From Value to Rightness: Consequentialism, Action-Guidance, and the Perspective-Dependence of Moral Duties 9780367702199, 9780367703844, 9781003146001

Citation preview

From Value to Rightness

This book develops an original version of act-consequentialism. It argues that act-consequentialists should adopt a subjective criterion of rightness. The book develops new arguments which strongly suggest that, according to the best version of act-consequentialism, the rightness of actions depends on expected rather than actual value. Its findings go beyond the debate about consequentialism and touch on important debates in normative ethics and metaethics. The distinction between criterion of rightness and decision procedures addresses how, why, and in which sense moral theories must be implemented by ordinary persons. The discussion of the rationales of “ought” implies “can” leads to the discovery of a hitherto overlooked moral principle, “ought” implies “evidence”, which can be used to show that most prominent moral theories are false. Finally, in the context of discussing cases that are supposed to reveal intuitions that favour either objective or subjective consequentialism, the book argues that which cases are relevant for the discussion of objectivism and subjectivism depends on the type of moral theory we are concerned with (consequentialism, Kantianism, virtue ethics, etc.). From Value to Rightness will be of interest to scholars and advanced students working in normative ethics and metaethics. Vuko Andrić is Lecturer at Bayreuth University, a researcher at the Institute for Futures Studies in Stockholm, and Feodor Lynen Research Fellow of the Alexander von Humboldt Foundation. His research focuses on consequentialism and has been published in international peerreviewed journals such as Acta Analytica, Ethical Theory and Moral Practice, Erkenntnis, Journal of Ethics and Social Philosophy, Journal of Global Ethics, Philosophical Studies, Philosophia, Ratio, Religious Studies, Thought, and Utilitas.

Routledge Studies in Ethics and Moral Theory Edited by Michael D. Waggoner

Virtue, Narrative, and Self Explorations of Character in the Philosophy of Mind and Action Edited by Joseph Ulatowski and Liezl van Zyl The Authority of Virtue Institutions and Character in The Good Society Tristan J. Rogers Getting Our Act Together A Theory of Collective Moral Obligations Anne Schwenkenbecher Love, Justice, and Autonomy Philosophical Perspectives Edited by Rachel Fedock, Raja Rosenhagen, and Michael Kühler From Value to Rightness Consequentialism, Action-Guidance, and the Perspective-Dependence of Moral Duties Vuko Andrić Practical Wisdom Philosophical and Psychological Perspectives Edited by Mario De Caro and Maria Silvia Vaccarezza Virtues, Democracy, and Online Media Ethical and Epistemic Issues Edited by Nancy E. Snow and Maria Silvia Vaccarezza For more information about this series, please visit: www.routledge.com/ Routledge-Studies-in-Ethics-and-Moral-Theory/book-series/SE0423

From Value to Rightness Consequentialism, Action-Guidance, and the Perspective-Dependence of Moral Duties

Vuko Andrić

First published 2021 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 Vuko Andrić The right of Vuko Andrić to be identified as author of this work has been asserted by his in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data A catalog record for this book has been requested ISBN: 978-0-367-70219-9 (hbk) ISBN: 978-0-367-70384-4 (pbk) ISBN: 978-1-003-14600-1 (ebk) Typeset in Sabon by Apex CoVantage, LLC

To Maya and Clara

Contents

Lists of Figures and Tables Acknowledgements

ix

x

Introduction

1

PART 1

Decision-Guidance

13

1 The Problem of Decision-Guidance

15

2 DP and Objective Consequentialism’s Travesty

of Decision-Guidance

27

PART 2

“Ought” Implies “Can”

43

3 In Defence of “Ought” Implies “Can”

45

4 Objective Consequentialism and Ignorance-Induced

Inability

62

5 Four Arguments Against Objective Consequentialism

Related to “Ought” Implies “Can”

77

6 “Ought” Implies “Evidence”

91

7 Is Subjective Consequentialism Compatible

With “Ought” Implies “Can”?

104

viii

Contents

PART 3

Intuitions About Perspective-Dependence

115

8 A New Perspective on the Perspective-Dependence

of Moral Duties

117

9 Objective Consequentialism’s Licensing Dilemma

140

Conclusion

159

References Index

162

169

Figures and Tables

Figures 3.1 When do I have a moral obligation to keep my promise on Wednesday? 3.2 Secondary duties that are compatible with Ought Implies Can 4.1 The structure of the Karpov case 4.2 Frances’s “cans” and “cannots” 4.3 Carlson’s suggestion 4.4 The solution 9.1 Rescue missions

51 53 64 65 71 72 148

Tables 2.1 Four constellations of the problem of decision-guidance 2.2 Three constellations of C2 8.1 Miners

32 32 132

Acknowledgements

This book is a revised, updated, and shortened version of my dissertation, which I completed at Saarland University under the excellent supervision of Christoph Fehige and Attila Tanyi, and with Roger Crisp as an exter­ nal reviewer. Most of the material was first developed in the context of the Workshops in Practical Philosophy at Saarland University, a delight­ ful event that is hosted twice per year by Christoph Fehige and Ulla Wes­ sels. Some of the material was first developed during a one-year stay at the University of Oxford, where Krister Bykvist provided extremely help­ ful comments on a regular basis. I am also very grateful to a number of other people for discussing large parts of the book and/or helping me significantly in other ways: Hannah Altehenger, Kevin Baum, Dorothee Bleisch, Paul Bowman, Annette Duff­ ner, Jan Gertken, Bernward Gesang, Svantje Guinebert, Thorsten Helfer, Frank Hofmann, Brad Hooker, Benjamin Kiesewetter, Sebastian Köhler, Jörg Löschke, Susanne Mantel, Erasmus Mayr, Stephan Padel, Oliver Petersen, Martin Peterson, Daniel Ramöller, Simon Rosenqvist, Helge Rückert, Tyler Paytas, Julius Schälike, Eva Schmidt, Thomas Schmidt, Peter Schulte, Rudolf Schüssler, Stephan Schweitzer, Bill Shaw, Tatjana Višak, Konstantin Weber, Christian Wendelborn, Christopher Woodard, and Joachim Wündisch. The material of the book (as well as related material that eventually did not make it into the book) was presented and discussed at numerous occasions, and I would like to thank all participants. Among the persons who provided helpful feedback at these events were Gustaf Arrhenius, Matthew Braham, Martin Brecher, Erik Carlson, Nicolas Espinoza, Simon Gaus, Tim Henning, Frances Howard-Snyder, Kent Hurtig, Jens Johansson, Anton Leist, Weyma Lübbe, Christoph Lumer, Leo Menges, Elijah Millgram, Andreas Müller, Alastair Norcross, Jonas Olson, Wal­ ter Pfannkuche, Peter Schaber, Walter Sinnott-Armstrong, Peter Stem­ mer, and Torbjörn Tännsjö. But there were more whose comments were helpful, and I apologize for the omissions (I should have kept lists). The events at which the material related to this book has been presented include conferences of the International Society for Utilitarian Studies

Acknowledgements

xi

in 2012 (New York University) and 2014 (Yokohama National Univer­ sity), conferences of the German Society for Analytic Philosophy in 2009 (Bremen University) and 2012 (Konstanz University), the conference of the German Society for Philosophy in 2014 (University of Münster), the Understanding Value conference in 2014 (University of Sheffield), and the Rocky Mountain Ethics Congress (Boulder, CO) in 2019. An earlier ver­ sion of the book was discussed at a book symposium at the University of Erlangen-Nuremberg in October 2019. In addition, material related to the book was discussed at workshops, colloquia, and seminars at Mannheim University, Konstanz University, the Humboldt University of Berlin, three academic retreats at Reichenau Island, meetings of the Ockham Society at Oxford University, and Higher Seminars at Stockholm University and Uppsala University. I would also like to thank Andrew Weckenman and Allie Simmons for their support with the production of the book as well as the reviewers for Routledge. Finally, I would like to thank my family for their support, my wife Maya and my parents Ursula and Dragan Andrić. Some passages of the book revise previously published material. Chap­ ter 4 is based on my “Is Objective Consequentialism Compatible with the Principle that ‘Ought’ Implies ‘Can’?”, published in Philosophia 44(1), 2016, pp. 63–77. Chapter 5 draws heavily on material from my “Objec­ tive Consequentialism and the Rationales of ‘Ought’ Implies ‘Can’”, published in Ratio 30(1), 2017, pp. 72–87. Section 6 of Chapter 8 has previously been published as a discussion note entitled “The Case of the Miners” in Journal of Ethics and Social Philosophy 7(1), 2013, pp. 1–8. Chapter 9 revises material from my “Objective Consequentialism and the Licensing Dilemma”, published in Philosophical Studies 162(3), 2013, pp. 547–566. I am grateful to the publishers – Springer and Wiley – for the permissions to reuse these materials, and I appreciate the licence agreement with the University of Southern California on behalf of Jour­ nal of Ethics and Social Philosophy.

Introduction

Life poses moral challenges. Countless pressing moral questions imbue private and public life. To name but a few: How much money must we donate to the poor? Should we legalize active euthanasia? Is it okay to raise and eat animals? How do your parental duties relate to your profes­ sional duties? Are abortions wrong? What we need in order to meet life’s moral challenges is a moral theory that yields convincing answers to these and other questions. Contemporary ethicists discuss competing moral theories, among them the group of act­ consequentialist theories. The most prominent act-consequentialist theory is: Classical Utilitarianism An act is right if and only if, and because, it leads to a sum total of happiness at least as large as any of its alternatives; otherwise the act is wrong. An act is obligatory if and only if, and because, it leads to a larger sum total of happiness than any of its alternatives.1 Classical utilitarianism yields answers to the moral challenges. Donating a certain amount of money to the poor is obligatory if and only if doing so leads to a larger sum total of happiness than any alternative. An abor­ tion is wrong if and only if there is an alternative that leads to a larger sum total of happiness. And so on. But the donation of which amount of money leads to a larger sum total of happiness than any alternative? Is there an alternative to abortion that leads to a larger sum total of happiness? These questions are hard to answer, not only because of the specific circumstances in which such questions arise, but also for a principled reason. The answers depend on the consequences of our actions, on the consequences our actions do or would have.2 But due to our cognitive limitations, we are often ignorant, mistaken, or uncertain about such consequences. Our cognitive limita­ tions give rise to the worry that consequentialist theories like classical utilitarianism are false because they are cognitively over-demanding. I shall argue in this book that this worry is well-founded.

2

Introduction

However, this does not mean that consequentialism in general is false. Theories like classical utilitarianism are versions of objective conse­ quentialism. According to objective consequentialism, the rightness and wrongness of actions depend on the consequences that actions do in fact have.3 The best-known alternative to objective consequentialism is sub­ jective consequentialism. According to subjective consequentialism, the rightness and wrongness of actions depend on foreseen or foreseeable consequences. According to a subjectivist version of classical utilitarian­ ism, for example, the rightness and wrongness of actions depend on the sum total of happiness that the agent foresees or can foresee the actions to have.4 This book attempts to show that subjective consequentialism is preferable to objective consequentialism. Arguments related to the worry of cognitive over-demandingness speak decisively against objective conse­ quentialism. But no such arguments speak against subjective consequen­ tialism. Moreover, subjective consequentialism coheres well with many independently attractive principles and positions in normative ethics. To prepare the grounds for my arguments, the remainder of the Introduc­ tion addresses topics that will be relevant in the following chapters. Section 1 gives a brief account of normative ethics, the background of our discussion. Section 2 delineates the topic of this book. Section 3 presents the approach I follow to address the topic. Section 4 provides a plan of the book.

1. Normative Ethics Let us start with a look at normative ethics, for consequentialist theo­ ries and doctrines belong to the subject matter of normative ethics. Nor­ mative ethics is a branch of the philosophical discipline of ethics that, roughly speaking, is concerned with what we ought to do.5 Apart from normative ethics, the philosophical discipline of ethics comprises meta­ ethics and applied ethics. In contrast to metaethics, normative ethics does not deal with questions such as what we mean by saying that an action is right but with which types of actions really are right and why they are right. While applied ethics is concerned with more particular questions such as whether abortion is wrong, normative ethics aims to provide a general account of what makes actions right or wrong. It is worth dwelling on the generality. Moral theories and doctrines state necessary and sufficient conditions for an act’s being right or wrong. They identify properties in virtue of which an act is right (or wrong, or obligatory – for brevity, I will henceforth often talk only about rightness). Different moral theories and doctrines identify different right-making properties. The general form of a moral theory is: Necessarily, an act is right if and only if the act has feature F. Necessarily, if an act is right, then the act is right because it has F.

Introduction

3

You can often find these two statements abbreviated and combined in the following way: Necessarily, an act is right iff (if and only if), and because, the act has feature F. For my purposes, it does not matter which phrase best captures the rela­ tion between a right act and its right-making features. Instead of talking about the “in virtue of” relation or using the term “because”, I will occa­ sionally deploy a couple of other expressions. For instance, I might speak of features that ground, provide, or constitute the rightness of an act or even of features to which rightness can be reduced. In stating different moral theories, I will for simplicity often omit the second sentence of the general form. Since it is the purpose of moral theo­ ries to explain the rightness of acts, it is clear that the right-hand side is the explaining part of the biconditional in the first sentence. Neither will I mention, except on rare occasions, the necessity operator.6 This leaves us with the following elliptical construction of the general form of a moral theory, which in my hope will make the text easier to read: An act is right if and only if the act has feature F. So far, I have suggested that normative ethics is about moral theories; and when I talked about “rightness”, this is naturally understood as moral rightness. However, normative ethics, as I understand it, is not only con­ cerned with the moral rightness of actions. Normative ethics also exam­ ines which actions are prudent (and why), what we have reason to do in general (and why), which character traits are virtues (and why), and which states of affairs are good (and why). More controversially, I think that normative political theory can be understood as a branch of norma­ tive ethics, too, a branch that examines which kinds of government are legitimate (and why) and which institutions are just (and why). Generally speaking, then, normative ethics as I understand it is concerned with nor­ mative theories. Normative theories have this general form: Necessarily, thing T has normative property N if and only if, and because, T has feature F. In line with the examples provided earlier, I use “normative” in a broad sense such that not only deontic (or prescriptive) properties (like being right, being obligatory, and being supported by reasons) but also evalu­ ative properties (like being good and being bad) qualify as normative properties. Moreover, I assume that normative relations (such as being obligatory for and being worse than) qualify as normative properties.

4

Introduction

2. The Topic This book adjudicates between objective and subjective consequentialism. The term “consequentialism”, as I use it, covers both a family of normative doctrines and a family of normative theories. Consequentialist theories are all and only those normative theories that entail a version of the consequentialist doctrine. What is the consequentialist doctrine? I will here follow the definition provided by Walter Sinnott-Armstrong (2019): Consequentialism Normative properties depend only on consequences. Classical utilitarianism is a consequentialist theory that entails a version of the consequentialist doctrine which applies to acts: Act-Consequentialism The normative properties of an act depend only on the act’s and its alternatives’ consequences. Act-consequentialism is just one of several versions of the doctrine of consequentialism. The best-known alternative to act-consequentialism is rule-consequentialism, which evaluates rules in terms of consequences and makes the rightness of acts depend on their relations to the best rules (Hooker 2016). Since I will in this book mainly be concerned with act-consequentialism, I will often allow myself to use the terms “con­ sequentialism”, “consequentialist”, and so on, in order to refer to act­ consequentialism, act-consequentialist, and so on. Where a different meaning is intended, this will be clear from the context. Objective and subjective consequentialism, as introduced earlier, are different versions of act-consequentialism. According to objective con­ sequentialism, the normative properties of an act depend only on the act’s and its alternatives’ actual consequences – the consequences these acts would (or do) have if they were (or are) performed. According to subjective consequentialism, as I understand it, the deontic properties of acts (i.e. a subset of the normative properties acts might have – cf. the end of the previous section) depend on expected or rationally expectable consequences. Objective and subjective consequentialism, as I understand them, dis­ agree about the deontic properties of acts, but not about their evaluative properties. Why? In this book, I assume a version of subjective conse­ quentialism that holds, just like objective consequentialism, that the evaluative properties of acts (their goodness, betterness, etc.) depend on actual consequences. Underlying this assumption is my conviction that deontic and evaluative properties differ fundamentally in that deontic

Introduction

5

properties, but not evaluative properties, are action-guiding and there­ fore depend on features of agents (in particular on their epistemic states and conditions – on the agent’s perspective, to use a short phrase). I am a subjectivist when it comes to deontic properties but an objectivist about value. However, since this book examines only the deontic properties of acts, I will not say much in defence of objectivism about values. So, this book addresses the debate over objectivism versus subjectiv­ ism about deontic properties (rather than all normative properties) in the context of act-consequentialism.7 We can narrow down the subject of our discussion even further. I have introduced subjective consequentialism as the position that the deontic properties of acts depend on expected or rationally expectable consequences. However, our discussion will focus only on rationally expectable, as opposed to expected, consequences. Expected consequences are those consequences that an agent actually believes her available acts to have. In contrast, by rationally expectable consequences I mean the consequences that it is reasonable to believe the available acts to have from the perspective of the agent, that is, given the evidence available to the agent. We can term the former position a purely subjectivist variant of subjective consequentialism and the latter an evi­ dentialist variant. I think that evidentialism is superior to pure subjectivism. For a vari­ ety of subjective consequentialism that focuses on actual beliefs would sanction acts dictated by beliefs that are bizarre, such as crazy religious beliefs. Suppose, for instance, that a lunatic believes – without evidence and contrary to the facts – that Satan will torture all humans forever if she does not shoot all the customers in a shopping centre (Bykvist 2010, p. 85). According to a purely subjectivist version of an otherwise plau­ sible moral theory, the lunatic has a moral duty to kill the customers.8 But that verdict is not plausible.9 So, I will focus on the evidentialist version of subjective consequentialism. We can now formulate the definitions of objective and subjective consequentialism that will concern us in this book: Objective Consequentialism An act is right if and only if the act’s outcome is at least as good as the outcome of any of the act’s alternatives; otherwise the act is wrong. An act is obligatory if and only if its outcome is better than the out­ come of any of its alternatives.10 Subjective Consequentialism An act is right if and only if it has at least as much expected value as any of its alternatives; otherwise the act is wrong. An act is obligatory if and only if it has higher expected value than any of its alternatives.

6

Introduction

By an act’s outcome, I mean everything that is the case if the act is per­ formed. An act’s expected value is the sum of its possible outcomes’ value– probability–products: Each act has possible outcomes. Each outcome has a value – assigned to it by a consequentialist theory of value – and is associated with a probability, i.e. a likelihood of its occurrence given the act is performed – assigned to it on the basis of what the agent has reason to believe.11 To calculate an act’s expected value, you have to multiply each possible outcome’s value with its respective probability and sum up all products. To simplify things, I will often just say that, according to objective con­ sequentialism, you ought to maximize actual value, whereas according to subjective consequentialism, you ought to maximize expected value. Let me conclude this section with some remarks about other issues that the reader might hope to but will hardly find addressed in this book. First of all, I will adjudicate only between objective and subjective con­ sequentialism, but there are other positions regarding the perspectivedependence of deontic properties: relativism, contextualism, and what I call the mixed view (according to which there are both objective and subjective deontic properties, and we need to disambiguate). For the most part, I am going to ignore these positions, pretending for simplicity that objectivism and subjectivism are the only two options.12 Second, I will neither engage in the debate over actualism and possibilism nor discuss related questions, even though these issues are sometimes relevant in the context of the debate over objective and subjective consequentialism.13 Finally, there are numerous issues that are relevant for the general understanding and assessment of consequentialism, such as the merits of different value theories, the debate over agent-neutrality versus agentrelativity, and maximizing versus satisficing versions of consequentialism (cf. Sinnott-Armstrong 2019). Lack of space does not allow me to address any of these issues at length. This book merely attempts to adjudicate between objective and subjective consequentialism as defined previously.

3. The Approach In order to find out whether act-consequentialists should be subjectivists or objectivists about rightness, we need to further our understanding of what rightness is. In order to do so, I suggest we look at the functions or purposes thought and talk about the rightness of acts can have. I see three main purposes. The first purpose is to guide your action. If you wonder about what to do, it is helpful to learn that it would be right or that you ought to perform a certain action and that it would be wrong to perform some other action. Since right, wrong, and ought are action-guiding concepts, the provision of action-guidance is the first function served by talk or thought about rightness. Accordingly, it is often held (though also sometimes disputed)

Introduction

7

that normative theories serve an action-guiding function, and I accept this view.14 Parts 1 and 2 of this book can be viewed as attempts to get clearer on the notion of action-guidance and assess objective and subjec­ tive consequentialism on that basis. Second, we often express blame by calling an action wrong. Many philosophers think that an agent can be blameworthy only for wrong actions (and praiseworthy only for right actions). One might even sug­ gest that talk about wrongness is reducible to talk about blameworthi­ ness. Like many act-consequentialists, I reject these views (cf. Shaw 1999, pp. 134–138). Accordingly, my case for subjective, and against objective, consequentialism is not based on the idea that a subjectivist criterion of rightness is better suited for the assessment of agents. (For this idea, see, e.g. Moore 1912, pp. 98–101.) Nonetheless, at some places in the book I will point out important implications of the view that an agent can be blameworthy only for wrong actions. Third, judging an act to be right or wrong can be used as shorthand for saying that the act is or fails to be, respectively, as strongly favoured by reasons as any of its alternatives. An act is obligatory, according to this façon de parler, if and only if it is more strongly favoured by reasons than any of its alternatives. I am sympathetic to this reductionist account that explains moral duties (the rightness, wrongness, and obligatoriness of actions) in terms of reasons, and I will assume it in some arguments. Why do I accept the claim that normative theories should be actionguiding and the explanatory primacy of reasons over duties but reject the alleged connection between wrongness and blameworthiness? After all, all three ideas are suggested by everyday moral discourse. My reason is that I view act-consequentialism primarily as a theory of practical reason and only secondarily as a moral theory and I think that the claims about action-guidingness and the primacy of reasons, but not the claim about wrongness and blameworthiness, fit well with this view. Let me elaborate. Everyday discourse gives rise to a distinction between what we morally ought to do and what we ought to do for non-moral reasons, especially for prudential reasons. Moreover, many people seem to think that what morality requires you to do can be at odds with what you ought to do all things considered, that is, when taking into account both moral and non-moral considerations. For instance, some people think that from a moral point of view, you ought not to fiddle your taxes, but the same people, at the same time, also think that you have most reason, all things considered, to evade taxes if you think that you can get away with it. By contrast, I am sympathetic to the view that impartial values lie at the foundation of the best theory of practical reason and that there is, accord­ ingly, no deep conflict between morality and practical reason. Conversely, I do not think that a (maximizing) version of act-consequentialism can be defended as the best moral theory unless it can also be defended as the best theory of practical reason. Prudence, on my favoured view, is just

8

Introduction

the part of practical reason/morality that is concerned with actions that affect only the agent herself. Now, when it comes to theories of practical reason, it seems plausible that talk about ought or right can be reduced to talk about what one has most reason to do all things considered. It seems strange to think that an act might have the property of being right all things considered in addition to its being most favoured by reasons. This is why I accept the explanatory primacy of reasons over duties. By contrast, when it comes to moral duties as they are recognized by common-sense morality, it is prima facie significantly less clear that talk about what one has a moral duty to do is reducible to talk about what one has (moral) reason to do, as the idea of supererogatory actions illustrates. The action-guiding function, too, seems less dubious in contexts of practical reason than in everyday moral contexts. If morality is conceptu­ alized in such a way that it is possible for you to have decisive all-things­ considered reason to violate a moral duty, then better think twice before basing your decisions on what you morally ought to do. By contrast, since reasons and “oughts” do not come apart in this way when it comes to what is required by practical reason, there does not seem to be a prob­ lem with basing your decisions on these requirements. Finally, I do not accept the claim that you can be blameworthy only for wrong actions, because if “wrong” is (primarily) understood along the lines of contrary to a requirement of practical reason, then I do not see good reasons for thinking that this claim is true. The alleged connection is most plausible if “wrong” is understood along the lines of contrary to a moral duty where “moral duty” refers a duty of positive morality (analo­ gous to positive law) (cf. Crisp 2006, Chapter 1). But I am not concerned with such duties (though I shall say a bit about them in Chapter 3). Should we consider act-consequentialism a moral doctrine at all? I think that we should – and I will do so in the remainder of the book. I am sympathetic to the idea that some (agent-neutral) act-consequentialist theory can, in a first step, be shown to be the best theory of practical reason.15 Once this has been accomplished, we can go on to establish that this version of act-consequentialism is, for two reasons, also the best moral theory. First, many philosophers think that moral requirements are “authoritative” in that they provide all-things-considered duties. This would be true of the requirements issued by an act-consequentialist theory that is the best theory of practical reason. Second, agent-neutral act-consequentialist theories, like utilitarianism, are impartial, and this coheres well with the widely observed impartial nature of morality.

4. The Plan of the Book The book examines whether the act-consequentialist criterion of right­ ness is objective or subjective. Part 1 (Chapters 1 and 2) considers the

Introduction

9

relevance of decision procedures to that question. Chapter 1 distinguishes what I call decision-guidance from other forms of action-guidance and introduces the problem of decision-guidance, which is the problem of how to decide what to do if you do not know what you ought to do. The problem of decision-guidance has often been accepted as a starting point for a widely recognized argument for subjective consequentialism: the argument from decision-guidance. The argument basically says that subjective consequentialism is preferable to objective consequentialism because only subjective consequentialism is decision-guiding. As it turns out, however, the argument from decision-guidance fails because subjec­ tive consequentialism has its own problem of decision-guidance. In Chapter 2, I present a solution to the problem of decision-guidance of both objective and subjective consequentialism. However, even though objective consequentialism escapes the problem of decision-guidance, my solution reveals that objective consequentialism faces a related problem, which I call the Travesty of Decision-Guidance Problem. The Travesty of Decision-Guidance Problem results from a more basic flaw of objective consequentialism, namely, its indifference towards uncertainty. Part 2 (Chapters 3–7) focuses on the principle that “ought” implies “can”. In Chapter 3, I introduce and defend this principle. Chapter 4 examines an argument that has been put forward by Frances HowardSnyder and according to which objective consequentialism is false because it is incompatible with the principle that “ought” implies “can”. I argue that Howard-Snyder’s argument fails. Although Howard-Snyder’s argument fails, the idea it is based upon sheds light on the deficiencies of objective consequentialism, as I argue in Chapter 5. While objective consequentialism is compatible with “ought” implies “can”, it is incompatible with the considerations underlying this principle. This allows for four new arguments, which build on the rationales for “ought” implies “can”: the Argument from Impossibility Comparisons, the Argument from Action-Guidance, the Argument from Blameworthiness and Fairness, and the Argument from Reasons. Chapter 6 develops a fifth argument against objective consequentialism, which goes as follows: Since important considerations that speak for “ought” implies “can” also support the principle that “ought” implies “evidence” and since there are no good objections to “ought” implies “evi­ dence”, we should accept this principle. According to “ought” implies “evidence”, lack of factual evidence is a defeater of moral obligation. It fol­ lows from this principle that objective consequentialism is false. Chapter 7 dwells on the principle that “ought” implies “can”. Subjec­ tive consequentialism has also been accused of being incompatible with this principle. I defend subjective consequentialism against the charge of being incompatible with “ought” implies “can”. The result of Parts 1 and 2 is that considerations of action-guidance do not count against subjective consequentialism but that several such

10

Introduction

considerations speak against objective consequentialism: the Travesty of Decision-Guidance Problem, the underlying problem of indifference towards uncertainty, and objective consequentialism’s incompatibility with the rationales for “ought” implies “can” and with “ought” implies “evidence”. Part 3 comprises Chapters 8 and 9 and focuses on intuitions about perspective-dependence. In Chapter 8, I propose a new perspective on the perspective-dependence of moral duties. According to the new perspec­ tive, we should decide the question whether moral duties are perspectivedependent by focusing on cases on the relevance of which objective and subjective consequentialists can agree. It turns out that the relevant cases favour subjective consequentialism. Thus, we get a further argument in favour of subjective consequentialism, which I call the New Perspective Argument. Objective consequentialists will deny that the relevant cases speak in favour of subjective consequentialism. They will try to mimic the intui­ tively plausible subjective consequentialist output. In Chapter 9, I argue that no attempt at such mimicry is successful. Objective consequentialists face what I will call the Licensing Dilemma. In the Conclusion, I sum up the result of this book, which is, in one sen­ tence, that subjective consequentialism is more plausible than objective consequentialism. I comment on what this means in terms of the overall plausibility of act-consequentialism and in particular with regard to the objection from cognitive over-demandingness. Finally, I ask which of the findings of this book would be interesting even if act-consequentialism were false. I conclude by listing these findings.

Notes 1. Classical utilitarianism is normally ascribed to Jeremy Bentham (1789/1961), John Stuart Mill (1861/1998), and Henry Sidgwick (1907). It is, however, controversial to which extent these philosophers really subscribed to classi­ cal utilitarianism as defined in the main text. Notice also that, here as well as throughout the book, I use “right” in the sense of “permissible”. 2. It seems natural, at least in most contexts, to refer to the consequences that acts do (or did or will) have if the acts in question are (or were or will be) performed and to the consequences that acts would have if the acts in ques­ tion are not performed. However, in order to avoid cluttering the exposition, I will for the most part ignore these differences. 3. For simplicity, I will ignore versions of consequentialism that focus on the objective probability with which actions lead to consequences. Such ver­ sions of consequentialism are arguably best classified as versions of objective consequentialism. 4. I use the names “objective consequentialism” and “subjective consequentialism” because these names are established in the debate. It would be more accu­ rate, I think, to speak of “objectivist” and “subjectivist” consequentialism. 5. For overviews of what normative ethics is and does, see Timmons 2002, esp. Chapter 1.

Introduction

11

6. I will make exceptions in Part 2 of the book because I will have to distin­ guish between necessity and possibility in my discussion of the principle that “ought” implies “can”. 7. There is a general debate over whether deontic properties depend on the per­ spectives of agents, cf. Part 3 of this book. Before its recent blossoming, the debate has seen important contributions by C.D. Broad (1952/53, Chapter 3), H.A. Prichard (1932), G.E. Moore (1912, Chapters 1 and 2), W.D. Ross (1939, Chapter 7), and B. Russell (1910, Chapter 1, Section 3). 8. Notice that I use “duty” and “obligation” synonymously throughout the book. Likewise, I will use expressions like “an act is obligatory” and “an act ought to be performed” interchangeably. 9. Similarly, as Zimmerman 2008, p.  14, points out, Hitler might well have done nothing wrong on pure subjectivism. 10. The definition assumes, for the sake of simplicity, that there are no incom­ mensurable values. If there are incommensurable values, we have to change this definition (and related definitions) such that an act is right if and only if the outcome is not worse than (rather than at least as good as) the outcome of any alternative. 11. Evidentialism is here assumed in the claim that expected value depends on what the agent has reason to believe, whereas according to pure subjectivism expected value depends on what the agent actually believes. Notice also that “expected value” is a widely used technical term. A more fitting phrase, in the case of evidentialism, would be “rationally expectable value”. However, even this expression is somewhat misleading. For in many cases, it is certain that the expected or rationally expectable value of an act will be different from the actual value that occurs if the act is realized – e.g., when an act has only two possible outcomes, a good one and an equally probable bad one such that the act’s expected value is neutral. 12. I touch on these positions (esp. on the mixed view), in particular in Part 3 of the book, but I do not discuss them at length. On relativism and contextualism, see Kolodny and MacFarlane 2010 and Henning 2014, respectively; on the mixed view, which is spelled out in different ways by different pro­ ponents, see, e.g., Dancy 2000, Parfit 2011, and Schroeder 2009. Another new perspective in the debate has been advanced by Martin Peterson (2013, Chapter 5), who proposes a combination of objective and subjective conse­ quentialism by assigning them to different moral dimensions, i.e., irreducible kinds of aspects. However, rather than solving any problems of objective or subjective consequentialism, this position is haunted by all problems of objective and subjective consequentialism; see Andrić and Tanyi (2016). 13. On actualism, possibilism, and related issues, see Timmerman and Cohen 2020. The actualism/possibilism debate is important, for example, in the context of criticisms of subjectivism that focus on the advice of bystanders who are better informed than the agent (cf. Kolodny and MacFarlane 2010; Kiesewetter 2011). Moreover, some versions of actualism (those defended in Sobel 1976 and Goldman 1976) might have resources to resist the arguments examined in Chapters 4–6 that are not available to other versions of actualism (like the version defended in Jackson and Pargetter 1986). Possibilists might have resources to resist these arguments if possibilism is conjoined with certain accounts of ability regarding compound actions (cf. Goldman 1978, Section III). This means that the arguments examined in Chapters 4–6 can be used to adjudicate between different versions of actualism and possibilism. 14. For extensive lists of references, see Smith 2018, esp. Chapter 3, footnotes 34 and 54, for the “pro action-guidance” view and footnote 46 for refer­ ences on the dismissive view. Smith distinguishes three views: the “pro

12

Introduction

action-guidance” camp is divided by her according to whether or not sup­ plementary rules are needed in order to implement moral theories. On this distinction, this book defends the “pro action-guidance but supplementary principles are needed” view, as will emerge in the first part of the book. 15. Promising arguments that are relevant in this context can be found, for example, in Fehige 2004, Huemer 2005, pp.  184–187, Lazari-Radek and Singer 2012; 2014, esp. Chapter 7, Kagan 1989, pp.  386–393, and Singer 1993, Chapter 12; 1995.

Part 1

Decision-Guidance

1

The Problem of Decision-Guidance

This and the next part of the book examine considerations that are related to action-guidance in order to adjudicate between objective and subjec­ tive consequentialism. The term “action-guidance” will be explained and elaborated at several places. In this and the next chapter, I am concerned with decision-guidance, which I understand to be a special kind of actionguidance. This chapter introduces what I call the problem of decisionguidance. Followers of objective consequentialism face this problem when they do not know how to decide what to do because they do not know which of the actions available to them maximizes actual value and is, hence, obligatory according to objective consequentialism. What is needed in order to solve the problem are rules that tell agents with cognitive limi­ tations how they ought to decide under ignorance and uncertainty. According to some philosophers, subjective consequentialism is supe­ rior to objective consequentialism because the first is decision-guiding but the second is not. However, I think that the problem of decisionguidance is not a problem exclusively for objective consequentialism. The problem also concerns subjective consequentialism, as Fred Feldman has shown. I will present and defend Feldman’s argument in the next section. The plan of the chapter is as follows. I elaborate on the problem of decision-guidance in Sections 1 and 2. Section 3 determines the relevance of the problem of decision-guidance for moral theories. The problem of decision-guidance will be distinguished from other problems that concern the action-guidingness of consequentialism in Section 4. Final remarks follow in Section 5.

1. The Problem of Decision-Guidance for Objective and Subjective Consequentialism Here are two quotes – the first from Shelly Kagan (1998, p. 64), the sec­ ond from Fred Feldman (2006, p. 50) – which introduce the problem of decision-guidance for objective consequentialism: Perhaps the most common objection to consequentialism is this: it is impossible to know the future. This means that you will never be

16

Decision-Guidance

absolutely certain as to what all the consequences of your act will be. . . . In fact lacking a crystal ball, how could you possibly tell what all the effects of your act will be? So how can we tell which act will lead to the best results overall – counting all the results? This seems to mean that consequentialism will be unusable as a moral guide to action. Utilitarians are attracted to the idea that an act is morally right iff it leads to the best outcome. But critics have pointed out that in many cases we cannot determine which of our alternatives in fact would lead to the best outcome. So we can’t use the classic prin­ ciple to determine what we should do. It’s not “practical”; it’s not “action-guiding”.

Objective consequentialism’s problem of decision-guidance seems to be this: Since you do not know the actual consequences of the acts available to you (after all, you would have to take into account all consequences until the end of days of any of the countless acts available to you), you do not know which of them are right and which are wrong. And therefore you cannot base your decisions on the objective consequentialist criterion of rightness, i.e. you do not know how to choose a right act. So essen­ tially, the problem is that you do not know how to decide; this is why I call it the problem of decision-guidance. James Lenman (2000) has illustrated how serious the ignorance under­ lying objective consequentialism’s problem of decision-guidance is. As a starting point, notice that all the consequences of an act are relevant according to objective consequentialism, however distant they may be. Two considerations are especially significant in this context. First, there are many identity-affecting actions. “These are actions that make a dif­ ference to the identities of future persons and these differences are apt to amplify exponentially down the generations” (Lenman 2000, p. 346). These actions have massive causal ramifications that cannot be predicted. Second, apparently insignificant actions can have huge impacts, so-called butterfly effects, because there are causal systems that are extremely sen­ sitive to very small and locally triggered changes (ibid., p.  347). Such systems include the weather and the finance markets. After illustrating the relevance of identity-affecting actions and sensi­ tive causal systems by means of some examples, Lenman concludes: The worry is not that our certainty [about the consequences of our actions] is imperfect, but that we do not have a clue about the overall consequences of many of our actions. Or rather – for let us be precise – a clue is precisely what we do have, but it is a clue of bewil­ dering insignificance bordering on uselessness – like a detective’s discovery of a fragment of evidence pointing inconclusively to the

The Problem of Decision-Guidance

17

murderer’s having been under seven feet tall. We may not be strictly without a clue, but we are virtually without a clue. (Lenman 2000, pp. 349f) Let us now turn to subjective consequentialism. In his article “Actual Util­ ity, the Objection from Impracticality, and the Move to Expected Utility”, Fred Feldman (2006) argues that the often-suggested move to expected value does not provide a cogent reply to the objection that objective con­ sequentialism is not action-guiding.1 I think that Feldman is right: sub­ jective consequentialism has its own problem of decision-guidance and many philosophers do not seem to be aware of this.2 Feldman’s basic thought is this: In order to base your decisions on the subjective conse­ quentialist criterion of rightness, you would have to proceed in any given choice situation in something like the following way: 1. 2. 3. 4. 5. 6. 7. 8. 9.

List your alternatives. List all the possible outcomes of the first alternative. For each outcome of the first alternative, write down its value. For each outcome of the first alternative, write down its probability given the alternative on your evidence. For each possible outcome of the first alternative, multiply the value of the outcome by its probability on the first alternative. Sum these products. This sum is the expected value of the first alternative. Repeat Steps 2–6 for each of the other alternatives. Identify the acts that maximize expected value. Perform one of those acts.3

The problem is that every one of these steps is problematic. The first step is problematic, as Feldman (2006, p. 58) explains, because there are millions of alternatives available to you in almost every choice situ­ ation. If you cannot list your alternatives, then you cannot perform the first step. Step 2, Feldman (2006, pp.  58f.) argues, is not less problematic. An ordinary act might have thousands of possible outcomes. But agents cer­ tainly cannot list thousands of outcomes of each available act. Notice also that subjective consequentialism in this regard seems to be worse than objective consequentialism. For while the objective consequentialist only has to list the consequences each act would in fact have, the subjec­ tive consequentialist has to list these consequences plus the other conse­ quences each act could have. Feldman presents similarly motivated worries for the other steps. Feldman is right that no human agent could follow the aforementioned instructions in order to determine how she ought to decide. However,

18

Decision-Guidance

Hilary Greaves (2016) has recently argued that Feldman’s argument fails nonetheless: Feldman (2006) argues that in fact we are more clueless about sub­ jective c-betterness [i.e., about an act’s expected value being higher than the expected value of another act] than about its objective ana­ logue. But Feldman’s argument assumes that in order to estimate which of two acts has the higher expected value (and by how much), we need to estimate what the expected value of each act is. This lat­ ter task is indeed massively more demanding, but (pace Feldman) is unnecessary. (Greaves 2016, p. 318, footnote 12) Greaves is right that in order to estimate which of two alternative acts has the higher expected value, we do not necessarily need to calculate any expected values. However, this criticism changes the subject. If I base my decision on my estimation about which of two acts has the higher expected value, then I am not basing it on the subjective consequen­ tialist criterion of rightness. Rather than following the rule “Do what maximizes expected value!”, I follow a rule along the lines of “Do what maximizes expected value according to your estimation!”. Instead of defending the usability of subjective consequentialism against Feldman’s objection, Greaves in fact suggests a different decision rule.4 I conclude, with Feldman, that subjective consequentialism has its own problem of decision-guidance.

2. A General Analysis of the Problem of Decision-Guidance How can we spell out the problem of decision-guidance in general terms? The first thing to notice is that the problem of decision-guidance results from factual, rather than normative, ignorance. The complaint is that someone who accepts and understands objective consequentialism does not know the consequences her options would in fact have. The problem is not how or why to become a follower of objective consequentialism in the first place. Nor does the problem concern details of the moral theory in question such as the contents of the best theory of value. Secondly, the person knows that she ought to do what would in fact have the best consequences, but she cannot tell or determine which of the acts available to her fits this description, that is, she cannot iden­ tify, among her alternatives, the one that would in fact have the best consequences. It is illuminating, I think, to point out that objective con­ sequentialism has this problem qua moral theory. Objective consequen­ tialism states the right-making features of acts in general terms: objective consequentialism tells you what it takes for acts in general to be right (namely, to maximize actual value), but it does not tell you which of the

The Problem of Decision-Guidance

19

acts available to you in a concrete situation are right (that is, it does not tell you which of the available acts maximize value). In order to find out whether a particular act a1 that you can perform in a concrete situation is obligatory, you have to subsume a1 under the general formula of a moral theory in the following way: Subsumption (1) [Moral theory:] An act is obligatory iff the act has feature F. (2) [Application of the theory to the concrete choice situation:] a1 has F. -------------------------------------------------------------------------------(3) [Information necessary and sufficient to make a decision:] a1 is obligatory. Let us locate the problem of decision-guidance within the subsumption steps. If you know that a1 is obligatory – that is, if you are able to infer (3) – then there is no problem of deciding what to do. Now since, as I have said earlier, we are dealing with factual rather than normative ignorance, it cannot be step (1) that the problem of decision-guidance is about. Rather, the problem of decision-guidance must be about step (2): the problem results from your inability to assign ought-making features to your various alternatives. Thirdly, the problem of decision-guidance results from factual igno­ rance rather than false factual beliefs. Holly Smith (2018) recently pro­ posed a solution to the problem of how to implement moral theories that attempt to take into account both problems (ignorance and false belief) simultaneously. In my opinion, this is a mistake. When you are mistaken about which of your alternatives have ought-making features, then you cannot, at the same time, be unable to assign ought-making features to your alternatives. A false assignment of ought-making fea­ tures is, after all, an assignment of ought-making features! This sug­ gests that the best way of solving the problem of decision-guidance is different from the best way of solving the problem of false factual belief. Fourthly, since the problem of decision-guidance is to choose among one’s available acts given one’s inability to assign ought-making, rightmaking, and wrong-making features to them despite one’s acceptance and understanding of a given moral theory, the problem can be expressed by means of this question: Q: Given that I do not know the ought-making, right-making, and wrong-making features of the acts available to me, which act ought I to choose?

20

Decision-Guidance

Perhaps you do not feel the pull of Q. It may be that you already accept some rules you use for decisions when you do not know the right-making or wrong-making features of your alternatives. The problem might thus appear to you to be trivial. Or you might wonder whether you really understand it, or whether there is a problem at all. For these reasons, I think the problem of decision-guidance can best be illustrated by means of a hypothetical scenario. Suppose you are an AI programmer. Your job is to get an intelligent robot going. Since you are a good person, you will equip the robot with what you think is a decent moral code. Let us say that the code consists of rules about prima facie duties to the effect that the robot ought not to lie, ought to obey its master, and so on, plus some fancy algorithm for solv­ ing conflicts between these rules. You turn the robot on and issue your first command: “Tell me, robot, is it raining?” The robot does not know if it is raining. But it has to obey the master; it has to answer your ques­ tion. Moreover, the answer has to be true since the robot is not allowed to lie. According to the programmed moral code, the robot has to say either “yes” or “no”. It ought to say “yes” if it is raining, and it ought to say “no” if it is not raining. But the robot does not know which of the two answers has the right-making feature of being true. To complete your job as an AI programmer, you have to supplement the robot’s moral code in order to make the robot capable of acting in this situation. The robot’s moral code is silent on what to do in this situ­ ation. What the robot seems to need is an additional programme that tells it what to do when it does not know the right-making features of its options. Now, to feel the pull of Q, imagine you happen to be in a situ­ ation comparable to that of the robot. That is, suppose you would not already accept rules that tell you what to do when you do not know the right-making features of your options. This is when you would ask Q. Fifthly, there are two constellations in which the problem of decisionguidance can manifest itself. As I said previously, the problem is with step (2) of the subsumption: You lack relevant beliefs about the oughtmaking, right-making, and wrong-making features of the acts available to you. One constellation is that you lack any belief about which oughtmaking, right-making, and wrong-making features your alternatives have. The second constellation is that you have uncertain beliefs to this effect, by which I mean that the contents of your beliefs have the form: The probability of a1 having F is p1, the probability of a2 having F is p2, . . ., the probability of an having F is pn; with 0 < p1, p2, . . . pn < 1.5 The AI robot can be used to illustrate either case. It is possible that the robot has not formed any belief about whether it is raining. But one could also imagine that the robot believes that, say, the probability is 0.5 that it is raining and 0.5 that it is not raining. In either case, the robot cannot

The Problem of Decision-Guidance

21

assign F (= being a truthful utterance and the obedience of the master) to the available acts (saying “yes” and saying “no”). Is there a third constellation? Perhaps. You might want to distinguish between different cognitive states in terms of their strengths or between degrees of belief. What about, for example, somebody who has “a slight tendency” to think that one of the acts available to her maximizes actual value? So far, I have only been concerned with what is sometimes called “outright belief” and the uncertainty I was speaking about features in the contents of beliefs rather than in their strengths. Perhaps, though, claims about strengths/degrees of beliefs or credences can be “translated” into claims about outright belief. For example, it might be justified to treat a person with a credence of c in the proposition that it is raining as if the person had an outright belief that there is a probability p that it is raining, where p would have to be a function of c. It is beyond the scope of this book to examine if this manoeuvre is possible, and I will ignore credences in what follows. However, three important points should be clarified and kept in mind. First, when I talk about “uncertain belief” I mean outright belief with a content of the aforementioned form. Second, I assume that humans form outright beliefs all the time and that such outright beliefs are often rational. Third, I assume that many beliefs (rationally) held by humans are certain (i.e. not uncertain in the content-related sense introduced earlier) even though they could turn out to be false. For example, I believe that I had oatmeal for breakfast rather than that it is extremely probable that I had oatmeal for breakfast. (This belief could turn out to be false – perhaps I was abducted by aliens and brainwashed into thinking that I had oat­ meal for breakfast.) Accordingly, I believe that telling somebody that I did not have oatmeal for breakfast would be lying, not that it would most likely be lying.6 This concludes my analysis of the problem of decision-guidance. Before I come to the distinction between this problem and closely related prob­ lems in Section 4, let me briefly explain why the problem is relevant at all for the assessment of moral theories.

