Consequentialism: New Directions, New Problems 019027011X, 9780190270117

Citation preview

Consequentialism

OXFORD MORAL THEORY Series Editor David Copp, University of California, Davis Drawing Morals: Essays in Ethical Theory Thomas Hurka Commonsense Consequentialism: Wherein Morality Meets Rationality Douglas W. Portmore Against Absolute Goodness Richard Kraut The Lewd, the Rude and the Nasty Pekka Väyrynen In Praise of Desire Nomy Arpaly and Timothy Schroeder Confusion of Tongues: A Theory of Normative Language Stephen Finlay The Virtues of Happiness: A Theory of the Good Life Paul Bloomfield Having It Both Ways: Hybrid Theories and Modern Metaethics Edited by Guy Fletcher and Michael Ridge Motivational Internalism Edited by Gunnar Björnsson, Caj Strandberg, Ragnar Francén Olinder, John Eriksson, and Fredrik Björklund The Meaning of ‘Ought’: Beyond Descriptivism and Expressivism in Metaethics Matthew Chrisman Practical Knowledge: Selected Essays Kieran Setiya Articulating the Moral Community: Toward a Constructive Ethical Pragmatism Henry S. Richardson Consequentialism: New Directions, New Problems Edited by Christian Seidel

Consequentialism NEW DIRECTIONS, NEW PROBLEMS

Edited by Christian Seidel

1

1 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America. © Oxford University Press 2019 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. CIP data is on file at the Library of Congress ISBN 978–​0–​19–​027011–​7 9 8 7 6 5 4 3 2 1 Printed by Sheridan Books, Inc., United States of America

Contents Acknowledgments List of Contributors 1. New Wave Consequentialism: An Introduction Christian Seidel

vii ix 1

PART I  |  Directions: Exploring New Wave Consequentialism 2. World-​Centered Value Jamie Dreier

31

3. Multidimensional Consequentialism and Population Ethics Martin Peterson

51

4. Consequentialism and Coordination: How Traditional Consequentialism Has an Attitude Problem Douglas W. Portmore 5. Fittingness Objections to Consequentialism Richard Yetter Chappell

71 90

PART II  |  Problems: Scrutinizing the New Wave’s Theoretical Basis 6. The Good of Consequentialized Deontology Monika Betzler and Jörg Schroth

115

7. The Restrictions of Consequentialism Jan Gertken

136

v

vi    Contents

8. Commonsense Consequentialism, Moral Rationalism, and Options Dale Dorsey

156

9. New Consequentialism and the New Doing-​Allowing Distinction Paul Hurley

176

10. Consequentialism, Rationality, and Kantian Respect Tim Henning

198

PART III  |  Scope: Assessing New Wave Consequentialism in Context 11. Consequentialism and Moral Responsibility Elinor Mason

219

12. Kantianism, Consequentialism, and Deterrence Steven Sverdlik

237

Index

259

Acknowledgments My first thanks go to all of the contributors to this volume for their commitment, inspiring ideas, and admirable openness throughout the project; it has been a pleasure and an honor to learn from them. I  am deeply indebted to Peter Ohlin, Emily Saccharin, Andrew Ward, and Isla Ng at Oxford University Press for their help and patience throughout this process; to an anonymous reviewer for insightful and constructive comments; to Konstantin Weber for unfailing editorial assistance; and to Gerhard Ernst for invaluable encouragement and advice from the start. Most of all, I would like to thank Ulrike for her unwavering support and for being what she is.

vii

Contributors Monika Betzler holds the chair for practical philosophy and ethics at the Ludwig-​ Maximilians-​Universität at Munich. Her main interests lie in moral psychology, theories of practical reason, and normative ethics. Her recent research focuses on the normative significance of close relationships and personal projects as well as on autonomy, weakness of will, and other challenges to our rational agency. She has a book in preparation on why personal projects matter, and her most recent coedited book is on familiar duties (2015). Richard Yetter Chappell is an assistant professor of philosophy at the University of Miami. His primary research interests concern the defense and development of consequentialism, effective altruism, and robust (nonnaturalist) normative realism. Chappell blogs at http://​www.philosophyetc.net about these and other topics (including philosophical zombies, peer disagreement, and personal identity). He has published articles in journals including Noûs, Australasian Journal of Philosophy, Philosophical Quarterly, and Philosophical Studies, and was co-​awarded the Rocky Mountain Ethics Congress 2013 Young Ethicist Prize. Dale Dorsey is a professor of philosophy at the University of Kansas and held fellowships at the Murphy Institute (Tulane University) and Australian National University. He has published widely on issues in normative ethics, metaethics, and political philosophy, and his research focuses on the intersection between morality, well-​being, and practical reason. He is the author of The Basic Minimum: A Welfarist Approach (2012) and, more recently, The Limits of Moral Authority (2016). He is currently working on a large-​form project on the nature of prudence and prudential rationality.

ix

x    Contributors

Jamie Dreier is a professor of philosophy at Brown University, where he has taught since 1988. Dreier has also taught at Monash University in Melbourne and has been the Harsanyi Fellow at the Research School of Social Sciences at the Australian National University. He is a founding editor of the online Journal of Ethics and Social Philosophy and an associate editor of Ethics. He writes about metaethics, especially expressivism and relativism; about meta-​metaethics, including the question of what the real difference is between metaethical theories that end up saying similar-​sounding things; about practical rationality and decision theory; and about the more abstract parts of normative ethics, including the structure of normative theories and the project of consequentializing traditionally deontological theories. Jan Gertken is a lecturer in philosophy at Humboldt-​Universität zu Berlin. His main research interests include issues in normative ethics and value theory, as well as questions concerning reasons and rationality. His book Prinzipien in der Ethik (2014) discusses the prospects of moral particularism. Currently, he mainly works on the “numbers problem” in ethics, that is, the question of whether the number of persons affected by one’s actions is of moral importance. Tim Henning is a professor of practical philosophy at the University of Stuttgart. He has worked on personal identity and autonomy (the topic of his first book, Person sein und Geschichten erzählen, which was awarded the Wolfgang-​Stegmüller-​ Award), on metanormativity, practical rationality, and the philosophy of language (the topics of his second book, From a Rational Point of View, 2018), on Kant (a German introduction to Kant’s ethics was published in 2016), and on various issues in normative ethics and applied ethics. His work has appeared in Ethics, Philosophical Review, Philosophy and Phenomenological Research, and Philosophical Quarterly. Paul Hurley is the Edward J. Sexton Professor of Philosophy at Claremont McKenna College, The Claremont Colleges. His research focuses on central questions in both normative ethics and metaethics and has appeared in numerous journals, including Ethics, Mind, Philosopher’s Imprint, Oxford Studies in Normative Ethics, Journal of Moral Philosophy, and Philosophical Studies. In addition he is the author of Beyond Consequentialism (2009). His current project is to demonstrate that the case for outcome-​centered moral theories such as consequentialism is rooted in outcome-​ centered accounts of reasons, values, attitudes, and actions, and, in challenging these accounts, to challenge consequentialism at its roots. Elinor Mason is a senior lecturer at the University of Edinburgh, having previously held positions at Arizona State University, the University of Colorado, Boulder, and most recently at Princeton University as a Laurence S. Rockefeller visiting fellow.

Contributors    xi

She has published on a wide range of topics in normative ethics (especially consequentialism), moral responsibility, and feminism. She has just completed a monograph on the connections between normative concepts (such as wrongness) and responsibility concepts (such as blameworthiness). Martin Peterson is the Sue G. and Harry E. Bovay Jr. Professor in the Department of Philosophy at Texas A&M University. Prior to coming to Texas A&M in 2014, he held academic posts at universities in the Netherlands (Eindhoven), UK (Cambridge), and Sweden (KTH Royal Institute of Technology). Peterson has published about sixty peer-​reviewed papers on normative ethics, applied ethics, and decision theory. His books include The Dimensions of Consequentialism, The Ethics of Technology: A Geometric Analysis of Five Moral Principles, An Introduction to Decision Theory, and Non-​Bayesian Decision Theory. Douglas W. Portmore is a professor of philosophy at Arizona State University. His research focuses on morality, rationality, and the relationship between the two. In his widely discussed book Commonsense Consequentialism (2011), he defends a version of consequentialism that unites commonsense moral convictions with a teleological conception of practical reasons. This version of consequentialism accommodates commonsensical ideas such as the idea that we’re not permitted to commit murder even to prevent two others from each committing murder, but it does so while holding on to the teleological notion that the deontic status of act is just a function of how its outcome ranks in comparison to those of the alternatives. Currently Portmore is working on his second book, Opting for the Best: Oughts and Options, which concerns the principle that agents ought always to perform their best option—​best, that is, in terms of whatever ultimately matters. Jörg Schroth teaches philosophy, with a focus on ethics, at the University of Göttingen. He is the author of Die Universalisierbarkeit moralischer Urteile (2001) and editor of Texte zum Utilitarismus (2016). For further information please visit his website, http://​www.ethikseite.de. Christian Seidel is a professor of philosophy at the Karlsruhe Institute of Technology. His main interests include personal autonomy (the topic of his book Selbst bestimmen:  Eine philosophische Untersuchung personaler Autonomie, 2016), issues in moral and political philosophy (consequentialism, moralism, distributive justice, egalitarianism, climate ethics, and the ethics of risk), as well as British moral and social philosophy in the nineteenth century ( John Stuart Mill and Henry Sidgwick). He was previously a lecturer at Friedrich-​Alexander-​Universität Erlangen-​Nürnberg and a postdoctoral fellow at the University of Zurich’s Centre for Ethics, where he worked on questions of intergenerational and global justice.

xii    Contributors

Together with Dominic Roser, he has authored a book on climate ethics (Climate Justice: An Introduction, 2017). Steven Sverdlik is Professor Emeritus of Philosophy at Southern Methodist University in Dallas, Texas, where he has taught since 1982. He has published papers on ethics, moral psychology, and action. His Motive and Rightness appeared in 2011. He is currently working in the philosophy of punishment. His most recent papers in this area are “The Origins of ‘The Objection,’ ” History of Philosophy Quarterly (2012); “Punishment and Reform,” Criminal Law and Philosophy (2014); “Giving Wrongdoers What They Deserve,” Journal of Ethics (2016); and “Public Policy and Philosophical Accounts of Desert,” in Routledge Handbook of Moral Epistemology (forthcoming).

Consequentialism

1 N E W WAV E CONSE QUE NTIA LISM

An Introduction

Christian Seidel

Consequentialism has been a focal point of discussion and a driving force behind developments in normative ethics ever since Bentham, Mill, and Sidgwick articulated (what is today believed to be) its most prominent version, classical utilitarianism.1 Indeed, major parts of the “history” of modern moral philosophy from then on—​the works of Moore, Ross, Anscombe, Foot, Rawls, Williams, Nozick, Hare, Nagel, Scheffler, Parfit, to name but a few—​can be understood in terms of fierce criticism and illuminating defenses of consequentialist moral theories. Until today, proponents of rival ethical perspectives (e.g., deontology, contractualism, and virtue ethics) continue to motivate and elaborate their accounts by contrasting them with consequentialism. But what, exactly, is consequentialism? This is a hard question, because “[i]‌t is not easy to identify any definition of consequentialism that would satisfy all those who have invoked that idea.”2 It may help to remember that “consequentialism” is

1 Bentham, An Introduction to the Principles of Morals and Legislation; Mill, Utilitarianism; Sidgwick, The Methods of Ethics. For the claim that Mill was actually a nonconsequentialist utilitarian, see Jacobson, “Utilitarianism without Consequentialism.” 2 Sen, The Idea of Justice, 217.

