As a crucial stage in the argument, we would modify Argument 2 in this fashion:

33

Argument 2 ∗ (1) If it is obligatory for one to refrain from doing A, then it is wrong for one to do A. (From OW.) (2) If it is wrong for one to do A, then one can do A. (From the ‘wrong’ implies ‘can’ analogue of K.) (3) Therefore, if it is obligatory for one to refrain from doing A, then one can do A.

I take the following principles, then, to be true or at least highly plausible: DO (“Deontic Obligatoriness”): If at t, S ought to perform A at t ∗, then at t S can do A at t ∗ and S can refrain from doing A at t ∗. DW (“Deontic Wrongness”): If at t, it is wrong for S to perform A at t ∗, then at t S can do A at t ∗ and S can refrain from doing A at t ∗. DR (“Deontic Rightness”): If at t, it is right for S to perform A at t ∗, then at t S can do A at t ∗ and S can refrain from doing A at t ∗. As I explained in the last chapter, ‘can’ in each of these principles refers to the “deontic can”; and ‘can’ in its deontic sense lends itself to at least two different interpretations, the moderate and the strong interpretation. One might worry about the fact that these conclusions do not, on their own, show that the action performed by the relevant agent (like Bob) in a Frankfurt-type case isn’t or can’t be wrong. For if ‘can’ in K expresses, for example, logical possibility, then Bob can (in his situation) refrain from stealing the car. To deal with this worry, the discussion regarding the requirements of K in Chapter 2 suggests that no matter what version of K is true, it will specify that if one ought to do something, then there shouldn’t be anything that prevents one from exercising the relevant abilities to do and to refrain from doing it; there shouldn’t, for example, be anything that prevents one from physically performing or refraining from performing that thing. In Frankfurt-type cases, although the agent can perform the relevant action, the counterfactual intervener (or some other factor like deterministic process P in the Mele/Robb example) ensures that he cannot refrain from performing it. In such a situation, though the agent has the relevant abilities to refrain from performing the pertinent action, he lacks opportunity to refrain from performing it; the intervener or some other factor guarantees that he cannot physically refrain from performing the germane action. Hence, in such a situation, the relevant action will lack a primary moral status – it won’t be obligatory, wrong, or right.

34

A crucial control requirement of deontic anchors has now been exposed: Such anchors require a sort of two-way control. Let’s distinguish two senses of ‘control.’ Agent S may be said to be in control of doing A if and only if (a) S can do A (S has “singular control” over A) or (b) S can do A and S can avoid doing A (S has “dual control” over A). Deontic anchors require dual control. As Bob in the Frankfurt-type case lacks dual control over stealing Ann’s car, his stealing of the car is not right, wrong, or obligatory. In contrast, many are inclined to believe that Bob is or can be blameworthy, and so morally responsible, for stealing the car although he has only singular control over stealing the car. If this is so, then we have reason to believe that the control requirements for deontic anchors and moral responsibility are asymmetric. Some might be unwilling to grant this asymmetry because they disagree either with the view that there is no requirement of alternative possibilities for moral responsibility or with the view that there is such a requirement for deontic anchors. Setting aside the former worry, in the following chapter, we will examine some avenues of resistance to the conclusion that deontic anchors require dual control. Specifically, in the present chapter I have presented three principles, OW, K, and CK, from which two further principles, WAP and WC, have been derived, and I have argued for CK. In the next chapter I give a defense of K and OW.

35

4 Control Requirements of Deontic Anchors: Some Objections

The upshot of the last chapter is that rightness, wrongness, and obligatoriness all require alternative possibilities. I have argued for this result by invoking a number of principles, including K and OW. These principles may, however, be regarded by some with suspicion. In this chapter, I first address various objections to K and then turn to concerns regarding OW. OBJECTIONS TO K AND REPLIES

I find principle K, particularly the moderate version, attractive, and in this I am not alone. K enjoys widespread support, the basis of which, I believe, is at least twofold. First, there is the deep-seated intuition had by many that if overall moral obligation expresses moral necessity or requirement, then one must be able to do the thing – one must have control over doing the thing – that one ought morally to do. Indeed, partly for this reason, K seems to have the status of a sort of axiom – a deontic one – that in the minds of many, any adequate substantive moral theory about what makes right acts right, wrong acts wrong, and obligatory acts obligatory, or any acceptable analysis of the concept of obligation, should “validate.”1 Second, as we have seen, some of the best theories about the nature of moral obligation, theories that, among other things, provide an analysis of the notion of obligation, have K as a theorem; K is, roughly, implied by these theories.2 Such a “theory-based” derivation of K meshes neatly with the initial intuition about control that one can’t be under a moral obligation to do something if one can’t do that thing. In addition, if one finds K attractive, then one should find no less attractive its cohorts that ‘wrong’ implies ‘can’ and ‘right’ implies ‘can.’

36

However, as the literature reveals, like any interesting philosophical principle, K also has its share of foes. It would be beyond the purview of this work to catalogue and discuss all of the major objections that have been leveled against K. Besides, it would be pretty pointless to do this as a lot of this work has been undertaken elsewhere. Further, as K has such widespread appeal, it would be profitable to see whether the sort of dual control that K and CK (or the wrongness analogue of K and OW) imply is required for moral obligation can be possessed by agents in deterministic or indeterministic worlds even granting the controversy over K. For these reasons, I shall be highly selective and confine attention to the sorts of criticism of K that I believe are particularly germane to the primary objectives of this work. These criticisms help, partly, to clarify aspects of my views on moral obligation. Moreover, the replies to each introduce principles or other elements that arguments yet to be presented will invoke. FRANKFURT-TYPE CASES AND K

We can appreciate why it might be thought that Frankfurt-type examples provide grist for the mill against K by reflecting on the following. I have argued that in his Frankfurt-type situation, (i) Bob’s stealing Ann’s car is not wrong; and as I explained, this argument is, among other things, fueled by K. It seems, in addition, perfectly reasonable to suppose that (ii) Bob is blameworthy for stealing Ann’s car. But now one might invoke the principle that (iii) a person is blameworthy for performing an action only if it is wrong for her to perform that action. The trouble is that (i), (ii), and (iii) are elements of an inconsistent triad; the conjunction of any two of these elements entails denial of the third, yet each seems true. If one is drawn toward accepting (ii) and (iii), then one will have to reject (i), presumably, by rejecting K.3 The first three arguments I consider below are each prompted by scrutiny of Frankfurt-type examples. My responses to them will make evident my preference that element (iii) rather than (i) of the inconsistent triad should be rejected. A Widerkerian Objection against K Considering something like the inconsistent triad, David Widerker (1991) opts (it appears) to reject the view that an agent (like Bob) in a Frankfurttype case is morally blameworthy for the germane deed. But his argument, with some modification, can easily be construed as an argument against

37

K and, hence, as an argument against element (i).4 Contrary to Frankfurt, Widerker submits that rejection of PAP, the principle that a person is morally responsible for what he has done only if he could have done otherwise, requires rejection of K. For according to Widerker, K entails the “part” of PAP that deals with moral blame: PAP2: An agent S is morally blameworthy for performing a given act A only if it is within S ’s power not to perform A. Widerker says that PAP2 follows from K, specifically, that if S ought to do other than A, then S can do other than A, when taken in conjunction with element (iii) of the inconsistent triad that Widerker claims is a necessary truth. He calls this element “B2” and parses it in this way: B2: An agent S is morally blameworthy for performing a given act A only if S has a moral obligation not to perform A. (1991: 223) As K and B2 entail PAP2, if PAP2 is false, and B2 is a necessary truth, then PAP2 cannot be renounced without renouncing K. But not only is B2 not a necessary truth, as I have argued in another work (1998a), it is false. I will offer considerations against B2 in due course, but for now, I confine my remarks to the following. Widerker claims that the moral-praise analogue of B2: B1: An agent S is morally praiseworthy for performing a given act A only if S has a moral obligation to perform A. is false “due to the possibility of an act being supererogatory” (1991: 223). Though there is disagreement on the analysis of the concept of the supererogatory, there is general consensus that a supererogatory action is morally optional in that it is not overall morally wrong or overall obligatory; it is morally significant; and it goes “beyond duty’s call.” A commonplace example of supererogatory behavior is that of the heroic soldier’s sacrificing his life by throwing himself on the grenade to save his friends (Urmson 1969: 63). It is standardly agreed that persons can be deserving of moral praise for their supererogatory deeds even though such deeds are not morally obligatory. Supererogatory actions, then, provide good grounds for rejecting B1. But if there are supererogatory acts, or at least if there can be such acts, and these suffice for impugning B1, then suberogatory acts, if there are some, or at least if there can be some, should also suffice for impugning B2. Suberogatory acts, the roughly symmetric flip sides of supererogatory

38

ones, are morally optional, not morally indifferent, and are offensive or fall short of decency. Mowing one’s lawn early Sunday morning, or annoying the person sitting next to one in the meeting by smoking although there are no rules forbidding smoking, may well qualify as suberogatory. (See Driver 1992: 288.) In yet another example, suppose Joy is aware that others are waiting for her table in the crowded caf e´, mistakenly believes she is doing wrong by lingering on, intentionally lingers on in light of the belief that she is doing wrong, but really suberogates by lingering on. Intuitively, I don’t see why, in this instance, Joy is not blameworthy for lingering on. Hence, if one is willing to grant that supererogatory acts cast B1 into question, I see little reason for one to deny that suberogatory acts cast B2 into question. It might be objected that there are different kinds of negative evaluation, not all of which are allotments of blameworthiness. So, for instance, perhaps Joy is, intuitively, condemnable for lingering on or not giving up her table, but she is not blameworthy. But this objection is far from decisive. I concede that there are different sorts of moral appraisal and that the conditions of satisfaction for some vary from those for others. However, without an adequate analysis of the various evaluative concepts like those of blameworthiness and condemnability, it is difficult to say conclusively which evaluations apply and which do not in the case under consideration. But even if we grant that, intuitively, Joy is condemnable, there is nothing about the case that suggests that, intuitively, she cannot also be blameworthy. In fact, there are strong grounds to suppose that Joy is deserving of blame. For blameworthiness, it appears, is intimately associated with the view that, in doing something for which one is blameworthy, one reveals what one morally stands for, as one acts in light of pertinent moral beliefs. This “revealing” (or “disclosing,” or, if one wants, “expressive”) aspect of blameworthiness is not something idiosyncratic to my view of responsibility. It finds expression, for example, as Gary Watson explains, in John Dewey’s work (Watson 1996: 227–8). Dewey writes: [W]hen any result has been foreseen and adopted as foreseen [by the agent], such result is the outcome not of any external circumstances, not of mere desires and impulses, but of the agent’s conception of his own end. Now because the result thus flows from the agent’s own conception of an end, he feels himself responsible for it. . . . The result is simply an expression of himself; a manifestation of what he would have himself to be. Responsibility is thus one aspect of the identity of character and conduct. We are responsible for our conduct because that conduct is ourselves objectified in actions. (1957: 160–1)

39

Watson goes on to distinguish between two perspectives or faces of responsibility, one of which he dubs the “self-disclosure” view. This view holds that an agent is morally responsible insofar as he has the capacity to choose ends freely and act in accordance with them. Watson says: The self-disclosure view describes a core notion of responsibility that is central to ethical life and ethical appraisal. In virtue of the capacities identified by the self-disclosure view, conduct can be attributable or imputable to an individual as its agent and is open to appraisal that is therefore appraisal of the individual as an adopter of ends. Attributability in this sense is a kind of responsibility. In virtue of the capacities in question, the individual is an agent in a strong sense, an author of her conduct, and is in an important sense answerable for what she does. (1996: 229)

We can reasonably assume that Joy’s conduct flows from her “own conception of an end” and is an expression of herself; a manifestation of what she would have herself to be. Further, as her action is freely performed in the absence of relevant ignorance and is attributable to her as its “author,” there is much about the case to support the intuitive judgment that Joy is deserving of blame for her action. Widerker claims that even if we rule out supererogatory acts, the asymmetry between praise- and blameworthy acts still holds for “praising the agent for performing the act implies that the agent had a moral obligation to perform the act, whereas blaming him for an act implies that he had a moral obligation not to perform it” (1991: 224). But here, again, there are concerns. On the one hand, perhaps the concepts of praise and blame in question are simply those of being praise- and blameworthy respectively. Widerker’s asymmetry would then amount to the view that when an agent is deserving of praise for performing an act, the agent has a moral obligation to perform that act, and when an agent is deserving of blame for performing an act, the agent has a moral obligation not to perform it. But surely this gives us no independent reason to affirm the proposed asymmetry that, as we are now construing it, implies that whereas one can be praiseworthy for an act even if that act is not obligatory, one cannot be blameworthy for an act unless that act is wrong. Consider, however, that perhaps the praise and blame in question comprise overt praise and blame respectively. Such praise or blame consists in reaction to, or treatment of, a person on account of some deed performed by that person, the reaction being positive (like a pat on the back) in the case of praise, and negative (like a scowl or a reprimand) in the case of blame. But though overt praise and blame frequently track obligatoriness

40

and wrongness respectively, sometimes there is divergence; and where there is, overt praise for an act does not imply that its agent ought to have performed it, and overt blame for an act does not imply that its agent ought not to have performed it. So, for example, if I discover that Kildare gives sick Tango medicine B in light of the belief that he is doing his moral duty by giving B, I can duly overtly praise Kildare for giving B (and not just for attempting to cure Tango), even though, sadly and unbeknownst to anyone at the time, it happens that giving B is the wrong course of treatment. Analogously, if I discover that Deadly gives Tango medicine A in light of the belief that he is doing wrong by giving A (imagine that Deadly wants to see Tango suffer), then even if, by sheer luck, giving A turns out to be best for Tango, I can duly overtly blame Deadly for giving A. In summary, it appears that the Widerkerian argument against K is not without problems. An Objection from Counterintuitiveness I have argued that K and OW entail WAP, and WAP implies that it is wrong for a person to perform an action only if she could have refrained from performing it. But then, as Bob in his Frankfurt-type situation cannot refrain from stealing Ann’s car, this fact and WAP entail that Bob’s stealing Ann’s car is not wrong, an entailment, it might be claimed, that jars against intuition. To appreciate the counterintuitiveness of this sort of result, simply imagine that the relevant agent is Jones and the relevant act she cannot refrain from performing in her Frankfurt-type predicament is killing White. It may be submitted that as Jones kills in cold blood, if anything is wrong, such killing is wrong. But if OW and K are both true (and hence WAP is true), then Jones’s killing is not wrong. The right conclusion to draw, it might then be proposed, is that, as OW is correct, K is the culprit. But this objection is questionable. Why believe that, in her circumstances, Jones’s killing is wrong? Perhaps what underlies this assumption is that Jones is clearly deserving of blame for her heinous deed, and that one can’t be deserving of blame for something that is not overall wrong. However, as I have indicated, while I accept the verdict of Jones’s being to blame, I (like others) reject the principle – Widerker’s B2 – that one is to blame for an action only if it is overall wrong. Indeed, Frankfurt-type cases themselves suggest that this principle is false. One may contest that a good enough explanation for the commonsense belief that Jones’s killing White in cold blood is wrong is simply

41

that Jones was responsible for and wholeheartedly endorsed her choice and subsequent action of killing. This explanation, though, is not without foibles. First, commonsense beliefs can be challenged by theoretical discoveries. Some analogies might help. The strategy I have utilized that invokes Frankfurt-type examples to establish a requirement of alternative possibilities for moral obligation or wrongness is, really, no different in spirit than, say, the strategy an indeterminist might utilize to discredit one moral that Frankfurt and others wish to draw from such examples, to wit, that freedom to do otherwise is not a requirement for moral responsibility. Suppose, for example, that an indeterminist believes that some version of PAP is true partly because it seems intuitively plausible and it has theoretical support.5 Suppose, now, that one advances a Frankfurt-type case to question PAP: One proposes that in such a case, the central player, Jones, is morally responsible for killing White. “Look,” one might say, “Jones wholeheartedly endorsed his choice and action; he killed in cold blood. So he must be responsible. What could be more commonsensical than this?” The indeterminist might retort that if the relevant version of PAP is true, and the case really is one in which Jones could not have done other than kill White in the right sense of ‘could not have done otherwise,’ then the commonsensical belief must be false. Or to take another parallel, suppose an analysis of the concept of obligation “verifies” a “no-conflicts” principle: It rules out the possibility of genuine conflicts of overall moral obligation. Then though common sense may sanction the belief that the brave soldier really has conflicting overall obligations, one to take care of his mother and the other to fight for his country, the no-conflicts principle and the facts imply otherwise. And if there is support for the analysis that entails the no-conflicts principle, the surprising implication that there are no genuine conflicts of obligation, an implication that may not mesh with commonsense beliefs, might in fact be true. A second concern with the proposed explanation of the commonsense belief that Jones does wrong by killing White in the Frankfurt-type case – the explanation that Jones was responsible for and wholeheartedly endorsed her choice of killing White – is the following. The proponent of the explanation grants that Jones is responsible for killing White and then exploits this fact to conclude that Jones must thereby do wrong in killing White. But this maneuver would be tenable if it were plausible to assume that the conditions that require satisfaction for one sort of moral appraisal to be apt are the very same conditions that require satisfaction for another variety of moral appraisal to be apt. But this assumption is not

42

true. Michael Slote (1990), for example, has proposed that many aretaic appraisals are possible even if agents lack freedom to do otherwise; but I have proposed that “deontic appraisals” when agents lack such freedom are not possible. Or, on the supposition that an act is overall evil if and only if were it performed, it would produce more intrinsic badness than intrinsic goodness, we can grant that the axiological appraisal that Jones does overall evil in killing White may well be accurate without conceding that, for example, Jones has a thoroughly bad character, or that Jones does wrong in killing White. Fischer against K One might still, however, suspect that Frankfurt-type examples hold the key to launching a potent attack against K. In response to the question of what justification could be offered for K, John Fischer writes: It is most natural . . . to say that the maxim is valid because if it were not, then there could be cases in which an agent ought to do X but in fact cannot do X (and never could do X ). Thus, given the connection [forged by Widerker’s B2] between its being the case that an agent ought to do X and the agent’s being blameworthy for not doing X, there could be cases in which an agent is blameworthy for not X-ing and yet he cannot X. And this seems unfair. (1999a: 124)

Fischer’s position seems to be this: A proponent of K might claim that (1) K is valid because if it were not, there could be cases in which S ought to do X but S cannot do X.

But then, assuming this principle: (2) S blameworthy for doing X [not doing X ] only if S ought not to [ought to] do X,

is true, it follows from (2) and K that: (3) S is blameworthy for not doing X (alternatively, S is blameworthy for not X-ing) only if S can do X.

In sum, Fischer’s suggestion is that the “most natural justification” for K is that: (4) K is valid because if it were not, there could be cases in which S is blameworthy for not X-ing even though S cannot do X.

43

Fischer, though, rejects (4) by invoking certain sorts of Frankfurt-type cases involving omissions: I believe that there are Frankfurt-type omissions cases that are relevantly similar to Frankfurt-type cases with respect to actions. That is, there are cases in which an agent is morally responsible for not X-ing although he cannot in fact X. Some of these are cases in which an agent is blameworthy for not X-ing and yet he cannot X. In fact, I believe that anyone who accepts the Frankfurt-type action cases must accept that there are such omissions examples. Thus, it is precisely the basic intuitions elicited by the Frankfurt-type cases which show that the most natural justification of the “ought implies can” maxim is faulty. It is therefore not ad hoc for anyone who accepts the standard interpretation of the Frankfurt-type cases to reject the “ought implies can” maxim. (1999a: 124–5, notes omitted)

But this objection against K also has its share of weaknesses. First, I agree with Fischer that certain sorts of Frankfurt-type examples involving omissions cast considerable doubt upon (4). (See, for e.g., Fischer and Ravizza 1997; and Haji 1992.) However, defenders of K who reject (2), which is just Widerker’s B2, can surely accept (1) – perhaps on the basis of accepting an analysis of obligation that entails K – without thereby being committed to accepting (4) as a justification for K. Second, recall that PAP2 is the version of the principle of alternate possibilities that says that a person is blameworthy for performing A [or for failing to perform A] only if she has the power not to perform [or to perform] A. It appears that Fischer’s objection rests on the presupposition that: (5) K is relevantly just like PAP2. So K stands or falls with PAP2; if PAP2 is falsified by a Frankfurt-type case, such a case must falsify K too.

For if one renounced (5), one might well concede that Frankfurt-type cases involving omissions call PAP2 into question without conceding that such cases call K into question. Why, though, believe that Fischer’s objection against K assumes that K stands or falls with PAP2? With (4) in mind, let’s reformulate his objection in this way: (1) If K is valid, it’s false that there could be cases in which S is blameworthy for not X-ing even though S cannot do X. (2) It’s not the case that it’s false that there could be cases in which S is blameworthy for not X-ing even though S cannot do X. (This is where appeal is made to Frankfurt-type cases involving omissions.) (3) Therefore, K is not valid.

44

But, given PAP2, we can recast line (1) in this way: (1∗ ) If K is valid, then PAP2 is true.

The rest of the argument would then unfold in this fashion: (2∗ ) It’s false that PAP2 is true. (3∗ ) Therefore, it’s false that K is valid.

(1∗ ), (2∗ ), and (3∗ ) surely strongly suggest that Fischer’s reasoning against K assumes that K stands or falls with PAP2. But K is not relevantly like PAP2. Rather, K – a “power” or “control” principle for deontic acts – is pertinently like the following control principle, BR: S is blameworthy (or, more generally, morally responsible) for performing A only if S has the power to perform A. BR simply affirms the connection between control and blameworthiness or moral responsibility; its gist is that one can’t be responsible for an action unless one has control over it. Not only is BR true – or at least it has very widespread support – BR is not falsified by Frankfurt-type examples involving actions. For in such examples, the central character of interest, like Jones, does seem morally responsible for, say, killing White, and, pretty clearly, Jones does have control over killing White. As Zimmerman has indicated, though, PAP is a contraction of BR and: FAP: It is within S’s power to perform [not to perform] A only if it is within S’s power not to perform [to perform] A. (Zimmerman 1996: 86) A Frankfurt-type example undermines FAP: Although it is within Jones’s power to kill White (indeed, she does kill White), it is false that it is within her power not to kill White. Alternatively, if (as Fischer concedes) a Frankfurt-type example undermines PAP2, FAP and BR entail PAP2, and BR is true, such an example thereby undermines FAP. But Frankfurttype examples leave OW intact, and the link, as Zimmerman indicates, between obligation and alternative possibilities is preserved by an appeal to OW and not FAP. Further, as Zimmerman stresses, there is no acceptable responsibility analogue to OW. In particular, the principle PB: S is morally praiseworthy for performing [not performing] A if and only if S is morally blameworthy for not performing [ performing] A, is clearly false (Zimmerman 1996: 86–7).

45

A Direct Threat against K from Frankfurt-Type Examples One might urge that Frankfurt-type examples seem to work just as well when applied to questions of obligation as when applied to questions of responsibility. If the presence of a counterfactual intervener, or some other condition, is insufficient to make it the case that the agent avoids blame, surely it is similarly insufficient to make it the case that the agent avoids wrongdoing. But here I would simply underscore, yet again, the fact that to deny that the agent acts wrongly is not to deny that he or she brings about something bad. Nor, of course, is it to deny that he or she does something that would be wrong if only there were other options available. Regarding the first of these considerations, appraisals of intrinsic, extrinsic, or overall goodness and badness are often conflated with those of rightness and wrongness, respectively. (For elaboration, see Haji 1998a: 43–4.) Unlike such appraisals, people often distinguish between appraisals of blameworthiness and wrongness. It is readily acknowledged, for instance, that one can do wrong without being blameworthy. In the Mele/Robb case, Bob brings about something bad in stealing Ann’s car. This judgment is true even if Bob has no alternatives. But then given conflation of the bad and the wrong, one might think that Bob does wrong and he does so even though he has no alternatives. Further, appraisals of responsibility, as I have suggested, disclose what an agent morally stands for in connection with performing an action (or intentionally failing to perform it). In this fashion, responsibility appraisals are “agent-focused.” An agent can disclose what she morally stands for in relation to an action she performs even when she lacks, as she would in a Frankfurt-type situation, alternative possibilities. So, for example, when one is morally blameworthy for an action, one expresses ill will toward another, and it is the expression of such ill will that (in part) sanctions the judgment that one’s moral worth with respect to the relevant deed has been diminished. Analogously, when one is morally praiseworthy for an action, one expresses good will toward another, and it is the expression of such good will that (in part) validates the judgment that one’s moral worth vis-`a-vis the relevant deed has been augmented. Such expression of good or ill will does not presuppose alternative possibilities. There is, then, much wisdom in Frankfurt’s thought (1969: 836) that since an agent in a Frankfurt-type case perceives her situation in just the manner in which she would if there were no counterfactual controller lurking in the wings, (agent-focused) appraisals of responsibility

46

should not, in such cases, be affected merely because the agent lacks alternatives. In contrast, appraisals of rightness, wrongness, and obligatoriness, or “deontic appraisals,” as I shall say, are more “act-focused.”6 They are not connected, in any essential fashion, with how the agent perceives the situation or with germane motivations or beliefs. An agent, for instance, may act “from” virtue on a certain occasion and (assuming the act is not virtue-deficient or not significantly so) be commendable for so acting, consistent with doing (deontic) wrong on that occasion. Whether an agent’s act instantiates a primary deontic property, then, is dissociated with, for example, whether the agent performs that action in the belief that it is wrong, or more generally, with the agent’s “moral perception” of her act. Being act-focused as they are, deontic appraisals turn primarily on conditions that the act itself must satisfy, and these conditions may not be met despite the agent’s believing otherwise or failing to believe that the conditions are not met. This discrepancy between appraisals of responsibility and those of deontic morality – the fact that the former are agent-focused whereas the latter are act-focused – gives us good reason to believe that the freedom requirements of deontic morality may well differ from those of moral responsibility. Hence, even though Frankfurt-type examples cast doubt on PAP, it is not clear that they cast doubt on K.

SELF-IMPOSED IMPOSSIBILITY AND K

Examples of self-imposed impossibility in which a person (freely) takes deliberate steps to cause herself to be unable to do something later present a serious challenge to K.7 Suppose, for instance, that Lottie’s spouse-tobe refuses to marry Lottie unless she gets rid of “Lottie’s little pest,” her baby. She believes that it is wrong to kill her baby but still settles on the decision to do so. To help with the execution of this vile decision, Lottie imbibes a pill, in the full realization that having taken it, she won’t be able to refrain from squeezing the trigger. Still, one might insist, Lottie ought not to squeeze the trigger (or kill the baby) when she does, even though she cannot, at the time, refrain from squeezing the trigger. Further, surely, it might be claimed, Lottie is blameworthy for her monstrous deed; she cannot get off the hook so easily by disabling herself in the relevant way. But if she ought to refrain from her deed at the time of its performance, it follows that K is false, since K implies that if Lottie ought not to squeeze the trigger when she does, then she can refrain from squeezing the trigger.8

47

I agree, though, with Zimmerman’s proposal that careful attention to the distinctions between immediate and remote control, and immediate and remote obligation enables us to meet this objection. As Zimmerman explains, regarding the former distinction, it is common to ascribe a double time-index to “can” contexts, one to the “can” itself and one to the action on which the “can” operates. So I may say that I can now be in Chicago tomorrow, but if I leave for Zanzibar this afternoon, I won’t be able to make the Windy City tomorrow. In the phrase-form, ‘S can at t do A at time t ∗,’ Zimmerman proposes, the ‘can’ is immediate if t is identical with t ∗ but remote otherwise. Thus, we should understand the claim that “ought” implies “can” as the claim that if S ought at t to do A at t ∗, then S can at t do A at t ∗. And just as control can be remote or immediate, so too can obligation. I may have an obligation to lecture now, or I may have an obligation to lecture this afternoon.9 These distinctions pave the way for responding to the objection that Lottie ought, at the time she takes the pill and then again when she squeezes the trigger, to refrain from shooting the baby even though, at the relevant times, she cannot refrain from shooting the baby. What is in fact true is that prior to taking the pill, Lottie ought to refrain from shooting the baby (in part, because she can, prior to taking the pill, refrain from shooting the baby). After taking the pill, though, and so no longer being able to refrain from the shooting, it is no longer true that Lottie ought to refrain, and that it is wrong for her to shoot the baby. Of course, we can maintain that Lottie ought, up to the time of taking the pill, to refrain from shooting her baby later, and as this (remote) obligation is not satisfied, a wrong is done. But this is consistent with holding that as of the time she takes the pill, Lottie’s not shooting her baby is not obligatory for her. (See, e.g., Zimmerman 1996: 98–100.)10 That part of the objection that Lottie is blameworthy for her gruesome deed even though she cannot refrain from performing it at the time that she performs it is correct. It is important to see that the proponent of K need not deny this. For the control or freedom-relevant requirements of blameworthiness are quite reasonably taken to be satisfied by Lottie. Regarding this requirement, a “tracing” principle of the following sort is plausible: An agent, S, is blameworthy for A-ing only if S exerts “responsibility-grounding control” (“RG control”) in A-ing; or S intentionally takes deliberate steps, over which she has RG control, to see to it that, when she performs A, she will lack RG control in A-ing; or S intentionally does something, B, over which she has RG control, and is aware that, if she then performs A, she will lack RG control in A-ing.

48

We think this sort of tracing principle is reasonable, in part, because the agent intentionally executes a plan in the belief that by carrying through with the elements in her plan, she will achieve her end, a component of which is her not being able to refrain from doing something. Lottie’s goal, for instance, is to kill her baby by shooting him. If we assume, simply for illustration, that freedom to do otherwise is the pertinent sort of control required for responsibility, and Lottie cannot refrain from squeezing the trigger, the tracing principle allows us to conclude that Lottie is blameworthy for squeezing the trigger. She is blameworthy as her action traces to earlier steps over which she has control, and she takes these steps in the belief that their execution will eventually lead to attainment of her goal by ensuring that she won’t be able to refrain from squeezing the trigger. Important lessons are brought to light from this case of self-imposed impossibility. First, Lottie is blameworthy for squeezing the trigger even though it is false that, as of the time she takes the pill, she ought not to squeeze it. (Thus, given that one ought not to do something if and only if it is wrong for one to do that thing, though Lottie is blameworthy for squeezing the trigger, her squeezing of the trigger is not overall wrong.) But then such cases also call into question Widerker’s B2, the principle that an agent is blameworthy for performing an act only if that agent has a moral obligation not to perform it. Why can’t it be, though, that the connection between blameworthiness and overall wrongness be preserved by the view that, though Lottie’s act of squeezing the trigger is not overall wrong at the time of its performance, it is “derivatively wrong” as its “derivative wrongness” traces to an earlier act that is overall wrong – her taking of the pill? Let’s clarify this proposal by paying attention to the times in B2. So far, we have been concerned with this version of B2: T1B2: S is blameworthy at t1 (and thereafter) for doing A at t1 only if it is morally wrong at t1 for S to do A at t1. I have argued that Lottie’s case undercuts T1B2. The proposal under consideration calls for an assessment of the following version of B2: T2B2: S is blameworthy at t1 (and thereafter) for doing A at t2 only if it is morally wrong at t1 for S to do A at t2. Lottie’s case leaves T2B2 unaffected. At a time just prior to the time at which Lottie takes the pill, for instance, it is morally wrong (as of that time) for Lottie to kill the child (or to squeeze the trigger) at some later time. If one believes that Lottie is to blame for killing the child and one

49

favors something like B2, then one can claim that it is not T1B2 but, rather, T2B2 that is the appropriate B2-like principle that (at least partly) undergirds the judgment that Lottie is blameworthy for killing the child. But I think T2B2 is incorrect as well. Simply suppose that Lottie is in a “global” Frankfurt-type situation. In a such a situation, the counterfactual intervener (say) watches over every action of Lottie’s, and will intervene and compel her to perform an action of a certain sort if she shows any signs (again, let’s assume) of not performing an action of that sort. Further, in such a situation, it turns out that Lottie always performs the “right” type of action, thereby obviating the need for intervention on the part of the counterfactual intervener. In a situation of this sort, it would seem that Lottie is blameworthy for killing the child even though it is not wrong, as of any time, for Lottie to kill the child at the time when she does kill the child. One might still not give up and propose that a significant link between overall wrongness and blameworthiness can be preserved if K is appropriately modified. Commenting on principle K, that one is morally required to do a given thing only if one is able to do the thing, David Copp, for example, submits that the “ability condition” needs qualification, “for there are cases in which a person causes himself to be unable to do something, by getting drunk, for instance, yet is nevertheless morally required to do it” (Copp 1997a: 444). A suggestion that surfaced in conversation with Copp is that perhaps the following gloss of K can be used to sustain the view that Lottie is to blame for squeezing the trigger, and that her act of squeezing the trigger is overall wrong, even after taking the pill: KC1: S ought at t to do A at t ∗ only if S has the power at t to do A at t,∗ or it would have been in S ’s power at t to do A at t ∗ had S not wrongly done something, B, at a time, t0, before t, the doing of which prevented S from A-ing at t.∗11 KC1, however, won’t do. Assume all other conditions for its being the case that S has an obligation at t (5:00 p.m.) to do A at t ∗ (6:15 p.m.), save the condition expressed in KC1, are satisfied in the following scenario so that if KC1 is met, S does have an obligation at t to do A at t.∗ We are to ponder whether this assumption is tenable, as it would be, if it were true that, given the facts of the scenario, as of 5:00 p.m., I ought to visit Vick in the hospital at 6:15. The relevant facts are the following: At 4:30 p.m., on the way to the hospital, I drop in at Cheers and get so drunk that I am no longer able to make it to the hospital. I don’t get drunk in order to avoid seeing Vick; I just get carried away with the cognac. Assume, further, that

50

had I not stopped at the bar, I would have visited Vick. As KC1 is met (its second but not first disjunct is satisfied), I have an obligation, as of the time I’m intoxicated and, hence, unable to make it to the hospital at 5:00 p.m., to visit Vick at 6:15 p.m. I find this result untenable. I’m inclined to think that, as of the time that I am inebriated, I have no obligation to visit Vick, though at an earlier time, I did have this obligation. As a remote obligation was not fulfilled, I did (remote) wrong. Further, I may well be blameworthy for failing to visit Vick. Here is another gloss of K, which is again meant to show that Lottie ought not to squeeze the trigger even after taking the pill: KC2: S ought at t to do A at t ∗ only if S has the power at t to do A at t,∗ or it would have been in S’s power at t to do A at t ∗ had S not wrongly done something, B, with the intention of preventing S from being able to do A at t ∗ at a time, t0, before t, the doing of which prevented S from A-ing at t.∗ Again, assume that if KC2 is met, S does have an obligation at t to do A at t ∗. But KC2 is problematic as well. Suppose I borrow Carl’s watch and promise to return it on Sunday. The question, this time, is whether it is true that, as of 8:00 a.m. on Sunday (t), I ought to return the watch at 10:00 a.m. (t ∗ ) on the same morning. I’m aware that the old watch that once belonged to Carl’s father is highly valued by Carl. I hear some nasty (but false) rumors about Carl at Cheers on Saturday night. I believe these nasty things. Later that night, I throw Carl’s watch into the river partly to prevent myself from being able to return it the next day. As KC2 is met (its second disjunct is satisfied), I have an obligation, even as of the time I’ve thrown the watch into the river, to return the watch to Carl at 10:00 on Sunday. Again, I find this result unacceptable. Surely, as an advocate of K, one wants to say that, as of the time I toss the watch into the river, I don’t have an obligation to return the watch to Carl since I can no longer return it. I did, though, earlier have an obligation to return it, and because this obligation is not fulfilled, a (remote) wrong has been committed. The second important lesson to be learned from Lottie’s case of selfimposed impossibility is that it provides further reason to believe that freedom or control requirements for blameworthiness, or more generally, moral responsibility, differ from those for deontic morality. For one can, it seems, be blameworthy for doing something even though one can’t refrain from doing it when one does it. But it is false that one ought not, at a time, to do something if at that time one cannot refrain from doing that thing.

51

We can conclude that neither the arguments galvanized by reflection on Frankfurt-type cases nor the argument from self-imposed impossibility give us adequate grounds to renounce K. Let’s now turn to OW. PEREBOOM ON OW

OW, remember, says that one has a moral obligation to perform [not to perform] A if and only if it is morally wrong for one not to perform [to perform] A. Derk Pereboom has recently taken issue with the implication of OW that if it is wrong for one to perform some act, then one has a moral obligation not to perform it. He explains that while one half of OW: 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 extremely plausible, the other half: 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, is not quite so obviously true. He writes: I cannot 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. Yet there are cases in which it seems quite intuitive 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 you say to an animal-abuser “You ought not to abuse that animal,” but then you find out that he has a psychological condition (which he could have done nothing to prevent) that makes animal-abusing irresistible for him, so that he cannot help but abuse the animal. From my point of view, there is an appreciably strong pull to admitting that the “ought” judgment was false, but there is relatively little to denying that abusing the animal is morally wrong for him. (2001: 147)

Pereboom’s position, then, is that of the two judgments: P1: S ought not to have abused the animal, and P2: It was wrong for S to abuse the animal, P2 but not P1 is true. I think, though, that there are concerns with this stance. First, I admit that arguments that rely centrally on intuition are always tricky because what is intuitive for one may not be so for another.

52

With this in mind, suppose one asks: “Why think it’s false that S ought not to abuse the animal?” A reasonable response might well be: “Well, S couldn’t (through no fault of S’s own, let’s suppose) resist doing what S did. But ‘ought not’ implies ‘can refrain from.’ So P1 is false.” If something along these lines is what underlies rejection of P1, one might well wonder why it is not also the case that ‘is wrong’ implies ‘can refrain from.’ After all, strictly intuitively, an alternative way of saying that “S was wrong to do A” is “S shouldn’t have done A.” But then why doesn’t ‘is wrong’ imply ‘can refrain from’? Second, one might be tempted to affirm P2 on the basis that abusing the animal is something bad. But while one might grant this axiological judgment of badness, there is no easy overlap between badness and wrongness. An act might produce a great deal of badness or just be bad but still not be overall wrong. In addition, if the abuser did do something that is bad, an alternative to Pereboom’s position is to reject both P1 and P2 but to endorse the pertinent judgment regarding badness. Third, I think the most forceful rejoinder to Pereboom’s position is that his stance assumes something that is not warranted without cogent argument: The freedom-relevant requirements for the primary deontic property of wrongness differ from those for the primary deontic property of obligatoriness. Simply put, if ‘ought’ implies ‘can’ and ‘can avoid,’ why is it not plausible to suppose that ‘wrong,’ too, implies these things?12 GENUINE MORAL DILEMMAS AND OW

A genuine moral dilemma, if there is one, involves a conflict of two or more overall obligations that cannot all be satisfied, although each can individually be satisfied. It might be said that those who accept the possibility of such dilemmas would surely wish to deny OW. This is because OW implies that it is obligatory for one to do something if and only if it is wrong for one to refrain from doing it. But proponents of genuine moral dilemmas, it might be claimed, would want to say that it can be both obligatory to do something and obligatory, rather than wrong, to refrain from doing it. But actually, this is not so. Someone who accepts the possibility of moral dilemmas is highly likely to accept OW. After all, the lesson that is typically derived from the claim that it is obligatory for one to do some act while also obligatory to refrain from doing it is that one will do wrong no matter what one does. As Michael Zimmerman (unpublished manuscript) has indicated, what proponents of dilemmas reject is not OW but rather the following:

53

OW∗: If it is obligatory for one to do something, then it is not wrong for one to do it. Incidentally, accepting OW does not require accepting the following: ONOTP: It is obligatory for one to do something if and only if it is not permissible (right) for one to refrain from doing it. For to say that something is not permissible is not to say that it is impermissible (that is, wrong). This is because it is possible for some act to be neither permissible nor impermissible, as would be the case if the agent had no alternative possibilities (see M. Zimmerman, unpublished manuscript).13 The two lines of reasoning I advanced (in Chapter 3) for the view that deontic anchors require dual control, both of which invoke principles K and OW, have survived proposed criticisms of these principles. If there is indeed a requirement of dual control for deontic anchors, then it seems that such anchors will be imperiled by determinism. This is the topic of the next chapter. APPENDIX: YAFFE ON K

We have seen that Widerker, contrary to Frankfurt, proposes that if Frankfurt-type examples are counterexamples to PAP2 (an agent S is morally blameworthy for performing a given act A only if it is within S’s power not to perform A), they are equally counterexamples to K, as K entails PAP2. Gideon Yaffe has recently taken issue with this proposal. I believe Yaffe’s reservations are mistaken. It will be instructive to see why they are so. At the heart of Widerker’s proposal is the premise that K entails PAP2 and he argues for the premise as follows: Argument 1 (1) An agent S is morally blameworthy for performing action A only if S has a moral obligation not to do A. (2) S has a moral obligation not to do A only if S has the power to do other than A. (From K.) (3) Therefore, S is morally blameworthy for A only if S has the power to do other than A. (From (1) and (2).)

Responding to this argument, Yaffe says that “while a principle similar to the Maxim [the Maxim is that part of K that says that one ought to do something only if one can do it] does imply a principle similar

54

to PAP, it is not the case that the Maxim implies PAP – at least not for the reasons that Widerker suggests” (1999: 219). The fault in the argument, Yaffe explains, is that it conceals a step that “requires the severest scrutiny” (219). To expose this step, he reformulates the argument in this way: Argument 2 (1 ) An agent is morally blameworthy for what she has done only if she had a moral obligation not to do as she did. (2 ) An agent was morally obligated not to act as she did only if she was morally obligated to act differently from the way she did act. (3 ) An agent was morally obligated to act differently from the way she did act only if it was within her power to act differently from the way she did act. (From K.) (4 ) Therefore, PAP2. (From (1), (2), and (3).)

Yaffe claims that the concealed step (2 ) is false because “An obligation not to perform a particular action does not, in general, generate an obligation to do something else instead” (220). This claim, in turn, is partially supported by the following example: Imagine that there is some act that I am obligated not to perform at midnight while I am soundly asleep. According to (2 ), the fact that I was morally obligated not to act in a certain way at midnight implies that I was morally obligated to perform a different action at that time. There is only one action that I can possibly be construed as performing at midnight – sleeping – and that action is not something (we can assume) that I was morally obligated to do, hence I cannot have discharged any moral obligation by doing it. If (2 ) is true, there must be some other morally obligatory action that I failed to perform at midnight, and, hence, I shirked a moral obligation at that time. Since I shirked no moral obligations at midnight, it follows that (2 ) is false. (220, note omitted)

The example, though, is controversial, as it is questionable whether any agent can have a moral obligation not to do something while sound asleep. This is especially true if we assume a version of K according to which one ought not to (or ought to) do something only if one can intentionally not do (or do) that thing. To explain, we have seen that one can do something only if one does not lack the opportunity to do it. Further, someone lacks the opportunity to do, say, A, only if there is something that prevents or would prevent a successful exercise of the germane ability or abilities to do A. It appears that intentional omissions or “negative doings” can be accomplished only by way of some “positive doing.” For example, if I

55

intentionally fail to turn on the light, I intentionally do something else, such as, for instance, remain seated, to see to the occurrence of the light’s not being turned on. Suppose (for reductio) that I ought to refrain from doing A while asleep. Then given the relevant version of K, I ought intentionally to refrain from doing A while asleep. But I can intentionally refrain from doing A only if I can intentionally do something else, B, to see to the occurrence of my not-A-ing. Now I can intentionally do this other thing, B, only if I don’t lack the opportunity to do B. While asleep, however, I do lack the opportunity intentionally to do B. For there is something – my being asleep – that prevents a successful exercise of the relevant abilities. So I can’t intentionally do B while asleep. It follows that it’s false that I ought to refrain from doing A while asleep. There is a further problem with Yaffe’s example. If the obligation not to do, say, A, at midnight is remote (which it presumably is), then it’s perfectly reasonable to say that this does require a remote obligation to do something, B, instead. So, for instance, if at 7:00 p.m. you ought not to shower at midnight, then, presumably, at 7:00 p.m. you ought to do something else instead at midnight. Despite the flaw in his reasoning, suppose we grant Yaffe’s point that an obligation not to perform a particular action does not generally generate an obligation to do something else instead. Still, Widerker need not admit defeat. For notice that PAP2 is ambiguous between PAP3 and PAP4: PAP3: S is morally blameworthy for doing A only if S could have refrained from doing A (where refraining from doing A does not entail that S could have done something other than A, like B, as opposed to refraining from acting at all). PAP4: S is morally blameworthy for doing A only if S could have done some other act, B, instead of A. It seems that if Frankfurt-type examples impugn PAP4, they impugn PAP3 as well. As this is so, Widerker could exploit the following argument to secure his conclusion that if Frankfurt-type examples impugn PAP3, they impugn K as well. Argument 3 (1∗ ) S is morally blameworthy for doing A only if S has a moral obligation not to do A. (2∗ ) S has a moral obligation not to do A only if S has the power to refrain from doing A. (From the corollary of K that says that ‘ought not’ implies ‘can refrain from.’)

56

(3∗ ) Therefore, S is morally blameworthy for doing A only if S has the power to refrain from doing A.

Anticipating this sort of move, Yaffe takes issue with the rationale for (2∗ ). He writes: Understood strictly, the Maxim says nothing at all about the nature of the abilities necessary for an agent to be under an obligation not to act in a particular way; the Maxim seems to apply only to cases in which the object of the ‘ought’ is an action, and in cases in which we are obligated not to act in some particular way, the object of the ‘ought’ is rather absence of action, a state of affairs that can be realized without any action at all. (1999: 221)

Yaffe admits that a principle much like K: New Maxim: If an agent ought not to do A then it is possible that the agent fails to do A, does imply this weak version of PAP: Weak PAP: An agent is morally blameworthy for what she has done only if it was possible that she should not have done as she did, and that Weak PAP is, arguably, impugned by Frankfurt-type examples. Notice that Yaffe’s New Maxim appears just to be the ‘ought not’ implies ‘can refrain from’ part of K, and that Weak PAP seems to be identical to PAP3. So one might wonder why Yaffe is troubled by Argument 3. Yaffe’s concern is that the New Maxim might not be so “deeply entrenched in our commonsense moral thought” (1999: 222) as is the Maxim. If it is, though, as deeply entrenched, it appears that he would be willing to accept Argument 3. In response to this concern, I believe that if we accept the Maxim, that is, if we accept the view that ‘ought’ implies ‘can,’ we should also accept Yaffe’s New Maxim, that is, we should also accept the view that ‘ought not’ implies ‘can refrain from.’ For what is deeply compelling about the Maxim is that it expresses a control principle about moral obligation. It simply highlights the fact that one can’t perform an act that is obligatory unless one has deontic control over that act. But if having an obligation to do something requires having control over doing that thing, so it would seem that having an obligation to refrain from doing something requires control over refraining from doing that thing. Without convincing explanation to the contrary, it would seem arbitrary to suppose that the sort of control required for having an obligation to do something differs from the sort of control required for having an obligation to refrain from doing something.

57

Besides, it would be difficult to see how the control requirements for obligations to bring about actions and omissions would differ in the face of the fact that an intentional omission can only be accomplished by way of intentionally bringing about some “positive” action. For the same sort of control that is involved in performing an action that one has an obligation to perform would be involved in performing the “positive” action that is required intentionally to bring about an omission. I remain, then, unpersuaded by Yaffe’s reply to Widerker. This lack of persuasion is consistent with the view that the Achilles’ heel in the Widerkerian argument against K is principle B2.

58

5 Determinism and Deontic Anchors

I have been preparing the way to initiate one of the principal arguments of this book. It can be summarized in this fashion. No one can perform an action that is right, wrong, or obligatory unless one could have done otherwise. As we have seen, this is because there is a requirement of alternative possibilities for deontic anchors. In addition, there are powerful reasons to believe that determinism is incompatible with anyone’s ever having freedom to do otherwise. It follows, straightforwardly, that if determinism is true, then no act can instantiate a primary deontic property. This new incompatibility thesis – that determinism is incompatible with deontic morality – is the topic of this chapter. It is noteworthy that this defense of the new incompatibility thesis, just like one prominent defense of the thesis that determinism is incompatible with moral responsibility, turns on the view that determinism expunges alternative possibilities. The cardinal elements of the most widely discussed line of reasoning over the past several years for the incompatibility of determinism and freedom to do otherwise are easy to grasp. As Peter van Inwagen explains in An Essay on Free Will: If determinism is true, then our acts are the consequences of the laws of nature and events in the remote past. But it is not up to us what went on before we were born, and neither is it up to us what the laws of nature are. Therefore, the consequences of these things (including our present acts) are not up to us. (1983: 16)

In what follows in this chapter, I begin by extracting one well-known version of the argument contained in the passage. It utilizes a modal “transfer” principle that sanctions a certain kind of transmission of unavoidability from a pair of propositions to a third. I believe that, allowing

59

for certain qualifications, this sort of modal argument is indeed sound. As I have little further to contribute to it, I simply lay out the argument and briefly summarize selective responses it has elicited and various replies to these responses. I then offer some general observations on why determinism is incompatible with deontic anchors. Finally, I consider criticisms of the new incompatibility thesis and explain why they fail. THE CONSEQUENCE ARGUMENT FOR THE INCOMPATIBILITY OF DETERMINISM AND ALTERNATIVE POSSIBILITIES

Determinism, recall, is the doctrine that for any given time, a complete statement of the temporally nonrelational (or “hard”) facts about the world at that time, together with a complete statement of the laws of nature, entails one unique future.1 The basic so-called Consequence Argument for the incompatibility of determinism and freedom to do otherwise appeals to a rule of inference that can be dubbed Transfer. The rule is designed to capture the intuition that if a person is powerless (or lacks control) over one thing, and powerless over that thing’s leading to another, then the person is powerless over that second thing. Let the abbreviation “Ns,t(p)” stand for: proposition p obtains and is unavoidable for agent S at time t; that is, p obtains, and S cannot at t perform any act such that if S were to perform it, p would not obtain. Transfer can now be stated in this way: If: Ns,t(p) and Ns,t(if p, then q), then: Ns,t(q).

Simple examples help to confirm the plausibility of this principle. Suppose, for instance, that the river floods, and I am powerless to prevent this. Imagine, further, that if the river floods, it destroys the bridge and that I am powerless to alter this truth. It seems to follow that the river floods and I am powerless to prevent the destruction of the bridge. A second component of the Consequence Argument is what Fischer has called the “Principle of the Fixity of the Past”: For any action y, agent S, and time t, if it is true that if S were to do y at t, some (temporally nonrelational or “hard”) fact about the past relative to t would not have been a fact, then S cannot do y at t (Fischer 1994: 9, 78). Finally, a third component is the “Principle of the Fixity of the Laws”: For any action, y, agent S, and time t, if it is true that if S were to do

60

y at t, some natural law which actually obtains would not obtain, then S cannot do y at t (Fischer 1994: 9, 67). We now turn to the argument. Let M designate the true proposition that Sid goes to the movies at t. Let L be a conjunction of the laws of nature, and F be a conjunction of all the temporally nonrelational facts about the world at a time prior to Sid’s existence. If determinism is true, then the facts of the past, F , and the laws of nature, L, entail that Sid goes to the movies at t. It follows that if Sid had refrained from going to the movies at t, then either the past or the laws of nature would be different. This, in turn, seems to imply that at least one of the following counterfactuals is true: If Sid had refrained from going to the movies at t, the past would be different; if Sid had refrained from going to the movies at t, the laws would be different. But Sid cannot do anything that would make it the case that the past or the laws are different. As Sid has no control over the past or the laws of nature, and no control over the past and the laws entailing one unique future, it follows (given Transfer) that Sid has no control over any of his actions in this sense of ‘control’: He cannot at t so act as not to go to the movies at t. Put a bit more rigorously, the argument amounts to this: (1) L obtains and L’s obtaining is unavoidable for Sid at t (i.e., Nsid,t(L)).

(1) is based on the Principle of the Fixity of the Laws. (2) F obtains and F ’s obtaining is unavoidable for Sid at t (i.e., Nsid,t(F)).

(2) rests on the Principle of the Fixity of the Past. (3) The conjunction of L and F obtains and this conjunction is unavoidable for Sid at t (i.e., Nsid,t(L and F)).

As Sid has no control over the past’s obtaining, and he has no control over the laws of nature obtaining, it seems reasonable that Sid (in his circumstances) has no control over the conjunction of these things.2 (4) L and F entail M.

(4) follows from the doctrine of determinism. (5) The complex proposition: [if (L and F), then M ] obtains, and its obtaining is unavoidable for Sid at t (i.e., Nsid,t(if (L and F), then M)).

(5) is based on the rule that if it is logically necessary that proposition P implies proposition q (that is, if p entails q ), then the proposition: if P then q, obtains, and is unavoidable for any agent at any time.

61

(6) Therefore, M obtains and is unavoidable for Sid at t (i.e., Nsid,t(M)).3

(6) is derived from an application of Transfer to (3) and (5).

SOME OBJECTIONS AND REPLIES

I want to highlight two of the more widely discussed objections to this argument. The first begins by distinguishing between the following counterfactuals: C1: If Sid had done otherwise (if he had refrained from going to the movies, for example), the past (or laws) would have been (indeed, would have to have been) different. C2: If Sid had done otherwise, this would cause the past (or laws) to be different.

C1, it has been claimed, is innocuous since all it means is that if Sid had, contrary to fact, acted otherwise, the actual world would have been different regarding its laws or past.4 That is, in a deterministic world, one would have exercised one’s abilities to do otherwise only if the past or the laws would have been different. C2, in contrast, is implausible as no one can cause a law-breaking event or initiate a causal sequence in the present that would result in some change in the past. Elaborating, “local miracle compatibilists” claim that if one were to act differently from the way in which one is causally determined to act, the different action would have been preceded by some law-breaking event (a “local” miracle) that would have allowed for the occurrence of that action. The law-breaking event would have been a violation of some actual law of nature but not of any law of its world. But these compatibilists deny that one can perform an action (in the actual world) at a time that involves a violation of a natural law (of the actual world) at that time or after. The performance of an action that did violate a natural law in this way would constitute or bring about a miracle, and this would, in the judgment of these compatibilists, be something incredible. Local miracle compatibilists insist that they are committed only to the weaker view that one can sometimes perform an action at t the performance of which would require violation of a natural law (of the actual world) prior to t.5 Similarly, “backtrackers” about the past (or “Multiple-Pasts Compatibilists”) deny that one can perform an action the performance of which would have caused the past – perhaps the entire history of the world – to have been different. But they believe that it is perfectly reasonable that

62

one can sometimes perform an act the occurrence of which would require the entire past to have been different. Just as there is an ambiguity between C1 and C2, there is an analogous ambiguity between two corresponding ability claims: A1: Sid has the ability to do something such that if Sid did it, the past (or laws) would have been different (would have to have been different). A2: Sid has the ability to do something such that if Sid did it, this could cause the past (or laws) to be different.