3. The Relevance of Decision-Guidance for Moral Theories The problem of decision-guidance is the problem of answering Q. As long as no answer has been provided, an agent who asks Q is left clue­ less. I think that, if an answer cannot be provided in the case of a moral theory, then this speaks against this theory. For moral theories ought to be decision-guiding for two reasons. Firstly, we certainly have an interest in moral theories being decisionguiding. We want to know how to decide in cases where we cannot assign right- and wrong-making features to the available acts. So, the decision­ guidingness of moral theories serves a practical purpose.

22

Decision-Guidance

Secondly, there is also a theoretical point. Different reactions to situa­ tions in which you ask yourself Q vary in rationality. For example, when you do not know whether it is right to tell a lie in a particular situation, it seems irrational, other things being equal, to try to draw an apple pie. A theory of decision-guidance attempts to explain the irrationality of your reaction by stating general facts about what makes decisions or actions irrational. The rationality or irrationality of an action, in this sense, is different from its moral rightness or wrongness (and different from its being, or failing to be, the thing that you ought to do all things considered). However, being rational and being irrational are normative properties of acts, and thus belong to the things that it is natural to expect moral theories to shed light on. Some qualificatory remarks with respect to the second point are in order. Just as the best moral theory could in principle turn out to be dif­ ferent from the best theory of practical reason (see the Introduction, Sec­ tion 3), so the best theory about what makes actions rational, in any given situation, could turn out to be different from both the best moral theory and the best theory of practical reason. In other words, there might be situations in which the moral thing to do and the reasonable thing to do are not the rational thing to do. For this reason, I think, moral theories need not themselves entail an account of what makes actions rational. Rather, there are in principle two ways to answer Q for any such theory. First, the theory can itself entail an answer to Q. Second, the theory might be supplemented with a so-called decision procedure that yields an answer to Q. To recap, the sense of “decision-guidance” I am after is such that a moral theory T fails to be decision-guiding if the following constellation is possible: (4) Although you understand and accept T and you want to do what is right according to T, and although you are a normally intelligent, sane human adult with a properly functioning mind, (5) you rationally ask Q, (6) T cannot be supplemented with a plausible decision procedure that answers Q. In short, a moral theory is not decision-guiding if it neither tells you how to decide in cases of factual ignorance nor can be supplemented with a plausible decision procedure that tells you how to decide in these cases. A final point: We have seen in the Introduction (Section 2) that purely subjectivist versions of otherwise plausible moral theories are implausible. Recall the lunatic who believes that Satan will torture all humans forever unless she shoots all customers in a shopping centre. The lesson was that moral theories ought not to derive what an agent has a moral duty to do from what the agent actually believes, but from the evidence available to her or from the facts. By contrast, in order to be itself decision-guiding

The Problem of Decision-Guidance

23

(rather than decision-guiding in virtue of being combined with a decision procedure), a moral theory would have to derive what an agent ought to choose from the agent’s actual beliefs. For if an agent is told how to decide merely on the basis of facts or evidence, then the problem of decisionguidance might well remain since the agent might be uncertain about what the facts are or what the evidence suggests. It follows that moral theories that are decision-guiding by themselves, without an additional decision procedure, must be purely subjectivist and, therefore, implausible. Hence, any plausible moral theory relies on an additional decision procedure in order to be decision-guiding. (This will be important later.)

4. Contrasting Decision-Guidance With Other Kinds of Action-Guidance Notice that all deontic moral theories are action-guiding in some sense, though not necessarily properly action-guiding. Even a moral theory that does not entail an answer to Q is action-guiding, though not by itself decision-guiding, simply in virtue of entailing deontic judgements. This is because expressions like “obligatory” and “being favoured by a reason” are essentially action-guiding (cf. Introduction, Section 3). In order to not conflate the problem of decision-guidance with other charges against objective consequentialism, it is important to distinguish between decision-guidance and other kinds of action-guidance. There are many objec­ tions to objective consequentialism that can be confused with the problem of decision-guidance but that concern other ways in which moral theories should arguably be action-guiding. Let us consider some such objections. (7) If you do not know what you ought to do according to a moral theory, then you cannot do what the moral theory requires you to do. (8) If you cannot do what a moral theory requires you to do, then the moral theory is not action-guiding. The challenge prompted by (7) and (8) concerns a moral theory’s com­ patibility with the principle that “ought” implies “can”. Chapters 4 and 7 deal with the compatibility of objective and subjective consequentialism with that principle. The argument based on the problem of decisionguidance does not presuppose that “ought” implies “can”. Here is a further sense of “action-guidance”: (9) If human agents cannot know which of their available acts are right, wrong, or obligatory according to a moral theory, then even the best humans will almost never succeed in doing what the moral theory requires them to do. (10) If even the best humans will almost never succeed in doing what a moral theory requires them to do, then the moral theory is not suf­ ficiently action-guiding.

24 Decision-Guidance

A critique along these lines has been put forward against objective con­ sequentialism by Dale E. Miller (2003). I take the underlying motive to be different from what motivates the problem of decision-guidance. An argument based on whether the best humans will in fact do what a moral theory requires would apparently quantify over long-term behaviour. The problem of decision-guidance, in contrast, focuses on single choice situa­ tions. This book will not be concerned with action-guidance in the sense suggested by Miller’s critique. Here is yet another sense of “action-guidance”: (11) If trying to achieve the aims a moral theory gives you has poor results in terms of these aims, then the theory is self-defeating. (12) If a moral theory is self-defeating, then it is not properly action-guiding. The self-defeatingness of objective consequentialism has been widely dis­ cussed by consequentialists and their critics (see, e.g. Parfit 1984, Part 1; Railton 1984). I shall ignore this problem. Here is another idea: (13) If you cannot know the facts that make your available acts right, wrong, or obligatory according to a moral theory, then the moral reasons that the theory claims to exist cannot have a deliberative function. (14) If the moral reasons claimed to exist by a moral theory cannot have a deliberative function, then the moral theory is not sufficiently action-guiding. According to objective consequentialism, the consequences an act would in fact have are its right-making or wrong-making features. On the assumption that right-making and wrong-making features are practical reasons (for or against the acts that have these features), objective con­ sequentialism says that the consequences of an act are the reasons for or against performing it. Since we cannot know the consequences of our acts, we cannot know our reasons either. This is problematic if we assume that reasons have a deliberative function such that we weigh our reasons in order to arrive at a decision. In Chapter 5, I will come back to the deliberative function of reasons. Finally, I shall argue in Chapter 6 that the lack of factual evidence is a defeater of obligation. This idea can also be spelled out in terms of “action-guidance”: (15) If you lack factual evidence about which of your available acts are right, wrong, or obligatory according to a moral theory, then the theory is not properly action-guiding.

The Problem of Decision-Guidance

25

According to (15), you need to have evidence in order to have a moral duty. In contrast, according to the sense of decision-guidance I presented in this chapter, moral theories need to be implementable. I argue for something like (15) in Chapter 6. There are two important differences between what I argue in Chapter 6 and what I claim in this chapter. The first difference is that the instructions issued by decision pro­ cedures are not moral duties. So, the requirements a moral theory needs to issue (by itself or in combination with a decision procedure) in order to be decision-guiding need not be moral duties whereas (15) is concerned with moral duties. The second difference is that according to (15), for a moral theory to be sufficiently action-guiding, it is necessary that there is evidence about what you ought to do according to the theory. In contrast, it would not be enough for a decision procedure to entail evidence-dependent instruc­ tions. For agents may lack belief or be uncertain about which evidence is available to them, as I have explained previously. A decision procedure, hence, has to be action-guiding in a stronger sense than is required from moral theories according to (15). It is not enough for a decision proce­ dure to instruct the agent “Do x if the evidence available to you indicates that . . .”. Rather, a decision procedure must just say “Do x” or “If you believe this or that, do x”. One might think that there is a further distinction, namely, that deci­ sion procedures range over decisions while duties range over (full-blown) actions, which can be understood as decisions/tryings plus bodily move­ ments. While I have so far written as if this were the case, I eventually disagree. As I will explain in Chapter 7, I think that at a fundamental level, moral duty also ranges over decisions rather than merely over (full­ blown) actions. However, in this and the next chapter I will for the sake of simplicity keep on referring to what is required by a decision proce­ dure by talking about what one ought to decide to do and referring to what one has a duty to do by talking about what one ought to do. The final point worth mentioning is that according to common-sense morality, lack of factual evidence is not a defeater of moral obligation but an excuse. The proposal that will be offered in Chapter 6 is, hence, revi­ sionary. By contrast, I take the demand that moral theories be decisionguiding (by themselves or when supplemented with a decision procedure) to be in line with common-sense morality.

5. Conclusion Sometimes, a follower of a moral theory might wonder how to decide because she cannot ascribe to the actions available to her what according to her theory are ought-making, right-making, or wrong-making features. In this chapter, I have argued that moral theories should either themselves entail instructions for this kind of case or be able to be supplemented

26

Decision-Guidance

with a plausible decision procedure. Moral theories that fall short on both counts lack sufficient action-guidingness in terms of decision-guid­ ingness. Furthermore, I have argued that moral theories can be decisionguiding by themselves only in virtue of implausibly claiming that moral duties depend on agents’ beliefs rather than facts or available evidence. It follows that only those moral theories are plausible that can be supple­ mented with a plausible decision procedure. Finally, I have distinguished different ways of understanding “action­ guidingness”. All moral theories are action-guiding in virtue of ascribing action-guiding (deontic) properties. But moral theories should also be action-guiding in several other respects: they should be decision-guiding in virtue of being supplementable with a plausible decision procedure and, as I will argue in Chapter 3, moral theories should be action-guiding in the sense of not violating “ought” implies “can” (i.e. the theories should not ask you to perform actions that you cannot perform). I also indicated that I will argue that moral theories should be action-guiding in two fur­ ther senses: the reasons that exist according to a moral theory must be able to serve a deliberative function and the duties that exist according to a moral theory must be such that you cannot lack factual evidence for the existence of these duties.

Notes 1. Feldman 2006, pp. 50f. Notice that Feldman focuses on a utilitarian version of consequentialism and, accordingly, on actual and expected utility rather than value. Note also that thoughts related to those presented in Feldman 2006 can be found in March and Simon 1958, Lindblom 1959, Smith 1988, and Frazier 1994. 2. Feldman (2006, pp. 74–77) mentions some of these philosophers. 3. Feldman 2006, p. 54. I substituted “expected value” for “expected utility”. 4. The decision rule Greaves suggests bears some resemblance to the deci­ sion procedure I defend in the next chapter. However, unlike my procedure, Greaves’s rule has problems that can be made clear in light of the criteria of adequacy for decision procedures that I suggest in the next chapter. Roughly, the problems concern, on the one hand, situations in which you cannot esti­ mate that one act’s expected value is higher than the expected value of its alter­ native and, on the other hand, situations in which the process of estimating the expected value of an act is too time-consuming, leading itself to costs in terms of expected value. 5. Notice that the probability mentioned here features in the content of your belief rather than in the belief’s strength or degree. I am only concerned with outright belief. Content-wise, outright beliefs can be certain or uncertain. However, on some views, a credence of x in a proposition just is, or at least goes hand in hand with, an outright belief that the probability for p is x. More on this later. 6. Without these assumptions, all plausible moral theories would suffer from the Travesty of Decision-Guidance Problem that I will present in the next chapter. While I think that the assumptions are plausible, they are, of course, contro­ versial and defending them goes beyond the scope of this book. For a very accessible overview of the relationship between outright belief and credence, see Jackson 2020.

2

DP and Objective Consequentialism’s Travesty of Decision-Guidance

In this chapter, I have two aims. First, a decision procedure will be proposed. The decision procedure is supposed to solve the problem of decision-guidance for all moral theories, including objective and subjective consequentialism. My second aim is to show that, although the proposed decision procedure gets objective consequentialism off the hook as far as the problem of decision-guidance is concerned, applying the decision procedure to objective consequentialism reveals a related problem of objective consequentialism, which I dub the Travesty of Decision-Guidance Problem. Subjective consequentialism, on the other hand, is in best order as far as decision-guidance is concerned. In Section 1, I propose criteria of adequacy for a decision procedure. In Section 2, I present a decision procedure that solves the problem of decision-guidance. I argue for this decision procedure in Section 3. The decision procedure will be applied to objective and subjective consequentialism in Sections 4–6. There, I also present objective con­ sequentialism’s Travesty of Decision-Guidance Problem. Section 7 concludes. Since I will in the following deal primarily with suggestions put forward by Fred Feldman (2012) and since I do not want diverging terminologi­ cal preferences to impede the understanding of my text, let me contrast my terminology with Feldman’s. While I talk about “moral theories” and “decision procedures”, Feldman talks about “two level theories” with “practical level components” and “theoretical level components”. My “moral theories” correspond to Feldman’s “theoretical level compo­ nents”, my “decision procedures” to his “practical level components”. What he calls a “two level theory”, I call “a moral theory plus a decision procedure”. Finally, when I talk about “what one ought to do” accord­ ing to a moral theory, as opposed to “what one ought to decide to do” according to a decision procedure, Feldman talks about “what one has a moral obligation1 to do” in contrast to “what one has a moral obliga­ tion2 to do”.

28

Decision-Guidance

1. Criteria of Adequacy for Decision Procedures Let me begin with a note of caution. Consequentialists often appeal to deci­ sion procedures in order to defend their theories against certain objections, such as the objections that consequentialism cannot accommodate special obligations and agent-centred restrictions. The line of argument normally put forward in these contexts is that, under realistic circumstances, human agents produce more good in the long run if they have certain motives or character traits or accept certain principles, even if these motives, charac­ ter traits, or principles sometimes make the agents perform actions that are wrong according to consequentialism (see, e.g. Crisp (1992), Hare (1981, Part 1), Parfit (1984, Part 1), Pettit and Brennan (1986), Railton (1984)). This line of argument is not relevant in our context. Rather, we are concerned with answering Q – the question introduced in Section 2 of the previous chapter. I will not examine how a decision procedure that answers Q is related to a decision procedure that makes the agent produce a maximum of goodness. I will here only examine the decision procedure that answers Q and argue that this decision procedure supports subjective rather than objective consequentialism.1 It seems plausible that we should have a different view of the other kind of decision procedure – the one having to do with the good consequences of motives, character traits, and principles – if we accept subjective rather than objective consequentialism. But this is a topic that lies beyond the scope of this book. With this clarification in mind, we can now return to our main subject and ask: What conditions must decision procedures meet to be plausible? Several criteria of adequacy are being discussed for decision procedures.2 Instead of discussing lists of criteria that have been suggested by others, I will suggest a list myself. First Criterion: Answering Q The first criterion can be justified by the very purpose a decision procedure has with regard to moral theories according to the analy­ sis presented in the foregoing chapter: a decision procedure ought to solve the problem of decision-guidance. In order to do so, a deci­ sion procedure must provide an answer to Q in all relevant situa­ tions, thereby making a moral theory implementable. So, a decision procedure must tell an agent in decision-guiding terms – that is, by referring to the acts available to the agent in such a way that she can identify them – how she ought to decide. Second Criterion: Leading to the Performance of Right Acts A decision procedure would satisfy the first criterion even if it com­ prised only one rule: “Decide to do whatever is the first thing that comes to your mind.” But as I have said earlier, the purpose of a

DP and the Travesty of Decision-Guidance

29

decision procedure for a moral theory is to help agents doing what is right according to the theory. So, the decisions sanctioned by deci­ sion procedures must make it likely, at least to some extent and under normal circumstances, that one ends up doing what is right accord­ ing to the moral theory one accepts. A corollary of this demand is that decision procedures can be better or worse in that they can be more or less effective with respect to leading to actions that are right according to the agent’s moral theory.3 Third Criterion: Universality Decision procedures must be universal in several respects. A decision procedure ought to be usable by followers of any moral theory. For how to implement a moral theory is not a substantial moral question, and therefore nothing different moral theories can provide different answers to. Moreover, a decision procedure should be usable by any (adult, sane, etc.) agent in any situation. This is suggested by the fact that the implementation of moral theories is a matter of rationality, and criteria of rationality apply universally.4 Fourth Criterion: Simplicity Decision procedures have to be simple for two reasons. The first rea­ son is that, as I have just claimed, the best decision procedure must be usable by any (sane, adult, etc.) agent. Since agents vary in their cognitive abilities, only a simple procedure will be universally rec­ ommendable. Secondly, following a decision procedure takes time. Suppose following a certain decision procedure takes so much time that the choice situation is over before you arrive at a decision. This decision procedure, it seems, is not sufficiently simple. For the purposes of this chapter, it is sufficient to concentrate on the four conditions I have suggested. I will now show that the decision proce­ dure suggested by Feldman does not satisfy my criteria of adequacy. The same is true of the decision procedures suggested by Holly Smith (2010, pp. 100–106; 2018, Chapters 10–12). However, since it would be boring to go through a second suggestion just in order to dismiss it, I will confine myself to discussing Feldman’s suggestion. Here is Feldman’s (2012, pp. 166–167) suggestion of a decision proce­ dure for a version of objective utilitarianism: Step One: Step Two:

Consider the acts that you take to be your alternatives – described in “helpful”, “action-guiding” terms. Consider, insofar as your epistemic state permits, what you take to be their values or perhaps just their relative values.

30

Decision-Guidance

Step Three: If you haven’t got useful information about the actual values of your alternatives, then consider how your views about the morality of risk apply to your present situation; and, in light of all this, Step Four: identify the acts in this particular case that seem most nearly consistent with the general policy of maximiz­ ing utility where possible while avoiding things that put people at excessive risk of serious harm; and then, Step Five: perform one of them.

In my view, Feldman’s proposal is way too complicated. Consider an agent who only has a few moments for choosing among her alternatives. One can easily imagine that the agent would not arrive at a decision by employing Feldman’s decision procedure before the choice situation is over. Let me illustrate this by commenting on several of the steps involved in Feldman’s decision procedure. As for Step One, Feldman explains in a footnote: Note that Step One does not require the agent to list all of her alter­ natives. There might be millions of them. In many cases it will be sufficient for the agent to consider whole groups of alternatives under suitable general descriptions. For example, suppose an agent has been asked to pick a number between one and one million. There is no need for her to consider picking one, picking two, picking three, etc. Since she will have no epistemic trouble in any case, she can describe her alternatives as a group by saying “I have to pick a number between one and one million.” This will be sufficiently action guiding. (Feldman 2012, p. 166, n. 21) Even with this clarification at hand, the decision procedure requires the agent to consider her alternatives. This is time-consuming and may well make the difference between being able to choose among one’s alterna­ tives on time and failing to do so. In another note, Feldman comments on Step Two: Step Two does not require that the agent consider each individual alternative separately. In the numbers case imagined in the previous footnote, she might simply consider that there is no number, n, such that her evidence gives her reason to suppose that picking n will yield more utility than picking any other number. (Feldman 2012, p. 166, n. 22) Again, this mitigates the problem but does not solve it. Considering rela­ tive values is time-consuming. So is considering that there is nothing spe­ cial about any of one’s alternatives.

DP and the Travesty of Decision-Guidance

31

Steps Three and Four are even more time-consuming. Step Three requires you to take into account your views about the morality of risk and apply them to your present situation, if you lack useful information about the values of your alternatives. One can easily imagine spending a whole day in an armchair, going through Step Three for a fairly complex choice situation, while one would have to arrive at a decision by noon. So, Step Three is also too time-consuming. The same is true of Step Four. Following Feldman’s proposal can easily lead to very bad decisions. It hence not only violates my fourth but also my second criterion of ade­ quacy. Moreover, Feldman’s proposal risks violating my third criterion of adequacy because it is not clear that every agent will be cognitively able to employ Feldman’s decision procedure. For not only is the procedure rather demanding in requiring agents to have views about the morality of risk, see Step Three. It also uses vague phrases that agents need to be able to interpret and apply to their situations, such as “most nearly consistent with the general policy of maximizing utility” or “excessive risk of seri­ ous harm” in Step Four. Finally, notice that the decision procedure cannot be applied to moral theories that entail a morality of risk. For it licenses the agent’s personal views about the morality of risk. This is a violation of my second criterion of adequacy. The lesson is that Feldman’s proposal lacks simplicity and universality and will often not lead to agents’ doing what they ought to do according to utilitarianism. My discussion of Feldman’s proposal focused on a ver­ sion of objective consequentialism because this is Feldman’s toy example. Feldman’s proposal has exactly the same difficulties when applied to other moral theories. I leave it to the reader to apply Feldman’s proposal and my critique to other theories.5

2. Proposal for a Decision Procedure Let us now turn to my own proposal. In Section 3 of the previous chap­ ter, I have distinguished between two constellations in which you fail to arrive at a decision: (i) Absence of belief: You lack belief about the right-making and wrongmaking features of your alternatives. (ii) Uncertainty/under risk: Although you have beliefs about right- or wrong-making feature of your alternatives, you do not assign these features to your alternatives with certainty. Notice that both problems can occur in one situation at the same time. This is the case when one does not lack belief about the ought-, right-, and wrong-making features with respect to all of one’s alternatives but only with respect to some of them while, as for those alternatives that one does assign ought-, right-, or wrong-making features to, the beliefs

32

Decision-Guidance

are uncertain. This leaves us with four problematic constellations, C1 to C4, as Table 2.1 illustrates. Let us start with the absence of belief problem and leave the uncer­ tainty problem for the next section; so we are now concerned only with C1 and C2. With respect to C2 we can distinguish between your hav­ ing beliefs about right-making features (which is entailed by your having beliefs about ought-making features), your having beliefs about wrongmaking features, and your having beliefs about both right-making and wrong-making features. This is illustrated in Table 2.2. The phrases “right-making feature” and “wrong-making feature” here refer to the features acts must have in order to be right or wrong all things considered or conclusively rather than prima facie or pro tanto. For illus­ tration, let us consider Kantianism, according to which what makes acts right is their being in accordance with the Categorical Imperative (CI). In C2.1, the agent believes with respect to at least one available act that this act is in accordance with CI and believes with respect to at least one available act that this act is not in accordance with CI, but the agent lacks such beliefs with respect to some other acts available to her. In C2.2, the agent believes with respect to at least one available act that this act is not in accordance with CI and does not believe with respect to any available act that this act is in accordance with CI. In C2.3, the agent believes with respect to at least one available act that this act is in accordance with CI and does not believe with respect to any available act that this act is not in accordance with CI. How ought the agent to decide? Here is my suggestion: Let {a1, . . . an} be the (relevant) set of the agent’s available acts described in decisionguiding terms, that is, as the alternatives occur to the agent rather than Table 2.1 Four constellations of the problem of decision-guidance.  

Complete absence of belief

Partial absence of belief

No absence of belief

No uncertainty Uncertainty

C1 –

C2 C3

–

C4

Table 2.2 Three constellations of C2.  

Belief about rightmaking features

No belief about rightmaking features

Belief about wrong­ making features No belief about wrong­ making features

C2.1

C2.2

C2.3

– (amounts to complete absence of belief)

DP and the Travesty of Decision-Guidance

33

from the viewpoint of an external observer. T, as before, designates the moral theory in question. DP Step 1 – Checking beliefs about right-making features (C2.1 and C2.3): If you believe with respect to some alternatives in {a1, . . . an} that these alternatives have right-making features according to T, pick from these alternatives in {a1, . . . an}! Step 2 – Checking beliefs about wrong-making features (C2.2): If you believe with respect to some alternatives in {a1, . . . an} that these alternatives have wrong-making features according to T, pick from the other alternatives in {a1, . . . an}! Step 3 – Complete absence of belief (C1): Pick any alternative in {a1, . . . an}! Let me make four explanatory remarks. First, by “picking”, I mean choosing at random. Second, “some alternatives” is to be read as: at least one alternative. Third, and related to this, Step 1 does not require you to consider all of your options. It is enough for you to believe with respect to at least one of your alternatives that this alternative has right-making features in order to be led to a decision by DP (the decision being to per­ form that alternative). Fourth, you only have to go through Step 2 in case Step 1 does not already lead you to a decision. Correspondingly, you only have to go through Step 3 in case Step 2 does not lead you to a decision. Fifth, Steps 2 and 3 are distinguished only for purposes of illustration. In principle, they could be combined into the instruction “pick one of your alternatives provided you do not think that this alternative is wrong”. For illustration, consider, again, our clueless Kantian. If the Kantian believes that one of her available acts is in accordance with CI (as in C2.1 and C2.3), then the Kantian ought to decide to perform that act, accord­ ing to DP (Step 1). If she believes this with respect to several alterna­ tives, then she can pick one of them. If the Kantian does not believe with respect to any available act that this act is in accordance with CI, then she has to apply Step 2 of DP: Does the Kantian believe that some of her alternatives are not in accordance with CI? If so (C2.2), she ought to pick from the other alternatives – from the alternatives, that is, of which she does not believe that they are not in accordance with CI. If the Kantian neither believes with respect to at least one of her alternatives that this alternative is in accordance with CI nor believes with respect to any of her alternatives that these alternative are not in accordance with CI (C1), DP (Step 3) leads her to pick among all alternatives.

34

Decision-Guidance

In this section, I have presented DP, a decision procedure that is sup­ posed to solve the problem of decision-guidance for all plausible moral theories. We can now turn to the discussion of DP and, in this context, of the constellations involving uncertainty, C3 and C4.

3. Discussion of DP The main reason why we should believe DP is that it probably better sat­ isfies the criteria of adequacy stated in Section 1 than any other decision procedure. DP satisfies the first criterion since it guides choices and thus answers Q. Since DP tells the follower of a moral theory T to choose what she thinks has right-making features and not to choose what she thinks has wrong-making features, DP tends to lead the agent to do what is right according to T, thus satisfying the second criterion. DP is very simple, uni­ versally applicable and immensely time-saving, thus satisfying criteria 3 and 4. All you have to know as a follower of T in order to be able to use DP is your beliefs about the right-making and wrong-making features of your alternatives. All you have to do is pick one of the acts you believe to have right-making features (Step 1) or of the acts you do not believe to have wrong-making features (Step 2) or of any of your alternatives (Step 3). What DP does presuppose, however, is that you know the beliefs you hold about the right-making and wrong-making features of your options. And maybe that your beliefs are consistent and fulfil certain other mini­ mal rationality requirements. But this does not seem to be problematic. Since moral theories only apply to rational agents, by which I mean agents who fulfil certain minimal rationality requirements, it is enough for deci­ sion procedures to be decision-guiding only for these agents. In any case, DP can even be applied, though not properly, by some agents who are irrational in the relevant sense. Unsurprisingly, the decisions such agents will arrive at via DP will themselves sometimes be irrational. A natural objection to raise against DP concerns uncertainty: DP only takes into account whether an agent believes with certainty (as far as the contents of beliefs are concerned) that acts have right- or wrong-making features, respectively, and ignores uncertain beliefs (i.e. outright beliefs of the form stated in Chapter 1, Section 2). Consider the case of a doctor who has to choose between two drugs and believes (in accordance with the evi­ dence available to her) that drug A has a 0.9 probability to kill the patient and a probability of 0.1 to heal him, while the other drug, B, has a 0.1 probability to kill the patient and a probability of 0.9 to heal him. Given that the patient will die if the doctor does not treat him at all, we will want to say that the doctor ought to give him drug B. But does it not follow from DP that, if we assume that killing the patient would be wrong while healing him would be right, the doctor is free to pick among A and B? It depends. If we do not make any further moral assumptions in addi­ tion to the claims that healing the patient would be right and killing him

DP and the Travesty of Decision-Guidance

35

would be wrong, then it follows indeed from DP that the doctor can pick between A and B. But we should make further moral assumptions. For it seems obvious that it would be wrong for the doctor to expose her patient to an unnecessary risk. We should assume, hence, that the doctor has a duty not to expose her patient to the risk. If the doctor accepts that she ought not to expose the patient to the risk, then it follows from DP that she ought to decide to give drug B. So, my claim is that DP rightly ignores degrees of risk and uncertainty and thus constellations C3 and C4. For how one ought to act in the light of uncertainty is a question that needs to be answered by moral theories, not by decision procedures. In addition, notice that a decision procedure that differentiates on the basis of uncertain beliefs will be more complex than DP and in turn do worse with respect to the criteria of adequacy. The plausibility of my suggestion depends on how we see risk. Is risk something our decision procedure must deal with? Or is risk, if it is rel­ evant, a right- or wrong-making feature? It seems to me that the latter is the case. Here are some arguments to this effect. The arguments are phrased in terms of risk, i.e. the probability that something bad will hap­ pen, because this seems to be the more important case from the view­ point of common-sense morality. However, corresponding claims apply to the other variant of uncertainty, namely, chance, i.e. the probability that something good will happen. First of all, it just seems intuitively plausible that risk, at least some­ times, is a wrong-making feature. The example of the doctor is a case in point. For another example, think of overtaking manoeuvres. If a risky overtaking manoeuvre happens to have a good outcome, we certainly do not conclude that it was right. The risk seems to make it wrong, regard­ less of its outcome. Secondly, there are of course different views on how risky one’s actions ought to be. In Feldman’s proposal, you simply ought to apply your own views on the morality of risk to the situation at hand. But subjectivism is not a plausible view about the moral relevance of risk. (Also, it would be strange to reject subjectivism with regard to other moral issues while endorsing it with regard to risk.) It seems more plausible that how risky your choices or actions ought to be is an issue proponents of competing moral theories can have reasonable disagreement about. This suggests that, if we look for a general decision procedure that we can add to all sorts of different moral theories, this decision procedure should be silent on risk. Thirdly, how risky your action ought to be has implications for other moral questions. For example, it seems to be a substantial moral issue how diligently you ought to check the evidence available to you. But it also has far-reaching implications, for example, when it comes to matters of distributive justice. How ought one to allocate medical goods, say, if one does not know their effects for sure? All these seem to be substantial moral questions that competing moral theories will answer in different

36

Decision-Guidance

ways. A decision procedure that fits all moral theories should not pre­ clude moral views on these issues. One might object that the major moral theories are rather silent about risk as a wrong-making feature. This is true. But it seems to me that this only shows that we must develop the existing moral theories (cf. Hansson 2018, Section 5). This can of course easily be done for many moral theo­ ries. Take a deontological theory according to which acts of a certain type, T, are wrong independently of their consequences. Deontological theories can naturally be supplemented with the thesis that an act is wrong if, and because, there is a certain probability that the act is a T-act. Notice again that it is an open question how large the probability is. Proponents of dif­ ferent theories might want to put the threshold at different levels. And they might want to distinguish between different act-types. It is also conceivable that some moral theorists want to hold with respect to certain wrong acttypes that the mere risk of their occurrence is not a wrong-making feature.6 In sum, DP seems to be the best universal decision procedures because it satisfies plausible criteria of adequacy for decision procedures and is silent, rightly, on the morality of risk. The discussion of DP has exposed the need to expand existing moral theories so that they determine the circumstances under which they take risk to be a wrong-making feature (or, for that matter, chance to be a right-making feature).

4. Applying DP to Subjective Consequentialism In this section, I apply DP to subjective consequentialism. Subjective con­ sequentialism, when combined with DP, seems to be a decision-guiding moral theory that takes into account uncertainty just the way it should. By contrast, as I will show in the next section, the combination of objective consequentialism with DP, though decision-guiding, exposes the inability of objective consequentialism to recognize the moral relevance of uncer­ tainty. However, let me start by pointing out which kind of uncertainty seems to be morally relevant when it comes to the rightness and wrong­ ness of actions. Three kinds of uncertainty come to mind. Firstly, there might be objective uncertainty. If determinism is false, then there are objective probabilities for events to happen. Objective uncertainty is independent from any epistemic limitations observers might have. For illustration, assume that you know all the present facts, including the laws of nature and the position of any particle in the universe. If there is an objec­ tive probability (larger than zero and smaller than one) for an event to hap­ pen, you will still be in a state of uncertainty as to whether it will happen. As far as human agents are concerned, objective uncertainty does not seem to be the kind of uncertainty a convincing morality of risk needs to take into account. For ordinary human agents know nothing, or next to nothing, about objective probabilities. Objective uncertainty, if there is any, occurs on the microphysical level and manifests itself only in chaotic

DP and the Travesty of Decision-Guidance

37

systems such as the weather, if at all. By contrast, the morally relevant risk we face in everyday choice situations is something we are all well acquainted with. There may be zero objective risk involved in drinking and driving at the same time in a particular situation. For if determin­ ism is true, it might be objectively certain that no accident will happen. In a morally relevant sense, though, drinking and driving is nonetheless dangerous. Secondly, there is purely subjective uncertainty, consisting in an agent’s actually being uncertain. This is not the kind of uncertainty we are after, either. For, as I already indicated in the Introduction (recall the lunatic shooter in Section 2), agents can have all sorts of bizarre beliefs. An agent might think, for example, that there is no risk involved in simultaneously drinking and driving in a particular situation. (And as we have seen in the previous paragraph, this belief may even be true as far as objective risk is concerned.) However, as I said previously, to drink and drive simulta­ neously is risky in a morally relevant sense. This shows that the purely subjective kind of uncertainty, as exemplified in the case of the agent who thinks that drinking and driving is not dangerous, is not morally relevant either as far as the morality of risk is concerned. The third kind of uncertainty is the relevant one: evidential uncer­ tainty. In an ordinary situation, the evidence suggests that it is dangerous to drink and drive simultaneously. Many accidents happen when people drink and drive, and there are convincing explanations why this is so. The morally relevant kind of uncertainty seems to be relative to the evi­ dence available to an agent, regardless of the agent’s actual beliefs and of the objective facts. I favour the evidentialist version of subjective consequentialism, as I have explained in the Introduction, Section 2. This version recognizes that evidential uncertainty is crucial when it comes to the rightness and wrongness of actions. This is just as it should be. As far as decisionguidance is concerned, the evidentialist version of subjective consequen­ tialism should be combined with DP. DP issues commands on the basis of purely subjective states (“Pick any alternative that you believe to have right-making features”, “Don’t pick an alternative that you believe to have wrong-making features”). This, we have seen in the previous sec­ tions, is also just as it should be.

5. Applying DP to Objective Consequentialism We can now apply DP to objective consequentialism. Let me start by put­ ting on the record that objective consequentialism in conjunction with DP is decision-guiding. The reason is that DP guides every agent’s deci­ sions, whatever the agent happens to believe concerning the right- and wrong-making features of her alternatives. Hence, DP solves objective consequentialism’s problem of decision-guidance.

38

Decision-Guidance

However, objective consequentialism now faces a new problem. The problem is that the decision-guidance provided is a farce. For a rational follower of objective consequentialism will almost never identify which of her alternatives have right-making features and which have wrongmaking features. As Lenman put it (see the quote in Section 1 of the previous chapter), we only have “a clue of bewildering insignificance bordering on uselessness” about the overall consequences of our actions. Accordingly, followers of objective consequentialism will often at most have uncertain beliefs (in the sense presented in Chapter 1, Section 2), or even lack any belief, regarding the right- and wrong-making features of their available acts. DP will, therefore, almost always guide the agent towards picking among all of her alternatives. For example, you might pick between killing someone and buying her an ice cream cone. Objec­ tive consequentialists will certainly not like this result. We can call this objective consequentialism’s Travesty of Decision-Guidance Problem. To repeat, the reason why rational human agents will almost never form certain beliefs about which alternative has right-making features and which has wrong-making features is that, given the incredibly numerous and far-reaching unidentifiable consequences each alternative has in ordinary choice situations and given that an agent has cognitive access only to a tiny fraction of these consequences, it is very irrational to form certain beliefs about which alternatives have right- or wrong-making features accord­ ing to objective consequentialism. This has nicely been demonstrated by Lenman; see Section 1 of the previous chapter. The only rational thing an objective consequentialist can do in ordinary choice situations is to with­ hold judgement or form uncertain beliefs with quite low probabilities. The objective consequentialist might try to respond by supplementing her theory with a value theory according to which risk (in evidentialist terms) of harm (or other bad things) is a bad-making feature of outcomes and thus, in turn, a (pro tanto) wrong-making feature of acts. (And by supplementing her value theory accordingly with claims about chance being a good- and right-making feature.) The idea is to evade the Trav­ esty of Decision-Guidance Problem in a way that can be illustrated with the following example. Suppose that there is miniscule evidence that one of your alternatives will lead to a larger sum of happiness than your other alternatives. For illustration, suppose your alternatives are open the door with a straight face and open the door with a smile. We know that there is a small tendency for people to be happier when you smile at them. If the objective consequentialist accepts that a miniscule chance that people be happier is itself a good-making feature of outcomes, then this changes the situation. For while there is, from the agent’s perspective, only a small chance that people be happier if she opens the door with a smile, the agent might believe with certainty that there is this small chance. Hence, the agent might come to believe that opening the door with a smile has a right-making feature that opening the door with a straight face lacks.

DP and the Travesty of Decision-Guidance

39

This solution does not work, though. Recall that right-making features relevant in our context are features that make acts conclusively, rather than merely pro tanto, right. A small chance that people be happier will not qualify as a right-making feature in this sense. Generally speaking, it is still only a tiny fraction of the overall consequences of one’s alterna­ tives that one has access to on this proposal. In addition, one might won­ der whether a theory of value ought to count uncertainty as a good- or bad-making feature of outcomes at all.7 Finally, notice that even with regard to agent-relative versions of objective consequentialism, it remains unreasonable to form beliefs about which of one’s alternatives have rightmaking features and which have wrong-making features, given their mul­ tiplicity and their far-reaching consequences. So, these suggestions are only drops in the ocean of epistemic uncertainty. Hence, the Travesty of Decision-Guidance Problem holds.8

6. Subjective Consequentialism and the Travesty of Decision-Guidance Problem In light of what I have just said, one might wonder: Does subjective con­ sequentialism have its own Travesty of Decision-Guidance Problem? The answer depends on whether it is possible for human agents to form cer­ tain beliefs about which of their alternatives have right-making or wrongmaking features according to subjective consequentialism. On subjective consequentialism, an agent’s belief that an act has a right-making feature is her belief that no alternative to this act has higher expected value. And an agent’s belief that an act has a wrong-making feature is her belief that the act has less expected value than at least one of its alternatives. For simplic­ ity, let us focus on rightness only. Can a human agent form a certain belief, in an ordinary choice situation, that an available act has at least as much expected value as any of its alternatives? Recall that an act’s expected value is the sum of the value–probability–products of the act’s possible outcomes. I see two reasons why one could think that an ordinary human agent can­ not form rational beliefs, and hence cannot normally form beliefs, about the expected values of her alternatives in an ordinary choice situation.9 The first worry is that the problematic claims about objective conse­ quentialism also apply to subjective consequentialism. As we have seen, there are countless, far-reaching consequences of ordinary acts. And human agents do not have cognitive access to these consequences. Now, as Feldman has shown (see Section 1 of the previous chapter), there are even more consequences to be taken into account according to subjec­ tive consequentialism. For according to subjective consequentialism, not only the actual but also all merely possible consequences count. If human agents know only a tiny fraction of an act’s actual consequences, then it seems plausible that they know a relatively even smaller subset of an act’s possible consequences, given that these entail very fancy possibilities.

40

Decision-Guidance

The response to this worry is that an act’s possible consequences are relevant, according to subjective consequentialism, only insofar as the agent’s evidence suggests some probability larger than zero that the con­ sequences obtain. Since one can speak about an agent’s “evidence” only insofar as it is possible to respond in a rational way by forming evidencebased beliefs, there cannot be evidence for an agent that is cognitively over-demanding. This means that the assignment of non-zero probabili­ ties to possible outcomes of an agent’s alternatives is itself a function of the evidence available to the agent and that, therefore, it is always epistemically possible for the agent to form rational beliefs about there being these outcomes and these likelihoods of their occurrences. A second problem a critic could stress is that forming beliefs in the sense of C1 and C2 requires, in the case of subjective consequentialism, that the agent forms beliefs about calculations, that is, about value-probability products and about the sums of these. In response, I will make three points. Firstly, it is important to recall that, as I have explained in Section 2 of the previous chapter, the issue of decision-guidance is only about fac­ tual uncertainty, not about normative uncertainty. When talking about decision-guidance, we have to imagine an agent who has all the details of a moral theory. So, if the best theory of value entails numbers, we have to focus on an agent who knows the numbers. Second, it is not clear whether the best version of subjective consequen­ tialism really assigns value and probability numbers to outcomes (and in turn expected value numbers to acts). Considering numbers may just be a useful device in moral theory. It was primarily for illustration that I have introduced the concept of expected value as a sum of products. If numbers are not really involved in assigning expected value, doing so might be much easier. The third and most important point is that we often have beliefs about something like expected values in everyday choice situations. True, there are only few situations in real life in which agents actually list the pos­ sible consequences and the likelihoods of these and then weigh them. But it seems plausible that we very often have implicit beliefs of the same sort. And we are often certain that these beliefs are correct. This suggests that we are able to form certain beliefs about expected values. The upshot is that, due to DP, neither objective nor subjective con­ sequentialism faces a problem of decision-guidance. Objective conse­ quentialism, though, is befallen by the Travesty of Decision-Guidance Problem. Subjective consequentialism has no such problem.

7. Conclusion Contrary to a widespread assumption, the problem of decision-guidance does not by itself justify the move from objective to subjective conse­ quentialism. For if there is a problem of decision-guidance for objective

DP and the Travesty of Decision-Guidance

41

consequentialism, then there is also a problem of decision-guidance for subjective consequentialism, as we have seen in the previous chapter. I hope, however, that I have solved the problem of decision-guidance for all plausible moral theories by introducing DP in this chapter. The real problem concerning decision-guidance for objective consequentialism is not that objective consequentialism is not decision-guiding. The real problem is that objective consequentialism, when combined with DP, sanctions your picking at random among your options. This is the Trav­ esty of Decision-Guidance Problem. And this problem results from the fact that objective consequentialism cannot accommodate uncertainty in a satisfactory way.

Notes 1. However, Chapter 9 contains a discussion of other kinds of decision procedures. 2. For candidate criteria, see Feldman 2012, pp.  153–160, and Smith 2010, pp. 72–73; 2018, Chapter 10.1. 3. Notice that, if we were looking for a decision procedure that promises to reconcile consequentialism with special obligations, agent-centred restrictions, and the like, this criterion would sanction decision procedures the posses­ sion of which is maximally conducive to value even if this does not lead to the performance of right actions. But we are concerned with the problem of decision-guidance. 4. Recently, a reason–response account of rationality has been defended by Ben­ jamin Kiesewetter (2017). While rationality is not the subject of this book, it is worth mentioning that, as Kiesewetter points out, a reason–response account of rationality supports subjectivism about deontic properties. 5. Feldman (2012, pp. 167–171) applies it to Rossianism and virtue ethics. 6. In Section 5, I argue that objective consequentialism cannot convincingly be expanded in this regard. But since this might be different for other standard moral theories (which are objectivist), it cannot be objected that DP is some­ how question-begging. 7. To many philosophers, objective consequentialism appears to be the default position because classic consequentialist theories such as classical utilitarian­ ism are objectivist. Classic theories of value, however, do not recognize uncer­ tainty as a good- or bad-making feature. In order to arrive at a convincing moral theory, the classic consequentialist should give up either objectivism or the view that uncertainty is not a good- or bad-making feature. Insofar as one takes the second to be a more central commitment than the first, subjectivism might become the default position because it is compatible with the traditional theory of value. 8. As mentioned in Section 3, it might well be possible to modify other tradi­ tional moral theories so as to make them accommodate uncertainty. Therefore, my argument against objective consequentialism does not (over-)generalize. 9. I also see one possible misunderstanding: A critic might think that by “certain belief” I mean “credence 1”. But this is not how I introduced “certain belief” in Section 2 of the previous chapter. Rather, I clarified that I am only concerned with outright beliefs and that the certainty/uncertainty refers to the contents of such beliefs.