1

2    Christian Seidel

a philosophical term of art that is invoked to serve a theoretical purpose:3 the term should be useful in moral theorizing, by marking a helpful distinction between classes of theories. Such a distinction is helpful if it guides us in choosing between theories on account of their advantages and disadvantages, i.e., if the distinction is closely connected to positive or negative features that are relevant for theory selection and that all theories falling on one side of the distinction share. The concept of consequentialism is thus an instrument to classify moral theories under some heading and to frame (and simplify) the discussion between competing theories in terms of general (positive or negative) features that all theories under this heading share: if we had shown that some candidate theory T is consequentialist, and if we had some general objection to (or attractive feature of ) a theory in virtue of its being consequentialist, then the objection (or attractive feature) would apply to T as well. One implication of this view is that consequentialism is not itself a moral theory but rather a class of moral theories that share some common property that makes them consequentialist. Which property? This will depend on the nature of the objection (or attractive feature) that is to be generalized to a class of theories. To illustrate:  When Anscombe coined the term “consequentialism” in 1958,4 her aim was to mark a distinction between theories of English moral philosophers from Sidgwick onward (“consequentialists”) and theories of earlier philosophers. The notable thing is that she explicitly placed Mill in the nonconsequentialist and Ross in the consequentialist camp,5 while from a contemporary perspective, we would classify them the other way round. What explains the difference is that Anscombe’s classification has a different point.6 She wanted to classify theories according to whether they allow actions, once they fall under a moral principle or (secondary) rule, to be treated (in practical deliberation) as if they were consequences—​objects of maximization and weighing—​ such that an unjust act (in spite of being generally prohibited and prima facie wrong under one aspect) might nevertheless be all things considered right if it produces the greatest balance of happiness or balance of prima facie rightness over prima facie wrongness. Anscombe thought that Mill did not believe this, while Ross (or rather every English moral philosopher from Sidgwick on) did.7 Her point of introducing 3 Similarly, Portmore, “Consequentializing Moral Theories,” 61; Jacobson, “Utilitarianism without Consequentialism,” 164; Brown, “Consequentialize This,” 751; Dreier, “In Defense of Consequentializing,”97; Portmore, Commonsense Consequentialism, 37. 4 Anscombe, “Modern Moral Philosophy,” 12f. 5 Anscombe, “Modern Moral Philosophy,” 9, 9n. 6 As has been pointed out by Diamond, “Consequentialism in Modern Moral Philosophy and in ‘Modern Moral Philosophy.’ ” 7 For evidence on Ross, see Ross, The Right and the Good, 41. Note that Anscombe was to some extent correct in picking the term “consequentialism,” since the terminology within the utilitarian tradition indeed begins



New Wave Consequentialism    3

the term “consequentialism” was to highlight the distinctive property that marks this difference and to advance an objection that applies to all theories with—​and in virtue of—​the highlighted property (viz., that in consequentialism so construed, there are no actions which are absolutely ruled out in advance of deliberation; everything might, in principle, be outweighed in some circumstances).8 If today we regard Mill, unlike Ross, as a consequentialist, this is presumably not because we think that Mill, rather than Ross, satisfies Anscombe’s distinctive property; rather, we think that other properties are distinctive of consequentialism. Our disagreement with her is rooted not in different views on a given concept’s extension but in different concepts. So our concept of consequentialism has evolved considerably, but what is our current concept? Given the connection between distinctions and objections (or attractive features), the conceptual change is intimately linked to and driven by debating consequentialism. Our current concept of consequentialism is therefore the result of attempts to identify a class of moral theories that are subject to (or can be defended against) a number of characteristic objections and that have (or fail to have) certain attractive features. The following section retraces how this debate and the concept of consequentialism developed. 1.1. Three Waves of Discussing Consequentialism

1.1.1 The First Wave: Consequentialism’s Conceptual Emancipation from Utilitarianism

The debate on consequentialism gained momentum in the 1970s, when a first wave of influential and forceful objections was raised against utilitarianism: that it cannot account for distributive justice because it “does not take seriously the distinction between persons”9; that its doctrine of negative responsibility triggers excessive demands that alienate agents from their most personal commitments and that attack their integrity;10 that it cannot accommodate rights but sacrifices everyone in to change with Sidgwick: Moore was the first to explicitly state his (version of a utilitarian) theory in terms of “effects” and “consequences” (Ethics, 35, 70). Though Sidgwick’s “official” statement of the utilitarian doctrine (The Methods of Ethics, 411) does not refer to consequences (or effects), it is quite clear that he thought of the utilitarian doctrine in terms of consequences throughout book IV (see, e.g., 416, 425, 429). Previous “official” statements of the utilitarian doctrine by Bentham or Mill did not refer to consequences (or effects) either, but rather to the promotion or production of happiness (Bentham, An Introduction to the Principles of Morals and Legislation, 11f.; Mill, Utilitarianism, 210). But note that Mill elsewhere claims that “[t]‌he morality of an action depends on its foreseeable consequences” (“Bentham,” 112). 8 The subsequent departure from Anscombe’s notion becomes evident, for instance, when Bernard Williams rejects the interpretation of nonconsequentialism as absolutism. See Williams, “A Critique of Utilitarianism,” 90. 9 Rawls, A Theory of Justice, 27. 10 Williams, “A Critique of Utilitarianism.”

4    Christian Seidel

the utility monster’s maw;11 that pleasurable experiences (which could be simulated by a machine) are not the only things of intrinsic value;12 and that it does not fit well with moral considerations rooted in agent-​relative reasons and values (e.g., that an agent’s killing someone herself is morally worse than her failing to prevent another person from killing someone).13 These objections, however, were primarily14 directed at maximizing (hedonic) act-​ utilitarianism. This is a very specific moral doctrine according to which an action ϕ is morally permissible if and only if (and because), among the alternatives available to the agent, ϕ maximizes total utility, where total utility is (i) the aggregate sum of (ii) equally weighted (iii) net well-​being (in its hedonic version: pleasures minus pains) of all morally relevant beings. Several moral philosophers noted that the canonical objections tend to lose their force when we consider moral theories that are close in spirit15 (or “structure”) to this utilitarian doctrine, but part company with some of its distinctive features, such as its exclusive focus on acts, its hedonistic and monistic axiology, or its impartial sum-​ranking. For instance, utilitarians tried to address implausible implications (e.g., that maximizing utility may require us to inflict horrendous harm or to abandon friends and projects) by developing versions of rule-​utilitarianism16 or motive-​ utilitarianism:17 They argued that utilitarianism determines the general rules of behavior or patterns of motivations that are best to have (in terms of utility), and that these rules and motives override a case-​by-​case utility calculation; what we ought to do in the allegedly counterintuitive cases is to continue to act in accordance with the utility-​maximizing rule or motive. On this account, an act’s deontic status no longer depends on whether it maximizes utility, but on whether it is recommended or demanded by rules or motives that maximize utility. The assessment of actions is

11 Nozick, Anarchy, State, and Utopia, 28–​42; the utility monster appears on 41. 12 Nozick, Anarchy, State, and Utopia,  42–​45. 13 Nagel, “The Limits of Objectivity”; Nagel, The View from Nowhere, chs. 8–​9. The idea of agent-​relativity has been carefully analyzed in McNaughton and Rawling, “Agent-​Relativity and the Doing-​Happening Distinction.” 14 Primarily, but not exclusively; see, e.g., Williams, “A Critique of Utilitarianism,” 81–​93. 15 The received view today is that utilitarianism is not only close in spirit to but by definition a version of consequentialism. See, e.g., Sen and Williams, “Introduction,” 4; Parfit, Reasons and Persons, 26; Bykvist, Utilitarianism, 19; and many others. For reasons why this might not be the best view, see Eyal, “Non-​Consequentialist Utilitarianism.” 16 Brandt, A Theory of the Good and the Right. See also Hare, Moral Thinking, for a combination of act-​and rule-​utilitarianism at different levels of moral thinking that is supposed to deflate the standard objections which, according to Hare, are based on the “commonest trick” of confusing both levels (“Ethical Theory and Utilitarianism,” 123). 17 Adams, “Motive Utilitarianism.”



New Wave Consequentialism    5

only “indirectly” utilitarian, because what makes actions right or wrong is not their maximizing utility but their conformity to something else that maximizes utility. Instead of changing the criterion of rightness and keeping the focus on (direct or indirect) evaluation of acts, other philosophers took a pluralist approach that allowed multiple forms of utilitarian (or, as it then began to be called, consequentialist) evaluations. For instance, Parfit made clear that such evaluations may apply to everything (not just acts),18 and that therefore consequentialists can weaken the force of objections (e.g., based on special duties toward one’s own children) by more nuanced descriptions of the moral landscape that allow for blameless wrongdoing and similar phenomena:  though giving a small benefit to one’s own child rather than a greater benefit to a stranger is wrong because it fails to maximize the good, the father is not blameworthy because he acts from a motive that maximizes the good. Railton addressed the alienation objection by endorsing a pluralistic axiology with several intrinsic values19 and by arguing that a form of sophisticated act-​consequentialism “can approve of dispositions, characters, or commitments to rules . . . that do not merely supplement a commitment to act for the best, but sometimes override it, so that one knowingly does what is contrary to maximizing the good.”20 And Sen21 showed (i)  how to accommodate rights by conceiving of them as features of outcomes that are subject to consequentialist assessments (thus introducing an early form of consequentializing) and (ii) how this nonwelfarist consequentialist evaluation allows rankings of outcomes that vary between persons (thus accounting for agent-​relativity).22 The result (or achievement) of this discussion was consequentialism’s conceptual emancipation from utilitarianism. It became increasingly clear that different components of standard utilitarianism have to be carefully distinguished and that different objections aimed at different components and do not easily generalize to structurally similar theories.23 At the same time, the question was whether there is some such “core” structural element of the (monistic, welfarist, agent-​neutral) utilitarian doctrine that—​although suitable for flexible combinations with pluralistic, nonwelfarist, or perhaps even agent-​relative axiologies—​is still susceptible to revised and more abstractly rephrased versions of the canonical objections, as

Parfit, Reasons and Persons, 24–​27,  31–​35. 19 Railton, “Alienation, Consequentialism, and the Demands of Morality,” 149. 20 Railton, “Alienation, Consequentialism, and the Demands of Morality,” 159. 21 Sen, “Rights and Agency.” 22 Sen called this approach “consequentialist evaluation” rather than “consequentialism” because he wanted to stick with the then common association of consequentialism with agent-​neutrality (“Rights and Agency,” 30). This association loosened later on in new wave consequentialism. 23 See, e.g., the introductory sentences in Sen, “Utilitarianism and Welfarism.” 18