Again, A1, it has been claimed, is an innocuous sense in which a deterministic agent might have the ability to make it the case that the past or the laws are different, whereas A2 is an incredible sense in which this might be so. A1 expresses harmless counterfactual ability, whereas A2 expresses incredible causal ability. It is then indicated that the relevant premises of the Consequence Argument are ambiguous between the two ways – the innocuous and the incredible – of understanding the pertinent counterfactual and ability claims. The argument is successful only if the pertinent premises are given the incredible reading. However, a compatibilist need not construe these premises in this fashion and can thus evade the argument. This objection, though informative and suggestive, is far from definitive. Among the many responses it has elicited, I find particularly compelling the following. When it comes to pondering what it is now in one’s power to do, one is pondering about what one has the ability to do in one’s present circumstances. It is little consolation to be told that, in one’s present situation, one does not have the ability to so act as to cause the past to be different or bring about a miracle by violating a law of nature. But it may be that one does have the ability to act in such a way that acting in that way would require that, under the circumstances, certain conditions obtain, including the conditions that the past be different or a prior local miracle occur. Given the circumstances in which we now find ourselves, the past is over and done with and the laws are set. If one cannot, in one’s current circumstances, do anything to bring about these conditions, it is dubious to suppose that one has the ability in question.6 Another cluster of objections against the Consequence Argument targets the Transfer Principle. Whether this principle is a valid inference rule depends on how its unavoidability operator ‘N ’ is interpreted. For instance, with ‘N ’ construed in the fashion in which it has been in the argument presented in the first section of this chapter, it has been claimed that transfer is susceptible to counterexample (see, for example,

63

Widerker (1987)). In addition (again, with ‘N ’ interpreted in the way in which it has been above), it has been argued that as transfer, in conjunction with another inference rule beyond suspicion, entails a principle that is false, transfer itself must be false (see Thomas McKay and David Johnson 1996).7 However, in support of the Consequence Argument, others have proposed reasonable replacements for transfer that evade the objections that plague the allegedly defective version. Alicia Finch and Ted Warfield’s substitute says: If p obtains and one cannot render it false that p obtains, and, necessarily, if p then q, then one cannot render it false that q (Finch and Warfield 1998: 521).8 The New Consequence Argument, invoking this replacement principle (“new transfer”), unfolds in this way: Necessarily, if the laws of nature and the past obtain, then so do all future truths. This is because if determinism is true, the future is just a logical consequence of the past and the laws of nature. Let RB stand for the proposition that you will read this book a year from now, and assume that RB is true. Then, necessarily, if the laws and the past obtain, so does RB. No one can render it false that the past and the laws obtain as the past and the laws are fixed. Hence, via new transfer, one cannot render it false that future truths, including RB, obtain. On an alternative strategy, the original form of transfer – if an agent is powerless over one thing’s obtaining, and there is an appropriate connection, over which the agent is also powerless, between this thing and a second, then the agent is powerless over the second thing’s obtaining – is retained but a different interpretation of N is offered. McKay and Johnson (1996), for example, suggest the following: Read Np as “p obtains and no human agent is able at any time to perform any action such that, were she to perform that action, it might not be the case that p.” With N so interpreted, transfer appears to be an inference rule both immune to counterexamples advanced against some versions of the principle in which N is differently interpreted and strong enough to sustain the Consequence Argument. To truncate a long story, it seems unpromising (though the argumentation is by no means settled here) to evade the Consequence Argument by rejecting transfer-like principles.9 To summarize, it has certainly not been my intention to pursue the many and varied objections that have been advanced against the Consequence Argument. However, it does appear that denials of the Consequence Argument (or variations of it) all come at the cost of jettisoning some highly reasonable and intuitively attractive principle. One might try,

64

for instance, to circumvent the argument by rejecting the Principle of the Fixity of the Past, or the Principle of the Fixity of the Laws, or transfer-like principles. It’s primarily for this reason that I believe that determinism is indeed incompatible with agents having alternative possibilities. At least this much is true: Any rejection of the view that determinism is incompatible with freedom to do otherwise is genuinely controversial. As its rejection is controversial, it would be instructive to learn what the view implies for the viability of deontic anchors. WHY DETERMINISM UNDERMINES DEONTIC ANCHORS

Deontic anchors, as we have argued, require dual control. The ‘ought’ implies ‘can’ principle K, which says that if one ought to do something, then one can do it, K’s complement, CK, the precept that if one ought to do something, then one can refrain from doing it, and OW, the tenet that one has a moral obligation to perform [not to perform] some action if and only if it is morally wrong for one not to perform [to perform] that action, entail that it is right, wrong, or obligatory for one to do something only if both one can do it and one can refrain from doing it. Alternatively, if just like ‘ought,’ ‘right’ and ‘wrong’ also imply ‘can,’ then again we can conclude that all three primary deontic properties require alternative possibilities. Our discussion of principle K in Chapter 2 should have left it clear that the relevant sense of ‘can’ invoked by this principle entails both ability and opportunity. Relatively weak versions of K entail a weak sense of ‘ability,’ where, roughly, to have such an ability to do something is just to have the capacity to do that thing. So, for instance, Trot, who is unaware of the safe’s combination, does have the weak ability to open the safe; or Waltz, who knows nothing about radios, but who is not physically disabled, or tethered to his seat, or otherwise in a position not to turn on the radio, does have the weak ability to switch it on. In contrast, relatively strong versions of K entail a stronger sense of ‘ability,’ where, loosely, to have such an ability to do something requires relevant skills and perhaps know-how. So, for instance, even if Waltz succeeded in turning on the radio, his turning it on would be something of a fluke. As he did turn it on, he did have the weak ability – the capacity – to turn it on, but we would not want to say that he had the pertinent skill or know-how; he lacked the strong ability to turn it on. In addition, we understood ‘opportunity’ broadly to include all circumstantial factors which would enable someone to exercise a relevant ability. On this view, someone lacks the opportunity, say, to perform a certain

65

action if there is something that prevents or would prevent a successful exercise of the relevant ability, regardless of whether this is something that is “external” to the agent (like a locked door) or something “internal” (like a state of unconsciousness). Turning for a moment to the dispute between compatibilists and libertarians over whether determinism threatens moral responsibility, I agree with Vihvelin’s insight that, once we realize that to have an ability is just to have a certain kind of capacity or skill, it should be uncontroversial that having abilities, including unexercised ones, is compatible with deterministic as well as indeterministic laws. For any ability we might think relevant to the question of freedom and moral responsibility – such as reasoning and deliberating about possible courses of action, making evaluative judgments, forming intentions, and so forth – there is no reason to think that deterministic causal laws (or the past) would deprive us of this ability. Libertarians should be understood as claiming that determinism deprives agents of the opportunity to do anything other than what they in fact do (Vihvelin 2000: 143). Similarly, regarding our concern over whether determinism threatens deontic morality, it would be implausible to suppose that deterministic causal laws deprive us of germane abilities to perform morally right, wrong, or obligatory actions. Rather, such laws (or the past) ensure that persons lack the opportunity to do other than what they in fact do. Now, no matter what versions of K and CK are true, these versions will specify that if one ought to do something, then one must have the opportunity to do that thing: There shouldn’t be anything that prevents one from exercising the relevant (weak or strong) abilities to do and to refrain from doing it; there shouldn’t, for example, be anything that prevents one from physically performing or refraining from performing that thing. In Frankfurt-type cases, although the agent can perform the relevant action, the counterfactual intervener ensures that he cannot refrain from performing it. In such a situation, though the agent has the relevant abilities (even volitional abilities) to refrain from performing the pertinent action, he lacks opportunity to refrain from performing it; the intervener sees to it that he cannot physically refrain from performing the germane action. Hence, in such a situation, the relevant action will lack a primary moral status – it won’t be obligatory, wrong, or right. The situation of agents in deterministic worlds is, in certain respects, just like the situation of agents in Frankfurt-type cases. In deterministic worlds, as in Frankfurt-type cases, though agents can perform pertinent actions (or intentionally fail to perform others), they lack the opportunity

66

to refrain from performing those actions (or to perform others). Hence, they lack dual control over the actions that they perform, and so their actions won’t have a primary moral status – they won’t (and can’t) be morally obligatory, wrong, or right. One might worry that unlike Frankfurt-type cases, determinism has actual prior effects on the agent’s capacities, abilities, character, and motives, whereas a merely counterfactual (but nonactual) intervener in Frankfurt-type cases does not have any actual effects on these things. So determinism, one might claim, does affect both one’s capacities or abilities and one’s opportunities. In response, I still retain deep reservations about how deterministic causal laws can detrimentally affect relevant abilities. For ponder the following: When advancing a positive account of free will, indeterminists can adopt many pertinent views of compatibilists, including for example, those about the causal connections among one’s values, best judgments, and intentions or decisions. Indeterminists of one brand will typically postulate indeterminacy at some point or points in the pathway of intentional action. They will, that is, theorize that some of the laws pertinent to the causal explanation of our choices or action will be nondeterministic or probabilistic ones. But surely nondeterministic causal laws do not detrimentally affect having abilities, including unexercised ones, required for free action. If they do not, it is a puzzle why deterministic causal laws should detrimentally affect these abilities. In any event, granting that determinism adversely affects both one’s abilities and opportunities to do otherwise simply strengthens the case for the view that determinism undermines deontic morality. We can conclude first that determinism and deontic morality are incompatible: If determinism is true, any act performed by any agent will fail to instantiate any of the primary deontic properties. Second, the truth of both K and CK (or the truth of both the wrongness analogue of K – ‘wrong’ implies ‘can’ – and OW) implies that if one has a moral obligation to do something, then one has to have two-way control over doing that thing. Alternatively, we can say that moral obligation requires “incompatibilist control,” or perhaps, more aptly, “dual control” in this sense: If an agent S ought to do something A in situation U , then in U , S has both the ability and opportunity to do A and the ability and opportunity to refrain from doing A. To amplify, suppose we grant that if one ought to do something, then it is true both that one can do that thing and one can refrain from doing that thing. Having granted this requirement of dual control, it is not sufficient merely to invoke an

67

hypothetical sense of ‘can.’ It would be incorrect, that is, merely to commit ourselves, for example, to the view that if S ought to do A in U, then S would do A in U if S wanted, or willed to, or chose to and S would do other than A in U if S wanted, or willed to, or chose to. For one thing, there are serious defects with hypothetical analyses of ‘can.’10 For another thing, as I have suggested, a person who does A in a Frankfurt-type situation is in a situation analogous to the one he would be in if determinism were true – in either situation he would lack the opportunity to do otherwise. In the former situation, the counterfactual intervener would ensure that he lacks the opportunity to will to do or to do otherwise, and in the latter, the laws and the past would ensure that he lacks the relevant opportunity. As refraining from doing something requires both the ability to refrain from doing it and the opportunity to refrain from doing it, in neither Frankfurt-type cases nor in deterministic worlds could one refrain from doing what one in fact does. S can lack the opportunity to refrain from doing other than A in U even if it is true that S would have refrained from doing other than A in U if S wanted or willed to. Further, without opportunity to refrain from doing A in U , S cannot refrain from doing A in U . Vihvelin indicates that Daniel Dennett (1984) suggests the following compatibilist account of ‘can do X ’ but shies away from claiming that this account amounts to the ‘can’ of ‘could have done otherwise.’ S can do X if (i) S has the ability to do X; (ii) S has the relevant mental abilities, for example, the ability to assess reasons for and against doing X and the ability to act according to S’s best judgment; (iii) none of the relevant abilities are impaired or malfunctioning (for instance, due to drugs, hypnosis, or extreme pain); and (iv) there is no impediment to S’s doing X (2000: 163, n. 13). But if S’s doing A is to have a primary deontic property, S must have both the ability and opportunity to do A and the ability and opportunity to refrain from doing A (or, as I’ll say, the ability and opportunity to do Not-A). In a Frankfurt-type case, the counterfactual intervener would ensure that S lacks opportunity to do Not-A; so condition (iv) of the compatibilist account of ‘can do X ’ would not be satisfied. Similarly, in a deterministic world, the laws and the past would ensure that S lacks opportunity to do Not-A. It is important to be clear that one need not suppose that factors that prevent a person from doing something need be of the same sort to be bona fide impediments to action. Being unconscious, just as assuredly as being shackled to his seat, would prevent the lifeguard from attempting to save the child. Although being unconscious is a different sort of

68

impediment than being shackled, both are impediments. Analogously, the counterfactual intervener in a Frankfurt-type case is a different sort of impediment to a person’s doing otherwise than the impediment to a person’s so doing that is imposed by the laws and the past. Nonetheless, surely this latter sort of impediment is a legitimate impediment. In all these cases, the impediments deprive the agents of relevant opportunity. So it seems, then, that with the laws and the past fixed, condition (iv) of the compatibilist account would go unsatisfied. A challenge can be issued to any person who wishes to be a compatibilist about determinism and deontic anchors: Develop a prima facie plausible account of ‘S can do X.’ Then show that in this very sense of ‘can,’ the agent can both do X and refrain from doing X in a Frankfurttype case, where X is the sort of action that the counterfactual intervener wants the agent to perform. I suspect that any such account of ‘can do X ’ will yield the result that in a Frankfurt-type case, the agent won’t be able to refrain from doing other than what she in fact does. This is because (i) such an account will, in all likelihood, include a proviso to the effect that there are no external or internal impediments that prevent the agent from doing X or from refraining from doing X; (ii) in Frankfurt-type cases, the counterfactual intervener ensures that the agent will lack opportunity to do otherwise even if she has the ability to do otherwise; so there will be an external impediment that prevents the agent from refraining from doing X; and (iii) refraining her doing X requires both the ability and the opportunity to refrain from doing X. In turn, if this is all correct, then we will have strong reason to believe that, in the proposed sense of ‘can,’ it won’t be true that the agent will be able to refrain from doing what she in fact does if determinism is true. For just as the counterfactual intervener is an external impediment to the agent’s doing other than what she does in a Frankfurt-type case, so the past and the laws are external impediments to the agent’s doing other than what she does if determinism is true. Compatibilists about responsibility and determinism have offered accounts of the sort of control required for moral responsibility that do not entail freedom to do otherwise. For example, Fischer and Mark Ravizza’s “reasons responsiveness” account provides a rationale of why agents in Frankfurt-type cases can have the right sort of control over their actions to be responsible for them (Fischer and Ravizza, 1998). Such accounts, illuminating as they might be to the type of control required for moral responsibility, cannot be adapted to function as appropriate accounts of the type of control required for deontic anchors. This is simply because deontic anchors require dual control whereas compatibilist reasons

69

responsiveness accounts are accounts of singular control. If an account of singular control for responsibility is correct, an agent in a Frankfurt-type case in which the intervener does not force the hand of the agent will (assuming all other conditions of responsibility are satisfied) be responsible for what she does (or fails to do) even though her action (or omission) will lack a primary deontic property. We can now draw a third conclusion. Several compatibilists, so-called semi-compatibilists, have argued that even if determinism undermines freedom to do otherwise, it does not thereby threaten moral responsibility, because freedom to do otherwise, as Frankfurt-type examples strongly suggest, is not the sort of control required for responsibility. Responsibility, they claim, requires only singular and not dual control. But dual control, entailing freedom to do otherwise, is required for deontic anchors. So whereas one can be a compatibilist about unfreedom to do otherwise and moral responsibility, it appears one cannot be a compatibilist about such unfreedom and deontic anchors. We are, thus, saddled with a new incompatibility thesis: Determinism is incompatible with deontic anchors.11 OBJECTIONS TO THE NEW INCOMPATIBILITY THESIS AND REPLIES

I now want to consider objections, newly advanced by Pereboom, to segments of my defense for the view that deontic morality and determinism are incompatible. Pereboom begins by reproducing my argument that OW and K entail WAP. He then notes that these principles, in conjunction with others, can be used to derive the result that no act in a determined world instantiates any of the primary deontic properties. Pereboom’s first concern is that the part of my argument that involves the derivation of WAP (WAP entails that there is a requirement of alternative possibilities for wrongness) has unintuitive results, and he recommends that the degree to which these results are unintuitive be weighed against how unintuitive it is to reject either OW or K. Pereboom finds especially unintuitive the result that a determined world is devoid of deontic morality (2001: 144–5). Here, however, Pereboom’s stance is indecisive. Pereboom defends a view he calls “hard incompatibilism,” which says that both determinism and its denial (indeterminism), with the exception of incompatibilist scenarios in which we are libertarian agent-causes (2001: xviii–xix), are incompatible with moral responsibility. This is, undoubtedly, a hard position. If one is willing to endorse this position and one finds nothing

70

unintuitive with, or is willing to accept the unintuitiveness of, its implication that both determined and nondetermined worlds (again, with the exception of worlds in which we are agent-causes) are devoid of responsibility, it is not clear why one should find so unintuitive the view that determinism is incompatible with deontic morality, especially in light of the fact that K and OW are, at least prima facie, very plausible. Second, Pereboom says that [T]he claim that actions are never obligatory, right, or wrong . . . is not the only unintuitive result of the argument that crucially depends on K and OW. The line that Haji is constrained to draw – because he endorses this argument – between judgments about goodness and those regarding rightness is also unintuitive. The following principle seems true: GR: Sometimes, actions that bring about the greatest good overall in worlds accessible to S are right for S. However, given that Haji wants to retain judgments about goodness, he must disavow GR, assuming causal determinism, and it is the argument based on K and OW that explains why he is committed to this result. (2001: 145, note omitted)

In response, an initial comment is simply that the judgments concerning goodness that I believe are possible in a determined world are judgments of the sort that an agent can bring about something that is overall (or intrinsically, or extrinsically) good or evil by doing something and not judgments or views analogous to GR. The former sorts of judgment can be true consistent with determinism’s being true. In addition, if one finds GR intuitively plausible, presumably one does so independently of having reflected on the view that there is a requirement of alternative possibilities for rightness, wrongness, and obligatoriness, and hence that determinism is incompatible with deontic morality. It is not clear, consequently, that having engaged in the relevant sort of reflection, one would still find GR intuitively attractive. Granted that GR does not flat out say that, if A is best for S, then A is right for S, some people might, nevertheless, claim that they find GR intuitive in virtue of accepting some such utilitarian principles as these: GR1: Given S’s alternatives at a time, if alternative A brings about the greatest (perhaps intrinsic) goodness, or more generally, the greatest deontic value, whatever such value consists in, then A is obligatory for S at that time, and

71

GR2: Given S’s alternatives at a time, if no alternative brings about more (perhaps intrinsic) goodness, or more generally, deontic value, than alternative A, then A is right for S then. Suppose S does A in a determined world, and by so doing brings about something good. One need not, if determinism is true, disavow principles like GR1 and GR2. One could, instead, opt for the view that if determinism is true, no act has genuine alternatives, and though some acts will produce goodness, they won’t, by the light of GR1 or GR2, be right or obligatory. Perhaps one will object that if A is S’s only option, then whatever its utility or deontic value, it does maximize such value, given that it has no alternatives. But then this would require disavowing GR1 and GR2 if determinism were true. This may well be so, but if actions with no alternatives are construed as maximizing utility (one wonders why they couldn’t equally be construed as minimizing utility), then GR1 and GR2 are not even remotely attractive. For we would then be burdened with the unacceptable implication of GR1 that a heinous act, such as a ruthless killing, would be obligatory (and thus right as well) provided that it had no alternatives. There are, in addition, strong reasons to renounce principles like GR1 and GR2 in favor of a “world utilitarian” principle, the gist of which is the following: W: An agent S ought to do an act A if and only if S can do A, S can refrain from doing A, and any accessible world in which S does not do A is denotically inferior to some accessible world in which S does A. (See, for instance, Zimmerman 1996: 26.) Unlike GR1, W does not, for example, imply that it can be the case that it is obligatory to perform a complex act (such as operating on a patient) but it is forbidden to perform some of its parts (like making an incision), and this is a strong card in its favor.12 In addition, axiological judgments of the form that some agent brings about something that is good by performing an action are perfectly consistent with W, even assuming the truth of determinism. A third worry of Pereboom’s concerns my view that, in a deterministic world, it can be true of agents that they are blameworthy without their having done anything wrong. Pereboom claims that, in “the context of a deterministic world” (2001: 145), this view violates another intuitive principle:

72

BW: Sometimes, when S is blameworthy for performing A, it was morally wrong for S to perform A. Pereboom writes: It is again the argument that crucially depends on K and OW that explains why Haji is committed to denying BW. Now, it may very well be that in the final analysis any view on these issues will have unintuitive consequences. However, it is not clear that a position that denies GR, BW, and that actions are sometimes morally obligatory, right or wrong while maintaining K and OW is superior to a view that, say, rejects OW and the claim that actions can be morally obligatory but accepts GR, BW, K, and that actions are sometimes right or wrong. (2001: 145)

Is BW intuitive though? First, let’s note that BW is ambiguous between BW1: Sometimes, when S is blameworthy for performing A in a deterministic world, it was morally wrong for S to perform A in that world, and BW2: Sometimes, when S is blameworthy for performing A in an indeterministic world, it was morally wrong for S to perform A in that world. Pretty clearly, Pereboom is concerned with BW1, which he submits is “very intuitive.” This submission, however, is deeply puzzling from the perspective of Pereboom’s hard incompatibilism. For surely such an incompatibilist should reject this principle (even though he might find it intuitive) because hard incompatibilism entails that both determinism and its denial are incompatible with blameworthiness. Unlike the hard incompatibilist, I reject BW1 because I believe that determinism is incompatible with deontic morality. If I am right about this, principle BW1 ought to be eschewed for more than one reason. Independently of begging the question either in favor of K and OW or against K and OW, why might one find BW1 intuitive? I think the most plausible reason hinges on presupposition of the truth of yet another principle we have encountered before: B2: S is blameworthy for performing [not performing] A only if it was morally wrong for S to perform [not to perform] A. For if blameworthiness requires wrongness, then BW is, of course, true. One might then draw the further conclusion that, as BW is true, BW1 should be true as well. This line of reasoning is not cogent, however. For one thing, as I explained in the last chapter, there are good initial

73

reasons to eschew B2, and a number of them are independent of the considerations that motivate the view that determinism and deontic anchors are incompatible. (Some of these reasons are further developed in Chapter 10.) Hence, Pereboom’s pivotal claim that it is “the argument that crucially depends on K and OW that explains why Haji is committed to denying BW” is disputable. For another thing, one can accept BW (perhaps because one accepts BW2 despite rejecting B2) without being committed to accepting BW1. This is, indeed, my position. (I should add that I do believe that there is a plausible rendition of the view that blameworthiness can survive determinism despite the latter’s eradicating wrongness (see Chapter 10 and Haji 1998a). But even if it were true that blameworthiness and determinism were incompatible, this would not be so because, roughly, blameworthiness requires wrongness (as B2 implies) and determinism undermines wrongness.) Also questionable is Pereboom’s proposal that a position that denies GR, BW, and that actions instantiate primary deontic properties (assuming determinism is true), while maintaining K and OW, is not obviously superior to a position that rejects OW and the claim that actions can be morally obligatory, but accepts GR, BW, K, and that actions are sometimes right or wrong (again, assuming determinism). It is so because the relevant version of BW, that is, BW1, is suspect, there is little reason to reject OW, and it is (as I explained) controversial whether GR is intuitively plausible. Fourth, Pereboom urges: [I]n this discussion it is important to keep in mind that we humans commonly presume that our actions are typically not causally determined. If this presumption were true, then K, OW, GR, and BW could all be maintained without any unintuitive consequences. But if we were to discover that causal determinism is true instead, then our easy acquiescence to these principles might need to be reconsidered. However, supposing that they cannot all be maintained, it seems arbitrary to privilege absolutely K and OW over GR and BW – and over the claims that actions are sometimes right and wrong and that judgments of moral obligation are sometimes true. (2001: 146)

Pereboom’s point about privileging “absolutely” K and OW over GR, the relevant version of BW (BW1), and the view that actions sometimes instantiate the primary deontic moral properties if determinism were true would be well motivated if, for example, K, OW, GR, and BW1 were equally defensible. But they are not. Since there are good grounds for rejecting BW1 (indeed, as I remarked previously, a hard determinist should

74

jettison BW1), it is questionable whether GR is as intuitively attractive as some people take it to be, and there are strong reasons for accepting K and OW, it is somewhat misleading to charge that I privilege, at the outset, K and OW over GR and BW1. Fifth, and finally, Pereboom suggests that, as he puts it, K may well be false for an action-guiding use of moral ought judgments. He explains that one role of such judgments is to guide actions. For example, we tell people that they ought not to steal to keep them from stealing. He then claims that K does not “harmonize” well with “an action-guiding point of view” of these judgments (2001: 147). He reasons as follows. When we engage in practical deliberation, we frequently believe that more than one option is possible for us: Suppose that causal determinism is true, and that hence no agent could ever have done otherwise. Nevertheless, we do not typically know in advance – before deliberation is complete or a decision has been made – which choice for action has been causally determined. Rather, . . . it is almost always true that from the epistemic point of view of the agent at the time of deliberation more than one option for which choice she will make is possible. That is, more than one such option is possible relative to (an appropriate subset of ) what she believes (or of what she should believe), in the sense that it is at least not ruled out by what she believes (or by what she should believe). Often, when one attempts to guide an agent by means of a moral “ought” judgment, it is the range of options for actions that are in this sense epistemically possible for the agent at the time of deliberation that one addresses. (2001: 147–8)

Pereboom then proposes that in such contexts of practical deliberation, “ought” judgments might well not only be practically rational to express but also generally true. If this is right, then K would be false for these uses of “ought.” One might hypothesize that K stands as a principle governing certain roles for moral “ought” judgments, but for this action-guiding function of moral judgment, K is best disavowed. (2001: 148)

If K is disavowed, then as it is a vital component of the argument for the incompatibility of determinism and deontic morality, that argument can be disavowed as well. Let’s limit discussion to two of Pereboom’s commitments: PB1: K, and derivatively WAP, do not “harmonize” with an actionguiding point of view, or the action-guiding role of moral ought judgments, and

75

PB2: K stands as a principle governing certain roles for moral ought judgments, but when the action-guiding function of such judgments is at issue, K should be disavowed, perhaps in favor of a principle (KAG) of this sort: S has a moral obligation to perform A only if S believes that a number of actions are alternatives for her and A is one of these actions; or simpler, S has a moral obligation to perform A only if S believes that S can do A. Both PB1 and PB2 are problematic. Regarding the former, an analogy might help. Judgments of blame- or praiseworthiness can be action guiding. But surely both compatibilists and incompatibilists about determinism and responsibility can agree that there is some principle (RC), whatever it precisely turns out to be, that expresses the sort of control one must have over an action in order for one to be blame- or praiseworthy for that action. Moreover, RC would presumably be consistent with the view that judgments of, say, blameworthiness have several different functions, including influencing action or gauging the moral worth of an agent with respect to a particular episode in her life. Reverting to K and judgments of deontic morality, as I see it, K is a principle of deontic control; it expresses the sort of control (or a component of control) that one must have over an action in order for that action to instantiate one or more of the primary deontic properties. Further, K is perfectly consistent with the view that “ought” judgments have a prescriptive or action-guiding function. I find PB2 more puzzling than PB1. First, reflect, for a bit, on KAG, the principle that S has a moral obligation to perform A only if A is an action that S believes S can do. Now either ‘moral obligation’ in KAG expresses deontic obligation – the normal garden variety concept of overall obligation – or it does not. If it does, then KAG is false. For there are many cases in which A is an alternative for S, S fails to consider A, perhaps because she believes falsely that A is not an alternative, or she selectively focuses on other alternatives at the expense of A, and A is obligatory for S. If ‘moral obligation’ in KAG does not express deontic obligation, then KAG is irrelevant to the issue of whether determinism is incompatible with deontic morality. Second, Pereboom’s line of reasoning strongly suggests that Pereboom accepts something of this sort: PB3: K is a principle that governs certain roles (functions) for moral ought judgments whereas KAG is a principle that governs other functions, specifically, the action-guiding one, of moral ought judgments.

76

But surely PB3 is controversial. Again, K is a “control principle” that specifies one condition requiring satisfaction for it to be the case that one has an obligation to do something. This is consistent with a moral “ought” judgment’s having several different functions. Even if we grant PB3, this does not bode any better for Pereboom. For the argument for the incompatibility of determinism and deontic morality is not in any way concerned with the action-guiding function of ought judgments. Further, that argument relies on K and not KAG. As far as I can see, Pereboom has not established that if PB3 is true, K is not the (or a) freedom-relevant or control component of deontic morality. Finally, consider the following variation of Strong K: S ought to do A only if S has the opportunity to do A, is physically and psychologically able to do A, has the relevant “know-how” to do A, and is not ignorant of germane facts (this last clause is meant, among other things, to capture the idea that S ought to do A only if S believes that A is an option). This version of K implies KAG, and together with OW, it entails WAP. In conclusion, it appears that the new incompatibility thesis that determinism is incompatible with deontic morality survives Pereboom’s objections. Traditionally, incompatibilists about freedom and responsibility have argued that freedom to do otherwise is required for moral responsibility, and that determinism undermines responsibility precisely because it is incompatible with such freedom. They have claimed that responsibility can secure a foothold only in an indeterministic universe in which people can make causally undetermined choices that, in turn, ensure that they have freedom to do otherwise. It’s worth exploring whether a world in which indeterministic choices occur can accommodate deontic anchors. This is what we will delve into in the second part of the book.

APPENDIX: SAKA ON ‘OUGHT ’ IMPLIES ‘CAN ’ AND DETERMINISM

Paul Saka (2000) has recently proposed that the ‘ought’ implies ‘can’ principle K is imperiled by determinism. Saka advances two modal arguments, one against our knowing that ‘ought’ implies ‘can’ and the other against its being the case that K is true, that, he claims, “share the same underlying structure” (93): According to the first, the Epistemic Argument, the ought-can principle is contradicted by the doctrine of determinism; but determinism may, for all we now

77

know, be true; and therefore we don’t know that ought implies can. According to the second, the Analyticity Argument, the ought-can principle is contradicted by the doctrine of determinism; said doctrine makes a synthetic claim; and therefore it isn’t analytic that ought implies can. (93)

The first argument would, at best, establish that we don’t know that ‘ought’ implies ‘can.’ Our failing to know that ‘ought’ implies ‘can,’ however, is consistent with its being the case that K is true. (I shall briefly revert to this concern below.) It is the second argument, it appears, that if sound, would undermine K. I will, though, in this appendix address both arguments. I aim to show that either is suspect. The crux of my criticism is that K is not “contradicted” by determinism. Rather, determinism is incompatible with its being the case that our actions instantiate one of the primary deontic properties of (overall) obligatoriness, rightness, and wrongness. In what follows, I summarize the two modal arguments, and I then explain why they fail. Prior to presenting the arguments, Saka distinguishes various interpretations of K. I limit discussion to the two arguments concerning K in which K is taken to be analytically true: In virtue of the concepts involved, if you ought to do A, then you are capable of doing A (93). Saka further explains that the Epistemic Argument comes in two versions, “one referring to full knowledge and the other to mere justification (all things considered)” (94), and that though the two versions are technically distinct, they are identical in all other relevant respects. In the interests of brevity, I address the full-knowledge version of the Epistemic Argument. This version of the Epistemic Argument proceeds by reductio ad absurdum from the initial assumption that we know that ‘ought’ implies ‘can’ (94). Introducing modal operators, read []P as “we know that P is true,” and read <>P as “for all we know, P is true.” The following remarks of Saka’s on determinism shed some light on this somewhat obscure “for all we know” operator. Saka claims that “the thesis of determinism, regardless of whether it’s true or false, appears to remain an unsettled issue. At any rate, there is no evident logical or factual contradiction, so far as we now know, in our asserting various versions of determinism; they are, given current knowledge, epistemically possible” (94). These remarks suggest that <>P amounts to something like: Given current knowledge, or relative to what we currently know about the world, P may well be true. The first argument unfolds in this fashion: (1) [](X ought to A →X can A). (We know that if X ought to A, then X can A.) (1) is the initial assumption. Saka seeks to convince us that this assumption, together with other premises, generates a contradiction.

78

(2) <>(∼X does not A → ∼X can A). (For all we know, if X does not do A, then X cannot do A.) The rationale for this premise is not transparent. I propose that (2) is based on certain alleged facts about determinism. Some of Saka’s remarks (see, for example, 96–8) suggest that Saka presupposes that determinism is incompatible with freedom to do otherwise. Given this presupposition and the further assumption that determinism is true, it follows that no one can do other than what one in fact does or fails to do. As relative to what we now know about the world, determinism may well be true; it seems to follow that for all we know, no one can do other than what one does or fails to do. (3) <>(X can A →X does A). (For all we know, if X can do A, then X does do A.) (From (2) and modal contraposition. The latter is simply the rule that <>(P →Q) is equivalent to <>(∼Q → ∼P ).) (4) <>[(X ought to A →X can A) and (X can A →X does A)]. (From (1), (3), and agglomeration. Agglomeration is the principle that ([]P & Q)→ <> (P & Q).) (5) <>(X ought to A →X does A). (For all we know, if X ought to A, then X does A.) (From (4) and the modal transitivity of → . Such transitivity, with <> as the operator, amounts to this: <>[(P →Q) & (Q →R)]→ <> (P →R ).)

Introducing a particular case, where for X we substitute “the president” and for A we substitute “always tell the truth,” Saka proposes: (6) [](X ought to A &∼X does A). (We know that the president ought always to tell the truth and yet does not.) Saka’s rationale for (6) is that there is a prima facie obligation not to lie that the president presumably violates at least on occasion. (95) (7) [] ∼ (∼ X ought to A v X does A). (By (6) and modal DeMorgan’s law.) (8) [] ∼(X ought to A →X does A). (By (7) and the modal definition of → .) (9) ∼<>(X ought to A →X does A). (It’s false that for all we know, whenever X ought to A, X in fact does A.) (9), Saka says, is explained by the fact that []P (we know that P is true) is equivalent to ∼<>∼ P (it’s not the case that, for all we know, P is false.)13

Saka is satisfied that the “logic of (3–5) and (7–9) is formally impeccable and that we are left with the substantive (1), (2), and (6) as candidates for rejection” (96). He opts to jettison (1). Cognizant that rejection of (1) does not amount to rejection of K, Saka claims that it is “pragmatically essential” for rational proponents of K to hold: (22): We know that if X ought to A then X can A, or We have justification for believing that if X ought to A then X can A. (102)

This is because if

79

ought-can proponents did not hold (22), it would be possible for them to say (24) If X ought to A then X can A, but we don’t justifiably believe it. While this position is possible for the unreasonable dogmatist, it deserves no philosophical attention. Thus any successful proof against (22) leaves legitimate participants of the dialectic unable to assert [K]. (102)

The structurally parallel Analyticity Argument takes the modal operators [] and <> as expressing analytic necessity and analytic possibility (“it is a conceptual truth that” and “it is not a conceptual falsehood that”) respectively. The argument proceeds in this way: (1 ) [](X ought to A →X can A). (It is analytically necessary that if X ought to A, then X can A.) (2 ) <>(∼X does not A → ∼X can A). (It is analytically possible that if X does not do A, then X cannot do A; i.e., the proposition, if X does not do A, then X cannot do A, is not false by definition.) According to Saka, (2 ) represents the claim of determinism. (3 ) <>(X can A →X does A). (From (2 ) and modal contraposition.) (4 ) <>[(X ought to A →X can A) and (X can A →X does A)]. (From (1 ), (3 ), and agglomeration.) (5 ) <>(X ought to A →X does A). (From (4 ) and the modal transitivity of → .)

Saka claims that we need to find a counterpart to premise (6) of the Epistemic Argument that holds analytically, and he proposes that for X, we substitute “murderers,” and for A we substitute “refrain from murder.” He comments that “It is an analytic truth that murderers ought to refrain from murder, for murder is by definition a kind of wrongful killing. It is also an analytic truth that murderers do not refrain from murder, for a murderer by definition is one who murders.” (98) (6 ) [](X ought to A & ∼X does A). (It is analytically true that murderers ought to refrain from murder and yet do not.)

Should we persist in seeing an existential presupposition in (6 ) that would invalidate its claim to analyticity, Saka suggests that we replace his schematic X with a quantificational structure (for all x, if x is a murderer, then x ought to refrain from murder . . . ). (7 ) []∼ (∼X ought to A v X does A). (By (6 ) and modal DeMorgan’s law.) (8 ) []∼ (X ought to A →X does A). (By (7 ) and the modal definition of → .) (9 ) ∼ <>(X ought to A →X does A). (From (8 ) and the equivalence of []P and ∼ <>∼P.)

80

Once again Saka’s verdict is that (1 ) ought to be rejected. By way of evaluation, one might, for instance, take issue with Saka’s rationale for premise (6) of the Epistemic Argument by calling attention to the fact that violation of a prima facie obligation not to lie need not amount to violation of an overall obligation not to lie. However, I wish to develop another line of resistance to either of the two arguments. I have suggested that if it is assumed that ‘ought’ implies ‘can,’ then it is very natural to assume that just like ‘ought,’ ‘right’ and ‘wrong’ also imply ‘can.’ For if its being obligatory to do something implies freedom to do that thing, then it is equally plausible to suppose that its being wrong or permissible to do something also implies the same sort of freedom to do that thing. Given this assumption, I have argued that rightness, wrongness, and obligatoriness all require alternative possibilities. Let’s recapitulate one part of that argument. Let KW be the “wrongness analogue” to K: KW: If it is wrong for one to do A, then one can do A. From: OW: It is obligatory for one to do A if and only if it is wrong for one to refrain from doing A, we derive: OW1: If it is obligatory for one to refrain from doing A, then it is wrong for one to do A. KW and OW1 entail that if it is obligatory for one to refrain from doing something, then one can do that thing. In other words, there is a requirement of alternative possibilities for obligatory acts. Analogously, there is such a requirement for wrong and right acts as well. Hence, if it is agreed that determinism is incompatible with alternative possibilities, then it cannot be maintained that determinism is nonetheless compatible with the case that one’s acts have one or more of the deontic properties of obligatoriness, rightness, and wrongness. We are now in the position to offer an alternative rejection of line (6) of the Epistemic Argument. If we assume line (1) of that argument, that is, if we begin with the assumption that K is true (even if only to derive a contradiction), as Saka does, then for the reasons explained, we can also assume that KW is true. KW and OW1, as we have seen, entail that there is a requirement of alternative possibilities for obligatoriness. Line (2) of the Epistemic Argument is reasonable, given determinism. But on the assumption that, relative to what we currently know about

81

the world, determinism may well be true, and that if determinism is true, no act can be right, or wrong, or obligatory, it follows that: (1∗ ): For all we know, it is false that the president ought always to tell the truth.

If (1∗ ), in turn, is true, then (2∗ ) is also true: (2∗ ) It is not the case that we know that the president ought always to tell the truth.

For if we know that P is true, then it is not the case that, for all we know, it is false that P is true. If (2∗ ) is true, though, then (6) is false. As for the Analyticity Argument, Saka, as noted, takes the [] operator to express analytic necessity, and []P is to be read as “It is a conceptual truth that P.” Conceptual truths are true in all possible worlds. If one were, for some reason to deny this, one could simply interpret K as the principle that in every possible world, if X ought to do A, then X can do A. Analogously, I shall take the <> operator to express truth in at least one possible world. (6 ), then, says that: (6 ): In every possible world, murderers ought to refrain from murder and yet do not. It will be helpful to translate (6 ) in this way: (6 ): In every possible world, murderers kill others but ought not to do so.

No act in a determined world can be right, or wrong, or obligatory. Again, this simply follows from the facts that there is a requirement of alternative possibilities for rightness, wrongness, and obligatoriness and that determinism rules out such alternatives. In a determined world, murderers do kill others, but it is false that they ought not to do so ( just as it is false that they ought to do so, or that it is wrong for them to do so, or that it is permissible for them to do so). I conclude that (6 ), too, is false since there are at least some worlds, determined worlds, in which it is false that murderers ought not to kill others. In summary, if we assume, as Saka does, that K is true and we assume, as Saka apparently does, that determinism is incompatible with freedom to do otherwise, then we can show, contrary to what Saka believes, that these assumptions do not generate the contradictions that Saka claims they do. It is much later in his paper that Saka even begins to acknowledge the possibility that “ought” is incompatible with determinism, but even then, it seems, he fails to get the point right. Saka says that a common response he finds to his two modal arguments is: “Of course the thesis that

82

ought implies can is incompatible with determinism. Determinists don’t recognize ought statements. Your proof is trivial” (101). Saka responds that this rejoinder is mistaken in two ways: In the first place, determinism is compatible with moral realism. Because hard determinists deny the existence of free will, it is true, they sometimes . . . deny that agents are accountable for their errors. But even when determinism is supplemented in this way, the position still falls short of denying that transgressors ought not do what they do. For to deny such ought statements is to withdraw an important force – social pressure – as one determining factor in others’ behavior. Assuming that hard determinists have social wants, it would go against their interests to deny ought statements. For this reason, I don’t think that hard determinists should deny ought statements; and in point of fact, I don’t think that they generally do. (101)

There are multiple errors here. One is that determinists should deny “ought” statements if it is granted that determinism effaces alternative possibilities. A second error is to assume that denial of “ought” statements is tantamount to withdrawing “an important social force – social pressure – as one determining factor in others’ behavior.” If Saka were right, then an ethics of virtue, seemingly not committed to deontic modes of appraisal, would lack the means to harness social pressure to influence others’ behavior. This is patently false. The general point is that other modes of moral appraisal, such as aretaic ones, or, for instance, appraisals concerning what it is intrinsically, instrumentally, or overall good or bad to do, can effectively be utilized to exert social pressure. Saka continues: Far more importantly, the triviality criticism misses the point of my proofs. I do not assume the truth of determinism, only its cogency. The mere possibility of determinism suffices to overthrow the necessity of “ought implies can.” Thus, even libertarians can accept my proofs. . . . (102)

Again, this is puzzling. We need not pass judgment on whether determinism is in fact true. Rather, if we assume that determinism along with principles like K, KW, and OW1 are true, then no act can be morally right, wrong, or obligatory. In the final section of the paper, Saka claims that some will insist that “the ought-can principle is validated by some connection between obligation and practical reason” (100). The issue is how this connection is to be understood. In one strategy, this connection is to be explained by giving “reductive analyses” of ‘ought’ statements. For example, Saka reports

83

that according to some theorists, the connection between obligation and practical reason reflects “a meaning equivalence between ought statements and imperatives” (100).14 Against this strategy, though, Saka argues that “available reductive analyses generally falsify ‘ought implies can’ ” (100). But there is a possibility that Saka has overlooked. Rather than attempting to validate K by constructing reductive analyses of it, one might resist such analyses altogether and opt instead for the alternative that, if K is validated, it will be so by some of our best theories of the concept of moral obligation. As we have noted, the recent promising theories of Feldman and Zimmerman, for example, have K as a theorem.

84

Part Two Indeterminism and Deontic Morality

6 Transition: From Determinism to Indeterminism

INTRODUCTION

In a deterministic world in which the future is not a garden of forking paths, no acts are right, wrong, or obligatory. I have argued that determinism undermines deontic anchors precisely because it rules out alternative possibilities that such anchors require. If this is so, it seems reasonable to suppose that if some of our actions are not causally determined and we have genuine alternative options, then there will be sufficient leeway to secure deontic morality. A hallmark of libertarianism is that free actions (if any) that we perform are not causally determined. Libertarian views have been and are generally conceived as views about the sort of freedom or control required for responsibility. However, to explore whether libertarian-like views can accommodate deontic morality, we can exploit the component of such views that allows for agents being able to do otherwise (on various occasions of choice). This is the topic of the second part of this book. Introducing some terminology will facilitate the discussion. An indeterministic theory specifies the species of control or freedom required for moral responsibility or deontic morality, and it entails that the sort of freedom required for either is incompatible with determinism. Traditional libertarian theories, for instance, all qualify as indeterministic theories. An R-libertarian view is the view that the kind of freedom necessary for moral responsibility is incompatible with determinism and that human beings can sometimes be or are morally responsible for what they do. Analogously, an M-libertarian view is the view that the type of freedom required for deontic morality is incompatible with determinism and that human beings can or do perform actions that have one or more of the primary

87

deontic properties. The guiding question of this part of the book can now be formulated succinctly in this way: Is there a defensible M-libertarian view? To tackle this question, a promising place to start is to look closely at leading R-libertarian theories. One contender, propounded by Carl Ginet (1990), postulates that the sort of control an agent must have over an action to be responsible for it is noncausal in nature. Ginet claims that there is a simple mental action – one lacking internal causal structure – that is a constituent of every action (15). Some simple actions, complete in themselves, are not part of any more complex action, as when one forms a mental image. They are to be distinguished from other mental events like unbidden dream images in virtue of an intrinsic phenomenological feature that Ginet dubs an “actish phenomenal quality” (13). Typically, such actions are (uncaused) volitions that, in turn, causally generate a person’s voluntarily exerting her body (23). Ginet insists that whereas all our actions, having at their core a simple mental action, are ultimately uncaused, they are not inexplicable. For it is possible to give a reasons explanation of pertinent actions by citing, for example, relevant desires or intentions that are elements of the volitional component of these actions. I will not pursue this theory further, for it has a prima facie difficulty not shared by other R-libertarian competitors: As Mele (1992), O’Connor (2000), and Clarke (in press) have indicated, if free actions have uncaused volitions at their core, and hence are not determined to occur by anyone or anything, they seem not to be under anyone’s control. If they are not under anyone’s control, and as control is a requirement of both moral responsibility and deontic morality, Ginet’s view seems to imply that no one can be morally responsible for one’s actions and no one can perform actions having one or more of the primary deontic properties. The fundamental difficulty with Ginet’s indeterminism, then, is that it is incompatible with the view that the sort of control required for responsibility or deontic anchors is, at bottom, causal in nature.1 A second sort of R-libertarian theory, the agent-causal theory, specifies that the kind of freedom required for moral responsibility is a function of agents possessing causal powers to make choices or perform actions without being causally determined to do so. On this view, the variety of causation implicated in an agent’s making a free choice is not reducible to causation among ordinary events, including events involving states of the agent or the agent. Rather, the sort of causation is an instance of a substance – the agent – directly causing a choice but not by way of any states or events.

88

Again, I set aside this sort of theory for a couple of reasons. First, its metaphysical commitments are more excessive than those of rival sorts that purport to get by with ordinary event-event causation. Second, and more importantly, if any of these leaner rivals can accommodate deontic anchors by, among other things, providing an account of free agency that allows for there being alternative courses of action genuinely accessible to the agent, suitably developed agent-causal theories should prove adequate in this respect as well. I confine attention, lastly, to what can be referred to as “modest R-libertarianism.” Two prominent features of this sort of incompatibilism are (a) that it makes no appeal to metaphysically exotic agents or forms of causation such as Kantian noumenal selves, Cartesian minds, or substance causation, and (b) that it settles for nondeterministic (or probabilistic) event causation at a point or points in the causal pathway of an agent’s choices or actions. A large attraction of modest R-libertarianism, its proponents claim, is that it limits our control over our choices only in a way that gives us no less control than we would have were determinism true, while making room for genuine alternative deliberative outcomes. (See, e.g., Clarke in press, 2000, 1995; Kane 1996a; Mele 1995: ch. 12.) If such a view of control is defensible, then it would appear that the direction in which our futures unfold would, in some deep sense, be truly up to us. To many, this is a captivating vision of free agency. Regarding our primary concern, this view of free agency may hold the key to accommodating deontic morality. SYNOPSIS

Part Two of this book evolves in the following fashion. After summarizing some general background on modest R-libertarian theories, I introduce Mele’s version of such a theory. In Mele’s theory indeterminacy in the actional pathway leading to an agent’s decision or other sort of action occurs relatively early at the juncture between the agent’s deliberations about what to do and the formation of a best judgment regarding what to do (Mele 1995: ch. 12). In standard nondeviant cases of intentional action, the agent decides on the basis of such judgment, say, to perform some action and then acts accordingly, intentionally performing the action. It is the best judgments, on this view, that are undetermined. I argue that this sort of modest libertarianism, even if it accommodates some deontic morality, accommodates far too little.

89

Next, I turn to a more robust form of modest R-libertarianism in which indeterminacy is located relatively late in the pathway, culminating in action at the juncture between, roughly, the consideration of reasons and the formation of intentions or the making of decisions. Here, it is the intentions or decisions that are undetermined. My verdict will be that this more robust form of libertarianism is hospitable to deontic morality. However, prior to being in a position to draw this verdict, a serious hurdle I previously thought could not be overcome (see Haji 1999a) will have to be cleared. It has been charged that the more robust version of R-libertarianism that I have distinguished succumbs to a powerful objection: In brief, if our choices are indeterministically caused, and it is the very intentions or decisions that are undetermined, then our choices are a matter of luck. For holding constant the conditions of the past that include the agent’s powers, values, deliberations, and character, the agent could have made one choice just as easily as she could have made another. Moreover, with nothing about the agent’s capacities, states of mind, reasons for action, and so on rationalizing this difference in outcome – as would be the case if the past were held fixed – the difference in outcome does seem to be a matter of luck. Luck of this sort appears to be incompatible with responsibility. I will argue that this luck objection does in fact have bite. This concession may seem suicidal to the project of accommodating deontic morality by appealing to elements of a robust modest R-libertarian theory. For luck, it appears, is not only incompatible with responsibility, it is incompatible with deontic morality as well. How could one, for example, do moral wrong if all of one’s actions were luck infected and so, it would seem, out of one’s control? Once we understand, however, why problems of luck undermine robust libertarian theories concerning the sort of control required for responsibility, we will see that an analogous objection cannot be launched against such theories when their resources are marshaled to account for the sort of control required for deontic morality. Specifically, in connection with robust R-libertarian theories and luck, I defend the view that there are factors that do not undermine an agent’s proximal control over her choices or actions but that do subvert responsibility because of their association with luck. Further, it is the very sort of proximal control postulated by robust R-libertarian theories and left unscathed by the luck objection that is required to accommodate deontic anchors. Elaborating, I contend that ascriptions of moral responsibility are agent focused: They disclose what an agent morally stands for in connection

90

with certain episodes – like the making of decisions – in her life. With indeterministic choice of the kind implied by robust modest R-libertarianism, there is no fact of the matter regarding what the agent morally stands for when she opts between competing choices or courses of action. For again, holding fixed the past that entails holding fixed the agent’s moral values, the agent could have made one choice, say, the prudential choice, just as assuredly as she could have made the competing choice, say, the moral one, and there would have been nothing in her past that would explain the difference in outcome. In contrast, ascriptions of deontic morality are act focused; they tell us more about the normative nature of the act itself. If such is indeed the case, then as long as an agent exerts appropriate proximal control in performing an action and was free to do something other than the act performed, the act could instantiate one or more of the primary deontic properties. R-LIBERTARIANISM

R-libertarianism is the view that some human beings choose and act freely and are morally responsible for at least some of what they do, and that free choice, free action, and moral responsibility are incompatible with determinism. Common to all forms of R-libertarianism of concern to us is the principle that an agent performs an action freely only if she was not causally determined to perform it. Agent S does action A and is not causally determined to do A when S does A if and only if S does A, and there is a possible world that is (a) pastwise indiscernible (with respect to “genuine” or nonrelational facts) from the actual one until just prior to either the moment at which S does A, or an earlier moment at which S does something else, B, like forming a judgment to A or an intention to A; (b) governed by the same laws of nature as the actual world; and (c) such that, in it, S does other than A.2 Alternatively, the notion of a causally undetermined action can be understood in this way: Two possible worlds diverge at a time, t, if and only if they are not duplicates at t but share the same past up to a certain point in time (see Lewis 1983). Two possible worlds diverge∗ at t if and only if they diverge at t and they have the same set of laws. Agent S does action A at t in world w and is not causally determined at t to do A in w if and only if there is a possible world, w ∗ , that diverges∗ from w at t or some time prior to t. The intuitive idea here is that an agent is not causally determined to perform an action at the time when she performs it if,

91

given the same laws and the past or near past, she could have performed some alternative instead. As Mele (1995: 208–9) and others like Robert Kane have explained, R-libertarians may grant that agents are sometimes morally responsible for actions that they are causally determined to perform. For example, Kane proposes that even though an agent may be causally determined to do A at time t and hence could not have done otherwise at t, he may still be morally responsible for A-ing at t if, partly through his own causally undetermined free actions, he has made himself into the kind of person who is causally determined to A at t (1996a: 39–43, 77–8). It is partly for this reason that Kane distinguishes between directly free actions and what we can call “derivedly free actions.” Directly free actions are free independently of the agent’s having had control over something that brought about these actions; their being free is not derived from the freedom of any other acts. Derivedly free actions are not directly free. R-libertarians, like compatibilists, agree that agents are morally responsible for their choices or actions only if they have control over them. Earlier we said that an agent has singular control over an action A if and only if she can do A, and that she has dual control over A if and only if both she can do A and she can avoid doing A. Unlike compatibilists, R-libertarians (of interest to us) insist that the sort of freedom or control required for moral responsibility for at least directly free actions is dual and not singular control.3 As actions that agents are not causally determined to perform are ones that their agents could have refrained from performing, given the natural laws and the past, it would seem that R-libertarians, unlike determinists, have the resources to accommodate deontic anchors. Is this so? Let’s examine our first variety of modest R-libertarianism, “Modest Meleian Libertarianism.” MODEST MELEIAN LIBERTARIANISM

Modest R-libertarian theories share a number of important features with the most promising compatibilist views on free action. For instance, they require that in order for an agent to choose or act freely, she must have the capacity to engage in practical reasoning and to guide her behavior in light of the reasons she has. In addition (as we remarked above), mirroring their compatibilist rivals, modest libertarian accounts avoid recourse to agents like Kantian noumenal selves or Cartesian minds and they eschew forms of causation with metaphysically extravagant commitments such as

92

agent-causation. Further, like their compatibilist competitors, they dictate that the choices – the decisions or intentions – for which an agent is responsible be the outcomes of causal processes. In particular, they require that a free decision (or action of a different type) is made for reasons, and its being made for reasons consists partly in its being caused in an appropriate, nondeviant way by the agent’s having those reasons. Modest R-libertarian accounts, though, differ from compatibilist ones in that they hold that at some point (or points) in the causal pathway to action, the causation among actional events be nondeterministic or probabilistic. This requirement is designed (partly) to meet the libertarian’s demand that free actions are not causally determined. The first version of modest libertarianism to be considered is a version that Randolph Clarke (2000: sec. 2) calls a “nonaction-centered” account. It locates indeterminism relatively early in the causal pathway of an action at some point prior to the making of a decision, and it posits that the events that are nondeterministically caused are not actions of any sort. This version of indeterminism seemingly enjoys an advantage over a more robust rival soon to be examined: It evades the charge that actions (such as making decisions) that are directly nondeterministically caused are a matter of control-undermining luck. Consider, as a representative example, one of Mele’s developments of this sort of Spartan indeterminism. Assume that there is some compatibilist account of control (CC) that does not entail nor preclude the existence of probabilistic causation in the causal sequence of events that culminates in full-blown intentional action. The sequence may involve some psychological basis for evaluative reasoning including the agent’s beliefs, values, and desires; an evaluative judgment made on the basis of such reasoning that recommends a course of action; an intention (or decision) formed (or arrived at) on the basis of that judgment; and an action executing that intention (or decision). Suppose Jones intentionally performs some action, A, and when he does, he satisfies the conditions of control for acting freely specified by compatibilist account CC. Assume that these conditions, if met, ensure that Jones has considerable deterministic agential singular control over the events leading from decisive best judgment to overt action. The variety of control at issue is proximal control. Mele (as indicated in Chapter 2) explains that proximal control concerns the direct causal production of agent-involving events such as the agent’s having specific values, desires, and beliefs, his making an evaluative judgment, his forming an intention or arriving at a decision, his executing an intention, and his performing an action. Thus, as Clarke elaborates, proximal control can be taken, first,

93

to constitute wholly or partly different types of direct actional control. An agent would exert this sort of control, for instance, in (nondeviantly and properly) forming an intention, something that qualifies as a mental action (2000: 26). Proximal control may, second, also have a nonactional form. Mele suggests that the event that is the making of an evaluative judgment by some agent would not be an action, and that an agent’s control over such an event would be a function of the way in which it was caused. An event of this kind would be under its agent’s proximal control to the extent that the (nondeviant) causal processes that produced it were free of influences such as compulsion, manipulation, and insanity (Mele 1995: 225; also see Clarke 2000: 26–7). To generate Mele’s brand of a nonaction-centered version of indeterminism, augment the “host” compatibilist theory, CC, with which one starts with the following sort of doxastic indeterminacy: Assume that it is causally undetermined whether certain of Jones’s nonoccurrent beliefs will enter into Jones’s deliberations about whether to A. On the view Mele suggests, in scenarios of the sort he considers, it is causally open, until a best judgment is made, what the agent will decide to do and do, for beliefs may nondeterministically come to mind at any point during deliberation and affect what judgment is reached. So it is causally open, until he has made a best judgment, whether Jones will form a decisive best judgment to A and in fact A.4 The introduction of such indeterminacy satisfies one central incompatibilist requirement for dual control: Jones’s A-ing is not causally determined. Still, such indeterminacy, Mele claims, need not erode singular control. It need not, because once the relevant beliefs have entered into his deliberations, Jones exercises significant deterministic agential singular control (of the type designated by CC) over the events leading from the formation of a decisive better judgment to action. For example, Jones may have considerable singular control over how carefully he deliberates in light of the beliefs, how much weight he gives to them, whether he deliberates in ways that violate his deliberative principles, and so on (Mele 1995: 216 and 1996a: 132–3). Two lessons proposed by Mele are worth recapitulating. First, a modest agent-internal indeterminism restricts our singular control over what happens only in areas in which we have no greater singular control were our world deterministic. This is so, Mele submits, because even in deterministic worlds, not all things that influence our deliberations, such as beliefs that come to mind, are things over which we have singular control. Second, this sort of indeterminism limits our singular control only in a

94

way that gives us no less singular control than we would have were determinism true, while making room for alternative deliberative outcomes (Mele 1999a: sec. 4, 1996a: 132, and 1995: 215–16). AN OBJECTION AND A REPLY

Prior to discussing whether Mele’s nonaction-centered libertarianism can accommodate deontic anchors, it is important to consider a recent, revealing objection to it. Fischer questions whether even this relatively weak type of libertarianism can provide the sort of control that would enable agents to be morally responsible for their deeds. If cogent, the objection would not bode well for prospects of this sort of libertarianism to sustain deontic anchors. Fischer begins his instructive assessment with these searching remarks: How can adding arbitrariness of the sort envisaged – the lack of determination of the beliefs that come to mind during deliberation – to a causally deterministic process yield genuine control? A libertarian . . . will contend that an entirely deterministic process does not contain genuine control by the relevant agent. How, then, can installing the sort of indeterminacy envisaged – indeterminacy as to which belief states will come to the agent’s mind – transform the sequence from one of lack of control to one containing control? This smacks of alchemy. (1999b: 140, note omitted)

Fischer suggests that his worry can be captured by drawing a distinction between an agent’s having control over what happens and its being the case that something different might have occurred: If an agent has genuine control in the sense of possessing alternative possibilities, he can make it the case that one path is followed, or another path is followed, in accordance with what he judges best and chooses. He can deliberately pursue one course of action, or deliberately pursue another; what path the world takes (at least in certain respects) is “up to him.” In contrast, when it is merely possible that something different have occurred, the path the world takes need not depend in the relevant way on the agent. . . . Mele admits that the precise sequence of doxastic states can have an effect on what the agent judges best (and then does); given that the sequence is not (entirely) determined by prior states of the agent (although it is constrained by such states), it is not clear that what the agent judges best and then does is genuinely up to him. (140–1)

Finally, regarding Mele’s suggestion that the agent typically does not control many of the doxastic states he undergoes even if determinism is true, Fischer proposes:

95

the compatibilist will point out that, even though the agent does not directly control what belief-states come to mind (in the sense of choosing them or willing them), they are envisaged as strongly connected to the agent’s prior states to the extent that they are a deterministic product of those past states. Under determinism, one’s prior states – desires, beliefs, values, general dispositions – determine the precise content and ordering of the subsequent doxastic states (that constitute deliberation), even if the agent does not directly control what doxastic states he will be in. . . . It may then be possible to argue that one does give up some measure of control, when one shifts from thinking of the doxastic sequence as deterministic to thinking of it as indeterministic: one gives up the notion that the states constituting one’s deliberations are an “outflowing” of the agent’s prior states in a strong sense. (141)

To respond to Fischer’s charge of alchemy – the charge that adding indeterminacy as to which belief states will come to mind in an otherwise deterministic sequence that, without the added indeterminacy, would be considered by the libertarian to be one in which an agent lacks control – it will be helpful to remind ourselves, once again, of Mele’s distinction between proximal and ultimate control. The former, as we have seen, concerns the direct causal production of agent-involving events. The latter, in contrast, involves the causal influence of agent-external events. In order for an agent S to have ultimate control over an event, say, S’s x-ing at a time t, where x-ing might, for example, be the making of a decision, there must be no time at which there are minimally causally sufficient conditions that do not include events or states internal to S, for S’s x-ing at t. In other words, to have ultimate control over S’s making some decision, roughly, there shouldn’t be conditions “external” to S that are minimally causally sufficient for S’s making that decision. Hence, agents could have ultimate control over their actions only if determinism is false. But proximal control, according to Mele, is compatible with determinism (1995: 211). Modest libertarians should not deny that an entirely deterministic “actional process” is one in which an agent lacks proximal control but need only deny that it is one in which the agent lacks ultimate control. Indeed, Mele insists that satisfaction of his compatibilist conditions for autonomous or responsible action, along with the proximal control that that involves, together with doxastic indeterminism, suffices for the agent’s having ultimate control over the pertinent action. On his view, then, as he says, ultimate control, rather than requiring the possession of any special “control power” beyond the powers required for satisfaction of compatibilist conditions for responsible action, “is something one has

96

in virtue of satisfying the compatibilist conditions and being suitably internally indeterministic” (1995: 213). Transformation of a deterministic actional process from one of lack of ultimate control to one containing such control by installation of the sort of internal indeterminacy that Mele recommends, should, consequently, not smack of alchemy. There still is, though, the question of whether internal indeterminacy erodes or diminishes responsibility-grounding control. It seems, contrary to Fischer, that it does not. For, suppose one accepts Mele’s claim that even if determinism is true, we are not in proximal control of many of our beliefs or thoughts that come to mind during deliberation. Compatibilists would do well to grant this claim. But even granting it, as compatibilists they have no reason to concede that we thus lack responsibility-grounding control over any of our decisions or actions. They might sensibly insist that if it is true that we frequently have no control over which beliefs come to mind prior to deliberative intentional action, it makes no difference to responsibility-grounding control whether the nonoccurrent beliefs that do come to mind are causally determined to do so or nondeterministically come to mind, given that once the beliefs come to mind, one can exert proximal control over them. Consider any point before or after an evaluative judgment is made at which beliefs nondeterministically come to mind. As long as it is true that we don’t have proximal control over such beliefs coming to mind even if determinism is true, and we have compatibilist proximal control over actions or choices that have as part of their etiology such beliefs, internal doxastic indeterminism, it appears, should not erode responsibility-grounding control. An alternative route to a similar moral is this: We can reasonably assume that for an agent to have control over her decisions, these decisions must be caused by her having germane reasons, and that constituents of these reasons will be pertinent beliefs. Modest R-libertarians are in a position to contend that the process that culminates in an agent’s decision includes the same events or states as would be included in an analogous deterministic decision-producing process that compatibilists favor with one difference: In the former type of process, the probabilistic causation of the agent’s decision by her having certain reasons will be governed by nondeterministic causal laws rather than deterministic ones. But probabilistic causation is still causation; and, further, it is not evident how the governing of various events in actional processes by nondeterministic causal laws diminishes proximal control over those events when the governing of analogous events in analogous processes by deterministic causal laws does not.

97

Fischer, as noted, suggests that even if we accept Mele’s claim that an agent typically does not directly control the beliefs that come to the agent’s mind if determinism is true, these beliefs are strongly connected to the agent’s prior states to the extent that they are a deterministic product of those past states. In virtue of being such products, he says, the states constituting one’s deliberations are an “outflowing” of the agent’s prior states in a strong sense. He thinks that it may, hence, be possible to argue that one does give up some measure of control when one postulates actional processes that include doxastic states or events that are nondeterministic, as it will no longer be true that such states are an outflowing of the agent’s prior states in a strong sense. The relevant issue, though, is surely one regarding whether the agent’s proximal control over states constituting the agent’s deliberations is undermined or attenuated in cases in which such states are not “strongly connected” to – they are not deterministic products of – the agent’s past states. Here, again, modest libertarians may plausibly maintain that if we typically have no proximal control over the coming to mind of beliefs that figure in our deliberations, then the fact, if it is one, that such beliefs (or appropriate belief states) are deterministic causal productions of our past states seems not to add anything to the proximal control that we do exert over these beliefs once they come to mind. Analogously, the fact, if it is one, that such beliefs or appropriate doxastic states are nondeterministic causal productions of our past states seems not to subvert the proximal control that we do have over the states once they come to mind. One might concur that placing indeterminism at the point of which beliefs come to mind does not detract from proximal control, and yet worry that it does detract from responsibility, as the following case supposedly shows. Smith has committed six burglaries over the past six months. When examined by a psychologist, he claims that he was raised to have high moral standards and a strong regard for the rights and property of others. He also claims that these considerations just never seem to come to mind when he is pondering the prospects of a new heist. If they had come to mind, he would have refrained from the heist. Now there are two subcases. In the first, the moral considerations did not come to mind prior to each of the heists as a result of indeterminism. Nothing in the agent’s (Smith1’s) antecedent character determined that they not come to mind. In the second subcase, the moral considerations did not come to mind as a result of the agent’s (Smith2’s) moral character determining that they not come to mind. One might be inclined to judge that Smith2 is responsible to a higher degree than Smith1, even though neither exerts

98

proximal control over which considerations fail to come to mind. This is because Smith2’s decision to commit the crimes is reflective of his moral character in a way in which Smith1’s decision to commit the crimes is not. Indeterminism, however, may not be what (partly) accounts for the difference in judgment, even granting the difference. Rather, provided the two subcases are fleshed out appropriately, the explanatorily relevant factor might be that Smith2 but not Smith1 exerts “indirect proximal control” over the pertinent considerations or beliefs. Suppose Smith is largely responsible for acquiring (and maintaining) his character by, among other things, exerting proximal control over factors central to its development. He trains himself so that, consciously or not, he will fail to take moral considerations into account when it suits him not to take them into account. Then, though his character later determines that various moral considerations not come to mind when they do not come to mind, we can say that Smith does exert indirect proximal control over the moral considerations that don’t come to his mind. Roughly, the idea is that such control is derivative from, or “traces” back to, the (direct) proximal control he did exert over the formation of his character that now determines that germane considerations don’t come to mind on germane occasions. The situation here would be somewhat analogous to one in which Smith freely takes a “blockage” pill that ensures that he will later ignore various moral considerations. He is directly free with respect to taking the pill (or at least with respect to his volition to take the pill) and indirectly free at later times with respect to ignoring moral considerations. Reverting to the two subcases, perhaps it’s the thought that Smith2 exerts indirect proximal control over his decisions whereas, Smith1 does not, that rationalizes the difference in the judgment regarding responsibility. Assume that Smith2 is responsible for his character; he has developed it in such a way that, when it comes to heists, he ignores moral considerations. Then though he lacks direct proximal control over its being the case that moral considerations fail to come to mind when they fail to come to mind, he has indirect proximal control over their failing to come to mind. He has such control, in part, in virtue of his character’s determining that these considerations fail to come to mind. In contrast, as Smith1’s failure to take the relevant considerations into account on the relevant occasions is the result of indeterminism, he lacks indirect proximal control over their failing to come to mind. So the two subcases don’t make it clear whether it is the presence or absence of indeterminism or the presence or absence of indirect proximal control that makes the difference to our verdicts.

99

To sort out this concern, we need to consider cases that are on equal terms with respect to considerations of both direct and indirect proximal control and that vary only insofar as indeterminism plays a role in one but not in the other. The following pair meets these requirements. In the first, though Smith3 is largely responsible for his character, he is not responsible for those aspects of it that determine that moral considerations not come to mind when they fail to come to mind. So Smith3 has neither direct nor indirect proximal control over its being the case that pertinent moral considerations not come to mind on pertinent occasions (though it is true that aspects of his character determine that they will fail to come to mind on the relevant occasions). Smith4’s case is just like Smith3’s, save that the moral considerations that don’t come to mind, on occasions when they don’t, is the result of indeterminism. I’m inclined to judge that Smith3 is no more responsible for his decisions than Smith4 is for his. Fischer’s criticisms of Mele, though, contain a genuine insight. Elsewhere, I have remarked that the doxastic indeterminacy of Mele’s modest R-libertarianism does make room for the agent’s having more than one physically possible future and for its being true that the agent could have judged, intended, and acted otherwise than she did. But such indeterminacy does little to persuade us that the agent ensures that she has more than one physically possible future, etc. Which beliefs enter the agent’s deliberations is undetermined, and as a result of such indeterminacy the forking paths future is guaranteed. But compatibilists may well urge that the agent contributes no more to control in this scenario with indeterminacy than she does in a fully determined world (1998a: 28). Fischer pointedly makes analogous observations. Perhaps it will be helpful to isolate what is legitimately acceptable in Fischer’s criticisms from what I believe is not in this way. A modest R-libertarianism of Mele’s stripe doesn’t adequately capture the libertarian’s concern with “ultimacy.” It doesn’t, some libertarians might claim, do adequate justice to the libertarian’s guiding principle that free will is the power of agents to be the “ultimate, buck-stopping originator[s]” of their actions (Strawson 1986: 26); and that for an agent to have libertarian ultimate control over a set of options at a given time is for that agent to be able to bring about voluntarily any one of the options she wills to, for the reasons she wills to.5 Still, distinguish issues of proximal control from those of ultimacy. If compatibilists are willing to grant that we can have proximal control over beliefs once they come to mind despite the fact that the coming to mind of typical such beliefs is something not in our proximal control, then, it seems, doxastic indeterminacy need not erode proximal

100

control. Libertarians can accept this lesson while retaining a healthy dose of skepticism about whether Mele’s brand of R-libertarianism is fully satisfying on the front of ultimacy.6 We can now finally tackle the issue of whether the kind of dual control that Mele’s modest agent-internal indeterminism allows is sufficient for deontic anchors.

MODEST MELEIAN LIBERTARIANISM AND DEONTIC ANCHORS

Suppose it is causally open at time t0, the time at which an agent, Jones, begins to deliberate about A-ing or B-ing, whether (i) he will form a decisive best judgment to A at t1, (ii) he will form an intention or arrive at a decision to A at t2, and (iii) he will in fact do A at t3, where t0 is earlier than t1 and t2 later than t1 but earlier than t3. The modest agentinternal indeterminist might claim that, consistent with the description of the case, as of t0, Jones morally ought, say, to make the decision to A at t2 and to A at t3. Confining our concerns to the sort of control – dual control – required for deontic anchors, this is because as of t0, Jones can decide to A at t2 and A at t3, and as of t0, Jones can refrain from so deciding at t2 and so acting at t3. There are, however, a number of concerns with this claim to accommodate deontic anchors. First, we can suppose that in Jones’s case, what Jones judges best is contingent upon which beliefs in a particular subset of his nonoccurrent beliefs come to mind. In this connection, the following passage of Mele’s is revealing: Suppose that a certain collection C1 of beliefs came to mind during Jones’s deliberation (where C1 is a segment of a much larger collection of beliefs that came to mind), even though it was not causally determined at any point that any member of C1 would come to mind. Suppose also that the coming to mind of these beliefs had some effect on what Jones judged best. Other nonoccurrent beliefs could have indeterministically come to mind instead, and if certain of them had come to mind, Jones would have reached a different judgment. (Mele 1995: 216)

The first concern is this. It is because what Jones judges best depends upon which beliefs in collection C1 come to his mind, and that the coming to his mind of these beliefs is not causally determined, that there is room for deontic anchors in Jones’s world. But Jones’s situation could have been different in a manner that would not have made room for deontic

101

anchors even in indeterministic worlds. For example, if the coming to mind of the beliefs from collection C1 had no effect on what Jones judged best, then even if the coming to mind of these beliefs were undetermined, it would have been false that, as of t0, Jones could both have decided to A at t2 and refrained from so deciding at t2. The principal point here is that it is a contingent matter whether persons in our world, even if this world is punctuated with “doxastic indeterminacy,” are like Jones with respect to whether specific nonoccurrent beliefs that are undetermined affect their deliberations. If such beliefs leave their deliberations unaffected, then such indeterministic worlds won’t open the doors to deontic anchors. A second major concern is that even if modest agent-internal indeterminism accommodates some deontic morality, it accommodates too little. In worlds with doxastic indeterminacy in which the undetermined beliefs that come to mind do affect our judgments, the pathway from best judgment through to overt action is deterministic. Suppose, at some point, t0, in his deliberations, Jones’s reasons cause him to make the judgment that he ought to A at t3. If he intentionally A-s at t3, there is no world, as of t1 – the time at which he makes the judgment to A – with the relevant past and laws held constant, in which he does other than A at t3. Hence, even though at t0 it may be obligatory, right or wrong, for Jones to A at t3, as of t1, it cannot be obligatory, right or wrong, for him to do whatever it is that he does at t3. We can put the point in this way: Modest agent-internal indeterminism may permit for remote obligations, rights, or wrongs but won’t accommodate immediate obligations, rights, or wrongs. In this way, it will severely restrict the “scope” of deontic anchors; for any action A, agent S, and time t, where t is a time later than the time at which S forms some decisive best judgment to do something – whether it is A or some other action – it will be false that as of t, it is obligatory, right, or wrong for S to do A at t. But things might even be worse. Zimmerman (1996: 99) has suggested that there cannot be remote obligation without immediate obligation. To motivate the plausibility of this view, suppose that on Friday I borrow some money that I promise to repay on Sunday. In virtue of promising, assume that I ought (on Friday) to repay the money (on Sunday). On Saturday night, though, I gamble away all the money. As of the time I lose the money, it is no longer true that I ought to repay on Sunday (as I cannot, as of then, repay). Rather, I face a new set of options, including, perhaps, apologizing and searching for other ways to discharge my debt. As the initial remote obligation was not satisfied, a remote wrong has been committed since the final time at which I ought to repay

102

(Saturday night prior to losing the money) does not coincide with the pertinent time of nonrepayment (Sunday). In this case, as Zimmerman indicates, there is a sense in which the remote wrong of nonrepayment is committed by virtue of the immediate wrong of losing the money (the remote wrong would not have been committed, in the manner in which it was committed, if the immediate wrong – losing at time t on Saturday night the money at t – had not been committed) (1996: 99). So perhaps a remote wrong is wrong only because something else is immediately wrong. But then if nonaction-centered modest libertarianism cannot accommodate immediate wrongs (or obligations), it won’t be able to accommodate remote wrongs (or obligations) either. Perhaps one might object that Mele’s version of modest libertarianism would allow for some events being immediately wrong or obligatory. A reasonable candidate, for example, might be the very making of an evaluative judgment. But even if we grant that such events could be immediately wrong or obligatory for an agent at a time, these events are not actions. So once again, nonaction-centered modest libertarianism would accommodate too little deontic morality. Further, an event like the making of an evaluative judgment is not, intuitively, the sort of event upon which the wrongness (or obligatoriness) of an action that is remotely wrong (or obligatory) is wholly parasitic. Hence, if there cannot be actions that are remotely wrong (or obligatory) without there being appropriate actions that are immediately wrong (or obligatory), then even if events that are not actions can be immediately wrong or obligatory if Mele’s nonaction-centered libertarianism is true, this fact will not support the further desideratum that the libertarianism in question open the doors to accommodating actions that are immediately wrong or obligatory. Perhaps these sorts of worries to accommodate deontic morality could be sidestepped by advancing a theory that located indeterminacy further along the pathway of events, for instance, between reasons and the making of decisions, culminating in action. We can judge whether this is so by turning to a more robust form of R-libertarianism than Mele’s.