Part 2

“Ought” Implies “Can”

3

In Defence of “Ought” Implies “Can”

In this part of the book, I discuss whether the ought verdicts implied by objective and subjective consequentialism are compatible with the prin­ ciple that “ought” implies “can” and with the considerations underlying this principle. In this chapter, I set the stage for that discussion by intro­ ducing “ought” implies “can” and defending this principle against three types of objections that have been raised against it. I will motivate “ought” implies “can” but not provide conclusive reasons for this principle because these reasons are the topic of Chap­ ter  5. The defence of “ought” implies “can” in this chapter is impor­ tant because if “ought” implies “can” were false, we would not have to bother whether objective and subjective consequentialism are com­ patible with it or with related principles such as “ought” implies “evi­ dence” (which will be introduced in Chapter 6). In defending “ought” implies “can” against objections, I make use of a general recipe. Given the success of the recipe’s application to the considered objections, I con­ clude that “ought” implies “can”, and I am optimistic that the recipe will enable proponents of “ought” implies “can” to defend their principle against future objections. In Section 1, I present the principle that “ought” implies “can” in some detail and I explain why I find it plausible. My recipe for defending “ought” implies “can” will be presented in Section 2. Sections 3–5 deal with three types of objections against “ought” implies “can”: straight­ forward objections (Section 3), the objection from self-imposed inability (Section 4), and sophisticated objections (Section 5). A conclusion fol­ lows in Section 6.

1. Ought Implies Can My aim in this section is twofold. First, I want to explain how I under­ stand the principle that “ought” implies “can”. Second, I state why I find the principle intuitively appealing.

46

“Ought” Implies “Can”

Here is my official wording of the principle, followed by some explana­ tory remarks: Ought Implies Can Necessarily, a person is morally obligated to φ only if the person can φ. I have four clarificatory remarks. Notice, first of all, that Ought Implies Can is an explanatory principle. Your inability to φ explains why you cannot have a moral obligation to φ. In its explanatory aspirations as well as in its being a necessary truth, Ought Implies Can is not different from other moral principles; cf. Section 1 of the Introduction. Secondly, “φ” ranges over actions as well as omissions. If you cannot refrain from killing me, then, according to Ought Implies Can, you can­ not have a moral obligation not to kill me. And if you cannot actively kill me, then you cannot be obligated to actively kill me. Thirdly, the “ought” in Ought Implies Can refers to all-things-considered or overall or conclusive moral obligations as opposed to prima facie or pro tanto moral obligations. Prima facie duties provide some moral grounds for doing something but can be overridden by competing prima facie duties. Prima facie duties are relative to some considerations whereas conclusive moral duties take into account all moral considerations and therefore cannot be overridden (cf. Ross 1930, pp. 18ff.). Finally, the “can” in Ought Implies Can does not only involve general ability but also opportunity. I may have the general ability to drive a car even if I lack a car. But to be able to drive a car in the sense of Ought Implies Can, I need to have access to a car (and a road, and so on). Why should we believe Ought Implies Can? Ought Implies Can is an intuitively appealing principle. To me, at any rate, the principle clearly seems to be true. Now, this is of course less of an argument than a report of my intuitions. However, I think it is worth mentioning that there are few if any non-trivial moral principles that, upon reflection, appear to be as plausible as Ought Implies Can. There are several candidate rationales for Ought Implies Can, which will be considered in Chapter 5. Here I confine myself to presenting what I take to be the most plausible rationale, because this rationale best explains, I think, why Ought Implies Can is intuitively so over­ whelmingly plausible and thus motivates the principle. The rationale in question also needs to be presented at this point because it figures in the recipe for defending Ought Implies Can, which I will present in the next section. The rationale I endorse is conceptual in nature. There seems to be a fundamental difference between deontic and evaluative vocabulary. Why do we use deontic terms like right, wrong, and obligatory at all? Why do we not confine ourselves to evaluative notions like good and bad?

In Defence of “Ought” Implies “Can”

47

Apparently, deontic notions essentially have an action-guiding function. We use them for classifying our actions – some are to be performed, others are to be omitted. Evaluative properties can properly be ascribed to all sorts of entities, whereas the properties of rightness and wrong­ ness primarily apply to actions or choices (cf. Smith 1986, p. 342). The occurrence of an earthquake, for example, might be bad but not wrong. Conceptual analysis thus seems to reveal an action-guiding function of thinking in deontic terms and, in turn, of moral theories that use these terms. A moral theory that contradicts Ought Implies Can fails to distin­ guish between, on the one side, what would be good and, on the other side, what we ought to do. It would be very good, for example, if I trav­ elled back in time and prevented World War II. But this does not mean that I have a moral obligation to do this. My main reason for believing Ought Implies Can, then, can be sum­ marized as follows. Ought Implies Can follows from the action-guiding function of deontic morality, which, in turn, is based on conceptual truths. This explains why Ought Implies Can is intuitively overwhelmingly plau­ sible. (I will say more on the action-guidance rationale of Ought Implies Can in Chapter 5.) Apart from its own appeal, Ought Implies Can is a plausible prin­ ciple because it provides convincing explanations of many moral state­ ments that prima facie seem to be correct.1 Notice, however, that the explanatory power of Ought Implies Can is not a rationale. For argu­ ments to the best explanation only tell us that we ought to believe some proposition but they do not explain why the proposition is true. Here are examples of statements that seem best explainable by Ought Implies Can: a) I am not morally obligated to travel back in time in order to prevent World War II. b) If I could travel back in time and prevent World War II, then I would be morally obligated to do this. c) Non-swimmers never have moral duties to rescue drowning persons by swimming to them. d) Swimmers sometimes have moral duties to rescue drowning persons by swimming to them. e) I am not morally obligated to write an article that explains how to cure AIDS. f) If I could write an article that explains how to cure AIDS, then I would be morally obligated to do this. Obviously, this list could easily be extended. So, there are countless apparently correct moral statements that seem to be best explainable by Ought Implies Can. Therefore, we should accept this principle.

48

“Ought” Implies “Can”

2. The Recipe for Defending Ought Implies Can My recipe for defending Ought Implies Can against objections is inspired by the strategy R.M. Hare has used for defending utilitarianism against the charge that utilitarianism is false because it sanctions slavery if slav­ ery happens to be utility-maximizing.2 Just as a case of utility-maximizing slavery is supposed to be a counterexample to utilitarianism, objections to Ought Implies Can are based on alleged counterexamples. Counterexamples in the case of Ought Implies Can would have to dis­ play the following two structural features: (i) an agent a has a moral obligation to φ and (ii) a cannot φ. In order to falsify Ought Implies Can by means of a counterexample, the critic must show both that the agent in question has a moral obligation to φ and that she cannot φ. Corre­ spondingly, the defender of Ought Implies Can has two possible targets: She can deny that the agent ought to φ or insist that the agent can φ. The recipe I recommend to proponents of Ought Implies Can consists of three steps: Step 1: Ask for Details The defender of Ought Implies Can should start by asking for details: (i) Why does the agent have a moral obligation to φ? (ii) Why is the agent not able to φ? Sometimes, either the answer to (i) or the answer to (ii) will be obvious. However, I suspect that at least one of the answers will reveal peculiarities of the case in question. Step 2: Check the Relevance of the Peculiarities If the case involves peculiarities that explain why the agent ought to φ, check this: given the peculiarities, is it plausible that the agent cannot φ? If the answer is “no”, the alleged counterexample fails. Accordingly, if there are peculiarities that explain why the agent cannot φ, ask yourself: given these peculiarities, is it more plausible that the agent has a moral duty to φ than it is that her act lacks deontic status (i.e. is not right, wrong, obliga­ tory, or favoured by reasons)? If not, the case in question is not really a counterexample to Ought Implies Can. Step 3: Consider Alternatives to the Ought-Statement At first glance, one might have the intuition that the agent ought to φ. But if the agent cannot φ and Ought Implies Can is true, then it is not the case that the agent ought to φ. The agent’s not having an obligation to φ is compossible with her having other moral duties, with – in short – her having a duty to ψ. And it is also compossible with the evaluative fact that the agent’s φ-ing would be best.3 If the defender of Ought Implies Can

In Defence of “Ought” Implies “Can”

49

can show that the judgement that an agent’s having a duty to ψ in con­ junction, as it might be, with the judgement that the agent’s φ-ing would be best is at least as plausible as the judgement that the agent ought to φ, then this consolidates that the alleged counterexample fails. In the following sections, I will show that the recipe allows for success­ ful defences of Ought Implies Can. In Sections 3 and 4, I will attempt to improve existing defences of Ought Implies Can against somewhat older objections by using the recipe. In Section 5, I will apply the recipe to a more recent objection, raised by Peter A. Graham. While I hope that the recipe will enrich the discussions Sections 3 and 4 refer to, it will have its genuine baptism of fire in Section 5.

3. The Straightforward Objection to Ought Implies Can Alleged straightforward counterexamples include sadists and kleptoma­ niacs who cannot refrain from torturing innocent people and stealing their goods (Blum 2000, p. 287; Kekes 1984, pp. 459–460). These peo­ ple, so the critics of Ought Implies Can argue, act wrongly although they are not able to refrain from their evil deeds. This would not be possible if Ought Implies Can were true. Hence, Ought Implies Can is false. Against the alleged straightforward counterexamples, I submit, the defender of Ought Implies Can can convincingly make use of the afore­ mentioned recipe: Step 1: Questioning the Inability It seems to be true that, in ordinary cases, people ought not to torture other people or to steal their goods. However, ordinary people can refrain from torturing and stealing. Why is it that the kleptomaniacs and sadists in the thought experiment cannot refrain from these acts? In real life, even people with mental diseases such as kleptomania and sadism can refrain from tor­ turing and stealing. The difference between them and normal people is that it is considerably harder for kleptomaniacs and sadists to refrain from steal­ ing and torturing. They are therefore partially excused for their wrong acts. Ought Implies Can does not rule out the possibility of ordinary kleptoma­ niacs and sadists being morally obligated to refrain from stealing and tor­ turing. It seems that, unlike their real-life counterparts, the kleptomaniacs and sadists in the thought experiment must have a thorough psychological compulsion that makes them unable to refrain from theft and torture. Step 2: The Relevance of Psychological Compulsion Given that the kleptomaniacs and sadists in the thought experiment, unlike real-life kleptomaniacs and sadists, cannot refrain from theft and torture, it does not seem clear anymore that they have a moral obligation to do so.

50

“Ought” Implies “Can”

This does not mean, of course, that it might be permissible for the extraordinary sadists and kleptomaniacs to torture other people and steal their goods. Rather, their acts seem to lack deontic status.4 Or at least, the judgement that theft and torture committed by people who are psycho­ logically forced to perform these acts lack deontic status is intuitively at least as plausible as the judgement that these acts are wrong. Hence, the objection fails. Notice that, if the blameworthiness of an agent for an act presupposes that the agent’s act is wrong, then the kleptomaniacs and sadists from the thought experiment are not blameworthy. But this is not counterintui­ tive. Given that the imagined sadists and kleptomaniacs lack control over what they are doing, they do not seem to be blameworthy. Step 3: The Badness of Psychological Compulsion and Related Duties The defender of Ought Implies Can can strengthen her case by point­ ing out that her view is compatible with the following judgements: “the kleptomaniacs’ and sadists’ compulsive acts are bad”, “the kleptomani­ acs and sadists ought to undergo psychotherapy”, “they ought to regret what they are doing”, and so on. The lesson is that mental diseases of the unusual kind under consideration only rule out the diseased persons having moral obligations to refrain from the compulsive acts. But the compulsive acts are still apt objects of moral assessment in other respects. In sum, the straightforward counterexamples to Ought Implies Can fail. It is not more plausible that compulsive acts of the kind described are wrong than that they lack deontic status. The judgement that the com­ pulsive acts lack deontic status is especially convincing when combined with judgements such as that the compulsive acts are bad, that the agents ought to undergo psychotherapy, and so on.

4. The Objection From Self-Imposed Inability Let us come to the phenomenon of self-imposed inability. This phenom­ enon has been used by Walter Sinnott-Armstrong (1984, p. 252; 1985) to argue against Ought Implies Can. The objection from self-imposed inabil­ ity can best be illustrated by means of an example. Suppose I promise to you on Monday that I will lend you some money on Wednesday. On Tues­ day, though, I use my entire assets for buying some worthless paintings. The critic of Ought Implies Can will try to point out that it seems that I have a moral obligation to lend you the money on Wednesday. And, she will add, it is not only between Monday and Tuesday that I seem to have this obligation, but also after Tuesday when I am no longer able to lend you the money. Hence, the phenomenon of self-imposed inability seems to falsify Ought Implies Can.

In Defence of “Ought” Implies “Can”

51

Against this line of argument, proponents of Ought Implies Can have pointed out that we need to distinguish between, on the one hand, the time of my having an obligation to lend you some money (the time indexed to the “ought”) and, on the other hand, the time for which I promised to lend you some money (the time indexed to the obligatory act).5 Put differ­ ently, we have to ascribe a double time-index to “ought”-contexts so that we get phrases of the form: “an agent oughtt1 to φt2”. Ought Implies Can is compatible with the view that I ought, between Monday and Tuesday, to lend you money on Wednesday. This is important because it shows that it would be a mistake to assume, as one could be tempted to think, that Ought Implies Can rules out any obligation whatsoever for me to fulfil my promise. The truth is that Ought Implies Can only rules out that I ought, later than Tuesday, to lend you the money on Wednesday. But as I have already mentioned earlier, the critic of Ought Implies Can will claim that it is also after Tuesday that I seem to have an obligation to lend you the money on Wednesday. The critic may well concede that there is an Ought-Implies-Can-friendly interpretation of the case in question. But the interpretation that is hostile to the principle, the critic will insist, is intuitively more appealing. The two views one may hold with regard to the case in question are depicted in Figure 3.1. The question we need to tackle now is this: Which of the two views is more appealing? Is it the Ought-Implies-Can-friendly view, according to which it is only until Tuesday that I am obliged to keep my promise? Or is it the Ought-Implies-Can-hostile view, according to which I ought to keep my promise even on Wednesday? It is at this point that I refer to the recipe for defending Ought Implies Can: Step 1: Questioning the Duty-on-Wednesday It is uncontroversial that, after Tuesday, I cannot keep my promise. Why is it, according to the Ought-Implies-Can-hostile view, that I have a moral

Ought-Implies-Can-friendly view

Ought-Implies-Can-hosAle view

Ame at which I make the promise

Ame at which I make myself unable to fulfil the promise

Ame at which the promised act would be performed t

Monday

Tuesday

Wednesday

Figure 3.1 When do I have a moral obligation to keep my promise on Wednesday?

52

“Ought” Implies “Can”

obligation, even after Tuesday, to keep my promise on Wednesday? Two factors seem to constitute my having a moral obligation even after Tues­ day, according to the Ought-Implies-Can-hostile view. Firstly, I have given a promise. This is, of course, a necessary condition for my having an obli­ gation to keep a promise. Second, and this is the decisive point, it would seem unfair, the proponent of the Ought-Implies-Can-hostile view will probably say, if I could free myself from the obligation simply by making myself unable to discharge it. Step 2: The Relevance of Self-Imposed Inability The important point the defender of Ought Implies Can needs to empha­ size is that it is not true that I can free myself from the obligation to keep my promise according to the Ought-Implies-Can-friendly view. I have this obligation. True, if the Ought-Implies-Can-friendly view is correct, I have this obligation only from Monday to Tuesday. But the important point is that my making myself unable to lend you the money on Wednes­ day does not free me from my obligation to borrow you the money on Wednesday. The critic might raise two objections. The first objection says that, despite what I have said, it is still true that I can free myself from having an obligation on Wednesday, if the Ought-Implies-Can-friendly view is correct. Secondly, if I do not have an obligation on Wednesday to lend you the money, the critic might worry, then I am not, as of Wednesday, blameworthy for not lending you the money. The first objection is correct in that, according to the Ought-Implies­ Can-friendly view, I can free myself from having an obligation on Wednes­ day to lend you the money on Wednesday. So what? The objection seems to presuppose that it is an issue of fairness whether I have a moral obliga­ tion or not. It seems to me, however, that the critic commits a category error at this point. It does not make sense, I think, to speak about the fair­ ness of moral obligations. A mistake possibly underlying that talk is the confusion of obligations and blameworthiness. It certainly seems to be a matter of fairness whether I ought to be blamed for an act or an omis­ sion. For blame is something we distribute. But moral duties – in contrast to, say, legal duties – are not something we distribute. Another mistake possibly underlying the idea that having a moral obligation is a matter of fairness is the confusion of moral demands with the demands issued by persons. It can certainly be unfair for a person to demand something. But since it is absurd to hold that moral demands are issued by persons, moral demands cannot be unfair in this way.6 Considerations of fairness, I conclude, do not apply to the question of whether I have, on Wednesday, an obligation to lend you the money on Wednesday. This leaves us with the second objection. Assume that blameworthi­ ness presupposes wrongness. Does the Ought-Implies-Can-friendly view

In Defence of “Ought” Implies “Can”

53

entail that I am not blameworthy, as of Wednesday, for not borrowing you the money on Wednesday? The answer is “no”. It is certainly not the case that you can be blameworthy for wrong acts only during your per­ formance of these acts. Obviously, persons are very often blameworthy for past acts. The war criminals of World War II, for instance, certainly deserved to be blamed (and punished) after 1945. In sum, the Ought-Implies-Can-friendly view entails that I can free myself from having an obligation after Tuesday. But on reflection, this implication is unproblematic. For one thing, it would be a category error to claim that it would be unfair if I did not have an obligation after Tues­ day. For another, my being able to free myself from having an obligation after Tuesday is compatible with my being blameworthy, as of Wednes­ day, for not lending you the money on Wednesday. Hence, the objection from self-imposed inability fails. Step 3: Secondary Duties and Badness If Ought Implies Can is true, then you cannot call me on Wednesday and truthfully say: “You have a moral duty to lend me the money today.” We have already seen that this implication of Ought Implies Can is compat­ ible with the judgement that, as of Wednesday, you can rightly blame me for not lending you the money. In addition, it is worth pointing out that Ought Implies Can is compatible with my having many secondary duties, that is, duties I have in virtue of the duty to keep my promises. It would be a mistake for the critic of Ought Implies Can to assume that I can free myself from any promise-keeping-related duties whatsoever just because, according to the Ought-Implies-Can-friendly view, I can free myself from having a duty, after Tuesday, to lend you the money. Ought Implies Can only implies that I do not have a moral obligation after Tuesday to lend you the money on Wednesday. This is compatible with my having a number of secondary duties such as the duty to apolo­ gize for not keeping my promise, the duty to earn money as fast as I can in order to then lend it to you, and many other duties (see Figure 3.2).

obligaAon not to make promises that I plan to break

Ame at which I make the promise

obligaAon not to make myself unable to fulfil the promise

obligaAons to apologize, to compensate, etc.

Ame at which I make myself unable to fulfil the promise

Ame at which the promised act would be performed t

Monday

Tuesday

Wednesday

Figure 3.2 Secondary duties that are compatible with Ought Implies Can.

54

“Ought” Implies “Can”

The bottom line is that reflection reveals all implications of the Ought­ Implies-Can-friendly view to be intuitively unproblematic. The objection from self-imposed inability fails.

5. The Sophisticated Objection to Ought Implies Can Finally, let us consider a sophisticated objection to Ought Implies Can, which has been put forward by Peter A. Graham: [I]f there are scenarios in which it is plausible that the moral per­ missibility of A’s φ-ing depends on the moral impermissibility of B’s ψ-ing but also for which it is not plausible that rendering B incapable of refraining from ψ-ing would render A’s φ-ing morally impermis­ sible, such scenarios would be counterexamples to [Ought Implies Can]. (Graham 2011, p. 344) Put in different words, Graham focuses on cases where the rightness of what A does is best explained by the wrongness of what B does and where B can­ not refrain from doing what she does. B, hence, ought (to refrain from what she does) although she cannot. This is incompatible with Ought Implies Can. Graham’s best example is a modified version of this case: TRANSPLANT: A surgeon has ten patients, each of whom will die of organ failure if he or she does not receive an organ transplant. The surgeon wants to save her patients and is convinced by philosophical arguments to the effect that it would be morally permissible to kill two people in order to save her patients. She notices that in another room of the hospital there are two innocent and unconscious tonsil­ lectomy patients who are perfect organ matches for her patients. The only means by which the hospital janitor, who is aware of the situa­ tion, can stop the surgeon from chopping up the two and redistribut­ ing their organs among the ten is by shooting her with his pistol. He does so and thereby kills her. (Graham 2011, p. 345) Graham claims that: (1) It is morally permissible for the janitor to kill the surgeon, and (2) if the janitor had not killed the surgeon, the surgeon would have morally impermissibly killed two people.

In Defence of “Ought” Implies “Can”

55

The important point, according to Graham, is that the best explanation of (1)’s truth is that (2) is true (Graham 2011, p. 346). Graham goes on to consider the following modification: TRANSPLANT (COMPULSION): Everything is as it is in TRANS­ PLANT except that the surgeon cannot refrain from killing the two because the ten are her grandchildren, and she is as compelled to save them as is the most severe kleptomaniac to steal. (Graham 2011, p. 346) It does not matter why the surgeon cannot refrain from killing the two. Graham invites the reader to drop the story about psychological compul­ sion and to add to TRANSPLANT whatever features they think are nec­ essary to transform it into a situation in which the surgeon cannot refrain from chopping up the tonsillectomy patients. It is not only in TRANSPLANT but also in TRANSPLANT (COM­ PULSION) that (1) is true and that the best explanation of (1)’s truth is that (2) is true. Now, since in TRANSPLANT (COMPULSION) the surgeon ought to refrain from killing the two people although she cannot refrain from killing them, Ought Implies Can is false. Graham considers and dismisses many alternative explanations for the truth of (1). I only want to mention one explanation that naturally comes to mind: the explanation that it would be bad, though not impermissible, if the surgeon killed the two. Graham’s response is that this suggestion fails because it would be good if ten rather than only two people died.7 Let us turn to my critique. I do not find the sophisticated objection to Ought Implies Can convincing. Let us accept that (1) and (2) are true in TRANSPLANT. It also seems plausible that (1) is true in TRANS­ PLANT (COMPULSION) and that what best explains the truth of (1) in TRANSPLANT (COMPULSION) is the same as what best explains the truth of (1) in TRANSPLANT. But I do not think that the truth of (2) is the best explanation for the truth of (1) either in TRANSPLANT or in TRANSPLANT (COMPULSION). And I think that (2) is not true for TRANSPLANT (COMPULSION). Hence, I submit, TRANSPLANT (COMPULSION) does not refute Ought Implies Can. In order to substan­ tiate my claims, let me use the recipe for defending Ought Implies Can. Step 1, Part 1: Questioning the Surgeon’s Duty It is uncontroversial that the surgeon cannot refrain from killing in TRANSPLANT (COMPULSION). The question we need to focus on is, then, why does the surgeon allegedly have a moral obligation in TRANS­ PLANT (COMPULSION)? Since I agree that the surgeon has a moral obligation to refrain from killing in TRANSPLANT, I shall start by

56

“Ought” Implies “Can”

presenting what I take to be the best explanation of the surgeon’s moral obligation in TRANSPLANT. One could think that the best explanation of the surgeon’s obligation focuses on the special relationship between the surgeon and her patients. This relationship might be thought to provide the surgeon with reasons to refrain from killing the tonsillectomy patients. Alternatively, one might think that there is a moral rule according to which it is not permissi­ ble for surgeons to harm their patients. Another candidate explanation focuses on the autonomy rights of patients. On the face of it, these and other explanations do not seem implausible. However, my conjecture is that none of these candidates is plausible as an ultimate normative expla­ nation. I am using the term “ultimate” here in the sense that a normative explanation e1 is ultimate if and only if there is no further normative explanation, e2, which is an explanation of e1. My claim is, hence, that even if one of the aforementioned candidates explains the surgeon’s obli­ gation, there is a deeper explanation. Why do I think that none of the mentioned candidate explanations qualifies as an ultimate explanation? The reason is that all candidate explanations are incompatible with there being a society that, morally permissibly, has a very different social practice as far as organ transplants are concerned. In such a society, it would be permissible to transplant the organs of two patients, even against their wills, in order to save ten people. It is possible, I submit, to have a society in which a surgeon mor­ ally permissibly kills two patients against their wills in order to save ten. This is compossible with the fact that it is morally wrong for a surgeon in our society to enforce the transplants. Since moral truths are necessary truths, I conclude that none of the mentioned candidate explanations for why it would be morally wrong for the surgeon to kill the two people is an ultimate explanation. I have to make good on the claim that there could be a society in which it would be right for a surgeon to kill two people against their wills and transplant their organs. For lack of a better name, let us call it the Donor Society. In the Donor Society, it goes without saying that you donate your organs, even when it costs you your life, whenever other people are in greater need of them. The difference between the Donor Society and our society is not that the Donors value their organs, or their lives, less than we do. Rather, the Donors are considerably less selfish than we are. For this reason the Donors are able to uphold a social practice that requires each of them to donate their organs whenever other people gain a greater benefit from having them. By a “social practice”, I mean, roughly, a set of social norms, that is, behavioural regularities among a society that shape the expectations of the society’s members.8 The overwhelming majority of Donors is, almost all of the time, very happy to have this practice. When a Donor has to give her life for other people, she certainly regrets, to some extent, that she has to do it. But the

In Defence of “Ought” Implies “Can”

57

situation is comparable, in principle, to that of a decent childless person in our society who willingly, though with mixed feelings, pays her taxes insofar as they serve financing the educational system. The childless per­ son has no self-directed reasons for accepting these tax regulations, for we assume that neither she herself nor her near and dear ones benefit from the educational system. Nonetheless, the childless person has inter­ nalized the tax rules and would justify her acceptance of the rules by referring to other-directed (altruistic or impersonal) reasons. She might be benevolent towards profiteers of the educational system, or having a well-financed educational system might be in line with the person’s con­ ception of the common good. It seems to me that the Donor Society has a perfectly legitimate social practice concerning organ transplants. It is important to notice that the Donors have, on average, expectations and attitudes towards organ transplants that, due to their social practice, are very different from our average expectations and attitudes. If two people in the Donor Society refuse to donate their organs in order to save ten people, the Donors regard these people in a way comparable to how we regard a selfish person who refuses to pay her taxes. If a surgeon kills the two unwill­ ing people, by deception or by force, in order to reallocate their organs, it seems to me that the surgeon acts rightly. First of all, it seems to be legitimate, if not even desirable, for a society to have the Donor social practice. (This does not mean, obviously, that such a practice would be feasible for our society.) Secondly, if such a practice is in place, it seems unfair for the two patients for selfish reasons to refrain from contribut­ ing their share. I conclude that the explanations envisaged earlier cannot be the ulti­ mate explanations of the surgeon’s duty in TRANSPLANT. For there are possible societies in which surgeons do not have this type of duty. What, then, is the ultimate explanation for the surgeon’s duty in TRANS­ PLANT? The decisive part of the answer suggests itself: The surgeon has her duty in virtue of living in our society. In our society, we have a social practice according to which people can decide on their own what happens to their organs. It should be hardly controversial that, given the extent to which we are selfish, a social practice of the sort upheld in the Donor Society is not feasible for us. (Though, it seems to me, it would be better to have such a practice and some steps towards it may be feasible.) Given the – probably by and large legitimate – practice of our society, the surgeon has a moral obligation to refrain from killing the two patients. One might ask how exactly our social practice concerning organ trans­ plants, assuming that it is by and large the best feasible practice, makes it wrong for surgeons to enforce organ transplants, in particular given act-consequentialism. This is a hard question. However, there are many explanatory accounts.9 I will not answer this question here. The crucial

58

“Ought” Implies “Can”

point is that eventually, any plausible moral theory will have to provide an answer to the hard question. Step 1, Part 2: Questioning the Janitor’s Duty Why is it morally permissible for the janitor to kill the surgeon in TRANSPLANT? The best explanation, I submit, is not that otherwise the surgeon would have impermissibly killed two people. I think that it is morally permissible for the janitor to kill the surgeon because we have a social practice of respecting the autonomy rights of patients and it is morally permissible for the janitor to be a reliable partner in this social practice. Participation in our social practice does not only require the janitor to refrain from infringing the autonomy rights of patients himself. It also requires him to prevent others from such infringements. If you are not yet convinced, notice that the same holds for other social practices. These practices arguably include promise-keeping, tax-paying, prohibitions against lying, property rights, and punishment. As a member of our society, you are not only required not to lie, to keep your promises, to pay your taxes, to respect the property rights of other people, and not to engage in activities that lead to the punishment of innocent persons, but you also have to intervene if some other person transgresses these social norms. So, the wrongness of the surgeon’s killing the tonsillectomy patients is not the ultimate, and by that not the best, explanation of the rightness of the janitor’s killing of the surgeon. Rather, the janitor is required, qua member of our society, to prevent violations of the autonomy rights of patients. Step 2: The Relevance of Our Social Practice We have seen that what best explains the surgeon’s duty to refrain from chopping up the tonsillectomy patients in TRANSPLANT is that she is a member of our society, which has the (by and large) best practice of autonomy rights regulating organ transplants we are (probably) able to uphold nowadays. Would it be wrong for the surgeon to kill the tonsil­ lectomy patients in TRANSPLANT (COMPULSION)? At first glance, it might seem so. But here is a plausible debunking story that explains this seeming. Social norms are, I have said earlier, behavioural regulari­ ties that shape our expectations. According to our social norms, the sur­ geon’s killing the two patients might be forbidden regardless of whether or not the surgeon can refrain from doing so. It would be a mistake to conflate wrongness-according-to-social-norms with moral wrong­ ness. True, often wrongness-according-to-social-norms guarantees moral wrongness. This is the case when it comes to the surgeon’s killing the two patients in TRANSPLANT. But this is not always the case. If an agent

In Defence of “Ought” Implies “Can”

59

cannot refrain from violating a social norm, it seems to me, her act is not morally wrong. The surgeon’s killing the tonsillectomy patients in TRANSPLANT (COMPULSION), it seems to me upon reflection, would not be wrong but only wrong-according-to-social-norms. Morally speak­ ing, the killing would lack deontic status. What about the janitor? We have seen that the best explanation for why it is permissible for the janitor to kill the surgeon in TRANSPLANT is that our social practice of autonomy rights does not only require agents to refrain from violations of autonomy rights but also requires agents to prevent violations of these rights by other people. The same explanation applies to the janitor’s permission to kill the surgeon in TRANSPLANT (COMPULSION). The requirement to prevent transgressions of social norms by other people does not presuppose that those people be able to refrain from the transgressions. To see this, consider other social norms. Think of persons who cannot keep their promises, refrain from lying, pay their taxes, respect other people’s property, or refrain from punish­ ing innocent persons. It clearly seems to be permissible-according-to­ our-social-norms (if not obligatory-according-to-our-social-norms) and hence prima facie morally permissible (if not prima facie morally obliga­ tory) to make these persons keep their promises, refrain from lying, pay their taxes, respect other people’s property, and refrain from punishing innocent persons. So, in TRANSPLANT (COMPULSION), social norms rather than the wrongness of the surgeon’s killing the tonsillectomy patients explain why it is permissible for the janitor to kill the surgeon. In sum, it seems upon reflection that the surgeon’s killing the tonsillec­ tomy patients in TRANSPLANT (COMPULSION) would not be wrong but rather lack deontic status. And the rightness of the janitor’s killing the tonsillectomy patients is best explained by social norms in both TRANS­ PLANT and TRANSPLANT (COMPULSION). Therefore, the sophisti­ cated objection to Ought Implies Can fails. Step 3: Social Duties, Prima Facie Duties, and Badness My claim that the surgeon’s killing the tonsillectomy patients in TRANS­ PLANT (COMPULSION) lacks deontic status might appear even more plausible if we add some further claims. First of all, as I have already said earlier, the surgeon’s killing the tonsillectomy patients is wrong-according­ to-our-social-norms. Put differently, the surgeon has a social duty to refrain from killing the two, although he does not have a moral duty. Second, one can qualify the claim that the surgeon’s killing of the ton­ sillectomy patients lacks deontic status by distinguishing between conclu­ sive moral wrongness and moral prima facie wrongness. Ought Implies Can only entails that the surgeon’s killing the tonsillectomy patients would be conclusively wrong. But Ought Implies Can does not rule out its being prima facie wrong. Depending on how the best moral theory

60

“Ought” Implies “Can”

describes the relation between social norms and moral duties, one might want to endorse that the surgeon’s killing the tonsillectomy patients is prima facie wrong. Thirdly, pace Graham, it might well be the case that, according to the best moral theory, the surgeon’s redistributing the organs would be bad, despite its leading to ten rather than only two people surviving the shortage of organs. Since social norms frame our expectations, emotions, desires, and moral beliefs, there are several possibilities to take them into account when ranking outcomes. One can hold, for example, that almost all members of our society would feel bad about the surgeon’s killing the tonsillectomy patients and desire it not to occur. As a consequence, the surgeon’s killing the two patients would be bad for us and in turn morally bad. Or one might argue that the two tonsillectomy patients may be more deserving of their organs than the other ten patients. If people getting what they deserve has intrinsic value, it might turn out that, all things considered, it would be bad if the surgeon redistributed the organs of the tonsillectomy patients.10 In sum, I have pointed out that there is a difference between moral duties and social duties. The sophisticated objection to Ought Implies Can does not take into account this difference. However, taking the difference into account leads to moral assessments of TRANSPLANT (COMPULSION) that are compatible with Ought Implies Can.

6. Conclusion I conclude that we ought to accept Ought Implies Can because the prin­ ciple is appealing in its own right and also yields plausible explanations of many apparently true moral judgements. The objections to Ought Implies Can fail upon reflection. For each objection, plausible views con­ cerning the cases in question are available to the proponent of Ought Implies Can, views that exhaustively describe what is going on from a moral point of view. These views are at least as appealing as the positions that contradict Ought Implies Can. In dealing with the objections, I have made use of a recipe that consists of three steps. Since the recipe seems plausible in itself and has proven successful at answering the three con­ sidered objections, I am optimistic that proponents of Ought Implies Can can use it to rebut future objections as well. In sum, then, the objections fail and it is reasonable to accept Ought Implies Can.

Notes 1. The explanatory appeal of Ought Implies Can is mentioned but dismissed in Graham 2011, pp. 372–379. 2. Hare 1979. A related strategy has been used for defending Ought Implies Can in Copp 2003 and Vranas 2007. Compare also, in a different context, Griffin 1977, esp. Section II.

In Defence of “Ought” Implies “Can”

61

3. Here the distinction between deontic and evaluative concepts, which I emphasized in the previous section, is important. 4. Against my contention that the acts of the extraordinary sadists and klep­ tomaniacs lack deontic status, one might object on grounds of the widely accepted principle that an act is permissible if and only if it is not impermis­ sible. I agree with this principle. In fact, I think that the behaviours of the extraordinary sadists and kleptomaniacs do not qualify as acts, because acts are something agents have control over. However, I call them acts for the sake of my opponent’s argument, since my opponent argues that acts can be wrong even if agents cannot refrain from performing them. I think my oppo­ nent’s position is naturally phrased in terms of acts rather than behaviour. So I will assume, for the sake of argument, that it is not true that an act is permissible if and only if it is not impermissible. 5. Zimmerman 1987, pp. 199f., and Streumer 2003. See also Goldman 1976, pp. 449f.; 1978, pp. 192f. 6. Why is it absurd to hold that moral demands are issued by a demander? The reason is that one can easily imagine situations that lack any candidate demander (God, society, the spaghetti monster) but that nonetheless involve moral demands. 7. Graham 2011, pp. 347f. That the surgeon’s sacrificing the “donors” would have good consequences – but seems to be wrong – is what makes TRANSPLANTlike cases a basis for objections against (evaluator-neutral) consequentialism. 8. For more on social norms, see Bicchieri, Muldoon, and Sontuoso 2018. 9. Let me name just a few accounts within an act-consequentialist framework. First of all, there are action-centred accounts such as, on the one hand, those focusing on group acts – see Mendola 2005; 2006a, part 1; 2006b, and Woodard 2019 – and, on the other, those focusing on expected value, see Kagan 2011. Action-centred accounts eventually build on the apparent para­ dox that arises if we assume that it is good if we as a group uphold a social practice but bad if each of us does. Secondly, act-consequentialists can adopt theories of the good according to which it is intrinsically good if people get what they deserve – see Feldman 1997 and Kagan 2012. In TRANSPLANT, the act-consequentialist might argue, the tonsillectomy patients deserve to keep their organs. The more complicated theory of the good developed in Pettit 2015 also promises to do the trick (but see my critique in Andrić 2019). Thirdly, act-consequentialists can endorse agent-relative theories, see Dreier 1993 and Smith 2003a. An agent-relative act-consequentialist can claim that it is good relative to the surgeon to do what is in line with the social practice of respecting organ autonomy. Fourthly, act-consequentialists can argue that, since our social practices frame our preferences and feelings, the surgeon’s killing the tonsillectomy patients would be wrong because it would fail to maximize well-being; see Fehige and Frank 2010, pp. 146–148, and Gesang 2003. 10. For references, see the previous note.

4

Objective Consequentialism and Ignorance-Induced Inability1

Some philosophers hold that objective consequentialism is false because it is incompatible with Ought Implies Can. Frances Howard-Snyder has used an innovative thought experiment to argue that sometimes you can­ not do what will in fact have the best consequences because you do not know what will in fact have the best consequences. This chapter’s aim is to show that Howard-Snyder’s argument for the incompatibility of objective consequentialism with Ought Implies Can fails. Erik Carlson has raised two objections to Howard-Snyder’s argument. I will show that, while Carlson’s first objection fails, his second objection ultimately suc­ ceeds (though not in the form he presents it). Moreover, I will provide a positive case for objective consequentialism being compatible with Ought Implies Can. Howard-Snyder’s argument will be presented in Section 1. In Section 2, I argue that Carlson’s first objection fails. Sections 3 and 4 deal with Carlson’s second objection. A summary follows in Section 5.

1. Howard-Snyder’s Argument The worry that objective consequentialism could be incompatible with Ought Implies Can has been known for a long time and recently been reinvestigated.2 At the outset of her article “The Rejection of Objective Consequentialism”, Howard-Snyder provides the following definition of objective consequentialism: Objective consequentialism is the view that the agent ought to per­ form an action of those available to her which will have consequences at least as good as the consequences of any other action available to her. (Howard-Snyder 1997, p. 241, footnote omitted) I take this definition to be equivalent to the one I presented in Section 2 of the Introduction:

Ignorance-Induced Inability

63

Objective Consequentialism An act is right if and only if the act’s outcome is at least as good as the outcome of any of the act’s alternatives; otherwise the act is wrong. An act is obligatory if and only if its outcome is better than the outcome of any of its alternatives. In the remainder of her article, Howard-Snyder (1997, e.g. on pp. 242, 243, 247) paraphrases objective consequentialism as the position accord­ ing to which agents ought to produce the best consequences. I think that Howard-Snyder’s own definition of objective consequen­ tialism suggests that objective consequentialism is compatible with Ought Implies Can. Since objective consequentialism judges only available acts to be right or wrong, it has built in the compatibility with Ought Implies Can. Objective consequentialism neither requires the non-swimmer to swim nor you to travel back in time, irrespectively of how good the con­ sequences would be. Nonetheless, Howard-Snyder thinks that she can demonstrate the incompatibility of objective consequentialism and Ought Implies Can by means of the Karpov case: Imagine that something important (say, your career) hangs on my winning a chess championship by defeating Karpov. Suppose I try to do so and fail. (Howard-Snyder 1997, p. 242) The crucial feature of the case is that the chess layperson – let us call her Frances – could not have beaten Karpov and therefore could not have produced the best consequences. Howard-Snyder (1997, pp. 242–243) has put forward the Karpov case as an analogy. The thought is that in ordinary choice situations, which series of choices will produce the best consequences is an even more complicated question than which series of moves will lead to victory over Karpov. Let us assume that not only good but the best consequences would have obtained if, and only if, Frances had beaten Karpov. And let us treat the Karpov case not as an analogy but as a candidate situation in which the agent cannot do what she ought to do according to objective consequen­ tialism. Howard-Snyder’s argument, then, can be summarized as follows: (1) According to objective consequentialism, you ought to produce the best consequences in all possible situations.3 (2) In some possible situations, you cannot produce the best consequences. (3) Ought Implies Can: Necessarily, a person is morally obligated to φ only if the person can φ.

64

“Ought” Implies “Can”

(4) Therefore, in some possible situations it is not true that you ought to produce the best consequences. (5) Therefore, objective consequentialism is false.4 Carlson’s first objection (which I discuss in Section 2) pertains to premise (2). Carlson’s second objection (see Sections 3 and 4) concerns premise (1). But before we turn to these objections, I will explain what I take to be the idea underlying Howard-Snyder’s argument. The argument’s central idea, I take it, is that ignorance can lead to inability, or more precisely that ignorance about the consequences of your options can make you unable to produce the best consequences. How could this idea work? Human agents often do not know the consequences of their options (cf. Chapter 1, Section 1). But how can Howard-Snyder get from this triviality to the conclusion that there are situations in which you cannot produce the best consequences? Two assumptions are needed. The first assumption is that your ignorance about the consequences of actions renders you ignorant about certain by-relations – such as those by-relations illustrated in Figure 4.1 for the Karpov case. If you do not know the consequences of your options, then you do not know by the performance of which act you would produce the best consequences. For instance, Frances does not know the consequences of the available chess moves. Let “e2–e4, etc.” designate the sequence of chess moves Frances has to play in order to beat Karpov. Frances does not know that it is by playing e2–e4, etc., that she would produce the best consequences. (To put it differently: Frances does not know that she has to play e2–e4, etc., in order to produce the best consequences; she does not know that her playing e2–e4, etc., is a way for her to produce the best consequences.) The general point, again, is that if you do not know the consequences of your options, then you do not know that certain by-relations hold. The second assumption is that the ignorance about by-relations leads to the inability to do what ranges at the higher levels of the chain of byrelations (see Figure 4.2). Frances can perform the relevant basic actions,5 that is, she can make the bodily movements that constitute her playing e2–e4, etc. And she can play e2–e4, etc. But since she does not know that

producing the best consequences beaAng Karpov by-relaAons playing e2-e4, etc. basic acAon

Figure 4.1 The structure of the Karpov case.

the box singles out Frances’s moral duty according to objecAve consequenAalism

Ignorance-Induced Inability

acAons Frances cannot perform

acAons Frances can perform

65

producing the best consequences beaAng Karpov playing e2-e4, etc. basic acAon

Figure 4.2 Frances’s “cans” and “cannots”.

by playing e2–e4, etc., she would be beating Karpov, she cannot beat Karpov. And in turn she cannot produce the best consequences. How can Howard-Snyder argue for the move from ignorance about by-relations to inability? Two ideas suggest themselves. One idea is to apply well-known theories about abilities to the Karpov case and show that these theories entail that Frances cannot produce the best conse­ quences. Another idea is to make use of intellectualism. Let us start with the second idea. Intellectualism is a view that connects knowledge-how and knowledgethat. As Jason Stanley explains: According to Intellectualism about knowing how, knowing how to do something is equivalent to knowing in what way one could do it. So, for example, you know how to ride a bicycle if and only if you know in what way you could ride a bicycle. But you know in what way you could ride a bicycle if and only if you possess some propositional knowl­ edge, viz. knowing, of a certain way w which is a way in which you could ride a bicycle, that w is a way in which you could ride a bicycle. (Stanley 2011, p. 209) If intellectualism is true, it follows that since Frances does not know that she has to play e2–e4, etc., in order to produce the best consequences, Fran­ ces does not know how to produce the best consequences. How does this support Howard-Snyder’s argument? If intellectualism is true, all HowardSnyder has to show is that Frances’s lack of knowledge-how to produce the best consequences leads to Frances’s inability to produce the best con­ sequences. This does not seem too difficult because talk of knowledge-how and talk of inability are intimately connected in everyday discourse. Intellectualism, however, is far from uncontroversial. Many philoso­ phers, influenced by Gilbert Ryle, believe that knowledge-how cannot be a species of knowledge-that because knowledge-how is an ability or a complex of dispositions, whereas knowledge-that is a relation between a thinker and a true proposition.6

66

“Ought” Implies “Can”

Jason Stanley and Timothy Williamson have recently defended intellec­ tualism on broadly linguistic grounds. They offer an influential critique of Ryle’s main argument as well as of other arguments for non-intellectualism (Stanley and Williamson 2001, pp.  412–416 and Section  3). However, Stanley’s and Williamson’s account is still hotly disputed (see the refer­ ences in Stanley 2011). Intellectualism has some intuitive appeal. Often you perform an action by performing another action and it seems true that you know how to perform the former action if and only if you know that you would per­ form it by performing the latter action. However, intellectualism is not very plausible when it comes to the performance of basic actions. Since you do not perform a basic action by performing another action, there is no by-relation you could have knowledge of in order to know how to perform a basic action. But since you can perform basic actions, it seems plausible to say that you know how to perform them. It would go beyond the scope of this chapter to discuss intellectualism.7 It seems clear that if intellectualism is true, or even true only with respect to non-basic actions, and I guess it might well turn out to be, then this lends support to Howard-Snyder’s argument. Let us then turn to the second way to connect ignorance about byrelations to inability. Here we can start with Howard-Snyder’s own ability-theoretical remarks. Howard-Snyder (1997, pp. 244–245) argues that an appropriate account of ability should make reference to why we are interested in it. According to her, we want to know how to under­ stand the “can” implied by the “ought” in Ought Implies Can. HowardSnyder connects “ought” to moral responsibility: To say that the agent ought to φ is to make the agent morally respon­ sible for φ-ing. To say that she ought to have φ-ed is to make her morally responsible for not φ-ing. (Howard-Snyder 1997, p. 245) Howard-Snyder goes on to argue that the attribution of responsibility makes sense only if the agent’s φ-ing is up to her, and that it is up to her only if her choosing to φ makes it at least very likely that she will φ. Hence, the “ought” in Ought Implies Can suggests that we should understand the “can” in such a way that an agent’s ability to perform an action consists in her choice’s making it very likely that she will perform the action. Howard-Snyder seems to have in mind a variant or a derivative of the so-called conditional analysis of ability, one of the three major accounts of ability.8 According to the conditional analysis, an agent can perform an action if and only if the agent would perform the action if she tried (or chose, willed, . . .) to perform the action. According to Howard-Snyder’s account, the truth of the conditional, though not sufficient, is neces­ sary for an agent’s ability to perform an action. Howard-Snyder (1997, p. 244) thus seems to endorse the following principle:

Ignorance-Induced Inability

67

Howard-Snyder’s Necessary Condition for Ability An agent can φ only if it would be very likely that the agent φ-ed if she tried (chose, etc.) to φ. How does this principle support the move from ignorance about byrelations to inability? The answer is, of course, that since Frances does not know that it is by playing e2–e4, etc., that she would produce the best con­ sequences and since there are countless series of chess moves Frances could play, Frances’s attempt to produce the best consequences would not make it very likely that she will produce the best consequences (cf. HowardSnyder 1997, p. 245). It follows from Howard-Snyder’s necessary condi­ tion for ability that Frances cannot produce the best consequences. Against Howard-Snyder’s necessary condition for ability, one could object that in order for an agent to be able to φ, it is not always necessary that the agent’s choosing to φ would make the agent’s φ-ing at least likely. For actually performing an action may be a sufficient condition for hav­ ing the ability to perform that action even if it is not true that an agent’s choosing to φ would make her φ-ing very likely. J.L. Austin has famously put forward the example of a golfer who sinks a difficult putt. Austin (1956a, p. 218) argued that “it follows merely from the premise that he does it, that he has the ability to do it, according to ordinary English”. Howard-Snyder (1997, p.  244) argues that actually performing an action is not sufficient for having the ability to perform the action. I will leave it open whether her argument is successful. Although the condi­ tional analysis seems at least to be a good rule of thumb for ascribing abilities or inabilities, respectively, there are more objections to it than the one just considered. However, insofar as other accounts of abilities also lead to the conclusion that Frances cannot produce the best conse­ quences, Howard-Snyder’s necessary condition for ability is dispensable. And as I will argue in the following section, other accounts of ability, in order to be convincing, indeed must lead to the conclusion that Frances cannot produce the best consequences. At this point, it is enough to point out that, since Frances does not actually beat Karpov but tries to do so and fails, it does not follow from “ordinary English” (as Austin put it) that Frances can beat Karpov.