6    Christian Seidel

Williams conjectured in his earlier critique of utilitarianism. It was around then that consequentialism—​the sought-​for core element—​took a life of its own, was recognizably distinguished from utilitarianism, and became the established focal concept in the discussion, e.g., when Foot noted that “what is most radically wrong with utilitarianism is its consequentialism.”24 It is not surprising that consequentialism’s conceptual emancipation was reflected in the title of influential books too.25 Given this emancipated concept of consequentialism (which was often a concept of act-​consequentialism), a consequentialist theory has two components: it first ranks outcomes (understood as states of affairs), which the available actions bring about, in terms of their impersonal (or agent-​neutral) value, from best to worst; it then determines an action’s deontic status on the basis of this ranking such that an action is right if and only if it produces the highest-​ranked outcome.26 As a first approximation, we might explicate the first wave’s concept of consequentialism as follows: A (logically consistent)27 moral theory T is consequentialist with respect to the rightness of acts if and only if T contains an instance of the following scheme: (C1)  For all acts x: The right x is the one that is ranked highest in terms of the (impersonal) value (or goodness) of the outcome x brings about (where the ranking ranges over all alternatives to x). 1.1.2 The Second Wave: Elaborations, Generalized Objections, and a Dialectical Impasse

Premised on this preconception of consequentialism, a diversified second wave of debate built up. It was marked by two connected threads. First, there was a refined differentiation within the consequentialist “research program.” Drawing on the groundwork of the first wave, consequentialists tried to gain a better understanding Foot, “Utilitarianism and the Virtues,” 197. This was also a major concern in Williams’s critique: “[S]‌ome undesirable features of utilitarianism follow from its consequentialist structure” (“A Critique of Utilitarianism,” 81). 25 Compare Scheffler’s The Rejection of Consequentialism and Consequentialism and Its Critics to Smart and Williams’s Utilitarianism and to Sen and William’s Utilitarianism and Beyond. 26 Compare Williams, “A Critique of Utilitarianism,” 87; Sen, “Utilitarianism and Welfarism,” 464; Scheffler, The Rejection of Consequentialism, 1; Scheffler, Consequentialism and Its Critics, 1. John Broome prefers a different analysis: on his narrower understanding, consequentialism is a claim about how the goodness of acts is determined. Then a theory is consequentialist as it is commonly understood if it is (1) teleological and (2) “consequentialist” in Broome’s narrower sense; i.e., if it claims that (1) an act’s rightness is determined by the goodness of acts (represented in an ordering) and that (2) the goodness of an act (and thus the ordering) is determined by the goodness of the prospects (roughly: possible outcomes, broadly construed) it leads to (Weighing Goods, ch.1; Weighing Lives, 31, 41). 27 For instance, a theory that contains an instance of (C1) but also claims that, e.g., the rightness of actions is never determined by the value of their consequences should not be counted as consequentialist. 24



New Wave Consequentialism    7

of consequentialism by elaborating important concepts. The aim was, of course, to respond to the standard objections against utilitarianism insofar as they had a bearing on consequentialism. The debates thus focused on developing a proper—​and often complex—​axiological basis (to resolve problems about, e.g., sadistic pleasures);28 on developing distribution-​sensitive consequentialism by addressing questions about the aggregation and distribution of the good (e.g., to rebut the implications of unweighted sum-​ranking in cases of acts that benefit the well-​off more than they harm the worse-​off );29 or on best responses to the problem of demandingness.30 By examining its basic notions of “outcome,” “consequence,” and “alternative,”31 consequentialism gained further flexibility:  consequences were construed very broadly so as to include everything that would be the case as a result of performing an act—​ which includes the fact that the act has been performed as well as facts about the history of that act. Sophisticated versions of indirect (rule-​)consequentialism32 and the option of sidestepping the problems associated with maximizing goodness by a form of satisficing consequentialism33 were intended to bring consequentialism even closer to common sense. With all these refinements, consequentialism moved far away from classical utilitarianism by dismissing the idea of maximizing the unweighted sum of a single good. Consequentialist theories became increasingly flexible, but also increasingly diverse. To conceive of them as members of a single family, these diverse theories had to fall under the same concept of consequentialism, and so consequentialism had to be understood at a higher level of abstraction. If we generalize (C1) and start with the idea that consequentialism identifies the right act with the act that maximizes value, and if we want to include different conceptions of outcomes (actual, foreseen, foreseeable, intended, expected, etc.) and the possibility of indirectly evaluating actions in terms of other things that are related to them (e.g., sets of rules that permit an action), then we get the idea that a theory is consequentialist if it says something like this: For all acts x: The right x is the one that is ranked highest in terms of the (impersonal) value in some type C of outcomes of something related to act x, G(x) (where the ranking ranges over all alternatives to G(x)).

28 Griffin, Well-​Being; Sumner, Welfare, Happiness, and Ethics; Feldman, Pleasure and the Good Life. 29 Broome, Weighing Goods; Parfit, “Equality and Priority”; Broome, Weighing Lives. 30 Kagan, The Limits of Morality; Mulgan, The Demands of Consequentialism. 31 Vallentyne, “Utilitarianism and the Outcomes of Actions”; Broome, Weighing Goods, ch. 5; Sosa, “Consequences of Consequentialism”; Carlson, Consequentialism Reconsidered. 32 Hooker, Ideal Code, Real World; Hooker et al., Morality, Rules, and Consequences. 33 Slote, Common-​sense Morality and Consequentialism.

8    Christian Seidel

In this explication, the rightness of act x is determined by the evaluative ranking through a specific function, which is maximizing. But to include satisficing and other nonmaximizing approaches, we should more generally consider arbitrary rightness functions.34 And if we want to include the idea that actions are assessed in terms of other moral categories than being right (e.g., being supererogatory, being wrong), we should consider not only rightness functions but more generally N functions (where N is some normative property). Something like the following seems to be a descriptively adequate explication of the second wave’s concept of consequentialism: A (logically consistent) moral theory T is C-​outcome G-​consequentialist with respect to normative property N of acts if and only if T contains an instance of the following scheme: (C2)  For all acts x: The normative property N of x is determined by a function F of a ranking in terms of the (impersonal) value in some type C of outcomes of something related to act x, G(x) (where the ranking ranges over all alternatives to G(x))—​and by nothing else. If G(x) is different from x, we have an indirect version of consequentialism.35 For instance, let G(x) be the collective internalization of a set of rules (a code) that permits or forbids act x. Then actual-​outcome rule consequentialism with respect to the permissibility of acts claims that we first rank the collective internalization of the various codes in terms of the value of their actual outcomes, and that the rightness of each act is then fully determined by a function of this ranking. If, e.g., the function is a maximizing one, it picks the highest-​ranked code and says that x is permissible if and only if this code permits x. Three things are noteworthy about (C2):  first, the assessment was more or less focused on acts; second, outcomes of acts were ranked in terms of value or goodness; and third, value and goodness were (implicitly or explicitly) understood as impersonal, impartial, or agent-​neutral.36 All three preconceptions were relaxed later on 34 Perhaps not fully arbitrary. A plausible constraint on the function is that it is “increasing” in the sense that if O1 is a better outcome than O2, then the act that brings about O1 will not be deontically inferior to the act that brings about O2. 35 Of course, G(x) can also be the identity function, thus allowing for direct evaluations of actions in term of the outcomes they bring about. 36 See, e.g., Scheffler’s characterization of consequentialism as “a moral doctrine which says that the right act in any given situation is the one that will produce the best overall outcomes, as judged from an impersonal standpoint” (Scheffler, Introduction to Consequentialism and Its Critics, 1), where “the rankings generated by the designated principle are not agent-​relative; they do not vary from person to person” (Scheffler, The Rejection of Consequentialism, 1). Sen’s perceived need to distinguish between consequentialism and consequentialist evaluation (noted in note 22) also suggests that agent-​neutrality was seen as a defining mark of consequentialism.



New Wave Consequentialism    9

in response to the second wave’s other thread: intimately connected to the refined differentiation of the consequentialist “research program” was the critics’ endeavor to generalize the traditional objections and to detach them from the special features of utilitarianism.37 For instance, Foot noted that certain implausible implications—​ that one is obliged to torture one victim if this is the only way to prevent an evil person from torturing five victims—​are implications “not of utilitarianism in particular but rather of consequentialism in any form.”38 Scheffler added a further step of generalization by pointing out that this objection and other traditional problems (in particular issues of demandingness) are rooted in the agent-​centered character of commonsense morality:39 common sense seems to allow for agent-​centered restrictions or constraints like Foot’s (it is sometimes impermissible to bring about the impersonally best outcome) and agent-​centered prerogatives or options (it is sometimes permissible not to bring about the impersonally best outcome), but both are incompatible with the consequentialist view that it is always required to bring about the impersonally best outcome. But Scheffler also noted that, on the other hand, agent-​centered restrictions are puzzling or paradoxical.40 They claim that the minimization of morally objectionable conduct is morally unacceptable, which seems hard to square with an intuitive conception of practical rationality: that it is rational to choose the option that best accomplishes the goal one deems desirable.41 The upshot of the second wave was a growing awareness of a fundamental dialectical impasse between consequentialism and competing (commonsense) theories.42 On the one hand, consequentialism seems to comply with an appealing conception of practical rationality, which supports the compelling idea “that it can never be right to prefer a worse state of affairs to a better.”43 But on the other hand, consequentialism seems to clash with intuitive judgments rooted in the agent-​centered character of commonsense morality. And while theories that include agent-​centered elements can vindicate such judgments, they conflict with the appealing conception

Foot nicely summarizes the connection between the two threads when she claims that “the modifications have never been able to catch up with the objections” (“Utilitarianism and the Virtues,” 197). Of course, recasting the objections at higher levels of abstractions matched and reinforced the trend for an increasingly abstract understanding of consequentialism. 38 Foot, “Utilitarianism and the Virtues,” 198. 39 Scheffler, The Rejection of Consequentialism. 40 For versions of the paradox, see Nozick, Anarchy, State, and Utopia, 30; Scheffler, The Rejection of Consequentialism, 133f., 141–​143. 41 Scheffler’s own “hybrid” view therefore rejects agent-​ centered restrictions but accepts agent-​ centered prerogatives. Given his understanding of consequentialism as fully agent-​neutral, this makes his view nonconsequentialist. 42 Compare Scheffler, Introduction to Consequentialism and Its Critics. 43 Foot, “Utilitarianism and the Virtues,” 62. 37

10    Christian Seidel

of practical rationality and thus appear paradoxical. So the problem was seen as one of reconciling agent-​centeredness and practical rationality. 1.1.3 The New Wave: Abstraction, Reasons, and Consequentialism’s Conceptual Emancipation from Agent-​N eutrality and Value