103

7 Robust Modest R-Libertarianism and Luck

I argued in the last chapter that Modest Meleian Libertarianism, despite its provision of dual control, is inhospitable to deontic anchors. In this chapter, I explore whether a more robust variety of libertarianism holds better promise for accommodating deontic morality. I defend the cogency of the objection that, if this sort of libertarianism is true, our choices and actions are a matter of luck undermining responsibility, in these steps: First, I propose a necessary condition of responsibility. In broad strokes and with qualifications to be noted, this is the condition that an agent is responsible for a choice only if there is an explanation (that need not be deterministic) in terms of prior reasons of why the agent made that choice rather than some other. Second, I provide a rationale for this condition that appeals centrally to aspects of the concepts of blameworthiness and responsibility in general. In particular, the rationale exploits the view that ascriptions of responsibility disclose what an agent morally stands for in relation to deeds for which she is responsible; responsibility requires a sort of “selfexpression.” If the luck objection does seriously threaten responsibility, then it seems that it should threaten deontic morality as well. If no agent can, for instance, be blameworthy for an action that is a result of luck, then equally, it would seem, no agent can, for example, do moral wrong by performing an action that is the outcome of luck. But a careful appreciation of why a robust R-libertarian view runs afoul of the luck objection prepares the way for showing that this initially plausible symmetry does not in fact obtain.

104

ROBUST MODEST R-LIBERTARIANISM

There are at least two different motivations to move from a relatively benign version of R-libertarianism of Mele’s variety to a more robust version defended, for example, by Kane. First, Modest Meleian Libertarianism, it seems, escapes the worry that the agent’s actions are a matter of luck, and this is a large point in its favor. All the same, despite Mele’s suggestion that his variety of libertarianism may be the best thing R-libertarianism can get, that variety does have a down side, or at least so other free will theorists might urge. The problem, as they see it, is that Mele’s modest libertarianism does not provide as full an account of moral responsibility as most libertarians want. This is because it fails to satisfy adequately a central tenet of libertarianism, that free will is the power of agents to be the ultimate, “buck-stopping” originators of their actions, and that for an agent to have libertarian control over a set of options at a given time is for that agent to be able to bring about voluntarily any one of the options, including its antecedent actional springs, she wills to, for the reasons she wills to. With the type of doxastic indeterminacy that is part of Mele’s view, which beliefs indeterministically come to an agent’s mind is not under the control of the agent, while the agent’s final choice is, thereafter, determined by these kinds of belief together with other background conditions. Hence, skeptics of Mele’s view might charge, the agent is not responsible for which beliefs come to mind and is only responsible in a compatibilist sense of ‘responsible’ for what occurs after they do come to mind. A second unrelated motivation to shift from a benign to a more robust species of libertarianism, as I have suggested, is that the robust version may fare better than the benign one in accommodating deontic morality. A robust modest R-libertarianism is, in Clarke’s terminology (Clarke 2000: sec. 2) an “action-centered” libertarianism. On this view, the event that is nondeterministically caused is a decision itself; the view is action centered since making a decision is performing a mental action. The recipe for generating a working version of this sort of libertarianism that I favor simply involves starting with our best compatibilist view of control (or freedom) and then adding to it the requirement that free decisions themselves are to be nondeterministically caused. The resulting robust (modest) libertarianism specifies that an agent’s responsibility- or deonticgrounding proximal control in making a decision consists in the decision’s being nondeviantly caused by apt agent-involving events. The degree of control that the agent exercises depends on which agent-involving events actually cause the decision and on their etiologies. It deserves emphasis

105

that on such a libertarian view, the factors that constitute an agent’s proximal control in making a decision are the very ones shared by this view and its compatibilist host: roughly, the decision’s being nondeviantly caused by deliberative processes with appropriate causal histories. So even this version of modest libertarianism offers nothing more in the way of proximal control that an agent would exercise in making a free decision than would be offered by its compatibilist host.1 But just like its less robust nonactioncentered cousin, the action-centered account allows for genuinely open futures: When a decision is freely made, the account implies that there remained, until the making of that decision, a chance that the agent would, instead, make an alternative decision for alternative reasons or that she would then make no decision at all. Although I have focused on decision (which is a type of mental action) in outlining an action-centered view, what has been described about the freedom of decision applies straightforwardly, with appropriate changes, to freedom of other sorts of action as well. To appreciate better this kind of R-libertarianism, it will be useful to adumbrate Kane’s elegant formulation of this sort of view. Free will, Kane says, is the “power of agents to be the ultimate creators (or originators) and sustainers of their own ends or purposes” (1996a: 4). A person is morally responsible for an action, on Kane’s view, only if he is ultimately responsible for it. To be ultimately responsible for an action, “the agent must be responsible for anything that is a sufficient reason (condition, cause, or motive) for the action’s occurring” (1999a: 106). The connection proposed by Kane between ultimate responsibility (UR) and alternative possibilities plants Kane firmly in the incompatibilist’s camp. As he explains, “UR does require that we could have done otherwise with respect to some acts in our life histories by which we formed our present characters. (I call these ‘self-forming actions’ (SFAs) . . . or at times self-forming willings)” (107). Ultimate responsibility for a decision is manifested at those times in life when we are “torn between competing visions of what we should do or become” (108). Self-forming actions, for instance, occur when there is an internal conflict in the agent between what the agent morally ought to do and what he prudentially ought to do (26–8). Kane claims that in such conflict situations an agent’s choice or decision to do something is often preceded by an effort of the agent’s will – a mental action – that, in conjunction with other actional elements, gives rise to the choice or decision. These efforts of will that terminate in choice in conflict situations are indeterminate. Their indeterminacy, Kane suggests, is analogous to the indeterminate position or momentum of a

106

microphysical particle (128). If agents’ choices or decisions are not determined in such situations, “they might choose either way, all past circumstances remaining the same up to the moment of choice” (127).2 Since the effort to resist temptation in one way or another is indeterminate, the choice that terminates the effort is undetermined (128). Kane emphasizes that whatever self-forming choice one makes in a conflict situation for which one is ultimately responsible satisfies the “plurality conditions” of plural voluntariness, plural rationality, and plural control.3 A choice is voluntary if it is done in accordance with the agent’s will (the agent’s predominant desires or motives) and it is not coerced or compulsive. It is plural voluntary if it is voluntary and there was at least one alternative choice open to the agent that would have been voluntary had it been made instead (109–10). A choice is plural rational if rational when made and there was minimally one alternative choice that the agent was able to make at that time that would have been rational had it been made then. In conflict situations, agents’ prudential choices are plural rational in that regardless of which choices the agents make, “the agents (r1 ) will in each case have had reasons for choosing as they did; (r2 ) they will have chosen for those reasons; and (r3 ) they will have made those reasons the ones they wanted to act on more than any others by choosing for them” (135). To have plural control over a set of options at a given time is to be able to “bring about any one of the options (to go more-than-one-way) at will or voluntarily at the time. That is, . . . 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 do it, and in such a manner that neither your doing it nor willing to do it was coerced or compelled” (111). A failed effort culminates in a choice of the enticing course of action, whereas a successful one culminates in the choice of what the agent deems right or best. The agent’s “past character and motives . . . influence the choice without determining it” (127). In sum, Kane argues for the intuitively attractive view that free will is composed of undetermined self-forming actions by virtue of which the agent “sets his will” in one direction and is thus the ultimate creator of some of his own purposes. ROBUST R-LIBERTARIANISM AND THE LUCK OBJECTION

Despite its many attractions, it appears that this sort of action-centered theory runs afoul of the luck objection.4 In the interests of generality,

107

I develop the objection independently of appealing to particular features of Kane’s theory. The crux of this objection can be highlighted by a simple illustration. Suppose Jones has reasons for and against smoking. After deliberation, he concludes that, whereas he morally ought not to smoke, all things considered, he has better reasons to smoke. He then straightaway decides to smoke in accordance with this best judgment and voluntarily and intentionally smokes. As his undetermined decision to smoke is nondeterministically caused by his best judgment and other actional events and states, even having made this best judgment, he could have decided to do something else instead. So suppose, holding constant the relevant past that includes Jones’s prior deliberations and his best judgment that he ought to smoke, we consider the nearest possible world in which Jones (or if we want, Jones’s counterpart, Jones∗ ) straightaway decides to refrain from smoking and then voluntarily and intentionally refrains from smoking. Here, apart from concerns about whether we can account for Jones∗ ’s having nonakratically refrained from smoking, it appears that it is a matter of luck that he decides as he does. As Mele has commented, if one agent smokes and the other refrains, “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 just is a matter of luck” (Mele 1999a: 280). Luck of this sort seems incompatible with moral responsibility. On one interpretation, the part of the objection that associates luck with lack of responsibility unfolds in this fashion: (1L) It is a matter of luck that Jones decides as he does. (2L) To the extent that some choice or action is a matter of luck for an agent, it is out of that agent’s control. (3L) If 1L and 2L are true, then Jones’s deciding as he does – his decision to smoke – is out of his control. (4L) If Jones’s deciding as he does is out of his control, then Jones is not morally responsible for his decision. (5L) Therefore, Jones is not morally responsible for his decision.

In responding to some of Fischer’s concerns about Mele’s libertarianism in the last chapter, I recommended that we prise apart concerns of ultimacy from those of proximal control and focus on the latter. In connection with the present argument, one might similarly propose that if we keep distinct the notions of ultimate and proximal control and direct attention to the latter, it appears that both Jones and Jones∗ exerted proximal control over their choices and ensuing actions. If they did, the sort

108

of luck that affects their scenario is not one that subverts proximal control over their choices. Hence, line (2L) can be rejected, assuming that ‘control’ in that line refers to proximal control, and that ‘luck’ denotes the same sort of luck found in play in Jones’s case. A problem of responsibility-subverting luck, though, still persists. It is intimately tied to lack of an explanation in terms of prior reasons of the difference in choices that Jones and his counterpart Jones∗ make. A preliminary version of the problem surfaces upon reflecting on the following thought experiment. God has a thousand times caused the world to revert to precisely its state at the moment just before Jones decides to smoke, and that on about half these occasions, Jones decides to smoke and acts accordingly (see van Inwagen 1983: 141). We can go so far as supposing (though this is not essential) that in each of the “reruns,” Jones was trying to do two competing tasks – he was trying to smoke and he was trying to refrain – and that whatever he eventually did, it was done voluntarily, intentionally, and for reasons.5 Yet, in half the reruns he decides to smoke and in the remaining he decides to refrain. Suppose, watching a playback of the recorded runs, Jones himself wonders: “I realize that I was trying to perform two competing tasks in each of the reruns; still, given relevant type-identical pasts, why did I decide one way in about half of them but another way in the others?”6 This would be a sensible question for a thoughtful person to ask. Further, I believe that it is lack of a contrastive explanation in terms of prior elements in an action’s trajectory – why Jones smoked rather than not – that fuels the intuition had by many that what Jones does in each of the reruns is a matter of responsibility-subverting luck. The absence of such an explanation does not entail that when Jones or Jones∗ makes the choices that he does, he lacks proximal control over his choices. So the idea I am proposing and wish to develop is that there can be (nonepistemic) factors that subvert responsibility even though they do not detrimentally affect an agent’s proximal control over various actional elements, like decisions, of hers. Against this sort of worry, Kane rejoins: The absence of an explanation of the difference in choice in terms of prior reasons does not have the tight connection to issues of responsibility one might initially credit it with. For one thing, the absence of such an explanation does not imply . . . that . . . [ Jones and Jones∗ ] (1) did not choose at all, nor does it imply that they did not both choose (2) as a result of their efforts, nor that they did not choose (3) for reasons (different reasons of course) that (4) they most wanted to choose for when they chose, nor that they did not choose for those reasons (5) knowingly

109

and (6) on purpose when they chose, and hence (7) rationally, (8) voluntarily, and (9) intentionally. None of these conditions is precluded by the absence of an explanation of the difference of choice in terms of prior reasons. Yet these are precisely the kinds of conditions we look for when deciding whether or not persons are responsible. (1999b: 239)

Kane’s challenge can be taken to be the following: Even if, in cases like the Jones/Jones∗ one, there is no explanation of the difference in choice in terms of prior reasons, this is perfectly consistent with all conditions of responsibility being met. It is consistent with, for instance, the agent’s having proximal control over his decisions or other actions, not being relevantly ignorant, and acting on “authentic” springs of action, roughly, springs of action that are “truly the agent’s own,” and not the product of unsolicited hypnosis. If all these conditions are met in Jones’s case and these conditions suffice for responsibility, how can lack of the relevant sort of contrastive explanation undermine Jones’s responsibility for his choices or actions?

PICKING UP THE GAUNTLET

To meet this challenge, notice that in Jones’s “rerun scenario,” the control Jones lacks in each rerun is antecedent control – proximal or otherwise – to see to it that, in that rerun, he smokes rather than that he does not and vice versa. For suppose, contrary to this claim, we assume that Jones did have antecedent proximal control over whether or not he would smoke in a particular rerun, he successfully exerted such control, and in virtue of exerting this control, he smoked in that rerun. But then, with fixed pasts, whatever proximal control Jones exerted over his having certain values, making a certain best judgment, forming a certain intention, and so on, Jones∗ , in a rerun in which he refrained from smoking, would have exerted as well. It would seem, consequently, that Jones∗ , too, just like Jones, in virtue of having successfully exerted the sort of antecedent proximal control over whether or not to smoke that Jones exerted, would have smoked in his rerun. Unlike Jones, though, Jones∗ refrained from smoking in his rerun. It follows that the initial assumption that Jones had antecedent proximal control over the relevant state of affairs is false. The point can be put in a different way. With fixed pasts, the difference in outcome in Jones’s and Jones∗ ’s cases appears to be merely a function of the indeterminacy in the actional pathways leading to choice. But it

110

would seem that no agent could exert proximal (or any other sort of ) control over such indeterminacy to ensure a particular outcome. One might attempt to resist the claim that Jones lacks antecedent control to bring it about that he decides to smoke, or that he smokes, rather than that he does not so decide or does not so act (or vice versa) in a particular rerun by reminding us that in an action-centered view, an agent’s freedom-level control stems from his performance of directly free actions. The freedom of such actions is not derivative from the agent’s having freely performed earlier actions or nonactional events that, in turn, brought about these actions. Rather, an agent’s control to ensure what he decides is centered on his action of deciding, not on something that occurs prior to that action and over which he lacks freedomlevel actional control. So the rejoinder, then, would be that if Jones does freely decide to smoke in a particular rerun, holding fixed the past, he did have, prior to making this decision, antecedent control to see to it that he decided one way rather than the other. He had such control in virtue of his being able to ensure what he would later intend by exerting direct freedom-level control in making whatever decision that he did make. But this view is not problem-free either. Clarke (2000: 26) explains rightly that at least part of what constitutes the exercise of direct control of an action (including a mental action such as deciding) is the proximal causation of that action – its being caused in a nondeviant way by the agent’s having certain desires, beliefs, and so on. We may safely assume that in the rerun(s) in which Jones decides to smoke, Jones exercises this type of direct control in making his decision to smoke. Assume that apart from this sort of direct proximal causal control, Jones exerts additional direct freedom-level control in making his decision to smoke. Now either exercising this additional causal control involves antecedent states of the agent or antecedent agent-involving events or it does not. On the one hand, if the latter is true, then it seems that we have abandoned modest libertarianism as the sort of control in question involves directly causing making a decision but not by way of any prior agent-involving states or events. It may be contested that the additional direct freedom-level control in making the decision is centered on the very action of deciding and not on something that occurs prior to that action and over which one altogether lacks freedom-level actional control (see, for e.g., Clarke 2000: 34). This is puzzling, however, given that the sort of freedom-level control in question is causal in nature, and that the event under consideration that is supposed to be the causal

111

outcome of the exertion of this sort of causal control, is the very action of deciding. On the other hand, if the former is true (that is, if the additional direct freedom-level control in deciding does include antecedent states or agentinvolving events), then, given fixed pasts, type-identical antecedent states or agent-involving events to the ones that played a role in Jones’s exercise of direct freedom-level control in his deciding to smoke, should also have played a role in Jones∗ ’s direct freedom-level control in his deciding as he did. But then, again, with fixed pasts, it would seem that if Jones decided to smoke in virtue partly of successfully exercising a type of direct freedom-level control in making the decision that he did, Jones∗ too should have decided analogously in virtue of successfully exercising this same sort of control. I conclude that Jones lacked antecedent proximal control to see to it that he would decide to smoke (or smoke) rather than that he would not so decide (or not so act). Lack of such antecedent control, in turn, suggests a general principle that rationalizes, in part, why Jones is not morally responsible for his choices and subsequent actions in each of the reruns. I formulate the principle first and then give a rationale for it. Assume type-identical relevant pasts in the reruns. So, for instance, on the supposition that in one of the reruns, Jones’s desire/belief pair DB that gives rise to a best judgment to refrain from smoking, that, in turn issues in the decision to refrain from smoking, is motivationally stronger than a “competing” desire/belief pair to smoke, we are to assume analogous type-identical actional antecedents (including a stronger desire/belief pair to refrain from smoking than a desire/belief pair to smoke) in a rerun in which Jones decides to smoke and smokes. The principle can now be formulated in this fashion: No Contrastive Control (NCC): If it is the case that (a) if agent S does action A, S will have proximal control in doing A; (b) if S does B, S will have proximal control in doing B; (c) the “antecedent actional elements” that would causally give rise to S’s doing A if S were to do A are type-identical to those that would causally give rise to S’s doing B if S were to do B; and (d) it is false that S has proximal or other antecedent control over seeing to it that S does A rather than that S does B or vice versa, then (e) S is not morally responsible for doing either A or B if S does either A or B. In Jones’s scenario, conditions (a) and (b) are both satisfied on the stipulation that ‘A’ stands for Jones’s refraining from smoking and ‘B’ for his smoking. Consistent with libertarian assumptions, (c) is true by hypothesis. At the time of smoking or refraining from smoking, of course, 112

in the reruns in which Jones smokes, his reasons to smoke prevail in the sense that having these reasons nondeterministically causes him to smoke, whereas in the reruns in which he refrains from smoking, his reasons to refrain from smoking prevail. Finally, as explained above, (d) is also true. Thus the consequent (e) appears reasonable. The fundamental idea principle NCC seeks to bring to attention is that there are nonepistemic factors that do not undermine an agent’s proximal control over his or her deliberations, best judgments, intentions, and so forth but nonetheless, subvert responsibility. It has been customary to suppose that at least two conditions require satisfaction for an agent to be morally responsible for her decisions or choices, an epistemic one, that says, roughly, that the agent not act in relevant ignorance; and a control or freedom condition that says, loosely, that the agent act freely.7 My proposal is that if NCC is true, then there is yet another condition which we can dub the “Contrastive Explanation Condition” (CEC). A preliminary formulation of this condition is this: PCEC: Agent S is responsible for decision or action A only if there is an explanation (that need not be deterministic) in terms of prior reasons of why S does A rather than something else (like refraining from doing A or some other thing, B). It is a condition of this sort that fails in the Jones/Jones∗ case. It is refined in the next section. REFINING PCEC

To get a better handle on CEC, it should be clarified at the outset that I am not denying (like Kane (1995) or Thomas Nagel (1986: 115–16)) that causal contrastive rational explanations are possible for putatively free choices not determined by prior events. Rather, the proposal is that an account of the freedom-relevant component of responsibility that fails to satisfy CEC is defective. To refine PCEC, let’s note a few things about contrastive explanations. In an instructive paper, Clarke (1996b) adapts Peter Lipton’s account of causal contrastive explanations for events in general to the task of providing an account of such explanations for rational, free choice. Lipton offers the following “difference condition” as a necessary condition for contrastive explanations: To explain why p rather than q, we must cite a causal difference between p and not-q consisting of a cause of p and the absence of a corresponding event in the history of not-q. (1991: 43)

113

Clarke argues persuasively that a construal of this account applies straightforwardly to contrastive explanations for choices, including explanations for choices in cases in which the choices to be explained are the result of indeterministic causal processes, and in which an explanation is desired of why an agent made choice p rather than not-p. According to this construal, to explain why an agent chose p rather than q (for example, why Sue chose to go for a bike ride rather than not), it must be the case that (i) there is some event, e, that is the cause of p; (ii) e bears an explanatorily relevant relation R to p; that is, e is explanatorily relevant to p’s occurrence; and (iii) had q occurred, nothing in the actual history of p’s occurring would have borne the same explanatorily relevant relation R to q that e bore to p. Clause (iii) is designed to capture Lipton’s insight that a causal contrastive explanation of why p rather than q requires that there be a cause of p that lacks a corresponding event in the history of q (see Clarke 1996b: 188–9). As an illustration, suppose Sue has reasons for and against going for a bike ride. After reflection she judges, in light of her reasons, that it would be better to go for a ride. On the basis of this judgment and the reasons for which she made this judgment, she chooses to go for the ride and does so. Assume that, even given her judgment that going for a ride would be better, it remained causally open to Sue to choose not to go; this judgment simply rendered it highly probable that she would choose to go for the ride (Clarke 1996b: 192–3). It is, consequently, false that, given her judgment, it had to happen that she choose to go. We can now ask: Why did she choose to ride her bike rather than not? Assuming that people generally choose what they judge to be better, the difference condition as adapted by Clarke, applied to this case, implies that (i) there is some event – Sue’s judging that going for a ride is better – that is a cause of her choosing to go for a ride; (ii) this judgment bears an explanatorily relevant relation to her actual choice (it causes and rationalizes her actual choice); and (iii) had she made the alternative choice, nothing in the actual history would have borne the same explanatorily relevant relation to that choice as her judgment does to her actual choice. Simply put, the judgment that she actually made caused and bears an explanatorily relevant relation to her actual choice that no actual occurrence would have borne to her choosing not to go, had she made that alternative choice. This causal difference, it seems, rationally explains the contrast in question. Sue’s case, thus, renders it highly probable that, at least in some cases, causal contrastive explanations are possible for choices not determined by prior events.

114

Now let’s refine PCEC. Clarke indicates that there are cases involving choice where a contrastive explanation is not possible but where the choice can be rationally explained. In his example of the several items in a vending machine, only two, a chocolate bar and the tortilla chips, interest Edna. The chocolate, Edna thinks, would be good and sweet, the chips good and salty, but neither would be better than the other. Edna chooses to buy the bar and does so. Clarke suggests, plausibly, that though her choice was not uniquely rational, it was not in any respect rationally deficient. And though we can provide an adequate rational explanation of it by citing her taste for the chocolate, it doesn’t appear that we can explain why she chose the chocolate rather than the chips; clause (iii) of the difference condition would fail (1996b: 195). The first refinement of PCEC is the restriction that it applies to intentional action or choice that does not involve rational indifference among the relevant alternatives as in Edna’s case. A second refinement is more involved. Revert to the Jones/Jones∗ case and let’s ask whether it is possible to give a contrastive explanation of why Jones∗ did not smoke rather than smoke. There is a complication regarding the event that, if there is one, supposedly causes and rationalizes his not smoking (or, in a variation of the case, his choice not to smoke). Two candidates for this event, it might be proposed, are (e1), Jones∗ ’s reasons to refrain, or (e2), Jones∗ ’s making his reasons to refrain prevail by choosing to refrain.8 If e1 is the relevant event, the difference condition implies that no contrastive explanation is possible, because the case will fail to satisfy clause (iii) of that condition. For had Jones∗ smoked, some event in the actual history – his having reasons to smoke – would have borne the same relevance to that outcome that e1 bears to Jones∗ ’s refraining to smoke (or his choice to refrain to smoke). In contrast, if e2 is the relevant event, clause (iii) of the difference condition would be satisfied. For had Jones∗ smoked, nothing in the actual history of his not smoking would have borne the same relevance – Jones∗ ’s making his reasons to smoke prevail by choosing to smoke – that e2 bears to his not smoking. Put another way, the difference condition requires us to find a cause of Jones∗ ’s not smoking that bears a certain relevance R to his not smoking, such that, in the actual history of his not smoking, there is no event that would have borne R to Jones∗ ’s smoking, had he smoked. Whereas e1 would not fit this bill, e2 would. One might, then, initially believe that, if we simply latch on to the correct “differentiating” event, a contrastive explanation of the sort that I am calling for is forthcoming even in the Jones/Jones∗ case.

115

However, there is a problem with identifying the relevant event with e2. If we want an illuminating constrastive explanation of why Jones∗ did not smoke rather than smoke that appeals to something like e2, we would want to know why Jones∗ made his reasons to refrain from smoking prevail rather than his reasons to smoke. But it appears that there is no contrastive explanation of this fact. I think that this is so because of the following: Suppose, for example, that Jones∗ judges that his reasons to refrain from smoking are better than his reasons to smoke and that he has a strong standing disposition to be motivated by the reasons that he has judged, in light of other reasons of his, to be better. We may, it might be urged, consequently have an answer to the second contrastive question as well. But then it would follow that Jones, who is Jones∗ ’s counterpart and who acts on antecedent actional springs type-identical to those on which Jones∗ acts, also judges that his reasons to refrain from smoking are better. As Jones, unlike Jones∗ , intentionally and supposedly freely smokes, his act of smoking would be irrational, in one sense of ‘irrational,’ since it is an act that is contrary to a better judgment of his. Presumably, though, modest libertarians would not want to be saddled with the burden that in cases like the Jones/Jones∗ one, free choice or action on the part of each is guaranteed only if one of the two acts irrationally. The libertarian’s view is that whichever choice Jones makes, it is one that is freely made and is rational. There is another, related motivation for refining PCEC. In the case in which the bike rider, Sue, makes the judgment that riding would be best and chooses, therefore, to ride the bike, there is a contrastive explanation of her choice even if it is also possible that she choose to refrain from riding under exactly the same antecedent conditions. But suppose that all the conditions are the same, including Sue’s making the judgment that riding would be best, and Sue chooses to refrain from riding the bike. In that case there seems to be no contrastive explanation of that choice. Again, one might claim that I am mistaken about this, for such an explanation is forthcoming if we isolate the correct differentiating event. The event (e3), it might be suggested, is Sue’s making her reasons to refrain from riding prevail by choosing to refrain. But then, one might wonder why Sue made these reasons prevail by so choosing rather than making the reasons to ride prevail by choosing appropriately. As I explained above, there would be, it seems, no contrastive explanation of this fact. If contrastive explanation is required for responsibility, then Sue would not be responsible for the choice to refrain from riding. Thus, while more than one choice – choosing to ride or choosing not to ride – is possible

116

given antecedent conditions, only one of those choices (her choice to ride if she made that choice) could be one for which Sue is responsible. This does not provide the sort of dual responsibility that libertarians want. Given these considerations, I propose that PCEC be amended in this way: CEC: Barring cases of intentional choice or action that involve rational indifference among salient alternatives (like Edna’s), agent S is responsible for choice or action A only if: (a) in cases in which there is a contrastive rational explanation of S’s choice or action A that appeals essentially to the notion of S’s making one set rather than another of S’s reasons prevail, there be a contrastive rational explanation (that need not be deterministic) of S’s making that one set rather than another of S’s reasons prevail; or ( b) in cases in which it is false that there is a contrastive rational explanation of S’s choice or action A that appeals essentially to the notion of S’s making one set rather than an alternative of S’s reasons prevail, there be a contrastive rational explanation (that need not be deterministic) in terms of prior reasons of why S does A rather than something else (like refraining from doing A or some other thing, B). SUPPORT FOR NCC

I think CEC is in the right ballpark and since it is motivated largely by NCC, we need to ask whether NCC itself is credible. I believe that the concepts of moral blameworthiness and that of responsibility in general help to bolster NCC. I start by sketching the notion of blameworthiness and then trace its implications for the tenability of NCC. Blameworthiness for a deed consists in a person’s being deserving of being judged in a certain way, the judgment recording the moral worth of the person with respect to that deed. A person of generally high moral standing or overall good character can be blameworthy for an action, and when she is, her moral worth is diminished as a result of its performance. Moral blameworthiness is frequently (but not always) closely associated with the moral beliefs a person thinks important in guiding her behavior. When deserving of blame for something, often (but again not always) the agent does what she takes to be in violation of the moral standards she endorses – those she may accept as authoritative in guiding and constraining her behavior and on which she typically relies to arrive at practical judgments or decisions about what to do. It is in virtue of the agent’s having done something she believes is below par that it is frequently fitting in instances of blameworthiness for the agent to have negative feelings or attitudes (like regret or remorse) toward herself and for other parties

117

to adopt appropriate negative attitudes toward her. But these feelings or attitudes are not necessary elements of blameworthiness. A person can be deserving of blame on account of something even though she feels nothing like remorse or guilt for doing it, and even though nobody else is aware that she is to blame for it, and hence even in the absence of any overt blame on the part of others. Pivotal to blameworthiness is the constituent that one is to blame for some action A only if one has a belief (even a dispositional one) that A is wrong or morally amiss; one has this belief irrespective of whether or not one cares about morality (one may, for instance, endorse prudential norms); one performs A despite the belief that it is morally wrong or amiss; and one believes (at least dispositionally) that one is performing A. Augmenting this epistemic or cognitive core are other conditions like being in appropriate control of the action one takes to be morally untoward. (For elaboration, see Haji 1998a.) As I mentioned in a prior chapter, my views on responsibility, in general, align themselves closely, in certain respects, with the account defended by Dewey and others that when one is responsible for what one does, one manifests or expresses in conduct what one would have oneself to be; one discloses in conduct one’s moral stance or commitment vis-`a-vis a particular episode in one’s life. Concerning blameworthiness, the view I have sketched resonates with this self-disclosure account. If at the heart of blameworthiness is the core that a person is blameworthy for an action only if she performs it in light of the belief that she is doing wrong, then when a person is indeed blameworthy for an action, there is a very real sense in which she “discloses” what she morally stands for in performing that action, as, roughly, when she acts, she does so in light of certain of her moral beliefs. Reconsider, now, the Jones/Jones∗ case. Entertain, once again, the thought experiment that God has a thousand times caused the world to revert to precisely its state at the moment just before Jones decides to smoke, and that on about half these occasions, Jones decides to smoke and acts accordingly. Assume, again, that in each of the reruns, Jones was trying to do two competing tasks, and that whatever he ended up doing, he would have done voluntarily, intentionally, and rationally. Suppose, as I have proposed, responsibility gauges the moral worth of an agent with respect to some episode in her life – a person discloses what she stands for when she is morally responsible for some deed. Then, given typeidentical pasts, when Jones does one thing in half or so of the reruns but

118

something else in the others, there is no saying what Jones stands for. It’s not as though, for instance, Jones acted akratically on half the occasions but with self-control on the others. If this were the case, appraisals of the moral worth of Jones with respect to what he did on each occasion would surely be possible. Nor is the problem just epistemic, stemming merely from not knowing whether a positive or negative mark gets recorded in Jones’s ledger of life on account of his deeds. Rather, there is no fact of the matter as to what mark gets entered, or more aptly, should get entered into his ledger. Perhaps, one might object, Kane would have an obvious answer to this concern in light of his response to the luck objection discussed earlier. Since Jones’s will is conflicted before he makes the choice, he ( Jones) in fact stands for different things (as all persons do when their wills are deeply conflicted). Jones is an avid smoker, but he is also someone committed to overcoming the habit. And in making the choice, Jones temporally chooses, for the time being at least, to go with and endorse one of the things he stands for rather than the other, while Jones∗ temporarily chooses and endorses the other thing he stands for. The disconnection between character, responsibility, and choice implied by the present concern need not necessarily obtain for either Jones or Jones∗ . Each might be responsible for choosing one of the competing things he “stands for” at this moment even though each chooses differently. In response, the current concern about luck revolves on a distinct variety of moral appraisal of the agent. The “stands for” locution is used to signal assessments pertaining to what agents reveal about themselves in relation to specific deeds that they perform, where what is revealed is something moral. Specifically, the sort of appraisal in question is an appraisal of an agent’s moral worth vis-`a-vis a particular episode in that agent’s life, and not an appraisal of the moral statuses of various options open to the agent, or some sort of normative appraisal of the different commitments of the agent. Nor is the “stand for” locution used merely as an indication of the fact that the agent endorses different commitments. In the conflict situation in which Jones finds himself, it is clear to Jones what he ought morally to do and what he ought, all things considered, to do; and that he cannot, in his situation, jointly fulfill these clashing demands. But prior to making a choice, under his circumstances, Jones has not yet augmented nor diminished his moral worth with respect to the choice that he will make. So it is misleading to claim that Jones “stands for” different things before making a choice. Prior to choice, he “stands for”

119

different things only in the undisputed sense that he is torn between two conflicting options. With this in mind, the kernel of my worry about luck can be brought out in this manner: If Jones, with one set of moral values, reasons, and motives, and fully cognizant of what morality demands of him, does one thing under one set of circumstances on one occasion, but something radically opposed, given those very same sorts of values, reasons, and motives, and knowledge of what morality demands of him, on another occasion under type-identical circumstances, there is a genuine puzzle as to what judgment is appropriate regarding his moral worth in connection with the choice that he makes. Is his moral worth augmented or diminished in relation to the choice that he makes, and what would be the basis for determining whether it is the one or the other, given opposed choices under type-identical conditions? Jones simply fails, on the pertinent occasion, to manifest what he would have himself to be; it is as if the “evaluative agent,” the agent who is the proper subject of responsibility appraisals, disappears from the picture. It’s no surprise, then, that there is bewilderment on Jones’s part when he puzzles about why he decided one way in half the reruns and a different way in the others. ROBUST R-LIBERTARIANISM AND DEONTIC MORALITY

To sum up, I have attempted to defend the view that the luck objection against certain versions of modest libertarianism does not get its bite from the thought that indeterministic choice erodes proximal control. I have conceded that even if an agent’s decision is nondeterministically caused, the agent may well have proximal control in making that decision and its subsequent translation into action. Rather, the luck objection is rooted in the idea that, when there is no contrastive explanation, deterministic or indeterministic, in terms of prior reasons why the agent made one choice rather than another (or why the agent made one rather than some alternative set of reasons prevail), disclosure of what the agent stands for vis-`a-vis what she did is not possible; the self-expressive requirement of responsibility is not fulfilled. Further, with no appropriate entry in her ledger of life, she cannot be morally responsible for her deed.9 Can, however, actions that result from such indeterministic choices be right, wrong, or obligatory? After all, there is a sense in which such actions are still luck infused, and if they are so infused, would it be credible to suppose that Jones did wrong, say, by smoking? Fortunately, there is reason to be hopeful. We should not ignore the fact that even if Jones’s smoking is luck infused, the kind of responsibility-undermining luck that

120

is associated with lack of a contrastive explanation of Jones’s choices or actions in terms of prior reasons leaves unscathed Jones’s proximal control over his decision to smoke and over his subsequent execution of that decision. This opens a window of optimism for the view that what Jones does can indeed instantiate one or more of the primary deontic properties. This thought is developed in the next chapter.

121

8 Robust Modest R-Libertarianism and Deontic Anchors

THE PROBLEM AND THE OUTLINES OF A SOLUTION

Just as one might reasonably deny that an agent can be responsible for performing an action if its occurrence is a matter of luck, so one might reasonably deny that an agent can perform an action that is morally right, or wrong, or obligatory if its occurrence is a matter of luck. How, for example, can one be under an obligation to open the magical safe if there is an equal probability of its opening or not opening even upon dialing the right combination? It might be replied that, assuming the person can (physically) dial any combination, is relevantly situated, has the cooperation of nature, and is lucky, the person can open the safe. After all, there is a world accessible to the person in which she succeeds in opening the safe even on the first try. Perhaps, it might be added, it would frequently be difficult to discharge our moral obligations if the accomplishment of our actions were luck infused, but that is a different concern than the one of whether there could be moral obligations in the first place. However, this sort of response is dubious. If moral rightness, wrongness, and obligatoriness require control, and if the safe’s opening is not even under one’s proximal control, then it is really not “up to” one at all whether the safe opens. Similarly, if the outcome of one’s intention to perform an action right away is thwarted, then even if one succeeds in acting as one intended, the outcome seems not to be “up to one” at all. Once again, the outcome’s occurring or failing to occur, whatever the case might be, is not under one’s proximal control; hence, it is sensible to suppose that it lacks a primary deontic property. But it is then an easy step to generalize and insist that robust indeterminism per se is inhospitable to deontic anchors. For, it might be charged,

122

if some robust indeterministic theory is true, then our actions are a matter of luck, and so they cannot instantiate primary deontic properties. However, once the problem of whether indeterminism is compatible with deontic morality is laid out in this fashion, then a solution suggests itself. There are crude indeterministic theories that do imply that all our choices or actions are indeed strictly out of our charge because we lack proximal control over their occurrence. But robust indeterministic theories of the sort, for example, defended by Kane and discussed by Clarke, do not have this implication. According to these theories, when an agent performs an action freely, the agent has proximal control over the occurrence of that action even though the agent could have done otherwise consistent with the past and the laws remaining the same. In addition, I have suggested that whereas ascriptions of moral responsibility are agent focused – they disclose what the agent morally stands for in relation to a certain choice or action – those of deontic morality are not. These latter ascriptions are act focused; they register morally salient features of the actions themselves. If this is indeed so, if there is no “disclosing” element that is a crucial component of act-focused deontic appraisals, then as long as an agent can exert appropriate control over an act, consistent with the agent’s having genuinely open alternatives, it would seem that the act could instantiate one or more of the primary deontic properties. Further, as far as I can see, “dual proximal control” should suffice for the sort of control an agent is required to have in performing an action if that action is to instantiate a primary deontic property. Robust modest libertarian theories of the sort discussed in the last chapter, then, should be able to accommodate deontic morality. Indeed, they seem tailor-made to do so. First, they meet the condition that there is a requirement of alternative possibilities for deontic morality; and second, they entail that free actions are ones over which an agent has proximal control. This strategy to accommodate deontic morality, though, rests on the claim that whereas ascriptions of responsibility are agent focused, those of deontic morality are act focused. The principal aim of this chapter is to provide a defense for this claim. In outline, the defense is the following. I first motivate a curious asymmetry having to do with “agency presuppositions” of moral responsibility and deontic morality: It appears that the agency presuppositions of the former are stronger than those of the

123

latter. Roughly, the idea is that whereas one cannot be responsible for an action unless it is caused by actional elements like beliefs, desires, and values that are “truly one’s own” or “authentic” – they aren’t, for example, implanted in one against one’s will by a nefarious being – one can perform an action that is right, wrong, or obligatory even though its causal antecedents are not truly one’s own. In what is to come, the precise nature of the asymmetry will, I hope, become more transparent. I will argue for the asymmetry via a somewhat circuitous route. First, I introduce a case involving what Kane has called “covert nonconstraining control” (CNC control). In such a case, an agent is controlled by another without being aware of being so controlled. The controller gets his way, not by constraining or coercing the relevant other against her will, but by manipulating her so that she willingly does what the controller desires (1996a: 65; and 1985: 33–51). Kane and others (see, for e.g., Watson 1987) have suggested that cases involving CNC control threaten compatibilist accounts of the kind of control required for moral responsibility. I will argue that if such cases are indeed a problem for compatibilist accounts of control, then they are equally a problem for incompatibilist accounts of control as well, again, when the sort of control in question is the sort required for moral responsibility. To establish one part of the asymmetry – the part having to do with the agency presuppositions of responsibility – I will then suggest that examples in which agents are covertly and nonconstrainingly controlled can profitably be regarded not as challenging conceptions of the kind of control required for responsibility, whether compatibilist or incompatibilist, but rather as motivating an “authenticity” condition of responsibility. This is, in broad strokes, the condition that one is responsible for a choice or action only if it issues from “authentic” beliefs, desires, values, and the like. To establish the other part of the asymmetry, the part having to do with the agency requirements of deontic morality, I will argue that an agent who is a victim of CNC control, though not morally responsible for her actions, may well do something that is right, wrong, or obligatory by those actions. Hence, though a covertly and nonconstrainingly controlled agent won’t be morally responsible for the pertinent action, this action of hers may, for example, be morally wrong. Finally, I will suggest that the view that ascriptions of moral responsibility are agent focused whereas those of deontic morality are act focused accounts for the asymmetry in agency presuppositions of moral responsibility and deontic morality. This gives us reason to endorse the view.

124

A CASE OF CNC MANIPULATION: PSYCHOHACKER

Compatibilists champion the view that determinism is compatible with free choice, free action, and moral responsibility. Incompatibilists, in contrast, deny that these freedoms and moral responsibility are compatible with the truth of determinism. Unlike compatibilists, they insist that if we are unable to choose or do otherwise, consistent with the past and the natural laws remaining as they were prior to choice or action, then we are not relevantly free. Further, they hold that as such freedom is required for moral responsibility, we are not morally responsible for anything we do if determinism is true. On these central issues that divide compatibilists and incompatibilists, many ingenious philosophical thought experiments have been designed by advocates of each side either in attempts to undermine the position of the other or to garner support for their own. One sort of case, generally viewed as yielding an incompatibilist moral, involves CNC control. Reflect on the following example that illustrates such control. Imagine Jenny’s being turned without her consent or knowledge into a psychological twin of the notorious computer hacker, Jim, by Max, the skilled psychosurgeon. Jenny awakens from the surgery with profound changes which, from her own inner perspective, she cannot but accept. The psychosurgery endows her with a new set of values, goals, preferences, and the like, while “erasing” ones she formerly had. Assume that these implanted elements are unsheddable.1 Catching the morning news, she learns about the new computing system installed in the Chase Manhattan Bank, and after some diligent work, she manages to transfer from an account in that bank a large sum of money into her own holdings. “Success!” exclaims Max with glee. In this case (“Psychohacker”), it has seemed to many that postsurgery Jenny is not responsible for her initial hacking offense despite the fact that she does not act in relevant ignorance when she commits the offense, and she exerts proximal control over actional elements that lead to her decision to commit the offense.2 The verdict that Jenny is not responsible for her offense, it appears, is crucially associated with her global manipulation. It is open to a libertarian like Kane to offer the following elucidation. Unlike Jim, Jenny is not responsible as she is not free in the deep sense of “free” required for moral responsibility: She is not the original creator of her ends or purposes. Bearing this explanation in mind, it may be thought that incompatibilists have the upper hand over compatibilists. For, first, incompatibilists might claim that an example like Psychohacker involving global manipulation of an agent undermines any prima facie reasonable compatibilist account of

125

freedom or control. For example, it would undermine an influential “hierarchical” conception of control according to which an agent exercises control over an action if, broadly, there is a right fit between her secondorder volition (her desire as to which first-order desire should move her to action) and the first-order desire that does in fact move her to action. (See, for instance, Frankfurt 1971.) Surely this fit, the incompatibilist can insist, could be produced directly in one agent like Jenny by someone with the resources of Max.3 The hierarchical approach would rule that such a manipulated agent would have control over her choices or actions, as the relevant psychological elements stand in the appropriate “control fulfilling” fit, even though it is clear that the agent would simply be a marionette of her manipulator. In addition, the incompatibilist might charge that this “wayward source problem” generalizes easily. It would, for instance, undermine various “reasons-responsiveness” accounts of control, again because a person’s basic parameters of reasoning, just as easily as her individual desires, could be the product of direct surreptitious manipulation.4 Second, at least some incompatibilists might insist that cases like Psychohacker involving CNC control strongly motivate the view that indeterminacy at strategic junctures in the causal pathway of events leading to choice and action is required to ensure that an agent is an ultimate originator of her choices, actions, purposes, and ends. If one cannot be responsible for one’s choices unless one is an ultimate originator of them, then the indeterminist can claim that cases like Psychohacker not only challenge compatibilism but also encourage serious consideration of libertarian accounts of control or freedom like the one advanced by Kane. Have (the relevant) incompatibilists, though, drawn the right moral from cases in which agents are covertly and nonconstrainingly controlled? I think not. It is true that the wayward source problem generalizes easily, but it also generalizes alarmingly: If scenarios involving CNC manipulation threaten compatibilist accounts of control, the sort required for responsibility, they equally threaten incompatibilist accounts of control. I start the remainder of this chapter with the assumption that Psychohacker and other cases of CNC manipulation do challenge various compatibilist accounts of control. Given this assumption, I show how Psychohacker presents a problem for robust modest R-libertarianism. I then summarize Harry Frankfurt’s compatibilist strategy to evade the problem of CNC control and argue that it is unsuccessful. Though unsuccessful, I believe that its suggestion that a root problem of various forms of CNC manipulation is that they threaten considerations of agency is fundamentally correct. I develop this suggestion and offer a

126

principle, drawing on considerations of agency, that in part explains why Jenny is not morally responsible for her choices upon recovering from surgery. This explanation strongly recommends reassessing the assumption that Psychohacker and some other cases of CNC manipulation undermine various conceptions of control, and it motivates an authenticity condition of responsibility. Last, I will turn to the agency presuppositions of deontic morality. ROBUST MODEST R-LIBERTARIANISM AND CNC MANIPULATION

The aim of this section is to show that if cases of CNC manipulation do indeed undermine various compatibilist accounts of control and responsibility, they also undermine robust modest R-libertarian theories. To set the stage for the argument, let’s recapitulate the main features of Modest Meleian Libertarianism. Assume that there is some compatibilist account of control (CC) that does not entail nor preclude the existence of indeterminacy in the causal sequence of events that culminates in fullblown intentional action. The sequence may involve these constituents: some psychological basis for evaluative reasoning including the agent’s beliefs, values, and desires; an evaluative judgment made on the basis of such reasoning that recommends a particular course of action; an intention formed on the basis of that judgment; and an action executing that intention. Suppose Jones intentionally does A, and his doing of A is indeterministically caused as a result of doxastic indeterminacy of the sort adverted to above (in Chapter 6) in the causal pathway of his A-ing. It is causally open whether certain nonoccurrent beliefs will come to his mind during his deliberations and, hence, whether he will form a decisive best judgment to A, and in fact A. Suppose, in addition, that the conditions specified by the compatibilist account of control, CC, are satisfied by his A-ing. These conditions ensure that Jones has considerable deterministic proximal agential control over the events leading from decisive better judgment through to overt action. The doxastic indeterminacy in the etiology of Jones’s A-ing need not erode such control. For once the beliefs that indeterministically come to his mind have entered into his deliberations, Jones exercises significant deterministic agential control (of the type specified by CC) over the events leading from the formation of a decisive better judgment to action. It is noteworthy that Mele’s account of the sufficient conditions for incompatibilist (or indeterminist) free action includes a condition that

127

precludes the sort of manipulating that Max does. Mele employs the locution “compulsion∗ ” to refer to compulsion not arranged by an agent (Mele 1995: 166). The condition that excludes global and other varieties of covert and nonconstraining manipulation says that the agent (like Jim) has no compelled∗ motivational states, nor any coercively produced motivational states. (See Mele 1995: condition 1 on 187, and 251–2.) In stark contrast, in his 1985 book and recently again (1996a: 142–3), Kane says that his indeterminism blocks the need for any such condition. It appears, then, that Kane does not think that he needs a condition like Mele’s precluding one to handle cases like Psychohacker. The discussion below casts doubt on Kane’s position. Reserve the label “Modest R-Libertarianism” to refer to a non–actioncentered libertarian view analogous to Modest Meleian Libertarianism in every respect save that it fails to include Mele’s precluding condition. The reasoning that if cases like Psychohacker call into question compatibilist accounts of control, they call into question action-centered (robust) R-libertarianism as well, will advance in these stages: In the first, we will note that indeterminacy leading to an agent’s action that occurs relatively early between the agent’s deliberations and formation of a best judgment regarding what to do, as in Modest R-Libertarianism, neither augments nor diminishes the agent’s proximal control over what she does in comparison to the proximal control she would exert over that same sort of action with no indeterminacy in its etiology. This, in turn, will enable us to draw the following conclusion: Suppose a globally manipulated agent, Jill, does A in a determined world, and she is not morally responsible for doing A. Then another agent who is also a victim of analogous manipulation, and who performs an action type-identical to Jill’s that issues from actional elements type-identical to those from which Jill’s action issues, should also not be morally responsible for A-ing, even if indeterminacy of the sort postulated by Modest R-Libertarianism is present in the pathway that culminates in her doing A. For both victims of manipulation exert the same sort of control over their A-ing, and if one is not morally responsible for A-ing, the other should not be either. In other words, the first stage will establish that if cases like Psychohacker undermine compatibilist accounts of control, they should also undermine Modest R-Libertarian control. In the second stage, I will argue that more robust forms of libertarianism than Modest R-Libertarianism such as Kane’s, which position indeterminacy relatively late in an agent’s actional pathway, fare no better with respect to the problem of global manipulation than does Modest

128

R-Libertarianism. Since cases like Psychohacker, as the first stage shows, call into question Modest R-Libertarianism, they should also call into question robust Kanean libertarian accounts. Let’s begin by confirming that if Psychohacker casts doubt on compatibilist notions of control, it should also cast doubt on Modest R-Libertarian control. Suppose Max turns Jenny into Jonsey, a psychological twin of Jones. Now consider two scenarios. In the first, free of any manipulation, Jones intentionally does A on “his own.” His A-ing results partly from certain beliefs of type B1, B2, and B3. Assume that it was not causally determined at any point that any of these beliefs would come to his mind; his action A is, thus, causally undetermined. Assume, also, that although Jones intentionally did A, he was also trying to do A∗ ; he was trying to do two competing tasks at the same time. And assume that if he had done A∗, his A∗ -ing would have resulted partly from beliefs that indeterministically came to his mind; so his A∗ -ing would have been causally undetermined. In the second scenario, suppose Jonsey, who is postsurgery Jenny, just like Jones, also performs A, and her A-ing, too, is not causally determined. Suppose, further, that her A-ing issues from actional elements type-identical to those that give rise to Jones’s A-ing, including beliefs of types B1, B2, and B3. Suppose, finally, that although Jonsey intentionally did A, she, too, like Jones, was trying to do A∗, and that if she had done A∗, it would have been causally undetermined because of antecedently occurring doxastic indeterminacy. I propose that the presence of indeterminacy in the actional pathway of her A-ing does nothing to dislodge the judgment that she is a marionette of Max and is thus not responsible for A-ing. To defend this proposal, notice, first, that though indeterminacy of the sort envisaged by the Modest Libertarian does permit for an agent’s having more than one physically possible future and for its being true that the agent could have judged, intended, and acted otherwise than she did, the presence of such indeterminacy (as noted Chapter 6) does little to persuade us that the agent ensures that he has more than one physically possible future, etc. Which beliefs enter the agent’s deliberations is indeterminate, and as a result of such indeterminacy the forking paths future is guaranteed. But the agent contributes no more to control in the scenario with indeterminacy than he does in a fully determined world. On the Modest Libertarian strategy, the threat of indeterminacy to control decreases as the model of control more closely approximates the model of control specified by CC, a model that stresses deterministic proximal control. But then the compatibilist who advocates CC can rightfully urge

129

that indeterministic initiation of action does not augment agential control. Of course, the compatibilist, as Mele proposes, need not charge that indeterminism, strategically located within the process of deliberation, erodes or diminishes control (see Mele 1995: ch. 12, and Clarke 1995): In sum, it appears that introduction of doxastic indeterminacy neither augments nor diminishes control in comparison to the sort of control that an agent would exercise in the absence of such indeterminacy if she were to exercise CC control over what she did. Armed with this lesson, compare Jonsey’s scenario with a third in which Jill, another victim of Max who has been turned by the eccentric scientist into a psychological twin of Jones, does A in a fully deterministic world. Again, suppose Jill’s A-ing issues from actional elements type-identical to those from which Jonsey’s do, including beliefs of type B1, B2, and B3. (Suppose, too, that Jill was also engaged in two competing tasks; if she had not intentionally done A, she would intentionally have done A∗ instead and her A∗ -ing would have been causally determined.) In the deterministic world, though it is false that the coming to Jill’s mind of her B1-, B2-, and B3-type beliefs is something that is in her control ( just as it is false that the coming to Jonsey’s mind of the relevant beliefs in her indeterministic world is something that is in her control), it is not causally open that they will not come to her mind. Although Jill exerts compatibilist control (CC control) over her Aing, we assume that she is not morally responsible for A-ing as she is a victim of Max. We are, remember, working under the assumption that cases of global manipulation subvert compatibilist accounts of control designed specifically to explain how agents in fully deterministic worlds can exercise control over their actions. But Jonsey is a psychological twin of Jill and she, too, just like Jill, is a product of Max. Further, Jonsey exerts no more nor less control over her A-ing than does Jill when Jill A-s. It would seem, consequently, that if Jill is not responsible for A-ing, in virtue of lacking control over A-ing by falling victim to Max, nor should Jonsey be despite Jonsey’s A-ing not being causally determined.5 The rejoinder that it is the possibility of Jonsey’s having more than one physically possible future – the possibility of her choosing and doing other than A – as a result of the doxastic indeterminism internal to her that makes the relevant difference between her case and Jill’s is unpersuasive. For simply add a Frankfurt-style counterfactual intervener to the second scenario: The intervener, who is also Max, desires that Jonsey chooses to A and A-s. He keeps vigil over Jonsey and will cause her to choose to A if and only if Jonsey, as a result of deliberation, forms

130

a judgment to choose to do something other than A. Jonsey, though, judges (on her own) that she ought to A, and A-s. In this Frankfurttype modification of the second scenario, although Jonsey is an indeterministic initiator of her action, she could not have chosen other than A, nor done other than A. She would, in these respects, be no different than Jill. Hence, if Jill is not responsible for A-ing because she is a marionette of Max and lacks relevant alternative possibilities, nor should Jonsey be, for she, too, is a marionette of Max and lacks relevant alternative possibilities.6 The mere presence of doxastic indeterminacy, located relatively early in the pathway leading to action, then, is insufficient to shield agents from responsibility-undermining CNC manipulation. Now, for the second stage of the argument, revert to robust modest R-libertarianism that locates indeterminacy relatively late in actional pathways at the juncture between the consideration of reasons and the making of decisions. Our goal is to show that if cases like Psychohacker undermine modest R-libertarianism, they undermine robust modest R-libertarianism as well. Suppose Max turns Jenny into Jasmine, another psychological twin of Jones. Jasmine finds herself in a conflict analogous, down to minute details, to the situation in which Jones found himself: She is tempted to smoke and she is also tempted to refrain. Suppose she reasons just like Jones, exerts an indeterminate effort of will type-identical to that exerted by Jones, decides to smoke, and smokes. Why should the mere fact that her decision to smoke is undetermined, given that it was caused by an indeterminate effort of will, render her responsible for her choice to smoke when she is simply one more victim of Max? For if victims of Max like Jonsey are not responsible for their choices when indeterminacy in the actional pathway of events leading to choice occurs relatively early in the pathway, and they are not responsible because they are victims of Max, how can they be responsible when indeterminacy occurs further up in the pathway ending in choice when they are still victims? We can grant that in her conflict, Jasmine’s effort of will that culminates in her decision to smoke is indeterminate and her resulting choice undetermined. But presumably the mental actions that are an agent’s efforts of will must, in some nontrivial way, depend causally on the agent’s antecedent actional elements like her values, desires, and beliefs. But in Jasmine’s case, as in Jonsey’s, the relevant antecedent actional elements have been heteronomously acquired. How, then, can Jasmine be morally responsible for her choice to smoke if we grant that Jonsey is not morally responsible for her choice to do A?

131

It might be objected that (setting concerns of luck aside), in cases in which indeterminacy occurs later in the etiology of action, unlike those in which indeterminacy occurs earlier on, the agent has control over how she responds to the unsheddable beliefs and desires implanted into her. This difference, it might be proposed, may well affect our responsibility ascriptions in these two sorts of cases even when the agents in them have been subjected to CNC manipulation. The objection, though, is not cogent. First, the sort of control at issue is, presumably, proximal (and not ultimate control). Perhaps, for example, it might be submitted that, with indeterminacy further along the actional pathway, it is in the agent’s power to diminish or augment the strength of a particular (implanted) desire, or to decide whether or not to take a particular (implanted) belief into account in further deliberation. But if a libertarian can avail herself of this strategy to escape the worry of CNC manipulation, a compatibilist, too, should be able to avail herself of it. As we remarked earlier, assuming determinism is true, once a belief has come to mind, one can exert proximal agential control over it. The moral here is that libertarians cannot both exploit this strategy to escape the problem of CNC manipulation and yet charge that compatibilists fall prey to worries of such manipulation on the ground that the strategy is unavailable to them. Second, this sort of strategy is itself problematic. For one thing, the very responses of the relevant agent, in cases in which indeterminacy occurs later in the actional pathway, may themselves be affected by the manipulation. So, for instance, the pertinent responses might be partially dependent upon the values of the agent, but these, in turn, may have been electronically induced. If the responses are so tainted, it is unclear whether the agent would be responsible for her subsequent choices that are mediated by the tainted responses. For another thing, responsibility, it seems, is an essentially historical phenomenon insofar as whether a person is morally responsible for a choice depends crucially on how its causal springs have been acquired.7 In particular, if not acquired under one’s own steam – if not “truly one’s own” – then they may well subvert responsibility. It is, I believe, concerns about this historical dimension of responsibility that really fan the flames of CNC-type objections to accounts of responsibility. Further, these concerns arise just as poignantly in cases in which indeterminacy occurs at pertinent junctions – relatively early or relatively late – along actional pathways. Prior to sketching an account of the conditions under which one’s springs of action are authentic, it will be instructive to examine Harry

132

Frankfurt’s attempt to show that global manipulation is really no threat to a particular conception of compatibilist control. HIERARCHICAL CONTROL AND CNC MANIPULATION

In a series of articles, Frankfurt defends the view that an agent is morally responsible for an action if, epistemic conditions of responsibility satisfied, there is conformity between his second-order volition (his preference as to which first-order desire should move him to action) and his will (the first-order desire that does in fact move him to action).8 In broad strokes, the thought is that an agent is morally responsible for his action insofar as there is an appropriate fit between what the person wills and what he wants to will. In later works, refining his theory, Frankfurt attempts to specify the way in which a person can “identify” with a particular desire. In one version, such identification requires that the agent form an unopposed higher-order desire to make some first-order desire his will and that he judge that no further consultation with other even higherorder desires would lead to a change in this endorsement. Of particular salience is Frankfurt’s contention that, to the extent that a person identifies himself with the springs of his actions, he takes responsibility for those actions and acquires moral responsibility for them; moreover, the questions of how the actions and his identifications with their springs are caused are irrelevant to the questions of whether he performs the actions freely or is morally responsible for performing them. (1988: 54)

In this and in other similar passages, Frankfurt makes it clear that as long as a person identifies wholeheartedly with his will, he should be held responsible for his action independent of the history of the identification. But then it seems that this account is vulnerable to the problem of CNC manipulation. As Frankfurt’s theory is a version of compatibilism, the existence of second-order mental states and processes of identification must be compatible with determinism.9 Since there is nothing mysterious about ordinary causal states and processes that are compatible with determinism, it is reasonable to assume that the theory presupposes a “naturalistic” account of consciousness and mental reflexivity, as opposed to some, say, Cartesian “nonnaturalistic” account. Consequently, if second-order volitions and identifications themselves do not involve anything more than ordinary causal states and processes, there is no reason why illegitimate causal influence – CNC manipulation, for instance – would not be possible on

133

the second-order level as well. In light of the presupposed naturalism, there is no reason why second-order reasons and reflexive identifications would have a privileged status by being immune to CNC manipulation. In his reply to this problem, Frankfurt distinguishes two cases in which a victim is manipulated by a “Devil/neurologist” (D/n) akin to Max. In the first, the D/n manipulates his subject on a continuous basis, like a marionette, so that each of the subject’s mental and physical states is the outcome of specific intervention on the part of the D/n (1988: 53). Regarding this case, Frankfurt says: [T]he subject is not a person at all. His history is utterly episodic and without inherent connectedness. Whatever identifiable themes it may reveal are not internally rooted; they cannot be understood as constituting or belonging to the subject’s own nature. Rather, they are provided gratuitously by an agency external to the subject. (1988: 53)

Frankfurt’s judgment that the victim “lacks all autonomy” and is hence presumably not morally responsible for his actions is, I believe, reasonable but his “history-insensitive” theory does not duly sanction this judgment. First, a relatively minor point: There is no reason whatsoever to believe that sophisticated CNC manipulation need lead to a history that is episodic and without connectedness; indeed, such manipulation may generate a life rich in “continuity and intelligibility [that is] essential to being a person” (1988: 53). The more potent worry is that, in the sense of ‘person’ presupposed by attributions of responsibility, Frankfurt’s own view is that a person is an agent with second-order volitions and the capacity for reflexive identification. (See, e.g., Frankfurt 1988: 16.) But then the D/n’s victim does qualify as an appropriate candidate for responsibility attributions, as Frankfurt himself allows that the victim’s “instantaneous states of mind may be as rich as those of a person; they may include second-order desires and volitions, or have even more complex structures” (1988: 53). The fact that identification is not “internally rooted” is irrelevant, given a history-insensitive view of the sort of agency presupposed by responsibility, to the status of an agent as a person. Frankfurt’s remarks that the victim’s having “no character or dispositions of his own” (emphasis added) undercut his own view that one “makes” one’s desires one’s own by identifying with them, on the supposition that identification is history-insensitive. In the second case, the D/n provides his subject with a stable character or program that, thereafter, determines the victim’s mental and physical responses to his external and internal environments without further

134

intervention on the part of the D/n. This sort of case is analogous to the one in which Max equips Jenny with the “stable character or program” of Jim. Frankfurt’s verdict is that there are no compelling reasons to deny that such an agent acts freely or is morally responsible for what he does. He says that the victim may become morally responsible, assuming that he is suitably programmed, in the same way others do: “by identifying himself with some of his own second-order desires, so that they are not merely desires that he happens to have or to find within himself, but desires that he adopts or puts himself behind” (1988: 53). But if pertinent identifications themselves are preprogrammed (and why couldn’t they be given a sophisticated enough programmer like Max?), this biting-thebullet response is hard to swallow.10 In addition, in the ensuing section I exploit Frankfurt’s insight, unveiled in his discussion of his first case, that examples such as Psychohacker that involve CNC manipulation can undermine responsibility by undermining agency, to support the view that the victim in Frankfurt’s second case is not morally responsible for his actions. NORMATIVE AGENCY AND CNC MANIPULATION

Drawing from work done elsewhere, my view, in outline, is that postsurgery Jenny is not responsible for her offense as this act is generated by an evaluative scheme that is not “authentic.” Beginning with the concept of an evaluative scheme, intentional deliberative action requires some psychological basis for evaluative reasoning. An agent’s deliberations that issue in a practical judgment about what to do, that in turn gives rise to a decision or intention, involve the assessment of reasons for or against action by appeal to the agent’s evaluative scheme. Such a scheme has the following constituents: (a) normative standards the agent believes (though not necessarily consciously) ought to be invoked in an assessment of reasons for action or beliefs about how the agent should go about making choices. The standards offer guidance within and across specific “domains.” So, for instance, Jenny might believe that financial decisions should be made on the counsel of her stockbroker, culinary choices in accordance with Julia Child’s recommendations, and moral choices on the basis of precepts of her religion. Norms need not be shared by agents. Unlike Jenny, James might believe that decisions across all domains should be made on the basis of some utilitarian moral principle. To be an apt candidate for moral responsibility, the normative principles must include a set of moral principles or norms; the agent must be minimally morally competent.