2. Carlson’s First Objection In developing his objections to Howard-Snyder’s argument, Erik Carlson discusses a situation involving a safe: We might compare the task of producing the best consequences with that of opening a safe to which you do not know the right combina­ tion. Suppose that the right combination is 448–961–5237. Because you can dial any ten-digit combination, you can dial this one. But

68

“Ought” Implies “Can” since you do not know that this is the right combination .  .  . you cannot open the safe. (Carlson 1999, p. 91)

Let us assume that the person who does not know the right combination – Erik – had only one chance to dial 448–961–5237 in order to produce the best consequences. Erik thus did not have time to find out what the right combination is. Erik tried to open the safe and failed. Provided this modi­ fication, the case of the safe is as supportive of Howard-Snyder’s argument as the Karpov case. Carlson’s first objection says that Erik could have opened the safe – just as Frances could have beaten Karpov and, in general, you can pro­ duce the best consequences.9 But why should one believe that Erik could have opened the safe? The answer, according to Carlson, is that he could have dialled 448–961–5237 and that, by doing so, he would have opened the safe. At this point, Carlson introduces two views on act-individuation and two corresponding derivatives of the conditional analysis: If you were to try to dial 448–961–5237 you would succeed in doing this. On the other hand, if you were to try to produce the best con­ sequences you would fail. Assume that we wish to claim that you can, nevertheless, produce the best consequences. If producing the best consequences and dialling 448–961–5237 are one and the same action, this claim is compatible with the following condition: A per­ son can perform an action a only if there is some description of a, such that if he were to try to perform a under this description he would succeed. If, on the other hand, producing the best consequences and dialling 448–961–5237 are different actions, the claim that you can produce the best consequences is still compatible with this condition: A person can perform an action a only if there is some action b, such that (i) he would do a by doing b, and (ii) if he were to try to perform b he would succeed.10 The challenge for Howard-Snyder is to point out why her necessary con­ dition for ability, according to which an agent’s trying to produce the best consequences makes it at least very likely that the agent will produce the best consequences, is preferable to Carlson’s alternatives. Against Carlson’s objection, Howard-Snyder argues that, although there is a sense of “can” in which it is true that Frances could have beaten Karpov, Frances could not have beaten Karpov in the sense required by Ought Implies Can in order for her to have had an obligation to beat Karpov. In support of this claim, Howard-Snyder points out that in everyday discourse we would assume that Frances could not have beaten Karpov.11 Howard-Snyder also alludes to the connection between Ought

Ignorance-Induced Inability

69

Implies Can and moral responsibility, which we encountered in the previ­ ous section. Howard-Snyder is right: There are strong common-sense intuitions to the effect that Frances could not have beaten Karpov in a morally rel­ evant sense and that, in consequence, it would be unfair to blame her for failing to do so. Hence, common-sense intuitions count against Carlson’s first objection. Howard-Snyder’s argument can be strengthened in two respects. Firstly, Eric Wiland (2005, p.  357) has added further examples of presumably ignorance-induced inabilities: you cannot write an article that explains how to cure AIDS, you cannot write the Great American Novel, you can­ not write the Islamic equivalent of Locke’s Letters on Toleration. Let us focus on your inability to write an article that explains how to cure AIDS. Given the astronomically good consequences your writing an article that explains how to cure AIDS would have, it seems that if you could write the article, you would be obligated to do so not only according to con­ sequentialism but also according to common-sense morality. This allows for the following argument: (6) If you could write an article that explains how to cure AIDS, then common-sense morality would require you to write it. (7) But you are not required to write the article according to common­ sense morality. (8) Therefore, you cannot write it. Here is a further reason to accept Howard-Snyder’s reply to Carlson. Howard-Snyder has not explicitly addressed Carlson’s challenge to show what is wrong with his alternatives to her necessary condition for ability. But this can easily be fixed. Any persuasive theory of ability has to do justice to our ordinary judgements about abilities and hence has to obey certain extensional constraints: It must not ascribe too many or too few abilities to agents (Maier 2018, Section 2.1). A theory of ability according to which you can write an article that explains how to cure AIDS ascribes too many abilities, as from the perspective of everyday discourse, and thus violates an extensional constraint. In sum, Carlson’s first objection fails. There are strong common-sense intuitions to the effect that Frances could not have beaten Karpov. A persuasive theory of ability has to take these intuitions into account. It follows that Frances could not have produced the best consequences.

3. Carlson’s Second Objection Howard-Snyder (1997, p. 245) discusses the following objection: Objec­ tive consequentialists “tell us to perform that action, of those we can perform, which will have the best consequences”. Frances can play e2–e4,

70

“Ought” Implies “Can”

etc. By implying that Frances ought to play e2–e4, etc., objective con­ sequentialism does not require anything impossible from Frances. This is, of course, the worry I mentioned already when presenting Howard­ Snyder’s argument. Howard-Snyder has a twofold reply to the objection. First, she claims that objective consequentialism by definition does not only require Frances to play e2–e4, etc., but also to do what has the best consequences. And Frances cannot do the latter. Second, if we mod­ ify objective consequentialism such that it does not anymore require Frances to do what has the best consequences but only to play e2– e4, etc., then objective consequentialism becomes a very undesirable moral theory. Why does Howard-Snyder think that we cannot do what objective consequentialism, by definition, requires us to do? Here is an instruction which objective consequentialism issues and which I cannot obey: “perform that action, of those you can perform, which will have the best consequences.” I cannot perform this action for the same reason I cannot beat Karpov. But . . . it is true by defini­ tion that objective consequentialism requires this of me. (Howard-Snyder 1997, p. 245) The details are important: According to Howard-Snyder, Frances can play e2–e4, etc., and her playing e2–e4, etc., would have the best conse­ quences. But Frances cannot obey the instruction to do what would have the best consequences. For Frances does not know that playing e2–e4, etc., is the action with the best consequences. Carlson has replied to the claim that Frances cannot do what has the best consequences. He argues that a distinction between properties of acts and act-types is crucial for rebutting Howard-Snyder’s argument: It might be thought, however, that the objective consequential­ ist involves herself in a contradiction, by denying that you ought to produce the best consequences. Is she not, after all, committed to precisely the claim that you ought to produce the best conse­ quences? The answer is yes and no. The question whether you ought to produce the best consequences is ambiguous. On one interpreta­ tion, it is the question whether there is some action that you ought to perform, which has the property of producing the best conse­ quences available. The answer to this question is, “Yes”, according to objective consequentialism. Your dialing 448–961–5237 is such an action. On the other interpretation, the question is whether you ought to perform an action of the act-type “producing the best con­ sequences”. The objective consequentialist’s answer to this question

Ignorance-Induced Inability

71

may well be no, since it may be the case that you cannot perform an action of this type. (Carlson 1999, p. 94, footnote omitted) If Frances cannot produce the best consequences, then, so the suggestion goes, objective consequentialism does not require her to perform an act of the act-type “producing the best consequences”. Objective consequen­ tialism only requires Frances to produce an act that has the property of “having the best consequences”. Since Frances can play e2–e4, etc., and Frances’s playing e2–e4, etc., has the property of having the best con­ sequences, Frances can perform the act that has the best consequences. Hence, objective consequentialism is compatible with Ought Implies Can. Carlson’s suggestion is illustrated in Figure 4.3.12 I think that Carlson is right in that, according to objective consequen­ tialism, Frances has a moral duty to play e2–e4, etc. However, I think that Carlson’s distinction between act-types and act properties is not sufficient for rebutting Howard-Snyder’s argument. First, Carlson fails to make clear what exactly it is that he has in mind. Is his view that (i) according to objective consequentialism, Frances has a moral duty to perform an act that has the property of having the best consequences or that (ii) according to objective consequentialism, Frances has a moral duty to play e2–e4, etc.? Howard-Snyder (1999) has responded to both possible interpretations. If we accept (i), then Howard-Snyder’s argument still goes through. For Frances does not know what she has to do in order to do what has the property of having the best consequences.13 If we accept (ii), then objec­ tive consequentialism does not require Frances to perform an act that has the property of having the best consequences but only to play e2–e4, etc. On this reading, objective consequentialism delivers an incredibly huge list of moral duties that apparently have nothing in common. Frances has the moral duty to play e2–e4, etc., I may have the moral duty to check my emails, you may have a moral duty to make two steps to the left, and

producing the best consequences beaAng Karpov playing e2-e4, etc. / doing what has the property of having the best consequences basic acAon

Figure 4.3 Carlson’s suggestion.

72

“Ought” Implies “Can”

so on. Since this list lacks coherence, objective consequentialism becomes an undesirable doctrine.

4. Carlson’s Second Objection Revisited I think that Carlson should have chosen (ii): The only moral duty objec­ tive consequentialism implies in Frances’s case is to play e2–e4, etc., cf. Figure 4.4. Does this mean that objective consequentialism becomes an undesirable, “huge-list” theory? I submit that objective consequentialists can stick to the established definition of objective consequentialism and nonetheless insist that Frances’s only moral duty is to play e2–e4, etc. How is this possible? The general definition of objective consequentialism, according to which an act is obligatory if and only if it has better consequences than any alternative, does not state a moral duty. Rather, it is a statement of necessary and sufficient conditions for an act’s being obligatory. It is an act’s having better consequences than its alternatives which makes the act obligatory. The point of the general definition of objective consequentialism is to explain what makes an act right or obligatory by stating the right-making or ought-making features. As we have seen in the Introduc­ tion (Section 1), it is the point of objective consequentialism qua moral doctrine to provide explanations of this sort. Thus, objective consequen­ tialism explains the rightness of Frances’s playing e2–e4, etc., by the fact that this action has the property of having the best consequences. Objective consequentialists can have their cake and eat it, too: They can endorse a general formula, which states the conditions under which acts are right, wrong, or obligatory, respectively, and at the same time deny that their theory issues any general instructions that agents would not be able to follow. Due to the general formula, the huge list of instruc­ tions issued by objective consequentialism does not lack coherence at all. At the same time, it does not make sense for Frances to complain that she did not know how to obey the instruction “do what has the

producing the best consequences / doing what has the property of having the best consequences beaAng Karpov playing e2-e4, etc. basic acAon

Figure 4.4 The solution.

Ignorance-Induced Inability

73

best consequences”, for objective consequentialism, properly understood, does not say that Frances has a moral obligation to do what has the property of having the best consequences. Of course, it does make sense for Frances to complain that she did not know which action she ought to perform according to objective consequentialism. But this is a different issue, which we will consider in the next chapter. As we have seen in Sec­ tion 1, Howard-Snyder’s argument is based on the idea to derive inability from lack of knowledge-how, not from lack of knowledge-which. A defender of Howard-Snyder’s argument could try to object: Since Frances did not know that objective consequentialism required her to play e2–e4, etc., she did not know how to do what objective consequentialism required her to do. Therefore, Frances was unable to do what objective consequentialism required her to do. But the reply is the same as before: According to objective consequentialism, you do not have a moral obligation “to do what objective conse­ quentialism requires”. In most situations, we do not know which act we have a moral obligation to perform according to objective consequentialism. But we know that it is an act with consequences at least as good as those of any of its alternatives. Here is a related objection. It seems to be trivially true that you ought to do what you ought to do. However, on the view defended here, appar­ ently, in cases like that of Karpov agents cannot do what they ought to do: While Frances can, and ought to, play e2–e4, etc., she cannot do what she ought to do – despite the fact that by playing e2–e4, etc., she would be doing what she ought to do. Now, if a theory implies that you can­ not do what you ought to do, the theory seems to be incompatible with Ought Implies Can after all. The objection focuses on this claim: (9) Necessarily, you can do what you ought to do. Does objective consequentialism, in the version I suggest, contradict (9)? It depends. Let me start by repeating the distinction between a coarsegrained and a fine-grained account of act individuation, which has been introduced in Section 3. On the coarse-grained account Frances’ playing e2–e4, etc., and her doing what she ought to do are different descriptions of one and the same act – just like her playing e2–e4, etc., and her beating Karpov. The proper object of obligation on this view of act individuation are not acts per se but acts under certain descriptions: not Frances’ beating Karpov – under this description of her act – but her playing e2–e4, etc., – under this description of her act. On the fine-grained view, by contrast, Frances’ playing e2–e4, etc., and her doing what she ought to do are not just different descriptions of the same act but different acts. Now, if (9) is

74

“Ought” Implies “Can”

meant to have the same truth-conditions as Ought Implies Can (the doc­ trine that, necessarily, a person ought to φ only if the person can φ), then the objective consequentialist can accept (9). If, however, (9) is meant to be identical, on a coarse-grained view of act individuation, with: (10) Necessarily, if you ought to perform an act, then you can do what you ought to do (under this description of your act), or, on a fine-grained view, (11) necessarily, if you ought to perform an act, then you can perform the act of doing what you ought to do, then, indeed, I think that the objective consequentialist should reject (9). At any rate, this chapter attempts to show that objective consequentialism is compatible with Ought Implies Can and can be silent on (10) and (11). But maybe this is too quick. Cannot something like (10) or (11) be derived from the triviality that you ought to do what you ought to do in conjunction with Ought Implies Can? The challenge for the objective consequentialist is to show that: (12) You ought to do what you ought to do is not trivially true on a reading that, when combined with Ought Implies Can, leads to (10) or (11). Let us start by disambiguating the claim that you ought to do what you ought to do. (12) might be understood as follows: (13) If you ought to perform an act, then you ought to perform this act. (13) is indeed trivially true. (13) implies, for instance, that, if Frances ought to play e2–e4, etc., then Frances ought to play e2–e4, etc. However, neither (10) nor (11) can be derived from (13) in conjunction with Ought Implies Can. (13) is compatible with objective consequentialism. Let us now come to the problematic interpretations of (12). Here is one interpretation, based on a coarse-grained view of act individuation, of the claim that you ought to do what you ought to do that is indeed incompatible with objective consequentialism in the version defended in this chapter: (14) If you ought to perform an act, then you ought to perform this act under the description “act which I ought to perform”. (14) is not trivially true. Objective consequentialists typically hold that the desire to produce good consequences should not always be the motive

Ignorance-Induced Inability

75

of your actions. From there it is only a small step to the claim that you should not always perform obligatory acts under this description. As long as the defender of Howard-Snyder’s argument does not offer an argu­ ment in support of (14), the objective consequentialist can dismiss (14) as question-begging. On the fine-grained view of act individuation, a natural understanding of (12) – apart from (13), which turned out to be unproblematic – is this: (15) If you ought to perform an act, then you ought to perform the act of doing what you ought to do. Of course, (15) in conjunction with Ought Implies Can is incompatible with the version of objective consequentialism suggested here. However, (15) is not trivially true either – for the same reasons that apply to (14). The bottom line is that the objective consequentialist can reject the claim that, necessarily, you can do what you ought to do, insofar as this claim contradicts objective consequentialism. At least, this is true if one tries to justify the claim on the basis of the alleged triviality that you ought to do what you ought to do in conjunction with Ought Implies Can.

5. Conclusion This chapter examined Howard-Snyder’s argument for the incompatibil­ ity of Ought Implies Can with objective consequentialism. I have argued that her argument fails. The argument fails because, although objective consequentialism tells you qua moral doctrine that an act is obligatory if and only if its consequences are better than the consequences of any alter­ native, objective consequentialism does not say that you have an obliga­ tion to do whatever has better consequences than any alternative. The former claim is a statement about necessary and sufficient conditions for an act being obligatory, not the expression of a moral duty. We have also seen a general reason for thinking that objective con­ sequentialism is compatible with Ought Implies Can. Objective conse­ quentialism ranges only over acts that you can perform (or omit) when assigning rightness or wrongness. I conclude that objective consequen­ tialism is compatible with Ought Implies Can.

Notes 1. This chapter is a slightly revised version of Andrić 2016. 2. As H.J. McCloskey (1973, p.  62) put it: “We are not truly free to do the right or obligatory act if we cannot in advance know what is the right act. ‘Ought implies can’, and ‘Can’ implies ‘Can know’. . . . Yet if utilitarianism is to remain the success theory, that it is actual consequences that count, then it is committed to the view that we may be subject to obligations of which it is impossible for us to be aware.” The topic has more recently been

76

“Ought” Implies “Can”

reinvestigated in Bergström 1996, Howard-Snyder’s 1997; 1999; 2006, Carlson 1999, Qizilbash 1999, Mason 2003, Miller 2003, Wiland 2005, and Moore 2007. 3. “All possible situations”, just like “some possible situations” in the next premise, only quantifies over situations that involve you as an agent. 4. A similar summary is given in Howard-Snyder 1999, p. 106. 5. An action is non-basic if and only if it is performed by performing another action. An action is basic if and only if it is not non-basic. The distinction between basic and non-basic actions has been put forward in Danto 1963 and Chisholm 1964. It is controversial whether there are basic actions and, if there are, whether they are bodily movements (as I assume), volitions (McCann 1974), or skilled actions (Baier 1971; Ripley 1974). 6. Ryle 1945; 1949, Chapter 2. For an instructive overview, see Fantl 2017. 7. Another objection suggests itself: Though knowledge how to beat Karpov at chess seems to require propositional knowledge, the knowledge appears to be about good chess moves and strategies rather than about chains of by-relations. However, the intellectualist might reply that if someone has the former knowledge, then, in order to know how to beat Karpov, she must be able to infer the latter knowledge. As an analogy, someone who knows how to open a safe must not only know the right combination but also which bodily movements she has to make in order to open the safe. 8. The three major accounts of ability are the conditional analysis (see, e.g. Moore 1912, Chapter 6, and Ayer 1954), a modal account (see, e.g. Lehrer 1976; Lewis 1976; Kratzer 1977; Hurley 2000), and the so-called new dis­ positionalism (see, e.g. Smith 2003b; Vihvelin 2004; Fara 2008). 9. Carlson 1999, p. 92. The objection has already been addressed in HowardSnyder 1997, p. 243. 10. Carlson 1999, p.  92, footnote omitted. The locus classicus for the finegrained view is Goldman 1970, Chapter 1. The loci classici for the coarsegrained view are Anscombe 1957, pp. 45–46, and Davidson 1963, p. 686. 11. Howard-Snyder 1999, p. 107, talks about how we do “in fact understand and use” the principle that “ought” implies “can”. 12. Should the box in Figure 4.3 (also or only) single out the basic action? Per­ haps. But for the question we are concerned with (is objective consequentialism compatible with Ought Implies Can?), this does not matter. 13. Moreover, as Howard-Snyder 1999, p. 109, points out, if an act type is noth­ ing but the set of all acts that have a certain property in common, Carlson may not even be able to uphold his distinction.

5

Four Arguments Against Objective Consequentialism Related to “Ought” Implies “Can”1

I have argued in the foregoing chapter that objective consequentialism is compatible with Ought Implies Can. However, I think that Frances Howard-Snyder was on to something when claiming otherwise. Do we really want our final verdict about the Karpov case to be that Frances can be morally obligated to play e2–e4, etc., given that Ought Implies Can rules out the possibility of her being obligated to beat Karpov? I suspect that many people feel uneasy with this verdict. Instead of focusing on Frances’s beating Karpov in contrast to her playing e2–e4, etc., in this chapter I will focus, for reasons to be stated in Section 1, on my writing an article that explains how to cure AIDS in contrast to my writing “O”, “n”, and so on. Ought Implies Can rules out the possibility of my having a moral duty to write an article that explains how to cure AIDS. But Ought Implies Can is silent on my hav­ ing a moral duty to write “O”, “n”, and so on, as the discussion in the previous chapter has shown. And objective consequentialism apparently implies that I ought to write “O”, “n”, and so on. But can this implication really be true, given that I cannot have an obligation to write an article that explains how to cure AIDS? In this chapter, I argue that, even though objective consequentialism is compatible with Ought Implies Can, it is incompatible with some con­ siderations underlying, or lending credence to, Ought Implies Can. The considerations are (i) intuitive plausibility, (ii) the action-guidingness of moral theories, (iii) considerations of blameworthiness and fairness, and (iv) the nature of moral reasons. Karpov-like cases enable us to see that objective consequentialism is incompatible with these considerations. The reason is that Karpov-like cases involve acts that cannot be obliga­ tory according to Ought Implies Can as well as closely related acts that are obligatory according to objective consequentialism, and we can sim­ ply compare these two acts with respect to the considerations underly­ ing Ought Implies Can. What we find, I submit, is that the acts that are obligatory according to objective consequentialism are as much at odds with the considerations underlying Ought Implies Can as the acts that cannot be obligatory according to Ought Implies Can.

78

“Ought” Implies “Can”

First of all, I will point out that the judgement that I ought to write “O”, “n”, and so on, is intuitively implausible, in light of the fact that I cannot have an obligation to write an article that explains how to cure AIDS. Secondly, objective consequentialism is not properly actionguiding as a moral theory, given that it implies that I ought to write “O”, “n”, and so on. Thirdly, objective consequentialism can be accused of not taking into account considerations of blameworthiness and fairness, if my not discharging a moral duty renders me a candidate for blame. Fourthly, objective consequentialism gives an inappropriate account of practical reasons, if my alleged moral duty to write “O”, “n”, and so on is supposed to bring along practical reasons to write “O”, “n”, and so on. This gives us four new arguments against objective consequentialism. Section 1 deals with intuitive appeal. Section 2 is about action-guidance. The topic of Section 3 is blameworthiness and fairness. I examine the nature of practical reasons in Section 4. I sum up the results in Section 5.

1. Just as Counterintuitive as Obligations to Perform Impossible Acts Ought Implies Can rules out the possibility of my having certain obligations – obligations that I intuitively do not have. Objective consequentialism, I submit, implies that I have obligations that are just as counterintuitive as some of the obligations ruled out by Ought Implies Can. This gives us the following argument against objective consequentialism: The Argument From Impossibility Comparisons (1) Objective consequentialism implies that I have obligations that are just as counterintuitive as some of the obligations ruled out by Ought Implies Can. (2) If a moral theory implies that I have obligations that are just as coun­ terintuitive as some of the obligations ruled out by Ought Implies Can, then the moral theory is false. (3) Therefore, objective consequentialism is false. Objective consequentialism, we have seen in the previous chapter, requires Frances to play e2–e4, etc. Howard-Snyder is right that the objective con­ sequentialist verdict in the Karpov case is extremely implausible. As we have seen, however, Howard-Snyder is wrong that objective consequen­ tialism’s implausible verdicts are incompatible with Ought Implies Can. For Frances can play e2–e4, etc. Ought Implies Can only rules out the possibility of Frances’ having an obligation to beat Karpov. Intuitively, however, it seems just as implausible that Frances has an obligation to play e2–e4, etc., as it seems that she has an obligation to beat Karpov. This suggests that objective consequentialism is false.

Four Arguments

79

Let me focus on another case. Recall an observation by Eric Wiland (2005, p. 357), which I have already alluded to in the previ­ ous chapter: Nearly every medical layperson has the physical and intellectual ability to make the series of letters, punctuation marks, and spaces that, taken together, form an article which explains how to cure AIDS. I am not morally obligated to write an article that explains how to cure AIDS because I cannot write such an article. What I can do, as we have seen in the previous chapter, is this: I can write every single letter the article that explains how to cure AIDS consists of. I can write: “O”, “n”, “_”, “H”, “o”, “w”, “_”, “t”, “o”, “_”, “C”, “u”, “r”, “e”, “_”, “A”, “I”, “D”, “S”, and so on. For brevity, let “O”, “n”, and so on designate the sequence of letters that constitutes the article on how to cure AIDS. The structure of the AIDS article case is similar to the structure of the Karpov case, which we have considered in the previous chapter. Why do I want to focus on the AIDS article case in addition to the Karpov case? The reason is that all plausible moral theories entail that I have a moral duty to write the AIDS article if I can do this whereas many moral theories do not entail that Frances ought to beat Karpov just because this happens to have the best consequences. It therefore might be more obvious that Ought Implies Can rules out an obliga­ tion that I otherwise would have. Ought Implies Can explains these judgements: (4) I am not morally obligated to write an article that explains how to cure AIDS. (5) If I could write an article that explains how to cure AIDS, then I would be morally obligated to do this. However, Ought Implies Can is silent on the following judgements: (6) I am not morally obligated to write “O”, “n”, and so on. (7) If I knew that by writing “O”, “n”, and so on I would be writing an article that explains how to cure AIDS, then I would have a moral obligation to write “O”, “n”, and so on. Intuitively, however, (6) and (7) are no less plausible than (4) and (5). If we accept (4) and (5), it would be strange for us not to accept (6) and (7). This shows, again, that objective consequentialism has an implication – that I have an obligation to write “O”, “n”, and so on – which is just as counterintuitive as a view ruled out by Ought Implies Can – that I have an obligation to write an article that explains how to cure AIDS.

80

“Ought” Implies “Can”

2. The Argument From Action-Guidance This section establishes the Argument from Action-Guidance, which claims that objective consequentialism is false because it is not properly action-guiding. First, I present a specific argument to that effect. Then I lay out a general argument. The specific argument to the effect that objective consequentialism is not properly action-guiding goes as follows: (8) A moral theory that requires me to write an article that explains how to cure AIDS is not properly action-guiding and, therefore, false. (9) A moral theory that requires me to write “O”, “n”, and so on is no more properly action-guiding than a moral theory that requires me to write an article that explains how to cure AIDS. (10) Objective consequentialism requires me to write “O”, “n”, and so on. (11) Therefore, objective consequentialism is false. As I have mentioned in Chapter 3, Section 1, Ought Implies Can can be derived from the assumption that a crucial function of moral theo­ ries consists in providing proper action-guidance. In the following, I will first remind us of why we should believe that moral theories should be action-guiding. Then, I will show that the desideratum of proper actionguidance explains both why I cannot have a moral obligation to write an article that explains how to cure AIDS, as (8) states, and why a moral theory that requires that I write “O”, “n”, and so on is no more properly action-guiding than a moral theory that requires that I write an article that explains how to cure AIDS, as (9) has it. Why believe that moral theories should be action-guiding? I mentioned two reasons before. First, we want to apply moral theories; we want them to tell us what the moral thing to do is and we want to implement moral theories by doing what they require.2 As I said at the outset of this book: Life poses moral challenges; and moral theories are supposed to yield convincing deontic judgements in order to equip us with knowledge about what ought to be done on which we can base our decisions. We have seen in Chapter 1 that for a moral theory to be action-guiding in this sense, the theory must be decision-guiding in the sense that, when the theory is supplemented with a plausible decision procedure, we must be able to implement it. The second reason is that moral theories must make sense of the deon­ tic vocabulary we use in everyday life. Terms like right, wrong, and oblig­ atory are essentially action-guiding, whereas evaluative notions like good or bad are not (see, e.g. Smith 1986, p. 342). As I already explained in Chapter 3, Section 1, evaluative properties can properly be ascribed to all sorts of entities, whereas the properties of rightness and wrongness

Four Arguments

81

primarily apply to actions or choices. Conceptual analysis thus reveals the action-guiding function of moral theories. The two reasons why moral theories should be action-guiding also explain why, as (8) states, a moral theory is not properly action-guiding if it requires me to write an article that explains how to cure AIDS. Firstly, if a moral theory tells us to perform an action that we cannot perform, such as writing an article that explains how to cure AIDS, the theory is of no help in situations in which we wonder how to act. The theory is pointless because we cannot implement it.3 Secondly, the theory fails to distinguish between what would be very good and what we ought to do. It would be very good, for example, if I wrote the AIDS article. But it seems false to say that I ought to write it. A corresponding argument for (9) suggests itself. According to (9), a moral theory that requires me to write “O”, “n”, and so on, is no more properly action-guiding than a moral theory that requires me to write an article that explains how to cure AIDS. Firstly, if we want a moral theory to tell us what to do, it is of no help if the moral theory implies that the right course of action is one that we cannot identify. I am at a complete loss if a moral theory tells me to write an article that explains how to cure AIDS because I do not know how to write such an article. I am left equally clueless if I know that there is a complex sequence of letters (“O”, “n”, and so on) that I ought to write but cannot identify. As far as practical decision-making is concerned, it does not make a difference if I can identify but not implement a theory’s instruction or if I cannot even identify the theory’s instruction. Either theory is unhelpful in real life if I want to do what I ought to. Let us come to the second reason why a moral theory that requires me to write “O”, “n”, and so on, is no more properly action-guiding than a moral theory that requires me to write an article that explains how to cure AIDS. While it seems appropriate to say that, in light of the results, it would be very good if I wrote the sequence of letters in question, it seems false to say that I have a moral duty to do so. A moral theory that requires me to write “O”, “n”, and so on, ascribes action-guiding proper­ ties (here: being someone’s moral duty) to actions that do not have them. This moral theory has the same kind of defect as a moral theory that requires me to write the article about how to cure AIDS. This concludes my substantiation of the specific argument. Let us now turn to the general argument. The general argument is based on an analy­ sis of what it takes for a moral theory to be properly action-guiding in a sense that underlies Ought Implies Can, which shall be developed now. As a starting point, recall that I have said, on the one hand, that we want to be able to base our decisions on a moral theory in order to meet life’s moral challenges. On the other hand, I have argued in Chapter 1 that for a moral theory to be decision-guiding, it is enough that the theory can be supplemented with a decision procedure that solves the problem

82

“Ought” Implies “Can”

of decision-guidance. How do these claims fit together? There seems to be a tension because we would like to base our decisions on a moral theory, rather than a moral theory combined with a decision procedure. I think that proper action-guidingness in the context of Ought Implies Can is related to, but not identical with, what it takes for a moral theory to be decision-guiding in the sense that I introduced in Chapter 1. There I argued that a moral theory fails to be decision-guiding if an ordinary human follower of the theory can rationally ask Q and the theory can­ not be supplemented with a plausible decision procedure that answers Q. Q is the question, “Given that I do not know the ought-making, right-making, and wrong-making features of the acts available to me, which act ought I to choose?” Now I want to suggest that for a moral theory to be properly action-guiding in the sense underlying Ought Implies Can, the theory must be decision-guiding for those of its fol­ lowers whose beliefs are in accord with the factual evidence available to them even if the theory is not supplemented with a decision procedure. Put differently, for a moral theory to be properly action-guiding, ordi­ nary humans who accept and understand the theory and whose factual beliefs are in line with the evidence available to them must be able to deduce from the theory itself how they ought to decide – they will not ask Q. For this to be the case, the follower of a moral theory must be able to identify (what according to the theory are) the right-making and wrong-making features of the acts available to her. In short, the sugges­ tion is this: (12) In order to be properly action-guiding, a moral theory must be decision-guiding without an additional decision procedure for those of its followers whose beliefs are in accord with the factual evidence available to them. The general version of the Argument from Action-Guidance has one fur­ ther premise: (13) Objective consequentialism is not decision-guiding without an addi­ tional decision procedure for those of its followers whose beliefs are in accord with the factual evidence available to them. (14) Therefore, objective consequentialism is false. (13) is clearly true. Since a follower of objective consequentialism cannot identify the sequence of letters under consideration (“O”, “n”, and so on), she does not know that writing “O”, “n”, and so on has an oughtmaking feature. Why believe (12)? We are concerned with the sense of action-guidingness that features in the justification of Ought Implies Can and that is connected to the desideratum to use moral theories themselves as decision-guides.

Four Arguments

83

Given this context, I can think only of two alternatives to my proposal. But both alternatives fail. The first alternative is the claim that for a moral theory to be properly action-guiding, followers of the theory must always be able to use the theory itself as a guide for their decisions in light of their actual beliefs. This alternative fails because, as I have explained previously (Chapters 1, Section 3; Chapter 2, Section 4; Introduction, Section 2), the sense in which moral theories should be action-guiding is not identical with the sense in which decision procedures should be decision-guiding. In order to be decision-guiding without an additional decision procedure, a moral theory would have to base what an agent has a moral duty to do on the agent’s actual beliefs. For otherwise, the problem of decision-guidance would occur whenever the agent lacks certain beliefs about the right- and wrong-making features of her options. However, a moral theory yields very implausible judgements if it bases what an agent ought to do on the agent’s actual beliefs. Therefore, it is not the case that moral theo­ ries should generally be decision-guiding without an additional decision procedure. The other alternative is the view that for a moral theory to be properly action-guiding the following must hold: We would be able to deduce our decisions from the theory itself if we knew all the relevant facts. On this view, moral theories themselves would not need to be decision-guiding for ordinary human persons but only for cognitively ideal agents, that is, for agents who possess all the factual knowledge and the reasoning capacities necessary for correctly inferring what the moral theory in ques­ tion prescribes.4 This suggestion fails because a cognitively ideal agent would be able, in the morally relevant sense, to write an article that explains how to cure AIDS. Therefore, Ought Implies Can, were it derived from considerations of action-guidance as understood in this suggestion, would fail to rule out the possibility of me being morally obligated to write the article on how to cure AIDS. But Ought Implies Can derives part of its appeal precisely from ruling out this possibility, as we have seen in Chapter 3, Section 1, and substantiated in Chapter 4, Section 2. So, it is not enough for a moral theory that is not supplemented with a decision procedure to be decisionguiding only for cognitively ideal agents. This leaves us with (12), the suggestion that a moral theory should be action-guiding in the sense that the theory, if not combined with a decision procedure, would nonetheless be decision-guiding for agents whose beliefs are in accord with the evidence. Not only should we accept this suggestion because we have eliminated the alternatives, but also because it seems plausible in its own right that people who are rational in the sense that their beliefs are in accord with the evidence available to them can correctly deduce their decisions from moral theories.

84

“Ought” Implies “Can”

In sum, objective consequentialism is false because it is not properly action-guiding. One can see this by focusing on objective consequential­ ism’s judgement that I ought to write “O”, “n”, and so on, which is no more properly action-guiding than the judgement that I ought to write an article that explains how to cure AIDS. Alternatively, one can see that objective consequentialism is not properly action-guiding by analysing the concept of action-guidance.

3. Blameworthiness and Fairness In this section, I put forward the Argument from Blameworthiness and Fairness: (15) If considerations of blameworthiness and fairness justify Ought Implies Can, then I cannot have an obligation to write an article that explains how to cure AIDS, because it would be unfair if I had this obligation. (16) My having an obligation to write “O”, “n”, and so on, would be equally unfair as my having an obligation to write an article that explains how to cure AIDS. (17) According to objective consequentialism, I have an obligation to write “O”, “n”, and so on. (18) Therefore, if considerations of blameworthiness and fairness justify Ought Implies Can, then objective consequentialism is false. I will start by explaining why one might think that considerations of blameworthiness and fairness justify Ought Implies Can. Then I will argue for (16). Ought Implies Can is often supposed to be derivable from consid­ erations of blameworthiness and fairness (cf. Kekes 1984; Stern 2004, pp. 46–48). This rationale needs to be scrutinized although I have some reservations about it and although I have already argued against certain aspects of this idea in Chapter 3, Section 2. So, this section should be read in a conditional way: “If you believe, despite the doubts that I have expressed, that Ought Implies Can can be derived from considerations of blameworthiness and fairness, then you should reject objective conse­ quentialism, because it is at odds with those considerations.” Correspond­ ingly, I have given the Argument from Blameworthiness and Fairness a conditional form: (18) does not claim that objective consequentialism is false but that objective consequentialism is false if considerations of blameworthiness and fairness justify Ought Implies Can. I will first present an idea that does not work and then show a more promising rationale, which will then be shown to be at odds with objec­ tive consequentialism. The first idea is to derive Ought Implies Can from the following principle:

Four Arguments 85

Unfair Blaming It would be unfair to blame people for the performance of acts that they could not have omitted.5 Unfair Blaming by itself does not provide a reason for adopting Ought Implies Can. For Unfair Blaming is compatible with the position that one can be obligated to perform acts that one cannot perform while one can­ not be blameworthy for not performing them. Ought Implies Can cannot be established on grounds of blameworthi­ ness, at least not solely on these grounds. Considerations of fairness yield a more promising rationale.6 According to some people, Ought Implies Can can be justified on grounds of the following principle: Unfair Obligation It would be unfair if a person ought to perform an act that she can­ not perform. Unfair Obligation, if true, would provide a rationale for Ought Implies Can. I have explained in Chapter 3, Section 2, that it is a category error to apply criteria of fairness to the conditions under which you have moral duties. However, some people disagree with my opinion. I want to show that, if you disagree with me and think that it is also a question of fairness which moral duties you have, then you should reject objective consequen­ tialism because it yields unfair duties. I think that people who believe in Unfair Obligation are motivated, at least in part, by thoughts about the fairness of blaming. Something like the following argument might be presented for Unfair Obligation by its fans: (19) A person is a candidate for blame if and only if the person acts wrongly. (20) It would be unfair if one could not avoid being a candidate for blame. (21) Therefore, it would be unfair if one could not avoid acting wrongly. (22) If a person ought to perform an act that she cannot perform, then the person cannot avoid acting wrongly. (23) Therefore, (Unfair Obligation:) it would be unfair if a person ought to perform an act that she cannot perform. As for (19) and (20): By a person’s being a candidate for blame with respect to some act, I mean that the person is blameworthy for the act if she does not have an excuse. (19) and (22) are intuitively plausible prin­ ciples. (20) has some appeal, in particular, if one assumes that the onus of proof for being excused lies with the candidate for blame.

86

“Ought” Implies “Can”

Let us come to (16). We assume that it would be unfair if I had the obligation to write an article that explains how to cure AIDS. (Recall that Ought Implies Can rules out my having this obligation and we are assum­ ing, for the sake of argument, that Ought Implies Can can be derived from Unfair Obligation. It follows that it would be unfair if I had the obliga­ tion to write the article on how to cure AIDS.) It seems that it would be equally unfair if I had the obligation to write “O”, “n”, and so on. After all, I would be just as clueless about the fact that “O”, “n”, and so on, is the sequence of letters I have to write as I would be about the fact that “O”, “n”, and so on, is the sequence of letters that constitutes the article on how to cure AIDS. In either case, it seems that it would not be up to me, in a morally significant sense, whether I comply with my duty. Corresponding thoughts apply to the connection between fairness and blameworthiness. Assume that it would be unfair if I could not avoid being a candidate for blame because I had the obligation to write the article that explains how to cure AIDS. Certainly, it would be equally unfair if I could not avoid being a candidate for blame because I had the obligation to write “O”, “n”, and so on. This establishes the Argument from Blameworthiness and Fairness.

4. The Argument From Reasons In this section, I attempt to establish the Argument from Reasons. According to this argument, objective consequentialism is false because it is connected to a false picture of practical reasons. The Argument from Reasons has two strands. The first strand is the Argument from Endless Tryings: Objective consequentialism is committed to the claim that you have reasons for actions that you will never perform, no matter how hard and long you try to do what you have reason to do. The second strand is the Argument from Pointless Deliberation: Objective consequentialism is committed to the claim that practical deliberation will lead you to the conclusion that you have most reason to perform actions that you will never perform, no matter how hard and long you try to do what you have reason to do. Bart Streumer has presented three innovative arguments for Ought Implies Can. I will use two of these arguments to show that objective consequentialism is false.7 But first, let me explain the general structure of Streumer’s arguments. Streumer has put forward arguments in favour of: Reason Implies Can There cannot be a reason for a person to φ if the person cannot φ. Reason Implies Can entails Streumer’s interpretation of:

Four Arguments

87

Most Reason Implies Can It cannot be the case that there is most reason for a person to φ if the person cannot φ.8 Since Streumer takes Most Reason Implies Can to have the same truthconditions as Ought Implies Can as he understands it, Reason Implies Can also entails Ought Implies Can as Streumer understands it.9 Thus, insofar as Streumer’s arguments for Reason Implies Can are convincing, we have a new rationale for Ought Implies Can. It follows from what I have said in the Introduction (Section 3) that I am sympathetic to Streumer’s assumption that Most Reason Implies Can has the same truth-conditions as Ought Implies Can. Consequentialism should be understood as the doctrine that you have most reason to do what has the best consequences, not only that it is your moral duty to do what has the best consequences. The first argument in favour of Reason Implies Can that Streumer puts forward is the Argument from Crazy Reasons. Streumer lists several claims that he takes to be obviously false. The falseness of these claims, Streumer (2007, pp. 358–362) shows, is best explained by Reason Implies Can. Thus we have good reason to think Reason Implies Can is true. Among the claims Streumer lists is the following: Susan’s Case Suppose that Susan is a moderately intelligent person. Given how important it is to save the lives of people who have deadly diseases, there is a reason for Susan to develop new medicines to save the lives of everyone who has a deadly disease, even though, if she tried to do this, she would be trying to develop these medicines until she dies, without ever succeeding. (Streumer 2007, p. 359) We have, of course, already encountered the idea underlying Susan’s Case: a medical layperson cannot write an article that explains how to cure AIDS. So, we can immediately see that the following claims are as absurd as Susan’s Case: Susan’s Case* Suppose that Susan is a moderately intelligent person. Given how important it is to save the lives of people who have deadly diseases, there is a reason for Susan to write “O”, “n”, and so on, since by doing so she would save the lives of everyone who has AIDS, even though, if she tried to find out which sequence of letters she has reason to write, she would be trying until she dies, without ever succeeding.

88

“Ought” Implies “Can”

I conclude that the considerations underlying the Argument from Crazy Reasons speaks as much for my not having an obligation to write “O”, “n”, and so on, as they speak for my not having an obligation to write an article about how to cure AIDS. In sum, we get: The Argument From Endless Tryings: (24) If you had the reasons objective consequentialism says you have, you have reasons for actions that you will never perform, no matter how hard and long you try. (25) You do not have reasons for actions that you will never perform, no matter how hard and long you try to do what you have reason to do. (26) Therefore, objective consequentialism is false. Another argument Streumer puts forward for Reason Implies Can is this: The Argument From Deliberation Suppose that Reason Implies Can is false and there are no other restrictions on the reasons there can be for a person. Then, when­ ever a person engages in rational deliberation about what to do, this person will not only have to judge which reasons there are to perform actions that she can perform, but will also have to judge which reasons there are to perform actions that she cannot perform. The person will then almost always have to conclude that she has most reason to perform an action she cannot perform, like travelling back in time and prevent the Second World War. But trying to do so is pointless. Thus, the results of practical deliberation would be pointless.10 Streumer then considers the objection that, although there are reasons for acts that one cannot perform, one does not have to take these reasons into account when one is engaged in practical deliberation. But, as Streumer (2007, p. 366) rhetorically asks, “if the only reasons that can ever make a difference to the result of a rational deliberation are reasons to perform actions that it is possible to perform, why should we believe that there are also reasons to perform actions that it is impossible to perform?” Streumer also refutes the objection that there can be reasons for a person to try to perform an act that it is impossible for this person to perform. For example, Jack can have a reason to try to run a marathon in less than 2 hours. These reasons for attempts which are doomed to failure exist, the objection says, because there can be reasons for a person to perform acts she cannot perform, like running a marathon in less than 2 hours. Streumer admits that there can be reasons for a person to try to perform an act she cannot perform. But he replies:

Four Arguments 89 [W]hat makes it the case that there is such a reason for Jack does not have to be that there is a reason for him to actually run a marathon in less than 2 hours. It could instead be that there is a reason for him to run a marathon as fast as he can, and that attempting to run a mara­ thon in less than 2 hours is the best way for him to run a marathon as fast as he can. Or it could be that he does not know whether he is able to run a marathon in less than 2 hours, and that attempting to run a marathon in less than 2 hours is the best way for him to find out whether he is able to do this. (Streumer 2007, p. 367) It should be clear then that the objection fails. Here is my parallel argument for Reason Implies Evidence: The Argument From Deliberation* Suppose that, whenever a person engages in rational deliberation about what to do, this person will not only have to take into account the reasons that speak in favour of acts that she can identify, but she will also have to take into account the reasons there are to perform acts that she cannot identify. For instance, I will judge that I have most reason to write “O”, “n”, and so on. But I cannot identify “O”, “n”, and so on as the action I have most reason to perform. Trying to do so will be unsuccessful. Thus, the results of my practical delibera­ tion would be pointless. One can raise an objection similar to the second objection against Streumer’s Argument from Deliberation. The objection says that one can have reasons to try to perform an act that one cannot identify as the act one has most reason to perform. For example, I can have a reason to try to write “O”, “n”, and so on. But Streumer’s reply can be transferred: It is true that I can have a reason to try to write “O”, “n”, and so on, but this does not show that I can have a reason to write “O”, “n”, and so on. My reason for trying can be explained, for instance, by the fact that I have a reason to find out whether I can write the article that explains how to cure AIDS. It seems to me that the Argument from Deliberation* is as strong as Streumer’s Argument from Deliberation. We can sum up this thought as follows: The Argument From Pointless Deliberation (27) If you had the reasons objective consequentialism says you have, practical deliberation would be a farce. (28) Practical deliberation is not a farce. (29) Therefore, objective consequentialism is false.