This was the starting point for two further developments in the recent debate on consequentialism, a “resort to abstraction” and a “turn to reasons,” which have shifted the debate considerably both in focus and in style: First, in response to the dialectical impasse, consequentialists have quickly identified the culprit—​agent-​ neutrality44—​and tried to model agent-​centeredness within a consequentialist framework. Continuing the second wave’s trend of arming consequentialist theories with increasingly complex axiologies, some consequentialists have appealed to agent-​ relative axiologies (and corresponding rankings) instead. This way, they claimed, any moral theory—​in particular those with agent-​centered characteristics—​can be couched in consequentialist terms such that both the theory and its consequentialist counterpart yield exactly the same deontic verdicts.45 A technical maneuver, known as “consequentializing,” is doing the trick here:  “[T]‌ake whatever considerations that the non-​consequentialist theory holds to be relevant to determining the deontic statuses of actions and insist that those considerations are relevant to determining the proper ranking of outcomes. In this way, the consequentialist can produce an ordering of outcomes that when combined with her criterion of rightness yields the same set of deontic verdicts that the non-​consequentialist theory yields.”46 In order to properly model agent-​centeredness, the consideration that is built into the consequences is not the agent-​neutral fact that, say, a lie has been told, but the agent-​relative fact that I have told a lie. This will yield agent-​relative rankings of outcomes: the outcome O1 in which I rather than you told a lie is ranked lower, relative to me, than the outcome O2 in which you rather than I told a lie, while O2 is ranked lower, relative to you, than O1. Such a theory does not count as consequentialist See, e.g., “[C]‌ onsequentialism’s counter-​ intuitiveness derives from its agent-​ neutrality” (Portmore, “Combining Teleological Ethics with Evaluator Relativism,” 107). 45 This is known as the deontic equivalence thesis (Portmore, Commonsense Consequentialism, 85), the extensional equivalence thesis (Dreier, “In Defense of Consequentializing,” 98), the representation thesis (Oddie and Milne, “Act and Value,” 44), or the consequentialist umbrella (Louise, “Relativity of Value,” 518). 46 Portmore, “Consequentializing Moral Theories,” 39. This maneuver is based on earlier work: Sen, “Rights and Agency”; Sen, “Evaluator Relativity and Consequential Evaluation”; Vallentyne, “Gimmicky Representations of Moral Theories”; Oddie and Milne, “Act and Value”; Dreier, “Structures of Normative Theories.” For a detailed discussion, see Portmore, “Consequentializing.” 44



New Wave Consequentialism    11

on the second wave’s notion of consequentialism, but on the new, expanded notion it does. Consequentializing thereby contributed to consequentialism’s conceptual emancipation from agent-​neutrality, with which consequentialism has traditionally—​either implicitly or definitionally, as we have seen—​been associated throughout the second wave, such that most of the previous “discussion of consequentialism has really been a discussion of agent-​neutrality.”47 Consequentialism was thus conceptualized at a more abstract level. This tendency was reinforced by two simultaneous trends: (i) advancing Parfit’s thought that consequentialism applies to everything, some theories have changed their focal points of assessment and consequentialist evaluation from acts or rules to motives, virtues, decision procedures, or attitudes.48 And (ii) others have changed the mode of assessment from one-​to multidimensional assessments that take into account several irreducible moral aspects,49 or from a single evaluative ranking to a dual-​ranking structure.50 In both cases, consequentialism becomes even more flexible: changing the focal point enables consequentialism to assess not just acts but arbitrary entities X (e.g., motives, dispositions, or beliefs) in terms of outcomes of something that is related to these entities (e.g., the outcomes of general compliance with a rule that permits an action x, the outcomes of those acts that are triggered by motive x, the outcomes of having a disposition x, etc.). And changing the mode of assessment enables them to determine the normative property N by a more complex structure than a single ranking in terms of a single aspect (viz., the goodness or value of these outcomes). These two trends fueled the resort to abstraction: generalizing (C2) in the light of these two developments, we get the idea that a theory is consequentialist if it says something like this, (C3)  For all x that are X: The normative property N of x is determined by a function F of a set of rankings {R1, . . . , Rn} in terms of the value in some type C of outcomes of something related to x, G(x) (where the ranking ranges over all alternatives to G(x))—​and by nothing else. 47 Broome, Weighing Goods, 5. The case for emancipating consequentialism from agent-​neutrality is also made in Dreier, “Structures of Normative Theories”; Louise, “Relativity of Value.” 48 See Kagan, “Evaluative Focal Points”; Pettit and Smith, “Global Consequentialism”; Driver, Uneasy Virtue; Sverdlik, Motive and Rightness; Driver, Consequentialism. For a critique, see Streumer, “Can Consequentialism Cover Everything?” 49 Peterson, The Dimensions of Consequentialism. 50 See Portmore, “Dual-​Ranking Act-​Consequentialism,” which generalizes an idea of Sider, “Asymmetry and Self-​Sacrifice.” The point is, roughly, that in order to model, e.g., supererogation, the deontic status of an act must be determined by two rankings: one in terms of moral reason, and the other in terms of reason, all things considered. An action is permissible if and only if its outcome is not outranked by any other action on both rankings. This allows for cases in which a permissible act ϕ has an outcome that is ranked higher than the outcome of any other permissible action on the moral ranking only; then ϕ is supererogatory.

12    Christian Seidel

This resort to abstraction has not been without critics. They have argued that one of its driving forces, consequentializing, does not fully succeed in establishing deontic equivalence with just any theory,51 or that even if it succeeds, it is a “gimmicky”52 rather than “promising”53 result with no significant implications for selecting moral theories.54 One influential line of criticism pointed out that the account of value or goodness that the resort to abstraction predominantly appeals to (in particular in consequentializing agent-​centered theories) is defective, incoherent, and incompatible with consequentialism’s compelling idea: that it is always permissible to bring about the most good—​and not the most good-​relative-​to someone.55 The other development triggered by the dialectical impasse in the second wave has been a “turn to reasons” within the recent debate on consequentialism:56 being in search of a reconciliation of agent-​centeredness with an intuitively compelling account of practical rationality, nonconsequentialists have analyzed more closely the relation between moral requirements, practical reasons, and rationality and have found the allegedly intuitively compelling account wanting, thereby forcing consequentialists onto the defensive.57 In keeping with the overarching paradigm in moral philosophy to frame issues in terms of reasons, consequentialists have reformulated their accounts in terms of (a teleological conception of ) reasons rather than value or goodness. For instance, the traditional compelling idea of consequentialism—​that it is always permissible to do what brings about the best state of affairs—​then becomes the claim that it is always permissible to do what brings about the outcome that one has better reasons to prefer above all other alternatives.58 The idea of an evaluative ranking is then explained as a ranking on Brown, “Consequentialize This.” 52 Nozick, Anarchy, State, and Utopia, 29n. 53 Portmore, “Combining Teleological Ethics with Evaluator Relativism.” 54 Sachs, “Consequentialism’s Double-​ Edged Sword”; Hurley, “Consequentializing and Deontologizing”; Tenenbaum, “The Perils of Earnest Consequentializing.” 55 Schroeder, “Not So Promising After All”; Schroeder, “Teleology, Agent-​Relative Value, and ‘Good.’ ” For replies, see Portmore, “Consequentializing Moral Theories”; Smith, “Two Kinds of Consequentialism”; Dreier, “In Defense of Consequentializing.” 56 The term “turn to reasons” most often denotes a comprehensive paradigm shift (triggered by developments in metaethics, the theory of normativity and the philosophy of action) to transform and discuss questions of moral philosophy in terms of practical reasons for action; see, e.g., Raz, Practical Reason and Norms; Parfit, Reasons and Persons; Dancy, Moral Reasons; Scanlon, What We Owe to Each Other; Skorupski, The Domain of Reasons; Parfit, On What Matters. Here, I use the term “turn to reasons” more narrowly to denote a change within the debate about consequentialism only. Plausibly, this turn mirrors the more comprehensive turn. For critical assessments on the more comprehensive turn, see Broome, “Reasons,” 28–​55; Thomson, Normativity, ch. 8; Väyrynen, “A Wrong Turn to Reasons?” 57 See, e.g., Hurley, Beyond Consequentialism. 58 Compare Portmore, “Consequentializing Moral Theories,” 50. In later work, Portmore suggests that the truly compelling idea of consequentialism is the combination of moral rationalism with a teleological conception of reasons (Commonsense Consequentialism, 33n18). 51



New Wave Consequentialism    13

the basis of the agent’s reasons for preferring outcomes,59 and the essence of (act-​) consequentialism becomes the claim that “an act’s deontic status is determined by the agent’s reasons for and against preferring its outcome to those of its alternatives, such that, if S is morally required to perform x, then, of all the outcomes that S could bring about, S has most reason to desire that x’s outcome obtains.”60 This account is intended to sidestep questions about agent-​relative value, which had been the focus of previous criticism of the consequentializing project: agent-​centeredness is modeled by appeal to agent-​relative reasons instead. Although it would be premature to speak of consequentialism’s conceptual emancipation from value, one might indeed wonder whether there remains any independent role for those concepts that were once essential to (if not defining of ) consequentialism—​value or goodness. 1.1.4 Taking  Stock

It seems that the dialectical impasse of the second wave has triggered an emerging new wave consequentialism that is characteristically associated with a “resort to abstraction” and a “turn to reasons.” In attempting to advance and reply to different objections, consequentialist moral theories are designed at much higher levels of abstraction, thus becoming extremely flexible and able to give more nuanced moral assessments of particular cases. Along the way, the concept of consequentialism itself has become increasingly general and abstract by continued incremental conceptual emancipations from certain characteristic marks.61 Accommodating the turn to reasons, we may introduce the variable V as the basis for the set of rankings to allow for the idea that outcomes are not ranked in terms of value but in terms of something else.62 If we modify (C3) correspondingly, then something like the following scheme seems to capture the abstract idea of consequentialism as it is employed in contemporary debates: A (logically consistent) moral theory T is V-based n-​rank C-​outcome G-​ consequentialist with respect to normative property N of entities X if and only if T contains an instance of the following scheme: Roughly, one state of affairs is better than another if the set of reasons to prefer the first is weightier than the set of reasons to prefer the latter. For a more precise explication, see Portmore, Commonsense Consequentialism, 35. 60 Portmore, Commonsense Consequentialism, 34. 61 As an indicator for the conceptual shift, note that Scheffler’s hybrid theory—​which is nonconsequentialist on his own understanding—​has been redescribed as falling under this new concept; see the discussion of “Schefflerian utilitarianism” in Portmore, Commonsense Consequentialism, 94–​98, 142–​149. 62 We could even allow for the idea that each ranking has its own basis; i.e., some rankings are in terms of value, others in terms of reasons, still others in terms of moral aspects (in the sense of Peterson, The Dimensions of Consequentialism). So we can think of V as a vector V = (V1, . . . , Vn) that gives each ranking Ri its basis Vi. 59