135

She must understand the concepts of rightness or obligatoriness and wrongness, and she must be able to appraise, morally, various choices or actions in light of the moral norms that are elements of her evaluative scheme. (There is no requirement that appraisals be fully considered, free of error, or even conscious.) (b) The agent’s long-term ends or goals he deems worthwhile or valuable. James, for example, may underscore his commitment to attempting to maximize overall happiness whenever he acts. (c) Deliberative principles the agent utilizes to arrive at practical judgments about what to do or how to act. For instance, James may believe that the best way to maximize utility is to rely on rules of thumb like “Keep your promises,” “Don’t cheat,” “Don’t steal,” etc. (d) Last, motivation to act on the basis of the normative standards in (a) and goals in (b) relying on the deliberative principles in (c). A full account of authenticity would include, among other things, a discussion of principles governing evaluative scheme authenticity (i) in global manipulation cases, and (ii) in cases involving agents like us who do not come into existence with evaluative schemes fully “in place,” but who acquire schemes gradually over time. Here, attention is confined to (i).11 To formulate principles that guide ascriptions of responsibility in cases like Psychohacker, I begin with the idea that to be an appropriate candidate for moral praise, blame, or responsibility in general, one must be a normative moral agent. I propose that it is a sufficient condition of an individual’s being such an agent at time t that the individual have at t (i) an evaluative scheme with the requisite moral elements – the agent is minimally morally competent; (ii) deliberative skills and capacities – for example, the agent has the capacity to apply the normative standards that are elements of its evaluative scheme to evaluating reasons; and (iii) executive capacities – the agent is able to act on (at least some of ) its intentions, decisions, or choices. An individual (like a baby) who fails to have deliberative or executive capacities will be able to exert much less control, if any, over its actions than an individual who does have such capacities. Construe condition (ii) as entailing that the agent is (at t) able to engage in genuine deliberation; his deliberative activities must meet the threshold of rationality below which such activities fail to count as bona fide deliberation. Normative agents have their evaluative schemes essentially: If E is the evaluative scheme of normative agent N, then having E is an essential property of N.12 One more concept requires elucidation before we can advance principles bearing on responsibility ascriptions in cases like Psychohacker. Let’s

136

say that an agent S broadly consents to pro-attitude (like a desire or value) PA’s being induced in S if and only if either S gives ordinary consent to PA’s being induced in S; or S would not object to PA’s being induced in S if, given S’s normal deliberative capacities, S were to reflect on PA’s being induced in S, under conditions S takes to be favorable or unobjectionable, to making decisions relevant to restructuring or reshaping S’s life. For example, suppose S freely consents to Max’s request to implant in S a cluster of pro-attitudes; then S has given broad consent. In another example, knowing that she is trying to reduce her intake of caffeine, suppose that by various persistent suggestions, Jake tactfully induces in Jasmine a desire to cut down on the number of colas she drinks in a day. Jasmine never gives overt free consent to the desire’s being thus induced in her. But assume that if she were thinking about modifying her long-standing habit of consuming large quantities of caffeine, under conditions she deemed unobjectionable to making decisions about such things, she would not object to the desire’s being induced in her in the manner in which Jake induces it in her. Again, Jasmine has given broad consent. Now we can introduce a principle about evaluative scheme unauthenticity germane to global manipulation cases: P1: If S is a normative moral agent N with evaluative scheme OriginalE at time t, and S acquires at t or after an evaluative scheme, NewE, via a process to which she did not give broad consent that either destroys N by destroying OriginalE (as in Psychohacker), or totally “represses” N, then NewE is not normative-wise authentic.13 Further, we add: P2: If an agent performs an action A, which arises from an evaluative scheme of hers that is not normative-wise authentic, then the agent is not morally responsible for performing A. The psychosurgery to which she did not give broad consent converts Jenny into a psychological twin of Jim, obliterating the property of being normative moral agent, say, Jenny∗ , and “replacing” that property with the one of being, say, normative moral agent Jim∗ . Hence, P1 and P2, together with relevant facts, imply that converted Jenny is not morally to blame for her offense. Principles P1 and P2 are meant to capture the suggestion that agency and authenticity of actional springs both affect ascriptions of responsibility. Some interim conclusions are in order. I have argued that the sort of CNC manipulation exemplified in Psychohacker – a case of global

137

manipulation – should just as surely be taken to undermine incompatibilist theories of control if they are taken to undermine compatibilist theories of control. Second, a comprehensive theory of responsibility involves a number of dimensions, including a control dimension (the agent must act freely or with control) and an epistemic one (roughly, the agent must not be relevantly ignorant of what he is doing). It has been customary to treat cases like Psychohacker as test cases for control. But this may be somewhat misguided. I recommend, instead, that they be used to cast doubt on accounts according to which the satisfaction of control and epistemic conditions is sufficient for responsibility. That these conditions are not sufficient, I believe, is one of the deep morals to be drawn from appropriate cases of CNC manipulation. If Jim has control of his actions, then as postsurgery Jenny is a psychological twin of Jim, she too must have control over her actions; they are, controlwise, on equal standing. This may be part of the reason why Frankfurt believes that cases of global manipulation do not undermine responsibility. This recommendation, if heeded, will help to refocus various debates concerning responsibility and control. For example, barring other difficulties with them, it would be misguided to charge that Frankfurt’s theory and Wolf ’s theory, say, are inadequate in virtue of their control dimensions falling prey to global manipulation cases; or that an advantage of indeterministic accounts of control is that they escape the problem of global manipulation. Third, considerations of agency are pivotal to ascriptions of responsibility – this isn’t news – but perhaps a lot more needs to be said about authentic selves as opposed to inauthentic ones, with an eye toward developing an adequate third dimension of responsibility.14 Now let’s turn to whether the actions of a CNC-manipulated agent like postsurgery Jenny are or can be morally right, or wrong, or obligatory. DEONTIC MORALITY AND CNC MANIPULATION

On one of the two possibilities of whether an agent like postsurgery Jenny (call her Jenny∗ ) can, for example, do moral wrong by stealing, the verdict is negative. Perhaps the strongest line of reasoning that sustains this verdict is that Jenny∗ ’s actions issue from actional springs – from elements of an evaluative scheme – that are not truly her own, and that it is only actions that issue from authentic springs that can instantiate one or more of the primary deontic properties of rightness, wrongness, or obligatoriness. This verdict would then lead to a sort of symmetry: Jenny∗ is not morally

138

responsible for stealing, as one is responsible for something only if it issues from an authentic evaluative scheme, and she does no wrong by stealing, as one’s action can be wrong (right, or obligatory) only if, again, it too issues from an authentic evaluative scheme. However, this line of reasoning is open to challenge. To begin with, an objector might first call attention to the fact that Jenny∗ intentionally stole in light of the belief that she was doing wrong. Next, it might be indicated that, presumably, any plausible normative ethical theory, like versions of utilitarianism or deontological theories, would rule that Jenny∗ did do wrong by stealing. The objector might then say that one shouldn’t conflate different sorts of moral appraisal: Whereas Jenny∗ did wrong by stealing, she is not ( perhaps) to blame for stealing. Jenny∗ s action, it might be urged, is relevantly analogous to that of a person who robs the strongbox in the throes of posthypnotic suggestion. Surely such a person, if she is totally in the dark about the hypnosis, is not morally responsible for the stealing, though she does wrong by stealing. The charge, then, would be that it is simply conflation of two different varieties of moral appraisal that leads to the verdict that a manipulated agent like Jenny∗ cannot do wrong. What, then, the advocate of the first possibility might ask, should be said about the crude android that successfully executes its program of robbing the strongbox? Here, there is no question about its being responsible for robbing, but does it do wrong by robbing? And if we deny that it does wrong, what is the relevant difference between its action and that of Jenny∗ ’s or the person who steals under posthypnotic suggestion? The germane difference couldn’t be that the android does not rob intentionally or that its “action” is not caused by beliefs or desires. For it might be proposed, we would not want to say of the tiger that kills the young child that its action lacks moral value or that it does wrong by killing, even though, arguably, its killing is an intentional action caused nondeviantly by appropriate beliefs and desires. There is, though, no compelling reason to deny that what the tiger does or what Jenny∗ does have moral disvalue. But there is, it seems, this salient difference between the tiger and Jenny∗ : The latter is a normative moral agent, whereas the former is not. Presumably the tiger is not, for example, minimally morally competent. Not having moral standards, it can’t apply these to evaluating reasons for action even if we assume that it does undertake crude evaluations of its reasons. One might plausibly propose that an entity or agent can perform actions that are right, wrong, or obligatory only if it is a normative moral agent, and one is a normative

139

moral agent only if one is minimally morally competent. One must, on this proposal, if one is to be a normative moral agent, have a basic grasp of the concepts of rightness, wrongness, and obligatoriness, and one must be able to use these concepts in appraisals of various things. Call this proposal the “moral competence model” of deontic moral agency. Although I am attracted to the moral competence model, I suspect that its requirements are too stringent: Things that fail to be minimally morally competent may, it seems, perform actions that can be right, wrong, or obligatory. Think, for instance, of hypothetical agents who are devoid of moral concepts, do perform a whole spectrum of intentional actions, and do undertake aesthetic or prudential evaluations of various things like policies, courses of action, paintings, architecture, and the like. Such beings would be normative agents of a particular variety: Though normatively competent, their normative competence would not include moral competence. These beings would have (i) an evaluative scheme with appropriate aesthetic and prudential elements but not moral elements, (ii) deliberative skills and capacities, and (iii) executive capacities. Intuitively, such beings may well perform actions that are morally wrong, although they wouldn’t know that these were morally wrong, or they wouldn’t perform them in light of the belief that they were doing moral wrong. With this sort of example in mind, one might propose, in contrast to the moral competence model, a normative competence model according to which an entity can perform actions that are right, wrong, or obligatory only if it is a bona fide normative agent of some sort; it must have an evaluative scheme that includes normative standards for evaluation, deliberative skills and capacities, and executive capacities. Such an entity would be able to undertake some sort of normative appraisal – one that needn’t be moral – of things like policies, actions, and so forth. Some might object that whereas this relatively weak model gets the case of the hypothetical beings right, one of its drawbacks is that it rules that the tiger’s act of killing the child is wrong. But this is controversial. It is debatable, for instance, whether tigers have anything like an evaluative scheme, even one that need not include moral standards of appraisal. I will refrain from deciding between the two models, though, of the two, I am drawn toward the weaker second. Either implies (together with relevant facts) that Jenny∗ does moral wrong by stealing. Let’s say that one is morally appraisable for an action if and only if one is either morally praiseworthy or morally blameworthy for its performance. I am inclined to believe that relatively stronger models than either of the two introduced will be implausible. So, for instance, a pretty obvious flaw of the condition

140

that an entity can do something that is morally right, wrong, or obligatory only if it can perform actions for which it is morally appraisable is that beings devoid of moral concepts but otherwise normatively sophisticated cannot be appraisable for what they do, though they may well do things that are right, wrong, or obligatory.15 My overall verdict is that there are fairly substantial grounds to resist the first possibility, the one that implies that Jenny∗ cannot do wrong by stealing. In the second possibility, a manipulated agent like Jenny∗ does wrong and so can do wrong by stealing. This option derives support from either the moral competence model or the weaker “normative competence” model. If this second option is correct – as I think it is – we are committed to a curious asymmetry: The kind of subversion of agency that we find in Psychohacker affects ascriptions of responsibility but leaves unblemished those of deontic morality. It should be emphasized that the concern here does not turn on considerations of control; assuming indeterminism, we can suppose that Jenny∗ has dual control over stealing – she has the ability to steal and the ability to refrain from stealing. Rather, what is at stake revolves around whether responsibility or deontic morality presuppose actions that issue from authentic actional springs. Responsibility for an action, it seems, does require that the action be caused by authentic actional elements – that is, by an evaluative scheme that is authentic with respect to norms – whereas one’s action’s being right, wrong, or obligatory does not require causation of that action by such elements. It is in this sense that the agency requirements of responsibility are stronger than those of deontic morality. This result prompts the obvious query: Why precisely is this so? The answer, I believe, is affiliated with the fact that responsibility, especially appraisability, tells us something about the moral worth of a person, whereas deontic morality – a person’s performing an action that is right, wrong, or obligatory – need not. In other words, the view that whereas responsibility ascriptions are agent focused but those of deontic morality are act focused accounts for the asymmetry in agency presuppositions of moral responsibility and deontic morality. Reconsider these aspects of the concepts of moral blame- and praiseworthiness: Moral blameworthiness for a deed is comprised of someone’s being deserving of being judged in a certain way, the judgment recording the moral worth of the person with respect to its performance. Blameworthiness for an action, in Zimmerman’s terminology, results in a “negative mark” being entered in the person’s “ledger of life” (Zimmerman 1988: 38). When blameworthy for an action, the person’s

141

moral worth is diminished by its performance. Analogously, praiseworthiness for an action results in a “positive mark” being entered in her ledger. When praiseworthy for an action, the person’s moral worth vis-`a-vis that action is augmented. Hence, if judgments of appraisability do reveal something about the moral worth of persons in relation to specific things they do, then it is no wonder – it is, indeed, sensible to expect – that it is only normative moral agents who can be morally appraisable and, more broadly, morally responsible for their conduct. In contrast, it appears that deontic evaluations are, or at least can be, much more act focused than agent focused. If by sheer accident and nonculpable ignorance, I trip the wire that sets off the explosive that brings down the house that Jack built, my moral worth with respect to tripping the wire has not been diminished, though I have, arguably, done wrong by tripping it. If alien amoral beings who are devoid of deontic concepts and who evaluate things either aesthetically or prudentially destroy Jack’s house because they believe it is an aesthetic atrocity, their moral worth with respect to the deed of destruction has not been tarnished, though again, arguably, this deed of theirs is morally wrong. ANTECEDENT CONTRASTIVE CONTROL, MORAL RESPONSIBILITY, AND DEONTIC ANCHORS

Let’s tie some ends together. In the last two chapters, I have attempted to show how it is that the objection from luck, raised against actioncentered robust forms of modest libertarianism (such as Kane’s), succeeds when moral responsibility is at issue but not when moral obligation is at issue. A key claim that I made is that such luck undermines not proximal control but authentic self-expression; whereas the former is necessary for both moral responsibility and moral obligation, the latter is necessary for moral responsibility only. In connection with Jones’s case (introduced in Chapter 7), I distinguished among (a) there is (some measure of ) an explanation of why Jones smokes, (b) there is (some measure of ) an explanation of why Jones∗ doesn’t smoke, and (c) there is (some measure of ) an explanation of why Jones smokes and Jones∗ doesn’t smoke.

I pointed out that the lack of a contrastive explanation (c) doesn’t imply that there are no contrastive explanations (a) and (b) and that (c) is lacking in the relevant context of indeterministic choice. There are two questions: Does the lack of (c) undermine moral responsibility? Does the lack of 142

(c) undermine moral obligation? I responded in the affirmative to the former but in the negative to the latter. But why the disparity? The answer, in a nutshell, is that lack of (c) indicates that neither Jones’s nor Jones∗ ’s choice adequately discloses what each agent stands for and that this undermines moral responsibility but not moral obligation. The lack of (c) does not indicate that neither Jones nor Jones∗ has the sort of (proximal) control over what he does that is needed for both moral responsibility and moral obligation, but it does indicate that the self-expressive requirement of moral responsibility cannot be satisfied. Some clarification will be helpful. I have proposed that there is an asymmetry between moral responsibility and moral obligation with respect to self-expression. Suppose this asymmetry is granted. One might still puzzle over why the lack of (c) should not be taken to indicate the absence of the sort of control necessary for both moral responsibility and moral obligation. Let me address the bearing of lack of an appropriate contrastive explanation on control and moral responsibility first. I submitted that authentic self-expression is part of the very concept of moral responsibility; when one is morally responsible for something, one discloses what one morally stands for in relation to that thing. But a necessary condition of authentic self-expression is that there be appropriate (c)-type contrastive explanations of the relevant behavior. One can’t, that is, disclose what one morally stands for with respect to a particular action without its being the case that there is a contrastive explanation, in terms of prior reasons, of why the agent performed that action rather than some other. The lack of an appropriate contrastive explanation, then, is indication of the lack of a sort of control required by moral responsibility (but not indication of lack of proximal control). The No Contrastive Control condition is designed to capture this lesson. Recall the condition: NCC: If it is the case that (a) if agent S does action A, S will have proximal control in doing A; (b) if S does B, S will have proximal control in doing B; (c) the “antecedent actional elements” that would causally give rise to S’s doing A if S were to do A are type-identical to those that would causally give rise to S’s doing B if S were to do B; and (d) it is false that S has proximal or other antecedent control over seeing to it that S does A rather than that S does B or vice versa, then (e) S is not morally responsible for doing either A or B if S does either A or B. It will be convenient to introduce this bit of terminology: If the conditions in clauses (a)–(d) of NCC are jointly satisfied when an agent 143

does either A or B, the agent lacks antecedent contrastive control in either A-ing or B-ing. In short, lack of a contrastive explanation (in relevant circumstances) implies lack of antecedent contrastive control. Without such control, one cannot authentically express oneself – one can’t disclose what one morally stands for in relation to an action. In turn, without such authentic self-expression, one cannot be morally responsible for one’s action. The argument can be streamlined in this way: (1) Self-expression is a requirement of moral responsibility. (2) Contrastive explanation is a requirement of self-expression. (3) Hence, lack of a germane contrastive explanation is sufficient for lack of responsibility. Addressing moral obligation now, the asymmetry between moral responsibility and moral obligation with respect to self-expression entails that whereas one can’t be morally responsible for an action unless one discloses what one morally stands for in performing that action, one can be morally obligated to perform an action even though it is false that, in performing it, one discloses what one morally stands for. Authentic selfexpression is not a constituent of the concept of moral obligation. We have seen that a necessary condition of authentic self-expression is that there be an appropriate (c)-type contrastive explanation of the relevant behavior. As there is no requirement of self-expression for moral obligation, there is little reason to believe that, put somewhat crudely, there is a requirement of contrastive explanation for moral obligation. In other words, one can be morally obligated to perform an action even though there is no contrastive explanation of why one performs that action rather than some other. Lack of a contrastive explanation for an agent’s action, then, does not indicate that the agent has no moral obligation to perform that action. Finally, as lack of a contrastive explanation for an agent’s action implies lack of antecedent contrastive control, and there is no requirement of a contrastive explanation for an agent’s action in order for that agent to have a moral obligation to perform that action, there is little reason to suppose that antecedent contrastive control is necessary for moral obligation. Proximal dual-control control appears to be what is needed – indeed, it appears to suffice – for moral obligation. Perhaps some might take issue with the view that there is as close a connection as I have proposed between lack of a contrastive explanation for an agent’s actions and lack of antecedent contrastive control on the part of that agent for that action; lack of the first, it might be suggested, does not imply lack of the second. Even if this is correct, the argument for the relevant asymmetry can be reformulated as follows. Authentic selfexpression is a requirement for moral responsibility, but lack of antecedent

144

contrastive control is sufficient for its being the case that this requirement is not satisfied. It is Jones’s lack of such control over whether or not he smokes in a particular rerun that precludes satisfaction of the selfexpression condition for moral responsibility. For it is in virtue of lacking such control that Jones cannot manifest what he would have himself to be. The essence of the pertinent argument is, thus, this: (1) Self-expression is a requirement of moral responsibility. (2) Antecedent contrastive control is a requirement of self-expression. (3) Hence, lack of antecedent contrastive control is sufficient for lack of responsibility. The rationale for (2), let’s remind ourselves, can be put as follows: If an agent (like Jones) lacks antecedent contrastive control in A-ing, then there can be situations in which though the agent does A under certain conditions, he does something opposed, B, under type-identical conditions. In such situations, authentic self-expression is not possible. In addition, if (under specified conditions) the agent successfully exerted such control in A-ing, then under type-identical conditions, he would have done A as well. This gives us reason to believe that antecedent contrastive control is necessary for authentic self-expression. Authentic self-expression, however, is not required for deontic appraisals. Hence, lack of antecedent contrastive control will not undermine deontic morality in virtue of undermining such self-expression. One might reasonably rejoin that even if moral obligation (or, more generally, deontic morality) doesn’t require authentic self-expression, it could be that it still requires antecedent contrastive control. Strictly, this is correct. But there appears to be little reason to affirm that obligation does require this sort of control, in the absence of something that is itself a requirement of moral obligation, and that is analogous to authentic self-expression in that it is not possible without antecedent contrastive control. What this thing could be is something that eludes me. Surely, if Jones (correctly) judges that he ought morally not to smoke but then acts contrary to this judgment, he does moral wrong in smoking even though there is no contrastive explanation of why he smokes rather than refrains (and, hence, even though he lacks antecedent contrastive control over his smoking). As long as he exerts proximal control over smoking and he could have refrained from smoking, it need not be denied that he did wrong by smoking. Still, one problem persists. Let A be Jones’s smoking and B his not smoking. How can it be that A is morally wrong and B morally obligatory when Jones cannot control which of these it is that he does? The present worry concerns the question, not whether one is in control of

145

which obligations one has, but rather whether one is in control of discharging whichever obligations one has. My view is that proximal control in performing an action when one performs it is all the control one needs in order for that action to instantiate some primary deontic property; the opposing view is that antecedent contrastive control is required. I have a good deal of sympathy with the opposing view; indeed, I formerly endorsed it (see Haji 1999a). However, I’m not quite so sure any more that it is correct. In the end, apart from the fact that deontic moral appraisals are act focused, it is intuitive considerations of the following sort which impel me toward my revised view. Ponder these two scenarios: In the first, of Jones’s alternatives A and B, Jones is aware that B is what he does in all the best worlds then accessible to him, he can do either A or B, he can control which of these he does, and whichever he does, he will have proximal control in doing it. Suppose Jones is dogged about doing B and he does B. Both my revised view and the opposing view yield the result that, in this scenario, Jones does what is obligatory. The second scenario is just like the first save that Jones∗ cannot control whether he does A or B. Again, Jones∗ is headstrong about doing B and he does B. The revised view once more implies that Jones∗ does what is obligatory. But the opposed view implies that, in this second scenario, A and B lack deontic normative statuses altogether. I believe this latter result is problematic as the control Jones exerts in doing B in the first scenario is of the very same sort that Jones∗ exerts in doing B in the second scenario. Or reverting to Jenny’s case involving global manipulation, it seems that after surgery Jenny’s offense is morally wrong despite the fact that the manipulation severely subverts her agency; her criminal behavior does not flow from actional components that are her own. There is some sense in which it is true that manipulated Jenny no longer controls her actions although she exerts (or can exert) proximal control in performing them. If we are willing to grant that deontic appraisals can be appropriate even when agency is so gravely undermined, it seems we should also be willing to grant that such appraisals can be fitting in scenarios like the second just described. The foregoing considerations in favor of the revised view will undoubtedly not persuade everyone. If the opposing view does win the day, then we will be saddled with the disconcerting position that both determinism and modest libertarianism are incompatible with deontic anchors. I close this section by acknowledging that there is a puzzle about deontic control that I find deeply perplexing. Opponents of the revised view I tentatively favor claim that, when our choices are nondeterministically

146

caused, there is a question about how it is that I may be said to control freely the actual choice I make. How is it up to me that, on a particular occasion of choice, one among two or more causally possible choices was made? The underlying constraint on control being proposed is that one has deontic control in bringing about a decision or action only if, given that two or more choices were causally possible, it was up to one which choice was made (or which pathway was taken), whichever way things go. It is not sufficient that one has relevant (proximal) control in making the choice that was made. In the revised view, proximal control in making a choice is all the control that is required for that choice to have one or more of the primary deontic properties. If one sides with the opponents on this issue, then if one cannot control which of, say, two things that one does, then neither of these has a primary deontic property; but this is not so on the revised view. I have suggested that the revised view has the edge over its competitor as both it yields intuitively acceptable results and deontic appraisals are act focused. But this doesn’t seem to me to be the whole story here. It may be that the opponents’ view is correct. Still, even if it is correct, I am at a loss to see why precisely it is that a decision or action cannot have a primary deontic property if it was not “up to one” that one made that decision or performed that action. CONCLUSION

I started this chapter with the suggestion that actions can instantiate primary deontic properties only if we have proximal control in performing them and we could have done something else instead just prior to their performance. I then proposed that action-centered modest libertarian theories can accommodate deontic morality. The argument for this proposal exploited the fact that such theories imply that when an agent performs a free action, the agent could have done otherwise, and that the free action performed is under the proximal control of the agent.16 However, action-centered modest R-libertarian theories, as we have seen, run afoul of the problem of luck: I have argued that if some such theory is true, then our actions are the result of responsibility-subverting luck. Specifically, if some such theory is correct, then an agent fails to disclose what she morally stands for in relation to an allegedly free action that she performs. We then found ourselves burdened with the difficulty of explaining why concerns of luck do not also subvert deontic appraisals if our actions are nondeterministically caused in the manner entailed by action-centered modest libertarian theories. Our explanation appealed,

147

centrally, to the view that ascriptions of responsibility are agent focused while those of deontic morality are act focused. This view, in turn, has been defended by reflection on the agency presuppositions of deontic morality and moral responsibility. My overall stance on the dilemma concerning freedom and deontic morality can, finally, be summarized in the following way. The dilemma, recollect, is this: If determinism is true, then no one has control – “deontic-grounding” control – over one’s actions. Similarly, if indeterminism is true, then no one has such control over one’s actions. Either determinism or indeterminism is true. Therefore, no one has deonticgrounding control over one’s actions. But it is morally obligatory, right, or wrong for one to perform some action only if one has deontic-grounding control over it. Hence, no action is such that it is morally obligatory, right, or wrong for one. While I endorse the first horn of the dilemma – I agree that determinism subverts deontic anchors – I (with reservations) reject the second horn. An action-centered modest M-libertarian theory can accommodate deontic morality. Although I have circumvented the dilemma by evading the second horn, the first horn is deeply troubling. Part of the reason for concern is that, for all we now know, our world is a determined one or a “near determined” one. The standard interpretation of quantum physics is compatible with indeterministic effects at the microphysical level “canceling out” at the macrophysical level involving control of our actions, with the result that our conduct is just as determined as it would be were the world fully deterministic. Semicompatibilists about determinism and responsibility have affirmed that determinism is no threat to responsibility despite determinism’s eliminating alternative possibilities as freedom to do otherwise is not required for moral responsibility. However, no such analogous position may be held regarding the relevance of determinism to questions of deontic morality. In particular, semicompatibilism is not a viable option with respect to deontic anchors. We cannot, that is, maintain the view that determinism is incompatible with its being the case that one could have done otherwise and yet compatible with its being the case that one’s act had one or more of the primary deontic properties. Still, one might wonder why it is of any consequence that no action can be right, or wrong, or obligatory in a determined world. What, if anything of significance, is at stake? It is this neglected issue that is the subject of discussion in the final part of this book.

148

Part Three Consequences of Being Deprived of Deontic Anchors

9 The Significance of the Possibility of Being without Deontic Anchors

TRANSITION AND SYNOPSIS

There are no deontic anchors in deterministic worlds. Some indeterministic worlds in which agents lack even proximal control over their decisions or actions will also be bereft of deontic anchors. Why should this matter? I explore this question in the present and in the subsequent four chapters. I begin in this chapter, by registering one cost of being without deontic anchors: A number of our reactive attitudes such as indignation, resentment, and forgiveness properly require that persons perform at least some actions that are wrong. A world devoid of deontic anchors, then, is a world devoid of the grounds for such attitudes. We value these attitudes or sentiments insofar as we believe that they have legitimate grounds. Hence, a world with these attitudes or sentiments but without rational grounds for them, would be a world deprived of something deemed morally valuable. In the next chapter (Chapter 10), I lay out an argument for the thesis that no one can be appraisable – that is, praise- or blameworthy – for any of one’s deeds in a world lacking in deontic anchors. The argument hinges on what I have dubbed the “Objective View” of blameworthiness and its praiseworthiness analogue: One is morally blameworthy for performing an action only if that action is morally wrong; and one is praiseworthy for performing an action only if that action is morally permissible or obligatory. Though many find it plausible, I believe that the Objective View is false. I consider proposed justifications for it and expose their deficiencies. I then cast doubt on this view by appealing to supererogatory and suberogatory actions. In Chapter 11, I examine the view that there are forms of virtue ethics that are possible in the absence of freedom to do otherwise. As aretaic

151

evaluations differ in kind from deontic ones, one might suppose that determinism does not threaten these forms of virtue ethics even though it undercuts deontic morality. I argue, however, that this supposition is mistaken. In addition, I introduce the notion that appraisals of deontic morality are autonomous. They are so in that they are distinct from other sorts of moral appraisal like aretaic ones; and their “place” in a full-fledged systematic moral theory that catalogues all genuine modes of moral appraisal such as aretaic, axiological, and deontic ones, and appropriate objects of such modes of appraisal like persons, acts, motives, and so on, cannot be “occupied” or “replaced” by other modes of moral appraisal. The autonomy of the deontic ensures that even if forms of virtue ethics are possible in a deterministic world, the world’s “grand ledger” of morality will be missing an essential constituent. A significant cost, then, of being without deontic anchors is that without such anchors deontic moral appraisals are not possible. In Chapter 12, I turn to the view that moral obligation is “overriding” in the sense that if moral obligation conflicts with other varieties of obligation such as legal or prudential obligation, all things considered, one ought to do what morality requires. But then, it might be claimed, with nothing morally obligatory in a world, as would be the case if that world were a determined one, there would be no final court of appeal to settle what one ought, all things considered, to do in cases in which varieties of obligation conflict. There is some truth to these claims, but they are also tainted with confusion. I aim to winnow truth from confusion. In addition, I introduce and ultimately reject two skeptical arguments that, if sound, would be of significance. One proceeds from the premise that the deontic “ought” is not overriding to the conclusion that no one is ever morally blameworthy for anything. The second proceeds from this same premise to the different, sweeping conclusion that there are no deontic anchors. Finally, in Chapter 13, I discuss what I take to be a further cost of being deprived of deontic anchors: In a world without such anchors, no moral agent is, in a way to be explained, a “source” of deontic morality. I conclude with some general remarks on the position (defended in Part II) that action-centered modest libertarian theories are hospitable to deontic anchors. Let’s start by examining two views – one advanced by Saul Smilansky and the other by Derk Pereboom – that attempt to convince us that lack of deontic anchors is not as momentous as one might initially envision. I believe that these views leave something to be desired. The discussion

152

will unveil one potentially significant cost of being deprived of deontic anchors: Lack of such anchors threatens the grounds for the very existence of various moral sentiments. SMILANSKY ON THE ETHICAL ADVANTAGES OF HARD DETERMINISM

In a recent paper, Smilansky explains that he takes hard determinism to be a combination of “non-libertarianism and incompatibilism” (1994: 356, n. 1). The hard determinist, he says, denies that there is libertarian free will, and he sees moral responsibility as incompatible with determinism. Even though hard determinism undermines moral responsibility – in this view determinism is incompatible with, for example, moral praise- and blameworthiness (1994: 356) – Smilansky proposes that hard determinism does enjoy some advantages. These are captured by the following theses advanced by Smilansky: Thesis1: Admirable moral action is still possible for hard determinists; and Thesis2: In “one sense the acceptance of hard determinism appears to create the possibility of a ‘purer’ morality than traditional free will based morality” (1994: 356).

To clarify and support these theses, Smilansky assumes, first, that “morality as such makes sense, i.e. that one need not repudiate all moral notions. If morality is impossible then obviously there are no ‘ethical advantages’ to hard determinism” (356). He next proposes that commonsense morality can be said to consist of two distinct parts: Firstly, certain things morally ought to be done (or not done). . . . This can be named the ‘substantive’ part. . . . Secondly, people often ought to get positive reactions such as praise when they do things that morally ought to be done (or not do what morally ought not to be done); and negative reactions such as blame when they do not do what morally ought to be done (or do what morally ought not to be done). This can be called the ‘accountancy’ part. . . . When a moral agent contemplates action, she normally has both aspects in mind. . . . A proper moral order consists not only of people behaving according to the substantive requirements of morality, but of their good or bad behavior being recognized for what it is. (357)

Smilansky contends that if hard determinism is true, the accountancy part of morality seems groundless. Hard determinism, he says, “severs the link between judging actions, say, good or bad, and being prepared

153

to praise or blame those who engage in them” (357). Thus, Smilansky believes that substantive moral requirements, if hard determinism is true, could not be seen as something that one could be blamed for not meeting. But he says that such requirements could be seen as “a specification of what, when deciding how to act, it would be best for us to seek to do” (358), and that this issue is a different and separate one from the accountability issue. The separation of the two issues – the substantive and accountability ones – Smilansky claims, presents us with the personal problem of what it is like “to be a person concerned with what ought to be done but not believing oneself to be praiseworthy (both for being concerned and for doing the right thing)” (358). Not directly engaging with this personal problem, Smilansky does address Theses 1 and 2: It seems . . . that an agent so construed [one concerned with what ought to be done but not believing oneself to be praiseworthy] is in one very curious sense a ‘purer’ moral agent than someone who is concerned with both parts of morality. For the hard determinist agent of the sort we are considering is not concerned with himself at all, he is solely focused on determining what he ought to do. By contrast, the ‘non-hard-determinist’ moral agent is inherently concerned with himself, with what action would make him praiseworthy or blameworthy in the eyes of others or of himself. . . . The hard determinist agent . . . enquires after the right thing to do and goes on to do it despite realizing that he is not praiseworthy for doing so and would not be blameworthy were he to act differently. He acts morally although he knows that this does not make him morally worthier than his neighbor, or even affect his moral worth in itself. If you like, he does not care about saving his moral soul, but about morality alone. (358–9, note omitted. See also Smilansky 2000: 234–41.)

Smilansky adds that, even if hard determinism is true, the substantive part of morality may vary in the sense that there is a “possibility of the existence of ‘morally pure’ hard determinist agents, both of consequentialist and nonconsequentialist persuasions” (1994: 360), despite disappearance of the accountancy part of morality. In sum, Smilansky urges that though determinism undermines moral responsibility, it has an ethical advantage in that it makes possible “morally pure” agents.1 Smilansky’s views, novel as they are, are not without difficulty. Start with his claim that if hard determinism is true, substantive moral requirements could be seen as a specification of what it would be best for us to seek to do. In one interpretation, this claim implies that there really are bona fide substantive moral requirements, moral obligations and prohibitions, in a deterministic world. But this, as we have seen, is false. In a second interpretation, we simply call what it would be best for us to seek

154

to do “moral requirements” in the knowledge that there really are no such requirements in a determined world. But this interpretation would undermine Thesis2 and at least one substantive construal of Thesis1. It would undermine Thesis2 as there would be no real moral requirements in a deterministic world and, consequently, hard determinism would not create the possibility of a purer morality than morality based on free will. It would undermine Thesis1, provided this thesis were taken to express a proposition that implied that actions in a deterministic world that were moral instantiated one or more of the deontic properties of rightness, wrongness, or obligatoriness.

PEREBOOM ON THE CONSEQUENCES OF DETERMINISM

Pereboom agrees with Smilansky (and many others) that a prominent feature of our ordinary conception of morality that would be undermined if determinism were true is that people deserve credit and praise when they willfully perform morally exemplary actions, and that they deserve blame when they willfully perform wrongful actions. He says that to deserve blame is to be morally liable to blame by deliberately choosing to do wrong. Determinism rules out one’s ever choosing to act wrongly, “for such choices are always produced by processes that are beyond one’s control” (1995: 32). However, Pereboom also argues for the view of direct interest to us that, though determinism undermines moral responsibility, it leaves intact “most moral principles and values” (22). I confine attention to three notable claims of Pereboom. First, Pereboom considers the suggestion that even if nobody is ever morally responsible for anything if determinism is true, it would still be best sometimes to hold people morally responsible. Such a view might be “justified on practical grounds . . . that thinking and acting as if people sometimes deserve blame is often necessary for effectively promoting moral reform and education” (32). But this option, Pereboom remarks, would leave the determinist thinking that someone is blameworthy when she also believes him not to be – an instance of theoretical irrationality – and would have her blaming someone when he does not deserve to be blamed – an instance of wrongdoing (32–3). Pereboom suggests that There is, however, an alternative practice for promoting moral reform and education which would suffer neither from irrationality nor apparent immorality. Instead of blaming people, the determinist might appeal to the practice of moral admonishment and encouragement. One might, for example, explain to

155

an offender that what he did was wrong, and then encourage him to refrain from performing similar actions in the future. . . . The hard determinist can maintain that by admonishing and encouraging a wrongdoer one might communicate a sense of what is right, and a respect for persons, and that these attitudes can lead to salutary change. . . . Likewise, although one could not justifiably think of one’s own wrongful actions as deserving of blame, one could legitimately regard them as wrongful, and thereby admonish oneself, and resolve to refrain from similar actions in the future. (33, note omitted. See also 2001: ch. 5, esp. 156–7)

The problem with this alternative, of course, is that it does suffer from irrationality. If determinism is true, nothing is right, wrong, or obligatory. Hence, the determinist cannot explain to the offender that what he did was legitimately wrong, and then discourage him to perform similar actions in the future. Nor can she, if determinism is true, communicate to others a sense of what is right or obligatory, for that matter. In sum, if the practice of moral admonishment and encouragement presupposes that actions can be right, wrong, or obligatory, a determinist cannot rightly engage in such a practice. One might, though, suggest that actions need only be intrinsically or perhaps overall good or bad for the practice of moral admonishment and encouragement to be “well-founded” and that actions can be intrinsically good or bad, or overall good or bad, even in deterministic worlds. But there are doubts with this suggestion. For one thing, if no action, and so no bad action, is morally wrong (if determinism is true), why should a person be so concerned or motivated to avoid its performance in the absence of things like institutionalized sanctions imposed to discourage people from performing acts of this sort? For another, assuming the possibility of deontic morality, it may well be that some acts that are overall or intrinsically bad are acts one morally ought to perform as these are the ones that one does in the deontically best worlds accessible to one as of the appropriate times. Further, there is no incoherence in the view that some acts that are overall or intrinsically good may well be wrong as none of these acts are the ones that one does in the deontic bests accessible to one as of appropriate times. Cognizant of these possibilities, one might have little motivation to avoid doing what is intrinsically or overall bad, or to fail to do what is intrinsically or overall good in deterministic worlds, again barring things like germane social or institutionalized sanctions or rewards. Second, Pereboom cautions that if we accept determinism, our belief that statements of the form: “although you did not choose x, you ought, morally, to have chosen x,” are sometimes true would be threatened,

156

as, he says, the central senses of ‘ought’ might well be undermined by determinism (1995: 36). He adds that if one can never choose otherwise than the way one actually does, given determinism, then it is false that one ever morally ought to choose otherwise. If it is never true that one ought to have chosen otherwise than the way in which one did, what then, it might be asked, would be the point of a system of moral ‘oughts’? (1995: 36). In response, Pereboom writes: [E]ven if moral ‘ought’ statements are never true, moral judgments, such as ‘it is morally right for A to do x,’ or ‘it is a morally good thing for A to do x,’ still can be. Thus, even if one is causally determined to refrain from giving to charity, and even if it is therefore false that one ought to give to charity, it still might be the right thing or a good thing to do. Cheating on one’s taxes might be a wrong or a bad thing to do, even if one’s act is causally determined, and hence, even if it is false that one ought not to do so. These alternative moral judgments would indeed lack the deontic implications they are typically assumed to have, but nevertheless, they can be retained when moral ‘ought’ statements are undermined. In addition, the various benefits of the system of moral ‘oughts’ can be recouped. For instance, when one is encouraging moral action, one can replace occurrences of ‘you ought to do x’ with ‘it would be right for you to do x,’ or with ‘it would be a good thing for you to do x.’ Discouragement of wrongful action could be revised analogously. (36, note omitted, and 2001: 143)

Again, I have reservations about these views. Determinism subverts deontic rightness and wrongness in addition to subverting deontic obligatoriness. Hence, in a deterministic world, judgments like ‘it is morally right for A to do x’ or ‘it is morally wrong for A to cheat on A’s taxes’ can’t be true. What of Pereboom’s proposal of recouping the various benefits of “the system of moral ‘oughts’ ”? When encouraging moral action, Pereboom suggests, one can replace prescriptions like ‘you ought to do x’ with ‘it would be right for you to do x’ or with ‘it would be a good thing for you to do x.’ Well, again, one couldn’t duly replace the ought prescription with ‘it would be right for you to do x’ as determinism subverts rightness. But wouldn’t replacement with ‘it would be a good thing for you to do x’ be appropriate? Here, circumspection is in order. On the one hand, if by ‘encouraging moral action,’ Pereboom intended ‘encouraging morally right action,’ then the suggested replacement won’t do as there is no necessary overlap between judgments of the right and judgments of the good. Moreover, in a world devoid of deontic anchors, it would be an error to encourage morally right action (or discourage morally wrong action). On the other hand, if by ‘encouraging moral action,’ Pereboom meant ‘encouraging intrinsically or overall good action,’

157

the replacement may well be appropriate. But I suspect that what would be “recouped” would not be a benefit “of the system of moral ‘oughts.’” For, as I have explained, even in a world with deontic anchors, actions that are intrinsically or overall good need not be ones that are morally right or obligatory, and those that are intrinsically or overall bad need not be morally wrong. In such a world, then, encouraging someone to do what is overall or intrinsically good, or discouraging one from doing what is overall or intrinsically bad need not be tantamount to encouraging one to do what is right (or obligatory) or discouraging one to do what is wrong, respectively. But then, too, there is little reason to suppose that in a world without deontic anchors, encouraging people to do what is overall or intrinsically good reaps the same sort of benefit that would be reaped by encouraging people to do what is right in a world with deontic anchors, and similarly with discouraging people to do what is overall or intrinsically bad. Perhaps we can express the point in this way: Presumably, among the “benefits” of a “system” of moral “oughts” is that there are legitimate moral standards – the true or correct set of basic moral principles that specifies necessary and sufficient conditions for rightness, wrongness, and obligatoriness – to which we can appeal to encourage people to perform acts deemed right or obligatory by those standards, and to discourage people to perform acts deemed wrong by them. In the absence of such a system, we may still be able to encourage people to do what is intrinsically or overall good, or discourage people from doing what is intrinsically or overall bad, without, in so doing, reaping the benefits of a system of moral “oughts,” as there is no neat overlap between the right (or obligatory) and the good, and the wrong and the bad. Pereboom considers the worry that if moral “ought” statements were never true, as would be so in a determined world, we would have no reason to do what is morally right (1995: 36). He claims that this view is mistaken as [O]ne might treat others with respect because one prefers to do so, or because one has resolved to do what is right, even if it is not the case that one ought to do so. If one has resolved to do what is right, by whatever motivation, one thereby has reason to act in accordance with this motivation. (36)

But again there is a problem. One cannot rationally, in a determined world, resolve to do what is right as nothing is right in such a world. A similar worry infects some of Pereboom’s views concerning “moral worth.” He says:

158

Although hard incompatibilism denies that human beings can have moral worth in the sense that they are morally praiseworthy for their actions, it need not renounce the notion of an agent’s moral worth altogether. . . . An agent can have moral worth by, for example, persistently doing what is right, even in situations in which there are strong countervailing pressures, by regulating her behavior by moral reasons, by having dispositions to examine her past behavior from the moral point of view, and by possessing a willingness to change her behavior when tendencies to immorality are recognized. (2001: 153)

However, in a world deprived of deontic anchors, no agent can persistently do what is right, nor possess a willingness to alter her behavior when tendencies to wrongdoing are recognized. Third, another cluster of issues that Pereboom discusses has to do with how it would be best to regard our reactive attitudes or feelings such as resentment, anger, gratitude, love, and forgiveness if determinism were true. These reactive attitudes are important, among other things, for interpersonal relationships. Some of them, such as forgiveness, seem to presuppose blameworthiness – we assume that the person forgiven for some deed is blameworthy for that deed. Pereboom claims that to the extent that the reactive attitudes do presuppose blameworthiness, they seem to be imperiled by determinism as determinism undermines blameworthiness. To the extent that our interpersonal relationships depend upon or involve the reactive attitudes, these relationships are also imperiled if these reactive attitudes are imperiled (1995: 40). However, Pereboom says that [T]here are certain features of forgiveness that are not threatened by hard determinism, and these features can adequately take the place this attitude usually has in relationships. Suppose your companion has wronged you in similar fashion a number of times, and you find yourself unhappy, angry, and resolved to loosen the ties of your relationship. Subsequently, however, he apologizes to you, which, consistent with hard determinism, signifies his recognition of the wrongness of his behavior, his wish that he had not wronged you, and his genuine commitment to improvement. As a result, you change your mind and decide to continue the relationship. In this case, the feature of forgiveness that is consistent with hard determinism is the willingness to cease to regard past wrongful behavior as a reason to weaken or dissolve one’s relationship. In another type of case, you might, independently of the offender’s repentance, simply choose to disregard the wrong as a reason to alter the character of your relationship. This attitude is in no sense undermined by hard determinism. (40)

In addition to his claim that forgiveness presupposes that the person being forgiven deserves blame, in this passage Pereboom suggests that forgiveness also presupposes that the person being forgiven has done wrong,

159

and the person doing the forgiving is willing to cease to regard the wrong done to him as a reason to weaken or dissolve his relationship with the person to be forgiven. Similarly, Jeffrie Murphy claims that forgiveness “essentially involves an attempt to overcome resentment,” and that resentment – and thus forgiveness – is directed toward responsible wrongdoing (Murphy and Hampton 1988: 20). Murphy adds that “if forgiveness and resentment are to have an arena, it must be where such wrongdoing remains intact – i.e., neither excused nor justified” (20). Jean Hampton also agrees that forgiving someone presupposes that the action to be forgiven was wrong (40, 54–5). Thus it seems that it would be an error, in connection with forgiveness, to suppose merely that the person being forgiven has done something that is bad – either, perhaps, intrinsically or overall bad – as it is possible that the person might have done something that is right or obligatory (assuming deontic anchors are possible) though intrinsically or overall bad. In like manner, Aristotle proposed that anger, when justified, essentially involves the correct judgment that one has been wronged and not just that something bad has befallen one (Rhetoric, Bk. II, 1378a30–2). “Essentially” here implies that without the presence of such a judgment, the psychological state in question is not anger. In a discussion on animal emotion, S. F. Sapontzis offers these necessary conditions for courageousness: S performs a courageous act only if S recognizes that the relevant situation in which S finds S’s self is potentially dangerous, and S intentionally risks harm to do what is right (1987: 19–22). Finally, in his probing and illuminating discussion of moral sentiments, Jay Wallace argues that the moral sentiments of resentment, indignation, and guilt are explained by beliefs about the violation of moral obligations construed as strict prohibitions or requirements, whereas other moral sentiments like anger or shame are explained by beliefs about the various “modalities” of moral value (1994: ch. 2). Wallace suggests that the members of “pure shame cultures,” if there are any such cultures, would lack the characteristic emotional resources of guilt, indignation, and resentment (38–9). If attitudes, or feelings, or states like forgiveness, indignation, resentment, and courageousness do presuppose correct deontic judgments – judgments to the effect that certain things are right, wrong, or obligatory – then as determinism undermines deontic anchors, it will also undermine the grounds for such attitudes, feelings, or states. Reflection on Pereboom’s and Smilansky’s discussion reveals the following general pattern of argumentation: Both are sympathetic to the

160

idea that determinism undermines blame- and praiseworthiness, and to the extent that various practices or attitudes presuppose correct judgments of blame- or praiseworthiness, determinism undermines these practices or attitudes. But it appears that both are insufficiently sensitive to the compelling proposal that many of these very practices or attitudes also presuppose correct deontic moral judgments; they presuppose, for example, that certain acts are indeed right, or wrong, or obligatory, and not just intrinsically or overall good or bad. But then as determinism undermines deontic anchors, it will also undermine the rational grounds for these very practices or attitudes. It is beyond the purview of this work to explore the fascinating arena of the importance of the reactive emotions such as resentment and indignation to our lives. I have to rest content with the conclusion that insofar as these emotions (again, assuming they are rationally grounded) are central, a world with no deontic anchors will be a world devoid of the benefits of these emotions. It’s time to examine a different sort of proposed threat of determinism: Determinism undermines responsibility – or at least moral praise- and blameworthiness – in virtue of undermining deontic anchors.

161

10 Determinism, Deontic Anchors, and Appraisability

Let’s remind ourselves of some terminology. To be morally appraisable for an action is to be deserving either of moral praise or blame for its performance. A common view is that determinism undermines appraisability. There are, though, different routes from the postulate that determinism is true to the conclusion that no one is appraisable for anything that one does. Three deserve special mention though I shall dwell at length on only one of them. The first begins with the premise that if determinism is true, then one lacks “genuine” (or libertarian) freedom to do otherwise. But “genuine” freedom to do otherwise, the second premise says, is necessary for appraisability. Hence, if these two premises are true, then determinism is incompatible with appraisability. As I have suggested in previous chapters, it is the second premise of this argument that many find problematic. A second argument starts with the premise that if determinism is true, then one cannot be an “ultimate originator” of one’s actions (as ultimate origination, requires, in part, freedom from control by the past, and one can have this sort of freedom only if there is some “indeterministic break” at appropriate junctures in actional pathways that culminate in action). But one is appraisable for an action, the argument continues, only if one is an ultimate originator of it. Hence, given these premises, it follows that determinism is incompatible with appraisability. Again, it is the second premise that is controversial. The third argument for the incompatibility of determinism and appraisability is predicated, like the first, on the assumption that if determinism is true, then we lack freedom to do otherwise. The subsequent essential premises are these:

162

(1) If no person can perform an action that is right, wrong, or obligatory, then no person can be appraisable for any of her actions. (2) No person can perform an action that is right, wrong, or obligatory (if determinism is true).

It is then concluded that determinism undermines appraisability. Premise (1) rests on what I have dubbed the “Objective View” of blameworthiness and its praiseworthiness analogue. The two principles that comprise this view are the following: Objective View: An agent is morally praiseworthy for performing an action only if that action is (overall and not just prima facie) morally obligatory or permissible (O1); and an agent is morally blameworthy for performing an action only if that action is (overall and not merely prima facie) morally wrong (O1). It may be useful to restate this view. We can distinguish objective moral obligation from subjective moral obligation in this way: Just as there is a difference between believing that something, p, is the case and its being true that p, so there is a difference between what one ought to do – one’s objective moral obligation – and what one believes one objectively ought to do – one’s subjective moral obligation. The Objective View can now be recast in this manner. One is praiseworthy for an act only if one had an objective (overall) moral obligation to perform it or it was objectively (overall) permissible for one to perform it; that is, praiseworthiness requires that one have done something (objectively) obligatory or permissible (O1). One is blameworthy for an act only if one had an objective (overall) obligation not to perform it; that is, blameworthiness requires that one have done something (objectively) wrong (O2). This view is to be distinguished from the view that each of blame- and praiseworthiness requires only that one have done something wrong, or right, or obligatory. The second premise, Premise (2), follows simply from one key result of the first part of this book, that there is a requirement of alternative possibilities for rightness, wrongness, and obligatoriness, together with the assumption that no one has freedom to do otherwise if determinism is true. If this argument is sound, then not only does determinism undermine appraisals of rightness, wrongness, and obligatoriness, but it also undermines appraisals of praise- and blameworthiness. Indeed, if the Objective View is true, then, necessarily, if determinism undermines

163

deontic anchors, it undermines appraisability as well. Once again, however, granting the assumption that determinism is incompatible with freedom to do otherwise, the argument has an Achilles’ heel: It is, I believe, the Objective View. In Part One of this book I gave preliminary reasons to doubt this view (see also Haji 1998a: ch. 8 and 1998b.). In this chapter I adopt a somewhat different approach: I critically examine various rationales for the Objective View and explain why they are unpersuasive. Then I develop one consideration (introduced in Chapter 4), having to do with the possibility of super- and suberogation, that tells against the Objective View. POSSIBLE JUSTIFICATIONS FOR THE OBJECTIVE VIEW

One proposed justification turns on the assumed impropriety of moral appraisals, either of the agent or behavior on the part of the agent, when the agent is an agent of a certain sort. Suppose, for example, that TJ, the neighbor’s retriever, saves baby Sid, who has crawled into the pool, from drowning. TJ, it might be claimed, is not morally appraisable for saving Sid as he has no grasp or understanding of moral concepts. In particular, he certainly does not act in light of the belief that he is doing anything right or obligatory by “retrieving” Sid. TJ just isn’t a normative moral agent. Equally, the judgment that it is obligatory or right for TJ to save Sid is also mistaken because such (legitimate) judgments, just like those of appraisability, presuppose that the agent in question is at least minimally morally competent. They presuppose that the agent must have some grasp of the concepts of rightness, wrongness, and obligatoriness and must be able to appraise (morally) various choices, courses of action, consequences of actions, and the like. It then seems reasonable, however, to hold that one cannot be blameworthy for something unless that thing is wrong, and one cannot be praiseworthy for something unless that thing is obligatory or right. This proposed rationale is defective, though. Let’s assume – although some have contested this (DeGrazia 1996, Dixon: personal correspondance, Johnson 1983, and Sapontzis 1987) – that TJ cannot be morally appraisable for saving Sid. Let’s assume, too (though, as I indicated in Chapter 8, this is controversial as well), that TJ’s saving Sid can’t be morally right, or wrong, or obligatory. Finally, let’s also grant the proposal that part of the support for these assumptions is that TJ is not minimally morally competent. Still, it certainly doesn’t follow from any of these things that the Objective View is true. All that follows is that correct judgments

164

of appraisability and judgments of deontic appraisals presuppose minimal moral competence on the part of the relevant agents. A second possible rationale for the Objective View starts with the suggestion that when no wrong has been done, the world really has not been made worst off.1 The world may be badly off in some respect – for example, a right might have been violated, a promise breached, an injury caused, a victim used as a mere means, or a prima facie duty not executed – but still, if one’s act were not in fact wrong, the world is not as badly off, it could be contended, as it would have been had one succeeded in doing wrong. The “objectivist” might add that even if one does something in light of the belief that one is doing wrong, and even if all other conditions of appraisability have been satisfied, there will be insufficient grounds for imputing blameworthiness if one does no wrong. An objectivist of this sort insists that deontic wrongness is not a function of belief in what is wrong or worst off and maintains that just as we should avoid conflating what is the case with belief in what is the case, so we should avoid conflating what is (objectively) wrong with belief in what is so wrong. A world devoid of objective wrongness, this variety of objectivist claims, is a world devoid of blameworthiness. Addressing, first, the suggestion that wrongness is not a function of belief in what is wrong and its corollary, that obligatoriness is not a function of belief in what is obligatory, although in the end I side with the objectivist in this metaethical concern, the suggestion has not gone unquestioned. So, for instance, H. A. Prichard held that of these two propositions: (A) that act which is best, and (B) that act which the agent believes to be best, the proper sense of ‘obligation’ is captured by (B) and not (A) (1949: ch. 2). But my primary worry with the stance of the objectivist of the sort under consideration is that even if Prichard were right, and wrongness is essentially a function of belief in what is wrong, surely, it seems, people could be appraisable for many of their deeds. As an illustration, suppose Al believed that, of all his alternatives, he was doing something that was both deontically worst and wrong by smoking on a particular occasion, and assume that he fulfilled all other conditions of appraisability. Then on the Prichard-like view that that act which the agent believes to be worst is wrong, Al intentionally, freely, and knowingly did wrong by smoking. Why is Al then not appraisable – specifically, blameworthy – for smoking, although, let’s assume, he acted in light of the false belief that his act was deontically worst? Zimmerman has insightfully proposed that even if what one ought to do is not what is best but what it is most reasonable to think (or believe) best,

165

still there is a distinction between doing what one ought (or ought not) to do (according to the Prichard-like account of obligation at issue) – that is, doing what is objectively wrong – and doing what one thinks (or believes) one ought (or ought not) to do or doing what is subjectively wrong. He says, “even if wrongdoing consists in not doing what is most reasonable to think best, still one must distinguish, on the one hand, doing wrong by not doing what is most reasonable in this respect and, on the other, doing what one thinks is wrong” (1988: 42). Suppose, sensitive to this distinction, a sophisticated objectivist claims that an agent does objective wrong when she does what she thinks (or believes) is worst or fails to do what she believes is best, and that she does subjective wrong when she does what she thinks (or believes) is wrong. Further, suppose that of Al’s only alternatives, A and B, Al believes A is worst but does not think A is wrong (as he doesn’t identify wrongness with being worst). He further believes that B is wrong, and also believes that B is best. (We can assume that he doesn’t identify rightness or obligatoriness with “bestness.”) The sophisticated objectivist would claim that if Al does A, he does objective wrong; and if he does B, he does subjective wrong. Suppose Al does B and all other conditions of appraisability have been met. Then the sophisticated objectivist, unlike the former sort of objectivist, can claim that though Al is not blameworthy for doing B, as he does no objective wrong in doing B, Al would have been blameworthy for doing A had he freely and intentionally done A (as his A-ing would have been objectively wrong). The sophisticated objectivist can, thus, circumvent the view that in a world devoid of objective wrongness of a certain sort – the sort that is not a function of belief in what is wrong – people cannot be appraisable for their conduct. However, the sophisticated objectivist’s position is plagued by its own worries. For one thing, it undercuts one of the key rationales for the Objective View with which we began: Acts for which one is blameworthy do in fact make the world worst off. For suppose B (in Al’s case) is not as good as A. Then A is best and B is worst (given that A and B are the only alternatives). Though the sophisticated objectivist would claim that (under appropriate conditions) Al would be blameworthy for A-ing (as Al would do objective wrong by A-ing), doing A would (in Al’s circumstances) not in fact make the world worst off. For another, suppose the sophisticated objectivist insists that by doing B, Al does not violate any moral requirement (as he simply does subjective and not objective wrong by B-ing), and that is why he is not to blame for B-ing. Then we have been given no justification for the Objective View; the question (against this view) has simply been begged.

166

The two proposed rationales that we have considered for the Objective View are, in summary, not cogent. Let’s now trace the implications of the possibility of super- and suberogation for the tenability of the Objective View. SUPEREROGATION, SUBEROGATION, AND THE OBJECTIVE VIEW

Commonsense morality presupposes, among other normative statuses, the supererogatory, roughly, permissible action that is beyond the call of duty; and the suberogatory, again, loosely, permissible action that falls short of decency.2 It also sanctions ascriptions of praise- and blameworthiness, respectively, to those who supererogate and to those who suberogate. In other words, commonsense morality condones the view that a person can be appraisable for super- or suberogating. If this is true, we can draw some revealing conclusions about the Objective View and certain features of the super- and the suberogatory. Although I cannot give full analyses of acts that fall into these two catagories, I will start by introducing fairly noncontroversial distinguishing characteristics of such acts. Building on considerations introduced in Chapter 4, I will then show that if commonsense ascriptions of being praise- or blameworthy for at least some super- or suberogatory actions (respectively) are correct, then the Objective View is mistaken. The discussion, in addition, will cast suspicion on the dictum that no action can be supererogatory unless its performance is praiseworthy, and that no action can be suberogatory unless its performance is blameworthy. The Supererogatory and the Suberogatory Stock illustrations of supererogatory behavior include the mailperson’s rescuing the child from the inferno at considerable risk to her own life, or the soldier’s sacrificing his life by throwing himself on the grenade to save his comrades.3 Many advocates of the supererogatory concede that the supererogatory is not confined to the saintly or heroic; unmomentous action, the doing of small favors, for example, can also qualify as going beyond duty. (See, e.g., Heyd 1982: 142). Despite controversy on just how the concept of the supererogatory is to be analyzed, there is a fair measure of agreement on these elements: A supererogatory action is morally optional in that it is not (overall) morally wrong or (overall) obligatory; it is not morally indifferent and so is morally significant (McNamara 1996:

167

420–1). Indeed, as Zimmerman remarks, it is especially valuable in some way, more so than any morally optional but nonsupererogatory alternative (1996: 234–5). Last, it goes beyond duty’s call. Regarding this third feature, Paul McNamara’s remarks are especially suggestive. He introduces the idea of doing the minimum morality demands: Suppose that in virtue of promising to get in touch with you, I become obligated to do so. Suppose . . . I can fulfill this obligation in two ways: by writing you a letter or by stopping by. . . . [M]y other obligations make me too busy permissibly to do both. Finally, suppose that, morally speaking, I put in a better performance if I pay you a visit rather that write you, even though either one is permissible. Then if I do the minimum morality demands, I will write rather than visit. . . . The important thing to note is that what is necessitated by meeting morality’s demands in a minimally acceptable way is not to be confused with doing what I am obligated to do. . . . [S]ome such notion is vital to the concept of supererogation. For if it is possible for me to discharge my obligations in a supererogatory way (in a better than minimal way), then it ought to be possible for me merely to discharge them in a minimal way – and vice versa. (1996: 425–6)

McNamara proposes that a supererogatory action is a permissible one whose performance is incompatible with doing the minimum (429). Suberogatory actions, the roughly symmetric counterparts of supererogatory ones, are morally optional, not morally indifferent, and are offensive or fall short of decency. In one of Julia Driver’s putative examples of a suberogatory act, a man who takes a seat on a train, thereby intentionally preventing two other people from sitting together despite availability of another seat, does no wrong but falls short of decency (1992: 286–7). Roderick Chisholm and Ernest Sosa propose that minor discourtesy (such as taking too long in a restaurant when others are waiting) may also be suberogatory (1966: 326). Assuming that there are minimal, better than minimal, and maximal ways of satisfying morality’s demands, something will be permissibly suboptimal, McNamara recommends, if it is permissible and precluded by doing the maximum (1996: 434). Whether permissible suboptimizing is equivalent to suberogation is something I won’t dwell on here save for the following. While suberogation implies permissible suboptimizing, it seems the reverse doesn’t hold. If paying a visit is better than writing a letter, and the latter is not wrong, then the latter is an instance of permissible suboptimizing. But it appears, we should not, for that reason alone, count it as an instance of suberogation, as what is not optimal may still be good; but ill-doing is bad.