90

“Ought” Implies “Can”

5. Conclusion By investigating the considerations underlying Ought Implies Can, we have found four new arguments against objective consequentialism: the Argument from Impossibility Comparisons, the Argument from ActionGuidance, the Argument from Blameworthiness and Fairness, and the Argument from Reasons. In Karpov-like cases, the considerations under­ lying Ought Implies Can rule out obligations like Frances’s obligation to play e2–e4, etc., or my obligation to write “O”, “n”, and so on, because there is no factual evidence to the effect that playing e2–e4, etc., or writ­ ing “O”, “n”, and so on, respectively, would have the best consequences. If there is no such evidence, there cannot be the obligations in questions. Moreover, I have examined the sense in which moral theories should be action-guiding, in the context of Ought Implies Can, and suggested a connection to the desideratum for decision-guidingness, which occupied us in Chapters 1 and 2. My suggestion is that in order to be properly action-guiding, a moral theory must be decision-guiding without an addi­ tional decision procedure for those of its followers whose beliefs are in accord with the factual evidence available to them.

Notes 1. This chapter is based on Andrić 2017. The most important revisions concern the elaborations on action-guidingness in Section 2. 2. For many references, see Smith 2018, esp. Chapter 3. 3. If we know that the theory asks us to do something that we cannot do, then we cannot even use DP, the decision procedure defended in Chapter 2, to implement the theory. For we will know that all actions that are available to us are wrong, and we will therefore not be able to carry out the second step of DP. 4. Carlson (2002) argues that moral theories need not be action-guiding for every person but at least for cognitively and conatively ideal agents. (See p. 74 for the definition of being cognitively ideal.) 5. This also holds for the omission of acts that one could not have performed. I will for simplicity’s sake not mention this variant in the following. 6. The talk about fairness can, arguably, be replaced by talk about desert or about justice. 7. I cannot think of an argument based on Streumer’s argument from tables and chairs (Streumer 2007, p. 362) that shows that objective consequentialism is false. 8. I have slightly rephrased Reason Implies Can and Most Reason Implies Can to fit my formulation of Ought Implies Can. 9. Streumer 2007. Ulrike Heuer (2010) has presented several objections to Streumer’s argumentation. But in my opinion, Streumer (2010) has convinc­ ingly answered all of them. 10. Streumer 2007, p. 365. I substituted “Reason Implies Can” for “(R)”.

6

“Ought” Implies “Evidence”

I have argued in the foregoing chapters that objective consequentialism, though compatible with Ought Implies Can, is incompatible with the rationales underlying Ought Implies Can. How can we make sense of this finding? I suggest that the rationales underlying Ought Implies Can also support another moral principle, which I call “ought” implies “evi­ dence”. Unlike Ought Implies Can, which only rules out that I have a moral duty to write an article that explains how to cure AIDS, the prin­ ciple that “ought” implies “evidence” explains why I cannot have a moral duty to write “O”, “n”, and so on. So, apart from the individual consid­ erations that speak against objective consequentialism, we get a general explanation for why objective consequentialism is false. The principle that “ought” implies “evidence” will be introduced in Sections 1 and 2 and defended against some objections in Sections 3 and 4. I sum up the results in Section 5.

1. The Standard View and Ought Implies Evidence In the previous chapter, I have argued that the considerations underlying Ought Implies Can rule out obligations such as Frances’s obligation to play e2–e4, etc., and my obligation to write “O”, “n”, and so on, because there is no factual evidence to the effect that playing e2–e4, etc., and writ­ ing “O”, “n”, and so on, would have the best consequences. If there is no such evidence, there cannot be the obligations in questions. Let me now suggest that we should accept a principle that is supported by at least some of the considerations that speak in favour of Ought Implies Can. The principle I have in mind, the principle that “ought” implies “evidence”, has way fewer proponents than Ought Implies Can. It is fair to say, I think, that Ought Implies Can, despite the criticisms we have seen in Chapter 3, is a widely accepted principle, whereas the prin­ ciple that “ought” implies “evidence” does not have many proponents. (Nonetheless, the number of critics of “ought” implies “evidence” is prob­ ably smaller than the number of critics of Ought Implies Can, because many philosophers do not even consider “ought” implies “evidence”.)

92

“Ought” Implies “Can”

Before I introduce the principle that “ought” implies “evidence” in more detail, let us focus on what we can call the standard view. Accord­ ing to the standard view, lack of evidence for the presence of morally relevant facts is not a defeater of moral obligation but an excuse – that is, a fact that makes it difficult to discharge one’s moral duty.1 Excuses are not defeaters of moral obligation. Rather, they merely defeat the blame­ worthiness of a wrong act or omission. They can do so partially or com­ pletely, depending on how difficult it is for a person to discharge her moral obligation. Thus, if a person has an excuse for not discharging her moral duty, either the person is not blameworthy at all or her blamewor­ thiness is diminished. On the standard view, a doctor acts wrongly by prescribing a lethal drug even if the doctor lacks evidence that the drug is lethal. Despite her acting wrongly, proponents of this view add, the doctor is not blame­ worthy. At least, this holds if we make certain assumptions. Suppose that the doctor tries to heal her patient and that the doctor lacks evidence that the drug is lethal. Moreover, suppose that the lack of evidence is not due to some negligence on the side of the doctor. Then the doctor is not blameworthy at all for killing her patient, according to the standard view, although the doctor acted wrongly by killing her patient. The doctor’s blameworthiness would only be diminished, rather than erased (and the doctor only partially, rather than fully, excused), if there had been some evidence for the drug’s being lethal, which the doctor had failed to take into account. In the following, I will ignore the distinction between partial and full excuses. What I have in mind when writing about the moral relevance of the lack of factual evidence are cases that would be considered cases of full excuse by proponents of the standard view. Here is the principle that I am advocating: Ought Implies Evidence Necessarily, a person is morally obligated at t1 to φ at t2 only if the evidence available to the person at t1 suggests that her φ-ing at t2 has an ought-making feature. Something like Ought Implies Evidence has been suggested by Lars Berg­ ström for reasons that are closely related to my reasons.2 Bergström, too, has used Ought Implies Evidence to argue that objective consequentialism is false: [I]t may be tempting to say that “ought” implies, not only “can”, but also “can know that one ought”. This would, however, be too strong. For one thing, we should not assume, without sufficient reason, that there is such a thing as moral knowledge. Rather, I suggest that “ought” (and “right”) implies “there is some property, which this action has,

“Ought” Implies “Evidence”

93

and which makes this action morally obligatory (or morally right), and which is such that it is possible for the agent to know or to have good reasons to believe that the action has this property”. In my opinion, this principle is just as plausible as the more well-known slogan that “ought” implies “can”. But if this principle is true, objective conse­ quentialism is false. Q.E.D.3 Let me elaborate on the contents of Ought Implies Evidence. First of all, just like in Ought Implies Can, φ ranges over acts as well as omissions in Ought Implies Evidence. Moreover, just like Ought Implies Can, Ought Implies Evidence is only concerned with conclusive moral obligations and not with moral prima facie obligations. Finally, we have seen in Chapter 3 that proponents of Ought Implies Can should distinguish between the time of the obligation and the time of the obligatory action or omission. Accordingly, Ought Implies Evidence distinguishes between the time of the obligation (t1) and the time of the obligatory action or omission (t2). It is assumed that the former is not later than the latter (t1 ≤ t2). Secondly, by “evidence” I mean, roughly, the information available to the agent that determines what the agent ought to believe. It does not matter whether the agent actually recognizes the information. It is argu­ ably enough for there being evidence for p that the agent has reason to believe p.4 Think of a doctor who misses to read vital instructions (provided at the canonical place) and prescribes a lethal drug. The vital instructions provide evidence for the drug’s being lethal although the doctor does not recognize it. Thirdly, I have to explain what I mean by an ought-making feature. I will begin by recapping a general idea, which I have already mentioned in previous chapters,5 and then deal with a complication. The general idea is that an ought-making feature is a property of an act (or an omission) that makes the act (or the omission) morally obligatory. I am talking here about conclusive rather than prima facie obligations. So, it is in virtue of an act’s (or an omission’s) having an ought-making feature that an agent has a conclusive moral duty to perform (or refrain from) the act. For illustration, consider classical utilitarianism according to which hap­ piness is the only intrinsic good. Suppose that of all acts available to me now, snapping my finger is the act that will lead to the largest amount of happiness. In that case, according to the classical utilitarian theory under consideration, snapping my fingers has an ought-making feature, namely, its leading to more happiness than any of its alternatives. Here is the complication. My claim is that lack of factual evidence sometimes is a defeater of moral obligation. I do not want to commit myself to the view that lack of normative evidence is a defeater of moral obligation (although I will argue later that this view might turn out to be true as well). The problem is that according to some moral theories, ought-making features are themselves normative. On the assumption that

94

“Ought” Implies “Can”

ought-making features are normative, Ought Implies Evidence would commit me to the view that lack of normative evidence is a defeater of moral obligation. Let me illustrate the problem. According to the utilitarian theory we considered a moment ago, the only ought-making feature an act can have is the act’s leading to more happiness than its alternatives. An act’s leading to more happiness than its alternatives is a factual ought-making feature. You do not need any moral knowledge in order to determine whether an act has this feature. All you need is empirical knowledge and a (non-moral) concept of happiness. Now consider, by contrast, a moral theory according to which you ought to treat people the way they deserve to be treated. The notion of desert is a normative notion.6 So, in order to determine whether an act has this ought-making feature, you need moral knowledge. Ought Implies Evidence hence would commit me to the claim that lack of normative evidence is a defeater of moral obligation. Here is my solution. In this chapter, I want the phrase “ought-making feature” to refer to the property of an act (or an omission) that makes the act (or the omission) morally obligatory in the case of moral theories according to which the properties that make an act (or an omission) mor­ ally obligatory are factual in nature. So, all I have said about the utilitar­ ian theory still holds. In the case of moral theories according to which the properties that make an act (or an omission) morally obligatory are nor­ mative in nature, I want the phrase “ought-making feature” to refer to the non-normative facts on which these properties supervene. As a welcome consequence, Ought Implies Evidence applies to these theories as well. To illustrate, let us assume that murderers deserve to be killed and that George is a murderer. The theory according to which you ought to treat people the way they deserve to be treated entails that you ought to kill George. Now suppose you lack normative evidence that murderers deserve to be killed. But the factual evidence available to you suggests that George is a murderer. Ought Implies Evidence, in this constellation, does not entail that you cannot be morally obligated to kill George. Your lack of normative evidence is irrelevant. This is just as it should be. Let us now turn to arguments in favour of Ought Implies Evidence. After that, we will consider objections to this principle.

2. Arguments for Ought Implies Evidence The rationales for Ought Implies Can almost equally strongly support Ought Implies Evidence, at least as far as they speak against objective consequentialism. So, let us reconsider the considerations from intuitive appeal, action-guidance, blameworthiness and fairness, and practical rea­ sons laid out in Chapter 5. First of all, Ought Implies Evidence has some initial intuitive appeal on its own. This is why W.D. Ross (1939, p. 156) simply accepted as an

“Ought” Implies “Evidence”

95

axiom in his Foundations of Ethics “that the act which a man in any situ­ ation ought to do is that which would be reasonable for him to do if he wanted to do his duty in that situation”. More important, though, is the following consideration. Recall that Ought Implies Can explains this pair of judgements: (1) I am not morally obligated to write an article that explains how to cure AIDS. (2) If I could write an article that explains how to cure AIDS, then I would be morally obligated to do this. However, Ought Implies Can does not explain this equally plausible pair of judgements: (3) I am not morally obligated to write “O”, “n”, and so on. (4) If I knew that by writing “O”, “n”, and so on, I would be writing an article that explains how to cure AIDS, then I would be morally obligated to write “O”, “n”, and so on. Ought Implies Evidence explains the second pair of judgements. This is a reason to accept this principle. In Chapter 5, Section 2, I have argued for the following account of action-guidance: (5) In order to be properly action-guiding, a moral theory must be decision-guiding without an additional decision procedure for those of its followers whose beliefs are in accord with the factual evidence available to them. Ought Implies Can can be understood as ruling out my obligation to write an article that explains how to cure AIDS on grounds of action-guidance. Correspondingly, Ought Implies Evidence can be understood as ruling out my obligation to write “O”, “n”, and so on for the same reasons. Ought Implies Evidence is a corollary of (5).7 So, considerations of actionguidance support Ought Implies Evidence. In Chapter 5, Section 3, I have explained that some people (though I do not count myself among them) believe that Ought Implies Can can be derived from considerations of blameworthiness and fairness. If so, we should accept that it would be unfair for me to have an obligation to write an article that explains how to cure AIDS. However, it would be equally unfair if I had an obligation to write “O”, “n”, and so on. But, again, Ought Implies Can does not rule out the latter obligation while Ought Implies Evidence does. This speaks in favour of Ought Implies Evidence. We can generalize the thought. As we have seen, some people believe in this principle, which makes them accept Ought Implies Can:

96 “Ought” Implies “Can” Unfair Obligation It would be unfair if a person ought to perform an act that she can­ not perform. However, the following principle, which suggests Ought Implies Evi­ dence, is not less plausible: Unfair Obligation* It would be unfair if a person ought to perform an act although the evidence available to the person at the time that she can perform the act does not suggest that the act has an ought-making feature. We have seen in the foregoing chapter that the following argument might also explain why some people believe Unfair Obligation: (6) A person is a candidate for blame if and only if the person acts wrongly. (7) It would be unfair if one could not avoid being a candidate for blame. (8) Therefore, it would be unfair if one could not avoid acting wrongly. (9) If a person ought to perform an act that she cannot perform, then the person cannot avoid acting wrongly. (10) Therefore, (Unfair Obligation:) it would be unfair if a person ought to perform an act that she cannot perform. However, here is an argument for Unfair Obligation* that is no less plausible: (6) A person is a candidate for blame if and only if the person acts wrongly. (11) It would be unfair if a person could avoid being a candidate for blame only by performing an act such that the evidence available to the person at the time that she can perform the act does not suggest that the act has an ought-making feature. (12) Therefore, it would be unfair if a person could avoid acting wrongly only by performing an act such that the evidence available to the person at the time that she can perform the act does not suggest that the act has an ought-making feature. (13) If a person ought to perform an act although the evidence avail­ able to the person at the time that she can perform the act does not suggest that the act has an ought-making feature, then the person could avoid acting wrongly only by performing an act such that the evidence available to the person at the time that she can

“Ought” Implies “Evidence”

97

perform the act does not suggest that the act has an ought-making feature. (14) Therefore, (Unfair Obligation*:) it would be unfair if a person ought to perform an act although the evidence available to the person at the time that she can perform the act does not suggest that the act has an ought-making feature. This argument for Unfair Obligation* is obviously as intuitively plausible as the earlier argument for Unfair Obligation. (6) figures in both argu­ ments. (13) is trivially true. (11) is not less plausible than (7). Assume that it would be unfair if I could not avoid being a candidate for blame other than by writing an article that explains how to cure AIDS. It would cer­ tainly be equally unfair if I could not avoid being a candidate for blame other than by writing “O”, “n”, and so on. Finally, we have considered some of Bart Streumer’s arguments from reasons for Ought Implies Can in Chapter 5, Section 4. Streumer derives Ought Implies Can from: Reason Implies Can There cannot be a reason for a person to φ if the person cannot φ. Reason Implies Can entails: Most Reason Implies Can It cannot be the case that there is most reason for a person to φ if the person cannot φ. Streumer takes Most Reason Implies Can to have the same truth-conditions as Ought Implies Can. In order to establish a corresponding rationale for Ought Implies Evidence, I have to show that we should accept this principle: Reason Implies Evidence There cannot be a reason at t1 for a person to φ at t2 if the evidence available to the person at t1 does not suggest that she has a reason to φ at t2. Reason Implies Evidence entails: Most Reason Implies Evidence It cannot be the case that there is most reason at t1 for a person to φ at t2 if the evidence available to the person at t1 does not suggest that she has most reason to φ at t2.

98

“Ought” Implies “Can”

If Most Reason Implies Can has the same truth-conditions as Ought Implies Can, then Most Reason Implies Evidence has the same truthconditions as Ought Implies Evidence. We have seen two considerations in Chapter 5, Section 4 that speak in favour of Reason Implies Can. The first consideration, illustrated by Susan’s Case, is that, if Reason Implies Can were false, there could be reasons for actions that an agent will never perform, no matter how long and hard she tries, actions such as my writing an article that explains how to cure AIDS. As Susan’s Case* illustrated, if Reason Implies Evidence were false, there could also be reasons for (other) actions that an agent will never perform, no matter how long and hard she tries, actions such as my writ­ ing “O”, “n”, and so on. Reason Implies Can alone would not rule out such reasons. This speaks in favour of Reason Implies Evidence. Second, we have considered the Argument from Deliberation for Rea­ son Implies Can. If Reason Implies Can were false, practical delibera­ tion could lead an agent to the conclusion that she has most reason to perform an action that she will never perform, no matter how hard and long she tries, actions such as my writing an article that explains how to cure AIDS. Since practical deliberation, properly done, cannot lead you to such a conclusion, we should accept Reason Implies Can. We have also considered the Argument from Deliberaton*, which shows that I cannot have a reason to write “O”, “n”, and so on, because other­ wise practical deliberation could, again, lead me to the conclusion that I have most reason to perform an action that I will never perform, no mat­ ter how hard and long I try. We can generalize the point: If Reason Implies Evidence were false, practical deliberation could lead me to the conclusion that I have most reason to perform an action that I will not perform, no matter how hard and long I try. Reason Implies Can alone would not rule out all such reasons. So, we should accept Reason Implies Evidence. In sum, all the considerations underlying Ought Implies Can that are at odds with objective consequentialism suggest Ought Implies Evidence. Ought Implies Can does not rule out obligations that seem worth being ruled out. Without Ought Implies Evidence, something seems to be amiss in our normative landscape. So, we should accept this principle. This gives us a further argument against objective consequentialism: objective consequentialism is false because it is inconsistent with Ought Implies Evidence.

3. The Objection From Lack of Normative Evidence In this and the next section, I will consider objections to Ought Implies Evidence. A critic of Ought Implies Evidence might argue as follows: You have argued that lack of factual evidence is a defeater of moral obligation. But if we accept this view, then we also have to accept the

“Ought” Implies “Evidence”

99

view that lack of normative evidence is a defeater of moral obliga­ tion. The latter view, however, has highly counterintuitive implica­ tions. Hence, we should not accept that the lack of factual evidence is a defeater of moral obligation. Does the acceptance of Ought Implies Evidence commit me to the view that lack of normative evidence is a defeater of moral obligation? Does it commit me, that is, to the following principle? Ought Implies Evidence* Necessarily, a person is morally obligated at t1 to φ at t2 only if the evidence available to the person at t1 suggests that she ought to φ at t2. According to Ought Implies Evidence*, for a person to have a moral obligation to perform an act, the evidence must not only suggest that the act has an ought-making feature (that is: that some descriptive facts about the act are true), but also that the act is obligatory (that is: that the ought-making features are ought-making). Why should it be problematic to accept Ought Implies Evidence*? Here is a case in point: The Case of Calvin Calvin has been raised in the antebellum era of the Deep South. Rac­ ism was pervasive, and everyone he knows believes that slavery is legitimate. More than that, one does not even question slavery. To Calvin it seems to be a self-evident truth that slavery is legitimate. If asked, he would insist that he has no more reason to question the legitimacy of slavery than he has for questioning that the sky is blue or that the world has existed for more than five minutes. Calvin owns a plantation with many slaves. He treats his slaves in a very brutal way and lets them fight to death in “Mandingo fights”.8 Given that Calvin has been raised from an early age in a thoroughly racist society, it seems that Calvin has no evidence that he ought to refrain from holding slaves. The problem, then, is that it seems to fol­ low from Ought Implies Evidence* that Calvin is not morally obli­ gated to refrain from holding slaves. But it is obvious that slavery is wrong and that Calvin ought not to have slaves. We can sum up the critique this way: The Objection From Lack of Normative Evidence (15) I f Ought Implies Evidence is true, then Ought Implies Evidence* is also true.

100

“Ought” Implies “Can”

(16) According to Ought Implies Evidence*, necessarily, a person is mor­ ally obligated to φ only if the evidence available to the person sug­ gests that she ought to φ. (17) Calvin is morally obligated not to keep slaves. (18) The evidence available to Calvin does not suggest that he ought to refrain from keeping slaves. (19) Therefore, Ought Implies Evidence* is false. (From (16), (17), and (18).) (20) Therefore, Ought Implies Evidence is false. (From (15) and (19).) (16) is true by definition. Denying (17) is not an option. This leaves us with two possible ways out for the proponent of Ought Implies Evidence. She might deny (15) or (18). It may well be that good arguments against (15) can be found. For instance, one might question whether sense can be made of Ought Implies Evidence* at all if metaethical cognitivism is false. However, my aim here is to show that (18) may well be false. I believe that suffering is bad. And all you need to do in order to know this is to look at instances of suffering or imagine suffering in a vivid way. If you vividly represent suffering, you see that it is bad. Now, there is plenty of suffering caused by slavery. All Calvin needs to do in order to see that suffering is bad is to open his eyes. Hence, there is plenty of evi­ dence available to Calvin that suffering is bad. This is, at the same time, evidence that keeping slaves is bad, since it causes suffering. Does Calvin also have evidence that he ought to refrain from keeping slaves? It is obvious, I think, that if something is bad then you have rea­ son to avoid it. Again, all Calvin needs to do in order to see that he has reason not to keep slaves is to look at the suffering that is caused by slav­ ery. It is all there! Suffering is obviously bad and hence obviously gives you reason to avoid it. Instances of suffering carry with them all moral information you need. Therefore, there is plenty of evidence that sug­ gests to Calvin that he has reason to refrain from keeping slaves. I take it that an agent has a moral obligation to φ if and only if the agent has most moral reason to φ (cf. Introduction, Section 3). Accordingly, I think that the evidence available to Calvin at the time that he can refrain from keeping slaves suggests that he ought to refrain from keeping slaves. This suggests that (18) is false. And I think that the point generalizes to all agents in all situations. Therefore, Ought Implies Evidence* is acceptable.

4. The Worry From Moral Uncertainty One might worry that solutions similar to my solution for the case of Calvin are not viable when it comes to cases where we ourselves are uncertain about what is good or bad, or right or wrong. A critic might say:

“Ought” Implies “Evidence”

101

Cases of genuine moral uncertainty cannot be solved by appeal to obvious moral truths since in such cases, ex hypothesi, there are no such truths. Your opinion about the case of Calvin notwithstanding, you might be uncertain about, say, whether well-being is the only value, whether it matters how the well-being is distributed, and many other questions. Hence it might well be that given your restricted evidence, you cannot figure out what you have a moral obligation to do. This contradicts Ought Implies Evidence*. Let us call this: The Worry From Moral Uncertainty (21) If Ought Implies Evidence* is true, then moral uncertainty is not possible. (22) Moral uncertainty is possible. (23) Therefore, Ought Implies Evidence* is false. I agree that I often do not know what my moral obligation is and that I have many open questions concerning the good. Moreover, normative ethics would be pointless if there were no moral uncertainty. So, there is moral uncertainty and (22) is true. I think, however, that (21) may well be false. My suspicion is that, although there is always sufficient evidence as to what I have a moral obligation to do, I sometimes fail to recognize what I morally ought to do, despite all the care taken. I think that, in principle, my relation to the morally wrong things I do is the same as the relation between Calvin and the morally wrong things he does. Calvin, we can assume, tries his best to find out what he ought to do. He may be a very conscientious person. Nonetheless, it seems, all the evidence that speaks against keeping slaves is available to him but he does not conclude that he ought not to keep slaves. Notice that my depiction of Calvin is not farfetched. I know many persons with absurd moral views who sincerely try to be decent persons and who might well give their best when they try to figure out whether their views are correct. My guess is that you know some people of this kind, too. And maybe some of my and your own views are in fact absurd, unbeknownst to us. In sum, I see no reason to think that there cannot be moral uncertainty if Ought Implies Evidence* is true. But this, obviously, raises another prob­ lem. One wonders what the explanation of my unsuccessful attempts to find out what I ought to do might be, given that all the relevant evidence is always available to me. After all, in non-moral affairs, agents tend to get things right if they give their best and there is sufficient evidence for concluding what is the truth.

102

“Ought” Implies “Can”

It is important to distinguish between moral and factual uncertainty. Some instances of what appears to be moral uncertainty might well in fact be instances of factual uncertainty. For example, if we do not know the consequences of some new drug or medical procedure, we are uncer­ tain as to how to assess the drug and procedure morally. But this uncer­ tainty roots in factual rather than moral uncertainty. It is only the latter, however, that is relevant here. The best explanation of genuine moral uncertainty, I tend to think, is psychological. First of all, there are many well-known biases when it comes to moral issues. To name just one major bias: We tend to be ego­ istic and see things in a way that allows us to favour our own interests over the interests of other persons. And think of all the cultural biases, which best explain why Calvin does not see the obvious, and, of course, of biases induced by religious education or supported by economic and political arrangements. These and other factors provide good explana­ tions of our confusion and controversy.9 Secondly, there may be factors that we do not yet know of. Think of things such as malnutrition that damages our brains. Given the many controversies among nutritionists, it might well turn out, eventually, that we all suffer from malnourished brains. There might be other completely unknown factors (radiation, genetic mutations) that influence our brains and our thinking. What I have said so far tacitly assumes that morality roots in reason. But if morality instead roots in sentiments, then this further strengthens my case.10 For it seems obvious that sentiments can easily be influenced and perverted by all sorts of factors. In sum, the worry from moral uncertainty seems to be on shaky ground. Cases such as the case of Calvin suggest that we have all the evidence we need to conclude what our moral duties are and that, nonetheless, there is moral uncertainty. In part, we can already explain this mismatch psycho­ logically. And it does not seem unreasonable to expect better and more complete explanations from future research.

5. Conclusion In this chapter, we have taken a second look at the considerations under­ lying Ought Implies Can. I have argued that these considerations also support Ought Implies Evidence and defended this principle against two objections. Since objective consequentialism is incompatible with Ought Implies Evidence, we have another argument for rejecting objective consequentialism.

Notes 1. For this understanding of the term “excuse” see, among others, Austin 1956b, esp. p.  20, McConnell 1989, pp.  437–438, Rosen 2003, passim, 2004,

“Ought” Implies “Evidence”

2.

3.

4. 5. 6. 7.

8. 9. 10.

103

pp. 297–298. For a criticism, see Wallace 1994, Chapter 5, and for a reply, see Zimmerman 2004. Bergström 1996, Section 3. Bergström seems to suggest, though, that Ought Implies Evidence is implied by Ought Implies Can and that, accord­ ingly, objective consequentialism is incompatible with Ought Implies Can: “‘Ought’ implies ‘can’, and ‘can’ implies ‘knows how’. Therefore, objective consequentialism implies that you know how to maximize value. But you don’t know how to maximize value. Therefore, objective consequentialism is false.” (p. 82). I argue against these claims in Chapter 4. Bergström 1996, p. 82, footnote omitted. A more recent article that explicitly deals with Ought Implies Evidence (though in a slightly different formula­ tion) is Graham 2011 (pp.  365–367). Graham, unlike me, argues against both Ought Implies Can (as we have seen in Chapter 3) and Ought Implies Evidence. A principle related to Ought Implies Evidence is defended in Kiese­ wetter 2016. I say “arguably” because one might hold that the evidence available to me regarding an issue does not give me reasons to form beliefs about this issue if I lack reasons to investigate the issue in the first place. Introduction, Section 1; Chapter 1, Section 2; Chapter 2, Section 2. Recall that I use “normative” as an umbrella term for both deontic and eval­ uative (see Introduction, Section 1). Suppose a moral theory says that I ought to perform some act even though the evidence available to me does not suggest that the act has an oughtmaking feature. This means that, even if my beliefs are in accord with the fac­ tual evidence available to me, I might ask Q: “Given that I do not know the ought-making, right-making, and wrong-making features of the acts avail­ able to me, which act ought I to choose?” This means that the theory is not properly action-guiding according to (5). The figure of Calvin is borrowed from Quentin Tarantino’s Django Unchained. For an interesting list of morally relevant biases, and its application to the case of the so-called repugnant conclusion, see Huemer 2008. For an overview of the debate between moral rationalism and moral sen­ timentalism with a historical focus, see Gill 2007. For a recent defence of moral sentimentalism, see Slote 2010.

7

Is Subjective Consequentialism Compatible With “Ought” Implies “Can”?

In this chapter, I argue that subjective consequentialism is compatible with Ought Implies Can. In Section 1, I present an argument to the effect that subjectivist accounts of moral duty are incompatible with Ought Implies Can. A well-known reply to the argument involves a commitment to what I will call the Volition Thesis. I will argue that subjective conse­ quentialists should accept the Volition Thesis. In Section 2, I propose a solution to the objection that the Volition Thesis does not get the subjec­ tivist account of moral duty off the hook as far as Ought Implies Can is concerned. In Section 3, I elaborate on the relevance of my solution for our understanding of subjective consequentialism. In Section 4, I deal with other objections to the Volition Thesis. In Section 5, I summarize the results.

1. The Challenge There is a considerable tradition of arguing that a subjectivist account of moral duty, according to which what an agent ought to do depends on her evidence rather than on the objective features of the situation, is incompatible with Ought Implies Can. Here is how Peter A. Graham put the charge: Take Bloggs. All of his evidence indicates both that if he presses one button, he will save ten people from being killed and that if he presses another button, he will save only five of the ten from being killed. Unbeknownst to him, he has been given a paralyzing drug that prevents him from being able to press the first button. He in fact presses the second button, motivated by his enmity for one of the ten. If there is a subjective sense of “ought”, then surely Bloggs did not act as he subjectively ought to have acted: what he subjectively ought to have done was press the first button – if things were as his evidence indicates that they are, morality would most surely dictate that he press the first button. This is true despite the fact that he can’t

Subjective Consequentialism

105

press the first button. So, [Ought Implies Can] interpreted with a subjective “ought” is false. (Graham 2011, p. 340) Classical proponents of the subjectivist account of moral duty, such as H.A. Prichard and W.D. Ross, have suggested a radical solution. Accord­ ing to them, moral duty does not range over (full-blown) actions, but over something like attempts or choices (Prichard 1932; Ross 1939, Chapter 7). On their view, Bloggs ought to have tried or ought to have chosen to press the first button. And this, in contrast to actually pressing the first button, was something Bloggs obviously could have done. Prob­ lem settled. Or so it might seem. Notice that Prichard’s and Ross’s solution contains a positive and a negative claim. The negative claim is that moral duty does not range over actions. The positive claim is: The Volition Thesis Moral duty ranges over attempts or choices (or something like that). I suggest that subjective consequentialists should accept the Volition Thesis but reject the negative claim involved in Prichard’s and Ross’s solution. What exactly is it that moral duty ranges over according to the Volition Thesis? I do not know. There are many plausible candidates: attempts, choices, decisions, self-exertions, intentions, wills, and so on. It is beyond this book’s scope to investigate the depths of philosophy of mind and action theory in order to dig up the appropriate mental state or activity. One might also wonder: “Aren’t attempts, choices, intentions, etc. themselves actions?” I will leave this question open, but briefly come back to it later, at the end of Section 3. If you think attempts, choices, intentions, etc. are themselves actions, I suggest distinguishing them from what I call full-blown actions like driving a car, killing a person, etc. Against Prichard’s and Ross’s solution, two objections have been raised. The first objection says that a subjectivist account of moral duty is incompatible with Ought Implies Can even if we assume that their view is true. According to the second objection, their view is highly implausible regardless of whether it renders the subjective account of moral duty compatible with Ought Implies Can. In the next sections, I will address these objections from the view­ point of a subjective consequentialist who accepts the Volition Thesis but rejects Prichard’s and Ross’s negative claim.

106

“Ought” Implies “Can”

2. The Volition Thesis and Ought Implies Can Does the Volition Thesis get the subjectivist account of moral duty off the hook as far as Ought Implies Can is concerned? A negative answer is suggested by Graham: [C]ases similar to those offered in Lehrer (1968) to refute tradi­ tional conditional analyses of “can” demonstrate that it can some­ times happen that unbeknownst to people they can’t even will to do things they may believe, and even have evidence that, they can do. For instance, . . . we might suppose that unbeknownst to Bloggs, and contrary to all his evidence, either he has been given a paralyzing drug, or he suffers from a severe phobia, which prevents him from even willing to press the first button.1 How can the proponent of the Volition Thesis respond to this objection? It is tempting to argue that what Bloggs ought to decide does not depend on what the evidence available to him indicates that he is able to decide but on what he really is able to decide. With respect to this response, however, Graham voices the following worry: [P]roponents of the subjective “ought” may think .  .  . that what one subjectively ought to do is constrained by one’s actual options, irrespec­ tive of one’s evidence about what options one has, but is still in some way dictated by one’s evidence concerning these options. . . . [A]ny such advocate of the subjective “ought” is under considerable pressure, it seems to me, to explain why one’s obligations are dictated by one’s evi­ dence in this merely partial way; such an advocate must explain why one’s obligations, if they are dictated by one’s evidence at all, are dictated completely by the agent’s evidence about all matters except what he or she can and cannot do. My suspicion is that a subjective “ought” only partially so determined by one’s evidence will lack a stable motivation. (Graham 2011, p. 341) Why will it lack a stable motivation? Graham (2011, p. 341, note 11) explains: The standard motivation for evidence dependence, after all, is the thought that people’s moral obligations must always be directly accessible to them . . . but what a person can do at a time is an objec­ tive matter and thus not always accessible to her. Let me start by pointing out my agreement with Graham when it comes to the original case of Bloggs, where Bloggs has been given a paralysing drug that prevents him from being able to press the first button (but not from

Subjective Consequentialism

107

being able to decide to press the first button). I agree that it would be ad hoc for this proponent of the subjectivist account of moral duty to claim that the act of pressing button 1 is not among Bloggs’s options. Certainly, if what an agent ought to do depends entirely on her evidence, then her options also must be individuated on the basis of the evidence available to her. However, we are now discussing what a proponent of the subjectiv­ ist account of moral duty who accepts the Volition Thesis can say in response to the modified case where a paralysing drug, or a phobia, pre­ vents Bloggs from being able to decide to press the first button. Is it ad hoc for this proponent of the subjectivist account of moral duty to claim that an agent’s options depend on what the agent really can do rather than on what she has evidence she can do? The answer, I submit, is “No”. To see why, note, as a starting point, that there is a crucial difference between the ability to perform an act and the ability to decide to perform an act. Persons often do not know whether they are able to perform an act. To borrow from Austin’s famous example, which I have mentioned in Chapter 4, Section 1: One might not know whether one is able to sink a difficult putt. By contrast, people do normally know whether they are able to make a decision. The modified case of Bloggs is very unusual, very distinct from everyday circumstances. Why is this relevant? Normally when people do not know whether they are able to perform an act, such as in the golf case, there is nothing wrong with the people in question. They are rational agents. This is dif­ ferent in the modified case of Bloggs. The paralysing drug or the phobia, respectively, makes him a deficient rational agent. And this seems true in general: People who do not know whether they can decide or intend or try to do something are agents with severe rational deficits. Given the plausible assumption that moral duties apply only to rational agents, the proponent of the Volition Thesis seems justified in denying that moral duties apply to agents insofar as they have severe rational deficits. And this means that it is not ad hoc for proponents of the Volitions Thesis to claim that the options an agent has depend on what the agent really can decide rather than on what she has evidence she can decide.2 I conclude that a subjectivist account of moral duty, if combined with the Volition Thesis, is compatible with Ought Implies Can. For the pro­ ponent of this combination will, for good reasons, insist that an agent’s options depend not on what the agent has evidence she is able to try or to decide but on what she really is able to try or to decide.

3. Subjective Consequentialism’s Deontic Iudicandum Subjective consequentialists should accept the Volition Thesis because oth­ erwise their theory would be incompatible with Ought Implies Can. Tradi­ tionally, however, subjective consequentialists wanted to assess acts – not just decisions, tryings, or the like. In this section, I point out that subjective

108

“Ought” Implies “Can”

consequentialists can do both: accept the Volition Thesis and assess acts (at least, some acts). As mentioned previously, subjective consequentialists should reject the negative claim of Prichard and Ross. My analysis also sheds light on an in-house debate among consequentialists about conse­ quentialism’s deontic iudicandum (cf. Kagan 2000; Pettit and Smith 2000). But most importantly, subjective consequentialism’s ability to assess not only decisions or tryings but also full-blown actions will prove useful in defending the Volition Thesis in the next section. Consequentialists can assess iudicanda either directly or indirectly. A direct consequentialist assessment is an assessment of something on the basis of its consequences. An indirect consequentialist assessment of something is an assessment on the basis of the thing’s relationship to another thing that, in turn, is assessed directly. To illustrate, consider act-consequentialism’s and rule-consequentialism’s assessments of actions. Act-consequentialists judge an act to be right or wrong on the basis of the act’s consequences – that is, traditionally, on the basis of how good the act’s consequences are. This direct assessment of actions can be depicted as follows: The good → actions3 According to rule-consequentialism, an act is right if and only if it is allowed by the best rules and wrong if and only if it is forbidden by the best rules. Rules are evaluated by rule-consequentialists directly, namely, on the basis of how good the consequences of (almost) everybody’s abiding by (or internalization of) the rules would be.4 Acts are judged indirectly by rule-consequentialists, namely, on the basis of their relationship to rules: The good → rules → actions I suggest that subjective consequentialists should distinguish between a direct and an indirect assessment in the following way: Decisions or attempts are to be assessed directly, that is, on the basis of their (rationally expectable) consequences. Sometimes a decision or an attempt results in a full-blown action – that is, in a change of state of affairs other than the agent’s deciding or attempting; most importantly, in bodily movements of the agent. In these cases, the full-blown action resulting from the decision (or attempt) inherits the decision’s (or attempt’s) deontic status: its right­ ness, wrongness, or obligatoriness. The indirect assessment of actions according to subjective consequen­ tialism can be depicted as follows (where “decisions” is a placeholder for whatever the appropriate deontic iudicandum might be – candidates include attempts, choices, volitions, and other things): The good → decisions → actions

Subjective Consequentialism

109

Sometimes, however, decisions and attempts do not lead to full-blown actions. In those cases, there is no full-blown action that inherits the deci­ sion’s or attempt’s deontic status. Suppose, for instance, that Bloggs, in the original case, decides or attempts to press button 1. Due to the para­ lysing drug, Bloggs’s decision or attempt does not lead to his pressing button 1. So, there is no full-blown action that inherits the rightness of Bloggs’s decision or attempt. I did not explain what I mean by a decision’s leading to an action or by an action’s resulting from a decision. As a consequence, one might won­ der what the account I recommend has to say about actions that causally result from a decision but are different from the action the agent decided to perform. This brings us to the subject of deviant causal chains. Suppose an agent has to choose between two acts. She decides to per­ form an act with the largest expected value, a1, because of a1’s having the largest expected value. The decision causally leads to the agent perform­ ing an act with very small expected value, a2. Had the agent decided to perform a2, this decision would have resulted in the agent performing a1. Which deontic status do a1 and a2 have? Let us distinguish three propos­ als regarding the deontic status of a2. (The deontic status of a1 can be determined accordingly.) Proposal 1: a2 is right because it inherits the deontic status of the agent’s decision. Proposal 2: a2 is wrong because the agent’s decision to perform a2 would be wrong. Proposal 3: a2 lacks deontic status in virtue of the deviant causal chain between the agent’s decision and a2. I accept Proposal 3. The agent does not have the kind of control over a2 that is necessary for a2 to have a deontic status. The agent has some control over a2: Had she decided differently, she would have performed another act. But this is not the relevant kind of control. Notice that Prichard and Ross subscribed to the negative claim, which I take to state that all full-blown actions lack deontic status, because they thought that agents lack control over full-blown actions.5 But things are not that simple. Rather, we should distinguish between direct and indirect control. An agent has direct control over an event if and only if the agent can decide whether the event takes place. An agent has indirect control over an event if and only if the agent can decide something else and whether the event occurs depends on the agent’s decision. In the case in question, the agent has only indirect control: Her performing a2 would not have obtained, had she decided to perform a2. It only obtained because she decided to perform a1. And her performing a1 did not obtain although she decided to perform a1. It would have obtained if, and only if, she had decided to perform a2. The kind of control an agent needs to

110

“Ought” Implies “Can”

have over an act in order for the act to have deontic status is, however, direct control. I hold this position for three reasons (which fit with my elaborations on Ought Implies Can in previous chapters). Firstly, there is no point in telling the agent that she ought to perform a1. For she did decide to perform a1. Nor is there a point in telling the agent not to perform a2. The agent’s only alternative to a2 was a1. So, considerations of action-guidance suggest that a2 lacks deontic status. Secondly, the agent does not seem to be responsible for her performing a2. All that one could have expected from her is to decide to perform a1. And she did decide to perform a1. This speaks against a2’s being wrong. How about a2’s being right? Here, the problem is that a2 has very small expected value. This makes it counterintuitive to call a2 right. Better then to go with Proposal 3 and judge that a2 lacks deontic status. Thirdly, the agent does not seem to have a reason against performing a2. She certainly deliberated correctly and accordingly decided to perform a1. She had a reason to decide to perform a1. Accordingly, had the agent decided to perform a2 because of a2’s very low expected value, it would seem counterintuitive to say that the agent had a reason for performing a1 – the act with the largest expected value, which she unwillingly would have ended up performing. To repeat, the account I defend here differs from the account Prichard and Ross suggested in an important respect. According to them, all an agent is ever responsible for, and hence the only kind of things that can be right, wrong, or obligatory, are decisions. In contrast, according to my account, full-blown actions can also have deontic status, that is, be right, wrong, or obligatory. They have deontic status whenever an agent has direct control over them. I conclude that subjective consequentialists should understand their position as a doctrine according to which acts are to be assessed indi­ rectly, namely, on the basis of the relationship the acts bear to decisions. An act inherits the rightness, wrongness, or obligatoriness of a decision if, and only if, the agent has direct control over the act. Otherwise the act lacks deontic status altogether. One might wonder whether the version of subjective consequentialism I defend should still be called act-consequentialism. I think that it should be called so for three reasons. First, it might well be that decisions or attempts also qualify as actions. If so, the deontic iudicanda are still acts. Second, decisions or attempts are at least a crucial component of actions. So, the deontic iudicanda are at least parts of acts. Third, it seems correct to say that decisions and full-blown actions have the same rationally expectable consequences. I have talked about inheritance, but one could reformulate my view and say that acts are being assessed not on the basis of a relation they have to decisions but on the basis of their own rationally expectable consequences (though only if they result from decisions in the right way). Be this all as it may, it is, ultimately, a terminological question. If you

Subjective Consequentialism

111

prefer, you can call my position decision-consequentialism. However, for the reasons mentioned I will go on calling it act-consequentialism.