14    Christian Seidel

(C)  For all x that are X: The normative property N of x is determined by a function F of a set of rankings {R1, . . . , Rn} in terms of something, V, which is related to some type C of outcomes of something related to x, G(x) (where the ranking ranges over all alternatives to G(x))—​and by nothing else.63 To illustrate: Suppose X is the category of acts, N is moral permissibility, n = 1, C are actual (rather than expected or foreseeable) outcomes, and G(x) is the collective internalization of a set of rules (a code) that permits or forbids act x. Then a value-​ based single-​rank actual-​outcome rule-​consequentialist theory with respect to the permissibility of acts claims: (CR)  For all acts x: The moral permissibility of x is determined by a function F of a single ranking in terms of the value of the actual outcomes of the collective internalization of a code that permits or forbids act x (where the ranking ranges over all collective internalizations of alternative codes that permit or forbid acts x)—​and by nothing else. This says that we rank all alternative codes that permit or forbid act x according to the value that their collective internalization actually brings about. Then, the function F could work on that ranking as follows: it picks the highest-​ranked code G *(x) and says that x is morally permissible if and only if G *(x) permits x. (CR) is thus a very cumbersome way of saying that an act is morally permissible if and only if the best code (i.e., the code whose collective internalization actually brings about most good) permits it. A different function F—​say, one that picks the best five codes and declares x to be morally permissible if and only if one of these codes permits x—​ would yield something like a satisficing rather than a maximizing version of that theory. And by varying the ranking basis (the V-​component), outcomes can also be ordered in terms of reasons to prefer rather than in terms of value, which yields a reason-​based (probably even more cumbersome) version of that theory. So what, then, is consequentialism? It is the class of (logically consistent) theories that contain an instance of the schematic (C) for some V, n, C, G, N, X. Theories falling under this concept turn out to be enormously flexible, which apparently enables 63 I do not claim that this captures every use of the term “consequentialism” in the contemporary debate; it is meant to be an explication (in Carnap’s sense), so some theories might drop out. Nor can I even try to relate this explication to others in the literature. Note that consequentialists might disagree about which kind of entities outcomes are, about how to individuate those outcomes, and about the nature of the “is determined by” part (is it some sort of substantial determination, an explanation, or a purely extensional claim about certain properties being necessarily coinstantiated?). The corresponding terms in (C) are supposed to be neutral with respect to such disputes.



New Wave Consequentialism    15

them to give more nuanced moral assessments, to meet a number of objections and to retain attractive features. But while this abstract concept of consequentialism might be adequate, descriptively, in accounting for the factual use of the term “consequentialism” in contemporary debates, it is a different question whether it is also useful with regard to the theoretical purpose of moral theorizing:  if the number of possible consequentialist theories is far greater than typically assumed, then one might wonder whether the resort to abstraction and the turn to reasons will eventually render consequentialism devoid of substantive moral content and will make the distinction between consequentialist and nonconsequentialist theories dispensable. This is a serious worry, in particular in the debate on the merits and drawbacks of consequentializing. To approach this higher-​level methodological question, we must look at the details of new wave consequentialism. And it is here, at the level of details, that the new wave also raises distinct questions and problems of its own:  Is it possible to successfully consequentialize commonsense morality (and in particular agent-​centered restrictions and options) without inheriting the problems of deontological theories? Is the resulting account of practical reason(s) and rationality tenable? And if consequentialists do not focus on the assessment of actions only, what does a morally fitting character or psychology of the consequentialist agent look like? How does new wave consequentialism deal with “long-​standing” puzzles in moral philosophy (e.g., different number choices or predictable preference changes)? And how does it account for key parts of commonsense moral practice (such as holding people morally responsible or punishing them)? This book addresses these questions and helps to assess the prospects and limits of new wave consequentialism.

1.2. Navigating the New Wave

The volume is organized in three parts with different foci. 1.2.1 Directions: Exploring New Wave Consequentialism

The contributions to part I explore new directions in new wave consequentialism. By drawing on insights from debates in value theory, moral psychology, and the foundations of normativity, they present refined conceptual frameworks that exemplify and add to (new wave) consequentialism’s enormous flexibility. While this flexibility so far mainly derives from the possibility of agent-​centered rankings of outcomes (rooted in appeals to agent-​relative value or agent-​relative reasons), the

16    Christian Seidel

contributions go beyond this and identify other resources to accommodate common sense, to meet common objections, and to address puzzling ethical questions. In his essay, Jamie Dreier generalizes the established concepts of agent-​centered and time-​centered value (which are crucial to the consequentializing project) and introduces a new, abstract concept—​world-​centered value—​as an extension of these. We use it when we hold, for example, that judged from the perspective of world A, A outranks a world B, while judged from the perspective of B, B outranks A. Dreier first shows how to make sense of this concept: while the standard account models agent-​centered (and time-​centered) views as generating different rankings of states of affairs, one for each agent (or for each time), Dreier suggests that we can also model these views as generating a single ranking of properties a person (or time) can have. On his account, the centering is built into the entities that are ranked: outcomes are understood as properties of an agent (or time), and these properties are modeled as sets of “centered worlds” (which are ordered pairs, , where a world is just a way the world can be, and the center picks out an agent or a time). And just as we can center value on agents or times, we can center value on worlds when judging something—​generating a different ranking for each world (as in the example above). Second, Dreier explores the benefits of centering value on worlds. By showing how the concept helps to illuminate consequentializers’ take on prohibition dilemmas (in which each of the available acts is impermissible) or cases from population ethics (in which the betterness relation appears to be intransitive), he concludes that although of limited practical value, the concept of world-​centered value plays a theoretical role important enough to give it some space in the consequentialist’s toolbox. Shedding new light on population ethics is also a major concern of the next essay. To accommodate common sense, Martin Peterson has developed a strategy that is very different from standard consequentialist approaches insofar as it changes the mode of consequentialist assessment:64 according to his multidimensional consequentialism, the deontic status of an act depends not on a single moral aspect (e.g., happiness) but on multiple irreducible moral aspects. In his contribution to this volume, Peterson explains and defends his theoretical approach, and—​like Dreier—​ explores its performance in the context of core issues of population ethics. He argues that multidimensional consequentialists can reasonably reject some of the key premises of puzzles about different number choices (such as the mere addition paradox): the mere addition of people (with lives worth living) is right with respect to one moral aspect, but it is not always entirely right, because it may be wrong with respect to another moral aspect.

Peterson, The Dimensions of Consequentialism. 64



New Wave Consequentialism    17

Instead of changing the mode of consequentialist assessment, there is another route to increased flexibility, as we have seen: changing the focal points of assessment from acts (or rules) to motives, virtues, decision procedures, or attitudes. This line is pursued by Douglas W.  Portmore in his contribution to the volume. He argues that traditional consequentialism has difficulty accounting for coordination problems—​cases in which the value of an action’s consequences depends on what the agent and others agents have done or will be doing. The reason is that in these cases, the relevant factor in determining whether the agent ought to ϕ is whether she has rational control over the other action—​a sort of control that we have over our reasons-​responsive yet nonvoluntary attitudes (e.g., beliefs, desires, intentions) and the things (e.g., voluntary acts) we influence via such attitudes. But if what an agent ought to do in coordination problems depends on what attitudes she ought to form, traditional consequentialism (which is mainly concerned with voluntary actions) has a problem. Portmore suggests an alternative, which is concerned not only with voluntary actions but also with nonvoluntary attitudes. According to rationalist consequentialism, whether an agent ought to ϕ depends on whether she would ϕ if she were now to form and perform all the acts and attitudes that she is required to form and perform. In shifting the focus from actions to attitudes, rationalist consequentialism may help to make consequentialism even more commonsensical. A question that naturally arises from such a turn to attitudes is: What will a morally fitting consequentialist agent look like, which attitudes (or psychology more broadly) will she have? This is the question Richard Yetter Chappell addresses in his essay. He, too, thinks that consequentialists should do more than just assess the deontic status of acts, because new work in the foundations of ethics (which extends the fitting attitudes analysis of value to yield a more general notion of normative fittingness) suggests that moral theories have implications for what will qualify as morally fitting attitudes (or as a morally fitting psychology more broadly)—​ implications that are relevant for assessing the truth of the theories. And there is a challenge lurking here for consequentialism:  Chappell shows that traditional character-​based objections to consequentialism can be revamped as fittingness objections, and that both standard and sophisticated consequentialist responses to these renewed objections fail. He goes on to argue that consequentialism can successfully be defended against paradigmatic fittingness objections by appealing to “well-​ calibrated” dispositions (which are desirable for their impact on the actions they tend to produce) and by distinguishing motivating from guiding dispositions: according to the objections, consequentialism has fittingness implications that involve defective guiding dispositions—​but in fact, consequentialist fitting motivating dispositions are perfectly compatible with truly fitting well-​calibrated guiding

18    Christian Seidel

dispositions. What is more, in suggesting a new argument against rule consequentialism, fittingness implications might even be helpful in settling disputes within the consequentialist camp. 1.2.2 Problems: Scrutinizing the New Wave’s Theoretical Basis

While part I presents refined conceptual frameworks for new wave consequentialism, the critical chapters in part II raise challenging fundamental problems for these frameworks, i.e., problems that target new wave consequentialism’s theoretical basis. As noted earlier, at the heart of this basis lies an ambitious maneuver to accommodate consequentialism with commonsense moral intuitions by “consequentializing” deontological theories (which are understood as reflecting common sense). This maneuver renders consequentialism very flexible, up to the point where any deontological theory is sucked up in “the consequentialist vacuum cleaner.”65 But when we have a closer look at the details, is this alleged flexibility a feature or a bug? In their critical discussion of consequentializing, Monika Betzler and Jörg Schroth opt for the latter. They argue that the consequentializing project is dialectically committed to claim that any (commonsensically plausible) deontological theory can be translated into a deontically equivalent consequentialist theory T which satisfies three criteria of adequacy:  (i) T retains the compelling idea of consequentialism (that it can never be right to prefer a worse state of affairs to a better one); (ii) T has no counterintuitive moral implications; and (iii) T avoids the paradox of agent-​ centered restrictions. Betzler and Schroth argue that consequentialized deontological theories fail to achieve this goal: since deontological intuitions are at odds with the compelling idea if the good is characterized agent-​neutrally, these theories have to abandon the original compelling idea; but the modified ideas that they appeal to instead—​an agent-​relative axiology or an agent-​relative interpretation of the teleological conception of reasons—​cease to be compelling, because they can no longer draw on the rationale that makes the original compelling idea so attractive: that the point of morality is to make the world as good a place as possible, agent-​neutrally conceived. Moreover, in appealing to an agent-​relative ranking of outcomes that is dependent on deontological intuitions, consequentialized theories provide no better moral orientation than deontological theories but have to swallow the latter’s disadvantages. So even if we accept that we can model agent-​centeredness within a consequentialist framework, it is still an open question why we should choose agent-​ centered consequentialist theories: What is their competitive advantage over deontological theories? McNaughton and Rawling, “Agent-​Relativity and the Doing-​Happening Distinction,” 168. 65