168

It may be contested that there are preliminary reasons to be suspicious of the category of the suberogatory. For, first, one might claim that the etymology of the term suggests that a suberogatory act is one in which an agent meets morality’s demands in a less-than-maximal way. To meet morality’s demands in such a way, though, is to act impermissibly. However, if one takes seriously the possibility that there are minimal, better than minimal, and maximal ways of satisfying morality’s demands, then it certainly does not follow that when one does in a less-than-maximal way what is demanded of one, one acts impermissibly. If we can meet morality’s demands in a variety of acceptable, yet better and better ways, we can also meet these demands in a variety of acceptable, yet worse and worse ways, and we can do less than meeting morality’s demands in the best or near-best way without doing wrong. Second, it might be objected that the supererogatory seems capable of adequately handling several of the purported examples of suberogatory conduct. Morality may not require me to give a seat to traveling companions or those waiting for a restaurant table, but it would be ordinary kindness to do so and failure so to act is therefore omission of something near the boundary between the obligatory and the supererogatory. That is not enough to make the permissible act “indecent,” but it can explain our moral concern, distaste, or uneasiness, especially if we are unsure on just which side of the obligation/supererogation border performing these easy acts lies. But this objection is not persuasive either. It treads on a confusion between the appraisal of an agent and the appraisal of aspects of actions of that agent: Suppose we grant that giving up a seat (in the relevant circumstances) is “something near the boundary between the obligatory and the supererogatory”; indeed, let’s suppose that it is supererogatory. When one supererogates, one does more than the minimum required by morality, and an act can so exceed the minimum – it can instantiate the property of going beyond the minimum – even though when performed, its agent fails to act “out of ” kindness or excessive kindness. In a case of this sort, the act would still be morally commendable as it exceeds the minimum, though its agent may not. Similarly, when one suberogates, one does in a less-than-maximal way what is demanded by morality; harking back to McNamara’s case, one puts in a morally better performance by visiting than by writing, but one writes rather than visits even though one could have visited. Here, the act itself is morally amiss in that it “falls below” the maximum, even though when performed, its agent may or may not have failed to act out of kindness or generosity. The moral here is simply that if an agent fails to act “out of ” kindness, we may have an explanation of

169

why we think less of the agent, but this need have no bearing whatsoever on why the act itself, even if permissible, is morally amiss. The category of the suberogatory captures this moral failing of the permissible act. Finally, some might protest that while the class of the supererogatory was an acknowledged class in “prior moral thought or theorizing,” the class of the suberogatory was not. The import of this protest is unclear, however. A class of actions can exist even though not acknowledged in previous “moral thought or theorizing.” Indeed, Chisholm seems initially to have become aware of the suberogatory through its status as a symmetric logical counterpart to the supererogatory (1963: 5). Furthermore, the suberogatory or at least something very much like the suberogatory does have an ancient lineage. Prevalent in Islamic ethics is a fivefold model of the normative categorization of acts that classifies acts as required, recommended, permissible, discouraged, or forbidden.4 Recommended acts, akin to supererogatory ones, are permissible acts for the doing of which there is a reward – they are commendable – but for the neglect of which there is no punishment. The class of discouraged acts bears a striking resemblance to that of suberogatory ones: They are permissible and though there is no punishment for their performance, their omission is rewarded. Kevin Reinhart, emphasizing, among other things, the ageold roots of discouraged and recommended acts and their prevalence in everyday life, reports that “It is noteworthy that Ansari [1972] finds that the five-fold system is implied in texts which predate the formal development of the system. It is reasonably clear, in any case, from the terminology and grammatical forms used (passive participle) that most of the terminology of the five-fold system is extra-Quaranic” (1983: 201, n. 21). In summary, each of the preliminary objections regarding the suberogatory can, I believe, be parried. The stage is now set for tracking the implications of going beyond the call and permissive ill-doing for appraisability. Appraisability, the Supererogatory, and the Suberogatory Starting with the supererogatory, let’s flesh out the case of the heroic mailperson. A baby is trapped in the burning building. The fire has reached a dangerous stage, many parts of the building have collapsed, many others are in flames, and thick suffocating smoke is funneling out. A mailwoman Marla passes by. Sizing up the situation, she judges that there is a small, highly risky chance of saving the baby. Acting partly on moral springs of action, she charges into the building, makes her way to the top floor,

170

and finds the baby still alive. On the verge of passing out and badly burned, Marla drops the child from one of the windows to safety below. (See Feldman 1978: 48 and McNamara 1996: 420.) McNamara comments that we could “easily imagine the fire to be such that we would not consider any nonfirefighters to have been obligated to go in. Yet we can also imagine that, although her action was very risky, it was not irresponsibly foolhardy. . . . So we can assume that her action was permissible but not obligatory” (1996: 420). Feldman adds that we would admire and praise such a mailperson for her heroism. “We think [s]he deserves a medal or some special commendation, or a reward of some sort.” Her action is “considered to be especially meritorious from the moral point of view” (1978: 48). A first lesson now discloses itself. If the mailperson’s action is in fact supererogatory and hence morally optional, and she is indeed deserving of praise for rescuing the baby, then it is simply false that one is to praise for an action only if that action is obligatory. This does not, of course, undercut principle O2 of the Objective View, which says that praiseworthiness requires that one have done something obligatory or permissible. Still, the lesson is suggestive. I’ll revert to O2 shortly. Similarly, assume that even though he (mistakenly) believes he is doing wrong by preventing the lovers from sitting together despite availability of another seat, Grumpy fails to give up his seat. Here, Grumpy may well be blameworthy for not giving up his seat even though, let’s assume, his not giving up his seat is suberogatory and hence morally optional. But then Grumpy’s case challenges the popular principle O1 that one is to blame for an action only if it is morally wrong for one to perform that action.5 Perhaps, however, it might be rejoined that suberogatory acts violate imperfect obligations even if they violate no perfect obligations. But then something like the view that one is deserving of blame for an action only if that action is wrong can be defended because some kind of obligation (even if only an imperfect one) is violated whenever an agent is blameworthy. There are, though, worries with this rejoinder. First, it is controversial just what the distinction between perfect and imperfect duties or obligations (here I use ‘duty’ and ‘obligation’ interchangeably) is supposed to consist of, as there are competing accounts of the distinction; moreover, some of these accounts are problematic.6 Second, even the most reasonable of these accounts, it seems, will not be relevant to our understanding of the suberogatory. For a general shortcoming of the proposal that suberogatory acts involve the violation of some sort of duty, perhaps imperfect ones, is

171

that suberogatory acts, though especially disvaluable, are not wrong and so violate no obligation or duty. Any acceptable conceptual framework of the normative statuses, including those of the super- and the suberogatory presupposed by commonsense morality, should, I believe, represent suberogatory actions as ones that are morally significant, optional, especially disvaluable, but not wrong.7 Let’s now return to O2. The concern here is that whereas blameworthy suberogation would undercut the claim that blameworthiness requires that one have done something wrong, and, similarly, praiseworthy supererogation would undercut the claim that praiseworthiness requires that one have done something obligatory, praiseworthy supererogation (of the sort described above) would not seem to undercut the claim that praiseworthy supererogation requires that one have done something permissible. This is strictly correct. But in the absence of good reason to believe otherwise, if we were to grant that blameworthy suberogation undercuts O2 (of the Objective View) and that praiseworthy supererogation undercuts the condition that one is morally praiseworthy for an action only if that action is (overall) morally obligatory, I see little reason to retain the view that praiseworthiness requires that one have done something obligatory or permissible. Further, we have assumed that Marla supererogates in saving the child. Assume, now, that Marla acts in exactly the same fashion in which she does when she supererogates – she rescues the child from the hungry flames – but she does so in a deterministic world devoid of moral anchors. In particular, assume that the actual sequence of events in the actional pathway that culminates in Marla’s supererogating (when she saves the child) is type-identical to, or relevantly just like, the actual sequence of events in the actional pathway that culminates in Marla’s rescuing the child in the deterministic world. Given this type-identity or relevant similarity, if Marla is praiseworthy for supererogating in saving the child, it appears she should be praiseworthy, too, for saving the child in the deterministic world. But in this world, her actions instantiate no primary deontic properties. Finally, we have already uncovered strong motivation to reject O1. If one lacks alternative possibilities, as one would if one were in a Frankfurttype situation, one could, it seems, be praiseworthy for one’s action even though it would not be right, wrong, or obligatory. We have, then, initial reason to conclude that the super- and suberogatory are incompatible with the Objective View that, recall (with temporal indices added), says:

172

OBV1: Agent S is morally blameworthy [praiseworthy] at time t and thereafter for doing A at t only if it is morally wrong [obligatory or permissible] at t for S to do A at t. This lesson will not, of course, perturb some like Gregory Mellema. He claims that an act is supererogatory if and only if its omission is not morally blameworthy, and its performance both fulfills no moral duty or obligation and is praiseworthy (1991: 13). He also claims that an act is an offence (i.e., it is suberogatory) if and only if it is “not forbidden, it is nevertheless blameworthy, and its omission is not praiseworthy” (11). Although I later contest Mellema’s contention that no act can be supererogatory unless one is praiseworthy for it, and that no act can be suberogatory unless one is blameworthy for it, Mellema’s characterization of going beyond the call and of permissive ill-doing evidences his rejection of one part (O2) of the Objective View. Resistance to this line of argument leading to renunciation of the Objective View is to be expected from those skeptical of the possibility of suberogation even if that of supererogation is acknowledged. These people may well accept the blame component (O2) of the Objective View – that one is to blame for an action only if it is wrong for one to perform it. So, for example, while accepting the possibility of supererogation, David Heyd seems reluctant to grant the possibility of suberogation. Really, Heyd questions the view that there is a neat symmetry between supererogation and offence, but his discussion can, I believe, plausibly be construed as an implicit attack on the possibility of suberogation (1982: 128–9). Taking certain liberties, one argument suggested by Heyd’s remarks can be reconstructed in this way: Let A (for reductio) be any act that is suberogatory. (1) Forbearance from doing evil is always praiseworthy.8 (2) When one refrains from performing A, one forbears from doing evil. (3) If (1) and (2), then the omission of A is praiseworthy. (4) Therefore, the omission of A is praiseworthy. (5) No act is suberogatory unless it is false that its omission is praiseworthy. (6) Hence, it is false that A is suberogatory. There are flaws in this argument however.9 (1) is clearly problematic. One could forbear from doing evil as a result of weakness of will, or cowardice, or promise of great reward with no thought whatsoever to doing right for right’s sake, and then it would be far from evident that one would be deserving of praise for one’s forbearance. In addition, (5), I believe, ought also to be eschewed, as I shall argue below. Second, Heyd is unconvinced that heinous or villainous acts of suberogation are possible. The worry this time is that the more villainous an act,

173

the less inclined we are to suppose that it is morally optional (128). Heyd may be right about this, though in the absence of a normative theory, it is difficult to ascertain whether no villainous act would qualify as suberogatory. This aside, this second reason certainly falls short of establishing that no acts of suberogation are possible. Finally, Heyd raises concerns about whether there are any examples of trivial acts of offence (129). Again, this is inconclusive as, on this issue, it is, once again in the absence of a normative theory, difficult to side with Heyd or with those like, for example, Chisholm, Driver, and Sosa, who believe that various trivial acts are indeed acts of offence. I ask those still reluctant to relinquish Objective View OBV1 to refocus attention on an appropriate Frankfurt-type case, like the Mele and Robb case, in which it certainly seems that Bob is morally blameworthy for stealing Ann’s car although this act of his is not wrong; or on scenarios (like the one involving Lottie) of self-imposed impossibility. In Lottie’s case, prior to taking the disabling pill, Lottie ought to refrain from shooting the child, in part, because she can then refrain from shooting the child. But after taking the pill, and so no longer able to refrain from the shooting, it is false that it is wrong for her to shoot the child, though, again, it certainly appears that she is blameworthy for the shooting. The “objectivist” might have reservations about some of the temporal restrictions in OBV1 and propose a more “liberal” view, OBV2, as a replacement: OBV2: S is morally blameworthy [praiseworthy] at t and thereafter for doing A at t only if, as of some time t∗, it is morally wrong [obligatory or permissible] at t∗ for S to do A at t. OBV2 gets Lottie’s case right as there are times (times before Lottie has taken the pill) at which it is wrong for her to shoot the child when she does. Since Lottie has done (remote objective) wrong, the objectivist might reasonably say, she is still to blame for the shooting. However, it seems that objectivists who subscribe to OBV1 may well be reluctant to commit themselves to OBV2 as this example suggests. A patient is ill, and there are two main courses of treatment: He can be given either two successive doses of medicine A or two successive doses of B. Treatment with B would cure the patient but would produce unpleasant side effects, treatment with A would be optimal, and mixing doses would kill the patient.10 Assume that, before treatment begins, the doctor ought to give A on Sunday and then again on Monday. Hence, as of Friday, prior to when treatment is supposed to commence, it is wrong for the doctor to

174

give B on Sunday. Suppose, unbeknownst to any others, someone gives B to the patient on Saturday night. As of this time, it is no longer right to give A on Sunday as giving A would kill the patient; rather, the doctor ought, as of this time, to give B on Sunday. Assume that the doctor, deliberately intending not to do the best for the patient, and believing that it is wrong to do so, injects the patient with what she takes to be the first dose of B on Sunday. Would objectivists agree that the doctor is to blame for giving B on Sunday? Many, I suspect, would say that she is not, for she ought to give B on Sunday. But then such objectivists cannot endorse OBV2, for this condition (along with all others required for appraisability, we may assume) is satisfied, with the result that the doctor is blameworthy for giving B. Of course, objectivists already committed to OBV2 need not be worried by this result and may in fact welcome it. Still, there are problems with OBV2. Suppose there is a wonder drug C that is beneficial to all when taken regularly except to those suffering from some rare disease D. Giving C to a person with D kills that person. Rick visits his doctor regularly for his C shot. His doctor, relying on very credible evidence she has, believes that Rick has D. Wanting to murder Rick, she gives him his regular dose of C when Rick next visits her for some C (we assume that the doctor believes she is doing wrong by giving C on that occasion). Luckily for Rick, there has been an error in diagnosis. Contrary to what the doctor believes, Rick is suffering from a disease that can be cured only by taking C. Giving the lucky-for-Rick dose of C results in Rick’s full recovery. Making allowance for moderate consequentialist considerations, although it is arguably obligatory for the doctor to give Rick the lucky dose of C, it seems that the doctor is to blame for giving this dose. She acts just as she would have, with the same sort of base intentions and calculated deliberation, had her giving C resulted in Rick’s death. Although the doctor is blameworthy, the condition in OBV2 is not satisfied, for we can reasonably assume that there is no time at which it is wrong for the doctor to give Rick some C. Notice that Rick’s case and the “double dose” case are analogous in vital respects: In either, it is obligatory for the doctor to administer the relevant drug even though she (falsely) believes that she is doing wrong by giving the drug, and in either, it is (partly) by the occurrence of a lucky series of events over which the doctor has no control that an action that it would have been wrong for the doctor to perform (giving a certain drug to a patient) is no longer wrong but obligatory. These things suggest that the fact that there was a time at which it was wrong for the doctor to

175

give B to the patient on Sunday in the double-dose case is irrelevant to an account of why the doctor is blameworthy for giving B on Sunday. In sum, I recommend that appraisability be detached from objective obligatoriness or wrongness. Rejecting an Alleged Asymmetry between the Super- and the Suberogatory An important ramification of such detachment is worth discussing. Mellema, like Heyd, believes that there is a significant asymmetry between supererogation and offence. According to Mellema, the asymmetry traces to the view that blame- and praiseworthiness admit of degrees. Mellema proposes that the more blameworthy an act is, the “more likely that its performance is morally forbidden”; but the more praiseworthy it is, “it is not the least plausible to argue that its performance becomes an increasingly prime candidate for the . . . obligatory. . . . The realm of the obligatory is certainly not the final destination of acts whose performances become increasingly praiseworthy” (1991: 196–7). This relates to the alleged asymmetry between going beyond the call and permissive ill-doing in this fashion: On the condition that no act is super- or suberogatory unless its performance is, respectively, praise- or blameworthy, Mellema’s moral seems to be that the more praiseworthy an act is, the greater the extent to which it goes beyond duty, and further, the greater the extent to which an act goes beyond duty, its status as supererogatory is not threatened. In contrast, the more blameworthy an act is, the more likely it is that its performance is wrong (call this “Mellema’s Thesis”) and hence the greater the extent to which it falls short of decency; but the greater the extent to which an act falls short of permissive ill-doing, its status as suberogatory is thereby threatened (as its status of being optional is thereby threatened). Frankfurt-type examples, though, challenge Mellema’s Thesis. To see this, start with Mellema’s contention that it is possible to imagine a blameworthy act becoming increasingly blameworthy as the circumstances of its performance are altered. If it is blameworthy to tell you a lie, perhaps it is more blameworthy if I know that your believing the lie will cause you to miss an important meeting; more blameworthy yet if I know it will cause you to take the wrong medication for your heart ailment; or cause you financial ruin, and so forth. (195)

I don’t doubt that blameworthiness admits of degrees, but I reject the view that the more blameworthy an act is, the more it “tends toward the realm of the forbidden” (197) as “Frankfurtizing” Mellema’s series of

176

cases involving lying confirms. Assume that, in each, the counterfactual intervener ensures that you cannot refrain from lying. Then your lying is not wrong in any of them (nor is it, of course, right or obligatory) even though you may be more blameworthy for lying in some than in others. Blameworthiness, after all, can vary in degree along other parameters such as whether the person wills to do wrong for wrong’s sake (Zimmerman 1988: 47) or the seriousness or likelihood of harm, if any, that is caused. Even if the Objective View is rejected, as I have argued that it should, one needn’t deny that a reason somewhat like the one advanced by Mellema can be given for an asymmetry between super- and suberogation: Going further beyond duty doesn’t threaten supererogation, but falling further short of decency does threaten suberogation. Praiseless Supererogation and Blameless Suberogation Reflection on detaching appraisability from the Objective View (partially) motivates the following (admittedly controversial) proposal of mine: One can super- and suberogate without meriting, respectively, praise and blame for doing so. Starting with the former, consider yet another valiant mailperson, Mia. Suppose Mia is heavily indebted to a ruthless loan shark and is in desperate need of cash to avoid broken bones. On her route one morning, she passes by the billionaires’ home, which is ablaze. Seeing the neighbors restraining the frantic parents, she quickly sizes up the situation and discerns her chance. “If I pull this off,” she mutters to herself, “I’ll be handsomely rewarded, and if I fail – well, it will still probably be far better than pulverized ribs!” So off she dashes into the inferno and the baby is saved. It is important to stipulate that Mia fails to act “from” or “out of ” duty or any sense that she is doing right; she intends to rescue the baby solely for the purpose of saving her own skin. It seems that Mia’s and Marla’s act are of the same normative ilk, specifically, that if Marla’s is supererogatory, so is Mia’s. The “deontic normative status” of an act, surely, all other things held constant, save change in the agent’s purpose or goal, does not alter with such change in purpose or goal (though appraisals of appraisability could alter). Marla, for example, may well have saved because she believed (mistakenly) that it was her moral duty to do so even though she supererogated. Similarly, Mia, I suggest, just like Marla, supererogates by saving the child, but unlike Marla, she is not deserving of praise for this deed, it appears. To drive home this point, simply assume that if Mia does the minimum morality requires, she pulls the fire alarm, which is a block away, and directs the fire personnel to the blaze, and doing the

177

minimum (in her circumstances) is incompatible with rescuing the child. At least this much, I believe, is true: Mia’s act “goes beyond” objective obligation just as Marla’s does, but whereas Marla is praiseworthy, Mia is not. Regarding Mia’s deed, it should be emphasized that, besides being praiseworthy, other laudatory normative appraisals are still apt. So we need not, for instance, deny that she did something good, or manifested courage or strength of will, in saving the child. But such assessments are all compatible with Mia’s not being praiseworthy for the rescue. In another sort of case, imagine a person who craves for the excitement to be had from performing highly dangerous and heroic deeds but who fails to perform such acts from any sense of duty or any sense of doing right for right’s sake. Here, too, the person may supererogate without deserving praise for doing so. Last, picture a person who, having served in Vietnam, finds nothing extraordinary or heroic in habitually performing acts that are in fact highly risky and dangerous but not foolhardy. If this hardened person fails to act out of any sense of moral duty, or in light of the belief that he is doing right for right’s sake, then again it appears that though he may often supererogate, he is not deserving of praise for going beyond the call. Let’s next turn to cases in which it is plausible to judge that one suberogates without meriting blame. To develop an example introduced earlier, suppose, aware that others are waiting for a table in the popular caf e, ´ Joy lingers on; but she fails to believe that she is doing any wrong by staying on and is justified in so failing. Here, I’m inclined to judge that, granting that Joy suberogates, she does nothing that is blameworthy although one might well judge that she is inconsiderate. Or to alter the case a bit, suppose, engrossed in her newspaper, Joy simply fails to notice that others want her table. Just as one can do wrong in the absence of being aware that one is doing so, so I propose, one can suberogate without being aware that one is doing so. But then it seems perfectly possible that our lingerer, unaware that her table is in demand, is not to blame for staying even though she suberogates; she may not act in light of any belief that she is doing wrong. Finally, suppose that just on the verge of losing herself in the cartoon section, Joy notices others waiting for her seat and, acting partly from a sense of doing right, gets up and leaves. Then I believe she has refrained from suberogating and may well be praiseworthy for doing so. Now for some conclusions. First, praiseless supererogation and blameless suberogation are definite possibilities. At least this is the case if, as I have assumed, super- and suberogation belong squarely in the deontic camp. I admit that some will disagree with this assumption, supposing, instead, that super- and suberogation are hybrid notions and that their

178

special value or disvalue have to do with praise- and blameworthiness. Perhaps these dissenters are drawn to this latter assumption, in part, because they see no reasonable alternative to account for the special value of supererogation and the special disvalue of suberogation. I suggest that a sensible alternative is simply that the special value of supererogation, in those cases excluding trivial actions in which we think supererogation is especially valuable, is to be explained, at least in part, by the fact that when one supererogates, one does in a more-than-minimal or muchmore-than-minimal way what is good and morally demanded. And the special disvalue of suberogation, in those cases in which we think suberogation is especially disvaluable, is a function of the fact that when one suberogates, one does in a less-than-maximal or much-less-than-maximal way what is bad and morally demanded. Second, being blameworthy for an act that is not objectively wrong and praiseworthy for one that is not objectively obligatory or right are also definite possibilities. The Objective View is, thus, imperiled. One might worry that there is a strict limit to the relevance and power of my discussion of whether super- and suberogation do indeed undermine the Objective View. The concern is this: If, as I believe, superand suberogation are matters of act-evaluation that themselves carry no implication regarding whether the agent is to be praised or blamed, then (it seems) they are strictly deontic categories. Two points follow. First, the discussion of praise- and blameworthiness for super- and suberogatory actions, while casting doubt on the Objective View as formulated above (see OBV1 or OBV2), casts no doubt whatsoever on yet another variant of the Objective View, according to which an agent is praise- or blameworthy for an action only if that action has some deontic status. This being the case, an appeal to super- and suberogation cannot by itself show the independence of moral responsibility from deontic morality. Second, if determinism precludes the possibility of an action’s being morally obligatory, right, or wrong, it would also seem to preclude the possibility of an action’s being super- or suberogatory. The second point is, without any doubt, correct because both superand suberogatory actions, if there are any such actions, are morally optional (or right). This has no bearing on whether moral responsibility is independent of deontic morality, however. But what of the first point? Distinguish between deontic statuses that have, as a component or components, one or more of the primary deontic properties of rightness, wrongness, and obligatoriness, such as the super- and the suberogatory; and deontic statuses that lack, as constituents, any of the primary deontic properties,

179

such as the amoral. (Amoral actions are not morally right, or wrong, or obligatory. In the Frankfurt-type situation, Bob’s stealing Ann’s car, for example, is an amoral act.) Call the former deontic statuses “dependent deontic statuses” and call the latter “independent deontic statuses.” The discussion of super- and suberogation in connection with the Objective View, I think, at least calls into question the principle that praiseworthiness requires that one have done something (objectively) obligatory or permissible and that blameworthiness requires that one have done something (objectively) wrong. But if this is so, I can think of no cogent considerations that would substantiate the proposal that praise- and blameworthiness require that one have done something with some dependent deontic status. If we consider a case in which an agent is blameworthy for performing a suberogatory action, for instance, it is highly unlikely that the agent is deserving of blame (in part) because the action performed is wrong (as it is not) or permissible. Similarly, in a case in which the agent is praiseworthy for performing a supererogatory action, it is implausible that the agent is deserving of praise (in part) because the action performed is obligatory (for it is not) or permissible. I suspect that praiseworthiness in most such cases is closely associated with the agent’s going beyond duty’s call and has nothing to do with the pertinent act’s being permissible. At best, then, it appears that the discussion of appraisability for super- and suberogatory actions leaves untouched the view that praise- and blameworthiness require that one have performed an action with some independent deontic status. This concession, though, does little to undermine the key moral that the super- and the suberogatory contribute substantially toward establishing the independence of moral responsibility from deontic morality. Third, rejection of the Objective View defuses the argument of primary interest considered for the incompatibility of determinism and appraisability that invokes this view. Fourth and finally, suppose that, contrary to what I have argued, the Objective View is correct. Robust R-libertarians can exploit this fact, in conjunction with the view that determinism subverts deontic anchors, to respond to the following sort of concern. Robust action-centered versions of R-libertarianism, it has been objected, offer nothing better in the way of control than what our best compatibilist competitors offer (see, e.g., Haji 1998a: 28–9). Discussing a type of action-centered modest R-libertarian view according to which indeterminism is required in the direct causation of decisions themselves – a free decision is such that there remained, until the making of that decision, a genuine chance that the agent would not make that decision – Clarke presents a nice summary of this objection:

180

The gist of the objection is that the view offers nothing better than what is available with a compatibilist competitor – an account that resembles it in every way except in the requirement of indeterminism. Consider a decision that meets the requirements of the modest libertarian account. There was a chance of the agent’s not making that decision. But what is wanted from an account of free will is an account of an agent’s control over which decisions he makes and which actions he performs, and the mere chance of not making a certain decision may appear not to enhance the agent’s control at all. The factors that constitute the agent’s control in making the decision that we are considering, it may be said, are just those factors that are shared by the modest libertarian account and a compatibilist competitor: the agent’s having and being able to exercise a capacity for rational self-governance, his decision’s being caused by his having certain reasons, and so forth. A decision that meets the requirements of a modest libertarian view, it may be objected, is actually made with no greater control than is a comparable decision meeting the requirements of the competing compatibilist account. (2000: 22)

But then, one might continue, what is offered by robust modest libertarianism is of no more value than what is offered by a competing compatibilist view. A response to this objection is suggested by Mele (1999a, b), implicit in Kane’s recent work (Kane 1996a and 1999a), and articulated by Clarke (2000, in press). The response is that, though this sort of libertarianism does not increase the degree of control that agents have over their decisions or actions, it does enhance the value of the agent’s control. Hence, action-centered modest libertarianism does offer something better than rival compatibilist views, and in an overall tallying of these views, it is the former that carries the day. Setting aside this response, I offer the following instead (or, if one is convinced by the response, supplement it with the following): Robust action-centered R-libertarian freedom (really, robust modest actioncentered M-libertarian freedom) that involves indeterminacy in the production of our decisions and other actions is of value, at least partly, because there is a requirement of alternative possibilities for morally right, wrong, and obligatory acts, and this species of libertarianism accommodates this requirement. Further, whereas the Objective View, if correct, poses a direct threat to appraisability if determinism is true (for if determinism is true, there are no deontic anchors), it fails to pose such a threat to appraisability if an action-centered R-libertarian view is correct (though such a view will itself, as I have argued, threaten appraisability). The discussion in the last few sections of this chapter should bring into sharp relief the fact that whether a loss of deontic moral appraisals would also cast into jeopardy appraisals of praise- and blameworthiness

181

is not as straightforward as has been commonly assumed. We can affirm, with some measure of confidence, that if the Objective View is true, then there is an easy route to the conclusion that without deontic anchors, no one can be appraisable for anything. But it is far from obvious that this view is true; indeed, I think it is false. A correct assessment of the “no appraisability without deontic anchors” maxim, necessitates, I believe, getting clear on the conditions that need satisfaction for appraisability and then examining whether any of these conditions entails that appraisability requires deontic anchors. This is a tall order. For what it is worth, I have elsewhere defended the view that one is morally appraisable for performing an action just in case one performs it freely (one has what I call “volitional control” in performing it), autonomously (i.e., it issues from an authentic evaluative scheme), and in the belief that one is doing something morally obligatory, right, or wrong (Haji 1998a). I have argued that none of these conditions presupposes “objective” rightness, wrongness, or obligatoriness. For this reason, it is not evident to me that a world without deontic anchors is a world in which one can’t be appraisable for what one does. Still, there is something clearly unsettling about the view that people could be appraisable for some of their conduct in a world devoid of deontic anchors. I believe that one source of this uneasiness is associated with the epistemic dimension of appraisability. Suppose it is true that a person cannot be blameworthy for an action unless she performs it in light of the belief that she is doing wrong. (See Haji 1998a, b; and Zimmerman 1988, 1997.) In a world devoid of deontic anchors (the actual world might be such a world), people may well harbor beliefs to the effect that certain things are right, others wrong, and yet others obligatory; and they may well perform actions in light of the belief that they are doing something morally wrong, right, or obligatory. The epistemic dimension of appraisability may, consequently, be satisfied by different people on numerous occasions in this world. Still, their moral beliefs would be irrational in that they would all be false. Ascriptions of appraisability in this world would, hence, be soiled with irrationality. THE OBJECTIVE VIEW AND THE SIGNIFICANCE OF ALTERNATIVE POSSIBILITIES

I want to digress, slightly, to consider an implication, if indeed it is even that, of being deprived of deontic anchors. The Objective View assumes a central role in this discussion.

182

There is an esteemed tradition that associates freedom or responsibility with control. Aristotle, for instance, suggests that a person is morally responsible for performing an action only if he had appropriate control over that action, and that one has appropriate control over an action only if one could have done otherwise (Nicomachean Ethics 1109b30–5; 1113b1–14). In alignment with the Aristotelian tradition, Joel Feinberg (like numerous other contemporary thinkers) has proposed that a lack of alternative possibilities undermines control, and that if all our options save one are closed, and we exercise that option, we don’t do so voluntarily (Feinberg 1980: 36–9). In this view, it is one of our deeply held convictions – one assumed to be true – that the sort of freedom or control required for responsibility is freedom to do otherwise. Other theorists (like Kane), again in keeping with the sentiments of another distinguished tradition that champions the view that we are responsible for our actions only if we are the “ultimate creators” of the ends or purposes that these actions express, propose that one is morally responsible for one’s actions only on the condition that one is ultimately responsible for them; and one is ultimately responsible for one’s actions only if one is a causally undetermined source of at least some of these actions. To be a causally undetermined source of some of one’s actions, in turn, requires that, with respect to these actions, one could have done otherwise. These two traditions, then, inspire two different views about the relevance of alternative possibilities for freedom or responsibility. According to the first, the sort of freedom or control presupposed by responsibility entails possession of genuinely accessible alternatives, and according to the second, alternative possibilities are relevant insofar as they are required for “ultimate origination” of our actions. Jay Wallace has enunciated a third view. He suggests that the debate between compatibilists and incompatibilists over why alternative possibilities matter to responsibility in the first place is essentially a normative one about the conditions that render it appropriate to hold people morally responsible (1994: 85). [Gary Watson (1996: sec. 5) suggests a somewhat similar view.] Incompatibilists, Wallace claims, typically contend that it would be unfair to hold people responsible for their actions if they lack alternatives of the sort that are irreconcilable with determinism,11 whereas compatibilists deny this. I want to suggest that aspects of this normative interpretation of the dispute between compatibilists and incompatibilists over the significance of alternative possibilities are fundamentally mistaken. The errors surface upon appreciation of the fact that

183

a determined world is devoid of deontic anchors and that the Objective View is problematic. Building on some Strawsonian themes, Wallace contends that to uncover the conditions that require satisfaction for persons to be morally responsible for their actions, we need to focus on facts about responsibility that make them dependent in the right way on our practice of holding people responsible (1994: chs. 2–4).12 He offers the following schema for understanding the conditions that make a person responsible: (N): S is morally responsible (for action x) if and only if it would be appropriate to hold S morally responsible (for action x). (91, 155) Moral norms of fairness, Wallace contends, set the standards of appropriateness in terms of which (N) is to be interpreted (94–5). This, in turn, suggests this explication of schema (N): (NR): S is morally responsible for action x if and only if it would be fair to hold S morally responsible for x. Wallace construes incompatibilists as charging that “it would be unfair (and hence [morally] wrong) to hold people responsible if determinism is true” (1994: 110). He says that one line of argument that leads incompatibilists to this position is the following. The Argument from Fairness (1) People do not deserve to be blamed or sanctioned for what they do if they lack alternative possibilities for action.

(1) is supposed to be a “subordinate” principle of fairness (110). (2) Determinism deprives people universally of alternative possibilities for action.

(2) might, for instance, be supported by some version of the Consequence Argument. (3) If (1) and (2), it would be morally wrong to blame people for what they do (if determinism is true).

(3) is based on the assumption that if people don’t deserve to be blamed for what they do, it would be wrong to blame them for what they do. (4) Therefore, it would be morally wrong to blame people for what they do (if determinism is true).

184

Given the close conceptual tie that Wallace believes exists between being responsible or blameworthy and holding one responsible or blameworthy that (N) is meant to capture, an alternative appropriate rendition of premise (3) would be: (3∗ ) If people do not deserve to be blamed or sanctioned for what they do if they lack alternative possibilities for action, and determinism deprives people universally of alternative possibilities for action, then holding people responsible, if determinism is true, would be morally wrong.

The conclusion of the argument, accordingly, would be that (4∗ ) Holding people responsible, if determinism is true, would be morally wrong.

Wallace rejects line (1) of this argument. As a prelude to this rejection, recognizing (1) to be a moral principle that is open to assessment, Wallace asks how claims about the content of our moral principles are generally defended (110). One strategy is to ground such principles in a “higher-order” moral principle, such as the principle of utility, or in a higher-order procedure that tests candidate moral principles for their acceptability, such as Kant’s categorical imperative. Wallace says, though, that for his purposes, he will attempt to resolve the issue about the significance of alternative possibilities without appealing to “ultimate” principles or procedures and opt instead for what he believes is the only alternative: invoking a “local application of the method of reflective equilibrium” (112–13). The method assesses proposed subordinate moral principles like premise (1) on the grounds of their ability “to account for” considered convictions in which we have a high degree of confidence. “One must ask,” he says, “whether a given principle is needed to explain or express such considered convictions, or whether it more successfully explains and expresses our considered convictions than other principles that are available” (113, note omitted). He proposes that the most promising application of this approach in the incompatibilist’s cause would start with a set of our considered judgments about the circumstances that we acknowledge to excuse or exempt people from moral responsibility in particular cases. Elaborating, he writes: It is our considered moral view that such conditions as physical constraint, coercion, insanity, hypnotism, behavior control, and childhood render it unfair to hold people morally responsible. The incompatibilist might try to show that the truth of determinism would render it unfair to hold people responsible in just the way that these kinds of conditions are already accepted as making it unfair to hold people responsible, in the concrete circumstances of the moral life. (114)

185

Wallace proposes that incompatibilists frequently argue that conditions such as physical constraint, coercion, hypnotism, and so forth are responsibility-subverting ones in virtue of the fact that they deprive agents of alternative courses of action. It is unfair, incompatibilists might contend, to hold a person responsible for something that she was compelled to perform, or that she performed under nonconsensual hypnosis, precisely because compulsion, hypnosis, and the other conditions, like insanity for example, that are regarded as standard excuses or exemptions render a person unable to act otherwise. But if this is the reason why the excusing or exempting conditions subvert responsibility, then determinism, too, just like these conditions, should subvert responsibility universally, on the reasonable supposition that determinism entails that no one has freedom to do otherwise (114–15). Wallace next launches into a penetrating discussion of the excuses and exemptions. An outcome of this complex discussion is the conclusion Wallace draws that it isn’t the fact that a number (but not all) of the excusing or exemption conditions deprive people of alternative possibilities that explains why these conditions are responsibility-subverting. Hence, line (1) of the argument from fairness is false. Rather, there is an alternative basic principle of fairness, Wallace contends, that “accounts” for all our concrete judgments of excuse and exemption. It will prove revealing to look at this alternative principle, but prior to doing so, some evaluatory remarks are in order. According to Wallace’s normative interpretation of the dispute that divides compatibilists and incompatibilists over the importance of alternative possibilities for moral responsibility, incompatibilists, drawing on a certain principle of fairness (line (1) of the argument from fairness), are to be regarded as defending the position that holding people morally responsible if determinism is true would be morally wrong. This construal of the incompatibilist’s position, however, is fundamentally flawed. For if determinism is true, there are no deontic anchors, and hence nothing is morally wrong. There is a more charitable interpretation of the incompatibilist’s position. In this interpretation, the incompatibilist might start with the premise that deontic anchors require alternative possibilities and then acknowledge that a determined world is devoid of deontic anchors (for if determinism is true, no one has freedom to do otherwise). Now the incompatibilist might invoke the Objective View to secure the skeptical conclusion that no one is appraisable for anything that one does if determinism is true. The Objective View, again, says that one is morally blameworthy (praiseworthy) for performing an action only if it is morally

186

wrong (right or obligatory) for one to perform that action. But as no one does anything that is right, or wrong, or obligatory in a determined world, no one can be appraisable for anything that one does if determinism is true. A shortcoming of this argument, though, as we have already noted, is that the Objective View is questionable. Interestingly, Wallace’s own account of how excuses or exemptions function – why excusing or exempting conditions undermine responsibility – invokes a principle of fairness that closely resembles the part of the Objective View that concerns blameworthiness. Let’s summarize some of the relevant aspects of that account. I’ll limit discussion to Wallace’s views on excuses. We will then be in a position to draw some general conclusions about the overall tenability of the normative interpretation. Wallace’s own interesting position is that we can account for the full range of excuses by showing that they are all conditions in which an agent has not violated any moral obligation because of the absence of choice (135). Appreciating this position requires understanding some of Wallace’s views on moral obligation. (To facilitate later exposition, in the following summary of Wallace’s views on obligation, the relevant points to which I will refer are numbered.) (1W) According to Wallace, moral obligations are supported by reasons of the sort that may be expressed in the form of moral principles. For instance, one’s obligation to pick up a friend at the airport might be supported by appeal to the fact that one has promised to pick up the friend and the principle that breaking promises is wrong (129). (2W) If one genuinely accepts a specific moral obligation for purposes of practical deliberation or practical discourse, there will be some justification one could cite in support of the obligation (129). (3W) The moral stance of holding people to moral obligations involves a commitment to the possibility of justifying the obligations by appealing to moral principles. In justifying the obligations, one identifies reasons that could move both the judge and the agent held to the obligations to comply with the obligations in question (130). (3W) has implications, Wallace says, for the content of moral obligations. Specifically, (4W), it suggests that what one is obligated to do must be the sort of thing that is susceptible to being influenced directly by reasons (131). Apparently, Wallace is committed to the following: (WO): Something, x, is a moral obligation for agent, S, only if x can be influenced directly by S’s reasons; or, only if x can be influenced or controlled directly by the reasons S would give to “support” x.

187

Wallace notes that the reasons expressed in moral principles are practical reasons, or reasons for action. He says (5W): This suggests that such reasons should regulate what we do, the bodily movements that we make. But it is not bodily movements considered merely as such that are subject to the direct influence of moral reasons. Rather it is bodily movements insofar as they manifest a choice that is made by the agent. . . . [I]t is only through the mediation of our choices that the reasons expressed in moral principles may influence . . . the bodily movements we make. This means that [6W] one can be said to have complied with a moral obligation only when there is present a relevant quality of choice. Someone who inadvertently bumps into me, thereby knocking me out of harm’s way, has in no sense complied with the obligation of mutual aid. . . . Similarly, one cannot be said to have violated a moral obligation in the absence of a relevant quality of choice. (132, note omitted)

(7W) If a bodily movement is not intentional, it will generally not express any choice that the agent has made, and so won’t provide grounds for thinking that a moral obligation we hold the agent to has been violated (133). For Wallace, excuses, in general, function by showing that the agent did not violate the moral obligations we accept after all (133, 135). For example, suppose I harm someone – I tread on her hand – but not intentionally. I didn’t even know the hand was there; nor could I reasonably have been expected to know that it was there. As I did not tread on the person’s hand intentionally, Wallace says that my act will not constitute a case of harming someone as the result of a choice to bring about such harm (133). But, given (7W), one violates an obligation to do something only as a result of a choice to do that thing. Hence, I haven’t violated any obligation, say of nonmaleficence, by treading on the person’s hand. Furthermore, Wallace argues that it would be unfair, in such a case, to hold me responsible – say, blameworthy – for treading on the person’s hand. This is because it is surely the case that people do not deserve to be blamed if they have not done anything wrong in the first place. . . . [T]hose who have not in fact done anything wrong clearly do not deserve to be subjected to the reactive emotions and the forms of sanctioning treatment that express them. This is a fundamental principle of desert; it expresses an abstract moral conviction in which reflective moral judges have the highest degree of confidence. (135)

Wallace concludes that his account of excuses would not provide incompatibilists with a basis for initiating an argument, from the truth of determinism, to the disturbing conclusion that if determinism is true, then no one is ever morally responsible for anything that one does. For

188

“it is doubtful in the extreme that the truth of determinism would entail that people never act on choices that violate moral obligations we accept” (135). This, in turn, he thinks would deal a severe blow to the principle of alternate possibilities, as the principle, contrary to what some incompatibilists believe, is not necessary to account for our concrete judgments of excuse. Wallace’s rich account of the excuses deserves careful study. My comments, given the concerns of this section, will be restricted to addressing the following. First, a preliminary observation: Suppose we simply grant (1W), (2W), and (3W). (4W), specifically, (WO), it seems, needn’t follow from them. (2W) and (3W) center on the notion of one’s accepting something as a moral obligation, the suggestion being, roughly, that if agent S accepts x as a moral obligation, S has some reasons S can (or perhaps could) cite to support the view that x is a moral obligation for S, and these reasons motivate S to comply with the obligation. But the notion of S’s accepting x as a moral obligation appears to be distinct from that of x’s being an obligation for S. It seems possible, for example, that x is a moral obligation for S, but that S doesn’t know this, or that S wouldn’t accept x as a moral obligation for S. Suppose, for instance, that some version of utilitarianism is true. Suppose it enjoins that S ought to kill the one to save the nation; it enjoins that S’s killing the one (K1) is a moral obligation for S. But suppose that S does not have reasons S could cite to support K1; or that even if S did have such reasons, these reasons wouldn’t move S to comply with the obligation. Then it appears that though K1 is an obligation for S, S doesn’t accept K1 as an obligation. My principal concern with Wallace’s account of excuses is that it ultimately relies on the “no blameworthiness without fault” principle that says that one is not deserving of blame for something unless that thing is wrong. I have reservations, as I have explained, about this principle. But let’s suppose that it is true. This would spell trouble for Wallace’s view that his account of excuses leaves no basis for securing the skeptical conclusion that determinism undermines blameworthiness (135–6). As we have seen, Wallace doubts that the truth of determinism would entail that people never act on choices that violate moral obligations we accept. But a determined world is devoid of deontic anchors. If determinism is true, people would continue to do things that express their choices, but they would never act on choices that violate moral obligations. The “no blameworthiness without fault” principle would, consequently, encumber Wallace with the result, anathema to any compatibilist like Wallace, that no one is blameworthy for anything in a determined world.

189

We can draw some general conclusions about Wallace’s normative interpretation of the relevance of alternative possibilities. First, it would, I think, be uncharitable to portray the incompatibilist as invoking subordinate principles of fairness to secure the wrongheaded conclusion that holding people responsible for what they do if determinism is true would be unfair and hence morally wrong. Second, either the “no blameworthiness without fault” principle is true or it isn’t. Suppose, on the one hand, that it is true. Then an incompatibilist might well exploit this principle, in conjunction with the fact that determinism subverts deontic anchors, to argue for the view that no one can ever deserve blame for anything that one does in a determined world. Such an argument need not hinge on any considerations of fairness. Rather, it would emphasize that alternative possibilities are required for deontic anchors and that determinism undermines such anchors. In addition, if the principle is true, then Wallace’s own account of excuses provides the basis for the incompatibilist’s skeptical conclusion that Wallace so wants to avoid. Suppose, on the other hand, that the “no blameworthiness without fault” principle is false. Then Wallace’s account of why excusing (and exempting) conditions subvert responsibility fails. This account is part and parcel of Wallace’s normative interpretation of the importance of alternative possibilities. Regardless of the truth of the “no fault without blameworthiness” principle, then, the normative interpretation does not fare well. A final comment on some of Wallace’s views merits mention. Construing the debate between compatibilists and incompatibilists over the relevance of alternative possibilities as a normative one, Wallace, as we have seen, inquires into how subordinate principles of fairness, like the principle that one is not deserving of blame for something if one could not have done otherwise, are to be evaluated. After eschewing an appeal to “higher-order” principles such as utilitarianism (or higher-order procedures), Wallace claims that the only reasonable option is assessment by way of a “local application” of the method of reflective equilibrium. An application of this method, it appears, convinces Wallace to endorse the “no blameworthiness without fault” principle that plays a crucial role in his account of excuses. But I believe that the method of reflective equilibrium, or at least something very much like it, yields rather different results. First, contrary to Wallace, it seems to call into question the “no blameworthiness without fault” principle. Second, principle K, which implies that one can’t be under an obligation to do something unless one can do that thing, is, in salient ways, like the “no blameworthiness without

190

alternative possibilities” principle. Both say or imply something about control. The metaethical inquiry with which we commenced, an inquiry that, at various points, involved testing principles like K against considered convictions, testing considered convictions against theoretical considerations, and subsequently making various adjustments in either convictions or theory – in short, an inquiry that involved the process of reflective equilibrium – revealed that K, CK, KW, and OW are highly plausible. I believe that they are true. But these ethical principles, in conjunction with other facts or theses, generate the result that there is a requirement of alternative possibilities for moral rightness, wrongness, and obligatoriness. If this is so, we have a powerful basis for calling into question Wallace’s normative interpretation of the importance of alternative possibilities for freedom or moral responsibility.13 Reverting to the Objective View, despite what I (and others) have said against it, I am confident that it will continue to have its staunch supporters; the rift between advocates and opponents of this view will remain wide open. I’ll wind up this chapter by suggesting what may, in the end, explain this difference in outlook.

TWO FACES OF RESPONSIBILITY

I mentioned earlier that in an insightful paper entitled “Two Faces of Responsibility,” Watson proposes that responsibility has two different parts, faces, or perspectives. The first, recollect, he calls the “attributability or aretaic” face. This face is often wedded to a conception of responsibility that Watson dubs the “self-disclosure view.” He introduces this view by citing a passage from a work of Dewey. Let’s remind ourselves of this passage: [W]hen any result has been foreseen and adopted as foreseen [by the agent], such result is the outcome not of any external circumstances, not of mere desires and impulses, but of the agent’s conception of his own end. Now because the result thus flows from the agent’s own conception of an end, he feels himself responsible for it. . . . The result is simply an expression of himself; a manifestation of what he would have himself to be. Responsibility is thus one aspect of the identity of character and conduct. We are responsible for our conduct because that conduct is ourselves objectified in actions. (Watson 1996: 227)

Watson explains that, according to this view, actions which “express” ourselves in the required sense are free, whatever their further causes; and to be a responsible agent is to have what Dewey calls “moral capacity,” the

191

power to “put various ends before the self ” and “to be governed in action by the thought of some end to be reached.” This capacity entails “freedom from the appetites and desires”; it is “the power of self-government” (227–8). Watson continues: The self-disclosure view describes a core notion of responsibility that is central to ethical life and ethical appraisal. In virtue of the capacities identified by the self-disclosure view, conduct can be attributable or imputable to an individual as its agent and is open to appraisal that is therefore appraisal of the individual as an adopter of ends. Attributability in this sense is a kind of responsibility. In virtue of the capacities in question, the individual is an agent in a strong sense, an author of her conduct, and is in an important sense answerable for what she does. (229)

The second face of responsibility Watson calls the “accountability face.” This face is concerned primarily with the practice or practices of holding one another responsible or accountable. Watson proposes that, putting aside the self-reflexive case, [H]olding responsible is a three-term relationship in which one individual or group is held by another to certain expectations or demands or requirements. . . . ‘Holding responsible’ can be taken as equivalent to ‘holding accountable.’ But the notion of ‘holding’ here is not to be confused with the attitude of believing (as in, ‘I hold that she is responsible for x’). Holding people responsible involves a readiness to respond to them in certain ways. . . . To be “on the hook” in . . . [cases in which one is held responsible] is to be liable to certain reactions as a result of failing to do what is required. To require or demand certain behavior of an agent is to lay it down that unless the agent so behaves she will be liable to certain adverse or unwelcome treatment. . . . I shall call the diverse forms of adverse treatment “sanctions.” Holding accountable thus involves the idea of liability to sanctions. (235–7)

I think there is a good deal of truth in both these perspectives. It is hard to deny that there is such a practice as holding accountable. There is also considerable plausibility to the view that one is responsible for an action only if that action is attributable to one as its agent. Concerning appraisability, the view that I have defended – which aligns itself closely to the self-disclosure perspective – is, as I summarized earlier, the following: Blameworthiness for a deed is comprised of someone’s being deserving of being judged negatively in a certain way, the judgment recording a diminution in one’s moral worth as a result of its performance. I have stressed that a person of overall good moral stature can, on occasion, be blameworthy for an action, and when she is, her moral worth is diminished

192

as a result of its performance. I have also claimed that at the heart of the analysis of blameworthiness is this core: An agent is morally to blame on account of performing some action A only if that agent has a belief (even a dispositional one) that A is wrong, she has this belief irrespective of whether or not she cares about morality, she performs A despite the belief that it is morally wrong, and she believes at least dispositionally that she is performing A. Praiseworthiness is to be understood in a similar fashion: It primarily gauges the moral worth of an agent on the basis of whether she has done something that she believes is morally obligatory or permissible. Finally, I have noted that when a person is appraisable for an action, she “discloses” what she morally stands for, as, roughly, she acts in light of certain of her moral beliefs. I suggest that appraisability per se be distinguished from holding appraisable, or as Watson would say, the practice of holding accountable. For it certainly seems possible that one be appraisable, say, blameworthy, for a deed without its being the case that unless one forbears from performing that deed, one will be liable to certain adverse or unwelcome treatment on the part of others (or oneself ). Now perhaps advocates of the Objective View either focus far too heavily on the accountability perspective of responsibility at the expense of the self-disclosure perspective or simply take responsibility to be exhausted by the accountability perspective. If one does either of these things, one will be drawn strongly to the Objective View. In that view, one is blameworthy for an action only if it is wrong for one to perform that action. This is in keeping with the part of the accountability perspective that says that holding accountable is a triadic relationship in which one individual or group is held by another to certain demands or requirements, or at least to those requirements that the relevant others take to consist in correct or “objective” moral requirements. Ignoring one face of responsibility at the expense of the other or conflating the two is, I believe, a mistake. In this chapter, I have looked at various suggestions about what would follow for appraisability if there were no deontic anchors. I have argued that these suggestions leave something to be desired. In the next chapter, I discuss Michael Slote’s view, and those of others closely related to Slote’s, that forms of virtue ethics “are possible in the absence of free will” (Slote 1990: 369). If such forms of ethics are indeed possible “in the absence of free will” in worlds of certain sorts, then again one might not regard as significant the fact that these worlds are devoid of deontic anchors.

193

APPENDIX: A DIRECT CHALLENGE TO THE OBJECTIVE VIEW

Certain cases directly challenge the Objective View (OBV1). These are, again, cases in which though the agent does no wrong by doing something, intuitively, the agent is blameworthy for doing that thing. In this appendix, I discuss one such case. Suppose that, through no fault of anyone, Ernie is incorrectly diagnosed as suffering from disease D. Deadly, who wants Ernie dead, gets wind of the mistaken diagnosis and discovers that injecting medicine A into a person afflicted with D kills the person. Deadly, though, is squeamish about needles and under normal circumstances is psychologically incapable of thrusting a needle into anyone. It is only under the special circumstances in which he has the purpose or intention of killing his bitter foe, Ernie, that he can force the injection containing A into Ernie. Suppose Deadly executes his intention to inject Ernie with A in light of the belief that he is doing wrong. But it so happens that had he not thrust the needle into Ernie when he did, Ernie would have died. Unbeknownst to Deadly, and luckily for Ernie, Deadly’s deed actually prevents Ernie’s death. Under these circumstances, it appears that Deadly is blameworthy for this state of affairs: (P ∗ ): Deadly’s injecting Ernie with A for the purpose (or with the intention of ) killing Ernie. It seems, though, that permitting for moderate consequentialist considerations, P ∗ is obligatory for Deadly (under the circumstances), as without the relevant purpose or intention, Deadly could not have thrust the needle into the patient.14 Notice that P ∗ entails the following: (Q∗ ): Deadly’s injecting Ernie with A. But this deontic principle is true or eminently reasonable: DP: If S ought to do p and q is a logical or causal consequence of p, then S ought to do q.15 As Deadly ought to do P ∗, and as P ∗ entails Q∗ , he ought to do Q ∗ too. Perhaps it will be objected that Deadly is blameworthy for attempting to kill Ernie as attempting to kill Ernie is wrong, but he is not blameworthy for P ∗ or Q∗ as these states of affairs are, in the circumstances and unbeknownst to Deadly, obligatory for Deadly. But why think that attempting to kill (in such circumstances) is always wrong? Surely not because of

194

some such view as that, generally, attempting to kill an innocent person is wrong; hence, Deadly’s attempt to kill Ernie must also be wrong. Whereas it might generally be wrong to kill an innocent person, it doesn’t follow from this that it is always wrong to kill an innocent person. I think those who believe that attempting to kill in such circumstances is wrong rely on the principle that it is wrong to attempt to do wrong. There are different versions of this principle. The more plausible version says that if an agent’s doing something is (or would be) wrong and that agent attempts to do that thing, then the agent’s attempt to do that thing is also wrong. But this version is, to say the least, controversial. Suppose that it would be wrong for Hera to become an addict. But faced with pressure from her associates and tensions in her unhappy home, she attempts to become an addict; she acts on the counsel of her peers that addiction to the psychoactive drug will “solve all worries.” However, in her attempts, she comes to realize that becoming addicted would be a grave mistake. Her experiences give her the fortitude to seek alternative aid for her ailments, and they also prompt her to help other unfortunate youths from becoming addicts. Here, again allowing for modest consequentialist considerations, it is not clear that Hera’s attempt to become an addict is wrong for Hera. In other instances, an attempt to do something that is wrong may lead to the discovery that that thing is wrong. Suppose, you invite a colleague over to your lab to discuss some results over coffee. Unbeknownst to either of you, the sugar cubes have been laced with poison. As you attempt to spoon a cube into your colleague’s mug, the cube accidentally falls into a beaker containing a chemical. It reacts in a distinct fashion. As a chemist, you know that it would only react in that fashion if it were coated with the poison. Here, placing the lethal sugar into the coffee is (or would be) wrong for you, but your attempt to do so is not, since this attempt, among other things, leads to the discovery that it would be wrong for you to dissolve the cube in the coffee. In the first case Hera does not (at least not initially) believe that becoming an addict is wrong; and in the second case you do not (initially) believe that sweetening the coffee with the sugar is wrong. Zimmerman (1997: 236–7) has suggested that the general problem with the first version of the “attempt principle” is that if one lacks the belief that doing something is wrong and is not to be blamed for not believing that this thing is wrong, then it is implausible to suppose that one’s attempt to do that thing is wrong even if one’s doing it is or would be wrong. Others might protest that, in Deadly’s case, Deadly deliberately does what he believes is wrong. But it is always wrong, it might be proposed,

195

for a person to do what he or she thinks is wrong.16 This suggests an alternative version of the attempt principle: It’s always wrong to attempt to do what one believes is wrong. Hence, as Deadly does do wrong by injecting Ernie with A (given that he believes that he will do wrong by injecting Ernie with A, and the proposed principle that it is wrong to do what one believes is wrong), Deadly may well be blameworthy for injecting Ernie. But then Deadly’s case, it might be concluded, does not impugn the Objective View (OBV1). If this line of reasoning is to be commissioned successfully to defend the view that Deadly’s injecting Ernie with A is wrong, it must sanction from this premise: PM1: Deadly (mistakenly) believes that injecting Ernie with A is wrong, but he still injects Ernie with A, the inference: PM2: Deadly’s injecting Ernie with A is wrong. But beliefs can, of course, be mistaken, and when they are PM2 need not follow from PM1. If Deadly’s belief is false, we can still fault Deadly for what he does consistently with maintaining that his deed is not in fact wrong. Deadly may, for example, be condemnable, and as I have proposed, he may well be blameworthy. But these “agent-directed” evaluations should not be conflated with the “act-focused” one that Deadly’s injecting Ernie with A is not (in the relevant circumstances) wrong. Finally, suppose we grant that Deadly is blameworthy for attempting to kill Ernie. We can then reason that attempting to kill Ernie is not wrong for Deadly. For it seems that the following state of affairs: A∗ : Deadly attempts to kill Ernie (by injecting Ernie with A) is true in any world in which P ∗ (the proposition that Deadly’s injecting Ernie with A for the purpose, or with the intention of, killing Ernie) is true. Indeed, one might plausibly claim that Deadly’s attempt to kill Ernie just is his injecting Ernie with A for the purpose or with the intention of killing Ernie. So P ∗ entails A∗. As P ∗ is obligatory for Deadly, and P ∗ entails A∗, then (given the deontic principle DP), A∗ is obligatory for Deadly. This gives us sufficient reason to reject the second version of the attempt principle. Deadly attempts to do what he takes to be wrong (his killing Ernie by injecting Ernie with A), but it is not the case that his attempting to kill Ernie (by injecting Ernie with A) is wrong for Deadly.

196

11 Virtue Ethics without Metaphysical Freedom

Slote begins an interesting paper with the remark that the question of whether human beings have free will has long been considered one of the most important in philosophy for, among other things, the assumption that free will is necessary to moral responsibility and vital to ethics. He goes on to argue, however, that there are forms of ethics that are possible in the absence of metaphysical freedom (1990: 369). In this chapter, I will start by considering a version of virtue ethics developed by Slote that he believes is compatible with determinism. I will follow by questioning some aspects of the compatibility claim. Then I will consider a sketch of another version of a virtue ethical theory offered by Watson. This theory does seem compatible with determinism, but I argue that the concepts of rightness, wrongness, and obligatoriness supported by this theory are far removed from our ordinary deontic understanding of them. So even if this version of virtue ethics survives pressure from determinism, it won’t sustain the view that deontic anchors (as I have been conceptualizing them) are possible in deterministic worlds. Further, if there are significant costs to not having such anchors, as indeed there are, such a virtue ethics will not be able to recoup these losses. Finally, I will end by arguing for the “independence” of deontic moral appraisals: Such appraisals are distinct from various other modes of moral appraisal such as, for example, appraisals of praise- or blameworthiness, and axiological ones concerned with the instrumental, intrinsic, or overall goodness of states of affairs. My suggestion will be that even if forms of virtue ethics are compatible with an absence of deontic anchors, a world without such anchors is a world in which the full spectrum of moral appraisal is not possible.

197

AN OUTLINE OF A VERSION OF VIRTUE ETHICS

One of Slote’s principal aims in From Morality to Virtue is to offer “foundations for a general account of a specific form of virtue ethics” that is sufficiently developed “to enable us to compare its merits with those of currently dominant approaches to ethics” (1992: xiv). In this section, I summarize some of the key constituents of this general account. Slote tells us that the idea of an ethics of virtue is commonly viewed as involving two distinctive elements: A virtue ethics must treat aretaic notions (like “good” or “excellent”) rather than deontic ones (like “morally right,” “morally wrong,” or “morally obligatory”) as primary; and it must put a greater emphasis on the ethical assessment of agents and their motives and character traits than it puts on the evaluation of acts and choices (1992: 89). The theory advanced (in his 1992 book) is an “agent-focused” and not an “agent-based” one in that the ethical status of acts is not entirely derivative, he says, from that of traits, motives, or individuals, even though traits and individuals are the “major focus of the ethical views being offered” (89). Importantly, Slote emphasizes that as the virtue ethics he develops will allow for the independent aretaic evaluation of acts, “it will not (at least in this respect) depart as radically from familiar act-oriented deontology and utilitarianism as a view like Martineau’s or Plato’s seems to do” (90). I will revert to this significant point below. Slote’s theory avoids making use of specifically moral notions. Its fundamental aretaic concepts are those of “a virtue or of an admirable trait, action, or individual” (90), and these concepts, Slote claims, in their common usage and application are not specifically moral. He submits that traits of character can qualify as virtues “through what they enable their possessors to do for themselves as well as through what they enable their possessors to do for others,” and so the ordinary employment of the aretaic notion of a virtue “gives fundamental evaluative significance to the wellbeing of the self (i.e., the agent of an act or possessor of a trait) and to the well-being of others” (91). He underscores the point that our ordinary thinking about the virtues is equally concerned with “virtues helping others and virtues helping their possessors, and this latter fact, assuming it is a fact, expresses or embodies a kind of equality or symmetry of concern for self and others” (101). Slote proposes that the primary positive term of appraisal in his theory is ‘admirable’ and the primary negative term, to express the opposite of admirability, is ‘deplorable’ (94–5). Slote defends a number of virtue rules or principles. Adumbrating some of them might foster a deeper appreciation of his views. First, he

198

says that a virtue like kindness intuitively belongs to the category of the other-regarding, and fortitude (mainly) falls into the category of the selfregarding, yet neither of these admirable traits seems in any obvious way to be incompatible with the cultivation and exemplification of virtues in the category to which it does not belong. He adds: “it seems plausible to claim that any trait falling into one of these two categories that is (largely) incompatible with admirable traits of the other category should not be regarded and not count as a virtue” (105, note omitted). Second, it is part of commonsense virtue ethics, Slote says, to assume that “people should be concerned with the happiness and virtue . . . both of others and of themselves, just as, less symmetrically, Kant advocates a practical concern with one’s own virtue . . . and with other people’s well-being or happiness” (111). And third, commonsense virtue ethics is not to be thought of as requiring or recommending that we concern ourselves with others and ourselves, and in regard both to virtue and to happiness, at every possible moment. “If common-sense virtue ethics considers it deplorable if one doesn’t do enough for others or for oneself, it has no tendency to regard it as deplorable if one doesn’t at every moment actively concern oneself with people’s happiness and/or possession of (the) virtue(s)” (113). In a chapter of fundamental interest to the concerns of this book entitled “Virtue Ethics, Imperatives, and the Deontic,” Slote proposes that close analogues of deontic notions like wrongness and obligation can be developed on a purely aretaic basis, and that virtue ethics can be expressed in terms of precepts recommending concern for oneself and others in the imperative mood. He first argues that aretaic notions like “good” or “bad,” and “admirable” or deplorable,” can have a force of requirement or prohibition – a practical or action-guiding force – similar to that of standard deontic notions like “obligatory” or “wrong” (159–63). In a pivotal passage, worth reproducing at length, he then offers virtue-ethical analogues of rightness, wrongness, and obligatoriness: [A]retaic ethics can easily express its own equivalent of rightness by speaking of what is “not deplorable (or criticizable).” . . . The virtue-ethical analogue of (moral) wrongness is deplorability or (some appropriate degree of ) criticizability, and to the extent rightness is conceived simply as the contradictory or negation of wrongness, that is, as permissibility or acceptability, it would appear that nondeplorability or non-criticizability is the virtue-theoretic equivalent of (moral) rightness. . . . To the extent that claims of moral wrongness are thought to justify “ought” claims and imperatives, we are entitled to claim that someone ought not to do something and to use a grammatical imperative to tell her not to do that

199

thing, once the claim that it is wrong has been justified; likewise, within the aretaic realm, the claim that something is deplorable can at the very least justify imperatives and “ought” claims that stand to more familiar, deontic imperatives and “ought” claims as deplorability stands to wrongness. Where a given act would be deplorable, we can say “don’t do it” or claim that one ought not to do it, and such utterances will be derivatively or analogically deontic, if we insist that aretaic notions are fundamentally non-deontic. Such an insistence may possibly be warranted even in the face of the earlier-given arguments for the equivalent force of (morally) aretaic and (morally) deontic claims, respectively, of badness/ deplorability and of wrongness. But in fact I am not sure how one can uphold the aretaic/deontic distinction while granting the equivalence in force of aretaic and deontic judgments, and so, although there is more, much more, that needs to be said on this issue, I think we should at this point at least mention the possibility – relative to what has been argued above – that there may be no significant fundamental distinction between aretaic and deontic notions. (164, note omitted)

Presumably, a virtue-ethical theorist will seek to establish the compatibility of an ethics of virtue and determinism by arguing that the following sorts of principles: VE1: Acts are right, admirable, or good to the extent they exhibit or express appropriate virtues (or traits) V1, . . . , Vn and VE2: Acts are wrong, deplorable, or bad to the extent they exhibit or express the opposite of virtues V1, . . . , Vn, or are somehow deficient in V1, . . . , Vn are perfectly compatible with determinism. Even in a fully determined world, someone can intentionally perform an act that expresses an appropriate virtue or that is deficient in that virtue. Hence, it may be thought, determinism does not threaten such an ethics; it doesn’t undermine appraisals of the sort derivable from relevant facts and principles like VE1 and VE2. However, the passage last cited points to a deep problem for Slote. Assume, as Slote theorizes, that there may be no significant fundamental distinction between aretaic and deontic notions. In the absence of such a difference, we would expect that the virtue-ethical theory that Slote outlines “validate” deontic principles that “govern” deontic notions. After all, couching a theory in aretaic terms can’t, it would seem, alter the “logic” of ought statements. So, for example, Slote himself concedes that “where a given act would be deplorable [wrong], we can . . . claim that one ought

200

not to do it” (164); the rule that ‘wrong’ implies ‘ought not to’ holds. Similarly, we would expect Slote’s theory to validate, among other principles, principle K, a basic precept that links control (or ability) and deontic obligatoriness. Indeed, in an insightful discussion of agent-based virtue ethical theories with the distinguishing mark that the ethical status of an act is entirely derivative from independent and fundamental aretaic as opposed to deontic characterizations of motives, character traits, and individuals (1997: 239), Slote entertains, for purposes of illustration, a very simple theory. It says: “(roughly) benevolence is the only good motive and acts are right, admirable, or good to the extent they exhibit or express benevolent motivation” and “wrong or bad if they exhibit the opposite of benevolence or are somehow deficient in benevolence” (243). Slote then writes: [A]ssuming only some reasonable form of free-will compatibilism, a benevolent agent is typically capable of choosing many actions that fail to express or exhibit her benevolence. . . . The person who expresses and exhibits benevolence in her actions performs actions that, in agent-based terms, may count as ethically superior to other actions she might or could have performed, namely, actions (perhaps including refrainings) that would not have expressed or exhibited benevolence. . . . By the same token, actions will count as wrong or contrary to obligation only if they exhibit bad or deficient motivation, and this means for one thing that agent-basing is entirely consistent with “ ‘ought’ implies ‘can’.” (243–4)

If, however, principle K is validated by a virtue-ethical theory, then K’s complement, CK, the principle that ‘ought’ implies ‘can avoid’ should, given our argument above (see Chapter 3), also be validated by that theory. Or if K is validated by a virtue-ethical theory, then KW (the principle that it is wrong to perform an act only if one can perform that act) should also be validated by the theory. KW and OW (the principle that it is obligatory for one to do something if and only if it is wrong for one to refrain from doing it) entail that there is a requirement of alternative possibilities for obligatory acts. Then, though, on the assumption that “aretaic judgments may well give rise to full-fledged (so-called) deontic claims about what one ought or ought not to do and to full-fledged ‘deontic’ imperatives” (1992: 164–5), there will be no deontic anchors in deterministic worlds. It may, Slote indicates, be the case that “aretaic notions are fundamentally non-deontic” and that “analogical or ersatz deontic notions can be derived from aretaic ethical concepts” (164). Setting aside the question of what precisely such derivations would consist of and what such analogical or ersatz notions would be, if these notions could in fact be derived from aretaic ones, then it would seem that analogical versions of K, CK, KW,

201

and OW should also be derivable from the pertinent notions. If so, once again, it would seem, determinism would threaten virtue-ethical deontic anchors. Still, a virtue-ethical theorist need not capitulate, for a theorist of this sort need not be committed to the view that there is no fundamental difference between aretaic and deontic notions. Let’s examine the broad contours of a theory that, it appears, has no such commitment. BROAD CONTOURS OF ANOTHER VIRTUE-ETHICAL THEORY

Exploring the structure of a virtue-ethical theory, Watson, much like Slote, proposes that, central to an ethics of virtue, is the claim of explanatory primacy. This is the claim that “how it is best or right or proper to conduct oneself is explained in terms of how it is best for a human being to be” (1993: 451). He adds that an ethics of virtue is not a particular claim about the priority of virtue over right conduct but the more general claim that action appraisal is derivative from the appraisal of character, or that the “basic moral facts are facts about the quality of character. Moral facts about action are ancillary to these” (453). Watson suggests that an ethics of virtue will have two components: a. Some version of the claim of explanatory primacy. b. A theory of virtue. (455)

He says that the most familiar versions of (b) are theories of an Aristotelian kind, and that an Aristotelian ethics of virtue will have the following basic structure: (1) The claim of explanatory primacy. Right and proper conduct is conduct that is contrary to no virtue (does not exemplify a vice). Good conduct is conduct that displays a virtue. Wrong or improper conduct is conduct that is contrary to some virtue (or exemplifies a vice). (2) The theory of virtue. Virtues are (a subset of the) human excellences, that is, those traits that enable one to live a characteristically human life, or to live in accordance with one’s nature as a human being. (455)

Elaborating, Watson explains that we may depict the appraisal of conduct on an Aristotelian ethics of virtue with the following schema: 1. Living a characteristically human life (functioning well as a human being) requires possessing and exemplifying certain traits T. 2. T are therefore human excellences and render their possessors to that extent good human beings.