4. Objections to the Volition Thesis In this section, I address several objections Terence McConnell has raised against Prichard’s and Ross’s view. My aim is to show that the objections are not worrisome when applied to the version of subjective consequen­ tialism I suggested in the previous section, although some of them seem troubling as far as other moral theories are concerned. The first objection McConnell (1989, p. 444) raises is this: [T]rying per se is a rather empty notion; trying is trying to do some act. And even if trying itself is an (internal) act, as some have argued, it is still always trying to do something else. So that which one is try­ ing to do seems to be central, and indeed ineliminable. It is true that trying or deciding is always trying or deciding to do some­ thing that qualifies as a full-blown action. In this sense, the full-blown action is ineliminable. This is true, but trivially so. The more important question is whether, and in which sense, the full-blown action might be central to attempts or decisions. In the context at hand, the only plau­ sible suggestion that comes to my mind is the following: It seems that, according to common-sense morality, you often have a moral duty to decide to φ or to try to φ because you have a moral duty to φ. Assume, for example, that you have a moral duty not to harm people. This duty seems to explain why you also have a duty to try not to harm people. However, this is all very different when it comes to the version of sub­ jective consequentialism that I advocate. What explains your duty to decide to maximize expected value is not, I submit, your duty to maxi­ mize expected value. It is the other way around. What explains the right­ ness of your maximizing expected value is, as I explained in the previous section, that your maximizing expected value results from your decision to maximize expected value. Your duty to decide to maximize expected value is not explainable by another duty. Your duty to decide to maxi­ mize expected value is explainable by the good: If consequentialism is correct, then the good provides reasons for decisions (and actions), and nothing else does. The aforementioned observation concerning common-sense morality is correct: According to common-sense morality, your duty to φ often explains why you ought to try to φ. Common-sense morality is imbued with deontological presuppositions. Some act-types are, ceteris paribus, wrong, according to deontology/common-sense morality. The wrong acttypes might include harming innocent people, lying, breaking promises, and eating the last cookie. If some types of acts are wrong, then this

112

“Ought” Implies “Can”

explains why you ought to try to avoid performing them. However, it would be a mistake for consequentialists to take their cue from deontolo­ gists. Consequentialists can put aside McConnell’s first objection with a clear conscience. However, McConnell (1989, p. 444) has another arrow in his quiver: Consider now a second objection to the claim that moral requirements range over attempts or tryings. Suppose that an agent believes that he ought to do A by noon. Suppose too that at 9:00 a.m. he tries (genu­ inely) to do A but fails. Fortunately, however, the agent will have several more opportunities before noon to try to do A again. But if oughtstatements ranged over attempts only, it would be hard to explain why in this situation the agent ought to continue trying to do A. Notice first that according to the version of subjective consequentialism that I defend, though moral duties range primarily over decisions or attempts, they do not range only over decisions or attempts but also over actions. McConnell’s central point, though, is that Prichard and Ross cannot explain why the agent in the case under consideration ought to try to do A after 9:00 a.m. As far as my version of subjective consequen­ tialism is concerned, however, there is no problem at all. If doing A has, if done by noon, the largest expected value (of all supposedly available actions), then deciding or trying to do A, under normal circumstance, also has the largest expected value (of all available decisions or attempts). Hence, the agent ought to try to do A, again and again, unless doing A, and in turn trying to do A, cease to have the largest expected value. McConnell’s second objection, then, also fails as far as my version of subjective consequentialism is concerned. We move on: A third difficulty concerns the assessment of others’ moral decisions. If moral requirements range over tryings, it seems that others cannot determine whether an agent has done what he ought to have done. Since trying to do a certain act, say A, is compatible with all sorts of external behaviour that does not involve doing A, others will be hard pressed to know in specific situations what an agent tried to do. In fact, even if others know that an agent ought to set himself to do A, and even if the agent does A, others still cannot know that the agent did what he ought to do because he may have tried to do something other than A. That the view in question renders an agent’s moral decisions virtually immune from the criticism of his fellows is a defect of that position because one of the important functions of moral theories is to assess the conduct of others. (McConnell 1989, p. 445. Footnote omitted.)

Subjective Consequentialism

113

My response to this objection is threefold. Let me start by pointing out that some of McConnell’s claims are exaggerations. It is simply not true that the moral theory endorsed by Prichard and Ross renders an agent’s moral decisions virtually immune from the criticism of his fellows. For the behaviour of people often reveals their attitudes – “by their fruits you shall know them”. In ordinary situations, we can conclude from the fact that someone φ-s that she tries to φ. For example, if someone robs a bank, we may infer that the person tries to rob the bank. From the fact that someone constantly brings about good states of affairs, we may infer that she constantly tries to bring about good states of affairs. There are, however, certain circumstances where a person φ-s although she does not try to φ. This brings me to my second point, though: Moral theories according to which moral duty ranges over full-blown actions rather than decisions or attempts face the same problem. Assume the bank-robber tries not to rob the bank but cannot refrain from robbing the bank (for whatever fancy reason). Then, certainly, the bank-robber’s responsibility is diminished – arguably to the point that she does not bear any responsibility at all. The assessment of the behaviour of agents is relevant when it comes to praise and blame. But when it comes to praise and blame, what an agent decides or tries to do is also relevant. Finally, notice that subjective consequentialists might deny that there is any relevant connection of the sort suggested by McConnell between wrongdoing and praise and blame. Such a connection is controversial (cf., e.g. Introduction, Section 3; Chapter 3, Section 4). In sum, McCon­ nell’s third objection also fails. Let us come to McConnell’s (1989, p. 444) final point: [I]f what agents are required to do is to try or to set themselves to do certain acts it seems to follow that there is no such thing as trying but failing to do one’s duty. The reason for this is that the notion of trying to try is puzzling, at best. [. . .] What I do mean to say is that we normally think that moral requirements are sometimes difficult to fulfill, and that sometimes we try but in spite of our best efforts fail to fulfill them. Our ordinary conception may be mistaken; but the burden of proof is on a view that departs from common sense. As far as this point is concerned, it is important to bear in mind that I assume that consequentialism is correct. The question we want to answer is whether a version of consequentialism that accepts the Volition Thesis is inferior to one that does not. What we need my version of subjective consequentialism to compare to, then, is objective consequentialism. Since according to objective consequentialism an act’s actual consequences (as compared to the consequences the act’s alternatives would have) deter­ mine whether the act is right or wrong, even the best possible humans

114

“Ought” Implies “Can”

act wrongly almost all the time if objective consequentialism is true (Miller 2003). The important point is that according to the relevant – i.e. consequentialist – alternative to my version of subjective consequentialism, even humans who constantly try as hard as they can to fulfil their moral duties fail to do their duty. Thus, both versions of consequentialism depart from common-sense morality, in opposite directions. Since my ver­ sion of subjective consequentialism does not fare worse than the relevant alternative, McConnell’s final objection does not go through either.

5. Conclusion I have argued that subjective consequentialism is compatible with Ought Implies Can. Subjective consequentialists should accept the Volition The­ sis. If they do so, they can keep on judging acts insofar as the acts result from attempts, decisions, or something like that. Subjective consequen­ tialism’s deontic iudicanda, however, are attempts, decisions, or some­ thing like that. This version of subjective consequentialism, I have tried to show, is immune against the critique generally put forward against moral theories that entail the Volition Thesis.

Notes 1. Graham 2011, pp.  340–341, note 10. The problem is also mentioned, for instance, in Feldman 2012, p. 159. 2. Similar considerations (though shorter and in a slightly different context) have been articulated in Hedden 2012, p. 354. Hedden (2012, pp. 352–354) points out, with reference to Bratman 1987 and to Anscombe 1957, that your abili­ ties to make decisions and to form intentions are, if you are a rational agent, wholly determined by your beliefs and desires. 3. This kind of depiction has been introduced in Kagan 2000. 4. On the complications alluded to in the parentheses, see Hooker 2000, Chapter 3. 5. See McConnell 1989, pp. 441–443, for quotes and references.

Part 3

Intuitions About Perspective-Dependence

8

A New Perspective on the Perspective-Dependence of Moral Duties

In this chapter, I argue for a new perspective on the debate over the perspective-dependence of moral duties. The first two parts of the book focused on objective and subjective consequentialism. But subjectivism and objectivism about moral duties can be defined independently of con­ sequentialism; cf. Introduction, Section 2. Subjectivism is simply the claim that moral duties are determined by the perspective of the agent (by her beliefs or, according to the variant that I prefer, by the evidence available to her), rather than by the facts. According to objectivism, moral duties depend on things as they are rather than as they appear to be. For illus­ tration, since Oedipus believes in accordance with the evidence that the woman he is marrying is not his mother, he is acting rightly according to the typical subjectivist. The typical objectivist disagrees: Since the woman married by Oedipus indeed is his mother, Oedipus is acting wrongly. The distinction between objective and subjective moral theories is well established. But philosophers usually fail to pay attention to other dif­ ferences among moral theories while being engaged in the debate over the perspective-dependence of moral duties.1 They just try to find the truth about perspective-dependence, period. In doing so, they take it for granted that the truth can be found independently of what we believe is the correct moral theory (the distinction between objective and sub­ jective theories notwithstanding) – may it be Kantianism, virtue ethics, consequentialism, or what-have-you. Their distinction between, on the one hand, the question whether moral duties are perspective-dependent and, on the other, which is the best moral theory, is reminiscent of the well-known distinction between normative ethics and metaethics: Many philosophers believe that the correct metaethical theory has no implica­ tions for what the best theory in normative ethics is, and vice versa. This chapter presents a – not substituting but complementary – alternative to the theory-independent approach to perspective-dependence: the New Perspective. The New Perspective is based on the idea that, if Kantian­ ism is correct, it might well be advisable to look for the truth about perspective-dependence at places different from those we should scru­ tinize if consequentialism is correct. According to the New Perspective,

118

Intuitions About Perspective-Dependence

when it comes to our attempt to find out whether moral duties are perspective-dependent, we should adopt a theory-laden perspective. We should not only ask, “Are moral duties perspective-dependent?” but also: “Are moral duties perspective-dependent relative to this or that moral theory?” If you think that there is nothing wrong with your marrying your mother, the New Perspective holds, it may not be instructive for you at all to focus on the Oedipus case in order to figure out whether moral duties are objective or subjective. For letting the Oedipus case determine your intuitions about perspective-dependence presupposes that there is something wrong with your marrying your mother. Here is the position I advocate: The New Perspective We should ask and answer the question of subjectivism versus objec­ tivism relative to specific moral theories and should let our intuitions in the matter be guided by cases that involve duties that are paradig­ matic for the respective moral theory. Based on the New Perspective, I will develop in this chapter what I call: The New Perspective Argument for Subjective Consequentialism (1) If the only duty paradigmatic for consequentialism seems to be subjective, then consequentialists have reason to accept subjective consequentialism. (2) Your duty in disaster situations is the only duty paradigmatic for consequentialism. (3) Your duty in disaster situations seems to be subjective. (4) Therefore, consequentialists have reason to accept subjective consequentialism. Premise (1) will be explained and established in Sections 1 and 2, where the New Perspective on perspective-dependence will be introduced (Section 1) and an objection to it addressed (Section 2). Sections 3 and 4 aim to estab­ lish premise (2). Section 3 determines the features a duty needs to have in order to be paradigmatic for consequentialism. Section 4 argues that only duties in disaster situations have these features. Premise (3) will be justified in Sections 5 and 6. In Section 5, I try to show that only a certain kind of disaster situations is informative with respect to the perspectivedependence of moral duties on consequentialism. This kind of disaster situ­ ation will be carefully examined in Section 6. Section 7 concludes.

1. In Support of Premise 1: The New Perspective What is the New Perspective on the perspective-dependence of moral duties? The New Perspective is a position in moral epistemology. It says

A New Perspective

119

that those cases involving moral duties that are paradigmatic for the moral theory we are interested in will reveal whether the theory is natu­ rally understood in an objectivist or in a subjectivist way. It may thus well turn out, say, that moral duties are naturally understood as being objec­ tive on Kantianism but subjective on consequentialism. But do moral theories not already tell us whether moral duties are perspective-dependent? What I mean by a “moral theory” in this chapter is, of course, a theory that does not already tell us whether moral duties are perspective-dependent. You can call it a “moral proto-theory” or a “moral doctrine”, if you like. The important point, however, is that we can formulate objectivist as well as subjectivist versions of Kantianism, virtue ethics, consequentialism, and so on. The objectivist and subjectiv­ ist versions of each respective theory have a lot in common – so much, I think, that we are able to distinguish cases involving paradigmatic moral duties from cases that do not involve such duties. And it is the former type of case we should let reveal our intuitions about perspective-dependence relative to moral theories. What do I mean by “paradigmatic” moral duties? I mean moral duties that, against the background of common-sense morality, seem to be best explainable by the respective moral theory. Moral theories identify right-, wrong-, and ought-making features of actions. Cases involve paradig­ matic moral duties if and only if the actions they are about seem to have the right-, wrong-, and ought-making features, from the viewpoint of common-sense morality, that actions of this kind are said to have by the respective moral theory. For convenience, I will occasionally talk about paradigmatic cases or paradigmatic situations, by which I mean cases that involve paradigmatic duties. To illustrate, if a case is about an action that seems to be right because it passes the Categorical Imperative, or is believed – in accordance with the evidence – to pass it, then the case involves a moral duty that is para­ digmatic for Kantianism. And thus, if we want to know whether moral duties are perspective-dependent and we think that Kantianism is correct, we can let the case determine our intuition as to whether moral duties are perspective-dependent according to Kantianism. Of course, things become problematic if there are several cases with paradigmatic duties where some of these seem to be objective and others subjective. In this case, it seems appropriate to conclude that the moral theory in question involves objective duties under some circumstances and subjective duties under others. However, the argument I present for subjective consequentialism holds that there is only one kind of case that involves duties that are paradigmatic for consequentialism. I shall there­ fore not do further research into this complication. Why should we adopt the New Perspective? The general methodologi­ cal assumption underlying the New Perspective is that moral intuitions are a guide to moral truth. With respect to perspective-dependence, this assumption amounts to the claim that if moral duties seem to be

120

Intuitions About Perspective-Dependence

perspective-dependent, then we have, ceteris paribus, reason to believe that they are. If they seem to be independent of the agent’s perspective, then we have, ceteris paribus, reason to believe that they are independent of her perspective. But why take into account only paradigmatic duties? As a starting point, notice that a proponent of a particular moral theory will not be happy to accept any old moral intuition as a reason to believe what seems to be the case. Common-sense morality contains very different kinds of moral duties. Some appear to fit well into a Kantian picture, others seem rather to be of a consequentialist nature, and so on. What should propo­ nents of a moral theory say about duties that apparently do not sit well with their theory? In general, there are two possible responses.2 The first response is to say that contrary to prima facie appearances, the moral theory in question can explain the duty under consideration, be it fully or partially. A paradigm example of this move has been based on the consequentialist distinction between a criterion of rightness and a decision procedure.3 Suppose that a consequentialist can explain why you ought not even to think about sacrificing your own children for the common good (under normal circumstances). This would seem to be an important step in harmonizing consequentialism with the existence of special obligations. Or take the ethical egoist who explains why your selfinterest gives you reasons to care about the welfare of others. Given such a type of explanation, the proponent of a moral theory can admit that the moral duties recognized by common-sense morality, or something very similar to those duties, exist even though they do not, prima facie, seem to fit into the theory’s picture of morality. We can call this kind of response the appeasement strategy. The second possible response is the confrontational strategy. It consists in dismissing intuitions about duties that do not square with the moral theory in question. The confrontational strategy can be based, for exam­ ple, on debunking explanations for why we have certain moral beliefs. It may be argued, for example, that untenable religious convictions are the only sources of some moral beliefs, such as the sanctity-of-life doctrine. However, debunking explanations may not even be necessary in order to be justified in dismissing a particular moral intuition. For if an oth­ erwise sufficiently compelling moral theory is at odds with a particular moral proposition that seems true, one may be justified to conclude that the proposition is false, its appearance of truth notwithstanding, simply because it contradicts the theory. There can be mixed strategies, of course, which contain elements of both the appeasement and the confrontational strategy. An ethical egoist may argue, for instance, that your altruist leanings are misguided insofar as they do not, in an indirect way, contribute to your own well-being. Or the consequentialist may dismiss the intuition that it is wrong to kill someone insofar as it cannot be derived from the moral duty to promote

A New Perspective

121

the good. In my opinion, the proponent of a moral theory will have to choose mixed strategies for the major part of non-paradigmatic situa­ tions she might be confronted with. For appeasement strategies typically, I think, explain intuitions about these situations only in part, at best. Be that as it may, the general point is that the appeasement and the con­ frontational strategies are the cornerstones of the battlefield on which moral theories are being defended against intuitions from common-sense morality that, at first glance, speak against them. I can now explain why the New Perspective states that only cases involving paradigmatic duties should be allowed to determine our intu­ itions about perspective-dependence. As a starting point, recall that actions in non-paradigmatic cases do not seem to have the right-, wrong-, and ought-making features the moral theory in question ascribes to them. As far as the respective non-paradigmatic cases are concerned, moral the­ ories seem to be false! The ways the proponent of a moral theory can react to non-paradigmatic cases are limited by the appeasement and the confrontational strategy. If the proponent of a moral theory chooses the confrontational strat­ egy, she thinks that common-sense morality is completely flawed as far as non-paradigmatic cases are concerned and that the alleged moral duty does not exist. Why should the proponent of an atheistic moral theory, say, care about whether the alleged duty associated with the sanctity­ of-life doctrine seems to be objective or subjective? If this duty does not exist, it would obviously be a mistake to let intuitions concerning cases that presuppose the existence of that duty determine our view about perspective-dependence. What about cases where the moral theorist chooses the appeasement strategy? There are three reasons not to rely on the intuitions associ­ ated with these cases. The first reason is that it is often hard to decide whether, and in how far, the appeasement strategy is successful. Insofar as it is not, the moral theorist will have to recede to the confrontational strategy and to deny any relevance of the case in question. And insofar as it is uncertain whether – and, if so, to which extent – an appeasement strategy is successful, it seems epistemically wrong for the proponent of a moral theory to base her beliefs about the perspective-dependence of moral duties on non-paradigmatic cases. The second reason has to do with the fact that appeasement strategies, as I said before, often aim only at partial explanations of intuitions con­ cerning non-paradigmatic cases. They often do not deny that the duties in question even upon reflection seem to have right-, wrong-, or oughtmaking features different from those stated by the moral theory. They rather consist in pointing out that you can reconstruct a moral duty rec­ ognized by common-sense morality despite the fact that this duty can­ not be grounded the way it appears to be grounded. According to the reconstruction, the duty might even have a slightly different content than

122

Intuitions About Perspective-Dependence

it has according to common-sense morality. This is especially clear in the case of the distinction between a criterion of rightness and a decision procedure. “Under normal circumstances, you ought not even to think about sacrificing your children for the common good”, the consequen­ tialist might say. But the full truth is that, according to her, “you ought to sacrifice your children nonetheless whenever this maximally promotes the good”. The important point is that if, from the perspective of the moral theory under consideration, the duty in question does not really have the features it appears to have against the background of common-sense morality, then intuitions concerning that duty generally do not seem trustworthy, including intuitions concerning perspective-dependence. At least as long as one cannot combine the appeasement strategy with something like a debunking explanation that explains the theory-hostile intuitions away by pointing to some source of error and shows that intuitions regard­ ing perspective-dependence are not affected by that error, it seems epis­ temically right for the proponent of a moral theory to ignore the latter intuitions. The third reason why it can be wrong for the moral theorist who chooses the appeasement strategy to take into account intuitions con­ cerning perspective-dependence that stem from non-paradigmatic cases is that whether or how a seeming moral duty can be reconstructed on a moral theory might often depend on whether moral duties are perspective-dependent according to the theory. The moral theo­ rist would thus get a circularity problem or an infinite regress if she attempted to let cases with non-paradigmatic duties determine her view on perspective-dependence. In sum, when we try to find out whether moral duties are perspectivedependent on a particular moral theory, we should focus on cases involv­ ing moral duties that are paradigmatic for that theory. If there is only one paradigmatic duty for a moral theory, we should conclude that, with respect to perspective-dependence, all duties postulated by the moral the­ ory are of the kind the paradigmatic duty seems to be.

2. In Support of Premise 1: The Primacy of Intuitions Objection This section addresses an objection against my claim that only intuitions concerning paradigmatic situations should be taken into account when determining whether moral duties are perspective-dependent according to a moral theory. In a sense, I have assumed a primacy of moral theory in the previous section. For I have argued that a proponent of a moral theory should take into account only intuitions that seem best explain­ able by her theory. Against this, someone could raise:

A New Perspective

123

The Primacy of Intuitions Objection On the New Perspective, moral theorists are free to take into account only paradigmatic duties, that is, only those moral duties that fit their respective theory. In effect, the New Perspective only tells us how convincing or unconvincing moral theories are when it comes to explaining certain duties, be they objective or subjective. But it should be the other way around. We should first determine which moral duties there are and whether they are objective or subjective – for this is the really important question! And then we should assess moral theories on the basis of how well they can explain this. For illustration, take the example of the duty not to kill someone. Accord­ ing to the New Perspective, consequentialists should dismiss intuitions to the effect that it is wrong to kill someone when they decide between objectivism and subjectivism. But my critic will say that we, as moral theorists who are looking for the truth about perspective-dependence, should not commit ourselves to, say, consequentialism. Rather, we should take seriously our intuition that it is wrong to kill someone. And if this intuition suggests that moral duties are, say, objective, then we should accept this as evidence that moral duties are indeed objective. If the exis­ tence of the duty not to kill someone, or its being objective, do not square with consequentialism, then so much the worse for consequentialism! My reply to the Primacy of Intuitions Objection comprises six points: •

Misunderstanding I agree that it does not matter, ultimately, whether moral duties are objective or subjective according to some moral theory. What ulti­ mately matters is whether moral duties are objective or subjective, period. However, I have explained that the New Perspective is a posi­ tion in moral epistemology. The New Perspective does not entail that the truth about perspective-dependence is theory-relative, or any­ thing like that.

•

Fruitlessness One motivation to accept the New Perspective is the thought that although the debate about the perspective-dependence of moral duties has been going on for quite a while, there is no general agree­ ment about whether moral duties are objective or subjective. This suggests trying a new path. The New Perspective has not yet been probed. (To be sure: I assume that I have put forward convincing arguments against objective consequentialism in the first two parts of this book. But this will not help those moral philosophers who think that consequentialism is false in the first place, be it objectivist or subjectivist.)

124

Intuitions About Perspective-Dependence

•

Supplementation The New Perspective does not require giving up the theory-independent search for the truth about perspective-dependence of moral duties. Fans of the theory-independent approach can use the New Perspec­ tive in addition to their favoured methods. A natural suggestion for this supplementation goes as follows: In addition to the theoryindependent arguments for subjectivism or objectivism, one can use the New Perspective in order to find out whether one’s favoured moral theory supports subjectivism or objectivism. So, the New Perspective need not to be seen as a rival to the theory-independent approach but can instead be regarded as a means to get additional arguments.

•

Alternative Regardless of whether the theory-independent approach succeeds in finding out the truth about perspective-dependence, the New Per­ spective promises to yield results that are interesting on their own. For it is certainly a question that is interesting for moral philoso­ phers whether a particular moral theory suggests that moral duties are objective or instead that they are subjective. For objective as well as subjective versions can be formulated of all major moral theories and we then know which one is the natural formulation.

•

Relevance The New Perspective can be deployed in arguments about the plau­ sibility of particular moral theories – even if the theory-independent approach is sufficient to settle the debate over objectivism and subjec­ tivism. For illustration, assume that the theory-independent approach is successful on its own and reveals that moral duties are objective. Suppose further that the New Perspective reveals that a certain moral theory, let us say consequentialism, suggests that moral theories are subjective. We then have a reason to believe that consequentialism is false. So, if the theory-independent approach to perspectivedependence is successful, the New Perspective is relevant for assess­ ing moral theories.

•

Incoherence The Primacy of Intuitions Objection suggests that we should take into account all common-sense moral intuitions, at least as far as they have not been debunked or disqualified in other ways. However, it is incoherent to accept a certain moral theory and, at the same time, take into account all intuitions stemming from common-sense morality when forming one’s moral beliefs. For some set of these intuitions always seems to be best explainable in a way that is at odds with whatever moral theory you might happen to accept and thus suggest that your moral theory is false. Therefore, proponents

A New Perspective

125

of moral theories should dismiss intuitions insofar as they cannot be assimilated by means of the appeasement strategy. Consequentialists should not take into account intuitions that there is a duty not to kill someone that goes beyond the duty to promote the good, and they should not take into account intuitions that presuppose a duty not to kill someone. (This point obviously only applies to people who accept a moral theory. Other people can ignore it.)

3. In Support of Premise 2: Characteristics of Your Consequentialist Duty How can one tell whether your duty in disaster situations is paradigmatic for consequentialism? The answer comprises two parts. First, I have to identify features of the moral duty you have on consequentialism – of your “consequentialist duty”, for short. These have to be right-making features objective and subjective consequentialists agree on. This job will be done in this section. Second, I will show that, from the viewpoint of common-sense morality, your duty in disaster situations – and only this duty – shares the right-making features characteristic of your consequen­ tialist duty. This is a task for the next section. Which are the features of your consequentialist duty that objective and subjective consequentialists can agree upon? There are at least three such features. These will suffice to single out disaster situations as the only cases involving duties that are paradigmatic for consequentialism. First of all, consequentialists accept that your consequentialist duty is about the promotion of the good. The good has traditionally been understood by consequentialists to be evaluator-neutral. This means that, according to the traditional view, states of affairs are better or worse, respectively, than their alternatives, independently of who the evaluator is. In contrast, according to evaluator-relative theories of the good, states of affairs are better or worse, respectively, than their alternatives, relative to some agent. As for an illustration, assume that either I murder someone or you murder someone. Other things being equal, which outcome is worse? An evaluator-relative theory of the good might imply that the first outcome is morally worse relative to me and the second is morally worse relative to you. According to the traditional view, by contrast, both outcomes are equally bad. Another traditional tenet of consequentialism is that your duty is agent-neutral. Roughly speaking, a duty is agent-neutral iff it is an equally weighty duty for everyone; and agent-relative otherwise. According to consequentialism, all possible agents are under all possible circumstances required to promote the good. (Be it by doing what can reasonably be held to be most effective, as subjective consequentialists hold, or by doing

126

Intuitions About Perspective-Dependence

what is indeed most effective, as objective consequentialists say.) By con­ trast, common-sense morality recognizes agent-relative duties. It says, for instance, that you have a duty to take more care of your children than of other people’s children. Other persons have a duty to take more care of their respective children than of yours. The third traditional tenet of consequentialism is that the good explains your duty. This follows from the very nature of consequentialism as a moral theory (cf. Introduction, Section 1). If you ought to bring about one outcome rather than another, this is, according to objective consequentialism, because the first outcome is better than the second. According to subjective consequentialism, it is because you are justified in expecting the first outcome to be better than the second. True, accord­ ing to subjective consequentialism, it is not only the good but also the evidence available to you that explains the content of your duty. But I take it that, according to objective as well as subjective consequentialism, the explanation for your having a duty at all is provided by the good alone.4 My argument assumes all three tenets of the traditional view: that the good is evaluator-neutral, that your duty is agent-neutral, and that the good fully explains your duty. As we will see in the next section, this is important when it comes to choosing between the duties recognized by common-sense morality. However, I think that arguments along the lines pursued here can also be developed for non-traditional versions of conse­ quentialism. In order to keep things simple and because I am sympathetic towards the traditional tenets, I will not rehearse those moves.

4. In Support of Premise 2: Disaster Situations and Paradigmatic Duties In this section, I argue that only disaster situations involve moral duties that are paradigmatic for consequentialism. Only disaster situations involve moral duties that are such that (i) they seem to be about the promotion of the good, understood in an evaluator-neutral way; (ii) they seem to be agent-neutral; and (iii) your having the duties seems to be best explainable by appeal to the good. Common-sense morality recognizes many very different moral duties. I think, however, that only the moral duties involved in disaster situations are paradigmatic for consequentialism. I am probably not able to address all kinds of moral duties recognized by common-sense morality, for I do not know of any complete list of these duties. I will address all duties that come to my mind, though. For illustrative purposes, let us start with an influential case that has been put forward by Graham Oddie and Peter Menzies. I have already alluded to the case in the introduction to this chapter:

A New Perspective

127

Oedipus Oedipus killed his father and made love to his mother. Accepting Greek, and conventional, views on the morality of patricide and incest, Oedipus acted wrongly on both counts. But that is not the whole of the story. For it was precisely in the attempt to avoid these very wrongs that Oedipus so acted, believing in the light of the best evidence available to him that he was successfully avoiding them. (Oddie and Menzies 1992, p. 512) It seems to me that Oedipus, if the case were relevant, would support objective consequentialism. For Oedipus seems to have violated moral duties, although he believed, in accordance with the evidence, to have acted rightly. Oedipus’s beliefs simply seem to have been mistaken rather than having rendered his actions right. (I will expand on these intuitions in Section 6.) Be this as it may, the alleged moral duties involved in Oedipus are clearly not paradigmatic for consequentialism. First of all, the duties are not about the evaluator-neutral good. One might argue that they concern the promotion of what is good-relative-to-Oedipus. It might be better­ relative-to-Oedipus if Oedipus does not kill his father and does not marry his mother. But evaluator-relative value does not fit into the traditional consequentialist picture of morality. Secondly, the alleged duties are agentrelative. You ought not to kill your father, not even under circumstances that would justify the killing of a person other than your father, and you ought not to marry your mother. Other people do not have these duties. They rather ought not to kill or marry their respective parents. Thirdly, even if one accepted that the duty is about the promotion of evaluatorrelative value (so that the case would in this respect correspond only to some non-traditional versions of consequentialism anyway), it would not be clear whether the evaluator-relative good explains Oedipus’s duty. Is it not more plausible to think that it is a fundamental duty not to kill or marry your parent, a duty not derivable from considerations of value? It might even appear that the agent-relative good, on the assumption that there is such a thing involved in Oedipus, is explainable by the duties, rather than the other way around. We may conclude that Oedipus is no good case for consequentialists who want to get clear on perspective-dependence. I can now be quick with a large number of moral duties recognized by common-sense moral­ ity that are also clearly not paradigmatic for consequentialism: Common­ sense morality recognizes moral duties not to harm people, not to lie, to keep your promises, to take care of your near and dear ones, and not to violate the rights of other people.5 All these duties are agent-relative. You ought not to harm people even if your harming people results in an overall smaller amount of harm. You ought not to lie even if, as a

128

Intuitions About Perspective-Dependence

consequence, more people will tell lies. You ought to keep your promises and to take care of your near and dear ones. And you ought not to vio­ late the rights of other people even if this minimizes the total number of rights-violations. Notice that all the mentioned duties arguably qualify as constraints and are as such incompatible with consequentialism, as Shelly Kagan explains: [A constraint is] a prohibition which rules out performing a certain type of act . . . even when performing an act of that type is necessary to bring about the best results overall. Now obviously enough, con­ sequentialist theories leave no room for moral constraints, since such theories hold that whatever act will lead to the best results is always permissible. But it is precisely at this point that commonsense moral­ ity diverges from consequentialism.6 So, what we are after is a duty recognized by common-sense morality that does not qualify as a constraint. How about the duty to help the needy? Here is a case suggested by Kagan: Suppose that someone is drowning in the lake, and the only way she can be saved is if I row out to her in a boat and pull her in. Should I do it? Presumably, the fact that my act would have a good result – it would save a life! – is one morally relevant factor in determin­ ing the rightness or wrongness of the act. In particular, if the act is permitted – or perhaps even morally required – this is in large part due to the very fact that my act would help to bring about this good result, and there is no other way to achieve it. The goodness or bad­ ness of the results is thus a morally relevant normative factor – a factor that can help determine the moral status of an act. (Kagan 1998, p. 17) I am not persuaded by Kagan’s analysis. Let us call the person who ought to save the drowning woman Shelly. Is Shelly’s duty paradigmatic for consequentialism? I think that Shelly’s duty is agent-relative, according to common-sense morality. Suppose that Shelly’s not saving the drown­ ing woman results in two people being saved by other agents. The reason may be that the other two agents are psychopaths who make their deci­ sions to rescue drowning persons dependent on Shelly’s decision not to save the drowning woman. On common-sense morality, Shelly still seems to be required to save the woman. Just as one is not allowed to commit a murder on common-sense morality even if, as a result, two other murders will not be committed by other agents, so one is not allowed not to help a person even if, as a result, two other people would get help. Thus, Shelly’s duty is not necessarily about maximizing the number of saved people. It

A New Perspective

129

is rather about seeing to it that he, Shelly, actively saves as many people as possible of those who are related to him in a certain way. At least two factors seem to play a role in this case according to common-sense morality that do not fit well with consequentialism. One is the distinction between doing and allowing. It does not seem to be enough to let other people save drowning persons. Rather, Shelly ought actively to save people. Second, not every drowning person has an equally strong claim against Shelly to be saved. People located nearer to him have a stronger claim. Shelly is not allowed to let the woman drown even if this (for whatever reason) happens to result in two women being saved in Africa. This does not sit well with the assumption that Shelly’s duty is about the promotion of evaluator-neutral value.7 I conclude that your duty to help the needy qualifies as a constraint, too. What we need is a duty to bring about optimal states of affairs. But this is hard to find. Here are two candidates. The first is the duty of a politician who works for a global organization and whose job is to bring about optimal results on the entire planet. The second candidate is the duty of a doctor who, again, works for a global organization. Arguably, in these cases the distinction between doing and allowing is irrelevant. And it also maybe that the politician and the doctor ought to care about everyone concerned in an equal way and ought thus to be concerned with outcomes which are best from an evaluator-neutral viewpoint. Thus, the politician’s and the doctor’s duties do not seem to be constraints but obligations to bring about the best results. However, the doctor’s and the politician’s duties are, according to common-sense morality, not agent-neutral. The doctor has the duty qua doctor, and the politician qua politician. Other people, if they do not occupy the roles of the doctor or the politician, do not have the duties in question accord­ ing to common-sense morality. Moreover, the roles the doctor and the politician play in society explain, at least partly, why they have the duties they have. The duties are thus not fully explainable only by appeal to the good.8 This concludes my survey of cases with non-paradigmatic duties. Con­ trary to what some consequentialists, such as Kagan, have argued, it seems to me that almost none of the many duties recognized by common-sense morality fit in perfectly with the consequentialist picture, including the duty to help other people. This result is in line with some findings of rule­ consequentialists like Brad Hooker. Hooker (2000, Chapter 1) thinks it a cardinal virtue of rule-consequentialism that rule-consequentialism coheres very well with common-sense morality. Now, Hooker attempts to show that almost all moral duties can be thought of as being provided by moral rules the general internalization of which would have the best con­ sequences. There is one problematic case according to Hooker and other rule-consequentialists, however, namely, disaster situations.9 In disaster situations, you ought not to follow the rules but to promote the good

130

Intuitions About Perspective-Dependence

directly.10 Given these remarks from rule-consequentialists, one might expect disaster situations to involve duties that are paradigmatic for act­ consequentialism. And this is indeed what one finds, as I shall show now. By “disaster situations” I mean situations that involve a disaster. In disaster situations, agents ought to save as many valuable things as they can. Think of an exploding power plant, for example, or of a catastrophic oil spill. Other disaster situations involve tsunamis or hurricanes. In these situations, agents ought to save what seems valuable from an evaluatorneutral standpoint: human lives, animal species, the natural environment, monuments, and works of art. Disaster situations can, of course, also involve things that are valuable relative to the agent. Or there may be situations where the agent has special obligations, and thus agent-relative duties, stemming, say, from promises or from personal relationships. However, since we are only interested in duties that are paradigmatic for act-consequentialism, let us focus on disaster situations that lack these specialities. Recall the three features of your consequentialist duty. The first feature is that your duty is about the promotion of the good. In disaster situa­ tions, as I have just explained, agents ought to save what is valuable from an evaluator-neutral standpoint: human lives, animal species, and so on. Secondly, your duty in disaster situations is agent-neutral. Anyone who faces a disaster ought to do what she can in order to save as much good as possible. Neither does it matter who the agent is nor in which way she promotes the good. Thirdly, the good seems to explain your duty. It is because human lives etc. are valuable that you ought to save them. In this section, I have argued that disaster situations, and they alone, involve duties that are paradigmatic for consequentialism. As for the “they alone”, I have dealt with all duties recognized by common-sense morality that came to my mind. Since I cannot exclude the possibility that I have overlooked something, my argument is tentative. Before we continue, let me briefly zoom out from the current discus­ sion to address a worry. If my claim that only disaster situations involve duties that are paradigmatic for consequentialism is correct, does this not pose a problem for consequentialism in the first place? After all, this claim suggests that common-sense morality and consequentialism do not cohere well, and this might be taken to speak against consequentialism. My response to this worry is twofold. First, let me just remind you of the appeasement strategy, confrontational strategy, and mixed strategies described in Section 1. Consequentialists can use these strategies to argue that the “gap” between consequentialism and common-sense morality is not worrisome. Of course, since the gap seems to be large, one might think that a lot of work needs to be done to defend consequentialism, whereas other moral theories might be more promising because they have much more in common with common-sense morality. However, and this is my second point, recall that I find consequentialism primarily appealing as

A New Perspective

131

a theory of practical reason, rather than as a moral theory; see Introduc­ tion, Section 3. This is why I am not troubled by the idea that consequen­ tialism is very revisionary when compared to common-sense morality.11

5. In Support of Premise 3: Eliminating Licensing Strategies In this section, I argue that only a certain kind of disaster situation is informative with respect to perspective-dependence. Why do I think that only a certain kind of disaster situation is informative? There is a prob­ lem with what we may call “simple disaster situations”. Here is a simple disaster situation: You think, in accordance with the evidence, that you will prevent a disaster by flooding a mine. So you flood the mine. Unfortunately, your flooding causes a disaster – without your flooding the mine, everything would have been fine. What is the problem with this case? According to the subjective conse­ quentialist, you ought to have flooded the mine. The objective consequen­ tialist holds that you ought not to have flooded the mine. But this is not the end of the story. Notice, first of all, that the simple disaster situation does not reveal clear intuitions about perspective-dependence. One might have the impression that there is something kind of wrong and something kind of right with your flooding as well as with your not flooding the mine – to put it as vaguely as possible. This, however, does not justify any view on perspective-dependence. For the intuitions are just too vague and thus allow for different interpretations. This brings us to a related point. There are some licensing strategies available to both the objective and the subjective consequentialist with respect to the actions they respectively judge to be wrong. By a “licens­ ing strategy” I mean the concession that something strongly counts in favour of an action combined with the insistence that this thing is not apt to make the action right. The subjective consequentialist can argue that your not flooding the mine, though wrong, would have been best, because it would have had the best consequences of all actions available to you. The idea is that actual consequences determine an act’s goodness while its rationally expectable consequences determine its rightness (cf. Introduction, Section 2). The objective consequentialist has a variety of strategies available in order to license your flooding the mine. She can, perhaps most saliently, distinguish between a criterion of rightness and a decision procedure and argue that your decision procedure ought to have led you to flood the mine, although your doing so was wrong. Both kinds of licensing strategies, those available to subjective consequentialists as well as those open to the objective consequentialists, seem promising. And both seem compatible with the intuitions – at least my intuitions,

132

Intuitions About Perspective-Dependence

which are inconclusive – in the simple disaster situation. Hence, the case is not informative with respect to perspective-dependence. The solution is to concentrate on disaster situations that exhibit a cer­ tain structure. The structure must not allow for the licensing strategies of the subjective or the objective consequentialist. Now, I do not know how to exclude the subjectivist strategy. But I claim to know how to avoid the objectivist licensing possibilities. Here is a disaster situation that does not allow for the objectivist strategies. It is the famous Miners case, which has been discussed by Derek Parfit and more recently has received much attention due to an article by Niko Kolodny and John MacFarlane. Here is how Kolodny and MacFarlane present the Miners case: Miners Ten miners are trapped either in shaft A or in shaft B, but we do not know which. Flood waters threaten to flood the shafts. We have enough sandbags to block one shaft, but not both. If we block one shaft, all the water will go into the other shaft, killing any miners inside it. If we block neither shaft, both shafts will fill halfway with water, and just one miner, the lowest in the shaft, will be killed.12 The main features of the case are represented in Table 8.1. The outcome ranking is obviously such that it would be best if all min­ ers were saved, it would be slightly worse if one drowned, and if all miners drowned, this would be by far the worst outcome. In short, with “≻” mean­ ing better than and “≻≻” meaning much better than: (all saved) ≻ (one drowned) ≻≻ (all drowned) According to objective consequentialism, we ought to save all miners. According to subjective consequentialism, given that the miners are equally likely to be in either shaft, we ought to block neither shaft. The objective consequentialist might try to guide our action towards blocking neither shaft although this act is wrong according to objective consequentialism. Here are some candidates as to how the objective consequentialist might try to license our blocking neither shaft to that effect:

Table 8.1 Miners.  

The miners are in shaft A

The miners are in shaft B

We block shaft A We block shaft B We block neither shaft

All saved All drowned One drowned

All drowned All saved One drowned

A New Perspective

133

• “Although we ought to block the shaft the miners are in, we nonethe­ less ought to have a decision procedure that will lead us to blocking neither shaft.” • “We have to distinguish between what we ought to do and what we ought to try. While we ought to save all miners, we ought to try to block neither shaft.”13 • “Our blocking neither shaft would be blameless wrongdoing.”14 I will argue in the next chapter that all these and similar strategies fail. In order not to interrupt the argument at this point, let us simply assume that my arguments in the next chapter are successful. After eliminating the objective consequentialist’s licensing strategies (or rather relying on my promise to do so in the next chapter), we can now let disaster situ­ ations reveal our intuitions. My intuition concerning Miners is that our blocking neither shaft is right. Moreover, our blocking shaft A and our blocking shaft B are each wrong – regardless of where the miners are. These intuitions are in harmony with subjective consequentialism. This is why we ought to accept subjective rather than objective consequentialism. (If you happen to have contrary intuitions, this might be due to your having read the influential article by Kolodny and MacFarlane. I will engage with this article in the next section.) I conclude that Miners is an instructive disaster situation because it dis­ plays a structure that does not allow for objectivist licensing strategies and thus reveals relevant intuitions. Our duty in Miners seems to be subjective.

6. In Support of Premise 3: Against Kolodny and MacFarlane15 In this section, I elaborate on my intuitions about Miners. I cannot, of course, argue you into the subjectivist intuitions about Miners. What I will do, though, is to examine Miners carefully, hoping that, at least on reflection, you share my intuitions. Similarly, I will present the aforemen­ tioned subjective consequentialist’s licensing strategy at more length and invite you to accept it. Why bother pumping intuitions? The reason is that Kolodny and MacFarlane use Miners to argue against modus ponens for indicative con­ ditionals and, in the course of this, against subjectivism. Kolodny and MacFarlane say that the outcome of our deliberation should be: (5) We ought to block neither shaft. They also want to accept: (6) If the miners are in shaft A, we ought to block shaft A. (7) If the miners are in shaft B, we ought to block shaft B.

134

Intuitions About Perspective-Dependence

We know that: (8) Either the miners are in shaft A or they are in shaft B. (6), (7), and (8) seem to entail: (9) Either we ought to block shaft A or we ought to block shaft B. However, (5) and (9) are incompatible. We seem to face a paradox. In order to avoid this paradox, Kolodny and MacFarlane argue for a somewhat radical solution: They reject modus ponens for indicative conditionals. I will first invite you to share my subjectivist intuitions concerning Miners by examining the case carefully and comparing it to Oedipus. Then I will provide an auxiliary argument for subjectivism about Min­ ers. Finally, I will elaborate on the subjective consequentialist’s licensing strategy. The Major Argument Subjectivists accept (5). But they reject (6) and (7) if we interpret them as inducing the aforementioned paradox. Since (5) is intuitively plausible, I can confine myself to rejecting (6) and (7) on their paradox-invoking interpretation.16 Which interpretation of (6) and (7) leads to the aforementioned para­ dox? This depends on how we understand (5). I assume that the “ought” in (5) refers to what we ought to do all things considered or overall. Here is a less ambiguous formulation of (5): (10) We ought, all things considered, to block neither shaft. We do not get a paradox if we interpret the “oughts” in (5), (6), and (7) as referring to what we merely prima facie or pro tanto ought to do. It is not paradoxical to be in a situation that is such that as far as some considerations are concerned we ought to perform an action while as far as other considerations are concerned we ought to refrain from that action. So, in order to run into a paradox we also need to understand the “oughts” in (6) and (7) as referring to what we ought to do all things considered. On this reading, (6) and (7) are identi­ cal to: (11) If the miners are in shaft A, we ought, all things considered, to block shaft A. (12) If the miners are in shaft B, we ought, all things considered, to block shaft B.