New Wave Consequentialism    19

While Betzler and Schroth focus on the general theoretical ambitions of consequentializing deontological theories, the next pair of chapters deals more specifically with the two distinctively deontological features of commonsense morality that were at the center of the second wave: agent-​centered restrictions and agent-​ centered options. In his essay, Jan Gertken addresses the attempt to consequentialize the former. He argues that the standard view, according to which the appeal to agent-​relative value is essential to a consequentialist account of restrictions—​a view that is widely shared among new wave consequentialist and their critics and that is also implicit in Betzler and Schroth’s arguments—​is mistaken: what matters, rather, is an agent-​centered conception of outcomes. Gertken shows (i) that by employing such a conception, it is possible to consequentialize restrictions without appealing to agent-​relative value, and (ii) that the attempt to consequentialize restrictions by appealing to agent-​relative value fails unless it employs in addition an agent-​centered conception of outcomes. Does this give new wave consequentialism another, more promising strategy for accommodating restrictions? In exploring how an agent-​ centered conception of outcomes can be supported and developed,66 Gertken remains skeptical about the merits and prospects of this strategy. Dale Dorsey draws a similarly skeptical conclusion with regard to options, albeit for different reasons. He argues that in trying to accommodate options, new wave consequentialism is committed to moral rationalism: the view that we are normatively required (or have most reason, all things considered) to do what we are morally required to do. But according to Dorsey, moral rationalism is (probably) false: since it is not a priori, it can be established only by substantive first-​order inquiry into the moral domain. This means that the truth of moral rationalism depends on the existence of moral options, and so justifying moral rationalism requires independent arguments for the existence of moral options. But given that any plausible rationale for moral options depends on moral rationalism in the first place, there are no such independent arguments. And if moral rationalism is (probably) false, then the project of consequentializing options (probably) fails, because it lacks a plausible rationale. A powerful argument for new wave consequentialism relies on moral rationalism combined with the so-​called teleological conception of reasons:67 if (i) the deontic status of ϕ is determined by the agent’s reasons for ϕ (so being morally required to ϕ implies having most reason to ϕ), and if (ii) the agent’s reasons for ϕ are determined Both Dreier and Gertken examine how to center value on agents by revising what counts as an outcome, but their suggestions are quite different: while Gertken continues to understand outcomes as states of affairs (and employs an agent-​centered filter on states of affairs), Dreier revises the type of entity that constitutes outcomes—​they are properties that a person might have. 67 See Portmore, Commonsense Consequentialism, 6f., 33f. 66

20    Christian Seidel

by her reasons for preferring ϕ’s outcome to those of the alternatives (so having most reason to ϕ implies having most reason to desire that ϕ’s outcome obtains), then the deontic status of ϕ is determined by the agent’s reasons for preferring ϕ’s outcome to those of the alternatives (i.e., being morally required to ϕ implies having most reason to desire that ϕ’s outcome obtains). While Dorsey’s essay brings up a justificatory challenge for the first main tenet of new wave consequentialism, moral rationalism, Paul Hurley levies an attack on the second main tenet, the teleological conception of reasons.68 In his essay, he focuses on what may be taken to form the basis69 for both restrictions and options:  the commonsense doing-​allowing distinction. In contrast to traditional consequentialism, new wave consequentialism frequently endorses this distinction and the associated claim that we sometimes have (decisive) moral reasons not to ϕ, although this allows some (agent-​neutrally) worse outcome to happen. But Hurley shows that the teleological conception of reasons commits new wave consequentialists to reject another commonsensical doing-​allowing distinction: that we sometimes have (decisive) reasons not to ϕ, although this allows some (evaluator-​relatively) worse outcome to happen. Hurley identifies a deeper dispute here: the teleological conception of reasons is grounded in certain outcome-​ centered accounts (of values, attitudes, and actions), while alternative accounts do not support but ultimately undermine the teleological conception of reasons (and hence the rejection of the new doing-​allowing distinction). Therefore, the status of the distinction as well as the prospects of new wave consequentialism depend on adjudicating between these competing accounts at a deeper, fundamental level outside the domain of normative ethics. Tim Henning further develops the line of fundamental critique initiated by Dorsey and Hurley. He challenges a seemingly intuitive picture of rationality that is often thought to support both traditional and new wave consequentialism.70 According to this picture (and the ensuing argument), consequentialist reasoning in ethics fits best with our convictions about rationality in nonmoral contexts, since it is structurally analogous to rational decision-​making in everyday life: both are forms of goal-​oriented, teleological decision-​making. But in attending to cases of predictable preference accommodation—​in which an agent can predict that performing ϕ (e.g., having a baby) will change her preferences such that she will like ϕ’s outcome 68 The locus classicus for (a critical discussion of ) the teleological conception of reasons is Scanlon, What We Owe to Each Other, 79–​87. Scanlon states the conception in terms of desirability or value of outcomes. For a discussion of how this relates to the statement in terms of reasons, see Portmore, Commonsense Consequentialism, ch. 3, esp. 58–​62. 69 For this view, see Kagan, The Limits of Morality, chs. 3 and 5. 70 As an anonymous reviewer suggested, this picture of rationality might be the main reason that stops new wave consequentialists from simply becoming nonconsequentialists.



New Wave Consequentialism    21

best—​Henning shows that the rationality of a choice is not always determined by features of the outcome of the chosen act.71 He discusses and rejects several consequentialist attempts to come to terms with cases of predictable preference accommodation (including Dreier’s proposal in his contribution to this volume). But if outcome-​based reasons cannot account for the rationality of choices in these cases, moral consequentialism is not just an obvious extension of an allegedly uncontroversial account of rational decision-​making. By contrast, Henning thinks that an adequate account of rationality might be nonconsequentialist in spirit by requiring a Kantian notion of respect. 1.2.3 Scope: Assessing New Wave Consequentialism in Context

While part II aims at new wave consequentialism’s theoretical basis, the essays in part III give a balanced assessment of new wave consequentialism’s limits and achievements in specific contexts of commonsense moral practice (such as holding people morally responsible or punishing them). Being sympathetic in spirit to consequentialist outlooks, these context-​specific discussions suggest unexpected revisions in our understanding of consequentialism’s essence (and of its relation to competing accounts). They also point to new paths consequentialists might explore to answer challenges from these contexts. Insofar as these paths go beyond “standard” new wave consequentialism (or even return to more traditional forms of consequentialism), the contributions to this part ask us to reconsider both the turn to reasons and the resort to abstraction that have been the driving forces behind new wave consequentialism. In her essay, Elinor Mason approaches a couple of common objections to consequentialism (concerns about demandingness, integrity, and epistemic requirements) from a fresh perspective. In contrast to the new wave’s attempts to answer these worries by a number of abstract maneuvers, Mason argues that consequentialists can silence these objections by attending to a hitherto neglected constraint on all moral theories: the responsibility constraint. It says that what is right or wrong must be something agents could reasonably be deemed responsible for. Mason shows, first, how traditional objections to consequentialism can be reconceptualized as appealing to what agents can reasonably be held responsible for and, second, how (both new wave and traditional) consequentialism can avoid the objections by embracing the responsibility constraint. Picking up the thread from Hurley’s contribution, her

In this sense, Henning’s argument is also directed against the teleological conception of reasons, at least if we understand rationality as essentially connected to (correctly responding to) reasons. 71

22    Christian Seidel

approach implies that consequentialism can accommodate a doing/​allowing distinction without building agent-​relativity directly into the theory. Assessing consequentialism (as a comprehensive moral theory) in the context of a specific domain of moral practice is the concern of the last essay, too. Historically, consequentialism (in particular utilitarianism) has often been linked with a special interest in designing social institutions.72 In his essay, Steven Sverdlik focuses on the institution of punishment and defends a consequentialist take on it against a Kantian objection: the consequentialist claim that punishment may permissibly be used to deter people from committing crimes has often been criticized for treating the offender simply as a means to benefit others. Sverdlik shows that the best interpretation of the formula of humanity (upon which this objection rests) does not defeat consequentialist approaches to punishment or support their retributivist alternatives. By showing that Kantianism and consequentialism do not fundamentally differ on the moral legitimacy of deterrence, Sverdlik contributes to a recurrent theme in debates on consequentialism (which is also a concern of Henning’s contribution): consequentialism’s relation to and delimitation from Kantianism. Since in this context consequentialism is superior to Kantianism in not setting an implausible ceiling on the severity of punishments, Sverdlik believes that consequentialism can resume its former position as the leading account of the moral foundations of punishment. 1.3. Prospects: Beyond the New Wave

Taken together, the contributions to this volume provide a detailed philosophical investigation of key issues in contemporary new wave consequentialism and a rigorous, informative analysis of the state of the (consequentialist) art. At the same time, they also point beyond that new wave insofar as several contributions pursue overlapping themes and identify a couple of emerging cross-​sectional topics for further exploration. For instance, questions about what counts as the outcome of an agent’s action and which kind of entities is the bearer of value—​which have been discussed in previous waves of the debate—​connect the essays of Dreier, Gertken, and Hurley; this might indicate the need for a more prominent examination of whether new wave consequentialism is compatible with just any stance on these fundamental questions. Similarly, how best to accommodate other people’s (in contrast to the agent’s own) actions in a consequentialist framework will be crucial to

Recall Bentham’s panopticon, Mill’s dedication to emancipatory reforms, and that even Sidgwick conceded that “[a]‌sincere Utilitarian . . . is likely to be an eager politician” (The Methods of Ethics, 495). 72



New Wave Consequentialism    23

deal with coordination problems (Portmore), restrictions (Gertken), and issues about responsibility (Mason): According to new wave consequentialists, should the agent disregard other people’s actions altogether (and redescribe the outcomes), or treat them as something she can control and is responsible for? Detailed further investigations are needed here as well as with respect to the most convincing combination of agent-​centered and time-​centered value (a challenge that is illustrated by the exchange between Dreier and Henning on predicted preference accommodation). And finally, Portmore’s and Chappell’s contributions can be understood as inviting new wave consequentialists to develop their paradigm with a distinctive focus on attitudes rather than acts. But perhaps, rather than evolving along the path of further sophisticated differentiation, the paradigm itself will be shifting:  New wave consequentialism has essentially been concerned with modeling agent-​centeredness through conse­ quentializing. While the consequentializing maneuver raises a number of already hotly debated questions (e.g., about whether it works at all, whether it is of any use, and whether there are good arguments for accepting agent-​centered consequentialist rather than deontological views), there are at least two deeper issues emerging from the discussion in this volume—​issues that call the new wave’s paradigm into question: First, insofar as many contributions from parts I and II either defend or challenge attempts to bring consequentialism closer to commonsense morality—​which is the declared aim of consequentializing—​there seems to be a tacit agreement that consequentialism is tenable only if it can successfully accommodate commonsense morality (and those of its features often championed by nonconsequentialists, like agent-​centeredness, pluralism, discontinuities in aggregation and weighing, etc.).73 But this ignores the other escape route from the dialectical impasse: that of challenging commonsense morality, and in particular challenging the coherence, reliability, and rationality of its agent-​centered elements. Ever since Sidgwick’s Methods of Ethics, this has been a major concern and a persistent motif in debates on consequentialism (often supporting defenses of counterintuitive normative implications), and attending to it again may help to advance the debate on new wave consequentialism in another direction, by addressing more directly the merits or drawbacks of agent-​centered and agent-​neutral approaches. Second, the critical assessments in part II suggest that the consequentializing maneuver also raises very substantial questions about key concepts (such as value, desire, reasons, rationality, actions, outcomes, etc.), and here, the fundamental worry is that the new wave’s answer to these questions might not be compatible with consequentialism’s other (perhaps compelling) theoretical tenets. But the I owe this observation to an anonymous reviewer. 73