202

3. Acting in way W is in accordance with T (or exemplifies or is contrary to T ). 4. Therefore, W is right (good or wrong). (459)

Suppose Gary is in a deterministic world and he possesses and exemplifies T. Suppose he gives alms to the poor and his giving alms is in accordance with T. Then the Aristotelian schema yields the result that Gary’s giving alms to the poor is right. So it seems that if such a virtueethical theory is correct, people will still be able to do things that are right, wrong, and presumably obligatory in deterministic (or in appropriate nondeterministic) worlds. But there are problems with this view as well. It appears, first, that the sort of deontic morality that the Aristotelian schema generates – specifically, its concepts of rightness and wrongness – is distinctly different from that of the understanding of such morality that has been presupposed in this work. This requires explanation. Consider the following example: Suppose Poppy has decided, after careful reflection, to develop irresistible desires to act in accordance with traits T, and in due time succeeds. Suppose, on a certain occasion, she is in a situation in which she judges it best – not just best for her, but best all things considered, where this allthings-considered judgment coincides with her moral judgment of what it is best for her to do in her current situation – not to act on her irresistible desires. Given the irresistibility of the germane desires, she ends up acting in accordance with T, and so does right according to the Aristotelian schema. However, it pretty clearly seems that, even if she cultivated the irresistible desires with a view to influencing her conduct as they now do, there is a sense in which these desires now control her; she does not control them. She is compelled by them against her better judgment. In the view of deontic morality that I have presupposed, rightness, wrongness, and obligatoriness require dual control. Even in compatibilist versions of control, persons whose actions issue from irresistible desires that are contrary to their best judgments do not control their actions.1 Hence, in the view of deontic morality that I have proposed, Poppy’s action is not wrong although it is in accordance with T. The Aristotelian schema, of course, yields a different verdict. It seems, in consequence, that according to this schema, it is not part of the concept of one’s doing moral right or wrong that one have dual control or a sort of compatibilist control over what one does. Second, there is an implication of the Aristotelian schema that even virtue theorists may be reluctant to accept. Again, let’s suppose that Poppy does A, A is in accordance with traits T, but A issues from irresistible

203

desires and is contrary to Poppy’s best judgment. We can suppose that an agent S in world w has minimal rational control over an action A of S’s only if, holding constant the proximal desire(s) that gives rise to A, there is a scenario (a possible world) w ∗ with the same natural laws as w, in which S has reasons to refrain from doing A that S judges, under minimal conditions of rationality, to be better than those S has for doing A, and S nondeviantly acts on these reasons. The conditions of minimal rationality invoked are relatively weak. Consider a normal, adult human being who judges that her reasons for refraining from doing A are better than her reasons for doing A. Suppose the judgment she makes is deemed reasonable. Whatever minimal conditions of rationality she satisfies to elicit the appraisal that her judgment is a reasonable one are the very conditions of rationality that are presupposed in the notion of someone’s having minimal rational control over some action. So, for instance, the person acts on relevant factual information available to her (though some of this information may well be mistaken), she does not act on manifestly inconsistent beliefs, she is not self-deceived, she isn’t the puppet of a CNC controller, she isn’t making any obvious mistakes in reasoning, and so forth. In Poppy’s case, when Poppy does A, A is not minimally responsive to Poppy’s practical reasoning; Poppy lacks even minimal rational control over A.2 Still, the Aristotelian schema would yield the result that Poppy’s doing of A is right. This is a result one may well want to resist. A virtue-ethical theorist might propose the modification that acting in accordance with T is right (or acting contrary to T is wrong) only if one has minimal rational control over the action that is in accordance with T (or that is contrary to T in the case in which the action is wrong). But then it would seem that the actions of a moral paragon like an omnibenevolent being who cannot but act benevolently would not be morally right – again, a result not congenial to a virtue theorist. There is a third worry with the Aristotelian schema. The schema does attempt to derive moral appraisals of action, appraisals of whether, for example, actions are morally right or wrong, from moral appraisals of character. But with moral appraisals of action, unless one has good reasons to believe otherwise, it would seem that precepts like K, CK, KW, and OW are true. And if they are true, no action of any person would be right, or wrong, or obligatory in deterministic worlds even if the Aristotelian schema were beyond reproach. Finally, there is the concern that some of the virtues themselves (for example, forgiveness, courageousness or, perhaps, even honesty) cannot be analyzed independently of the deontic notions of rightness, wrongness,

204

or obligatoriness, but such deontic notions presuppose the freedom to do otherwise.3 In summary, I believe that there are reasons to be suspicious of the view that forms of virtue ethics such as those of Slote or Watson are possible in the absence of “metaphysical freedom.”

THE MANY FACES OF MORALITY AND THE AUTONOMY OF DEONTIC APPRAISALS

I want to end this chapter by providing reasons to support the thesis that deontic appraisals – appraisals to the effect that actions or states of affairs are morally right, wrong, or obligatory – are autonomous. Roughly, the autonomy-of-deontic-appraisals thesis says the following: (i) There are a plurality of possible moral evaluative assessments, or if one wants, there are a plurality of possible modes of moral evaluation such as deontic evaluation, aretaic evaluation, appraisability evaluation, evaluation regarding whether one has done anything instrumentally, intrinsically, or overall good (or evil), and so forth. (ii) Deontic evaluation is distinct from the other modes of moral evaluation (a feature of such distinctness is that such evaluation does not stand or fall with various other modes of moral evaluation). And (iii) a world devoid of moral (deontic) anchors is a world in which an important mode of moral appraisal – deontic appraisal – is missing. (I believe that aretaic appraisals are, in this respect, also autonomous.) One significance of this thesis is that it is an error to overlook the fact that deontic appraisals are one among, and distinct from, other possible modes of moral appraisal, and to suppose, for instance, that if an account of acting from virtue can successfully be defended, then deontic judgments (or judgments of appraisability) are, in some manner, inappropriate, superfluous, or illegitimate. By way of preliminary clarification of the thesis, we would do well to learn from Zimmerman’s perspicacious observation that “a person is in principle open to a variety of moral assessments and . . . he may score high on one scale and low on another, so that ambivalence is called for” (1993: 384). In his lucid discussion on the complexity of moral assessments, Zimmerman says: First, we can distinguish between what is evaluated: a person, a person’s character, a trait, a motive, an action (this list isn’t meant to be exhaustive).Then we can distinguish between modes of evaluation. Here we meet with a wealth of available terminology. On the positive side: “good,” “right,” “praiseworthy,” “laudable,”

205

“virtuous,” “admirable,” “commendable,” and so on; on the negative side: “bad,” “wrong,” “blameworthy,” “culpable,” “vicious,” “reprehensible,” “reproachable,” and so on. At the extreme, we could apply each of these epithets to each of the listed objects of evaluation and claim that in each case we get a distinct judgment. . . . Further complexity would be introduced by allowing for the possibility that a certain object of evaluation (an act, say) merits a certain epithet (“right,” say) in one respect or for one reason but doesn’t merit this epithet or merits its contrary (here, “wrong”) in another respect or for another reason. . . . Now I don’t wish to plead for such extremism. . . . [Whether we should accept the distinctness of all such judgments] would require a full-fledged, systematic moral theory, one which provides criteria for the application of each epithet to each object. (1993: 385–6)

Suppose we are settled on what it is for a person to act from (moral) virtue – what it is, that is, for a person’s act to express a virtue – and for what it is for a person’s act to be virtue deficient. Consistent with this supposition, it would seem that other sorts of moral appraisal could well be appropriate.4 So, for instance, on the assumption that Gary’s giving alms to the poor results from virtue, we can still legitimately ask whether Gary is deserving of praise for giving alms, or whether Gary gives alms in all the best worlds accessible to him at the relevant time (again, we simply leave the axiological question of what “bestness” consists in open). It certainly seems possible for a person not to do the best he can (in this manner) and still act from virtue, or for a person to succeed in doing the best he can (in this manner) and his act’s being virtue deficient. Both sorts of assessment, the deontic and the aretaic, seem perfectly legitimate and the one doesn’t, in any way, invalidate or nullify the other. The two sorts of assessment are autonomous. As another illustration, suppose Doctor Getemwell is responsible for the treatment of a certain patient. All the evidence indicates that the patient is suffering from a rare and dangerous disease, “Malady,” that can be cured easily and safely by administration of one dose of medicine A. Giving readily available B, though, to a patient who suffers from Malady instantly kills the patient. Suppose, doing the very best she can to save her patient, Getemwell gives the patient A. We can assume that Getemwell acts from virtue – her giving of A expresses various virtues such as, for example, those of conscientiousness, kindness, and benevolence – and is not virtue deficient at all or at least not virtue deficient with regard to the moral virtues. But suppose, through no fault of anyone, and given the current state of medical technology, the diagnosis is incorrect. The patient is in fact suffering from a different disease, Malaise, which causes a patient to

206

exhibit symptoms almost indistinguishable from those of Malady. Suppose, finally, that giving B to a patient who suffers from Malaise safely cures the patient, but giving A to a patient stricken with Malaise kills the patient. Although Getemwell acts from virtue in giving A, giving A is not what she ought morally to do, as in all the best worlds accessible to her as of the time just prior to giving any medicine, she gives B to the patient. Nor, in such a case, would we judge that the good doctor is blameworthy for giving A. One might object that one can be led to affirm the autonomy of the deontic (and the aretaic) only at the expense of being insufficiently sensitive to certain other theses, perhaps metaphysical ones, that are presupposed by, or motivate, the different moral evaluative modes. A case in point would be Aristotle’s teleological metaphysics, which seems to have played a central role in his conceptualization of virtue-ethical appraisals. As Stephen Darwall explains, Aristotle believed that every natural thing has a purpose or “final cause” that simultaneously explains how it actually acts and provides a normative standard for evaluating its activity or development. According to Aristotelian teleological science, how a thing actually behaves can be explained by its inherent goal or telos, and its telos also specifies how it should behave (1998: 194). Aristotle’s teleological metaphysics underpins his ethical views: The chief good required to structure all human choice – human flourishing – is an end inherent in our nature; and what we should do, be, and become is implicit in what we are (195). Darwall indicates what he takes to be a striking contrast between Aristotle and Mill: Even if, as Aristotle and Mill both believed, human beings are profoundly social creatures whose happiness substantially includes the happiness of others, there can nonetheless be genuine conflicts of interest between them. At least in our imperfect circumstances, it can happen that a more flourishing life for some comes at the cost of a less flourishing life for others. When confronted with such a situation, we have to choose. . . . For . . . Mill, this is the kind of situation that gives morality its point. . . . For Aristotle, however, the ultimate and unquestionable standpoint of practical deliberation is the agent’s own flourishing, since it is the end that is inherent in her own metaphysical nature. . . . This is the natural view to take if one holds to a teleological metaphysics, since a teleological structure provides a metaphysical guarantee that individual interests will harmonize. A common telos ensures that human life does not present us with a fundamental collective action problem. It ensures that mutual disadvantage cannot result from everyone’s acting for his own good. Consequently, no conception of another normative perspective (morality) that governs and overrides self-interest for mutual advantage is required.

207

The flourishing of each is guaranteed to interlock with the flourishing of all. Without teleology, however, there is no such assurance. (196)

Let’s suppose that Aristotle’s teleological presuppositions are sound. Grant, in particular, that “human life does not present us with a fundamental collective action problem.” Darwall may well be right that, in consequence, no conception of an evaluative perspective – “morality” – that governs and overrides self-interest for mutual advantage is required – required in the sense that such a perspective is needed to help us deal with collective action problems. Still, it doesn’t follow at all that there is no bona fide alternative evaluative moral perspective that appraises actions deontically. We can legitimately ask of an Aristotelian agent who, on a certain occasion, acts from virtue – acts on that occasion as her telos “dictates” that she should – whether her acting in that way is consistent with the way in which she acts in all the best worlds then accessible to her. But what would be the point or purpose of such deontic appraisals, it might well be asked, in an Aristotelian teleological world? The point would simply be that such an appraisal would constitute a different sort of moral appraisal than an aretaic one, just as appraising the Aristotelian agent for whether she has done something that is overall good or best, given her alternatives, would be a different sort of moral appraisal than whether she has acted from virtue. Morality has many evaluative dimensions. Maybe some of them have some sort of “deep” substantive “point”; whether or not this is so, we err by overlooking the complexity of the many faces of morality. Finally, one might attempt to deny the autonomy of the deontic by defending this sort of virtue-ethical theory. First, one would develop what Watson calls a “theory of virtue.” Among other things, this would single out a list of the moral virtues. Then one might clarify what it is to act from moral virtue; one would defend the conditions that require satisfaction for an act to express a virtue and for an act to be virtue deficient. Presumably, such an account would allow for the possibility that one’s act could express a certain moral virtue (say, that of honesty), but also be virtue deficient (say, by falling short of expressing the virtue of kindness or generosity). One might, for instance, develop an overall “virtue metric” that tells us whether, of two or more acts that express virtues, which is best in light of the virtue metric. Finally, one might propose this sort of basic moral principle: BMP: An act A is obligatory for S if and only if S performs A out of virtue, and, given the virtue metric, A is virtuewise best.

208

BMP is a very rough general schema. I am not concerned with working out its details. Suppose something like BMP could be defended. Wouldn’t this then show that aretaic appraisals – acting from virtue appraisals – and deontic appraisals amounted to one and the same thing, and that the two sorts of appraisal were not autonomous? Not at all. For suppose S does A from virtue and A is virtuewise the best; it is best according to the virtue metric. Then a principle like BMP would dictate that A is obligatory for S. Although A would be obligatory for S, we could still legitimately ask: Did S fail (or excel) morally in some nondeontic aretaic respect? Suppose A expressed the virtue of honesty but was “deficient” in the virtue of kindness. Surely we could legitimately judge that in doing A, S did something that is aretaically deficient; A is “kindness deficient.” An analogy might help. Let the “I-utility” of an act be the result of subtracting all the units of intrinsic evil that would be produced as a result of performing that act from all the units of intrinsic goodness that would result from its performance. Suppose BMP∗, a rival of BMP, dictates that an act A is obligatory for S if and only if, of S’s alternatives, A maximizes I-utility.5 Even if BMP∗ were true, deontic appraisals and what we can call axiological ones would still be distinct. If, of S’s alternatives, A maximizes I-utility, then A is obligatory for S. We can still legitimately inquire into whether, for example, S had a set of alternatives of which A is a member, such that some other member, if performed, would have maximized overall goodness; or some other member, if performed, would have produced less intrinsic evil. Further, there would be a point to such axiological appraisals. Suppose S judged that, in the circumstances, doing what is overall good (doing B) is better than doing what is intrinsically best (doing A). Then although S would have failed to do what is obligatory, S’s doing of B would have some moral merit. I see no reason whatsoever to deny that the latter sort of appraisal is a legitimate sort of moral appraisal. Both sorts have their place in an overall, comprehensive, ethical view. If aretaic and deontic appraisals are autonomous, then however rich the aretaic appraisals of whatever the subject of moral evaluation in the world – persons or traits – happens to be, that world will be deficient in a vital mode of moral evaluation if it is devoid of deontic anchors. This point is of sufficient importance to merit some final comments. Without deontic anchors we cannot appropriately make deontic moral appraisals of the sort that some things are morally right, wrong, or obligatory. This is a significant loss. Of course, as I have suggested, with no deontic anchors there is no bar to some state of affairs being just as good as, better, or worse than others, where the notion of goodness may either

209

be that of instrumental, intrinsic, or overall goodness. So, for example, if we assume a simple hedonistic axiology according to which the sole intrinsic good is pleasure and the sole intrinsic evil pain, even in a world devoid of deontic anchors, some states of affairs can be more “pleasure involving” than others and hence be intrinsically better than others. But it is one thing for states of affairs to be appraised, as it were, axiologically, that is, appraised for their “goodness value” but quite another for these to be appraised deontically, that is, appraised for whether they are right, wrong, or obligatory. Suppose, for example, that whatever state of affairs an agent brought about on a certain occasion, it was causally determined that she did so. Suppose, further, that the agent believed she could have brought about either of two states of affairs on that occasion. From the fact that she brought about the one that is intrinsically worse, it does not follow that she did any wrong. Had she brought about the intrinsically better one instead, it would not have followed that she would have brought about the state of affairs that it would have been morally permissible for her or morally obligatory for her to bring about. The situation here would be somewhat parallel to the one in which blameworthiness for an action required, let’s assume, freedom to do otherwise, and an agent deliberately did what he took to be wrong. If determinism undermines freedom to do otherwise, then the control-relevant requirements of blameworthiness, given our assumption, have not been satisfied by the agent. The agent, then, would not be blameworthy for what he did, even though he may have intentionally done what he did in light of the belief that he was doing wrong. Perhaps it might better help to appreciate what is involved with a loss of deontic anchors if we think of the following: Assume, first, that actions can, in a certain world, be right, wrong, or obligatory. A person in such a world might supererogate on some occasions, suberogate on others, and do the minimum morality requires without doing what he is obligated to do on yet others. Thinking back to McNamara’s example, I promise to get in touch with you, and in virtue of promising, become obligated to do so. I can discharge this obligation either by writing to you or by visiting you, but I can’t permissibly do both. Suppose I put in a better performance morally by visiting rather than by writing, although either is permissible. I do the minimum morality requires – I write. But in so doing, the writing is not itself obligatory, for “morality demands only that I contact you and that I don’t both write and visit, whereas doing the minimum that morality demands includes these plus writing you” (McNamara 1996: 426). In such a world, on some occasions I can meet

210

morality’s demands in a number of acceptable yet worse and worse ways without doing wrong. Although I don’t put in the best performance by writing, writing is morally permissible for me, and if I don’t contact you at all, I do wrong. Now suppose that I am in a world without deontic anchors and I find myself in similar situation with similar options. Writing would still be worse than visiting you, but it is false that it would be permissible. In addition, presumably, not contacting you would be worse yet than either visiting or writing, but it is false that it would be wrong. I cannot, in consequence, be condemned for doing wrong by not contacting you though I would have done better by visiting. In summary, among other modes of appraisal, aretaic, deontic, and axiological appraisals all have their place in a full-fledged ethical theory. A world in which one of these sorts of appraisal is not possible because, for example, it lacks deontic anchors, is one in which a full moral story – a full moral appraisal – of the sort that would be possible in a world in which each of these sorts of appraisal were possible, is not forthcoming. Prior to exposing what I believe is another legitimate cost of being deprived of deontic anchors, I wish to examine one final proposal, having to do with the alleged “overridingness” of moral obligation, regarding what would be lost without deontic anchors. This is the topic of the next chapter. APPENDIX: KEKES ON K

I have either advanced or argued for, among other things, the following views: (a) A multiplicity of moral appraisals like deontic or aretaic ones are possible and these should not be conflated. (b) The principle of alternate possibilities is problematic. (c) The Objective View – an agent is morally blameworthy (praiseworthy) for performing an action only if it is morally wrong (right or obligatory) for that agent to perform that action – is suspect. And (d) there is an asymmetry in control requirements for deontic anchors and moral responsibility. Whereas deontic morality requires dual control, moral responsibility, it appears, does not. These views can be enlisted to evaluate various proposals, principles, or theses that are relevant to this work. In this appendix, attention is directed yet again to principle K. I examine John Kekes’s reservations about K (really, a K-like principle) and explain why they are not cogent. Of the five arguments I consider, each relies either implicitly or explicitly on one or more of (a), (b), (c), or (d).

211

In a paper entitled “‘Ought Implies Can’ and Two Kinds of Morality” (1984) Kekes dubs a K-like principle “The Principle” and he interprets it as follows: A person is morally obliged to do something only if it is in his power to do it or not to do it (1984: 459). The inclusion of the “not to do it” clause distinguishes this principle from K. The arguments that Kekes advances against The Principle, though, if cogent, would equally impugn K. Having noted this, I shall treat Kekes’s arguments in what follows as arguments against K. Kekes adds that “if someone is so obliged [i.e., obliged to do something in the manner captured by The Principle], then it is appropriate to approve of his doing it and disapprove of his not doing it” (459). He then advances two cases as a prelude to questioning The Principle. The first is a slightly altered episode in Styron’s Sophie’s Choice. A mother and her two children await transportation to a concentration camp. She has a chance to save one of them by giving the child to an obliging stranger and has a few hours to decide which child to save. She agonizes over the choice but is unable to make it. By the time the decision must be made, she has collapsed as the psychological burden was too great for her. Subsequently, all three die. Kekes remarks: There are strong reasons for saying that the mother ought to have made a decision and thereby saved one of her children. . . . Due to weakness, understandable as it is, she failed. She could not make the decision, yet, all the same, she ought to have made it. (459)

In a second case, a mentally ill sadist has entrapped dozens of people and tortured them to death. He knows that others think of what he is doing as evil, but he does not care. He realizes, too, that his victims suffer; indeed, he tortures them precisely because their suffering delights him. Assume that it is not in the sadist’s power to stop doing what he does. “Surely,” Kekes claims, “we should say that he still ought not to do it? But if we accept The Principle, we cannot say this” (460). Kekes concedes that not everyone will agree with his judgment that the mother ought to have made the decision or that the sadist ought to have stopped the entrapment and torture of innocent persons. For, Kekes says, it might be objected that if the mother literally could not have made the decision and the sadist literally could not have stopped his vile activities, “moral judgments are then inappropriate, and so The Principle can be maintained after all” (460). Kekes, though, responds that a reasonable person may be left unimpressed by this argument. He may say that kind and strong people are obviously morally better than cruel and weak ones, and that this is true independently of how they came to acquire

212

their virtues and vices. . . . Of course, disapproval should be mitigated by extenuating circumstances. A person who is vicious by choice should be judged more severely than one who cannot help being vicious. . . . However, extenuation is needed only when wrong-doing is present. Cruelty and weakness are thus proper objects of moral disapproval. The Principle holds that such disapproval is always inappropriate, and that shows that The Principle is faulty. (460–1).

It appears, then, that one of Kekes’s arguments against K amounts to this: Argument 1 (1) If K is true, then disapproval in the two cases (the mother’s and the sadist’s) is inappropriate. (2) It is false that disapproval in the two cases is inappropriate. (3) Therefore, K is not true.

The rationale for (1) apparently rests on what Kekes calls the “strong version” of The Principle. It says that “A person can be approved or disapproved of for what he does only if it is in his power to do it or not to do it” (462). (Kekes explains that the weak version of The Principle plays a secondary role. “Disapproval should be mitigated by extenuating circumstances, and the weak version of The Principle expresses one possible ground for extenuation” (462).) There are, however, difficulties with this argument. First, an advocate of K, contrary to what Kekes seems to suppose, can agree that kind and strong people are morally better than cruel and weak ones consistent with maintaining K. Both aretaic and deontic judgments are bona fide moral judgments. From the fact that a deontic judgment is not appropriate, it does not follow that an aretaic one is not as well. Similar things can be said about moral disapproval. We can certainly disapprove of cruelty, and legitimately so, in cases like the ones Kekes describes even though the agents do no wrong. One proper object of disapproval is cruel behavior; another is the vice of the agents. In such cases, one can express disapproval of the cruel behavior that issues from the pertinent vices. Second, against line (1), K does not say anything about when moral approval or disapproval is appropriate; K is a principle concerning deontic control. This is, of course, not to deny that in cases like the mother’s or the sadist’s, although the agents do no wrong (assuming they lack deontic control over their behavior), others may well be under a moral obligation to express disapproval of the agents’ cruel behavior.

213

Finally, defenders of K can invoke Frankfurt-type cases to question (1). One moral of such cases is that an agent can be blameworthy for performing an action even though the agent could not have refrained from performing it. Advocates of K should rightly insist that in Frankfurt-type cases, what the agent does lacks a primary deontic status – what is done is not morally right, wrong, or obligatory. Yet overt blame, or some other appropriate form of disapproval, could surely be in order in these cases. In recent works, Kekes once again takes up his case against K. Prior to addressing his fresh attacks, some stage setting will be helpful. One of Kekes’s central concerns in “The Reflexivity of Evil” (1998) is to cast doubt on the view that agents whose evil actions are not autonomous are not morally responsible for these actions. Beginning with a brief explanation of key terms, Kekes says that human actions are evil if they cause serious and morally unjustified harm to other human beings. Harm is serious if it interferes with the functioning of a person as a full-fledged agent. Causing such harm is morally unjustified if it is avoidable, undeserved, and not the best way of preventing even greater harm. Actions are autonomous “if their agents (a) choose to perform them, (b) their choices are unforced, (c) they understand the significance of their choices and actions, and (d) they have favorably evaluated the actions in comparison with other actions available to them” (1998: 217). He adds that actions of which any one or more of (a), (b), (c), or (d) is not true are nonautonomous. Kekes proposes that the claim: NAMAR: Agents are not morally responsible for their nonautonomous evil actions, contradicts “widely shared and strongly held moral judgments” (221). For if NAMAR were true, we would have to deny that agents who are “undogmatic, sensitive, and altruistic, and whose virtues lead them to benefit others habitually and predictably, but . . . are acting nonautonomously,” are morally responsible for good actions that issue from their virtues (221). Specifically, if NAMAR were true, Kekes proposes, we would have to deny that (i) it is morally better to be undogmatic, sensitive, and altruistic than to be dogmatic, insensitive, and ruthless; and (ii) virtues and virtuous actions are morally better than vices and evil actions (222). To the question, “What could lead reasonable and morally committed people to such absurdity?” (222), Kekes’s answer is that it is acceptance of mistaken principle K. A preliminary interjection is appropriate. Even if NAMAR is true, (i) and (ii) need not be denied. Again, we should remember the key

214

lesson that there are many different sorts of moral appraisal and that the conditions under which one sort of appraisal is germane need not coincide with those under which another sort of appraisal is germane. Perhaps agents who habitually perform nonautonomous good actions are no more responsible for these actions than agents who habitually perform nonautonomous evil actions are responsible for these evil actions. Still, as we have argued, when, for instance, we ponder such agents, various aretaic judgments like those expressed by (i) and (ii) may well be perfectly legitimate. Turning now, to Kekes’s additional arguments against K, Kekes claims that K is mistaken because: The principle ascribes moral responsibility to agents for the way in which they come to perform their actions, but not for the effect the actions have on others. The principle is thus preoccupied with causes at the expense of effects, and that is what is wrong with it. (225)

Elaborating, Kekes proposes that one ultimate purpose and justification of morality is to prevent evil and to promote good (225). If we keep this aim of morality clearly in sight, then we should realize that K is “misused as a criterion for the ascription of moral responsibility” (226). For if it is the prevention of evil that centrally matters to morality, then even agents whose evil actions are nonautonomous, Kekes suggests, can be morally responsible for their evil deeds, despite the fact that: The effect of . . . [K] is to place nonautonomous actions outside of moral concern, thus exempting their agents from moral responsibility. Whatever restraint morality can then exercise is thereby removed from nonautonomously acting agents. This is why . . . [K] actually fosters evil. (227)

To bolster his position, Kekes adduces a number of cases. One involves a sixteenth-century witch-hunter whose task is to find, interrogate, prosecute, and burn witches. A devout Christian, he is convinced that witches are possessed by the Devil, and he sees himself as God’s servant, fulfilling his duty of obeying God’s decrees by extirpating evil. “He has no doubts about his beliefs; he is supported in his activities by the highest authorities; and he is sincere, dedicated, and as just as he can be in following ecclesiastic law” (219). Kekes’s judgment is that the witch-hunter is morally responsible for burning and torturing several women even though his evil actions, given some simple presuppositions, are nonautonomous (219). In another case, a ruthless and highly successful businessman has achieved what he has through dedication, keen competitiveness, and unwavering

215

expediency. Following in his father’s footsteps, he is merciless to his competitors, sacrifices others when it is necessary, and observes moral prohibitions only when they serve his purpose. He “sees morality, friendship, love, and loyalty . . . most often as handicaps that reasonable people will overcome” (219–20). He believes that strong and realistic people all act as he does, and that those who question his belief are “weak, or insincere and cunning competitors trying to prevail over him” (220). Again, Kekes proposes that even if we assume that the ruthless businessman’s evil actions are nonautonomous (“The choice was forced on these agents; they lacked the means for the critical evaluation of their conduct; and they did not understand the significance of their habitual and predictable actions” (220)), the businessman is responsible for his nonautonomous evil deeds. Reminiscent of his discussion in his earlier work on The Principle, Kekes distinguishes two versions of K, the legitimating and the degree assigning. He writes: Both versions are used as criteria for ascribing moral responsibility. But the legitimating version is used to determine whether the ascription of moral responsibility is at all legitimate, whereas the degree-assigning version is used to determine the degree of responsibility that is legitimate. The legitimating version marks a threshold, below which the ascription of moral responsibility is unjustified and above which it is justified. The degree-assigning version marks gradations of moral responsibility which may be justifiably ascribed above the threshold. (229)

Kekes believes that the legitimating version (“LK”) but not the degreeassigning version is mistaken. For our purposes, discussion will be confined to LK save for the following passing remarks. Kekes believes that the degree-assigning version should be used to gauge degrees of moral responsibility, and he suggests that greater autonomy creates greater moral responsibility, lesser autonomy, and lesser moral responsibility (230). Perhaps this suggestion about assigning degrees of responsibility is ultimately defensible, but it is difficult to see what K has to do with it. One may well accept this suggestion without abandoning K. One prominent argument against LK appears in this passage: Consider . . . the dogmatic . . . [and] ruthless, but nonautonomous agents discussed earlier. Are they morally responsible? If the legitimating version of the principle were correct, it would follow from it that, since they lack autonomy, it would be unjustified to hold them morally responsible. Due to their circumstances, innate dispositions and formative experiences, they could not have acted otherwise than they did. If ought implies can, and if they could not, then it is unjustified to

216

say that they ought not to have acted the way in which all influences on them prompted them to act. The dogmatic witch-hunter . . . and the ruthless businessman are therefore not morally responsible for their evil actions and vices, and should not be condemned as wicked. And this surely is an unacceptable moral view. (229)

This passage suggests that LK is or at least implies: LK1: S is morally responsible for doing A only if S could have done other than A. The argument against LK, then, is this: Argument 2 (1) If LK1 is true, then the witch-hunter and the ruthless businessman are not morally responsible for their evil deeds. (2) It’s not the case that the witch-hunter and the ruthless businessman are not morally responsible for their evil deeds. (3) Therefore, LK1 is not true.

Given (3), it follows that LK is not true if LK1 is its implicate or LK1 is identical to LK. There are problems with this argument, though. First, as I have proposed elsewhere (Haji 1998a), if belief is essential to blameworthiness – if it is true that one cannot be blameworthy for performing an action unless one performs that action in light of the belief that one is doing wrong – then it is not evident that characters like the witch-hunter and ruthless businessman are blameworthy for their evil deeds. A second problem is that LK1 is just PAP (the principle of alternate possibilities) and PAP is problematic. A third fundamental problem is that K is a principle about deontic control and not a principle for the ascription of moral responsibility. In other words, LK1 (or LK) are diverse from K. Perhaps, though I am unsure about this, Kekes might have had the following in mind. Suppose one accepts: OB: S is morally blameworthy for doing A only if A is morally wrong. As (ONO): An action A is morally wrong for a person S if and only if S ought not to do A, and S ought not to do A only if S can refrain from doing A (as K implies), we can derive from OB, ONO, and K:

217

LK2: S is morally blameworthy for doing A only if S can refrain from doing A. It is open to Kekes to invoke the following argument against LK2: Argument 3 (1) If LK2 is true, then the witch-hunter and the ruthless businessman are not morally blameworthy for doing A. (2) It’s false that the witch-hunter and the ruthless businessman are not morally blameworthy for doing A. (3) Therefore, LK2 is not true.

The major problem with Argument 3, however, is that OB, as we have discussed in preceding chapters, is false, since OB is equivalent to the blame component of the Objective View. Finally, let’s consider a couple of variations of an argument against K suggested by Kekes in his 1997 book, Against Liberalism. In a section entitled “The Principle: ‘Ought’ Implies ‘Can,’” Kekes says: If it is true that an agent ought to act in a certain way, . . . principle [K] asserts, then it must also be true that the agent could act in that way. In other words, it is a condition of being morally responsible that the agent be able to discharge the responsibility. It follows that if it is impossible for an agent to fulfill a moral responsibility, then it is illegitimate to ascribe the responsibility to the agent. The principle thus presupposes that the agent has sufficient control over the action in question. The presence of such control is necessary for moral responsibility, whereas its absence exempts agents from it. The specifications of the required control . . . will look pretty much like the conditions of autonomy. . . . Ought implies can, then, translates into the claim that only autonomous agents are morally responsible. (52)

In one interpretation, the last sentence in this passage suggests that Kekes takes K to be (or maybe to imply) the following principle: LK3: S is morally responsible for doing A only if A is an autonomous action of S’s. When we combine some of the things that Kekes says in this passage with his assumption that the witch-hunter and the ruthless businessman are morally responsible for their pertinent deeds, the following argument against LK3 suggests itself: Argument 4 (1) If LK3 is true, then the witch-hunter and the ruthless businessman are not morally responsible for doing A.

218

(2) It’s false that the witch-hunter and the ruthless businessman are not morally responsible for doing A. (3) Therefore, LK3 is not true.

Again, one worry with Argument 4 is that line (1) is controversial. A second worry is that that LK3 is not identical to K. Kekes may have been led to believe that both LK3 and K amount to one and the same principle as (i) both deontic morality and moral responsibility require control – one’s action can’t be morally right, wrong, or obligatory and one can’t be morally responsible for that action – unless one has control over it; and (ii) the types of control required for deontic morality and moral responsibility are “pretty much like (those specified by) the conditions of autonomy” (52). It is a presupposition of clause (ii), in other words, that the sort of control required for moral responsibility is of the very sort that is required for deontic morality. I have argued, though, that this is false; there is, in fact, an asymmetry in control requirements for moral responsibility and deontic morality. In a second interpretation, the passage in the book contains a more complex argument than Argument 4. It’s this: Assume that (i) the witchhunter, who believes that innocent Ann is a witch, burns Ann, and is morally responsible (blameworthy) for doing so. Further, assume (ii) for reductio that K is true. Finally, assume (iii) that OB and ONO are true. Argument 5 unfolds in this fashion: Argument 5 (1) If the witch-hunter is morally blameworthy for burning Ann, then the witchhunter ought not to burn Ann. (2) If the witch-hunter ought not to burn Ann, then the witch-hunter can refrain from burning Ann autonomously. (3) It is false that the witch-hunter can refrain from burning Ann autonomously.

Therefore, (4) It is false that the witch-hunter is morally blameworthy for burning Ann. (5) The witch-hunter is morally blameworthy for burning Ann. (6) Therefore, K is false.

Let’s briefly review the premises. Line (1) is based on OB; a person is blameworthy for doing something only if that person ought not to do it. Line (2) rests on the view that if one ought not to do something, one has control over not doing that thing, and the sort of control is “autonomous control.” That is, one ought (ought not) to do A only if

219

one can do (can refrain from doing) A autonomously. Line (3) is predicated on certain facts of the case. Line (4) is an intermediate conclusion. Line (5) is an assumption of Kekes. It should be fairly obvious that there are several things wrong with this argument: OB is suspect; (5) is questionable; and (2) seems to invoke K, something illegitimate in an argument against K. I conclude, therefore, that none of the arguments of Kekes against K considered in this appendix is sound.

220

12 On the Connection between Morality’s Dethronement and Deontic Anchors

Of all the different sorts of obligation such as prudential, legal, or moral, moral obligation has been thought by many from Plato down to several contemporary thinkers to be the most stringent or “overriding.” The moral “ought’s” claim to supremacy – its claim to overridingness – it may be thought, kindles another concern with being deprived of deontic anchors. In rough strokes, the issue is this: In a world with no deontic anchors, no nonmoral variety of obligation would be overriding, as it is moral obligation that is overriding, and this sort of world is devoid of such obligation. But then there would be no ultimate standard to which one could appeal to settle what one “plain ought” to do in “conflict situations” where, for example, legal obligation conflicts with prudential obligation, or prudential obligation conflicts with duties of love. For the moral “ought” is the “overarching” ought, the standard that dictates what one plain ought to do in conflict situations, but a world without deontic anchors is a world without moral obligation. Hence, loss of deontic anchors would be accompanied by loss of an overarching standard – the final court of appeal for settling normative conflicts of the relevant sort. Consequently, practical reason would, in a manner of speaking, be “fragmented.” Aspects of this concern ring true to me; others are rife with confusion. To sift truth from confusion, let’s start with some fundamentals. THE OVERRIDINGNESS THESIS

The Overridingness Thesis says that all in (that is, overall ) moral obligations (or moral “oughts”), as opposed to prima facie ones, always trump all other all in normative obligations such as prudential or legal ones. According

221

to this thesis, if, as of some time t, an agent S ought morally to do A at t ∗ (where t ∗ may be identical to or later than t), and S ought (say legally) to do B at t ∗, but S cannot do both A and B at t ∗ , then (as of t) S “plain ought” to do A at t ∗.1 The Overridingness Thesis, then, as I understand it, is a thesis about the relative normative importance of verdicts delivered by various special normative standpoints such as morality, etiquette, law, self-interest, and so on. We can suppose that each of these is a standpoint regarding some concept of obligation. It should be evident that in many cases these concepts conflict. So, for example, moral obligation may conflict with prudential or legal obligation. In a specific case, a person may realize that she has a moral obligation to do one thing, a prudential obligation to do another, and a legal obligation to do yet something else, but may legitimately ask: “I’m aware what my specific obligations are and that they conflict; but just what plain ought I to do?” When a person asks this question, she is asking about which of these obligations takes “precedence over” or “overrides” the others. To elaborate, imagine being in a situation in which one is aware that the dictates of morality and those of long-term self-interest conflict.2 In this, as in all situations, morality passes judgment on what one ought morally to do or what it is morally permissible for one to do, and in this particular conflict, morality forbids the action that is obligatory from the point of view of self-interest. Further, and this is directly pertinent to the concern registered in the opening paragraph of this chapter, though this action is condemned by morality, morality itself, it seems, cannot deliver a verdict about the relative “normative importance” of what it, as opposed to what self-interest, recommends; it can simply tell us what is right, wrong, obligatory, or gratuitous from the point of view of its own special moral perspective. Analogously, the standard of self-interest dictates what one ought to do to maximize personal welfare, and in the conflict situation, it will proscribe the action prescribed by morality. But even so, self-interest cannot rule on whether its prescriptions are normatively more important than those of morality’s. Like morality, it is a “self-contained” standard, restricted to dictating what is right, wrong, obligatory, or perhaps gratuitous from the point of view of its own special perspective. It appears, then, that we cannot make sense of the question of whether the verdicts of morality override those of self-interest, or for that matter, of other special standpoints unless we suppose that there is an overarching normative standard, one regarding an all-inclusive concept of obligation, that passes judgment on the relative normative stringency of the special standpoints.

222

I follow Copp (1997b) in calling such a standard “Reason,” and I will say that Reason dictates what we “plain ought” to do. Copp plausibly suggests that Reason would have the properties of “comprehensiveness” and “supremacy.” He explains that Reason would take the verdicts given by all the special standpoints in any situation where various obligations conflict, “evaluate these verdicts without any question-begging; and . . . produce an overall verdict as to what the agent is to do. . . . [Reason] would be ‘comprehensive’ ” (1997b: 94). Further, he says that Reason would be the normatively most important standard for evaluating such verdicts and the choice of how to act. “Hence, an agent ought simpliciter to comply with its overall verdict . . . it would be ‘supreme’” (1997b: 94–5). ‘Reason,’ then, as I am using the term, does not, for example, refer to the standard which says, roughly, that rationality consists in maximizing long-term self-interest. Nor does it refer to a set of powers or capacities like the powers to deliberate, to make practical judgments, to form intentions or to arrive at decisions, or to reason practically. For further clarification of the Overridingness Thesis, consider what could be said in its support. First, some might submit that a normative standard qualifies as a morality only if it is overriding.3 (This should be distinguished from the superficially similar submission that a standard is a morality only if it is one to which we attribute “overriding importance.” For it is possible for us mistakenly to attribute overriding significance to a standard that is not in fact overriding.) This is a dubious view, however. It has, to begin with, some untenable consequences. For example, if it turned out that the specific or special standard now identified with self-interest were overriding, then the special standard now identified with morality would fail to qualify as morality! Further, if no normative standard were overriding, then the view would implausibly imply that no standard qualifies as a moral one. Another of its difficulties is that the identification of morality with any standard that is overriding (even assuming some standard to be overriding) would not conceptually guarantee that standpoints now identified with, say, self-interest and morality, harmonize in a manner in which it is not possible for the two to conflict. One could, then, in situations in which conflicts did occur still coherently ask: What ought one to do from a standpoint that provides an ultimate assessment of the relative normative stringency of (what is now taken to be) morality and self-interest? Finally, it is unclear what would secure the view that nothing can qualify as a morality unless it is overriding other than the unhelpful question-begging claim that for some normative “ought” or obligation to be a moral one, it must be supreme.

223

Second, one might attempt to champion the Overridingness Thesis by exploiting Bernard Williams’s distinction between morality and ethics. According to Williams, ethics, in broad strokes, is concerned with how one should live, whereas morality, a particular part of ethics, is concerned with universalizable and impartial obligations (1985: 174–96). Answers to how one should live, Williams claims, will take into account much more than just moral concerns. So while Williams’s “morality,” as Williams admits, can be overridden by nonmoral considerations such as aesthetic ones, ethics cannot. This may well be true but the distinction between morality and ethics leaves untouched the fundamental issue of overridingness, which can equally be expressed in these terms: When that part of ethics which is morality conflicts with another part, such as art or self-interest, which is overriding? Third, it might be thought that Lawrence Becker’s distinction between special and general conceptions of morality sustains the Overridingness Thesis. Special conceptions of morality interpret morality as just one point of view “among many that a rational agent might consider” (Becker 1986: 17). This conception can conflict with other distinct points of view in terms of which one might choose to act, like etiquette or egoism. In contrast, the general conception of morality identifies morality with what ought to be done all things considered. “Moral oughts, under the general conception, are by definition, overriding in the sense that they are ‘final.’ What we ought to do all things considered cannot possibly be superseded in the way that judgments from a special point of view can be” (28). Becker’s distinction between special and general conceptions of morality, though illuminating, does not, I believe, lend support to the Overridingness Thesis. This is because it is not clear that the all things considered “ought” (assuming there is such an “ought”) is what I have termed “plain ought.” For if these two “oughts” were identical, then it would be the case that, under any circumstances, if one plain ought to do something, one ought also, all things considered, to do that thing. Now, under the condition that no standpoint is overriding, plain “ought” and Becker’s all things considered “ought” would (I think) yield the same prescriptions in all cases. But suppose, for instance, that morality (in the special sense) were overriding. Then the two varieties of “oughts” under consideration could yield different results. Here is a revealing passage from Becker: Suppose . . . we can give a reasoned justification for holding that the special moral point of view is overriding. . . . This . . . falls back on what is, in all but name, the general conception of morality. What we are to do, in each case, is what reason

224

recommends. In some cases that will be something moral (in the special sense). In other cases, it will be something prudent or self-interested or altruistic. (24)

Becker sensibly suggests that what one ought to do, all things considered, will vary across situations. In some, the general conception of morality will recommend that one do the moral thing (in the special sense); in others, it will recommend that one do the prudential or self-interested thing. If, however, there were an overarching standpoint like Reason, and if Reason pronounced morality in the special sense to be overriding, then it would be true that one plain ought to do what morality (in the special sense) recommends in any situation. Plain “ought” prescriptions could, thus, diverge from all things considered “ought” prescriptions. Hence, it seems, plain “ought” should not be conflated with Becker’s all things considered “ought.” We should guard against an important ambiguity in asking whether morality is overriding. We could, in raising this query, either be asking: From the all-things-considered point of view, ought one always to do what the special conception of morality recommends? Or, quite differently, from the point of view of Reason or plain “ought,” ought one always to do what the special conception of morality recommends? It is the latter question that is of concern in this chapter. To tie some ends together, the issue of overridingness is one about the verdicts delivered by an overarching standpoint, Reason, which is comprehensive and supreme, regarding the relative normative significance of the overall verdicts of special standpoints like morality and self-interest. It is implausible, for the reasons given, to identify the standard of Reason with that of any special standpoint like morality or with Becker’s general conception of morality. Reason is, in this sense, independent, and its being independent leaves open the possibility of whether it necessarily rules that the verdicts of morality and not those of some other special standpoint are overriding.4

A CONFUSION REVEALED

We can now expose a gross confusion in the concern that loss of deontic anchors goes hand-in-hand with loss of a standard that is overarching. Assuming (what I ultimately believe is false) that there is such a normative standard as Reason, this standard is not to be identified with the standard of deontic morality. Hence, a world devoid of deontic anchors need not, in virtue of being so deprived, be a world devoid of Reason. So, it seems, there should still be a final court of appeal in such a world that could pass

225

judgment on which, if any, of the specific or special obligations in that world reigns supreme. Nevertheless, one might still raise a concern analogous to the mistaken one that loss of deontic anchors results in loss of a standard (Reason) that is overarching. The concern, this time around, would be that loss of deontic anchors results in loss of a specific standard that is overriding – a standard judged supreme by the lights of the overarching perspective of Reason. For, it might be argued, if a normative standard is overriding – if it is judged to be so by Reason – then, presumably, it is necessarily overriding. Suppose the actual world is not determined; nor is it of an indeterministic variety that is bereft of deontic anchors. Rather, it is a kind of world (let’s assume) that can accommodate deontic anchors. Then in such a world, morality would be overriding, and hence, necessarily morality is overriding. So even if a particular world is devoid of deontic anchors, perhaps because it is a determined one, as long as there are deontic anchors in some world, moral obligation is overriding. If so, a world without deontic anchors suffers a significant loss: No sort of obligation in that world is overriding. To assess this modified concern, its vital presupposition of morality’s – the deontic moral “oughts” – being overriding requires careful scrutiny. In the next two sections I first examine arguments that appeal to cases in which it is alleged it is unreasonable to suppose that morality overrides other considerations like those of self-interest, advanced by Slote, and more recently, by Kekes, against the Overridingness Thesis, and I explain why they are inconclusive. Subsequently, granting the Overridingness Thesis presupposes an overarching standpoint, I discuss various shortcomings of this standpoint. I then argue that even if there is an overarching standpoint, that standpoint itself will probably dethrone morality by ruling against its claim to supremacy. SLOTE ON OVERRIDINGNESS

Slote construes the Overridingness Thesis as the thesis that “there cannot be any (overall) justification for doing what is morally wrong” (1983: 84). Attempting to refute the thesis, he provides examples in which an agent takes himself to be doing wrong but “feels justified in doing wrong, and is willing to stand by what he does” (86). Slote stresses that in the relevant examples, the agent believes that he must do wrong, and claims that the ‘must’ here expresses “a sense of justification – rather than (merely) of physical necessity or psychological compulsion” (86). Slote explains that,

226

in this sort of example, we have grounds to think that the agent fails to act out of psychological compulsion, believes that what he does is wrong, and also believes that he is nonmorally justified, and not merely that he has an excuse, for acting as he does. One cornerstone example of Slote’s is this: A father may deliberately mislead police about his son’s whereabouts, even knowing that the son has committed a serious crime and even while acknowledging the validity of the local system of criminal justice. He may feel he mustn’t let the police find his son, but must instead do everything in his power to help him get to a place of safety, even though he is also willing to admit that there can be no moral justification for what he is doing. Parental love can lead someone to act this way and allow him to feel justified in doing so. . . . (86)

Slote’s argument against the Overridingness Thesis, implicit in this passage, seems to amount to this: Argument 1 (1) The father believes (truly) that he ought morally not to mislead the police. (2) The father believes (truly) that, acting “out of ” parental love, he ought to mislead the police. (3) If (1) and (2), then morality does not override “all opposing considerations.” (4) Therefore, morality does not override “all opposing considerations.”

The “ought” in line (2) does not signal any moral duty or imperative. Rather, we can suppose that it expresses an obligation, or at least some prescriptive element like an obligation, associated with acting out of love, that is somewhat analogous to the moral obligation of acting out of moral duty. The “obligation” here, then, signifies an imperative stemming from some appropriate nonmoral normative standard from which the father acts when he misleads the police. Judging from her commentary on this example, Marcia Baron, it appears, would reject line (2). She suggests that in order for the father’s case to refute the Overridingness Thesis, it must be one in which the father “believes he is acting wrongly (morally speaking) and without moral justification, but that in some other sense he is justified” (1986: 559). The problem, she says, is that she cannot “make sense of the example in such a way that the devoted parent does feel justified in misleading the police, yet not morally justified” (560). So, for instance, if it were claimed that the father would later hate himself for cooperating with the police, and blame himself if his relationship with his son were hopelessly strained, and hence that with so much at stake, it would be morally permissible

227

for the father to lie, the justification for misleading the police would be a moral one. Any such rejection of the second premise is suspect, however. The father acts “out of ” or “from” parental love, and where such love is concerned, moral obligation may not matter at all.5 Slote himself remarks that “moral justification may well be the furthest thing from the mind of the parent who deceives the authorities. . . . Realistically speaking, what makes . . . [the father] feel justified and willing to stand by what he has done is not (the possibility of ) an overall moral justification in terms of the morality of traits and motives, but rather the particular danger to his son.” (1983: 92) To this, Baron rejoins: The dichotomy that Slote draws between an overall moral justification and the danger to his son is, if not false, misleading: the devoted parent may be thinking particularly about the danger to his son, but the thought may perform its motivational role against the background of other beliefs which, together with it, provide an overall moral justification. While he is not reciting it to himself, it may still be the basis of his willingness to “stand by what he has done. . . . ” (1986: 561, note omitted)

Baron’s view that it is possible for the parent to have germane background moral beliefs that “provide an overall moral justification” for misleading the police is reasonable. It is, though, no less reasonable to suppose that it is also possible that the parent lacks such background beliefs. Indeed, reflecting on our day-to-day dealings with others, it seems natural to suppose that an agent in many or most situations is not wedded to any one evaluative standpoint. One does not usually commit oneself to acting on (or despite), for instance, moral considerations or, say, those deriving from the heavy hand of tradition no matter what the situation in which one finds oneself. It seems, rather, that (at least in ordinary life) many of us reveal a flexibility in our values, taking moral considerations to be more important in some situations than, for instance, legal or prudential or aesthetic ones, but reversing our commitments in different situations, and in yet others paying no heed to morality at all but acting “out of ” or despite love or friendship. Further, it seems that in many situations in which a person acts out of love, the person simply lacks (or may lack) any dispositional beliefs that provide an overall moral justification for her action. A mother who gives up one of her own kidneys to her child who would not otherwise survive may never have reflected on the morality of kidney donation and may harbor no beliefs at all that provide an overall moral justification for her deed

228

stemming from love or care. Hence, unlike Baron I am willing to grant line (2). As I see it, the problem lies in (3). Suppose the father has a true and justified belief that he ought (morally) not to mislead the police, and that he also has a true and justified belief that he ought (acting out of love) to mislead the police.6 So grant that the father believes that he must (morally) refrain from misleading the police, but that he also must (from the perspective of love) mislead the authorities. What we have is a conflict in which obligations of different varieties – “justified obligations” we might say – cannot be jointly executed. But such a conflict does not imply that the “ought” of love overrides the “ought” of morality. The proponent of Argument 1 cannot, on pain of begging the question, assume that, in the father’s particular circumstances, love is weightier than morality, even granting that the father feels he must act out of love at the expense of thwarting what he takes to be his moral duty. There is a different interpretation of Slote’s argument on which he does not need the premise that the father’s belief is true, but only the premise that the father believes he plain ought to mislead the police. Perhaps Slote is assuming (plausibly) that if the Overridingness Thesis is true, it is necessarily true, and that if so, it would be conceptually incoherent to deny it. On this reading, Slote need merely claim that the father’s belief is not incoherent; he doesn’t need to claim it is true. This alternative interpretation of the argument distills to Argument 2 (1) (2) (3) (4) (5)

It is false that it is incoherent to deny the Overridingness Thesis. If (1), it is false that the Overridingness Thesis is necessarily true. Therefore, it is false that the Overridingness Thesis is necessarily true. If the Overridingness Thesis is true, then it is necessarily true. Therefore, it is false that the Overridingness Thesis is true.

Argument 2, however, is problematic as well. Its chief shortcoming is that the concept of incoherence, in line (1), is left unclear. On the one hand, if (1), the premise that it is false that the denial of the Overridingness Thesis is incoherent, entails that it is false that the Overridingness Thesis is true, then though (2) is true, (1) begs the question against the thesis. On the other hand, if (1) entails or is consistent with the truth of the Overridingness Thesis, then premise (2), on the assumption that (4) is true, fails. Still, a proponent of the argument might insist that the examples Slote gives are supposed to be ones in which a person in the story takes a

229

position which implies that the Overridingness Thesis is false, and where no incoherence is obvious. Then the burden of proof is supposed to be shifted to the defenders of the Overridingness Thesis. If the thesis is necessarily true, the examples contain some incoherence. But where is it? It might, to supplement these claims, be proposed that to say that it is incoherent to deny the Overridingness Thesis, amounts roughly to saying something like “the denial of the thesis entails either a contradiction or a thought that is unintelligible.” If Slote is right, no contradiction or unintelligible proposition is to be found in the case of the father. This rejoinder, though, is not conclusive either. The story of the father is certainly intelligible and appears to be free of contradiction. In particular, an opponent of the argument can grant that it is intelligible to suppose that the father (in his circumstances) believes that considerations of love override those of morality ( just as it is intelligible to suppose that some desperate mother believes that she ought morally to have saved her drowning child even though she could not have done so). However, opponents committed to the Overridingness Thesis should insist that this sort of inference Slote wishes to draw on the basis of the father’s story does contain a contradiction: The father’s case is a case where certain (personal) considerations override the recommendations of morality (Slote 1983: 86–7). The opponent’s position here would be relevantly analogous to the position of those committed to deontic principle K that one ought to do something only if one can do it, when challenged with the case of the desperate mother. Confronted with this case, an advocate of principle K (which if true is, presumably, necessarily true) will insist that one cannot, with propriety, conclude from this case that ‘ought’ does not imply ‘can,’ even though the pertinent story involving the mother and her drowning child is intelligible and contains no obvious contradiction.7 It seems, in summary, that though Argument 1 and Argument 2 are highly suggestive, they do not establish their conclusions. KEKES ON OVERRIDINGNESS

Consider, now, Kekes’s concerns about overridingness. To convince us that the Overridingness Thesis is false, Kekes presents two cases in which agents “clearly and unambiguously” conduct themselves immorally and yet have “weighty nonmoral reasons for acting the way they do” (1992: 66). Kekes explains that “unlike previous arguments (like those of Slote’s) against morality’s being overriding, these cases cannot be

230

reinterpreted as involving conflicting moral requirements. Nor can they be written off as the products of an impoverished conception of morality, since, on any reasonable view of morality, these cases would have to be regarded as involving immorality” (66). I’ll focus on Kekes’s intriguing story of the porcelain collector. Imagine a man whose ruling passion in life is a collection of porcelain figurines produced in Bohemia in the eighteenth century. There are psychological reasons why he has come to be the way he is. He had been a clumsy, insecure, ugly, lonely, and stupid child who was given a figurine on a birthday. Acquiring the relevant facts, he gradually taught himself to be a connoisseur. We encounter him as a middle-aged eccentric whose life is centered around the treasures: The collection is the focus of his emotional life; his chief preoccupation is with protecting it and adding to it; and such human contacts as he has all focus in some way or another on the collection. . . . He lives in Prague, and the events going on around him – the . . . German occupation, . . . the murder of the Jews, the communist purges . . . – impinge on his life merely as potential threats to the collection, or as opportunities to enlarge it by judicious purchases from those who need money and have the goods. He casually collaborates with whomever happens to be in power, and he is quite willing . . . to exploit the latest wave of victims. . . . From the moral point of view, the collector is despicable. His obsession has made him violate common decency; he is a spineless accomplice of great crimes, a supporter of vicious regimes, and an exploiter of innocent victims. . . . But from a nonmoral point of view . . . [what] reason could he have given for thinking that the collection was more important than decency? . . . [T]he most important consideration is an appreciation of what the figurines meant to the collector. . . . [T]hey were his life. His identity, the integrity of his personality, his attitude toward the world, the meaning and purpose of his life were inseparably connected to the collection. As some aborigines carry their souls in a box, so the collector’s soul was in the figurines. Their destruction would have meant the destruction of the psychological props of his life, and without them he would have been lost. (66–8)

This engaging case suggests the following argument against the Overridingness Thesis: Argument 3 (1) There are moral reasons for the collector not to “violate common decency.” (The collector shouldn’t (morally) be an accomplice of great crimes, or a supporter of vicious regimes, or an exploiter of innocent victims.) (2) There are weighty nonmoral reasons for the collector to have violated common decency. (As Kekes stresses, the collector’s “soul was in the figurines.”)

231

(3) The collector’s nonmoral reasons to violate common decency were more stringent than his moral reasons not to violate common decency. (4) If (1), (2), and (3) are true, then the Overridingness Thesis is false. (5) Therefore, the Overridingness Thesis is false.

The third and fourth premises require some explanation. I assume that there are moral and nonmoral reasons for acting without committing myself to any substantive conception of the nature of such reasons. Kekes’s defense of (3) unfolds in this fashion. He proposes that the collector can appeal to a “nonmoral good” – that of having a minimally acceptable life – as a reason for his immoral conduct. Elaborating, Kekes explains: If the circumstances of one’s life produce a conflict between this nonmoral good and any particular moral good, then there is no reason why the moral good should necessarily override the nonmoral one, and there is a powerful reason why the nonmoral value may take precedence. . . . The reason why for many people the nonmoral good of having a minimally acceptable life may override the moral good of being a good moral agent is that for many people, not being the stuff of which martyrs are made, the first is necessary for the second. (69)

It seems, then, that the reasoning undergirding line (3) is this: The collector has the goal of living a minimally decent life. To achieve this goal, either he can walk the morally straight path or he can “violate common decency.” Given the circumstances in which he finds himself, he couldn’t achieve this goal by doing what morality requires; indeed, if he were to follow the prescriptions of morality, he wouldn’t survive. So to live a minimally decent life, the collector would, in his circumstances, do better by acting on nonmoral reasons than he would by acting on moral ones. Hence, the collector’s pertinent nonmoral reasons are weightier than his pertinent moral ones. Finally, line (4) of the argument appears to be based on this sort of construal of the Overridingness Thesis: Necessarily, if an agent (like the collector) has moral reasons to do A in particular circumstances, and nonmoral reasons to do B in those circumstances, and the agent cannot (in those circumstances) do both A and B, the moral reasons are “stronger”; further, in virtue of the moral reasons being stronger, morality is overriding. If, however, Kekes is right that the collector’s nonmoral reasons for acting immorally are weightier than his moral reasons for conducting himself morally, then the Overridingness Thesis must be false.8 By way of evaluation, it is reasonable to think that, given his goal of living a minimally decent life, the collector would do better for himself

232

by violating common decency. Hence, I am willing to grant that, with the aim of securing this goal, acting on nonmoral reasons would be better than acting on moral ones, and in this sense, I admit that the conductor’s nonmoral reasons are “stronger” than his moral ones. Still, none of this impugns the Overridingness Thesis. To see why, let’s reemphasize that this thesis says that if an agent ought morally to do something A, at a time, and ought (nonmorally) to do something else B, at that time, but can’t then do both, then (from the overarching standpoint of Reason) the agent plain ought to do A. Assume Reason exists. Then it seems perfectly possible that Reason rules that (of all special standpoints) morality is overriding, and so, all things considered, the collector plain ought not to violate common decency, even though toward achieving his goal of living a minimally decent life, the collector would do better by shunning morality. If a standpoint is overriding, then overridingness is a feature it has in virtue of its being assessed as supreme by an overarching standpoint, and not in virtue of the guidance (if any) it provides for securing certain goals. The situation in which the collector finds himself is relevantly similar to one in which, for example, it is morally wrong for Gauguin to desert his family to paint in the South Seas, even though it is in his (let’s assume) self-interest to do so; by deserting his family, he would mature into a great painter. We can grant that, in the service of securing his goal of developing into a great painter, Gauguin would do better for himself by acting on self-interest than he would by acting morally, but this is perfectly compatible with morality’s overriding self-interest.9 Further, grant Kekes’s suggestion that a prerequisite for some persons to be morally good agents is to live minimally decent lives; and that a prerequisite for some people to live minimally decent lives is to engage in immorality. These suggestions, if correct, at most sanction the inference that a prerequisite for some people to be morally good agents is to engage in immorality, and not the inference that some nonmoral reasons are weightier than moral ones. Perhaps all Kekes wished to show is that there are various nonmoral goods, like living a minimally decent life, which are such that, with an eye toward securing such goods, some agents would do better by thwarting the requirements of morality and conducting themselves immorally. Relative to achieving such goods, nonmoral reasons may, then, well be better than moral ones. As I have argued, though, none of this undermines the Overridingness Thesis. I now want to address what I think are legitimate concerns about overridingness.