A New Perspective

135

But (11) and (12) seem implausible. Let us focus on (11). Suppose that the miners are in shaft A and we block that shaft. Have we done what we ought to do, all things considered? The answer seems clear: By blocking shaft A without knowing where the miners were, we exposed the miners to an irresponsible risk. Therefore, we failed to do what we ought to have done.17 Similar claims apply to (12) and shaft B. If you are not yet convinced, it might be instructive to compare the Miners situation to the Oedipus case. The interesting feature of the Oedi­ pus case is that the ex ante judgements and the ex post judgements about Oedipus’s acts come apart. Ex ante, that is, at the moment of Oedipus’s acting, killing the man and marrying the woman seem to be right acts from Oedipus’s perspective, because Oedipus does not identify the man as his father and the woman as his mother. Ex post, that is, after becoming aware of all relevant facts, Oedipus judges his acts to be wrong. The acts only appeared to be right, he might say, while they were in fact wrong. This is very different in Miners. There, blocking neither shaft seems to be the right thing to do both from our ex ante and from our ex post perspective. Given that we do not know ex ante where the miners are, it seems wrong to block either shaft. We ought to block neither shaft. Now, if we block shaft A and thereafter become aware that this is where the miners were, it is still natural – from our ex post perspective – to think that our action, though it resulted in the survival of ten miners, was too risky and therefore wrong, all things considered. The same holds, mutatis mutandis, if we block shaft B and then learn that the miners are there. Now suppose that we block neither shaft and then learn that the min­ ers were in shaft A. The correct judgement still seems to be that our act, though it resulted in the death of one miner, was right, all things considered. The lesson is that Miners seems to be different from Oedipus in that, in the latter case, the evidence available to the agent has no bearing on what the agent ought to do all things considered, whereas in the former case our evidence determines what we ought to do all things considered. Even if we come to know all relevant facts, we will stick to our judgement that, all things considered, we ought to have blocked neither shaft and it would have been wrong to block either shaft. I conclude that (11) and (12) are wrong. The correct verdicts with respect to Miners seem to be: (13) We ought, all things considered, not to block shaft A, even if the miners happen to be there. (14) We ought, all things considered, not to block shaft B, even if the miners happen to be there. To sum up, (6) and (7) invoke a paradox only if we interpret them as (11) and (12). But we should dismiss (11) and (12). The correct verdicts

136

Intuitions About Perspective-Dependence

in Miners seem to be (13) and (14). Thus, Miners intuitively speaks in favour of subjectivism.18 The Auxiliary Argument These judgements receive further support from intuitions about blame­ worthiness. As we have seen in several chapters, many philosophers hold: (15) An agent is blameworthy for an action only if she ought, all things considered, not to have performed it. It would clearly be appropriate for us to feel guilty if we saved all miners by luck. And bystanders should blame us. So, given that we can save all miners only by luck: (16) We would be blameworthy for saving all miners. It follows from (15) and (16) that: (17) We ought, all things considered, not to save all miners. Since (17) entails that (11) and (12) are false, we have an additional argu­ ment for thinking that Miners does not lead to a paradox. (16) is highly plausible. However, I have repeatedly expressed doubts in this book concerning (15). Nonetheless, since many philosophers believe (15), I think the auxiliary argument was worth mentioning. The Subjective Consequentialist’s Licensing Strategy Kolodny and MacFarlane (2010, p. 118) say that (6) and (7) “naturally occur to one in the course of deliberation, and they seem perfectly accept­ able”. But as we have seen, this is not true if we identify (6) and (7) with (11) and (12). However, since “ought” is ambiguous, there are other propositions we can identify (6) and (7) with. The subjective consequen­ tialist will suggest an interpretation of (6) and (7) on which they indeed naturally occur to one in deliberating about what to do. A natural judgement to make is that it would be desirable from a moral point of view if we blocked the shaft the miners are in. We can express this by saying that we “ought” to block the shaft the miners are in, but this “ought” does not refer to what we ought, all things considered, to do. It should rather be understood as expressing an ideal, though not an ideal it would be permissible for us to strive for if this means that we try to save all miners. On a plausible reading of (6) and (7), the “ought” is not an “ought to do” but an “ought to be”.19 “We ought to save all miners” can most sensibly be understood as the value judgement that it would be

A New Perspective

137

best, on balance, if we saved all miners. On this interpretation, (6) and (7) are identical to: (18) If the miners are in shaft A, our blocking shaft A would be the over­ all best thing to happen. (19) If the miners are in shaft B, our blocking shaft B would be the over­ all best thing to happen. (18) and (19) seem to be true. Although we ought, all things considered, not to save all miners, because we could do so only by luck, it would be best if we saved all miners. Why do (6) and (7), understood in the sense of (18) and (19), naturally come to mind in deliberating about what to do? The reason is that it is natural to weigh the rationally expectable outcomes of one’s available acts in situations like Miners. Here is how we might reason: Blocking neither shaft would result in one dead miner. Can we avoid this? Blocking shaft A will have the best result, if shaft A is where the miners are. If they are in shaft B, shaft A ought not to be blocked. Blocking shaft B will have the best result if the miners are there. Otherwise, it would be better not to block shaft B, for the result would be ten dead miners. Unfortunately, we do not know where the miners are. Here we are deliberating about what we ought, all things considered, to do. In this context we weigh the consequences of our options in light of their probabilities. So, it seems true that (6) and (7) naturally occur to us in deliberating about what to do and that they seem perfectly acceptable – but only if we identify (6) and (7) with (18) and (19). This, the subjective consequentialist will argue, explains the intuitive appeal of (6) and (7). The bottom line is that Kolodny and MacFarlane are wrong in arguing against subjectivism on the basis of Miners. The subjective consequen­ tialist has a persuasive explanation of our considered moral judgements both concerning what we ought to do and concerning what ought to be the case.

7. Conclusion In this chapter, I have put forward the New Perspective Argument for subjective consequentialism. The argument is based on the New Perspec­ tive on the perspective-dependence of moral duties. Participants in the debate between objectivism and subjectivism about moral duties typi­ cally do not distinguish between different moral theories but argue as if the issue were independent of which moral theory you accept. This chap­ ter presented an alternative. The New Perspective leads consequentialists

138

Intuitions About Perspective-Dependence

to focus on disaster situations because these are paradigmatic cases for consequentialism. The duty involved in these cases seems to be subjective. And consequentialists should rely on that seeming.

Notes 1. See, for instance, Bykvist 2009, Björnsson and Finlay 2010, Kiesewetter 2011, Kolodny and MacFarlane 2010, p.  115, fn. 2, Prichard 1932, Ross 1939, pp.  146–167, Smith 2010, and Zimmerman 2006a, p.  578; 2006b, pp. 329–330; 2008, pp. 3–5; 2014. Notice that some of these texts are offi­ cially not concerned with moral duties but with what one has most reason to do or with what one ought to do, all things considered. This does not matter, however, if we understand act-consequentialism (primarily) as a theory of practical reason, cf. Introduction, Section 3. Some texts deal exclusively with consequentialism rather than with moral duty in general but none of these proposes an argument among the lines of the argument I put forward for subjective consequentialism in this chap­ ter. See, for instance, Feldman 2006, Howard-Snyder 1997, Jackson 1991, Moore 1912, pp. 98–101, Oddie and Menzies 1992, Parfit 1988, and Pers­ son 2008. 2. The following is inspired by Hare 1979. 3. See Railton 1984, and Hare’s levels of moral thinking (1981, Part 1). Notice that the decision procedure suggested by these authors is not primarily a device for implementing moral theories, unlike the kind of decision procedure that I have defended in Chapter 2, and I am not endorsing the distinction between a criterion of rightness and a decision procedure along the lines sug­ gested by these authors. My aim here is to illustrate the idea, not to assess it. 4. Maybe a better conceptualization is that, according to subjective consequen­ tialism, the good alone features in the substantial explanation of your duty whereas not only the good but also evidence features in the explanation that accommodates both substantial and formal considerations. I lack the space for investigating this issue further. At any rate, objective and subjective con­ sequentialists agree that the good features centrally in the explanation of your duty. 5. Examples of duties recognized by common-sense morality are provided in Kagan 1998, Chapters 3 and 4. Among other duties, Kagan mentions con­ straints against doing harm, lying, promise-breaking, and constraints from spe­ cial obligations, conventions, and duties to oneself. Ross’s prima facie duties are also a good indication of the contents of common-sense morality. Ross mentions a duty to keep your promises, a duty to right a previous wrong you have done, a duty to return services to those from whom you have accepted benefits, a duty to promote a maximum of aggregate good, and a duty not to harm others. See Skelton 2012 for the references. My list in the main text con­ tains what I take to be the most obvious duties recognized by common-sense morality together with the duty not to violate the rights of other people, which functions as a kind of placeholder for less salient contents of common-sense morality, like Ross’s duty to right a previous wrong or to return services. 6. Kagan 1998, p. 72. The locus classicus regarding constraints is Nozick 1974, Chapter 3. 7. That traditional consequentialism is more demanding than common-sense morality has often been observed, especially since Singer 1972. See, for instance, Kumar 1999, Mackie 1977, pp. 129–134, and Miller 2004.

A New Perspective

139

8. This is at least one reason why the famous case of Dr. Jill in Jackson 1991, pp.  462–463, which has been used in order to argue for subjective conse­ quentialism, is no optimal fit. 9. “I suggest that (at least) lying and promise-breaking can be morally right, when not to do them would allow a disaster we could have prevented to occur”, Hooker 2000, p. 131. 10. There is a complication that we may ignore for the purposes of this chapter: whether your duty to promote the good in disaster situations is best thought of as an overriding duty or as an exception clause that is built into the other rules. 11. If consequentialism is first and foremost a theory of practical reason rather than a moral theory, would it not be fruitful to adjudicate between objective and subjective consequentialism by letting our intuitions be guided by cases that involve duties that are paradigmatic for consequentialism understood as a theory of practical reason? This question suggests an alternative to the New Perspective. Unfortunately, I lack the space to explore this issue here. 12. Kolodny and MacFarlane 2010, pp. 115–116; cf. Parfit 1988, pp. 2–3. 13. For this kind of suggestion, see Persson 2008. 14. As for blameless wrongdoing, see Parfit 1984, pp. 31–35. 15. This section is based on Andrić 2013b. 16. Kolodny and MacFarlane (2010) provide two arguments against subjectiv­ ism. Their first argument says that “the loss of (6) and (7) is already a sig­ nificant cost” (p. 118, I have replaced “(2)” and “(3)” with “(6)” and “(7)”). Second, they put forward a disagreement-based argument (pp. 119f.). I am only concerned with their first argument here since my aim is to show that Miners does not support the conclusion that Kolodny and MacFarlane seek to establish. For a defence of subjectivism against the disagreement-based argument, see Kiesewetter (2011). 17. Kolodny and MacFarlane could wonder at this point as well as at several other passages throughout this section whether my argument presupposes that modus ponens is valid for indicative conditionals. However, there are two reasons why the assumption that modus ponens is valid is not a prob­ lem for my argument. First, since Kolodny and MacFarlane want Miners to motivate their case against modus ponens in the first place, I am within my dialectical rights to rely on modus ponens. Kolodny and MacFarlane certainly do not themselves assume that modus ponens is invalid when they discuss Miners. Rather, they try to show that the most plausible judgments concerning Miners commit us to the rejection of modus ponens. Second, this section does not aim to contribute to the discussion of Kolodny and Mac­ Farlane’s reasons for the rejection of modus ponens but to the debate over the perspective-dependence of moral duties. In this context, it seems legiti­ mate to assume, along with the majority of the people engaged in the debate over perspective-dependence, that modus ponens is valid and to ask, on this assumption, whether Kolodny and MacFarlane are right when they dismiss subjectivism with respect to Miners. 18. Interestingly, Parfit spoke out in favour of subjectivism when he discussed Miners: “I have claimed that, when we are deciding what to do, we should ask what is subjectively right. In most contexts, this is what ‘right’ means. Similarly, when we are assigning blame, we should be concerned with sub­ jective wrongness” (Parfit 1988, p. 4). So, Kolodny and MacFarlane depart from Parfit’s verdict about the case. 19. For an examination of the difference between “ought to do” and “ought to be”, see Schroeder 2011.

9

Objective Consequentialism’s Licensing Dilemma1

In this chapter, I make good on the previous chapter’s claim that objective consequentialist’s licensing strategies concerning the Miners case fail. In Section 1, I introduce a third version of consequentialism in addition to objective and subjective consequentialism. The reason for introducing a third view is that, as I shall argue, objective consequentialism is in danger of collapsing into the third view in the case of some licensing strategies. In Section 2, I present what I call the Licensing Dilemma. This chapter claims that objective consequentialism is on the horns of the Licensing Dilemma. In Sections 3–10, I substantiate this claim. Section 11 concludes.

1. Objective Consequentialism, Subjective Consequentialism, and the Mixed View In this section, I introduce a rival to objective and subjective consequen­ tialism. According to objective consequentialism, to repeat, an act is right if and only if it leads to an outcome that has at least as much value as the outcome any other act would lead to, wrong if and only if it is not right, and obligatory if and only if it leads to an outcome that has more value than any outcome the act’s alternatives would lead to. According to subjective consequentialism, an act is right if and only if it has at least as much expected value as any other act, wrong if and only if it has less expected value than at least one of its alternatives, and obligatory if and only if it has more expected value than any of its alternatives. There is an alternative to objective and subjective consequentialism that I call the mixed view. The mixed view consists of the claim that both objective and subjective consequentialism are false plus the conjunction of what I call weak subjective consequentialism and weak objective consequentialism: Weak Subjective Consequentialism An act is subjectively right if and only if it has at least as much expected value as any other act, subjectively wrong if and only if it has less expected value than at least one of its alternatives, and

The Licensing Dilemma

141

subjectively obligatory if and only if it has more expected value than any of its alternatives. Weak Objective Consequentialism An act is objectively right if and only if it leads to an outcome that has at least as much value as the outcome any other act would lead to, objectively wrong if and only if it is not right, and objectively obligatory if and only if it leads to an outcome that has more value than any outcome the act’s alternatives would lead to. Weak subjective and weak objective consequentialism are derivatives of subjective and objective consequentialism. Contrary to the original posi­ tions, the derivatives are consistent with one another because they do not judge acts to be right or wrong simpliciter, but subjectively or objectively right or wrong, respectively. Proponents of the mixed view can there­ fore endorse weak subjective and weak objective consequentialism at the same time without contradicting themselves.

2. The Licensing Dilemma This section presents the Licensing Dilemma, which befalls objective con­ sequentialists who try to apply licensing strategies to Miners. Recall that in Miners, which I have presented in Chapter 8, Section 5, our evidence suggests that the miners are trapped either in shaft A (probability 0.5) or in shaft B (probability also 0.5). If we block the shaft the miners are in, we save them all. If we block the wrong shaft, they all die. If we block neither shaft, we save nine of them. According to objective consequentialism, the only right thing for us to do is to block the shaft the miners are in. But there is a problem. If we tried to act rightly according to objective consequentialism, we would risk the deaths of nine miners. Intuitively, our risking nine deaths seems to be wrong. Of course, according to objective consequentialism it is not right for us to risk their deaths – it is right for us to save all ten miners. However, since we cannot decide to save all ten miners (under that description of the deci­ sion’s content), we are left clueless. Thus, objective consequentialists have to somehow supplement their position in order to guide our action. And they had better do so without recommending our trying to save all miners. In contrast to objective consequentialism, subjective consequentialism obviously succeeds in guiding our action in an appropriate way. Blocking neither shaft has a higher expected value than blocking shaft A and than blocking shaft B. Therefore, according to subjective consequentialism, it is right for us to block neither shaft. Notice that, as I have argued in Chapter 8, Section 6, the subjective con­ sequentialist output in Miners is quite appealing. I suppose that objective

142

Intuitions About Perspective-Dependence

consequentialists will want to supplement their theory so that it does not only guide our action but also does so appropriately. Appropriate actionguidance in Miners means that objective consequentialism will somehow license our blocking neither shaft. I think that objective consequentialism has to license our blocking neither shaft because that action intuitively seems to be right. According to objective consequentialism, blocking neither shaft is the wrong thing for us to do. Therefore, “licensing” the action that produces most expected value has to mean something different for objective conse­ quentialism than judging it to be right. How can a proponent of objective consequentialism spell out the term “licensing”? There are many different ideas about how objective consequentialism can license the action that has most expected value. What I want to show in the following sections is that all the ideas exhibit one of two serious faults: Either the licensing is too weak, that is, our action is not guided towards blocking neither shaft; or the guidance of our action towards blocking neither shaft is, in some respect or other, ad hoc. It is beyond the scope of this chapter to present a deductive argument for the inescapability of the Licensing Dilemma. Rather, I will consider eight strategies that spell out the idea of licensing our blocking of neither shaft and show for each of them that either the problem of appropriate action-guidance remains unsolved or the solution is ad hoc. I will begin with five strategies that license the blocking of neither shaft by focusing not on this act but on some other iudicanda (Sections 3–7). Then I will consider three strategies that license the act of blocking neither shaft by assigning a positive status to this act, albeit not the status of being right (Sections 8–10).

3. “We Ought to Decide to Block Neither Shaft” A very straightforward way for the objective consequentialist to license our blocking of neither shaft consists in giving us the following piece of advice: “While you ought to block the shaft the miners are in, I recognize that you do not know which shaft that is. Neither do I. In this situation of uncertainty, you ought to decide to block neither shaft.” Thus, the objective consequentialist distinguishes between the right action on the one hand and the right decision on the other hand. This distinction is supposed to enable the objective consequentialist to provide us with appropriate action-guidance.2 Distinguishing right actions from right decisions commits the objec­ tive consequentialist to the acceptance of two criteria of rightness, one criterion for actions, the other for decisions. In order to provide appro­ priate action-guidance for us, the objective consequentialist’s criterion of rightness for decisions has to be a subjective criterion of rightness. That is, according to the objective consequentialist’s criterion of rightness for

The Licensing Dilemma

143

decisions, a decision is right if and only if it maximizes expected value. If the objective consequentialist endorsed an objective criterion of right­ ness for decisions, then we would have to decide to block the shaft with the miners. Adopting an objective criterion of rightness for decisions would, regarding the problem of appropriate action-guidance, not be an improvement of objective consequentialism compared to just holding an objective criterion of rightness for actions. Let me now state my criticism. If you already accept the objective cri­ terion for actions, there seems to be nothing apart from overcoming the problem of appropriate action-guidance that speaks in favour of adopt­ ing a subjective rather than an objective criterion for decisions. Thus, the move seems ad hoc. The charge of ad hocness can be backed by the following two con­ siderations. The first consideration is that, in everyday moral discourse, it seems to be generally accepted that either you ought to decide to act rightly or your right decision guarantees your acting rightly. In our every­ day moral discourse, the rightness of an action and the rightness of a decision for that action cannot come apart. By contrast, according to the view under consideration, we ought to decide to act wrongly. This is completely at odds with everyday discourse. Secondly, notice that the view under consideration includes an objective criterion of rightness for actions as well as a subjective criterion of right­ ness for decisions. This is an unsatisfactory compromise because whatever the objective consequentialist dislikes about the subjective criterion of rightness for actions also speaks against the subjective criterion of right­ ness for decisions. And, accordingly, whatever the objective consequential­ ist likes about her objective criterion of rightness for actions also seems to speak in favour of the objective criterion of rightness for decisions. Here is an example of what the objective consequentialist dislikes about the subjective criterion of rightness for actions. Subjective con­ sequentialism ennobles the misleading evidence that gives rise to wrong beliefs about what will produce most value. For example, subjective con­ sequentialism sanctions the misleading evidence that gives rise to Oedi­ pus’s wrong beliefs. (We, by contrast, only have limited knowledge rather than misleading evidence in Miners.) This problem speaks against the subjective criterion for decisions, too, because the rightness of a decision can also be a function of misleading evidence. To sum up, the objective consequentialist adopting a subjective crite­ rion of rightness for decisions in order to license our blocking neither shaft leads to a highly unattractive ad hoc position. The combination of the subjective criterion for decisions with the objective criterion for actions even seems to be incoherent insofar as what counts in favour of the objective criterion for actions speaks against the subjective crite­ rion for decisions. (The same is true, mutatis mutandis, for what counts against the subjective criterion for actions and the objective criterion for

144

Intuitions About Perspective-Dependence

decisions.) Thus, the combination of objective consequentialism with a subjective criterion of rightness for decisions sits on the second horn of the Licensing Dilemma.

4. “Our Decision Procedure Will Tell Us to Block Neither Shaft” Instead of adopting a subjective criterion of rightness for decisions, the objective consequentialist may want to stick to her objective criterion of rightness for actions exclusively. The idea now is to somehow license our decision to block neither shaft without judging the decision to be right. This is supposed to be achieved by the recommendation of a decision procedure that does not qualify as a criterion of rightness for decisions. The decision procedure tells us to block neither shaft so that (under nor­ mal circumstances) we will decide to block neither shaft and put this decision into action. At the same time, objective consequentialists can either be silent on our decision to block neither shaft or even judge it to be wrong. Either way, objective consequentialism implies that it is right for us to block the shaft with the miners and objective consequentialism does not imply that it is right for us to decide to block neither shaft.3 I consider two ideas to ground the distinction between the criterion of rightness for actions and a decision procedure. According to the first idea, the distinction between the criterion of rightness and a decision procedure can be based on general observations. According to the other, having a decision procedure can be justified on objective-consequentialist grounds. I will consider the first idea in this section and the second in the next one. Can the distinction between the criterion of rightness and a decision procedure convincingly be based on general observations? As I have explained in Chapter 1, all plausible moral theories must be combined with a decision procedure in order to be action-guiding for moral agents with cognitive limitations like the ones human beings exhibit. In order to be capable of acting rightly according to sufficiently complex moral theo­ ries, human moral agents must have decision procedures that enable them to match their available action tokens with the allegedly right action types. Thus, it seems that the distinction between the criterion of rightness and a decision procedure can be grounded on general observations. But this distinction between the criterion of rightness and a decision procedure does not help the objective consequentialist in Miners. Firstly, I have argued in Chapter 2 that DP is the only appropriate decision proce­ dure for all moral theories. DP, however, simply allows us to pick among the shafts. Secondly, even if you are not convinced by my arguments for DP, I think that any other at least halfway plausible account of a deci­ sion procedure will not help the objective consequentialist either. For the question is not whether we ought to have a decision procedure in order

The Licensing Dilemma

145

to enable us to perform the action that is right according to objective consequentialism. Such a decision procedure will not be of help for us when it comes to saving the miners. As things stand, we need a deci­ sion procedure that tells us to do the wrong thing by blocking neither shaft. Such a decision procedure goes beyond the decision procedure that allows you to match available action tokens with right action types. It cannot be justified in the same way. With the solution at issue, objective consequentialism thus sits on the first horn of the Licensing Dilemma: It fails to guide our action in an appropriate way.

5. “It Is Right for Us to Have a Decision Procedure That Tells Us to Block Neither Shaft” Let us now turn to the idea that, according to objective consequentialism, it is right for us to have a decision procedure that tells us to block neither shaft. For this idea to work, adopting and maintaining a decision proce­ dure have to qualify as actions. For actions are the iudicanda of objective consequentialism and the objective consequentialist wants to judge on purely objective-consequentialist grounds that we ought to have – adopt or maintain – a decision procedure that tells, and thus leads, us to block neither shaft.4 This idea, however, does not work. If having a decision procedure counts as an action that is right according to objective consequentialism, then having a decision procedure must maximize actual value. What kind of decision procedure would maximize actual value? It seems that objec­ tive consequentialism can only recommend having a decision procedure that leads the agent to perform exclusively actions that maximize actual value. This means for us that we ought to have a decision procedure that results in our blocking the shaft the miners are in. But we cannot decide to adopt the decision procedure that leads us to maximize actual value (under that description of the decision’s content), just as we cannot decide to block the shaft with the miners (under that description of the decision’s content). Therefore, we are not appropriately action-guided and objective consequentialism faces the first horn of the Licensing Dilemma. But perhaps I have been too fast. An objective consequentialist could object that – under normal circumstances – there may be no decision procedure available to us that would lead us to block the shaft with the miners. Furthermore, the best available decision procedure may well be one that would lead us to block neither shaft. Such a decision proce­ dure would be best under normal circumstances, because we would on the whole realize more value than we would if we had any other available decision procedure. I do not buy the objection. For I think a decision procedure is avail­ able that would lead us to perform all and only acts that maximize actual value. It is a decision procedure that simply leads to all and only those

146

Intuitions About Perspective-Dependence

decisions that result in us maximizing actual value, whatever contents the decisions may have. It is true that this decision procedure is not available in the sense that we could choose it under the description “decision pro­ cedure that leads us to maximize actual value”. But it is also true that we cannot choose to block the shaft with the miners under that description. The objective consequentialist nevertheless holds that blocking the shaft the miners are in is the right action for us to perform. Since the objective consequentialist thereby presupposes that this action is available to us, the objective consequentialist would have to justify why she wants to employ the notion of availability in another sense when it comes to the action of adopting and maintaining the aforementioned decision proce­ dure. I see no such justification. The objective consequentialist could reply that the sequence of all decisions that lead an agent to maximize actual value on every occasion does not qualify as a decision procedure. Arguably, a decision procedure is something that can be used by an agent, like an instrument, but the sequence of all decisions that result in maximizing actual value cannot be chosen and cannot be used in order to maximize actual value. The crucial point seems to be that the agent does not derive the decisions in question by using a unified procedure, like applying a set of rules of thumb, and that the sequence of decisions therefore does not constitute a decision procedure. Moreover, the best (genuine) decision procedure that can be used by us arguably is the one that tells, and thus leads, us to maximize expected value. Therefore, we ought to have a decision procedure that tells us to block neither shaft. But this objection does not work either. Let us assume that the best avail­ able decision procedure for us is one that leads us to maximize expected value. Then it simply is not true that we, according to objective consequen­ tialism, ought to have the best available decision procedure. For this deci­ sion procedure would lead us to perform actions that, according to objective consequentialism, are wrong, such as blocking neither shaft rather than the shaft with the miners. We would realize more value without having any decision procedure at all and by just acting rightly on every occasion. There is another reply available to the objective consequentialist. The reply’s starting point is the claim that the best (genuine) decision proce­ dure is one that leads the agent to maximize expected value. However, this does not take the objective consequentialist to the licensing of our blocking neither shaft. The objective consequentialist needs further prem­ ises for a full-blown argument. Here are the best that come to my mind. The first premise rests on the claim that, even though we can maximize actual value in every choice situation, we, as a matter of fact, often would not do so if we tried. Even as devoted consequentialists, without any motivational shortcomings, we would, due to our cognitive limits, often perform wrong actions – such as, out of ignorance, not choosing the right shaft. The first premise, then, says that:

The Licensing Dilemma

147

(1) We would, in the long run, realize more value if we had a decision procedure that would lead us to maximize expected value than if we had any other decision procedure or no decision procedure at all. Now, the objective consequentialist needs a further premise that says something about the normative significance of the first premise. For she has not established yet that, according to objective consequentialism, it is right for us to have the best decision procedure. Establishing this is challenging because one could argue that, according to objective con­ sequentialism, we ought to maximize actual value on all occasions, and since our doing so is incompatible with our having a decision procedure that effectively leads us to maximize expected rather than actual value, it is wrong for us to have such a decision procedure. Thus, the objective consequentialist needs a normative premise along the following lines: (2) Whether an act is right, wrong, or obligatory according to objec­ tive consequentialism does not depend on the consequences an agent could bring about if she performed the act, but on the consequences the agent would bring about if she performed the act. It follows from (1) and (2) that according to objective consequentialism, we ought to have a decision procedure that would lead us to maximize expected value. To illustrate this proposal, let us concentrate on a simplified scenario. Suppose there are two rescue groups, 1 and 2. The members of group 1 have a decision procedure that leads them to always try to maximize actual value whereas the members of group 2 have a decision procedure that leads them always to try to maximize expected value. Now sup­ pose that each group faces four Miners-like situations, one after another. Under normal circumstances, group 1 would save only 20 miners whereas group 2 would save 36 miners, see Figure 9.1. Therefore, so the argument goes, the objective consequentialist can recommend the expected-value decision procedure over the actual-value decision procedure. And similar claims apply to the comparison of the expected-value decision procedure and any other decision procedure or no decision procedure at all. There are various problematic aspects about this line of thought that cannot be discussed in this chapter.5 One could wonder, for example, whether the justified decision procedure would be of any help at all for us when it comes to Miners. After all, we could call into question what our decision procedure suggests to us and resist the suggestion when it becomes obvious that maximizing expected value will not maximize actual value. However, I will confine myself to just one point and assume, for the sake of the argument, that the best decision procedure would effectively lead us to block neither shaft. Let us assume that our behav­ iour has no long-term consequences and that our rescue mission is the

148

Intuitions About Perspective-Dependence

40

30

20

10

0 Start of Mission 1

Mission 1 Completed

Mission 2 Completed

Mission 3 Completed

Mission 4 Completed

Miners group 1 could save Miners group 2 would save Miners group 1 would save

Figure 9.1 Rescue missions.

last thing that we do in our life. This assumption renders the first premise false because it is no longer true that there are long-term profits associ­ ated with having a decision procedure that leads us to maximize expected value. To see in detail how the assumption that the rescue mission is our last action undermines the objective consequentialist’s argument, we can distinguish two cases. Either we have already adopted a decision pro­ cedure that leads us to maximize expected value, or we do not yet have such a decision procedure. In the former case, the question is whether, according to objective consequentialism, it is right for us to maintain the decision procedure. In the latter case, the question is whether, according to objective consequentialism, it is right for us to adopt the decision pro­ cedure. Since there will not be any choice situations in our life after the rescue mission, the answers to both questions depend solely on whether we would choose the right or the wrong shaft if we tried to save all min­ ers. In case we would choose the right shaft, it would not be true that we ought to maintain or adopt, respectively, the decision procedure. But, of course, we do not know whether we would choose the right or the wrong shaft. Therefore, we do not know whether we ought to have (adopt or maintain, respectively) a decision procedure that leads us to block neither shaft. Our action is not guided at all and thus not appropriately guided. Objective consequentialism is still left on the first horn of the Licensing Dilemma.

The Licensing Dilemma

149

6. “Our Blocking Neither Shaft Is Blameless Wrongdoing” Instead of distinguishing between the criterion of rightness and a decision procedure, the objective consequentialist could focus more on our char­ acter and present the following idea, which is based on a thought that is known under the label “blameless wrongdoing”. Under normal circum­ stances, so the objective consequentialist could say, it would be best for us to be motivated to maximize expected value and thus to block neither shaft, because having such motives would, under normal circumstances, realize most actual value. Moreover, having these motives would be most conducive to value even if it would result in our acting wrongly. Thus, our blocking neither shaft in virtue of our good motives is, on balance, desirable from a consequentialist point of view. How can the idea of blameless wrongdoing be justified? Imagine we were a rescue team that is willing to risk the lives of the miners just for achieving perfect results. Even if we assume that we always succeed in performing actions that maximize actual value and even if Miners-like cases actually do not occur, our fellow human beings would be scared to face an emergency situation if we are the rescue team. This would make outcomes worse according to any plausible theory of value. We would be more conducive to value by being different persons, who take less risk.6 Notice that the idea to license our blocking neither shaft on grounds of blameless wrongdoing has three attractive features. These features make the idea more promising than the licensing attempt that is based on the distinction between the criterion of rightness and a decision procedure. At least, this seems to be the case in the light of the criticisms provided in the previous sections. First of all, it is perfectly in line with the consequentialist’s commitment to the promotion of value to license our blocking neither shaft insofar as it results from our having the best motives. If we are, on balance, more conducive to value by having certain motives that lead us to act wrongly than by having other motives that lead us to act rightly, then, from a consequentialist perspective, we ought to have the former motives, even if they result in our acting wrongly. By contrast, our having a decision procedure is desirable, from a consequentialist perspective, only insofar as it results in our promoting value through actions. The second feature is related. It also concerns the way in which having certain motives can promote actual value. Having a decision procedure is conducive to value just insofar as the actions it results in are conducive to value. Therefore, a decision procedure can only be recommended on consequentialist grounds for persons who will perform at least one future action. In contrast, having certain motives can be conducive to value in other ways than by being mediated through actions. Thus, we could be most conducive to value by being motivated to block neither shaft, even if our motives will not lead us to perform actions that promote value on future occasions and even if we will not act again at all. For this reason, I

150

Intuitions About Perspective-Dependence

cannot simply rephrase the scenario I invented at the end of the previous section. Thirdly, while a decision procedure can immediately be adopted or dropped, it takes time to develop or lose motives. Thus, if Miners is our last rescue mission and we have already developed the motive to maxi­ mize expected value, we may not be able to drop this motive. This is another reason why I cannot outline a situation similar to the one that I introduced in order to argue against the action of having a decision pro­ cedure that leads us to maximize expected value. Despite its promising features, the idea to license our blocking neither shaft on grounds of blameless wrongdoing does not work – and here is why. The idea provides no solution to the problem of appropriate actionguidance. The crucial point is that it might not always be possible or it might not always be best for us to have the motive to block neither shaft. We can illustrate or modify Miners to the effect that it is not possible or not best, respectively, for us to have the motive to block neither shaft. As a result, we will not be appropriately action-guided and objective conse­ quentialism faces the first horn of the Licensing Dilemma. For a case where it is not possible for us to have the motive to block neither shaft, imagine that we fail to have this motive by the time we have to decide on what to do about the miners. We cannot at that time choose to have the motive because developing motives takes some time. It could be that we, as a first step, have to go to psychiatrists in order to cure our obsession with actions that maximize actual value. This would take some time. But, of course, we have to decide now because otherwise the flood waters will flood both shafts. Thus, we are left without action-guidance. One can imagine several cases where it is not best for us to have the motive to block neither shaft. Consider the possibility that we are very skilful or just lucky when it comes to hiding our motive to maximize actual value. In that case, our commitment to maximizing actual value would not give rise to mistrust. Arguably, the best consequentialist theory of value says that the outcome would on balance be best if we maxi­ mized actual value and people think that we would not do so by risk­ ing their lives. In this case, we would again be left without appropriate action-guidance. Of course, one can also modify Miners in ways that are more fanciful. Perhaps a demon or a mad scientist somehow caused our motive to have no consequences apart from those that are mediated by actions. Then our blocking neither shaft simply would again not be blameless but blame­ worthy wrongdoing. At the bottom line, the objective consequentialist cannot always license our blocking neither shaft with the idea of blameless wrongdoing. Cou­ pled with this licensing manoeuvre, objective consequentialism is left on the first horn of the Licensing Dilemma.

The Licensing Dilemma

151

7. “We Are Not Blameworthy If We Block Neither Shaft” The phrase “blameless wrongdoing” is inappropriate, because the issue is not whether an agent is blameworthy or to be blamed for wrong actions but rather whether an agent will sometimes realize more value on the whole if she has a motive that leads her to not maximize actual value. In contrast, the view under consideration in this section really is about whether an agent is blameworthy for wrong actions. You can sometimes read that, while subjective consequentialism is the appropriate basis for the assess­ ment of agents, objective consequentialism is the appropriate theory for the assessment of actions.7 Agents should be assessed, so the suggestion goes, as praiseworthy or blameworthy depending on whether they act rightly or wrongly according to subjective consequentialism. Thus, if we block nei­ ther shaft, we are not blameworthy, because we maximize expected value. It is in this sense that our blocking neither shaft is supposed to be licensed. The aforementioned considerations, in my view, are fallacious. For one thing, it is doubtful whether agents, according to consequentialism, should be assessed in the alleged way. Furthermore, the proposed assess­ ment would not solve the problem of appropriate action-guidance. The alleged standard for assessing agents is doubtful because the right­ ness of praising or blaming an agent may have nothing to do, according to consequentialism, with whether the agent has acted rightly or wrongly. Arguably, on consequentialism, the blameworthiness of agents is associ­ ated with social practices more than it is with wrongdoing. Or the con­ sequentialist might want to hold that calling an agent “praiseworthy” or “blameworthy” is just another way of saying that it is right to praise or blame the agent. Or she may want to dismiss the notions of praiseworthi­ ness and blameworthiness altogether. Any of these views is problematic for the licensing strategy under consideration. Maybe doubts concerning the proposed link between blameworthiness and praiseworthiness on the one side, and right and wrong actions on the other, arise in particular from the thought that blaming as well as praising an agent are actions and thus are to be judged either by their actual consequences (according to objective consequentialism) or by their expected consequences (according to subjective consequentialism).8 It might be right, for example, to praise someone who gave more money to the poor than his fellow citizens because the praising has the best (actual or expected) consequences, although the praised person did act wrongly (according to consequentialism), for she could have given more money than she actually did. It seems that this person is praiseworthy although she failed to maximize expected value. But even if we put aside this objection, the suggestion under consider­ ation does a poor job in Miners. Assume that we are blameworthy if we block neither shaft. So what? This has no relevance for our decision which shaft to block. We would have to block neither shaft if we were primarily

152

Intuitions About Perspective-Dependence

concerned with not being blameworthy. But that is not our concern. Instead, as moral agents we simply want to do the morally appropriate thing. Thus, the problem of appropriate action-guidance remains and the suggestion under consideration sits on the first horn of the Licensing Dilemma.

8. “We Ought to Block Neither Shaft” Let me now consider licensing strategies that assign a positive status to the act of blocking neither shaft, albeit not the status of being right. The boldest of these strategies consists in modifying objective consequentialism and distinguishing between, on the one hand, the action that it is right (or wrong) for an agent to perform and, on the other hand, the action that the agent ought (or ought not) to perform. On the basis of this distinction, Miners can be analysed as follows: It is right for us to block the shaft with the miners because blocking this shaft has the best consequences. At the same time, we ought to block neither shaft because blocking neither shaft maximizes expected value. Is the proposal convincing? I do not think so. There is nothing that counts in favour of distinguishing between rightness and oughtness in this context apart from overcoming the problem of appropriate actionguidance. The distinction therefore qualifies as an ad hoc solution to the problem of appropriate action-guidance. Two considerations back the accusation of ad hocness. The first con­ sideration can be expressed as a question: How can the proponent of objective consequentialism defend her subjective criterion of oughtness against an objective criterion of oughtness and at the same time defend her objective criterion of rightness against a subjective criterion of right­ ness? It seems to me that all that can be said in favour of a subjective cri­ terion of oughtness undermines the arguments in favour of an objective criterion of rightness, and vice versa. Secondly, there is a close conceptual link in everyday language between “being right” and “ought to be done”. This link seems to consist of two tenets: (1) if an action is right, then it is not the case that the agent ought not to perform it; (2) if an agent ought to perform an action, then that action is right.9 According to the view under consideration, we ought to act wrongly. This is clearly at odds with the second tenet. My conclusion is that it is not reasonable to combine an objective cri­ terion of rightness with a subjective criterion of oughtness. The combina­ tion seems ad hoc. Objective consequentialism is left on the second horn of the Licensing Dilemma.

9. “It Is Subjectively Right for Us to Block Neither Shaft” In Section 1, I introduced weak subjective consequentialism. According to weak subjective consequentialism, an act is subjectively right if and

The Licensing Dilemma

153

only if it maximizes expected value. The view under consideration in this section is the conjunction of objective consequentialism and weak subjec­ tive consequentialism. According to this view, we ought to block the shaft with the miners and we ought subjectively to block neither shaft. Is the distinction between subjective rightness and rightness helpful for us? It depends. According to objective consequentialism combined with weak subjective consequentialism, we ought subjectively to block neither shaft while we ought to block the shaft with the miners. But what does this mean for our decisions? Is it more important to act rightly than to act subjectively rightly? Are we licensed to act subjectively rightly just in case we do not know what we ought to do? Or can we ignore rightness altogether and base our decisions on subjective rightness alone? Let us consider the different possibilities one by one and start with the view that it is always most important to act rightly even if we do not know which action is the right one. This interpretation is faced with the first horn of the Licensing Dilemma. Since we do not know where the miners are, the problem of appropriate action-guidance remains unsolved. Here is a second extreme interpretation. It says that, whenever you cannot identify the right action, it is more important to act subjectively rightly than to act rightly. This interpretation, in my view, does not qual­ ify as a version of objective consequentialism. It rather seems to be a kind of mixed view, combining a weak subjective criterion of rightness for the sake of action-guidance with an objective criterion of rightness. Between the two interpretations, there is a lot of middle ground, for there are different amounts of weight that we can attach to acting sub­ jectively rightly on the one hand and to acting rightly on the other hand. In Miners, it seems reasonable to care more about not bringing about the death of ten miners than about not acting wrongly by blocking neither shaft. Therefore, it seems reasonable for us to attach more weight to not acting subjectively wrongly. But in order to hold a middle-ground solu­ tion and not the mixed view, you have to assign some weight to not act­ ing wrongly. For instance, an objective consequentialist might propose that we had to flip a coin between blocking shaft A and blocking shaft B, given a modified Miners case in which blocking one of the shafts saves all miners and blocking the other has, for whatever reason, a negligible effect, killing only some of the miners, perhaps two or three. Is one of the middle-ground solutions convincing? I do not think so. The middle-ground solutions do not just include the objective and the subjective criteria of rightness, but a third criterion of rightness that deals with how to weigh the other two criteria in order to determine what it is right to do conclusively. This is confusing and the middle-ground solu­ tions seem to be theoretically overloaded. But the flaw just mentioned can be eradicated. The middle-ground solutions can just drop the subjective criterion of rightness because this criterion’s attractiveness resulted solely from its action-guiding function

154

Intuitions About Perspective-Dependence

that it does not fulfil anymore. Instead of weighing the subjective right­ ness (or wrongness) of an action against its wrongness (or rightness), the middle-ground solutions can now talk of weighing the expected value of an action against its wrongness (or rightness). The now vacant notion of the subjective criterion of rightness can thus be assigned to the criterion of rightness that tells one how to act in light of the expected value of one’s options and their rightness (or wrongness). Let us consider the restated middle-ground solutions. A middle-ground solution has to determine a threshold for when to maximize expected value and when to take one’s chance of acting rightly. How can this threshold be determined? It seems to me that intuition-pumping by means of consideration of different cases will not do. If you accept the basic commitment of consequentialism to the promotion of value, then there simply is no place for putting acting rightly on an equal footing with the promotion of value. Consider Miners. As consequentialists, we are concerned with the goodness of states of affairs. If, in addition, we were concerned with our acting rightly according to objective consequentialism for its own sake, then we would eventually approve losses or at least probable losses of value. Therefore, the middle-ground solutions do not match the very idea of consequentialism. Each of them is ad hoc. The middle-ground solu­ tions are on the second horn of the Licensing Dilemma.

10. “It Is Rational for Us to Block Neither Shaft” Here is the last proposal for objective consequentialism. It combines the objective criterion of rightness with considerations of instrumental ratio­ nality. Let us therefore call the proposal the rational-licence view. Accord­ ing to the rational-licence view, it is rational for us to block neither shaft, although doing so is wrong. There are two versions of the rational-licence view. They differ in their explanations of why it is rational for us to block neither shaft. The first version says that it is rational for us to block neither shaft because it is more rational to be certain to do a minor all-things-considered wrong than to risk a major all-things-considered wrong by trying to act all-things-considered rightly. According to the second version, it is rational for us to block neither shaft because this maximizes expected value.10 It seems to me that the second version is more attractive than the first one. You can see this by considering the following two questions: 1) In which circumstances does the risk of doing a major wrong make it rational for an agent to do a minor wrong? 2) Why is it rational to do a minor wrong in these circumstances?

The Licensing Dilemma

155

The answer to the first question is that it is rational to do a minor wrong if and only if the expected value of doing a minor wrong is at least as great as the expected value of trying to do the right thing. There are no reasonable alternatives to this answer. The salient alternative is to hold that doing the right thing has a value of its own. Thus, there would be cases in which it would be rational to try to do the right thing even if the trying had an expected value that is smaller than the expected value of doing a minor wrong. But this alterna­ tive position is not reasonable. For it would assign right actions the status of a fetish. Moral agents who followed this view would in the long run do less good than agents who always maximize expected value. Moreover, I cannot see how the proponent of the alternative position could determine the “rational weight” of acting rightly in a non-arbitrary way. Proponents of the rational-licence view should hold, then, that acting rightly does not carry a rational weight of its own and therefore it is rational to do a minor wrong if and only if the expected value of doing a minor wrong is at least as great as the expected value of trying to do the right thing. I have just argued for this claim. I needed to argue for it because the claim is not obvious. The first version of the rational-licence view is misleading in that it can lead you to think that acting rightly has a value of its own. This speaks against the first and in favour of the second version of the rational-licence view. Now consider the second question. Why is it in some circumstances rational to do a minor wrong instead of trying to act rightly? The answer is, of course, that its larger expected value makes it rational to do a minor wrong instead of trying to act rightly. Thus, my second argument against the first version of the rational-licence view is once more that this ver­ sion is misleading. It suggests that you can fully explain a moral agent’s doing a minor wrong in the circumstances under consideration with the fact that it is too risky to try to do the right thing. But the exhaus­ tive and ultimate explanation of the agent’s action is that it maximizes expected value. Thus, the second version of the rational-licence view is more appealing, for it allows for stronger explanations than the first one. I take these thoughts to establish that a proponent of the rationallicence view should accept the second version of this view according to which it is rational for us to block neither shaft because this maximizes expected value although it is wrong for us to block neither shaft. In what follows, I will critically assess the rational-licence view. In order to be distinguishable from the mixed view, the rational-licence view has to insist that calling our blocking neither shaft rational is some­ thing different from calling it subjectively right. But I think that there are good reasons to call our blocking neither shaft not only rational but also subjectively right. The refusal to do so makes the rational-licence view an ad hoc solution to the problem of appropriate action-guidance.

156

Intuitions About Perspective-Dependence

What speaks in favour of calling our blocking of neither shaft not only rational but also subjectively right? First of all, in everyday morality, the notion of rightness has, at least among other things and in some way, an action-guiding function. According to the rational-licence view, rational­ ity has an action-guiding function. According to the best formulation of the rational-licence view, the rightness or wrongness of our action has nothing to do with what it is rational for us to do and thus is not linked with action-guidance at all. Rather, rationality has to do only with the expected value of our available actions. This is at odds with everyday morality. In order to link rightness and action-guidance, one should call our blocking neither shaft not only rational but also subjectively right. Especially since, as cases such as Oedipus on the one hand and Miners on the other suggest, the everyday notion of rightness is ambiguous between objective and subjective rightness. Secondly, according to the rational-licence view blocking neither shaft is what we ought to do for moral considerations that, as one can argue, qualify as moral reasons of some kind. It seems strange not to call an action right, at least in some sense, if one ought to perform it for moral considerations. Thus, analyses of the concepts that are connected with rightness, rationality, and morality further strengthen the first point. Last but not least, moral intuitions suggest that our blocking of neither shaft is right. Correspondingly, blocking the shaft the miners are in intui­ tively seems to be too risky and therefore wrong. Thus, moral intuitions strengthen the case for calling our blocking neither shaft right. Is there something that speaks in favour of calling our blocking neither shaft rational but not subjectively right? Michael Smith offers an argu­ ment. He takes Frank Jackson’s remark that the “ought” of subjective con­ sequentialism is the “ought most immediately relevant to action” to mean that the “ought” of subjective consequentialism refers to what an agent can appropriately be held responsible for doing (Smith 2006, p.  141). Smith (2006, pp. 142–145) then claims that we cannot more legitimately expect agents to maximize expected instead of actual value than we can expect them to maximize value instead of value-as-they-see-it. This holds because some agents live in oppressive cultures and are therefore epis­ temically cut off from facts about what is of value. If an agent maximizes expected value-as-she-sees-it instead of expected value, then she is acting wrongly according to subjective consequentialism, albeit through no fault of her own. From this, Smith concludes that the concept of a right action must abstract from what we can legitimately hold agents responsible for. “It is therefore . . . a concept that incorporates two ideals – one on the evaluative side of things and the other on the non-evaluative side of things – that properly captures the concept of right action” (Smith 2006, p. 145). If Smith is right, then the subjective criterion of rightness is unstable because on the non-evaluative side of things it focuses on what one can expect an agent to believe while on the evaluative side it focuses on facts

The Licensing Dilemma

157

and abstracts from what one can expect an agent to believe. By contrast, the criterion of rationality focuses on both sides on what one can expect an agent to believe. So, you cannot substitute the talk of rationality by the talk of rightness. It is a sheer coincidence that our blocking neither shaft does not only maximize value-as-we-see-it but does also maximize expected value. There are various metaethical assumptions in Smith’s argument that one could disagree with. Sections 3 and 4 of Chapter 6 have presented answers to Smith’s worries. I will not repeat them here but instead focus on one crucial point: I do not think that the rationale for the subjective criterion of rightness consists in the belief that an action can be right only if the agent can be held responsible for not performing it. Rather, what speaks in favour of the subjective criterion of rightness is its provision of appropriate action-guidance by pointing to features of acts that seem to be right-making. So, it would not undermine the subjective consequen­ tialist criterion of rightness if it were true that agents sometimes cannot be held responsible for failing to recognize value. This is why Smith’s argument does not convince me. Perhaps the proponent of the rational-licence view will want to state another reason for her refusal to call the rational action the subjectively right one. It is argued by some philosophers that the mixed view faces some problems on its own.11 Therefore, it might not be rational for the objective consequentialist to switch to the mixed view. It is not the purpose of this chapter to argue for the mixed view. But notice that, even if the mixed view were untenable, then this would not mitigate the problem of the rational-licence view. For the refusal to adopt the subjective criterion of rightness just for the sake of eschewing the problems of the mixed view would still be ad hoc. I see no other argument that speaks in favour of the rational-licence view. The view fails because calling our blocking of neither shaft ratio­ nal while refusing to call it subjectively right is ad hoc. As a result, the rational-licence view sits on the second horn of the Licensing Dilemma.