24    Christian Seidel

question is: the theoretical tenets of which consequentialism? If consequentializing works, there are really two kinds of consequentialism involved here:  traditional (agent-​ neutral) consequentialism and new wave (agent-​ centered) consequentialism. Both kinds have quite different commitments on substantial issues about value, reasons, rationality, actions, etc. To point out that new wave (agent-​centered) consequentialism’s commitments are incompatible with some of the tenets of traditional (agent-​neutral) consequentialism is not a defeating objection, but just shows that you cannot have the attractive features of both. So again, one possible future direction to continue (and perhaps advance) the discussion is to address these commitments more directly and assess their respective merits and drawbacks explicitly in terms of a debate between agent-​centered and agent-​neutral theories, rather than in terms of consequentialism versus nonconsequentialism. Remembering the theoretical purpose of terms of art, the former distinction may turn out to be more useful—​from a theoretical point of view—​in moral theorizing74 by shifting the issue from (reductionist) metaphysical claims about which kind of entities determine rightness and wrongness to structural claims about which kind of deontic verdicts a theory is willing to accept. Finally, some contributions point beyond the new wave in suggesting another kind of reorientation that is motivated from a more practical point of view:  the essays in part III focus on important domains of our moral practice and (implicitly) ask which kind of work a (consequentialist) theory is supposed to do in these domains. And as Mason and Sverdlik suggest, it might turn out that when properly understood, more traditional consequentialist approaches can get us pretty far. It is thus an open question whether the new wave’s turn to reasons and resort to abstraction are helpful in addressing, from a consequentialist perspective, those questions that really matter to us. It would therefore be premature to draw definite conclusions about the new wave’s prospects: Will we witness an even more abstract and sophisticated differentiation of the project, a paradigm-​shifting reorientation of the discussion in terms of agent-​centered versus agent-​neutral (rather than consequentialist versus nonconsequentialist) theories, or a return to more traditional consequentialist approaches? One thing, though, seems rather sure: as the new wave rolls, consequentialism remains a focal point of discussion and a driving force behind developments in normative ethics.

As has been suggested by others, e.g., Piper, “A Distinction without a Difference”; Dreier, “Structures of Normative Theories”; Louise, “Relativity of Value”; Dreier, “In Defense of Consequentializing”; Dougherty, “Agent-​Neutral Deontology.” For a related proposal on abandoning the term “deontology,” see Timmermann, “What’s Wrong with ‘Deontology’?” 74



New Wave Consequentialism    25

References Adams, Robert Merrihew. “Motive Utilitarianism.” Journal of Philosophy 73 (1976): 467–​481. Anscombe, G. E. M. “Modern Moral Philosophy.” Philosophy 33 (1958): 1–​19. Bentham, Jeremy. An Introduction to the Principles of Morals and Legislation (1789), edited by J. H. Burns and H. L. A. Hart, with a new introduction by F. Rosen. Oxford: Oxford University Press, 1996. Brandt, Richard B. A Theory of the Good and the Right. Oxford: Oxford University Press, 1979. Broome, John. “Reasons.” In Reason and Value:  Themes from the Moral Philosophy of Joseph Raz, edited by R. Jay Wallace, Philip Pettit, Samuel Scheffler, and Michael Smith, 28–​55. Oxford: Oxford University Press, 2004. Broome, John. Weighing Goods: Equality, Uncertainty and Time. Oxford: Basil Blackwell, 1991. Broome, John. Weighing Lives. Oxford: Oxford University Press, 2004. Brown, Campbell. “Consequentialize This.” Ethics 121 (2011): 749–​771. Bykvist, Krister. Utilitarianism: A Guide for the Perplexed. London: Continuum, 2010. Carlson, Erik. Consequentialism Reconsidered. Dordrecht: Kluwer, 1995. Dancy, Jonathan. Moral Reasons. London: Blackwell, 1993. Diamond, Cora. “Consequentialism in Modern Moral Philosophy and in ‘Modern Moral Philosophy.’” In Human Lives: Critical Essays on Consequentialist Bioethics, edited by David S. Oderberg and Jaqueline A. Laing, 13–​38. Houndmills, Basingstoke, UK: Macmillan, 1997. Dougherty, Tom. “Agent-​Neutral Deontology.” Philosophical Studies 163 (2013): 527–​537. Dreier, James. “Structures of Normative Theories.” The Monist 76 (1993): 22–​40. Dreier, Jamie. “In Defense of Consequentializing.” In Oxford Studies in Normative Ethics, vol. 1, edited by Mark Timmons, 97–​119. Oxford: Oxford University Press, 2011. Driver, Julia. Uneasy Virtue. Cambridge, UK: Cambridge University Press, 2001. Driver, Julia. Consequentialism. London: Routledge, 2012. Eyal, Nir. “Non-​Consequentialist Utilitarianism.” Éthique et économique/​Ethics and Economics 11 (2014): 34–​53. Feldman, Fred. Pleasure and the Good Life: Concerning the Nature, Varieties, and Plausibility of Hedonism. Oxford: Oxford University Press, 2004. Foot, Philippa. “Utilitarianism and the Virtues.” Mind 94 (1985): 196–​209. Griffin, James P. Well-​Being: Its Meaning, Measurement and Moral Importance. Oxford: Oxford University Press, 1986. Hare, Richard M. “Ethical Theory and Utilitarianism.” In Contemporary British Philosophy, edited by H. D. Lewis, 113–​131. London: Allen & Unwin, 1976. Hare, Richard M. Moral Thinking:  Its Levels, Method and Point. Oxford:  Oxford University Press, 1981. Hooker, Brad. Ideal Code, Real World: A Rule-​Consequentialist Theory of Morality. Oxford: Oxford University Press, 2000. Hooker, Brad, Elinor Mason, and Dale E. Miller, eds. Morality, Rules, and Consequences:  A Critical Reader. Edinburgh: Edinburgh University Press, 2000. Hurley, Paul. Beyond Consequentialism. Oxford: Oxford University Press, 2009. Hurley, Paul. “Consequentializing and Deontologizing:  Clogging the Consequentialist Vacuum.” In Oxford Studies in Normative Ethics, vol. 3, edited by Mark Timmons, 123–​153. Oxford: Oxford University Press, 2013.

26    Christian Seidel Jacobson, Daniel. “Utilitarianism without Consequentialism:  The Case of John Stuart Mill.” Philosophical Review 117 (2008): 159–​191. Kagan, Shelly. The Limits of Morality. Oxford: Oxford University Press, 1989. Kagan, Shelly. “Evaluative Focal Points.” In Morality, Rules and Consequences: A Critical Reader, edited by Brad Hooker, Elinor Mason, and Dale E. Miller, 134–​155. Edinburgh: Edinburgh University Press, 2000. Louise, Jennie. “Relativity of Value and the Consequentialist Umbrella.” Philosophical Quarterly 54 (2004): 518–​536. McNaughton, David, and Piers Rawling. “Agent-​ Relativity and the Doing-​ Happening Distinction.” Philosophical Studies 63 (1991): 167–​185. Mill, John Stuart. “Bentham” (1838). In The Collected Works of John Stuart Mill, vol. 10: Essays on Ethics, Religion and Society, edited by John M. Robson, 75–​116. Toronto:  University of Toronto Press, 1985. Mill, John Stuart. Utilitarianism (1861). In The Collected Works of John Stuart Mill, vol. 10: Essays on Ethics, Religion and Society, edited by John M. Robson, 203–​260. Toronto: University of Toronto Press, 1985. Moore, G. E. Ethics:  The Nature of Moral Philosophy (1912), edited by William H. Shaw. Oxford: Oxford University Press, 2005. Mulgan, Tim. The Demands of Consequentialism. Oxford: Oxford University Press, 2001. Nagel, Thomas. “The Limits of Objectivity.” In The Tanner Lecture on Human Values, vol. 1, edited by S. M. McMurrin, 75–​139. Salt Lake City: University of Utah Press, 1980. Nagel, Thomas. The View from Nowhere. Oxford: Oxford University Press, 1986. Nozick, Robert. Anarchy, State, and Utopia. Oxford: Blackwell, 1974. Oddie, Graham, and Peter Milne. “Act and Value: Expectation and the Representability of Moral Theories.” Theoria 57 (1991): 42–​76. Parfit, Derek. Reasons and Persons. Oxford: Oxford University Press, 1984. Parfit, Derek. “Equality and Priority.” Ratio 10 (1997): 202–​221. Parfit, Derek. On What Matters. Vol. 1. Oxford: Oxford University Press, 2011. Peterson, Martin. The Dimensions of Consequentialism:  Ethics, Equality and Risk. Cambridge, UK: Cambridge University Press, 2013. Pettit, Philip, and Michael Smith. “Global Consequentialism.” In Morality, Rules and Consequences: A Critical Reader, edited by Brad Hooker, Elinor Mason, and Dale E. Miller, 121–​133. Edinburgh: Edinburgh University Press, 2000. Piper, Adrian M.  S. “A Distinction without a Difference.” Midwest Studies in Philosophy 7 (1982): 403–​435. Portmore, Douglas W. “Combining Teleological Ethics with Evaluator Relativism: A Promising Result.” Pacific Philosophical Quarterly 86 (2005): 95–​113. Portmore, Douglas W. “Consequentializing Moral Theories.” Pacific Philosophical Quarterly 88 (2007): 39–​73. Portmore, Douglas W. “Dual-​ Ranking Act-​ Consequentialism.” Philosophical Studies 138 (2008): 409–​427. Portmore, Douglas W. “Consequentializing.” Philosophy Compass 4 (2009): 329–​347. Portmore, Douglas W. Commonsense Consequentialism:  Wherein Morality Meets Rationality. Oxford: Oxford University Press, 2011.