233

SHORTCOMINGS OF THE OVERARCHING STANDARD – REASON

As I have explained, the Overridingness Thesis presupposes the existence of an overarching normative standard. To advance discussion, although I am highly skeptical about its existence, I have simply granted that such a standard exists and I have called it “Reason.” Reason takes the verdicts given by other special standpoints regarding any situation where the agent needs to act (or bring about some state of affairs), compares and evaluates them in an impartial non–question-begging way, and produces an overall verdict as to what the agent plain ought to do. Assume that Reason assigns numbers expressing normative importance or stringency to these special standpoints. It does so, presumably, on the basis of comparing the verdicts of these standpoints. Call such a number, assigned to some special standpoint, the standpoint’s “priority number.” The higher the priority number, the more normatively important the standpoint. Assume, finally, that a standpoint with a higher priority number always takes “normative precedence” over one with a lower priority number. For example, if morality has a higher priority number than prudence, moral “oughts” always trump prudential ones. Now suppose, initially, that the verdicts of special standpoints are indeed comparable in such a way as to make possible an assignment of (comparable) priority numbers. Even assuming comparability, we can still inquire into precisely which feature of the verdicts is being compared. It would be pretty unhelpful to say “strength of the competing ‘oughts’” without explaining what this amounts to. One plausible alternative might be that the overarching standpoint compares different values. If, in the interests of simplicity, we assume that normative obligations entail maximization of certain values, one sort of “ought” or obligation would be more important than another just in case the value it is concerned to maximize is deemed more important (by the lights of Reason) than the sort of value the other is concerned to maximize. Suppose, for instance, that morality requires that we maximize intrinsic (moral) value, whereas self-interest requires that we maximize some other sort of value (perhaps “welfare value”). Then in a situation in which the requirements of morality and self-interest conflict, morality would “override” self-interest just in case the sort of value morality is concerned to maximize is more important than the sort of value self-interest is concerned to maximize. However, two evident shortcomings of this view of what Reason compares are that the view presupposes that different sorts of values are indeed comparable and

234

that there is an adequate account of what it is for one value to be more important than another. A second view of what is being compared is the following: Assume that each person has a certain “basic goal” – that of living a “minimally decent life.” Assume that such a goal can be fulfilled in degrees. Then one might propose that the verdicts of one standpoint (like morality) are more important than those of another (like self-interest), if always abiding by the verdicts of morality helps achieve, to a greater degree, this goal than always abiding by the verdicts of self-interest. But this “conducivenessto-achieving-certain-goals” view, as I suggested above, also has a serious problem. It seems fairly obvious that living a minimally decent life might, in various cases, best be achieved by abiding by the prescriptions of morality at some times, while at others, doing what self-interest requires. The view in consideration, then, would rule, implausibly, I believe, that neither standpoint is overriding. Further, even if one’s life (given one’s circumstances) would go best for one were one always to abide by self-interest, this seems perfectly compatible with morality’s being more normatively important than self-interest. A third possible view about what sort of thing Reason compares in arriving at an overall verdict of importance is also problematic. In this view, Reason compares the judgments delivered by people as to which standpoint they take to be more important, or care for more. The idea, then, would be that if people take morality to be more important than self-interest, then morality would override self-interest. Apart from the obvious concern that there is no agreement on the relevant judgments of importance among people, a serious deficiency of this third view is that it ignores the reasonable distinction between what people take to be overriding and what is in fact overriding. Undoubtedly, there are other accounts of what precisely Reason compares to arrive at an overall verdict regarding what an agent plain ought to do, some, perhaps, more promising than the ones canvassed. My objective, in this section, has simply been to point out the need to defend such an account if we are to make sense of overridingness at all.

ON REASON’S DETHRONING MORALITY

Suppose the hurdle of what Reason compares has been overcome. Let’s finally revert to the issue of whether morality is indeed overriding by dwelling on the following possibilities. Again, in the interests of furthering

235

discussion, I take as unproblematic the existence of Reason. We cannot, of course, assume that some special standpoint is in fact overriding in the absence of any verdict delivered by Reason. It is, then, quite possible that Reason will weigh, as equally important, two (or more) special standpoints. In light of this possibility, I propose these “metamoral” principles as elements of Reason: Reason1: It is (as of any time) plain obligatory for an agent to do the act required by the special standpoint whose priority number is higher than that of alternative special standpoints. Reason2: It is (as of any time) plain right for an agent to do the act required by the special standpoint whose priority number is no lower than that of alternative special standpoints. Let’s assume, first, that the special standpoints (or their verdicts) to be assessed by Reason are in fact comparable with respect to their normative importance. In this case, Reason (we assume) will assign appropriate priority numbers to the various special standpoints. Then there are three salient possibilities: (a) the priority number of morality is higher than that of any other special standpoint; (b) the priority number of morality is lower than that of some other standpoint whose number is highest; and (c) the priority number of morality and that of some other standpoint or standpoints tie for first place; they are equal, but higher than the priority numbers of all other special standpoints. If (a) were correct, Reason1 would rule that morality is overriding, and if (b) were correct, it would rule that some other standpoint is overriding. Finally, if (c) were to win the day, Reason2 would deliver the verdict that morality is not overriding. Independently of what Reason recommends, I don’t see any plausible way of establishing which of these three options, if any, is correct. Assume, second, that the special standpoints (or their verdicts) which are to be evaluated by Reason are not comparable. (Or at least assume that the standpoint of morality (or its verdicts) is not comparable to at least one other special standpoint (or its verdicts)). In this case, option (d), Reason2 would once again rule that morality is not overriding, as, assuming incomparability, each standpoint is such that its priority number is no lower than that of any alternative standpoint. Now I think (d) is much more plausible than (a), (b), or (c), for the simple reason that it is much more plausible to suppose that the special standpoints (or their verdicts) are incomparable than it is to suppose that they are comparable. Minimally, I believe it is much more plausible to suppose that the standpoint of

236

morality is incomparable with at least one other special standpoint than it is to suppose that the standpoint of morality is comparable with all others. And here I rely for support, first, on the well-entrenched skepticism regarding the prospects of defending an account of those features in virtue of which one standpoint is superior (in terms of normative importance) to another, and second, simply on intuitive judgments. To use one of Slote’s examples again, acting “out of ” love, the father refuses to disclose the whereabouts of his guilty offspring to the authorities, although he concedes that he morally ought to. It is, I believe, implausible to suppose that the dictates of morality and those of love are comparable in any fashion that would bear fruitfully on the issue of overridingness; we have, here, two strikingly different “value systems.” (Analogous things can be said about morality and self-interest.) Similarly, the aesthetic standpoint may require Gauguin to act in a manner proscribed by morality, but I am hard pressed to see how these two fundamentally different standpoints are comparable when it comes to assessing normative importance. SOME INTERIM CONCLUSIONS

We can now draw some interim conclusions. We started with the concern that loss of deontic anchors would entail loss of an overarching normative standard that passes judgment on the verdicts of special normative standards. As the overarching standard, even if there is one, is not identical to that of the special standard of deontic morality, this concern is mistaken. This mistaken concern gave way to another: Loss of deontic anchors would lead to loss of a normative standard – that of deontic morality – deemed overriding by the lights of an overarching standard (Reason). The assumption here is that Reason would rule that the special standard of deontic morality is overriding, and hence that, if a world is devoid of deontic anchors, no special normative standard in that world would be overriding. But this concern is misguided as well since the standard of Reason itself (assuming that such a standard exists), it seems, would dethrone morality. There is, however, a kernel of truth in the concern that lies at the heart of the view that with no overriding or overarching standard, something of significance would be lost. This kernel of truth, I now argue, is associated not with loss of deontic anchors per se but with loss of an overarching standard, or its being the case that the deontic “ought” is not overriding. There is, I believe, no such standard as Reason. More cautiously, at least this much seems true: Determined worlds will be devoid of deontic anchors and “anchors of Reason.” For presumably, if one plain ought to

237

do something, then both one can do it and one can refrain from doing it. But then, given a line of reasoning analogous to the line of reasoning that shows that determined worlds are devoid of deontic anchors, we should be able to show that such worlds have no anchors of reason either: Nothing in these worlds is plain obligatory, plain right, or plain wrong. If there is no overarching normative standard, it follows that no special normative standard is overriding. Now for the kernel (or kernels) of truth. What would be the significance of any normative standard’s failing to be overriding? One of the most salient would be that we would indeed be saddled with a kind of skepticism about normativity. To wit, there would be no ultimate standard to which one could appeal to settle what one plain ought to do in conflict situations. Practical reason would be fragmented. What of morality’s failing to be overriding? What follows? At least these things.10 First, as Slote and Kekes believe, there could be occasions on which there is overall justification – an “all things considered” justification, for example – for doing what is morally wrong. Wrongdoing, on these occasions, would not be contrary to reason. Second, a variety of moral criticisms lose some of their bite. To illustrate, suppose various precepts of a legal system fail to coincide with morality’s dictates. We could still criticize these precepts from the moral point of view, but these criticisms, I believe, would be less potent than they would be were morality the final court of appeal. Indeed, it appears that one reason why we care deeply about convergence between moral and legal dictates rests on the implicit presupposition of the supremacy of morality. Whereas it is not out of the ordinary to criticize a legal system or theory for sanctioning instances of immorality, it is far less common to fault an otherwise plausible moral theory on the grounds that some of its prescriptions (in conjunction with relevant facts) conflict with those of some legal system. This asymmetry in assessments seems essentially tied to the supposition that morality is supreme. As for individuals, one can imagine an agent reasoning in this manner: “I realize that in executing my legal obligation, I’ve thwarted my moral one, but why should this be such a bad thing? After all, I care more about the law than I do about morality and it isn’t that the moral ‘ought’ is more normatively important than the legal one. That an act is morally wrong is not a definitive reason against doing it if morality is not supreme.” Third, there are varieties of blameworthiness. For instance, one can be legally blameworthy for violating a certain law even if one is not morally

238

to blame for doing so, or one can be intellectually blameworthy for paying insufficient attention to an important mathematical proof even though one is not morally nor legally to blame for doing so.11 It is sensible to suppose that of the different varieties of blameworthiness, moral blameworthiness is the most “damaging.” It weighs most heavily in one’s “ledger of life” that records the normative worth, or perhaps, better, varieties of normative worth, of an agent on account of a particular action she performs (or fails to perform). But it is unclear why moral blameworthiness should be thought to be most telling against a person’s normative worth vis-`a-vis some action of hers in the absence of morality’s being supreme, unless one is willing to entertain the proposal that the stringency of blameworthiness is a function of the importance agents attribute to normative standpoints like the moral or the legal standpoint. OVERRIDINGNESS AND CULPABILITY

Suppose morality – the deontic “ought” – is not overriding, either because there is no such standard as Reason (in which case no normative “ought” would be overriding) or because Reason, if there is such an overarching standard, dethrones morality. Then it might be thought that this fact can be used as a vital plank in an argument, different and in some ways far simpler, from the one I have developed in past chapters for the skeptical conclusion that there are no deontic anchors. This is the argument to be addressed in the remainder of this section. It will be helpful, I think, to approach it in a slightly roundabout way by first sketching a case in which a particular action of a person issues from her “decisive best judgment” about what she should do but where this judgment fails to coincide with her moral judgment about what she should do. It would appear that in such a case the person is (or can be) morally blameworthy for what she does. Let’s start with the notion of a decisive best judgment. Mele explains that “A judgment made or held by an agent is a decisive best or better judgment . . . if and only if it settles in the agent’s mind the question what (from the perspective of his own desires, beliefs, etc.) it is best or better to do given his circumstances – and best or better, not just in some respect or other (e.g., financially), but simpliciter” (1995: 15). Suppose Sal does not believe that morality is overriding; she believes, in fact, that the Overridingness Thesis is false. Assume that she is aware that she ought morally in her situation to do X, but all things considered, she forms the decisive best judgment that she ought to do Z. Suppose she endorses the practical rule that, ideally, she should always

239

act in conformity with her all things considered decisive best judgments. Suppose Sal freely and intentionally Z-s. Is she morally blameworthy for Z-ing? It certainly seems so as, partly, she freely and intentionally Z-s in the absence of responsibility-subverting factors like relevant force, ignorance, or manipulation, despite the belief that it is morally wrong for her to do so. Sal’s case, however, may encounter the following sort of resistance. It might first be charged that the case presupposes the truth of this principle: Presupposition: It is possible that some agent S believe that action A is morally but not all things considered wrong. Second, Presupposition, it might additionally be charged, exposes the following serious difficulty with the case. Moral theorists like Neil Cooper (1981), R. M. Hare (1963: 161,168–9), and Nowell-Smith (1954: 306– 10) appeal to overridingness to distinguish moral judgments and principles from other varieties of normative judgments and principles. They hold, roughly, that an agent S accepts a principle P as a moral principle only if S is unwilling to allow P to be overridden by other principles. Under the reasonable assumption that accepting a principle as a moral one entails believing that it is a moral one, their view seems to be that one believes that some principle is a moral one only if one believes that it cannot be overridden by other nonmoral normative principles. (See, e.g., Cooper 1981: 98–101 and Hare 1963: 168–9.) We can restate the crux of this view that is relevant to formulating the objection to Sal’s case in this way: (P1): An agent S believes that an “ought” requirement R, one that enjoins that one ought, or ought not to, do something is a moral one only if S believes that R is overriding. Further, these theorists additionally seem to maintain that (P2): An agent S believes that an “ought’ requirement R is overriding only if S believes that whenever R conflicts with other nonmoral normative “ought” requirements like, for example, legal, prudential, or decisive best ones, S plain ought to do what R requires. (See, e.g., Cooper 1981: 98–101.) From (P1), (P2), and the assumption that if S believes that R is a moral requirement, then, given the object of this belief, S believes either that she ought morally to do something or that she ought morally not to do something, we derive:

240

(P3): An agent S believes that S ought morally not to do something (or that it is morally wrong for S to do something) only if S believes that whenever this “ought” requirement conflicts with other nonmoral normative “ought” requirements, S plain ought to act in accordance with this “ought” requirement. In the example involving Sal, though, I claimed that although Sal believes that she ought morally not to do Z, she believes that all things considered, she ought to do Z. But then it is sensible to assume, and let’s so assume, that Sal believes that moral “ought” prescriptions or proscriptions are not (P2)-overriding. So though she believes that she morally ought not to do Z, she does not believe that she plain ought not to do Z; in other words, though she believes that she morally ought not to do Z, she does not believe that the requirement that she ought not to do Z is (P2)-overriding. Now the objection to Sal’s case should be evident: If (P3) is true, the case is incoherent. This objection, though, is questionable. One way in which it can be challenged is by constructing a plausible scenario in which though someone believes that an “ought” requirement R is a moral one, she does not believe that it is overriding; she does not, that is, believe that it is what she plain ought (or ought not) to do. Let’s embellish Sal’s case. Suppose Sal believes that she ought morally in her situation to do X. Can we tell a cogent story about how it is reasonable that she can believe this consistent with not believing that she plain ought to do X ? Well, suppose Sal has reflected on the Overridingness Thesis. She reaches the ensuing conclusions. First, there are difficulties with the notion of overridingness itself. The Overridingness Thesis, she observes, presupposes the existence of an overarching standpoint whose “function” is to take the verdicts given by other special standpoints regarding any situation where the agent needs to act, compare and evaluate them in a non–question-begging way, and produce an overall verdict regarding what the agent plain ought to do. Sal is highly skeptical about the existence of any such standpoint, for among other reasons, that such a standpoint, if it did exist, would be a normative one, and, in consequence, there would be a question about the stringency of its very verdicts. Would there not have to be an even more inclusive overarching standard that gauged the verdicts of all special standpoints and that of plain “ought” as well? But then what about the stringency credentials of this even more inclusive overarching standard itself, and so on? An ugly regress seems unavoidable.12 Second, echoing what I summarized in a preceding section, Sal reasons that even

241

granting the existence of an appropriate overarching standard, Reason, there is no guarantee that it will rule that the standpoint of morality is normatively supreme; indeed, in her view, Reason itself may dethrone morality. Sal concludes that she has good reasons to believe that morality is not overriding. And if this is so, no incoherence need infect the view that though Sal believes that she morally ought not to do X, she does not believe that this “ought” requirement is (P2)-overriding. A NEW PAIR OF SKEPTICAL ARGUMENTS

Perhaps, reflecting on the failure of the last objection against Sal’s case, one might suggest that there is a direct connection between blameworthiness and overridingness that challenges the verdict that Sal is to blame for Z-ing. Consider these principles: (P4) (Blame/Wrongness): S is blameworthy for doing A only if S morally ought not to do A. (P5) (Overridingness/Wrongness): S morally ought not to (ought to) do A only if this “ought” requirement is plain overriding, where “ought” requirement, R, is plain overriding if and only if whenever R conflicts with other normative “ought” requirements, such as legal, prudential, or decisive best ones, S plain ought to do what R requires. (P4) and (P5) entail: (P6): S is blameworthy for doing A only if S morally ought not to do A (call this proscription “MO”) and moral proscription MO is plain overriding. Now one might argue in this way. Argument 4 (4.1) It is false that morality – the deontic “ought” – is plain overriding. (4.2) An agent S is blameworthy for doing A only if S morally ought not to do A (MO) and MO is plain overriding. (4.3) If 4.1 and 4.2 are true, then no agent is morally blameworthy for anything. (4.4) Therefore, no agent is morally blameworthy for anything. (Hence, Sal is not blameworthy for Z-ing.)

4.1 is simply based on the view that either there is no such standard as Reason, or if there is, Reason itself dethrones morality. 4.2 is P6, and P6 is derived from P4 and P5. This skeptical argument, however, is problematic. P4, vital in this argument to the rationale for P6, is just the part of the

242

Objective View concerning blameworthiness, but the Objective View is suspect. Despite the failure of Argument 4, one might utilize P5 in an attempt to undermine deontic anchors: Argument 5 (5.1) It is false that morality – the deontic “ought” – is plain overriding. (5.2) An agent S morally ought not to (ought to) do A only if this “ought” requirement is plain overriding (where “ought” requirement R is plain overriding if and only if whenever R conflicts with other normative “ought” requirements, like legal, prudential, or decisive best ones, S plain ought to do what R requires). (5.3) If 5.1 and 5.2 are true, then no action of any person is wrong or obligatory for that person. (5.4) Therefore, no action of any person is wrong or obligatory for that person.

Presumably, a corollary of P5 could be invoked in an argument similar to Argument 5 to derive the conclusion that no action of any person is morally right for that person. One could then draw the overall conclusion that there are no deontic anchors. Although I agree with 5.1, I think that 5.2 is questionable. 5.2 says that an “ought” requirement is a moral one only if it is overriding or normatively supreme as judged by the overarching standard of plain “ought.” Alternatively, 5.2 can be expressed in this way: The prescriptions of a normative standard qualify as moral ones only if the overarching standpoint of Reason (plain “ought”) rules that that standard is overriding. I find 5.2 questionable for the reasons that I discussed in various sections above. To reiterate some of these, 5.2 has some untenable consequences. Harking back to Sal’s musings, if there is no such overarching standard as Reason, then it appears that no normative standard can be overriding, but it is unclear how this, in turn, would establish that no normative standard qualifies as a morality. Further, even if there were an overarching standard, the standard itself may rule that morality is not overriding. Suppose that from the point of view of Reason, it turned out that the special standard now identified with self-interest were overriding. Then the special standard now identified with morality would fail to qualify as morality, something that is hard to swallow. Finally, we have seen that various considerations advanced by people like Williams and Becker that might be taken to support 5.2 are not free of their own difficulties. I conclude that Argument 5 fails as well. We might well, though, alert ourselves to the following: If 5.2 (or something like it) can successfully be defended, then the deontic “ought’s” failing

243

to be overriding will have some pretty drastic implications for whether deontic anchors are possible in the first place. In this and the preceding three chapters, I have considered concerns having to do with declared costs or detrimental consequences of the possibility of being deprived of deontic anchors. I have argued that a number (though not all) of these concerns are well founded. In the next (and final) chapter, I propose what I believe is one additional legitimate cost.

244

13 Concluding Remarks

Being deprived of deontic anchors does exact serious costs. I have discussed some of these in the past few chapters. One is that, without deontic anchors, there would be no grounds for reactive attitudes such as resentment and forgiveness. A second is that, in the absence of deontic anchors, deontic appraisals of rightness, wrongness, and obligatoriness would lack propriety. A third cost is affiliated with concerns of overridingness. If the moral “ought” is overriding, and its verdict is definitive in settling what one plain ought to do in situations in which different sorts of obligation conflict, we will be dispossessed of this ultimate standard of appeal if there are no deontic anchors. One might be, as I am, skeptical about morality’s being overriding. Determinism, though, threatens overridingness itself in another way. As I explained, the Overridingness Thesis presupposes the existence of an overarching standard, “Reason,” that passes judgment on the relative normative stringency of special standards of obligation such as moral or prudential obligation. The “plain ought” dictates of Reason, it seems, just like those of the deontic moral “ought,” require alternative possibilities. But if this is so and determinism effaces alternative possibilities, determinism would undermine plain “ought” prescriptions as well. In this chapter, I first single out an additional cost of being deprived of deontic anchors: In a world devoid of such anchors, we are not, in a way to be explained, “sources” of deontic morality. I then conclude with some remarks on the view (proposed in Part Two) that certain robust forms of libertarianism can accommodate deontic morality.

245

ON NOT BEING SOURCES OF DEONTIC MORALITY

A price to be paid if there are no deontic anchors is that we would not be sources of deontic morality. Assume that determinism is true and, thus, that there are no deontic anchors. Then whatever decisions we arrive at or actions we perform make no “deontic difference” whatsoever in the world. Many of us value being in control of the following aspect of our moral lives. We want it to be the case that it is in our power to avoid doing wrong, to discharge our obligations, and even to supererogate. When this sort of power is more or less wrenched from our hands as, for example, would be the case if faced with dire consequences for failing to comply with the evil wishes of a fanatic, we feel that something of great value has been lost. The situation with no deontic anchors, as far as loss of this sort of control is concerned, is far worse. In the scenario in which we are threatened but in which deontic anchors are possible, we can still retain a measure – even though perhaps only a negligible one – of control. Our decisions and subsequent actions do make a difference to how things go, and they can make a deontic difference. We could supererogate, perhaps by turning hero, and refuse to do wrong. But with the truth of determinism our decisions and actions make no deontic difference at all to how things unfold. It is true that had we not made the decisions that we did, things would have gone differently if determinism were true. Still, and this is what is worthy of emphasis, things would not have gone “deontically” differently: No matter what decisions we made or actions we performed, they wouldn’t, under the circumstances, be right, or obligatory, or wrong. Let’s say that one has deontic control over an action when one has the power to see to it that that action is either right, wrong, or obligatory. One has the power to see to it that some action is, for example, wrong, if one can intentionally do wrong in performing it. (An analogous condition holds with respect to the power to see to it that an action is right or obligatory.) We, or at least many of us, value having deontic control over our actions; we value being the sorts of agents who can make a deontic difference in the way things go. That many of us value being such agents is not, I think, controversial. Why exactly, though, we value being agents of this sort is not transparent. Perhaps the value is wholly derivative: We value having dual control over our actions because we believe that if we have such control, the action is really “up to us,” and the value of deontic control is entirely derivative from the value for us of dual control. It is not something over and above. Or perhaps those who value deontic control value it intrinsically; they think that it is good in itself to have such control. Or alternatively,

246

maybe those who value deontic control value in it instrumentally; they believe that exercising such control is a means to molding or crafting our moral characters or our “moral life.” Determinism usurps deontic control and hence it usurps something of value to many of us. Significant losses, in conclusion, cannot be avoided if we are deprived of deontic anchors.

CLOSING COMMENTS

Let’s summarize the crux of the results of the preceding chapters: First, if determinism is true, then there are no deontic anchors. The verdict regarding the truth of determinism is still not in, and for all we know, our world is a deterministic one. Hence, for all we know, our world is devoid of deontic anchors. Determinists may be quick to interject that what I have shown, at best, is that in a determined world, there would be no deontic anchors if deontic moral principles such as K, CK, OW, and KW are true. But these principles, they might urge, ought to be rejected. However, their rejection comes at a high price. These principles, I believe, stand or fall together: Any normative moral theory that validates K should also validate the others, and if any such theory is true, then determinism will threaten deontic anchors. Second, being deprived of deontic anchors wrests high costs. Although I have been at pains to stress that other modes of moral appraisal such as those of intrinsic, extrinsic, or overall goodness are not obviously imperiled by determinism – the various modes of moral appraisal do not stand or fall together – being without deontic anchors leaves us impoverished in sundry and significant ways. Finally, I have argued that the troubling dilemma with which we began can be skirted: If either determinism or indeterminism is true, then there are no deontic anchors. Either determinism or indeterminism is true; and so there are no deontic anchors. I have proposed that indeterministic worlds of a certain variety can accommodate deontic anchors. I confess that evasion of the dilemma in the direction in which I have suggested leaves an acrid aftertaste. For I am still of the bend of mind that appraisability is compatible with determinism but incompatible with the sort of indeterminism hospitable to deontic anchors. There is a sense, then, in which I cannot both have my cake and eat it too. I realize that the thesis that some versions of causal indeterminism are consistent with deontic morality but not with moral responsibility

247

will strike many as being intuitively problematic. For it would seem that if some version of causal indeterminism were compatible with deontic morality, it would thereby be compatible with moral responsibility. Let me conclude with some remarks on this thesis. Clarke has forcefully argued that versions of action-centered (robust) modest libertarianism, in which the very decisions or intentions of agents are underdetermined, do not succumb to the luck objection, at least when that objection is meant to question whether such theories can accommodate moral responsibility. Clarke contrasts a case in which the indeterminism is in the production of a (basic) action with one in which there is indeterminism between a basic action and an intended result that is not itself an action. In the first case, Peg decides to keep a promise, and her decision to keep the promise is nondeterministically caused by her having certain reasons for doing so, including her making the moral judgment that she ought to keep the promise. Prior to making this decision, Peg recognizes and is tempted to act on reasons of self-interest to break her promise, and it is true that, until she decided to keep her promise, there was a chance that here deliberative process would have terminated in a decision not to keep the promise. Had the decision not to do the moral thing been made by Peg, that decision would have been nondeterministically caused by here having reasons of self-interest. In the second case, you attempt to hit a target by throwing a ball at it, and you succeed. The ball’s striking the target is not itself an action, and as Clarke stresses, you exercise control over this event only by way of your prior action of throwing the ball. Reflecting on these two cases, Clarke writes: Now suppose that, due to certain properties of the ball and the wind, the process between your releasing the ball and its striking the target is indeterministic. Indeterminism located here inhibits your succeeding at bringing about a nonactive result that you were (freely, we may suppose) trying to bring about, and for this reason it clearly does diminish the control that you have over the result – it constitutes control-diminishing luck. But the indeterminism in Peg’s case – and the indeterminism required by the sort of event-causal libertarian view at issue here (i.e., the relevant sort of robust modest libertarianism) – is located differently. It is located not between an action and some intended result that is not itself an action, but rather in the direct causation of the decision, which is itself an action. The control that an agent exercises in making a decision does not (typically) derive at all from any prior attempt on her part to bring about that decision. In the ball-throwing sort of case, the explanation of why the indeterminism constitutes control-diminishing luck is that it inhibits the agent’s bringing about a nonactive

248

result that she is actively trying to bring about. But that explanation is not available in the second kind of case. (in press: sec. 2.2)

One way to question the judgment of some, that Peg is morally responsible for making the decision to keep her promise, is by invoking the consideration, advanced in previous chapters, that the relevant sort of action-centered libertarianism cannot accommodate the kind of selfexpression required for moral responsibility. But others might propose a different path of resistance, and this path has a bearing on whether the kind of robust libertarianism at issue can accommodate deontic anchors. One might argue that, in the second case, your control over the ball’s hitting the target is not diminished in virtue of the fact that you intend a certain result but that that result may not obtain. Rather, it is diminished because, roughly, there are other factors (such as wind velocity or wind direction), over which you lack control, that influence whether the ball will hit the target. Reverting to Peg’s case, suppose her decision to keep the promise is nondeterministically caused by, among other things, a prior better judgment of hers. If decisions themselves are undetermined, then it seems that whether Peg’s better judgment to keep the promise will result in the corresponding decision to keep the promise is also influenced by factors over which Peg has no control. The factor isn’t anything quite analogous to the wind in the first case; rather, it is just the indeterminacy in the production of her decision. So just as you can’t exert control over its being the case that the wind influences the ball’s trajectory, so, it might be claimed, Peg can’t exert control over its being the case that the indeterminacy (that is a feature of the causal pathway, or elements of that pathway, that culminates in the decision to keep the promise) influences whether Peg intends or makes a decision in accordance with her better judgment. And this is so even though Peg’s decision is a mental action that is nondeterministically caused by prior actional elements. To this it might be replied that the wind is not part of the causal pathway’s segment between your throwing the ball and the ball’s hitting the target, in the way in which the indeterminacy is part of the causal pathway’s segment between Peg’s forming the better judgment to keep the promise and Peg’s deciding to keep the promise; the wind is something “external” to the ball-throwing pathway, whereas the indeterminacy is something “internal” to Peg’s pertinent actional pathway. But this, it might be rejoined, seems strictly irrelevant. The core of the issue is that just as there is a factor (the wind) over which you have no causal control, that influences the outcome of a prior action of yours (your throwing the ball),

249

so there is a factor (the indeterminacy) over which Peg has no causal control, that influences the outcome or effect of a prior activity of hers (Peg’s forming a better judgment to keep the promise). So Peg’s decision to keep the promise, it might be claimed, is luck-infused. In addition, it might be proposed that this sort of luck is incompatible not only with moral responsibility but also with deontic morality. I believe that at the heart of the present objection to the position regarding Peg’s being responsible for her decision to keep the promise is a concern of ultimate control. Peg is not, in any obvious way, in (ultimate) control of the indeterminacy that is part of the causal process that results in her decision. In virtue of lacking such control, some free-will theorists may well want to insist that Peg is not morally responsible for her decision, irrespective of that decision’s having been appropriately caused by the having of certain reasons by Peg. Similarly, insisting on an analogous requirement of ultimate control for deontic morality, proponents of the present objection can be construed as advancing the view that, as Peg lacks ultimate control over, say, her keeping of the promise, her keeping of the promise cannot be morally right, or wrong, or obligatory. If this construal of the objectors’ concerns about indeterministic choice, that ultimate control is discerned to be the culprit in connection with both moral responsibility and deontic morality is correct, then it is open to others to propose the alternative that proximal and not ultimate control is required for deontic morality. It should be evident that I side with these others. These reflections on ultimate control naturally prompt the question whether there is an account of freedom that accommodates both moral responsibility and deontic morality. I advance the following conjecture. Let’s say that an agent has genuine alternatives at a time just in case she could at that time have performed two or more alternatives, consistent with the past and the natural laws remaining fixed. An agent has dual directional control if and only if, just prior to the moment at which she performs some action, mental or otherwise, she had genuine alternatives, and she determines which of these she performs. Assume that deciding to keep the promise and deciding to break the promise are genuine alternatives for Peg. Suppose Peg determines which of these alternatives she performs by exercising dual directional control. If she decides to keep the promise, she brings about this decision and not the alternative. Assume, too, that no matter which decision she makes, she exercises proximal control in bringing about the decision. Suppose that she does decide to keep the promise. Then it seems that the way would be open to support the dual claims that Peg is morally responsible for

250

the decision to keep the promise and that her making the decision can instantiate some primary deontic property. I have my strong suspicions, though, that any libertarian theory that allows for dual directional control (as well as proximal control) will have to appeal to agent causation. In closing, I hope that my attempt to come to grips with the new dilemma concerning freedom and deontic anchors has at least succeeded in mapping out and exposing terrain yet to be fully explored, and that it will impel others to try their hand at resolving or dissolving this intriguing puzzle.

251

Notes

CHAPTER 1

1. Among others, Ginet 1990, Fischer 1994, and van Inwagen 1983 have argued for the view that if determinism is true, then there is exactly one physically possible universe. 2. For a spirited defense of this view, see Kane 1985, 1996a. See also McKenna 1998. 3. Schlossberger 1992: 4–7, 79, 101, 112, 117–18 has argued at length for the view that moral responsibility is not essentially a function of the agent’s being free or acting freely. For responses to Schlossberger, see Zimmerman 1995: sec. 2.4. 4. See, e.g., Fischer 1982, 1994, Fischer and Ravizza 1998, and Haji 1998a. 5. Partisans of this view include Smith 1983, 1991, Widerker 1991, 2000, Oakley and Cocking 1994, Fields 1994, Copp 1997a, and Wallace 1994. 6. For some recent literature on alternative possibilities and moral responsibility, see Fischer 1999a, 2000, Haji 1999b, Kane 2000, Mele and Robb 1998, Pereboom 2000, Widerker 2000, and Zagzebski 2000. 7. See, e.g., Fischer 1999a, Henderson 1966, Kekes 1997, and 1998, Mann 1983, Richman 1983, Shaw 1965, Sinnott-Armstrong 1985, 1988, Saka 2000, Stocker 1971, Widerker 1991, and Yaffe 1999. 8. I first gave this argument in Haji 1994. CHAPTER 2

1. On plain ought, see, for example, Copp 1997b and Feldman 1986: 212–5. 2. See Zimmerman 1996: ch. 5 for a recent analysis according to which prima facie obligation is to be understood in terms of absolute obligation. 3. As Zimmerman notes, Feldman makes very similar remarks as well. See Feldman 1986: 24–5. 4. Bruce Waller claims that moral acts like the rescue of children or friends “require the right intent, the proper motive” (1998: 100). To convince us that many humans and nonhumans can form and act upon intentions that they cannot “conceptually order and explain” (100), he says that talk of “intentions is another rich source of verbal confusion. . . . To act morally, I must genuinely intend to rescue my child

252

5. 6. 7. 8.

(rather than intend to push the child from a ledge or intend to preserve the lottery ticket clutched in its grubby little fist)” (100). This passage suggests that Waller is committed to the view that one ought morally to do something, A, or it is right for one to do A, only if one can do A intentionally, or one has the intention to do A. But his example is controversial. Suppose, intending to kill the child, I push the child from the ledge. Unbeknownst to me, there is a safety net under the ledge that breaks the child’s fall, and the child is saved. Had I not pushed the child off the ledge when I did, the child would have been killed by a falling rock. Arguably, allowing for modest consequentialist considerations, I have done (unintentional) right in pushing the child off the ledge despite not having the right motive. It is much more plausible to suppose that one acts from virtue or duty in doing A only if one is aware that one is doing A and one does A intentionally. For some discussion of acting from virtue, see Chapter 11. Recent discussion against this version of internalism is to be found in Audi 1997: esp. ch. 10 and Mele 1996b. McCloskey 1973 also seems to believe that ignorance of facts can be obligation subversive. Personal correspondence. See also Vihvelin 2000: 141–3. Personal correspondence. See also Vihvelin 2000: 142.

CHAPTER 3

1. John Locke described something very much like a Frankfurt-type example in Locke [1690] 1984: bk. 2, ch. 21, sec. 8–11. Many accept Frankfurt’s verdict that Frankfurt-type cases do indeed render highly plausible the view that agents can be morally responsible for what they do even though they could not have done otherwise. See, for example, Fischer 1994, Haji 1999b, Hunt 1996, Mele and Robb 1998, Pereboom 2000, Ravizza 1994, Stump 1996a, Zimmerman 1988: sec. 4.1, and Zagzebski 2000. Many others disagree with Frankfurt’s verdict. See, for instance, Ekstrom 1998a, Ginet 1996, Lamb 1993, Kane 1996a, b, 2000, McKenna 1997, 1998, Otsuka 1998, Rowe 1991: ch. 5, Vihvelin 1998, Widerker 1995a, b, 2000, and Wyma 1997. Critical discussion of Mele and Robb’s particular Frankfurttype case is to be found, among other places, in Kane 2000, Pereboom 2000, 2001, Widerker 2000, and Zagzebski 2000. A reply to these (and other) critics is contained in Mele and Robb’s manuscript, “BBs, Magnets, and Seesaws: The Metaphysics of Frankfurt-Style Cases.” Pereboom’s construction of a Frankfurt-type example is particularly engaging (see Pereboom 2000: 128–31). In his case, the counterfactual intervener makes the agent (in a nondeterministic world) decide to cheat on his tax deduction if he doesn’t (nondeterministically) do so on his own. The agent’s psychology is such that a causally necessary (but not causally sufficient) condition for the agent to decide not to cheat is that a moral reason occur to him with a certain force. Such a reason’s occurring to the agent is the cue for intervention. The ability to identify a causally necessary condition for nondetermined choice prior to the choice is, in principle, possible in a nondeterministic world. 2. Equally, one could consider a world in which S has no alternatives to anything that S does in that world. S would, in such a world, be in a “global Frankfurttype situation.” Global Frankfurt-type cases have been discussed, among others, by

253

3. 4. 5.

6. 7.

Double 1989, Fischer and Ravizza 1998: 230–7, Kane 1996a: 142–3, 1996b: 84–8, Mele 1995: 141, Schoeman 1978, and D. Zimmerman’s paper, “Doing and Time: An Essay on Autonomous Agents and Personal Histories.” David Copp and Jarett Leplin independently raised this sort of concern. On alternatives, see Bergstrom 1971, 1976, and Aqvist 1969. Some, however, might take issue with this last contention. Susan Wolf, for example, has argued that right (or good) actions and wrong (or bad) actions are asymmetric with respect to the requirement of alternative possibilities for moral responsibility: Whereas one can be responsible for a right (good) action that one could not have avoided performing, one cannot be morally responsible for a wrong (bad) action that one could not have avoided performing (Wolf 1980, 1990). As I have argued elsewhere (Haji 1998a: 43–4, 65–8), though, I am unpersuaded by Wolf ’s asymmetry thesis. Additional criticisms of this thesis can also be found in Fischer and Ravizza 1992 and Mele 1995: 162–4. See Feldman 1986: esp. ch. 2. This sort of theory is also defended by Zimmerman (1996). Feldman 1986: 37. Suppose that for every world accessible to s at t, there is a better one then accessible to s as well. To handle this possibility, Feldman advances the following as the official version of (MO): as of t, s ought to see to the occurrence of p if and only if p occurs in some world accessible to s at t, and it is not the case that not-p occurs in any accessible world as good as or better than that one (1986: 38).

CHAPTER 4

1. For a recent, vigorous defense of K, see Zimmerman 1996: ch. 3. 2. Feldman’s (1986) and Zimmerman’s (1996) theories have K as a theorem. 3. Copp (1997a) rejects (ii). I have responded to Copp in Haji 1998a: 54–62, and in Haji 1998b. See, in addition, the appendix in Chapter 10 of this work. 4. It should be cautioned that, in the article cited, it isn’t clear that Widerker in fact accepts PAP rather than rejecting it (and hence K also). 5. Kane seems to take this position in Kane 1996a. 6. This claim may be disputed. For although we say things like “Moe did what was wrong,” and this appraisal seems to be act focused, we also say things like “Moe was wrong to do what he did,” and this appraisal seems to be agent focused. Still, there is a significant difference in the two. We readily agree that it can be the case that though Moe did wrong (or right), he was not blameworthy (or praiseworthy) for what he did. It is less commonly acknowledged that an agent can be blameworthy (or praiseworthy) for doing something that is not wrong (or right). 7. Such cases are discussed in Henderson 1966: 106, Richman 1983: 85, Robinson 1971: 252, Sinnott-Armstrong 1988: 116–17, and Stocker 1971: 314. 8. Zimmerman discusses a case relevantly like Lottie’s in his 1996: 95–6 and 1987. 9. See Zimmerman 1987: 199–200 and 1996: 96–7. On time relativization of ought statements, see, e.g., Feldman 1986, Smith 1976, and Zimmerman 1996. 10. One may object that the whole notion of remote obligation is imperiled by the following: If Lottie does do wrong, when exactly does she do wrong? At every

254

time when she has the remote obligation (to refrain from killing the baby later), she does not then kill her baby then, nor does she then kill her baby later. So at all these times, she does no wrong. Indeed, at all these times, it appears that she satisfies her remote obligation. Further, when she does kill on, say, Sunday, she doesn’t then do wrong since as of then (on the view being proposed) she no longer has an obligation to refrain from killing then. So there is no time at which she violates an obligation not to kill at the later time; and so there is no time, contrary to what has been assumed, at which Lottie commits the (relevant) wrong. The right response, I believe, is again suggested by Michael Zimmerman. He proposes that we distinguish between the time at which an agent does wrong (or goes wrong) and the time at which the wrong occurs (1996: 100–5). Without going into technical details, the wrong that is done (killing the baby) occurs on Sunday, even though by then it is no longer wrong for Lottie to refrain from killing the baby. Although Lottie doesn’t do wrong with respect to the killing on Sunday, the time at which she does do wrong (or goes wrong) is the time at which she takes the pill and thereby ensures that she cannot refrain from killing on Sunday. So if, on the one hand, in asking, “when does Lottie do wrong?” one is asking, “when does the wrong occur?” the answer is that it occurs on Sunday. If, on the other had, one is asking, “when does Lottie do (or go) wrong?” the answer is that she goes wrong at the time she takes the pill. There is no incoherence, consequently, in endorsing these views: (i) Lottie ought, up to the time of taking the pill, to refrain from killing the baby later; (ii) there are times, for example, times prior to taking the pill, at which Lottie does satisfy her remote obligation to refrain from killing later; (iii) there is a time at which Lottie goes wrong or does relevant wrong. This is the time at which she takes the pill; and (iv) there is a time at which the pertinent wrong (killing the baby) occurs. 11. I don’t intend to suggest that Copp is committed to anything like KC1. 12. Pereboom (in correspondence) has advanced two further considerations against OW. First, he suggests that there is an aspect of the concept of being wrong that is relevantly similar to the concept of being illegal. He explains that, apparently, it would be illegal for me to speed even if I’ve been brainwashed into doing so and, as a result, could not do otherwise, and analogously, it seems that it would be morally wrong for me to exceed the speed limit even though I’ve been brainwashed and so couldn’t but exceed the limit. Here again, though, there is room to maneuver. If a jurisdiction sanctions strict liability laws, then it can be the case that, in that jurisdiction, one could do legal wrong even if one lacked the ability to conform to the requirements of these laws. Further, there may be good moral justification for the laws in question. But it is doubtful in the extreme whether there are analogous “strict liability” moral laws, justified on moral grounds, according to which it can be wrong for one to do something that one cannot avoid doing. A proponent of K should insist that, if one cannot do something, then it is false that one ought to do that thing; obligation presupposes control. But if this is conceded, then it is unclear why one should accept the view that it can be wrong for one to do something that is unavoidable for one, given that wrongness, just like obligation, presupposes control. Second, Pereboom proposes that, perhaps, what we mean when we say that it is wrong for S to do A is that it would be wrong for S to do A. But then ‘wrong’

255

wouldn’t imply ‘can,’ since all that would have to be the case for it to be wrong for S to do A is that in the nearest possible world in which S does A, it is wrong for S to do A. One problem with this proposal, however, is that if ‘wrong’ is understood in the manner suggested, ‘ought,’ too, it would seem, should be understood analogously. But if ‘ought’ were so understood, one would be committed to denying K. A second problem has to do with reasons for accepting the proposed analysis or meaning equivalence of ‘wrong.’ It could be that it is the doctrine of “actualism” that impels one to the suggested equivalence. Actualism is, roughly, the view that S ought to do A if and only if what would happen if S did A is deontically superior to what would happen if S performed any alternative. Actualism is to be contrasted with “possibilism,” which is the doctrine that, roughly, S ought to do A if and only if what could happen if S did A is deontically superior to what could happen if S performed any alternative. Whether actualism is true, though, is highly controversial. For relevant literature on actualism and possibilism, see, for example, Feldman 1986, Goble 1993, Greenspan 1978, Jackson and Pargetter 1986, Smith 1976, 1978, Sobel 1976, Thomason 1981, and Zimmerman 1996. 13. Not surprisingly, some people have attempted to reject not OW but K by appealing to genuine moral dilemmas. Roughly, the argument here is as follows: Suppose there are genuine moral dilemmas. Then there can be situations in which (ignoring temporal indices) an agent, S, ought to do A and ought to do B but can’t do both A and B, i.e., the agent can’t do the conjunctive act (A and B). As S ought to do A and ought to do B, the agglomeration principle: AGP: If O(A) and O(B), then O(A and B),

implies that S ought to do the conjunctive act (A and B). If K is true, then S can do this conjunctive act. But S cannot, in S’s dilemmatic situation, do this conjunctive act. Hence, it is concluded that K must be false. (This species of argument is carefully laid out by Gowans in Gowans 1987: 20–2 and by McConnell in Gowans 1987: 55–7. It is invoked by Lemmon (in Gowans 1987) to discard K.) Some people have questioned the agglomeration principle (AGP). (See, e.g., Hudson 1986: 16–18, van Fraassen 1987: 146, and Williams 1987: 132–3.) Even if one accepts this principle, the major premise of this line of reasoning against K – that there are genuine moral dilemmas – is highly controversial. For arguments in favor of such moral dilemmas, see, for example, Marcus 1987, Sinnott-Armstrong 1988, Stocker 1990, and Williams 1987. For arguments against such dilemmas, see, for instance, Conee 1987, Donagan 1987, Feldman 1986: sec. 9.1, McConnell 1987, and Zimmerman 1996: 217–25. CHAPTER 5

1. See Fischer 1994: 111–20 on hard facts. 2. To avoid defending (3) by invoking the following invalid agglomeration principle: If Ns,t(P) and Ns,t(Q), then Ns,t(P & Q) (see Gallois 1977, McKay and Johnson 1996, and Slote 1982), one can simply note that Nsid,t(F and L) holds in the particular sort of case under consideration, and then seek an explanation of why this is so. For thoughts on this issue, see O’Connor 2000: 10–15.

256

3. This sort of argument is articulated and defended by van Inwagen 1983 (see, for example, 68–70). Similar versions are to be found in Fischer 1994, Ginet 1983 and 1990, Lamb 1977, Warfield 1996, and Wiggens 1973. Nelson Pike 1965 offers a theological version of the argument that is discussed by (among others) Fischer 1989, Hoffman and Rozenkrantz 1980, and Vihvelin 1998. 4. Versions of this reply have been offered by many, including Fischer 1983, Lewis 1981, Saunders 1968, and Vihvelin 1988, 1991. There is an informative discussion of this reply in Kane 1996a: 44–52. 5. The plausibility of denying the view that we are powerless over the natural laws may depend partly on the nature of the laws and the nature of causation. For a defense of compatibilism by appeal to a nonnecessitarian view of laws, see Berofsky 1987: esp. chs. 8 and 9, and Swartz 1985: ch. 10. Ekstrom 1998b argues that compatibilists have reason to reject counterfactual approaches to analyzing causation or, conversely, incompatibilists have reason to endorse some counterfactual theory of causation. Her view is that the counterfactual account of causation gives the incompatibilist a way to block successfully the local miracle compatibilist’s reply to the Consequence Argument. 6. For developments of this sort of reply, see, e.g., Ekstrom 2000: 49–50 and O’Connor 2000: 17. Other replies include the following. Carl Ginet has proposed that local miracle compatibilism is questionable because it is committed to the existence of cases in which it is not problematic both that an agent is able to do something and her doing it would require violation of a natural law just prior to the time at which the action was performed. But there are, Ginet suggests, many cases in which it appears that an agent cannot do something precisely because the agent’s doing that thing would require violation of a natural law just prior to her doing that thing (Ginet 1990: 113–14.) Fischer has questioned whether a merely counterfactual power over the past or the laws gives us grounds for rejecting the Consequence Argument. He suggests that there are intuitive grounds for inferring from “if S were to do x, the laws (past) would be different” to “S cannot do x” even if we stipulate that the sense of ‘ability’ in question is the counterfactual one (Fischer 1994: 67, 78). 7. Specifically, McKay and Johnson interpret the “power necessity” operator, N, in this way: “Np” abbreviates “p and no one has, or ever had, any choice about whether p.” Three principles involving this N operator are Alpha: It is logically necessary that p implies Np; The Transfer Principle: {Np & N(if p then q), implies Nq}; and The Agglomeration Principle: Np and Nq implies N(p & q).

McKay and Johnson construct a counterexample against the Agglomeration Principle. As this principle is a consequence of the Transfer Principle and Alpha, and Alpha is beyond question, McKay and Johnson reject the Transfer Principle. 8. Widerker himself (1987) proposes a principle of this sort. Timothy O’Connor (2000: esp. 3–15) has insightful contributions on the Transfer Principle. In a recent illuminating paper, Erik Carlson 2000 distinguishes three formally different versions of the Transfer Principle and eight different interpretations of ‘has a choice about’;

257

this, he says, yields twenty-four substantially different Transfer principles. Of these twenty-four principles, Carlson argues that only four are not controversially valid. 9. Another strategy to impugn the Transfer Principle maintains that this principle is selective in the inferences it licenses. In this approach, pursued by Michael Slote 1982, there are various types of necessities like epistemic and deontic for which appropriately formulated analogues of the Transfer Principle fail. These failures are used to motivate rejection of the Transfer Principle when it is applies to the type of necessity – “unavoidability necessity” or “power necessity” – that is involved in the Consequence Argument, by assimilating power necessity to the other forms of necessity for which analogues of the Transfer Principle fail. So in this strategy, one can grant that one is powerless over the laws of nature and the past, and powerless over these entailing one unique future, but deny that it then follows that one is powerless over the future. Slote’s argument has been criticized by Fischer 1994: 29–44; and O’Connor 1993. 10. J. L. Austin 1966 described cases in which though it is true that some person could do something, it is false that she would do that thing if she chose to do so. In one of his best-known examples, Austin imagined himself a golfer standing over a three-foot putt. Suppose he had made such putts in several but not in all relevantly similar situations in the past. Then it seems perfectly consistent to say, Austin proposed, that he could make the putt on this occasion, although he might miss (on this occasion). So his ability to make the putt does not entail that he would make the putt whenever he chose to do so. Keith Lehrer 1968 offers a different sort of counterexample against the hypothetical analysis in consideration. There can be cases, Lehrer claims, in which although a person cannot do something, it is true that she would do it if she chose to do so. Suppose, for instance, that as a result of the traumatic experience of having been bitten by a snake in her youth, Leila has developed an extreme psychological aversion to snakes. Her aversion renders her unable to choose to pick up the harmless snake in her biology class. And (partly) as she can’t choose to pick up the snake, she cannot pick up the snake. Yet, Lehrer argues, it might be true that if she were to choose to pick up the snake, she would do so. So whereas Leila cannot pick up the snake because of her aversion, the conditional analysis yields the result that she can so act. It is, hence, implausible to propose that ‘S can refrain from doing A in U ’ simply amounts to ‘S would refrain from doing A in U had S wanted, or willed, or chose to do something else instead.’ Objections against conditional analyses of ‘can’ are also to be found in Anscombe 1971, Chisholm 1966, Foot 1966, and Kane 1996a: ch. 4. 11. In The Natural Selection of Autonomy Bruce Waller, it seems, accepts principle K (1998: 59). In addition, he believes that alternative possibilities are required for “natural autonomy” (or freedom). This sort of autonomy, what he calls “natural autonomy-as-alternatives,” is exercised, he says, by both humans and non-humans (like white-footed mice), and is “tightly linked to environmental shaping” (1998: 11). Waller claims that one performs some action freely or with natural autonomy only if one could have done otherwise, and, favoring a hypothetical interpretation of ‘could have done otherwise,’ he adds, one could have done otherwise if one had chosen otherwise:

258

When pursuit of alternatives is recognized as an effective natural strategy – for mice as well as humans – then the compatibilist “hypothetical” account of alternatives becomes more appealing. We do not want freedom for choices with no causal antecedents, freedom from all environmental contingencies. . . . To the contrary, . . . we (humans and white-footed mice) want to be able to act otherwise if we choose otherwise; that is, we want other options available when we experience different circumstances in our changing environment. (11)

Waller believes that such alternatives “cannot carry the weight of moral responsibility” (10) and he bids good riddance to this sort of responsibility in the natural world. Yet, he thinks, in such a world all kinds of other moral assessments are possible, including aretaic assessments and deontic ones of the sort that one ought morally to feed one’s children (1998: 59–60, 73). So Wallace seems to espouse an asymmetry thesis interestingly different from the one I have argued for, to wit, although one cannot be morally responsible for anything that one does in the natural world, one can do things that are right, wrong, or obligatory in such a world, or one can act from virtue or duty in such a world. But I think we should not lose sight of the following. Suppose Waller’s “scientific-naturalist world view” (34) is a view that has, as a constituent, the thesis that determinism is true. Then as Waller accepts principle K, he should also accept principles CK and KW for the reasons advanced in Chapter 3 of this book. But these principles (in conjunction with other relevant claims), as we have seen, imply that a determined world is devoid of deontic anchors and, hence, deontic assessments are not possible in the natural world if it is a determined one. 12. In formal problems with principles like GR1, see, e.g., Feldman 1997: 3–4 and ch. 1 and Castaneda 1968. 13. Saka adds that lines (3), (5), (7), and (8) can be established by appeal to the following rules of inference (together with apt classical tautologies): (R1) [] (P →Q) []P Therefore, []Q

(R2) [] (P →Q) <>P Therefore, <>Q

For example, according to Saka, from the tautology (P & ∼ Q) → ∼ (∼ P v Q), we have [][(P & ∼Q) → ∼ (∼ P v Q)]. This combined with [](P & ∼Q) from line (6) allows rule (R1) to yield []∼ (∼ P v Q), i.e. line (7). 14. So, for instance, Saka reports that according to one “meaning-equivalence” view, (16) is equivalent to (17) (cf. Kant 1797, Hare 1952, and Gensler 1996).

(16) You ought to take off your shoes. (17) Take off your shoes! But this reduction . . . contradicts “ought implies can”. . . . [T]he simple TRANSFORMATIONAL ANALYSIS treats (17) as equivalent to (18) (e.g. Stockwell et al. 1973). But if this is true, then (16) is equivalent to (18) which is absurd.

(18) You will take off your shoes. (2000: 100–1)

259

CHAPTER 6

1. Other proponents of noncausal accounts of free action include Goetz 1998 and McCann 1998. There is an instructive, critical discussion of such accounts in Clarke (in press). 2. See van Inwagen 1983: 3. I have somewhat modified van Inwagen’s characterization of undetermined acts. 3. It is arguable that some R-libertarians can consistently reject PAP and so not insist that the control required for moral responsibility is dual. For example, Eleonore Stump 1996b suggests that Aquinas may have been a libertarian of this sort. Also see Zagzebski 2000. 4. I have described a doxastic version of Modest Meleian Libertarianism. Mele makes clear that other analogous versions are possible: Other indeterministic scenarios may also be considered. For example, one may explore the benefits and costs of its being causally undetermined which of a shifting subset of desires come to mind at a time, or which of a changing segment of beliefs and agent actively attends to at a given time, or exactly how an agent attends to various beliefs or desires of his at certain times. (1995: 221, note omitted)

Laura Ekstrom 2000 offers a modest R-libertarian account that is analogous to Mele’s doxastic account in its positioning of indeterminacy. In her account, the notion of preference rather than that of best judgment plays a central role. A preference, Ekstrom says, is a desire “formed by a process of critical evaluation with respect to one’s conception of the good” (2000: 106). She further maintains that the formation of such a preference is an action. Indeterminism, in Ekstrom’s view, is involved in the formation of these preferences. A decision is free, she proposes, if it is brought about in an appropriate way by an active formation of a preference favoring that decision, and that preference formation (an action), in turn, results from an uncoerced exercise of the agent’s evaluative faculty. Specifically, the formation of the preference is nondeterministically caused by the considerations (which are inputs of the agent’s evaluative faculty) taken up in deliberation (109, 114). Others who have proposed or discussed similar versions of libertarianism include Dennett 1978, Fischer 1995, Kane 1985, and Nozick 1980. 5. Kane stresses this point. See, e.g., Kane 1996a: 111. 6. In fairness to Mele, he remarks that what some libertarians might prize with his version of libertarianism over compatibilist accounts is that his modest libertarianism offers them a species of agency that gives agents a kind of independence and an associated kind of explanatory bearing on their conduct that they would lack in any deterministic world. Agents of the imagined sort would make choices and perform actions that lack deterministic causes in the distant past, and for this reason, their choices and actions would have their ultimate sources in them. Mele adds that traditional libertarians who want more than what his modest libertarianism offers in the way of ultimacy must show that “what they want is coherent. And that requires showing that what they want does not entail or presuppose a kind of luck that would itself undermine moral responsibility” (1999a: 291–2). Perhaps Mele’s point can be recast in this way: Modest libertarianism of Mele’s variety can

260

accommodate some concerns about ultimacy; libertarian views (like Kane’s) that make room for more robust ultimacy by, for example, insisting that decisions are undetermined, will run afoul of problems of luck. So there is an unhappy tradeoff between concerns of ultimacy and responsibility-undermining luck.

CHAPTER 7

1. Perhaps agent-causal theories would offer more in the way of control. For thoughts on this view see Clarke 1996a. A dissenting opinion is to be found in van Inwagen 2000. 2. Kane withdraws this “all past circumstances remaining the same” condition when the choice is preceded by some indeterminate event. His view is that “Exact sameness (or difference) of possible worlds is not defined if the possible worlds contain . . . indeterminate events of any kinds” (1996a: 172). 3. Kane claims that directly free choices must satisfy the plurality conditions. 4. Versions of the Luck Objection have been discussed by Double 1991: 140, Mele 1999a, b, 1998, Pereboom 2001, Strawson 1994, and Waller 1988. Van Inwagen (2000: sec. II) has recently argued that agent-causal libertarian theories also succumb to the luck objection. 5. Kane introduces this doubling of efforts – the agent’s attempting intentionally and voluntarily at the same time to do two incompatible things – in Kane 1999a, b. This kind of irrationality on the agent’s part may pose a threat to free action. Minimally, it saddles Kane with the troubling view that irrationality of the relevant sort is an inevitable ingredient of many free choices or actions. 6. Here, I set aside the worry of Kane’s that with indeterminate efforts, neither exact sameness nor exact difference is “defined” (1996a: 171). For critical discussion of this point, see, e.g., Clarke 1999, Haji 1999b, Loewer 1996, Mele 1999a, esp. 278–80, and O’Connor 1996. 7. Some theorists have proposed a third condition that says, in broad strokes, that the agent not act on actional springs that are not “truly her own” or “unauthentic.” See, e.g., Fischer and Ravizza 1998: chs. 7, 8 and Haji 1998a. Various incompatibilists champion other conditions like the condition of ultimate responsibility. In Kane’s view, to be ultimately responsible for an action, “the agent must be responsible for anything that is a sufficient reason (condition, cause, or motive) for the action’s occurring” (1999a: 106). 8. This condition is suggested by some of Kane’s remarks on libertarian free choice. See, e.g., Kane 1996a: 134–5, 187. In the context of indeterministic choice, Kane’s notion of an agent’s making reasons (R) for a choice prevail by choosing for those reasons is problematic. For if these reasons prevail as a result of making the pertinent choice, the fact that the agent made this very choice can’t duly be explained by those reasons being made to prevail. For a crisp articulation of this worry, see Clarke 1999: 35–6. For an analogous point about Kane’s earlier views that appeared in Kane’s Free Will and Values, see Mele 1995: 207–8. 9. “Ledger views” of the notion of responsibility are held by Feinberg 1970: 30–1, 124–5, Gert and Duggan 1979: 119–217, Glover 1970, Haji 1998a: 9–10, and Zimmerman 1988: 38–9. Carl Eliot 1996: 65–6 also appears to favor a ledger view. There are alternatives. So, for example, Peter Strawson suggests that we can

261

understand moral responsibility in terms of certain social practices. Strawson 1962 argues that, when members of a society regard a person as a responsible agent, they react or see it fitting to react to that person with a characteristic set of feelings and attitudes – the so-called reactive attitudes – such as resentment, love, gratitude, forgiveness, respect, and indignation. For an instructive discussion of Strawson’s account, see Russell 1992. Jay Wallace 1994 defends an original and sophisticated variation of the Strawsonian account. Rejecting the Strawsonian approach, Marina Oshana proposes that “when we say a person is morally responsible for something, we are essentially saying that the person did or caused some act (or exhibited some trait of character) for which it is fitting that she give an account” (1997: 77). In Gary Watson’s “self-disclosure” view, the concept of moral responsibility entails that an agent is morally responsible insofar as he has the capacity to choose ends freely and act in accordance with such choices (Watson 1996). CHAPTER 8

1. The notion of unsheddability here is Mele’s notion of practical unsheddability. See Mele 1995: 153–4. So, for instance, Jenny’s implanted values are unsheddable in that, given her psychological constitution at her world, she can neither uproot nor attenuate these values. Such unsheddability, unless otherwise stated, is assumed in all subsequent cases involving CNC manipulation. 2. Similar cases involving global manipulation have been discussed by various authors including Dennett 1984, Double 1989, and Mele 1995: esp. ch. 9. Some might be concerned about whether postsurgery Jenny is identical to presurgery Jenny. In response to this concern, Walter Glannon 1998: sec. IV has argued that our practices of holding people morally and criminally responsible require only a low threshold of psychological connectedness and bodily continuity. He agrees with Dan Brock and Allen Buchanan, who write: There are powerful pragmatic reasons for maintaining a social consensus according to which degrees of psychological continuity which meet or exceed a particular threshold are sufficient for an unqualified judgment of personal identity, a judgment which carries with it maximal moral force. . . . So from the claim that the psychological continuity upon which personal identity depends is a matter of degree, it does not follow that personal identity judgments should be replaced by judgments about varying degrees of psychological continuity and correspondingly moral judgments concerning the varying weights of rights and obligations. (1989: 180)

Further discussion on personal identity in “global transformation cases” can be found in Mele 1995: 75 and n. 22. An important point stressed by Mele is that the pre- and postpsychosurgery persons may be strongly psychologically connected in Parfit’s sense: The number of direct psychological connections between them “is at least half the number that hold, over every day, in the lives of nearly every actual person” (Parfit 1986: 206). Jennifer Radden 1996: esp. ch. 3 provides instructive examples from abnormal psychology of multiple or successive selves which, in some respects, resemble Psychohacker. 3. An instructive criticism of this sort can be found in Fischer and Ravizza 1994. See also Slote 1980 and Stump 1993.