11. Conclusion I have argued that objective consequentialism cannot escape the Licensing Dilemma. Either our action is not guided towards blocking neither shaft. Or the guidance of our action towards blocking neither shaft is ad hoc in some respect or other. Objective consequentialism cannot license the intuitively right thing in a persuasive way. This is the final defeat of objec­ tive consequentialism – and the winner is: subjective consequentialism.

Notes 1. This chapter is based on Andrić 2013a. Besides some minor corrections, the main difference is that in that paper I unfortunately focused on the case

158

2.

3. 4.

5.

6. 7. 8. 9.

10. 11.

Intuitions About Perspective-Dependence in Jackson 1991, pp.  462–463, when presenting objective consequential­ ism’s Licensing Dilemma. Now that I endorse the New Perspective on the perspective-dependence of moral duties, which I defend in the previous chap­ ter, I think that consequentialists should not let the Jackson case guide their intuitions about perspective-dependence. Fortunately, however, what I have said with respect to the Jackson case concerning objective consequentialist’s licensing strategies applies also to the Miners case. Persson (2008) distinguishes between an objective criterion for what we ought to do and a subjective criterion for what we ought to try. This seems to be just another formulation of the view under consideration. I think that the underlying idea becomes clear in this quotation: “We could roughly dis­ tinguish between the ‘internal’ and the ‘external’ side of an action which is intentional under some description. The internal side is what we decide, intend and try to do, while the external side is what we in fact do, intention­ ally or unintentionally, what changes we in fact cause in our attempt” (Pers­ son 2008, p. 351). I think that my argumentation in this section can easily be adapted to any distinction between an objective criterion for the external and a subjective criterion for the internal side of actions. The distinction between a criterion of rightness for actions and a decision procedure has been very popular since Bales 1971. The view under consideration in this section can be understood as based on the idea that, as Tännsjö (1995) formulates it, “if, consistently, we try to maximize expected utility, then, in the long run, probably, we end up with better results than consistently applying any conceivable alternative strat­ egy” (p. 589). (Tännsjö himself, though, defends the view that I consider in Section 10, cf. Tännsjö 1995, pp. 589–590. For a criticism of the ideas men­ tioned in Tännsjö 1995, see Bergström 1996, Section 6.) The position that an agent ought to perform an act if and only if it is an option such that what would happen if the agent performed it is better than what would happen if she did not perform it is called actualism. Actualism is defended, for example, in Jackson and Pargetter 1986. The main alternative to actualism is possibilism. Possibilism says that an agent ought to perform an act if and only if it is an option such that what could happen if the agent performed it is better than what could happen if she did not perform it, whereby “could” refers to what the agent can do (cf. Zimmerman 2008, Chapter 3.1). Concerning the issue of “blameless wrongdoing”, see Parfit 1984, pp. 31–35, and the related discussions concerning “global consequentialism” in Pettit and Smith 2000 or “evaluative focal points” in Kagan 2000. This idea has often been mentioned since Moore 1912, pp. 98–101. For straightforward consequentialist accounts of blaming and praising, see Norcross 2006, pp. 224–227, and, especially, Shaw 1999, pp. 134–138. Against the assertion of the first connection, one could hold that if an action is right, then it ought to be done. But this is not true. For there might be more than one right action and you can, and thus ought, only choose one of them. Moreover, according to common-sense morality, there are supererogatory actions, which are right but it is not the case that you ought to perform them. The first version is mentioned in Bykvist 2010, pp. 90–91; the second seems to be defended in Smith 2006, pp. 145–147. A prominent objection to the mixed view has been put forward in Zimmer­ man 2008, pp. 6–7.

Conclusion

The conclusion contains two sections. In Section 1, I summarize the main findings of this book. The main message of Section 1 is that subjective con­ sequentialism is superior to objective consequentialism. I also go one step further and allude to some considerations that speak in favour of subjec­ tive consequentialism being true. However, since these considerations are far from being conclusive (exploring them in depth would require at least a further book), I might be mistaken, and act-consequentialism might turn out to be false. If act-consequentialism were false, many insights from this book would still be interesting, as I hope to show in Section 2.

1. The Case for Subjective Consequentialism This book has defended the claim that subjective consequentialism is superior to objective consequentialism. Act-consequentialists should be subjectivists rather than objectivists about moral duty for a number of reasons, which fall into three categories. First, subjective consequentialism, combined with DP, is appropriately decision-guiding whereas the decision-guidance of objective consequentialism is a travesty, because objective consequentialism cannot take into account the moral signifi­ cance of uncertainty in satisfying ways (Chapters 1 and 2). The second category of reasons is related to Ought Implies Can. Moral theories are false if they are incompatible with Ought Implies Can (Chapter 3). Although both objective and subjective consequentialism are compatible with Ought Implies Can (Chapters 4 and 7), objective consequentialism is incompatible with four considerations underlying Ought Implies Can (Chapter 5) as well as with the plausible principle Ought Implies Evidence, which is supported by the same considerations that underlie Ought Implies Can (Chapter 6). The third category concerns case-based intu­ itions. Intuitions concerning relevant cases support subjective rather than objective consequentialism (Chapters 8 and 9). Arguing for the superiority of subjective over objective consequentialism is one thing. Arguing for the truth of subjective consequentialism is another. To show that subjective consequentialism is true, I would have to

160

Conclusion

show, on the one hand, that subjectivism is not only superior to objectiv­ ism but also to other positions regarding the perspective-dependence of moral duties, like the mixed view, relativism, and contextualism. While I have not done that, I think that some arguments against these posi­ tions can be developed on the basis of the material presented in this book – for example, the intuitions that I have defended with regard to the Miners case do not seem to fit well with these positions. Be that as it may, a full-blown defence of subjective consequentialism would also need to rebut objections against act-consequentialism. If the argu­ ments for subjective consequentialism are successful, then the objection that act-consequentialism is cognitively over-demanding fails. But other objections remain, for example, objections related to the phenomena of supererogation, special duties, and justice. I think that the most promis­ ing way of defending act-consequentialism against such objections con­ sists in defending act-consequentialism primarily as a theory of practical reason and only secondarily as a moral theory (cf. Introduction, Section 3). For then, to the extent that act-consequentialism contradicts wide­ spread moral intuitions, these contradictions are arguably not so much a problem for act-consequentialism as they are for the moral intuitions. Showing this, though, is a task for another occasion.

2. What If Act-Consequentialism Is False? If act-consequentialism, contrary to my expectations, turns out to be false, some findings of this book are still relevant. In this section, I list the most important findings that are independent of the truth of act-consequentialism. First of all, I have analysed the concept of decision-guidance and sug­ gested a decision procedure, DP, in combination with which moral theo­ ries are decision-guiding (Chapters 1 and 2). My arguments for DP are independent of act-consequentialism and DP could be combined with whatever turns out to be the best moral theory. The discussion of DP has revealed, as a side effect, that the best moral theory should take into account the significance of uncertainty. Many existing moral theories fail in this regard. We have to improve them. Secondly, I have argued for Ought Implies Can and presented a recipe for defending it in Chapter 3. The recipe had its litmus test when I used it for rebutting a sophisticated objection to Ought Implies Can. Chapter 3, too, remains fully intact if act-consequentialism is false. Thirdly, what I have said about the compatibility of act-consequentialism with Ought Implies Can, Ought Implies Evidence, and their common ratio­ nales in Chapters 4–6 can be taken as a basis for examining the compatibility of other moral theories with these considerations. Furthermore, the finding that Ought Implies Evidence is an often-overlooked plausible moral prin­ ciple is completely independent of the truth of act-consequentialism. I

Conclusion

161

have also argued that, first appearances notwithstanding, the view that “ought” implies not only factual but also normative “evidence” might after all not have counterintuitive implications. (Yet, I do not commit myself to the position that “ought” implies normative “evidence”. Per­ haps a crucial difference between factual and normative evidence can be sustained.) Fourthly, I hope that I have furthered our understanding of actionguidance in ways that are independent of whether act-consequentialism is true. Chapter 1 (esp. Section 4) has distinguished and delineated sev­ eral kinds of action-guidance. Moral theories are action-guiding in a trivial sense if they contain action-guiding terms like “ought” or “right”. Chapters 1 and 2 have introduced and examined the concept of decisionguidance. Chapters 3–7 have examined the kind of action-guidance that underlies Ought Implies Can and Ought Implies Evidence. Chapter 5, Section 2, has suggested a conceptual link between decision-guidance and the kind of action-guidance associated with Ought Implies Can and Ought Implies Evidence: in order to be action-guiding in the sense related to Ought Implies Can, a moral theory must be decision-guiding without an additional decision procedure for those of its followers whose beliefs are in accord with the factual evidence available to them. Fifthly, Chapter 7 has examined the compatibility of subjective conse­ quentialism with Ought Implies Can and subjective consequentialism’s deontic iudicanda. This chapter does not become worthless if act­ consequentialism is false either. For the compatibility with Ought Implies Can and the normative relations between decisions and actions are rel­ evant topics for any subjectivist moral theory, be it act-consequentialist or not. Finally, I have put forward a new perspective on the perspectivedependence of moral duties in Chapter 8. The validity of the New Per­ spective, a method in moral epistemology, is entirely independent of whether act-consequentialism is true. It can be used for examining the perspective-dependence of moral duties according to any moral theory. This can be fruitful in several ways, as I have explained in Chapter 8, Section 2.

References

Andrić, V. (2013a): “Objective Consequentialism and the Licensing Dilemma”, Philosophical Studies 162(3), pp. 547–566. ———. (2013b): “The Case of the Miners”, Journal of Ethics and Social Phi­ losophy 7(1), pp. 1–8, URL = . ———. (2016): “Is Objective Consequentialism Compatible with the Principle That ‘Ought’ Implies ‘Can’?”, Philosophia 44(1), pp. 63–77. ———. (2017): “Objective Consequentialism and the Rationales of ‘“Ought” Implies “Can”’”, Ratio 30(1), pp. 72–87. ———. (2019): “Consequentialism and Robust Goods”, Utilitas 31(3), pp. 334–342. Andrić, V. and A. Tanyi (2016): “Multidimensional Consequentialism and Risk”, Ethical Theory and Moral Practice 19(1), pp. 49–57. Anscombe, G.E.M. (1957): Intention, Oxford: Basil Blackwell. Austin, J.L. (1956a): “Ifs and Cans”, Proceedings of The British Academy 42, pp. 107–132. ———. (1956b): “A Plea for Excuses: The Presidential Address”, Proceedings of the Aristotelian Society. New Series 57, pp. 1–30. Ayer, A.J. (1954): “Freedom and Necessity”, in his Philosophical Essays, London: Macmillan, pp. 271–284. Baier, A. (1971): “The Search for Basic Actions”, American Philosophical Quar­ terly 8, pp. 161–170. Bales, R.E. (1971): “Act-Utilitarianism: Account of Right-Making Characteris­ tics or Decision-Making Procedure?”, American Philosophical Quarterly 8(3), pp. 247–265. Bentham, J. (1789/1961): An Introduction to the Principles of Morals and Legis­ lation, Garden City, NY: Doubleday. Bergström, L. (1996): “Reflections on Consequentialism”, Theoria 62, pp. 74–94. Bicchieri, C., Muldoon, R., and A. Sontuoso (2018): “Social Norms”, in: E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2018 Edition), URL = . Björnsson, G. and S. Finlay (2010): “Metaethical Contextualism Defended”, Eth­ ics 121(1): 7–36. Blum, A. (2000): “The Kantian versus Frankfurt”, Analysis 60, pp. 287–288. Bratman, M. (1987): Intentions, Plans, and Practical Reason, Stanford, CA: CSLI. Broad, C.D. (1952/53): Ethics, edited by C. Lewy, Dordrecht: Martinus Nijhoff, 1985.

References

163

Bykvist, K. (2009): “Objective versus Subjective Moral Oughts”, in: L.-G. Johansson, R. Sliwiński, and J. Österberg (eds.), Logic, Ethics, and All That Jazz: Essays in Honour of Jordan Howard Sobel, Uppsala: Uppsala University, pp. 39–65. ———. (2010): Utilitarianism: A Guide for the Perplexed, London: Continuum. Carlson, E. (1999): “The Oughts and Cans of Objective Consequentialism”, Util­ itas 11(1), pp. 91–96. ———. (2002): “Deliberation, Foreknowledge, and Morality as a Guide to Action”, Erkenntnis 57, pp. 71–89. Chisholm, R.M. (1964): “The Descriptive Element in the Concept of Action”, The Journal of Philosophy 61, pp. 613–624. Copp, D. (2003): “‘Ought’ Implies ‘Can’, Blameworthiness, and the Principle of Alternate Possibilities”, in: M. McKenna and D. Widerker (eds.), Moral Responsibility and Alternative Possibilities: Essays on the Importance of Alter­ native Possibilities, Burlington, VT: Ashgate, pp. 265–299. Crisp, R. (1992): “Utilitarianism and the Life of Virtue”, The Philosophical Quarterly 42, pp. 139–160. ———. (2006): Reasons and the Good, Oxford: Oxford University Press. Dancy, J. (2000): Practical Reality, Oxford: Oxford University Press. Danto,A.C. (1963): “What We Can Do”, The Journal of Philosophy 60, pp. 435–445. Davidson, D. (1963): “Actions, Reasons and Causes”, The Journal of Philosophy, pp. 685–700. Dreier, J. (1993): “The Structure of Normative Theories”, The Monist 76, pp. 22–40. Fantl, J. (2017): “Knowledge How”, in: E.N. Zalta (ed.), The Stanford Ency­ clopedia of Philosophy (Fall 2017 Edition), URL = . Fara, M. (2008): “Masked Abilities and Compatibilism”, Mind 117, pp. 843–865. Fehige, C. (2004): Soll ich?, Stuttgart: Reclam. Fehige, C. and R.H. Frank (2010): “Feeling Our Way to the Common Good: Utilitarianism and the Moral Sentiments”, The Monist 93(1), pp. 141–165. Feldman, F. (1997): “Adjusting Utility for Justice: A Consequentialist Reply to the Objection from Justice”, in his Utilitarianism, Hedonism, and Desert, Cam­ bridge: Cambridge University Press, pp. 154–174. ———. (2006): “Actual Utility, the Objection from Impracticality, and the Move to Expected Utility”, Philosophical Studies 129(1), pp. 49–79. ———. (2012): “True and Useful: On the Structure of a Two Level Normative Theory”, Utilitas 24(2), pp. 151–171. Frazier, R.L. (1994): “Act Utilitarianism and Decision Procedures”, Utilitas 6(1), pp. 43–53. Gesang, B. (2003): Eine Verteidigung des Utilitarismus, Stuttgart: Reclam. Gill, M.B. (2007): “Moral Rationalism vs. Moral Sentimentalism: Is Morality More Like Math or Beauty?”, Philosophy Compass 2(1), pp. 16–30. Goldman, A.I. (1970): A Theory of Human Action, Princeton, NJ: Princeton Uni­ versity Press. Goldman, H.S. (1976): “Dated Rightness and Moral Imperfection”, Philosophi­ cal Review 85(4), pp. 449–487. ———. (1978): “Doing the Best One Can”, in: A.I. Goldman and J. Kim (eds.), Values and Morals, Dordrecht: Reidel.

164

References

Graham, P.A. (2011): “‘Ought’ and Ability”, Philosophical Review 120(3), pp. 337–382. Greaves, H. (2016): “Cluelessness”, Proceedings of the Aristotelian Society 116, pp. 311–339. Griffin, J. (1977): “Are There Incommensurable Values”, Philosophy and Public Affairs 7(1), pp. 39–59. Hansson, S.O. (2018): “Risk”, in: E.N. Zalta (ed.), The Stanford Encyclopaedia of Philosophy (Fall 2018 Edition), URL = . Hare, R.M. (1979): “What Is Wrong with Slavery?”, Philosophy and Public Affairs 8(2), pp. 103–121. ———. (1981): Moral Thinking: Its Levels, Method, and Point, Oxford: Clarendon. Hedden, B. (2012): “Options and the Subjective Ought”, Philosophical Studies 158, pp. 343–360. Henning, T. (2014): “Normative Reasons Contextualism”, Philosophy and Phe­ nomenological Research 2014, pp. 593–624. Heuer, U. (2010): “Reasons and Impossibility”, Philosophical Studies 147, pp. 235–246. Hooker, B. (2000): Ideal Code, Real World, Oxford: Clarendon. ———. (2016): “Rule Consequentialism”, in: E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2016 Edition), URL = . Howard-Snyder, F. (1997): “The Rejection of Objective Consequentialism”, Utili­ tas 9(2), pp. 241–248. ———. (1999): “Response to Carlson and Qizilbash”, Utilitas 11(1), pp. 106–111. ———. (2006): “It’s the Thought That Counts”, Utilitas 17, pp. 265–281. Huemer, M. (2005): Ethical Intuitionism, New York, NY: Palgrave Macmillan. ———. (2008): “In Defence of Repugnance”, Mind 117(468), pp. 899–933. Hurley, S. (2000): “Responsibility, Reason, and Irrelevant Alternatives”, Philoso­ phy and Public Affairs 28(3), pp. 205–241. Jackson, E.G. (2020): “The Relationship Between Belief and Credence”, Philoso­ phy Compass 15(6), pp. 1–13. Jackson, F. (1991): “Decision Theoretic Consequentialism and the Nearest and Dearest Objection”, Ethics 101, pp. 461–482. Jackson, F. and R. Pargetter (1986): “Oughts, Options, and Actualism”, in: The Philosophical Review 95, pp. 233–255. Kagan, S. (1989): The Limits of Morality, Oxford: Oxford University Press. ———. (1998): Normative Ethics, Boulder, CO: Westview. ———. (2000): “Evaluative Focal Points”, in: B. Hooker, E. Mason, and D.E. Miller (eds.), Morality, Rules, and Consequences: A Critical Reader, Edin­ burgh: Edinburgh University Press, pp. 134–155. ———. (2011): “Do I Make a Difference?”, Philosophy and Public Affairs 39(2), pp. 105–141. ———. (2012): The Geometry of Desert, Oxford: Oxford University Press. Kekes, J. (1984): “‘Ought’ Implies ‘Can’ and Two Kinds of Morality”, The Philo­ sophical Quarterly 34, pp. 459–467. Kiesewetter, B. (2011): “‘Ought’ and the Perspective of the Agent”, Journal of Ethics and Social Philosophy 5(3).

References

165

———. (2016). “You Ought to φ Only If You May Believe That You Ought to φ”, Philosophical Quarterly 66(265), pp. 760–782. ———. (2017): The Normativity of Rationality, Oxford: Oxford University Press. Kolodny, N. and J. MacFarlane (2010): “Ifs and Oughts”, Journal of Philosophy 107, pp. 115–143. Kratzer, A. (1977): “What ‘Must’ and ‘Can’ Must and Can Mean”, Linguistics and Philosophy 1, pp. 337–355. Kumar, R. (1999): “Defending the Moral Moderate: Contractualism and Com­ mon Sense”, Philosophy and Public Affairs 28(4), pp. 275–309. Lazari-Radek, K. de and P. Singer (2012): “The Objectivity of Ethics and the Unity of Practical Reason”, Ethics 123(1), pp. 9–31. ———. (2014): The Point of View of the Universe: Sidgwick and Contemporary Ethics, Oxford: Oxford University Press. Lehrer, K. (1968): “Cans Without Ifs”, Analysis 29, pp. 29–32. ———. (1976): “‘Can’ in Theory and Practice: A Possible World Analysis”, in: M. Brand and D. Walton (eds.), Action Theory, Dordrecht: D. Reidel, pp. 241–270. Lenman, J. (2000): “Consequentialism and Cluelessness”, Philosophy and Public Affairs 29(4), pp. 342–370. Lewis, D. (1976): “The Paradoxes of Time Travel”, American Philosophical Quarterly 13, pp. 145–152. Lindblom, C.E. (1959): “The Science of ‘Muddling Through’”, Public Adminis­ tration Review 19, pp. 79–88. Mackie, J.L. (1977): Ethics: Inventing Right and Wrong. Harmondsworth: Penguin. Maier, J. (2018): “Abilities”, in: Stanford Encyclopedia of Philosophy (Spring 2018 Edition), E.N. Zalta (ed.), URL = . March, J.G. and Simon, H. (1958): Organizations, New York: Wiley. Mason, E. (2003): “Consequentialism and the ‘Ought Implies Can’ Principle”, American Philosophical Quarterly 40(4), pp. 319–331. McCann, H. (1974): “Volition and Basic Action”, The Philosophical Review 83, pp. 451–473. McCloskey, H.J. (1973): “Utilitarianism: Two Difficulties”, Philosophical Studies 24, pp. 62–63. McConnell, T. (1989): “‘“Ought” Implies “Can”’ and the Scope of Moral Require­ ments”, Philosophia 19, pp. 437–454. Mendola, J. (2005): “Consequentialism, Group Acts, and Trolleys”, Pacific Philo­ sophical Quarterly 86(1), pp. 64–87. ———. (2006a): Goodness and Justice: A Consequentialist Moral Theory, Cam­ bridge: Cambridge University Press. ———. (2006b): “Multiple-Act Consequentialism”, Noûs 40(3), pp. 395–427. Mill, J.St. (1861/1998): Utilitarianism, edited by R. Crisp, New York: Oxford University Press. Miller, D.E. (2003): “Actual-Consequence Act Utilitarianism and the Best Possible Humans”, Ratio 16(1), pp. 49–62. Miller, R.W. (2004): “Beneficence, Duty and Distance”, Philosophy and Public Affairs 32(4), pp. 357–383. Moore, E. (2007): “Objective Consequentialism, right actions, and good people”, Philosophical Studies 133, pp. 83–94.

166

References

Moore, G.E. (1912): Ethics, Oxford: Oxford University Press 1966. Norcross, A. (2006): “The Scalar Approach to Utilitarianism”, in: H.R. West (ed.), The Blackwell Guide to Mill’s Utilitarianism, Oxford: Blackwell Publish­ ing, pp. 217–232. Nozick, R. (1974): Anarchy, State, and Utopia, Oxford: Blackwell. Oddie, G. and P. Menzies (1992): “An Objectivist’s Guide to Subjective Value”, Ethics 102, pp. 512–533. Parfit, D. (1984): Reasons and Persons, Oxford: Clarendon Press. ———. (1988): “What We Together Do”, Unpublished. ———. (2011): On What Matters, Oxford: Oxford University Press. Persson, I. (2008): “A Consequentialist Distinction Between What We Ought to Do and Ought to Try”, Utilitas 20(3), pp. 348–355. Peterson, M. (2013): The Dimensions of Consequentialism: Ethics, Equality and Risk, New York: Cambridge University Press. Pettit, P. (2015): The Robust Demands of the Good: Ethics With Attachment, Virtue, and Respect, New York: Oxford University Press. Pettit, P. and G. Brennan (1986): “Restrictive Consequenitalism”, Australasian Journal of Philosophy 64, pp. 438–455. Pettit, P. and M. Smith (2000): “Global Consequentialism”, in: B. Hooker, E. Mason, and D.E. Miller (eds.), Morality, Rules, and Consequences: A Critical Reader, Edinburgh: Edinburgh University Press, pp. 121–133. Prichard, H.A. (1932): “Duty and Ignorance of Facts”, in his Moral Obligation, Oxford: Clarendon, 1949, pp. 18–39. Qizilbash, M. (1999): “The Rejection of Objective Consequentialism: A Com­ ment”, Utilitas 11(1), pp. 97–105. Railton, P. (1984): “Alienation, Consequentialism, and the Demands of Moral­ ity”, Philosophy and Public Affairs 13(2), pp. 134–171. Ripley, C.R. (1974): “A Theory of Volition”, American Philosophical Quarterly 11, pp. 141–147. Rosen, G. (2003): “Culpability and Ignorance”, Proceedings of the Aristotelian Society 103, pp. 61–84. ———. (2004): “Skepticism About Moral Responsibility”, Philosophical Perspec­ tives 18, pp. 295–313. Ross, W.D. (1930): The Right and the Good, Oxford: Oxford University Press. ———. (1939): Foundations of Ethics, Oxford: Clarendon Press. Russell, B. (1910): Philosophical Essays, London: Longmans, Green & Co. Ryle, G. (1945): “Knowing How and Knowing That”, Proceedings of the Aristo­ telian Society 46, pp. 1–16. ———. (1949): The Concept of Mind, London: Hutchinson. Schroeder, M. (2009): “Means-End Coherence, Stringency, and Subjective Rea­ sons”, Philosophical Studies 143, pp. 223–248. ———. (2011): “Ought, Agents, and Actions”, Philosophical Review 120(1), pp. 1–41. Shaw, W.H. (1999): Contemporary Ethics. Taking Account of Utilitarianism, Oxford: Blackwell Publishers. Sidgwick, H. (1907): The Methods of Ethics (Seventh Edition), London: Macmillan. Singer, P. (1972): “Famine, Affluence, and Morality”, Philosophy and Public Affairs 1(3), pp. 229–243. ———. (1993): Practical Ethics, Cambridge: Cambridge University Press.

References

167

———. (1995): How Are We to Live: Ethics in an Age of Self-Interest, Amherst, NY: Prometheus Books. Sinnott-Armstrong, W. (1984): “‘Ought’ Conversationally Implies ‘Can’”, The Philosophical Review 93, pp. 249–261. ———. (2019): “Consequentialism”, in: Stanford Encyclopedia of Philosophy (Summer 2019 Edition), E.N. Zalta (ed.), URL = . Skelton, A. (2012): “William David Ross”, in: Stanford Encyclopedia of Philoso­ phy (Summer 2012 Edition), E.N. Zalta (ed.), URL = . Slote, M. (2010): Moral Sentimentalism, Oxford: Oxford University Press. Smith, H.M. (1986): “Moral Realism, Moral Conflict, and Compound Acts”, The Journal of Philosophy 83, pp. 341–345. ———. (1988): “Making Moral Decisions”, Noûs 22(1), pp. 89–108. ———. (2010): “Subjective Rightness”, Social Philosophy and Policy 27(2), pp. 64–110. ———. (2018): Making Morality Work, Oxford: Oxford University Press. Smith, M. (2003a): “Neutral and Relative Value After Moore”, Ethics 113, pp. 576–598. ———. (2003b): “Rational Capacities, Or: How to Distinguish Recklessness, Weakness, and Compulsion”, in: S. Stroud and C. Tappolet (eds.), Weakness of Will and Practical Irrationality, Oxford: Clarendon Press, 2003, pp. 17–38. ———. (2006): “Moore on the Right, the Good, and Uncertainty”, in: T. Horgan and M. Timmons (eds.), Metaethics After Moore, Oxford: Clarendon Press, pp. 133–148. Sobel, J.H. 1976. “Utilitarianism and Past and Future Mistakes”, Noûs 10(2): 195–219. Stanley, J. (2011): “Knowing (How)”, Noûs 45, 207–238. Stanley, J. and T. Williamson (2001): “Knowing How”, The Journal of Philoso­ phy 98, pp. 411–444. Stern, R. (2004): “Does ‘Ought’ Imply ‘Can’? And Did Kant Think It does?”, Utilitas 16, pp. 42–61. Streumer, B. (2003): “Does ‘Ought’ Conversationally Implicate ‘Can’?”, Euro­ pean Journal of Philosophy 11, pp. 219–228. ———. (2007): “Reasons and Impossibility”, Philosophical Studies 136, pp. 351–384. ———. (2010): “Reasons, Impossibility and Efficient Steps: Reply to Heuer”, Philosophical Studies 151, pp. 79–86. Tännsjö, T. (1995): “In Defence of Theory in Ethics”, Canadian Journal of Phi­ losophy 25, pp. 571–594. Timmerman, T. and Y. Cohen (2020): “Actualism and Possibilism in Ethics”, in: Stanford Encyclopedia of Philosophy (Fall 2020 Edition), E.N. Zalta (ed.), URL = . Timmons, M. (2002): Moral Theory: An Introduction, Oxford: Rowman & Littlefield. Vihvelin, K. (2004): “Free Will Demystified: A Dispositionalist Account”, Philo­ sophical Topics 32, pp. 427–450. Vranas, P.B.M. (2007): “I Ought, Therefore I Can”, Philosophical Studies 136(2), pp. 167–216.

168

References

Wallace, R.J. (1994): Responsibility and the Moral Sentiments, Cambridge, MA: Harvard University Press. Wiland, E. (2005): “Monkeys, Typewriters, and Objective Consequentialism”, Ratio 18, pp. 352–360. Woodard, C. (2019): Taking Utilitarianism Seriously, Oxford: Oxford University Press. Zimmerman, M.J. (1987): “Remote Obligation”, American Philosophical Quar­ terly 24, pp. 199–205. ———. (2004): “Another Plea for Excuses”, American Philosophical Quarterly 41, pp. 259–266. ———. (2006a): “On the Fulfillment of Moral Obligation”, Ethical Theory and Moral Practice 9(5), pp. 577–597. ———. (2006b): “Is Moral Obligation Objective or Subjective”, Utilitas 18, pp. 329–361. ———. (2008): Living With Uncertainty: The Moral Significance of Ignorance, Cambridge: Cambridge University Press. ———. (2014): Ignorance and Moral Obligation, Oxford: Oxford University Press.

Index

abilities: ignorance-induced inability 64–9, 73; theories of abilities 11n13, 66–7, 76n8; see also objections to Ought Implies Can act-types 36, 70–1, 111 action-guidance: action-guidance and decision-guidance 9, 15, 24–26, 82–3; action-guidance and Ought Implies Can 46–7, 80–4; action-guidance and Ought Implies Evidence 95; action-guiding concepts 6–7, 29, 47; action­ guidingness of moral theories 7–8, 11n14, 16, 23–6, 80–4, 90, 95, 103n7; appropriate action-guidance 80, 82, 142–3, 145, 150–3, 155, 157; action-guiding properties 4–5, 26, 47, 80–81; see also deontic actions: basic and non-basic 66, 76n5; decisions and full-blown actions 25, 105, 108–10, 113; compound actions 11n13; identity-affecting actions 16; see also individuation actualism 6, 11n13, 158n5 agent-centred restrictions 28, 41n3 agent-neutrality 8, 125–6, 129–30; see also evaluator-neutrality agent-relativity 39, 61n9, 125–130; see also evaluator-relativity Anscombe, G.E.M. 76n10, 114n2 Argument from Action-Guidance 9, 80, 82, 90 Argument from Blameworthiness and Fairness 9, 84, 90 Argument from Crazy Reasons 87–8 Argument from Deliberation 88–9, 98 Argument from Endless Tryings 88 Argument from Impossibility Comparisons 9, 78, 90

Argument from Reasons 9, 86, 90 Austin, John L. 67, 102n1, 107 Ayer, A.J. 76n8 Baier, Annette 76n5 Bales, R. Eugene 158n3 belief: absence of belief 31–3; factual belief 5, 23, 26, 37, 83, 117; content of belief 20–1, 26n5, 34, 41n9; degrees/strength of belief 21, 26n5, 35, 41n9; outright belief 21, 26n5, 26n6, 34, 41n9 Bentham, Jeremy 10n1 Bergström, Lars 76n2, 92, 103n2, 103n3, 158n4 biases 102, 103n9 Bicchieri, Cristina 61n8 Björnsson, Gunnar 138n1 blameless wrongdoing 133, 139n14, 149–51, 158n6 blameworthiness: blameworthiness and fairness 77–8, 84–6, 90, 94–7; blameworthiness and wrongness 7–9, 50, 52–3, 113, 136, 151–2; excuses 92, 136, 151–2 Blum, Alex 49 Bratman, Michael 114n2 Brennan, Geoffrey 28 Broad, C.D. 11n7 by-relations 64–7, 76n7 Bykvist, Krister 5, 138n1, 158n10 candidate for blame 78, 85–6, 96–7 Carlson, Erik 62, 64, 67–72, 76n9, 76n10, 76n13, 90n4 category error 52–3, 85 causes: sensitive causal systems 16; deviant causal chains 109 certainty see uncertainty

170

Index

character traits 3, 28 Chisholm, Roderick 76n5 cognitive: cognitive abilities 29; cognitive access 38–9; cognitive limitations 1, 36, 144, 146; cognitive over-demandingness 1–2, 10, 40, 160; cognitive states 5, 29; see also uncertainty cognitively ideal agent 83, 90n4 Cohen, Yishai 11n13 common-sense morality 8, 25, 35, 69, 111, 114, 138, 158n9; common­ sense morality and paradigmatic moral duties 119–31 compulsion 49–50, 55, 58–60 conditional analysis 66, 68, 76n8 consequentialism: act­ consequentialism 4, 7–8, 110–11, 130, 159–61; definition of consequentialism 4–6, 11n10, 62–3, 70, 72; consequentialism and blame 7, 151–2, 158n8; rule­ consequentialism 4, 108, 129–30 constraints 128–9, 129n5, 129n6; see also agent-centred restrictions control 50, 61n4, 109–10 Copp, David 60n2 credences see degrees/strength of belief Crisp, Roger 8, 28 Dancy, Jonathan 11n12 Danto, Arthur 76n5 Davidson, Donald 76n10 decision guidance 9, 40–1, 9, 15–24, 27–8, 32, 34, 37, 41n3, 83; see also action-guidance decision procedures 27, 34–6, 41n1, 41n3, 122, 158n3; criteria of adequacy for decision procedures 28–31; decisions procedures as supplements for moral theories 22–3, 25–6, 80–3, 90, 95, 160–1; decision procedures and licensing strategies 131, 133, 138, 144–50 defeaters 9, 24–5, 92–4, 98–9 deliberation 24, 26, 86, 88–9, 98, 133, 136–7 deontic: deontic judgements 23, 80; deontic properties 3–6, 11n7, 26, 41n4; deontic status 48, 50, 59, 61n4, 109–10; deontic terms 46–7, 61n3, 103n6; see also iudicanda deontology 36, 111–12 desert 90n6, 94

discourse 7, 65, 68–9, 143 DP 33–41, 90n3, 144, 159–60 Dreier, Jamie 61n9 egoism 120 epistemic see cognitive evaluative 4–5, 46–8, 61n3, 80, 156; see also value evaluator-neutrality 61n7, 125–30 evaluator-relativity 125, 127 evidence: evidence and cognitive overdemandingness 40, factual evidence 9, 24–6, 82, 90–95, 98–9; 103n7, 161; normative evidence 93–4, 98–9, 161 evidentialism 5, 11n11, 37–8 expected value 5–6, 11, 61, 109–12, 140–3, 146–57; expected value and evidentialism 11n11; expected value and decision-guidance 17–18, 26n4, 39–40; expected value and social practices 61n9 Fantl, Jeremy 76n6 Fara, Michael 76n8 Fehige, Christoph 12n15, 61n9 Feldman, Fred 15–18, 26n1, 26n2, 26n3, 27, 29–31, 35, 39, 41n2, 41n5, 61n9, 114n1, 138n1 Finlay, Stephen 138n1 Frank, Robert H. 61n9 Frazier, Robert L. 26b1 Gesang, Bernward 61n9 Gill, Michael B. 103n19 Goldman, Alvin I. 76n10 Goldman, Holly S. 11n13, 61n5; see also Smith, Holly M. good see value Graham, Peter A. 49, 54–55, 60, 61n7, 103n3, 104–6, 114n1 Greaves, Hilary 18, 26n4 Griffin, James 60n2 Hansson, Sven Ove 36

Hare, R.M. 28, 48, 60n2, 138n2,

138n3 Hedden, Brian 114n2 Henning, Tim 11n12 Heuer, Ulrike 90n9 Hooker, Brad 4, 114n4, 129, 139n9 Howard-Snyder, Frances 9, 62–71, 73, 75, 76n2, 76n4, 76n11, 76n13, 77–8, 138n1

Index Huemer, Michael 12n15, 103n9 Hurley, Susan L. 76n8 ignorance 15–16, 18–19, 22, 64–7, 69, 146 individuation 68, 73–5, 76n10, 107 intellectualism 65–6, 76n7 intentions 105, 114n2, 158n2 iudicanda 107–8, 110, 114, 142, 145, 161 Jackson, Elizabeth G. 26n6 Jackson, Frank 11n13, 138n1, 139n8, 156, 158n1, 158n5 justice 25, 90n6, 160 Kagan, Shelly 12n15, 15, 61n9, 108, 114n3, 128–9, 138n5, 138n6, 158n6 Kantianism 32–3, 117, 119–20 Karpov 63–5, 67–73, 76n7, 77–9, 90 Kekes, John 49, 84 Kiesewetter, Benjamin 11n13, 41n4, 138n1, 139n16 knowledge-how 15–6, 65–6, 72–3, 76n7, 81, 103n2 Kolodny, Niko 11n12, 11n13, 132–4, 136–7, 138n1, 139n12, 139n16, 139n17, 139n18 Kratzer, Angelika 76n8 Kumar, Rahul 138n7 Lazari-Radek, Katarzyna de 12n15 Lehrer, Keith 76n8, 106 Lenman, Jimmy 16–17, 38 Lewis, David 76n8 licensing: Licensing Dilemma 10, 140–42, 144–5, 148, 150, 152–4, 157, 158n1; licensing strategies 131–4, 136, 140–1, 152, 158n1 Lindblom, Charles E. 26n1 MacFarlane, John 11n12, 11n13, 132–4, 136–7, 138n1, 139n12, 139n16, 139n17, 139n18 Mackie, J.L. 138n7 Maier, John 69 March, James 26n1 Mason, Elinor 76n1 McCann, Hugh 76n5 McCloskey, Henry John 75n2 McConnell, Terence 102n1, 111–14 Mendola, Joseph 61n9 Menzies, Peter 126–7, 138n1

171

metaethics 2, 100, 117, 157 Mill, John Stuart 10n1 Miller, Dale E. 24, 76, 114 Miller, Richard W. 138 Miners case 132–7, 139n16, 139n17, 139n18; see also licensing strategies mixed view 6, 11n12, 140–1, 153, 155, 157, 158n11, 160 Moore, Eric 76n2 Moore, George Edward 7, 11n7, 76n8, 138n1, 158n7 moral intuitions 119, 124, 156, 160; appeasement strategy 120–2, 125, 130; confrontational strategy 120–1, 130 moral rationalism 103n10 moral sentimentalism 103n10 moral theories: implementation 12n14, 19, 25, 28–9, 80–1, 90n3, 138n3; paradigmatic duties of moral theories 118–23, 125–30, 138, 139n11 motives 28, 149–50 Muldoon, Ryan 61n8 New Perspective Argument 10, 118, 137 Norcross, Alastair 158n8 Normative: normative as encompassing deontic and evaluative 3–5, 103n6; normative doctrines and theories 4; normative ethics 2–3, 10n5, 101, 117; normative explanations 3, 56; normative properties 22 Nozick, Robert 138n6 Objection from Lack of Normative Evidence 98–9 objections to Ought Implies Can: objection from self-imposed inability 49–54; straightforward objection 49–50; sophisticated objection 54–60 Oddie, Graham 126–7, 138n1 Oedipus 117–8, 127, 134–5, 156 Ought Implies Can 9, 10, 11n6, 23, 26, 45–7, 60; compatibility with objective consequentialism 9, 62–3, 68, 71, 73–5, 103n2; compatibility with subjective consequentialism 9, 104–7; rationales of Ought Implies Can 9–10, 46, 77–8, 91, 94; recipe for defending Ought Implies Can

172

Index

48–9; see also objections to Ought Implies Can Ought Implies Evidence 9–10, 24–5, 92–5, 103n2, 103n3, 159–61; arguments for Ought Implies Evidence 95–8; objections against Ought Implies Evidence 98–102 ought-making features see rightmaking features Parfit, Derek 11n12, 24, 28, 132,

138n1, 139n12, 139n14, 139n18,

158n6

Pargetter, Robert 11n13, 158n5 Persson, Ingmar 139n13, 158n2 Peterson, Martin 11n12 Pettit, Philip 28, 61n9, 108, 158n6 possibilism see actualism prescriptive see deontic Prichard, H.A. 11n7, 105, 108–13, 138n1 prima facie duties 20, 32, 46, 59–60,

93, 134, 138n5

Primacy of Intuitions Objection 122–4 pro tanto duties 38–9; see also prima facie duties Probability: objective probability

10n3, 36–7; probabilities and

uncertain beliefs 20–1, 26n5, 34,

36, 141; risk and chance 35–6,

38–9; value-probability-products 6,

17, 39–40

problem of decision-guidance see decision-guidance promises 50–3, 58–9, 111, 127–8, 130, 138n5, 139n9 prudence 3, 7–8 Qizilbash, Mozaffar 76n2

question Q 19–23, 28, 34, 82, 103n7

Railton, Peter 24, 28, 138n3

rational-licence view 154–7 rationality 22, 29, 34, 41n4, 156–7

Reason Implies Can 86–8, 90n8, 90n10, 97–8 Reason Implies Evidence 89, 97–8 responsibility 66, 69, 110, 113, 156–7

right-making features 2–3,

18–20, 24–5, 31–7, 82, 103n7;

right-making features and

consequentialism 18, 24, 38–9, 72,

125, 157

Ripley, Charles 76n5 risk: excessive risk 30–1; morality of

risk 30–1; risk as a wrong-making

feature 35–8, 135, 141, 149–50,

155–6

Rosen, Gideon 102n1

Ross, W.D. 11n7, 41n5, 46, 94, 105,

108–13, 138n1, 138n5 Russell, Bertrand 11n7 Ryle, Gilbert 65–6, 76n6 Schroeder, Mark 11n12, 139n19

self-defeatingness 24

Shaw, William H. 7, 158n8

Sidgwick, Henry 10n1

Simon, Herbert 26n1

Singer, Peter 12n15, 138n7

Sinnott-Armstrong, Walter 4, 6, 50

Skelton, Anthony 138n5

Slote, Michael 103n10

Smith, Holly M. 11n14, 19, 26n1, 29,

41n2, 47, 80, 90n2, 138n1; see also

Holly S. Goldman

Smith, Michael 61n9, 76n8, 108,

156–7, 158n6, 158n10

Sobel, Jordan Howard 11n13

social norms see social practices

social practices 56–9, 61n9, 151

Sontuoso, Alessandro 61n8

special obligations 28, 41n3, 56, 120,

130, 160

Stanley, Jason 65–6

Stern, Robert 84

Streumer, Bart 61n5, 86–9, 90n7,

90n9, 90n10, 97

subjectivism: pure subjectivism 5,

11n9, 11n11; see also evidentialism

subsumption 19–20

supererogation 8, 158n9, 160

Tännsjö, Torbjörn 158n4

Tanyi, Attila 11n12

Tarantino 103n8

theory of practical reason 7–8, 22,

130–1, 138n1, 139n11, 160

time: consumption of time 26n4,

29–31, 150; double time-index 51,

93; time travel 47, 63, 68

Timmerman, Travis 11n13 Timmons, Mark 10n5 transplant case 54–61 Travesty of Decision-Guidance Problem 9–10, 26n6, 27, 38–40

Index uncertainty 1, 9–10, 16, 20, 25, 26n5,

31–41, 83, 142; moral uncertainty

100–102; see also risk

utilitarianism 8, 16, 26n1, 29, 31, 48,

75n2; classical utilitarianism 1–2, 4,

10n1, 41n7, 93–4

value: impartial value 7; incommensurable values 11n10; objectivism about value 6; value theory 6, 38, 61n9, 125; see also expected value Vihvelin, Kadri 76n8

173

virtue ethics 41, 117, 119

Volition Thesis 104–8, 111–14

Vranas, Peter B.M. 60n2

Wallace, R. Jay 103n1

Wiland, Eric 69, 76n2, 79

Williamson, Timothy 66

Woodard, Christopher 61n9

wrong-making features see right-

making features Zimmerman, Michael J. 11n9, 61n5, 103n1, 138n1, 158n5