262

4. Reasons responsiveness accounts are defended, for example, by Fischer and Ravizza 1998, Haji 1998a, Wallace 1994, and Watson 1987. Fischer and Ravizza have proposed a way of responding to worries about manipulation. Their contention would be that, given the sort of manipulation involved in Psychohacker, the “mechanism” issuing in the agent’s behavior would not be the “agent’s own mechanism.” See Fischer and Ravizza 1998: esp. chs. 7, 8. 5. I have assumed that whereas Jill deterministically A-s, Jonsey nondeterministically A-s, and that Jill’s A-ing results from actional elements type-identical to those from which Jonsey’s A-ing results. One might worry whether Kane ought to allow this example, as in the former case, Jill’s pertinent beliefs and desires are not causally undetermined as to which course of action Jill will perform. But the germane beliefs and desires are causally undetermined in the case of Jonsey. So Jonsey’s beliefs and desires, it might be objected, could not be type-identical to those of Jill’s. I think, though, that the example ought to be allowed. Assume that in a deterministic world, it is determined that certain beliefs will come to Jill’s mind, and these beliefs (together with other actional elements) will deterministically cause her to A. In an undetermined world, it might be that pertinent beliefs nondeterministically come to Jonsey’s mind and they then deterministically cause her to A. On this option, it isn’t clear why the beliefs that cause Jonsey to A cannot be type-identical to those that cause Jill to A in every respect germane to deliberation and to the causal production of A-ing. The only difference between the two sets of beliefs is that whereas Jill’s beliefs deterministically come to Jill’s mind, Jonsey’s beliefs indeterministically come to her mind. Further, this difference does not imply that beliefs of the first sort can’t be type-identical to beliefs of the second sort. Alternatively, it might be that in an undetermined world, pertinent beliefs nondeterministically come to Jonsey’s mind and they then nondeterministically cause her to A. Still, one wonders why the very same sort of belief (weighted in the same type of way by its agent) cannot both probabilistically (in one scenario) cause its agent to A; and deterministically (in a second scenario) cause its agent to A. From the fact that one belief (B1) probabilistically causes its agent to A, and a second (B2) deterministically causes its agent to A, it doesn’t appear to follow that B1 and B2 can’t be beliefs of the same type. 6. Kane argues that at least some Frankfurt-style cases are incoherent. (See, for example, Kane 1996a: 142–3 and 1999a: 84–8.) Widerker (1995a: 247–61, 1995b: 113–18) has adduced an argument similar to Kane’s. For responses to Kane, see Fischer 1999a, Haji 1999b, Mele and Robb 1998, and Pereboom 2001: 18–28 and 2000. 7. For a defense of this view, see, e.g., Fischer and Ravizza 1998: ch. 7, Haji 1998a: chs. 6, 7, and Mele 1995. 8. See, e.g., Frankfurt 1992, 1988. 9. Frankfurt 1988: 25 says that his theory is neutral with respect to the problem of determinism. But such “neutrality” implies that his hierarchical view is compatible with determinism. 10. Many thanks to Stefaan Cuypers for discussion of these issues. For further thoughts of Cuypers on Frankfurt’s hierarchical account of responsibility and manipulation, see Cuypers’s article, “The Trouble with Harry.” David Zimmerman has recently developed a criticism that builds on the idea that a victim of CNC manipulation,

263

11. 12.

13.

14. 15.

16.

in Frankfurt’s view, can “take responsibility” for externally induced volitions simply by bringing himself to identify with them wholeheartedly and decisively, despite their causal origins. Zimmerman argues that this view of “taking responsibility” commits Frankfurt to the morally repellent Stoic thesis that “acting freely . . . is a matter of resigning and adapting oneself to necessity” (2000: 25). For an account of (ii), see Haji 1998a: 126–39. Note that ‘normative agent’ is a technical term. One should not confuse, say, Starr with a normative agent and then scoff at the untenable suggestion that Starr has his evaluative scheme essentially. Presumably, Starr could have had a very different evaluative scheme than the one he in fact has. For additional discussion on the notion of normative agency, see Haji 1998a: 118–19. Repression of a scheme might occur, for instance, in a case in which though the agent possesses the elements of that scheme, the agent cannot “access” those elements. An example involving repression is to be found in Haji 1998a: 6–7. For discussion on authenticity, see Fischer and Ravizza 1998, and Mele 1995. For this reason, I disagree with Patricia Williams’s claim that it is not possible to develop “theories of prescriptive evolved ethics which do not suffer from internal, logical contradictions” (Williams 1993: 238). In support of this claim, she says that “the minimum requirement for beings to be ethical is that they can legitimately be blamed, praised, and held responsible for their actions”; further, “to deserve praise or blame or to be held responsible for their actions . . . [they] must be able to self-reflect, . . . [they] must be able to weigh and deliberate among various options” (234) and they must be able to make and act on choices based on such deliberations. The robust modest M-libertarian theorist need not, of course, be committed to the view that free actions are preceded by the agent’s exerting efforts of will or exerting efforts of will to do two obviously incompatible things. I see little reason to augment an M-libertarian theory (or, for that matter, an R-libertarian theory) with prior efforts. CHAPTER 9

1. Ted Honderich contends that although determinism is incompatible with moral responsibility and with retributive attitudes in general, as these things presuppose that agents causally originate actions, it is not incompatible with rightness and wrongness (1988: 525–30, esp. 529). I do not, of course, accept the latter claim of Honderich’s. Similarly, Bruce Waller submits that It is not difficult to construct examples that challenge the widespread belief that there can be no morality without moral responsibility. For example, if you are convinced of the truth of utilitarianism, then you can believe in it just as confidently while also believing that you deserve no praise for your enlightened moral beliefs and that benighted deontologists deserve no blame. (1990: 152, 170–4.)

Again, I disagree with Waller if the remarks in this passage are meant to imply that in a determined world, a utilitarian can rationally affirm that it is possible for actions to instantiate primary deontic properties.

264

CHAPTER 10

1. Here, and in the rest of this section, I use ‘worst’ and ‘best’ liberally. Following Zimmerman 1996: 14, in saying or implying that one ought to do the best one can, or one ought not to do what is nonbest or worst, or, of all of one’s alternatives, one of them makes the world worst off, I am only committing myself to the claim that there is some way in which what one ought to do is superior to any other of one’s alternatives. 2. I use ‘supererogatory,’ ‘action beyond duty,’ and ‘action beyond the call’ interchangeably, as I do ‘suberogation,’ ‘offence,’ ‘permissive ill-doing,’ and ‘falling short of decency.’ 3. For the first, see Feldman 1978: 48 and for the second, see Urmson 1969: 63. 4. Relevant discussion of the Islamic fivefold model can be found in Carney 1983 and Reinhart 1983. 5. Proponents of the Objective View include Copp 1997a, Fields 1994, Moore 1984, Oakley and Cocking 1994, Smith 1983, 1991, Wallace 1994, and Widerker 1991, 2000. Baier 1995: 322, 325 and Honderich 1988: 431 also appear to endorse the Objective View. Zimmerman 1988, 1997 and Milo 1984: ch. 1 reject the Objective View. So, it appears, does Brandt 1958. Widerker has recently proposed that a person is morally blameworthy for doing A if and only if in doing A he acted in a morally wrong way, and in the circumstances it was morally reasonable to expect him not to have done A (2000: 192). This alternative possibilities condition should be rejected if O1 (of the Objective View) is defective. Further discussion related to this condition is contained in n. 13 below. 6. For example, in one interpretation, a perfect duty ought always to be satisfied, whereas an imperfect one only ought sometimes to be satisfied, (See Hill 1971: 56–7.) In a second interpretation, a perfect duty is fully specific whereas an imperfect one leaves it to the agent to determine how that duty is to be satisfied. A stock illustrative example involves what Kant took to be the imperfect duty to act charitably. This duty can be satisfied either by being charitable to Larry, or to Moe, or to Curly, and so on. One might think of this duty as a disjunctive one. (See, e.g., Mellema 1991: 33–7 and Stocker 1968: 61.) In a third interpretation, a perfect duty is a duty to perform some act (or to refrain from performing some act) whereas an imperfect duty is a duty to adopt a certain maxim for action. See Hill 1971: 57. 7. A theoretical framework that supports this claim is defended by McNamara (1996). 8. Heyd actually claims that forbearance from doing evil is always a right thing to do. (This, I believe, is questionable: Sometimes, when performance of any of one’s alternatives would result in (overall) evil, one may well be obligated to bring about the one that would produce the least evil.) I have taken the liberty of replacing ‘right thing to do’ with ‘praiseworthy.’ 9. I don’t take Heyd to be committed to this argument. My position is simply that the relevant text suggests this sort of argument. 10. Feldman gives this example, among other things, to illustrate the view that obligations can change with the passage of time. See Feldman 1986: 11–12. 11. Gary Watson 1996: 239–40 considers similar views. 12. Wallace’s account of responsibility is a variation of the Strawsonian account.

265

13. I proposed in Chapter 3 that an appeal to theoretical considerations may help in assessing principles such as K and CK. One might well wonder whether this same sort of strategy could be successfully pursued to defend PAP. Specifically, if we grant, for the sake of argument, that Frankfurt-type examples are possible – there are cases in which an agent freely does something that she could not have avoided doing – could we not argue on the basis of theoretical reasons favoring the demand for alternative possibilities, that agents in such examples are not, for example, morally blameworthy? Widerker has recently developed a defense of PAP (the “W-defense”) along these lines (see Widerker 2000: 191–4). To advocates of the view that an agent may well be blameworthy for what he does in appropriately described Frankfurt-type examples, Widerker poses the question central to the W-defense: “What, in your opinion, should he have done instead?” (191). Widerker writes: The W-defense points to an important reason why it would be unreasonable to judge an agent morally blameworthy in . . . [a Frankfurt-type situation]. When we consider someone morally blameworthy for a certain act, we do so because we believe that morally speaking he should not have done what he did. This belief is essential to our moral disapproval of his behavior. Sometimes, however, such a belief may be unreasonable. This happens in a situation in which it is clear to us that the agent could not have avoided acting as he did. To expect in that situation that the agent should not have done what he did is to expect him to have done the impossible. By implication, considering him blameworthy because he has not fulfilled this unreasonable expectation would be unreasonable. (191–2)

Widerker concludes that the W-defense suggests the following general constraint on ascriptions of moral blame:

(PAE) An agent S is morally blameworthy for doing A only if in the circumstances it would be morally reasonable to expect S not to have done A. (192) He explains that by ‘morally reasonable,’ he means morally reasonable for someone who is aware of all the relevant moral facts pertaining to S’s doing A (199, n. 26). Notice, however, that it is only to advocates of the Objective View (in particular, to proponents of O1) that the W-defense may prove appealing. Widerker proposes that when we consider someone morally blameworthy for a certain act, we do so because we believe that morally speaking he should not have done what he did and that this belief is essential to our moral disapproval of his behavior. If ‘should,’ in this proposal, expresses deontic obligation, then the proposal straightforwardly begs the question in favor of O1. I suggest, instead, that when we consider someone morally to blame for a certain act, we do so because we believe that he acted in light of the belief that he was doing wrong. One can so act even if one could not have avoided doing what one did. 14. Both moderate and strong K entail that if S ought to do A, then S is psychologically capable of doing A. In the circumstances, Deadly is psychologically capable of injecting the patient with A. Without the relevant intention or purpose, Deadly is psychologically incapable of injecting the patient with A. For other examples involving similar psychological capacity or incapacity, see Mele 1992: 87–8.

266

15. DP is a theorem of theories like Feldman’s (1986) and Zimmerman’s (1996). One should be heedful that DP requires qualification in order to avoid Good-Samaritan-type paradoxes. The following modification seems adequate: If S cannot do p without doing q (perhaps because q is a logical consequence of p), and if S can refrain from doing q, then, if S ought to do p, S ought also to do q. DP is discussed, among other places, in Zimmerman 1996: sec. 2.3 and in Feldman 1990. 16. I thank David Copp for recommending that I consider this rejoinder. CHAPTER 11

1. Frankfurt argues that addicts can be responsible for actions that issue from irresistible desires with which they identify. See, for example, Frankfurt 1971. I discuss this view of Frankfurt in Haji 1998a: 70–5. 2. An agent can lack minimal rational control in A-ing consistent with that agent’s being morally competent (as specified by the moral competence model of moral agency), or normatively competent (as specified by the normative competence model of moral agency). 3. The worry that some of the virtues can’t be understood apart from deontic notions as well as other criticisms of virtue theories that tie right and wrong to virtue and vice are discussed by James Rachels in his 1999: 189 ff. 4. For an account of acting from virtue, see Audi 1995. 5. For our purposes, we can safely overlook the formal incoherence of this sort of theory. For discussion of the incoherence see, for example, Castenada 1968 and Feldman 1986. CHAPTER 12

1. ‘Plain ought’ is a term Feldman introduces in his discussion of overridingness. See Feldman 1986: 212–15. Needless to say, the Overridingness Thesis is controversial. Some opponents include Copp 1997b, Foot 1978, Kekes 1992, Slote 1983, Williams 1981, and Wolf 1982. 2. I assume that such conflicts are possible. 3. This sort of view is defended by Neil Cooper 1981 and Michael Smith 1994: esp. 182–4. 4. David Copp 1997b argues that there is no such standard as Reason. His argument, I believe, turns essentially on whether it is possible for Reason to have the property of being normatively most important without its having this property being dependent in any way on the assessment by any other standard of Reason’s relative importance. If Copp is right that there is no such standard, it does not, of course, follow that there can be nothing like practical reasoning which relies, among other “inputs,” on specific normative standpoints to generate “outputs” like practical judgments about what to do. 5. For some interesting discussion on acting “from” love, as opposed to acting “from” moral obligation, see Frankfurt 1994. 6. There is a complication here. On the one hand, the father may believe that he ought (morally) not to mislead the police, and he may believe that he is justified in believing this, but the justification may indeed not be one, or at least not be a

267

7.

8.

9. 10. 11. 12.

good one. On the other hand, the father may believe that he ought morally not to mislead the police, and that this belief of his is justified. It’s the latter, I believe, Slote has in mind in his use of ‘must’ in the relevant contexts. The point here is that a proponent of the ‘ought’ implies ‘can’ principle K, when challenged with the example of the desperate mother, will not be moved by this sort of argument: (1) It is false that it is incoherent to deny K. (2) If (1), it is false that K is necessarily true. (3) Therefore, it is false that K is necessarily true. (4) If K is true, then K is necessarily true. (5) Therefore, it is false that K is true. Argument (1) would be supported by a tale in which the desperate mother couldn’t save her drowning child as she couldn’t reach him in time. An element of the story would be that the mother sincerely believed that she had a moral obligation to save her child. If one were to claim that this belief of the mother’s does contain a contradiction, then a proponent of the Overridingness Thesis should similarly claim that the father’s belief that love overrides morality also contains a contradiction. Strictly, a proponent of the Overridingness Thesis need simply be committed to the view that cases in which obligations of different sorts conflict involve conflicts between different overall verdicts as to what agents ought to do, and not conflicts between different kinds of reasons. Useful discussion on a somewhat fictionalized Gauguin’s case can be found in Slote 1983 and Williams 1981. For some other implications, see Copp 1997b: sec. 7 and Kekes 1992: esp. 74. For elaboration, see Haji 1998a, ch. 11 and Wolf 1982. David Copp 1997b develops this sort of criticism.

268

Glossary and List of Principles

To assist readers unfamiliar with standard terminology in the literature on freedom and responsibility, and with technical terms used or introduced in this work, I provide some definitions. I also list principles K, CK, OW, PAP, and WAP. GLOSSARY

Action-Centered Modest R-Libertarianism (or Robust Modest R-Libertarianism): A version of modest R-libertarianism that locates indeterminism relatively late in the causal pathway of an action, and it posits that the events that are nondeterministically caused are actions (for example, the making of decisions). Antecedent Contrastive Control: The control an agent has to bring it about that she performs some action rather than that she does not and vice versa in virtue of exercising control, proximal or otherwise, prior to the occurrence of the action performed. Compatibilism: The thesis that the sort of freedom required for moral responsibility (and, I might add, deontic anchors) is compatible with the truth of determinism. Deontic Act (or a Deontic Anchor): Any act that instantiates one or more of the primary deontic properties. Deontic Morality: The set of deontic acts. Determinism: The thesis that, for any given time, a complete statement of all the (nonrelational or “genuine”) facts about that time, together with a complete statement of the laws of nature, entail one unique future. Dual Control: An agent has dual control over performing an action if and only if she can perform that action and she can avoid performing that action.

269

Incompatibilism: The denial of compatibilism. Traditional incompatibilists deny that the sort of freedom required for moral responsibility is compatible with determinism. Others might want to deny that the sort of freedom required for deontic anchors is compatible with determinism. Indeterminism: The thesis that determinism is false. (The occurrence of a single causally undetermined event in a world suffices for the falsity of determinism at that world.) Indeterministic Theory: A theory that specifies the type of freedom or control required for moral responsibility or deontic morality and it entails that the sort of freedom required for either is incompatible with determinism. M-Libertarianism: The thesis that some human beings can or do perform actions that have one or more of the primary deontic properties, and that the type of freedom required for deontic morality is incompatible with determinism. Modest R-Libertarianism: R-libertarianism conjoined with the view that the sort of freedom necessary for responsibility does not require metaphysically exotic agents (such as Cartesian minds) or exotic forms of causation (such as agent causation), but does require nondeterministic event causation at a point or points in the causal pathway of an agent’s choices or actions. (Modest M-libertarianism is an analogous view about the sort of freedom required for deontic anchors.) Nonaction-Centered Modest R-Libertarianism: A version of modest R-libertarianism that locates indeterminism relatively early in the causal pathway of an action at some point prior to the making of a decision, and it posits that the events that are nondeterministically caused are not actions of any sort. Primary Deontic Properties: The moral properties of rightness, wrongness, and obligatoriness. Proximal Control: The sort of control (exerted by an agent) concerning the direct causal production of agent-involving events such as the agent’s having specific values, desires, and beliefs, making an evaluative judgment, and forming an intention. R-Libertarianism: The thesis that some human beings choose and act freely and are morally responsible for at least some of what they do, and that free choice, free action, and moral responsibility are incompatible with determinism. Singular Control: An agent has singular control over performing an action if and only if she can perform that action. Ultimate Control: The sort of control involving lack of influence of agentexternal events. For an agent S to have ultimate control over event, S’s x-ing at time t, where x-ing might be the making of a decision, there must be no time at which there are minimally causally sufficient conditions that do not include events or states internal to S for S’s x-ing at t.

270

PRINCIPLES

K As of time t, an agent S ought morally to do something A at time t ∗ (where t may either be t ∗ or a time later than t ∗) only if S can, as of t, do A at t ∗; and, as of t, S ought not to do A at t ∗ only if S can, as of t, not do A at t ∗. PAP A person is morally responsible for what he has done only if he could have done otherwise (Frankfurt 1969: 829). OW Agent S ought to do (not to do) A if and only if it is morally wrong for S not to do (to do) A. WAP It is morally wrong for S to do (not to do) A only if S can refrain from doing (do) A. CK: If one ought to do A, then one can refrain from doing A (and if one ought not to do A, then one can do A).

271

References

Ansari, Zarar Ishaq. 1972. “Islamic Juristic Terminology Before Shafii: A Semantic Analysis with Special Reference to Kufa.” Arabica 19: 255–300. Anscombe, G. E. M. 1971. Causality and Determinism. Cambridge: Cambridge University Press. Aqvist,Lennart.1969.“Improved Formulations of Act Utilitarianism.” Noˆus 3: 299–323. Aristotle. 1941. Nicomachean Ethics. In Richard McKeon, ed., The Basic Works of Aristotle. New York: Random House. Aristotle. 1941. Rhetoric. In Richard McKeon, ed., The Basic Works of Aristotle. New York: Random House. Audi, Robert. 1997. Moral Knowledge and Ethical Character. New York: Oxford University Press. Audi, Robert. 1995. “Acting from Virtue.” Mind 104: 449–71. Austin, J. L. 1966. In Bernard Berofsky, ed., Free Will and Determinism. New York: Harper & Row. Baier, Kurt. 1995. The Rational and the Moral Order: The Social Roots of Reason and Morality. Chicago and La Salle, IL: Open Court. Baron, Marcia. 1986. “On Admirable Immorality.” Ethics 96: 557–66. Becker, Lawrence. 1986. Reciprocity. London: Routledge & Kegan Paul. Bergstrom, Lars. 1976. “On the Formulation and Application of Utilitarianism.” Noˆus 10: 121–44. Bergstrom, Lars. 1971. “Utilitarianism and Alternative Actions.” Noˆus 5: 237–52. Berofsky, Bernard. 1987. Freedom from Necessity: The Metaphysical Basis of Responsibility. New York: Routledge & Kegan Paul. Brandt, Richard. 1958. “Blameworthiness and Obligation.” In A. I. Melden, ed., Essays in Moral Philosophy. Seattle: University of Washington Press. Brock, Dan, and A. Buchanan. 1989. Deciding for Others: The Ethics of Surrogate Decision Making. Cambridge: Cambridge University Press. Carlson, Erik. 2000. “Incompatibilism and the Transfer of Power Necessity.” Noˆus 34: 277–90. Carney, Frederick. 1983. “Some Aspects of Islamic Ethics.” Journal of Religion 63: 159–74.

272

Castaneda, Hector-Neri. 1968. Analysis 28: 141–2. Chisholm, Roderick. 1966. In Bernard Berofsky, ed., Free Will and Determinism. New York: Harper & Row. Chisholm, Roderick. 1963. “Supererogation and Offence: A Conceptual Scheme for Ethics.” Ratio 5: 1–14. Chisholm, Roderick, and E. Sosa. 1966. “Intrinsic Preferability and the Problem of Supererogation.” Synthese 16: 321–31. Clarke, Randloph. In Press. “Libertarian Views (II): Critical Survey of Noncausal and Event-Causal Theories.” In R. Kane, ed., The Oxford Handbook of Freewill. New York: Oxford University Press. Clarke, Randolph. 2000. “Modest Libertarianism.” Philosophical Perspectives 14: 21–45. Clarke, Randolph. 1999. “Free Choice, Effort, and Wanting More.” Philosophical Explorations 2: 20–41. Clarke, Randolph. 1996a. “Agent Causation and Event Causation in the Production of Free Action.” Philosophical Topics 24: 19–48. Clarke, Randolph. 1996b. “Contrastive Rational Explanation of Free Choice.” Philosophical Quarterly 46: 185–201. Clarke, Randolph. 1995. “Indeterminism and Control.” American Philosophical Quarterly 32: 125–38. Conee, Earl. 1987. “Against Moral Dilemmas.” In Christopher Gowans, ed., Moral Dilemmas. New York: Oxford University Press. Cooper, Neil. 1981. The Diversity of Moral Thinking. Oxford: Clarendon Press. Copp, David. 1997a. “Defending the Principle of Alternate Possibilities: Blameworthiness and Moral Responsibility.” Noˆus 31: 441–56. Copp, David. 1997b. “The Ring of Gyges: Overridingness and the Unity of Reason.” In A. F. Paul, F. D. Miller, Jr., and J. Paul, eds., Self-Interest. Cambridge: Cambridge University Press. Darwall, Stephen. 1998. Philosophical Ethics. Boulder, CO: Westview Press. DeGrazia, David. 1996. Taking Animals Seriously: Mental Life and Moral Status. Cambridge: Cambridge University Press. Dennett, Daniel. 1984. Elbow Room: The Varieties of Free Will Worth Wanting. Cambridge, MA: MIT Press. Dennett, Daniel. 1978. “On Giving Libertarians What They Say They Want.” In D. Dennett, Brainstorms. Montgomery, VT: Bradford Books. Dewey, John. [1891] 1957. Outlines of a Critical Theory of Ethics. New York: Hillary House. Donagan, Alan. 1987. “Consistency in Rationalist Moral Systems.” In Christopher Gowans, ed., Moral Dilemmas. New York: Oxford University Press. Double, Richard. 1991. The Non-Reality of Free Will. New York: Oxford University Press. Double, Richard. 1989. “Puppeteers, Hypnotists, and Neurosurgeons.” Philosophical Studies 56: 163–73. Driver, Julia. 1992. “The Suberogatory.” Australasian Journal of Philosophy 70: 286–95. Ekstrom, Laura Waddell. 2000. Free Will: A Philosophical Study. Boulder, CO: Westview Press. Ekstrom, Laura Waddell. 1998a. “Protecting Incompatibilist Freedom.” American Philosophical Quarterly 35: 281–91.

273

Ekstrom, Laura Waddell. 1998b. “Freedom, Causation, and the Consequence Argument.” Synthese 115: 333–54. Elliott, Carl. 1996. The Rules of Insanity: Moral Responsibility and the Mentally Ill Offender. Albany: State University of New York Press. Feinberg, Joel. 1980.“The Interest in Liberty on the Scales.” In Joel Feinberg, Rights, Justice, and the Bounds of Liberty: Essays in Social Philosophy. Princeton, NJ: Princeton University Press. Feinberg, Joel. 1970. Doing and Deserving. Princeton, NJ: Princeton University Press. Feinberg, Joel. 1968. “Supererogation and Rules.” In J. J. Thompson and G. Dworkin, eds., Ethics. New York: Harper & Row. Feldman, Fred. 1997. Utilitarianism, Hedonism, and Desert. Cambridge: Cambridge University Press. Feldman, Fred. 1990. “A Simpler Solution to the Paradoxes of Deontic Logic.” Philosophical Perspectives 4: 309–41. Feldman, Fred. 1986. Doing the Best We Can. Dordrecht, The Netherlands: D. Reidel. Feldman, Fred. 1978. Introductory Ethics. Englewood Cliffs, NJ: Prentice-Hall. Fields, Lloyd. 1994. “Moral Beliefs and Blameworthiness.” Philosophy 69: 397–415. Finch, Alicia, and T. Warfield. 1998. “The Mind Argument and Libertarianism.” Mind 107: 515–28. Fischer, John. 2000. “The Significance of Free Will by Robert Kane.” Philosophy and Phenomenological Research 60: 141–8. Fischer, John. 1999a. “Recent Work on Moral Responsibility.” Ethics 110: 93–139. Fischer, John. 1999b. “Alfred R. Mele, Autonomous Agents: From Self-Control to Autonomy.” Noˆus 33: 133–43. Fischer, John. 1995. “Libertarianism and Avoidability: A Reply to Widerker.” Faith and Philosophy 12: 119–25. Fischer, John. 1994. The Metaphysics of Free Will. Oxford: Blackwell Publishers. Fischer, John. 1989. “Introduction: God and Freedom.” In John Fischer, ed., God, Foreknowledge, and Freedom. Stanford, CA: Stanford University Press. Fischer, John. 1983. “Incompatibilism.” Philosophical Studies 43: 127–37. Fischer, John. 1982. “Responsibility and Control.” Journal of Philosophy 89: 24–40. Fischer, John, and M. Ravizza. 1998. Responsibility and Control: A Theory of Moral Responsibility. Cambridge: Cambridge University Press. Fischer, John, and M. Ravizza. 1997. “Responsibility, Control, and Omissions.” Journal of Ethics 1: 45–64. Fischer, John, and M. Ravizza. 1994. “Responsibility and History.” Midwest Studies in Philosophy 19: 430–51. Fischer, John, and M. Ravizza. 1992. “Responsibility, Freedom, and Reason.” Review, Freedom within Reason by Susan Wolf. Ethics 102: 368–89. Foot, Philippa. 1978. “Are Moral Considerations Overriding?” In Philippa Foot, Virtues and Vices. Oxford: Blackwell. Foot, Philippa. 1966. In Bernard Berofsky, ed., Free Will and Determinism. New York: Harper & Row. Frankfurt, Harry. 1994. “Autonomy, Necessity, and Love.” In Hans Friedrich Fulda and Rolf-Peter Horstmann, eds., Vernuftbegriffe in der Moderne. Stuttgart: KlettCotta.

274

Frankfurt, Harry. 1992. “The Faintest Passion.” Proceedings and Addresses of the American Philosophical Association 66: 5–16. Frankfurt, Harry. 1988. The Importance of What We Care about. Cambridge: Cambridge University Press. Frankfurt, Harry. 1971. “Freedom of the Will and the Concept of a Person.” Journal of Philosophy 68: 5–20. Frankfurt, Harry. 1969. “Alternate Possibilities and Moral Responsibility.” Journal of Philosophy 66: 829–39. Gallois, Andre. 1977. “Van Inwagen on Free Will and Determinism.” Philosophical Studies 32: 99–105. Gensler, Harry. 1996. Formal Ethics. London: Routledge and Kegan Paul. Gert, B., and T. J. Duggan. 1979. “Free Will as the Ability to Will.” Noˆus 13: 197–217. Ginet, Carl. 1996. “In Defense of the Principle of Alternative Possibilities: Why I Don’t Find Frankfurt’s Argument Convincing.” Philosophical Perspectives 10: 403–17. Ginet, Carl. 1990. On Action. Cambridge: Cambridge University Press. Ginet, Carl. 1983. “In Defense of Incompatibilism.” Philosophical Studies 44: 391–400. Glannon, Walter. 1998. “Moral Responsibility and Personal Identity.” American Philosophical Quarterly 35: 231–49. Glover, Jonathan. 1970. Responsibility. London: Routledge & Kegan Paul. Goble, Lou. 1993. “The Logic of Obligation, ‘Better,’ and ‘Worse.’ ” Philosophical Studies 70: 133–63. Goetz, Stewart. 1998. “A Noncausal Theory of Agency.” Philosophy and Phenomenological Research 49: 303–16. Gowans, Christopher. 1987. “Introduction: The Debate on Moral Dilemmas.” In Christopher Gowans, ed., Moral Dilemmas. New York: Oxford University Press. Greenspan, Patricia S. 1978. “Oughts and Determinism: A Response to Goldman.” Philosophical Review 87: 77–83. Haji, Ishtiyaque. 2001. “Control Conundrums: Modest Libertarianism, Responsibility, and Explanation.” Pacific Philosophical Quarterly 82: 178–200. Haji, Ishtiyaque. 2000. “Indeterminism, Explanation, and Luck.” Journal of Ethics 4: 211–35. Haji, Ishtiyaque. 1999a. “Moral Anchors and Control.” Canadian Journal of Philosophy 29: 175–203. Haji, Ishtiyaque. 1999b. “Indeterminism and Frankfurt-Type Examples.” Philosophical Explorations 2: 42–58. Haji, Ishtiyaque. 1998a. Appraisability: Puzzles, Proposals, and Perplexities. New York: Oxford University Press. Haji, Ishtiyaque. 1998b. “Frankfurt-Pairs and Varieties of Blameworthiness: Epistemic Morals.” Erkenntnis 47: 351–77. Haji, Ishtiyaque. 1998c. “On Morality’s Dethronement.” Philosophical Papers 27: 161–80. Haji, Ishtiyaque. 1994. “Changing Obligations and Immutable Blameworthiness.” Theoria 60: 48–62. Haji, Ishtiyaque. 1992. “A Riddle Regarding Omissions.” Canadian Journal of Philosophy 22: 485–502. Hare, R. M. 1963. Freedom and Reason. Oxford: Clarendon Press.

275

Hare, R. M. 1952. The Language of Morals. Oxford: Clarendon Press. Henderson, G. P. 1966. “ ‘Ought’ Implies ‘Can.’ ” Philosophy 41: 101–12. Heyd, David. 1982. Supererogation. Cambridge: Cambridge University Press. Hill, Thomas E., Jr. 1971. “Kant on Imperfect Duty and Supererogation.” KantStudien 72: 55–76. Hoffman, Joshua, and G. Rozenkrantz. 1980. “On Divine Foreknowledge and Human Freedom.” Philosophical Studies 37: 289–96. Honderich, Ted. 1988. A Theory of Determinism: The Mind, Neuroscience, and LifeHopes. Oxford: Clarendon Press. Hudson, Stephen D. 1986. Human Character and Morality. London: Routledge and Kegan Paul. Hunt, David. 1996. “Frankfurt-Counterexamples: Some Comments on the Widerker-Fischer Debate,” Faith and Philosophy 13: 395–401. Jackson, Frank, and R. Pargetter. 1986. “Oughts, Options, and Actualism.” Philosophical Review 95: 233–55. Johnson, Lawrence. 1983. “Can Animals Be Moral Agents?” Ethics and Animals 4: 50–61. Kane, Robert. 2000. “Responses to Bernard Berofsky, John Martin Fischer, and Galen Strawson.” Philosophy and Phenomenological Research 60: 157–67. Kane, Robert. 1999a. “On Free Will, Responsibility, and Indeterminism: Responses to Clarke, Haji, and Mele.” Philosophical Explorations 2: 105–21. Kane, Robert. 1999b. “Responsibility, Luck, and Chance: Reflections on Free Will and Responsibility.” Journal of Philosophy 96: 217–40. Kane, Robert. 1996a. The Significance of Free Will. New York: Oxford University Press. Kane, Robert. 1996b. “Freedom, Responsibility, and Will-Setting.” Philosophical Topics 24: 67–90. Kane, Robert. 1995. “Two Kinds of Incompatibilism.” In Timothy O. Connor, ed., Agents, Causes, and Events Essays on Indeterminism and Free Will. New York: Oxford University Press. Kane, Robert. 1985. Free Will and Values. Albany: State University of New York Press. Kant, Immanuel. [1797] 1987. “Moral Duties.” In Christopher W. Gowans, ed., Moral Dilemmas. New York: Oxford University Press. Kekes, John. 1998. “The Reflexivity of Evil.” In Ellen Frankel Paul, Fred D. Miller, Jr., and Jeffrey Paul, eds., Virtue and Vice. Cambridge: Cambridge University Press. Kekes, John. 1997. Against Liberalism. Ithaca, NY: Cornell University Press. Kekes, John. 1992. “On There Being Some Limits to Morality.” In Ellen Frankel Paul, Fred D. Miller, Jr., and Jeffrey Paul, eds., The Good Life and the Human Good. Cambridge: Cambridge University Press. Kekes, John. 1984. “ ‘Ought Implies Can’ and Two Kinds of Morality.” Philosophical Quarterly 34: 459–67. Lamb, James. 1993. “Evaluative Compatibilism and the Principle of Alternate Possibilities.” Journal of Philosophy 90: 517–27. Lamb, James. 1977. “On a Proof of Incompatibilism.” Philosophical Review 86: 20–35. Lehrer, Keith. 1968. “ ‘Cans’ without ‘If ’s.’ ” Analysis 29: 29–32. Lemmon, E. J. 1987. “Moral Dilemmas.” In Christopher Gowans, ed., Moral Dilemmas. New York: Oxford University Press.

276

Lemos, Ramon. 1980. “Duty and Ignorance.” Southern Journal of Philosophy 24: 301–12. Lewis, David. 1983. “New Work for a Theory of Universals.” Australasian Journal of Philosophy 61: 343–77. Lewis, David. 1981. “Are We Free to Break the Laws?” Theoria 47: 113–21. Lipton, Peter. 1991. Inference to the Best Explanation. London: Routledge. Locke, John. [1690] 1984. Essay Concerning Human Understanding. Oxford: Clarendon Press. Loewer, Barry. 1996. “Freedom from Physics: Quantum Mechanics and Free Will.” Philosophical Topics 24: 92–113. Mann, William. 1983. “Dreams of Immorality.” Philosophy 58: 378–85. Marcus, Ruth Barkan. 1987. “Moral Dilemmas and Consistency.” In Christopher Gowans, ed., Moral Dilemmas. New York: Oxford University Press. McCann, Hugh J. 1998. The Works of Agency: On Human Action, Will, and Freedom. Ithaca, NY: Cornell University Press. McCloskey, H. J. 1973. “Utilitarianism: Two Difficulties.” Philosophical Studies 24: 62–3. McConnell, Terrance. 1987. “Moral Dilemmas and Consistency in Ethics.” In Christopher Gowans, ed., Moral Dilemmas. New York: Oxford University Press. McKay, Thomas, and David Johnson. 1996. “A Reconsideration of an Argument against Compatibilism.” Philosophical Topics 24: 113–22. McKenna, Michael. 1998. “Moral Theory and Modified Compatibilism.” Journal of Philosophical Research 23: 441–58. McKenna, Michael. 1997. “Alternate Possibilities and the Failure of the CounterExample Strategy.” Journal of Social Philosophy 28: 71–85. McNamara, Paul. 1996. “Making Room for Going beyond the Call.” Mind 105: 415–50. Mele, Alfred. 1999a. “Ultimate Responsibility and Dumb Luck.” Social Philosophy and Policy: 16: 274–93. Mele, Alfred. 1999b. “Kane, Luck, and the Significance of Free Will.” Philosophical Explorations 2: 96–104. Mele, Alfred. 1998. Review of Robert Kane, “The Significance of Free Will.” Journal of Philosophy 95: 581–4. Mele, Alfred. 1996a. “Soft Libertarianism and Frankfurt-Style Scenarios.” Philosophical Topics 24: 123–41. Mele, Alfred. 1996b “Internalist Moral Cognitivism and Listlessness.” Ethics 106: 727–53. Mele, Alfred. 1995. Autonomous Agents: From Self-Control to Autonomy. New York: Oxford University Press. Mele, Alfred. 1992. Springs of Action. New York: Oxford University Press. Mele, Alfred, and D. Robb. 1998. “Rescuing Frankfurt-Style Cases.” Philosophical Review 107: 97–112. Mellema, Gregory. 1991. Beyond the Call of Duty: Supererogation, Obligation, and Offence. Albany: State University of New York Press. Milo, R. 1984. Immorality. Princeton, NJ: Princeton University Press. Moore, Michael. 1984. Law and Psychiatry: Rethinking the Relationship. Cambridge: Cambridge University Press.

277

Murphy, Jeffrie, and J. Hampton. 1988. Forgiveness and Mercy. Cambridge: Cambridge University Press. Nagel, Thomas. 1986. The View from Nowhere. New York: Oxford University Press. Nowell-Smith, P. H. 1954. Ethics. Baltimore: Penguin Books Nozick, Robert. 1980. Philosophical Explanations. Cambridge, MA: Belknap Press of Harvard University Press. Nozick, Robert. 1974. Anarchy, State, and Utopia. New York: Basic. Oakley, Justin, and D. Cocking. 1994. “Consequentialism, Moral Responsibility, and the Intention/Foresight Distinction.” Utilitas 6: 201–16. O’Connor, Timothy. 2000. Persons and Causes. New York: Oxford University Press. O’Connor, Timothy. 1996. “Why Agent Causation?” Philosophical Topics 24: 143–58. O’Connor, Timothy. 1993. “On the Transfer of Necessity.” Noˆus 27: 204–18. Oshana, Marina. 1997. “Ascriptions of Responsibility.” American Philosophical Quarterly 34: 71–83. Otsuka, Michael. 1998. “Incompatibilism and the Avoidability of Blame.” Ethics 108: 685–701. Parfit, Derek. 1986. Reasons and Persons. New York: Oxford University Press. Pereboom, Derk. 2001. Living without Free Will. Cambridge: Cambridge University Press. Pereboom, Derk. 2000. “Alternative Possibilities and Causal Histories.” Philosophical Perspectives 14: 119–37. Pereboom, Derk. 1995. “Determinism al Dente,” Noˆus 29: 21–45. Pike, Nelson. 1965. “Divine Omniscience and Voluntary Action.” Philosophical Review 74: 27–46. Prichard, H. A. 1949. Moral Obligation. Oxford: Clarendon Press. Rachels, James. 1999. The Elements of Moral Philosophy. 3rd ed. New York: McGrawHill. Radden, Jennifer. 1996. Divided Minds and Successive Selves. Cambridge, MA: MIT Press. Ravizza, Mark. 1994. “Semi-Compatibilism and the Transfer of Non-Responsibility.” Philosophical Studies 75: 61–93. Reinhart, Kevin. 1983. “Islamic Law as Islamic Ethics.” Journal of Religious Ethics 11: 186–203. Richman, Robert. 1983. God, Free Will, and Morality. Dordrecht, The Netherlands: D. Reidel. Robinson, Richard. 1971. “Ought and Ought Not.” Philosophy 46: 193–202. Ross, William David. 1930. The Right and the Good. New York: Oxford University Press. Rowe, William. 1991. Thomas Reid on Freedom and Morality. Ithaca, NY: Cornell University Press. Russell, Paul. 1992. “Strawson’s Way of Naturalizing Responsibility.” Ethics 102: 287–302. Saka, Paul. 2000. “Ought Does Not Imply Can.” American Philosophical Quarterly 37: 93–105. Sapontzis, S. F. 1987. Morals, Reason, and Animals. Philadelphia: Temple University Press.

278

Saunders, John Turk. 1968. “The Temptations of Powerlessness.” American Philosophical Quarterly 5: 100–8. Schlossberger, Eugene. 1992. Moral Responsibility and Persons. Philadelphia: Temple University Press. Schoeman, Ferdinand. 1978. “Responsibility and the Problem of Induced Desires.” Philosophical Studies 34: 293–301. Shaw, P. D. 1965. “ ‘Ought’ and ‘Can.’ ” Analysis 23: 118–20. Sinnott-Armstrong, Walter. 1988. Moral Dilemmas. Oxford: Basil Blackwell. Sinnott-Armstrong, Walter. 1985. “ ‘Ought to Have’ and ‘Could Have.’ ” Analysis 45: 44–8. Slote, Michael. 1997. “Agent-Based Virtue Ethics.” In Roger Crisp and Michael Slote, eds., Virtue Ethics. New York: Oxford University Press. Slote, Michael. 1992. From Morality to Virtue. New York: Oxford University Press. Slote, Michael. 1990. “Ethics without Free Will.” Social Theory and Practice 16: 369–83. Slote, Michael. 1983. “Admirable Immorality.” In Michael Slote, Goods and Virtues. Oxford: Clarendon Press. Slote, Michael. 1982. “Selective Necessity and the Free Will Problem.” Journal of Philosophy 79: 5–24. Slote, Michael. 1980. “Understanding Free Will.” Journal of Philosophy 77: 136–51. Smilansky, Saul. 2000. Free Will and Illusion. Oxford: Clarendon Press. Smilansky, Saul. 1994.“The Ethical Advantages of Hard Determinism.” Philosophy and Phenomenological Research 54: 355–63. Smith, Holly. 1991. “Varieties of Moral Worth and Moral Credit.” Ethics 101: 279– 303. Smith, Holly. 1983. “Culpable Ignorance.” Philosophical Review 92: 543–71. Smith, Holly. 1978. “Doing the Best One Can.” In Alvin I. Goldman and Jaegwon Kim, eds., Values and Morals. Dordrecht, Holland: D. Reidel. Smith, Holly. 1976. “Dated Rightness and Moral Imperfection.” Philosophical Review 85: 449–87. Smith, Michael. 1994. The Moral Problem. Oxford: Basil Blackwell. Sobel, Jordan Howard. 1976. “Utilitarianism and Past and Future Mistakes.” Noˆus 10: 195–219. Stocker, Michael. 1990. Plural and Conflicting Values. Oxford: Clarendon Press. Stocker, Michael. 1971. “ ‘Ought’ and ‘Can.’ ” Australasian Journal of Philosophy 49: 303–16. Stocker, Michael. 1968. “Supererogation and Duties.” In Nicholas Rescher, ed., Studies in Moral Philosophy. Liverpool: Basil Blackwell. Stockwell, Robert, P. Schachter, and B. Partee. 1973. The Major Syntactic Structures of English. New York: Holt, Rinehart, and Winston. Strawson, Galen. 1994. “The Impossibility of Moral Responsibility.” Philosophical Studies 55: 5–24. Strawson, Galen. 1986. Freedom and Belief. Oxford: Oxford University Press. Strawson, Peter. 1962. “Freedom and Resentment.” Proceedings of the British Academy 48: 1–25. Stump, Eleonore. 1996a. In J. Jordan and D. Howard-Snyder, eds., Faith, Freedom, and Rationality. Lanham, MD: Rowman & Littlefield.

279

Stump, Eleonore. 1996b. “Persons: Identification and Freedom.” Philosophical Topics 24: 183–214. Stump, Eleonore. 1993. “Sanctification, Hardening of the Heart, and Frankfurt’s Concept of the Will.” In John Martin Fischer and Mark Ravizza, eds., Perspectives on Moral Responsibility. Ithaca, NY: Cornell University Press. Swartz, Norman. 1985. The Concept of Physical Law. Cambridge: Cambridge University Press. Thomason, Richmond H. 1981. “Deontic Logic and the Role of Freedom in Moral Deliberation.” In Risto Hilpinen, ed., New Studies in Deontic Logic. Dordrecht, The Netherlands: D. Reidel. Urmson, J. O. 1969. “Saints and Heroes.” In J. Feinberg, ed., Moral Concepts. Oxford: Clarendon Press. van Fraassen, Bas. 1987. “Values and the Heart’s Command.” In Christopher Gowans, ed., Moral Dilemmas. New York: Oxford University Press. van Inwagen, Peter. 2000. “Free Will Remains a Mystery.” Philosophical Perspectives 14: 1–19. van Inwagen, Peter. 1983. An Essay on Free Will. Oxford: Clarendon Press. Vihvelin, Kadri. 2000. “Libertarian Compatibilism.” Philosophical Perspectives 14: 139–66. Vihvelin, Kadri. 1998. Review of John Martin Fischer, The Metaphysics of Free Will. Noˆus 32: 406–20. Vihvelin, Kadri. 1991. “Freedom, Causation, and Counterfactuals.” Philosophical Studies 64: 161–84. Vihvelin, Kadri. 1988. “A Modal Argument for Incompatibilism.” Philosophical Studies 1988: 227–44. Wallace, Jay. 1994. Responsibility and the Moral Sentiments. Cambridge, MA: Harvard University Press. Waller, Bruce. 1998. The Natural Selection of Autonomy. Albany: State University of New York Press. Waller, Bruce. 1990. Freedom without Responsibility. Philadelphia: Temple University Press. Waller, Bruce. 1988. “Free Will Gone out of Control.” Behaviourism 26: 149–67. Warfield, Ted. 1996. “Determinism and Moral Responsibility Are Incompatible.” Philosophical Topics 24: 215–26. Watson, Gary. 1996. “Two Faces of Responsibility.” Philosophical Topics 24: 227–48. Watson, Gary. 1993. “On the Primacy of Character.” In Owen Flanagan and Amelie Oksenberg Rorty, eds., Identity, Character, and Morality: Essays in Moral Psychology. Cambridge, MA: MIT Press. Watson, Gary. 1987. “Free Action and Free Will.” Mind 96: 145–72. Widerker, David. 2000. “Frankfurt’s Attack on the Principle of Alternative Possibilities: A Further Look.” Philosophical Perspectives 14: 181–201. Widerker, David. 1995a. “Libertarianism and Frankfurt’s Attack on the Principle of Alternative Possibilities.” Philosophical Review 104: 247–61. Widerker, David. 1995b. “Libertarian Freedom and the Avoidability of Decisions.” Faith and Philosophy 12: 113–18. Widerker, David. 1991. “Frankfurt on ‘Ought Implies Can’ and Alternative Possibilities.” Analysis 51: 222–4.

280

Widerker, David. 1987. “On an Argument for Incompatibilism.” Analysis 47: 37–41. Wiggens, David. 1973. “Towards a Reasonable Libertarianism.” In T. Honderich, ed., Essays on Freedom of Action. Boston: Routledge & Kegan Paul. Williams, Bernard. 1987. “Ethical Consistency.” In Christopher Gowans, ed., Moral Dilemmas. New York: Oxford University Press. Williams, Bernard. 1985. Ethics and the Limits of Philosophy. London: Fontana. Williams, Bernard. 1981. “Moral Luck.” In Moral Luck. Cambridge: Cambridge University Press. Williams, Patricia. 1993. “Can Beings Whose Ethics Evolved Be Ethical Beings?” In Matthew H. Nitecki and Doris V. Nitecki, eds., Evolutionary Ethics. Albany: State University of New York. Wolf, Susan. 1990. Freedom within Reason. New York: Oxford University Press. Wolf, Susan. 1982. “Moral Saints.” Journal of Philosophy 19: 419–39. Wolf, Susan. 1980. “Asymmetrical Freedom.” Journal of Philosophy 77: 151–66. Wyma, Keith. 1997. “Moral Responsibility and Leeway for Action.” American Philosophical Quarterly 34: 57–70. Yaffe, Gideon. 1999. “ ‘Ought’ Implies ‘Can’ and the Principle of Alternate Possibilities.” Analysis 59: 218–22. Zagzebski, Linda. 2000. “Libertarian Freedom and Alternate Possibilities.” Philosophical Perspectives 14: 231–48. Zimmerman, David. 2000. “Making Do: Troubling Stoic Tendencies in an Otherwise Compelling Theory of Autonomy.” Canadian Journal of Philosophy 30: 25–53. Zimmerman, Michael. 1997. “A Plea for Accuses.” American Philosophical Quarterly 34: 229–43. Zimmerman, Michael. 1996. The Concept of Moral Obligation. Cambridge: Cambridge University Press. Zimmerman, Michael. 1995. “Responsibility Regarding the Unthinkable.” Midwest Studies in Philosophy 20: 204–23. Zimmerman, Michael. 1993. “A Plea for Ambivalence.” Metaphilosophy 24: 382–9. Zimmerman, Michael. 1988. An Essay on Moral Responsibility. Totowa, NJ: Rowman & Littlefield. Zimmerman, Michael. 1987. “Remote Obligation.” American Philosophical Quarterly 24: 199–205.

281

Index

ability, 22–3, 68 volitional, 23 act, deontic definition of, 3, 269 virtue ethical analogues, 199–201 actualism, 53n.12 agency, moral moral competence model of, 139–40 normative, 136–7, 136n.12 normative competence model of, 140 agent causation, 88, 251 alternative possibilities lack of requirement of for responsibility, 26 requirement of for obligatoriness, 31–2, 33–4 requirement of for rightness, 31–2, 33 requirement of for wrongness, 27–31, 32–3 Ansari, Zarar I., 170 antecedent contrastive control, 143–4, 269 and deontic morality, 142–7 and moral responsibility, 142–7 appraisability, moral, 141, 162 distinguished from holding appraisable, 192–3 and supererogatory action, 167–76 appraisals aretaic, 198–205, 206, 208 the autonomy of deontic, 205–11

deontic, 205, 206, 208 a plurality of moral, 42–3, 205–6 Aqvist, Lennart, 29n.4 Aristotle, 160, 183 and teleological metaphysics, 207–8 and virtue ethics, 202–5 asymmetry in agency presuppositions of responsibility and deontic anchors, 8, 123–4 in control requirements for responsibility and deontic anchors, 7, 35, 51 in the self-expressive requirement for responsibility and deontic anchors, 142–7 Audi, Robert, 19n.5, 206n4 Austin, J.L., 68n.10 Baier, Kurt, 171n.5 Baron, Marcia, 227–8 Becker, Lawrence, 224–5 Bergstrom, Lars, 29n.4 Berofsky, Bernard, 62n.5 blameworthiness, 117–8, 141–2 and overridingness, 339–44 overt, 40 varieties, 238–9. See also suberogatory action Brandt, Richard, 171n.5

283

Brock, Dan, and A. Buchanan, 125n.2 Burms, Arnold, xiii “can” hypothetical analysis, 68n.10, 70n.11 sense of in K, 16–24 Carlson, Erik, 64n.8 Carney, Frederick, 170n.4 Castaneda, Hector-Neri, 72n.12, 209n.5 causal determinism doctrine of, 1, 60, 269 incompatibility of with deontic morality, 65–70 incompatibility of with freedom to do otherwise, 59–65 causally undetermined action, 91–2 Chisholm, Roderick, 168, 170 “CK” (ought–can refrain) principle, 28, 29–31, 271 Clarke, Randolph, xiii, 88, 88n.1, 89, 93, 105, 106n.1, 109n.6, 111, 113, 114–15, 115n.8, 130, 181, 248–9 compatibilism, 269 Conee, Earl, 54n.13 Consequence Argument, 60–2 and different senses of ‘ability,’ 62–3 and transfer principle, 60, 63–4 Cooper, Neil, 223n.3, 240 Copp, David, xiii, 7n.5, 15n.1, 28n.3, 37n.3, 50, 50n.11, 171n.5, 196n.16, 221n.1, 223, 225n.4, 238n.10, 241n.12 covert nonconstraining control, 124, 125 and deontic morality, 138–42 and a hierarchical account of responsibility, 133–5 and normative agency, 135–8 Cuypers, Stefaan, xiii, 135n.10 Darwall, Stephen L., 207–8 Davenport, John, xiii De Grazia, David, 164 Dennett, Daniel, 68, 94n.4, 125n.2 deontic control, 246 deontic morality

autonomous appraisals of, 152 definition of, 3, 269 and dilemma concerning freedom, 3, 148 and incompatibility with determinism, 9, 65–70 lack of self-expressive requirement for, 144 not being sources of, 246–7 requirement of alternative possibilities for, 31–3 and semi-compatibilism, 148 summary of costs of being deprived of, 151–2, 245 virtue ethical analogue, 191–200 Dewey, John, 39, 118 Dixon, Beth, 164 Double, Richard, 28n.2, 107n.4, 125n.2 Driver, Julia, 39, 168 dual control, 35, 92, 269 duty, imperfect vs. perfect, 171, 171n.6 Ekstrom, Laura Waddell, 26n.1, 62n.5, 63n.6, 94n.4 Elliott, Carl, 120n.9 explanation, contrastive and deontic morality, 144 and the luck objection to modest R-libertarianism, 117–20 for nondeterministically caused action, 113–17 and responsibility, 143–4 Feinberg, Joel, 120n.9, 183 Feldman, Fred, 15n.1, 17n.3, 30n.6, 30–1, 31n.7, 36n.2, 48n.9, 53n.12, 54n.13, 72n.12, 167n.3, 171, 174n.10, 194n.15, 209n.5, 222n.1 Fields, Lloyd, 7n.5, 171n.5 Finch, Alicia and Ted Warfield, 64 Fischer, John, M., xiii–xiv, 1n.1, 6n.4, 8n.6, 26n.1, 43–4, 60n.1, 62nn.3,4, 94n.4, 95–6, 131n.6 Fischer, John and Mark Ravizza, 6n.4, 28n.2, 29n.5, 44, 69, 113n.7, 126nn.3,4, 132n.7, 138n.14

284

Foot, Philippa, 222n.1 Frankfurt, Harry G., 25, 26, 26n.1, 46, 126, 133nn. 8,9, 133–5, 203n.1, 228n.5 Frankfurt-type examples, 26 and deontic morality, 34–5 and determinism, 66–9 global, 28n.2 and moral responsibility, 25–6 Gallois, Andre, 61n.2 Gensler, Harry, 84n.14 Gert, B., and T.J. Duggan, 120n.9 Ginet, Carl, xiii, 1n.1, 26n.1, 62n.3, 63n.6, 88 Glannon, Walter, 125n.2 global manipulation, 125n.2, 128 Glover, Jonathan, 120n.9 Goble, Lou, 53n.12 Goetz, Stewart, 88n.1 Gowans, Christopher, 54n.13 Greenspan, Patricia S., 53n.12 Hampton, Jean, 160 Hare, R.M., 240 Henderson, G.P., 9n.7, 47n.7 Heyd, David, 167, 173nn.8,9, 173–4 Hill, Thomas E., Jr, 171n.6 Hoffman, Joshua, and G. Rozenkrantz, 62n.3 holding responsible, 184–5, 192 Honderich, Ted, 154n.1, 171n.5 Hudson, Stephen D., 54n.13 Hunt, David, 26n.1 incompatibilism, 270 indeterminism, 2, 270 action-centered, 105–7, 269 doxastic, 101, 105 nonaction-centered. 93–5, 270. See also libertarianism indeterministic theory, 87, 270 Jackson, Frank, and R. Pargetter, 53n.12 Johnson, Lawrence, 164

“K” (‘ought’ implies ‘can’) principle direct threat to from Frankfurt-type examples, 46–7 Fischer’s objection to, 43–5 formulation, 4, 14, 271 and genuine moral dilemmas, 54n.13 Kekes’s objections to, 211–20 and know-how, 16–20 and knowledge that, 20–1 moderate version (“Moderate K”), 22–3 and motivation, 18–19 objection from counterintuitiveness to, 41–3 Pereboom’s objections to, 75–7 and physical ability, 16, 22–3 and psychological ability, 21 Saka’s objections to, 77–84 and self-imposed impossibility, 47–52 strong version (“Strong K”), 23–4 Widerkerian objection to, 37–41 Kane, Robert, xiii, 2n.2, 8n.6, 9n.7, 28n.2, 42n.5, 62n.4, 63n.6, 89, 92, 94n.4, 100n.5, 106–7, 107nn.2,3, 109nn.5,6, 109–10, 113, 113n.7, 115n.8, 119, 128, 130n.5, 131n.6, 181 Kant, Immanuel, 84n.14 Kekes, John, 9n.7, 211–20, 222n.1, 238n.10 Lamb, James, 26n.1, 62n.3 Lehrer, Keith, 68n.10 Lemmon, E.J., 54n.13 Lemos, Ramon, 20 Leplin, Jarett, 28n.3 Lewis, David, 62n.4, 91 libertarianism, 87, 91–2 failure of nonaction-centered versions to accommodate deontic morality, 101–3 M-, 87–8, 270 modest action-centered, 105, 269 modest Kanean, 106–7 modest nonaction-centered, modest R-, 89, 93, 270 R-, 91, 270

285

Lipton, Peter, 113 Locke, John, 26n.1 Loewer, Barry, 109n.6 Mann, William, 9n.7 Marcus, Ruth Barkan, 54n.13 McCann, Hugh J., 88n.1 McCloskey, H.J., 20n.6 McConnell, Terrance, 54n.3 McKay, Thomas and D. Johnson, 61n.2, 64, 64n.7 McKenna, Michael, xiii, 2n.2, 26n.1 McNamara, Paul, 167–8, 171, 172n.7, 210 Mele, Alfred, xiii, 19n.5, 26, 28n.2, 29n.5, 88, 89, 92, 93–5, 94n.4, 96, 101, 101n.6, 107n.4, 108, 109n.6, 125nn.1,2, 127–8, 130, 132n.7, 138n.14, 181, 194n.14, 239 Mele, Alfred, and David Robb, 8n.6, 26n.1, 131n.6 Mellema, Gregory, 171n.6, 173, 176–7 Mill, John Stuart, 207 Milo, R., 171n.5 Modest Meleian Libertarianism, 92–101 and CNC manipulation, 129–31 and failure to accommodate deontic morality, 101–3 Fischer’s objections to, 95–101 Moore, Michael, 171n.5 Murpy, Jeffrie, 160 Nagel, Thomas, 113 “NCC” (no contrastive control) principle, 207–8, 143 Novell-Smith, P.H., 240 Nozick, Robert, 94n.4 Oakley, Justin, and D. Cocking, 7n.5, 171n.5 Objective View, 151, 163 and blameworthiness, 38–9, 49–51, 72–4, 171, 179–80 direct challenge to, 194–6 and praiseworthiness, 171, 172, 179–80 rationales for, 164–7

obligation, moral and “can,” 16–24 fundamentals of, 14–16 objective, 163 overriding of, 14–15, 152, 221–5 prima facie, 15 requirement of alternative possibilities for, 31–2, 34 subjective, 163 O’Connor, Timothy, 61n.2, 63n.6, 64n.8, 88, 109n.6 opportunity, 22, 68 Oshana, Marina, 120n.9 Otsuka, Michael, 26n.1 “ought” all things considered, 224–5 ideal, 15 plain, 14, 221, 222, 224–5, 234, 236, 240–2. See also obligation, moral overridingness and culpability, 239–44 and deontic anchors, 225–6 Kekes’s objections to, 230–3 problems with, 234–7 Slote’s objections to, 226–30 thesis of, 221–5 “OW” (obligation-wrong) principle, 4, 271 and genuine moral dilemmas, 53–4 Pereboom’s objections to, 52–3, 53n.12 “PAP” (alternative possibilities) principle, 25, 54–5, 56–7, 271 Widerker’s W-defense of, 191n.13 Parfit, Derek, 125n.2 Pereboom, Derk, 8n.6, 26n.1, 52–3, 53n.12, 70–7, 107n.4, 131n.6, 155–61 novel version of Frankfurt-type example, 26n.1 Pike, Nelson, 62n.3 possibilism, 53n.12 praiseworthiness, 142, overt, 40–1. See also supererogatory action Prichard, H.A., 165

286

primary deontic properties aretaic analogues of, 199–200 definition of, 3, 270 proattitudes, unsheddability of, 125, 125n.1 proximal control, 23, 93–4, 96, 270 direct, 99 indirect, 99 sufficiency of for deontic morality, 123, 145–6 Rachels, James, 205n.3 Radden, Jennifer, 125n.2 Ravizza, Mark, 26n.1 reactive attitudes, 10, 151 anger, 160 forgiveness, 159–60 subversion of by determinism, 159–60 reflective equilibrium, 185, 190–1 Reinhart, Kevin, 170, 170n.4 responsibility, moral concept of, 120n.9 and contrastive rational explanation, 117–20, 143–4 and dilemma concerning freedom, 1–3 ledger view, 119, 120n.9, 141–2 and luck, 5, 107–10 and normative agency, 135–8 self-expressive requirement of, 10, 118–20, 143, 191–3 ultimate, 106. See also blameworthiness; praiseworthiness Richman, Robert, 9n.7, 47n.7 Robb, David, 26 Robinson, Richard, 47n.7 robust modest R-libertarianism, 105–7 and CNC manipulation, 131–3 and contrastive explanation, 109–13 and deontic morality, 120–1, 122–3 Kanean, 106–7 Luck Objection to, 107–10, 117–20 Rowe, William, 26n.1 Russell, Paul, 120n.9 Saka, Paul, 9n.7, 77–84, 79n.13, 84n.14 Sapontzis, F., 160, 164

Saunders, John Turk, 62n.4 Schlossberger, Eugene, 2n.3 Schoeman, Ferdinand, 28n.2 self-imposed impossibility, 47 and K, 47–52 semi-compatibilism, 6, 7, 70, 148 Shaw, P.D., 9n.7 singular control, 35, 92, 270 Sinnott-Armstrong, Walter, 9n.7, 47n.7, 54n.13 Slote, Michael, 61n.2, 64n.9, 126n.3, 193, 197, 198–201, 221n.1, 226–30, 229n.6, 233n.9, 237 Smilansky, Saul, 153–5 Smith, Holly, 7n.5, 48n.9, 53n.12, 171n.5 Smith, Michael, 223n.3 Sobel, Jordan Howard, 53n.12 Sosa, Ernest, 168 Stocker, Michael, 9n.7, 47n.7, 54n.13, 171n.6 Strawson, Galen, 100, 107n.4 Strawson, Peter, 120n.9 Stump, Eleonore, 26n.1, 92n.3, 126n.3 suberogatory action, 38–9, 167, 168 blameless, 178 objections to, 169–70 and the Objective View, 39, 171 supererogatory action, 38, 167–8, 170–1 and an asymmetry with suberogtory action, 176–7 and the Objective View, 170–3 praiseless, 177–8 Thomason, Richmond H., 53n.12 ultimate control, 96, 250, 270 ultimate origination, 162, 183 Urmson, J.O., 38, 167n.3 van Frassen, Bas, 54n.13 van Inwagen, Peter, 1n.1, 59, 62n.3, 91n.2, 106n.1, 107n.4, 109 Vihvelin, Kadri, 22–3, 22n.7, 23n.8, 26n.1, 62n.4, 66, 68

287

virtue ethics Slote’s agent-focused version, 198–201 subversion of by determinism, 200–5 Watson’s account of, 202–5 Wallace, Jay R., 7n.5, 120n.9, 126n.4, 160, 171n.5, 183–91, 184n.12 Waller, Bruce, 18n.4, 70n.11, 107n.4, 154n.1 “WAP” (wrongness-alternative possibilities) principle, 4, 271 Warfield, Ted, 62n.3 Watson, Gary, 39–40, 120n.9, 126n.4, 183, 183n.11, 191–2, 197, 202–3 “WC” (wrong-can do) principle, 28, 29 Widerker, David, 7n.5, 8n.6, 26n.1, 37–41, 38n.4, 54, 55, 56, 64, 64n.8, 131n.6, 171n.5, 191n.13 Wiggens, David, 62n.3 Williams, Bernard, 54n.13, 222n.1, 233n.9, 224

Williams, Patricia, 141n.15 Wolf, Susan, 7, 24n.5, 221n.1, 239n.11 wrongness and belief in what is wrong, 165–6 and principle CK, 27–8 requirement of alternative possibilities for, 27, 31–2 virtue ethical analogue, 199 Wyma, Keith, 26n.1 Yaffe, Gideon, 54–8, 252n.7 Zagzebski, Linda, 8n.6, 26n.1, 92n.3 Zimmerman, David, 28n.2, 135n.10 Zimmerman, Michael J., 2n.3, 15n.2, 16, 17, 17n.3, 21, 26n.1, 30n.6, 31, 36nn.1,2, 45, 47n.8, 48, 48nn.9,10, 53n.12, 54n.13, 102, 103, 120n.9, 141, 165–6, 165n.1, 168, 171n.5, 177, 182, 194n.15, 195, 205–6

288