New Wave Consequentialism    27

Railton, Peter. “Alienation, Consequentialism, and the Demands of Morality.” Philosophy and Public Affairs 13 (1984): 134–​171. Rawls, John. A Theory of Justice. Cambridge, MA: Harvard University Press, 1971. Raz, Joseph. Practical Reason and Norms. London: Hutchinson, 1975. Ross, W. D. The Right and the Good (1930), edited by Philip Stratton-​Lake. Oxford:  Oxford University Press, 2002. Sachs, Benjamin. “Consequentialism’s Double-​Edged Sword.” Utilitas 22 (2010): 258–​271. Scanlon, T. M. What We Owe to Each Other. Cambridge, MA: Harvard University Press, 1998. Scheffler, Samuel, ed. Consequentialism and Its Critics. Oxford: Oxford University Press, 1988. Scheffler, Samuel. Introduction to Consequentialism and Its Critics, edited by Samuel Scheffler, 1–​13. Oxford: Oxford University Press, 1988. Scheffler, Samuel. The Rejection of Consequentialism:  A Philosophical Investigation of the Considerations Underlying Rival Moral Conceptions (1982). Revised edition. Oxford: Oxford University Press, 1994. Schroeder, Mark. “Not So Promising After All: Evaluator-​Relative Teleology and Common-Sense Morality.” Pacific Philosophical Quarterly 87 (2006): 348–​356. Schroeder, Mark. “Teleology, Agent-​Relative Value, and ‘Good.’” Ethics 117 (2007): 265–​295. Sen, Amartya. “Utilitarianism and Welfarism.” Journal of Philosophy 76 (1979): 463–​489. Sen, Amartya. “Rights and Agency.” Philosophy and Public Affairs 11 (1982): 3–​39. Sen, Amartya. “Evaluator Relativity and Consequential Evaluation.” Philosophy and Public Affairs 12 (1983): 113–​132. Sen, Amartya. The Idea of Justice. Cambridge, MA: Harvard University Press, 2009. Sen, Amartya, and Bernard Williams. “Introduction: Utilitarianism and Beyond.” In Utilitarianism and Beyond, edited by Amartya Sen and Bernard Williams, 1–​21. Cambridge, UK: Cambridge University Press, 1982. Sen, Amartya, and Bernard Williams, eds. Utilitarianism and Beyond. Cambridge, UK: Cambridge University Press, 1982. Sider, Theodore. “Asymmetry and Self-​Sacrifice.” Philosophical Studies 70 (1993): 117–​132. Sidgwick, Henry. The Methods of Ethics (1907), 7th ed., edited by John Rawls. Indianapolis: Hackett, 1971. Skorupski, John. The Domain of Reasons. Oxford: Oxford University Press, 2010. Slote, Michael. Common-​sense Morality and Consequentialism. London: Routledge & Kegan Paul, 1985. Smart, J. J. C., and Bernard Williams. Utilitarianism: For and Against. Cambridge, UK: Cambridge University Press, 1973. Smith, Michael. “Two Kinds of Consequentialism.” Philosophical Issues 19 (2009): 257–​272. Sosa, David. “Consequences of Consequentialism.” Mind 102 (1993): 101–​122. Streumer, Bart. “Can Consequentialism Cover Everything?” Utilitas 15 (2003): 237–​247. Sumner, L. W. Welfare, Happiness, and Ethics. Oxford: Oxford University Press, 1996. Sverdlik, Steven. Motive and Rightness. Oxford: Oxford University Press, 2011. Tenenbaum, Sergio. “The Perils of Earnest Consequentializing.” Philosophy and Phenomenological Research 88 (2014): 233–​240. Thomson, Judith Jarvis. Normativity. Chicago: Open Court, 2008. Timmermann, Jens. “What’s Wrong with ‘Deontology’?” Proceedings of the Aristotelian Society 115 (2015): 75–​92.

28    Christian Seidel Vallentyne, Peter. “Utilitarianism and the Outcomes of Actions.” Pacific Philosophical Quarterly 68 (1987): 57–​70. Vallentyne, Peter. “Gimmicky Representations of Moral Theories.” Metaphilosophy 19 (1988): 253–​263. Väyrynen, Pekka. “A Wrong Turn to Reasons?” In New Waves in Metaethics, edited by Michael Brady, 185–​207. Houndmills, Basingstoke, UK: Palgrave-​Macmillan, 2011. Williams, Bernard. “A Critique of Utilitarianism.” In Utilitarianism: For and Against, by J. J. C. Smart and Bernard Williams, 75–​150. Cambridge, UK: Cambridge University Press, 1973.

I

Directions EXPLORING NEW WAVE CONSEQUENTIALISM

2 WO R L D -​C ENT E RE D  VA LUE

Jamie Dreier

The term “consequentialist” is sometimes reserved for theories like utilitarianism that are agent-​neutral and temporally neutral, but in recent years its meaning has been expanded to include theories that retain some of the structure of common sense that utilitarianism leaves behind.1 These structural features include centering on agents and centering on times:  the way to accommodate them in the consequentialist framework is to allow for agent-​centered value and time-​centered  value. I have a concept, which I’m calling “world-​centered value,” that seems to be a natural extension of some concepts we already have in ethical theory, and also seems useful for understanding some confusing ideas I’ve found in the work of other philosophers. I’ll be explaining it as well as I can and employing it to explicate the confusing ideas. If I am right that this concept makes sense, and that it can be useful in understanding the ideas by placing them in the consequentialist framework, then world-​centered value earns its keep and should find a place next to agent-​centered and time-​centered value in the ethical theorist’s toolbox.

1 For the older use, see, for example, Kagan, The Limits of Morality. For the more expansive, see Dreier, “Structures of Normative Theories.” The most elaborate use of the kind of centering I  have in mind is in Portmore, Commonsense Consequentialism.

31

32    Jamie Dreier

I will not be claiming or arguing that some world-​centered theory is true. Instead, my aim is to show that some theoretical thoughts about the world of value can be made sense of by understanding them to be claims about world-​centered value. The first section discusses a suggestion of Doug Portmore and Campbell Brown’s about the consequentialist representation of prohibition dilemmas. In the middle section I try to add some intuitive flesh to the formal skeleton of centered value. In the third section I then use the idea to develop Larry Temkin’s theoretical device of essentially comparative value. 2.1. Consequentializing Prohibition Dilemmas

Douglas Portmore likes consequentialism. That’s largely (but not entirely) because, like me, he takes a very broad view of what counts as consequentialist. Portmore thinks a consequentialist theory consists of an ordering of outcomes and a collection of deontic verdicts that are functions of the order of alternatives in the situation. Permitted acts are those that are not outranked by any alternative.2 The main burden of his Commonsense Consequentialism is to show how much of common sense can be accommodated by some form of consequentialism.3 Some people do not count as consequentialist any theory that admits agent-​ centered value.4 But plenty of agent-​centered views tell us to attend only to consequences, or the expectation of some quantity present in consequences. For instance, egoistic hedonism tells us each to maximize the expectation of our pleasure in the consequences of our choices. Nothing else matters to what we ought to do, according to egoistic hedonism, but this feature of the consequences. Egoistic hedonism is a paradigmatically consequentialist theory, and a paradigmatically agent-​ centered one. One tactic for making consequentialism safe for common sense, then, is to center value on agents. What does it mean to center value on agents? There are two ways to think about centered value. One is to take there to be many better than orderings, one for each agent. Outcomes are ranked from perspectives, so that the outcomes that are better from your perspective are not the same as those that are better from mine. This is Portmore’s model. The other way is to build the centering into the things that are

2 I know this characterization is telegraphic. For elaboration, see Portmore’s Commonsense Consequentialism; Brown, “Consequentialize This” (2011); or my “In Defense of Consequentializing.” 3 A whole lot, it turns out. 4 Scheffler, famously, in the introduction to Consequentialism and Its Critics. Also Kagan, Limits of Morality. See also Schroeder, “Teleology, Agent-​Relative Value, and ʻGood.’ ”



World-Centered Value    33

ordered rather than multiplying orderings.5 Instead of states of affairs as outcomes, we can think of outcomes as properties that a person might have or lack, or situations that a person can find herself in. Even if you and I occupy the same state of affairs, our situations are different; though my 1994 self is the same person as my current self, my situation in 1994 was very different from my current one. Egoistic hedonism is the view that the better situations are the more pleasant ones. More pleasant for the agent? That would be redundant. A pleasant situation to be in is one that is more pleasant for the one in it. It may be very unpleasant to be in a situation in which everyone but you is extremely happy. In competitive circumstances, you and I might aim at “the same thing,” meaning the situation you aim to be in is the one I aim to be in, but we couldn’t both achieve it anymore than Argentina and Germany could both achieve what they’re both aiming at when they’re playing each other in the World Cup Finals. To accommodate certain other aspects of common sense, we have to center also on times. For instance, it is plausible that at any time it is better for bad experiences to be over than still to come.6 If that’s true, there must be different rankings of states of the world at different times; alternatively, the experiences thought of as properties of times could be ranked.7 For example, Tuesday afternoon has the property that my pain is still ahead of me, whereas Wednesday afternoon has the property that my pain is over and done with. When I am glad the pain is over, my relief is a fitting reaction to the good property I have: being at a time with the over-​and-​done-​with property. Since many time-​centered values are also agent-​centered, we could rank properties of persons-​at-​times once, or have a ranking for each  pair. Having centered on times and persons, we might look for other centers for value. Are there plausible candidates? We might center on places. One reason that’s natural is that persons and times are elements of indices in semantics—​we need them to give a content to indexicals, for example—​and places are also elements of indices. As far as I can tell, centering on places is useless in axiology. True, common sense tells us

5 In his contribution to this volume, Jan Gertken suggests a way of centering value on agents that also revises which sorts of things are outcomes, but his way is quite different from the one I describe in the text. 6 Plausible: See Prior, “Thank Goodness That’s Over.” Problematic: Dougherty, “On Whether to Prefer Pain to Pass.” 7 I believe this is the correct answer to the question posed by Tim Henning in his contribution to this volume about how a consequentialist should think about his cases involving predictable preference accommodation. Henning wants to know how a consequentialist can account for the way a person’s preferences at a time can affect what it is rational for her to do even though they do not affect the consequences of the act. The answer is that a person’s preferences at a time may affect the ranking of outcomes that is relevant to the assessment of her actions. The value is time-​centered, and the time relevant to the deontic status of an act (including, perhaps, its rationality) is the time of the act itself.

34    Jamie Dreier

that the distant needy have less claim on our help than the nearby, but the strength of their claim is probably not really a matter of physical distance.8 The other important index element is a world. And there is another abstract reason to suspect that worlds are candidates for useful centers: theirs is the third important dimension through which value can be spread and aggregated. How valuable a policy’s outcome is seems to be a function of how valuable it is for each person; how good a university’s growth plan is may be a matter of how good it is each year; the value of an investment is its value in the different ways the world might turn out.9 Parameters for worlds tend to be useful in axiology, where parameters for persons and times are useful. But that’s just a hunch, so far. Is there any use to centering value on worlds—​and does it even make any sense? I will argue that it does make sense, and that there is enough theoretical use to be made of it to give it some space in our toolbox. Portmore suggests a use. In Commonsense Consequentialism he asks whether a consequentialist theory can accommodate the notion of a “prohibition dilemma” in which each of the alternatives available is prohibited. It looks doubtful. An act is prohibited when it has some alternative whose outcome outranks it, for Portmore. But surely there are no sets of alternatives all of whose outcomes are outranked by some other. So there must be an outcome highest ranked out of all outcomes of available acts, and bringing it about isn’t prohibited. But Portmore thinks we can find a way of modeling prohibition dilemmas after all. Following a strategy of Campbell Brown’s,10 Portmore adopts a model according to which the ranking of outcomes depends on which act is performed. Here’s what he says about what a consequentialist theory can say when the only available acts, A1 and A2, are prohibited because their associated outcomes, O1 and O2, are bound to be outranked: “To do so, the consequentialist must hold that whether O1 outranks O2 or vice versa depends on whether I perform A1 or A2.” He elaborates, “[T]‌he consequentialist must hold that O1 outranks O2 if and only if I perform A2, and that O2 outranks O1 if and only if I perform A1.”11 The suggestion is Brown’s, the statement is Portmore’s, but the idea that this move centers value on worlds is mine. I have to clarify. Let’s start by asking exactly what Portmore means by his biconditionals. We commonly say things of the form “Which of these outcomes is better depends on this other thing,” perhaps on which choice we make, and we don’t mean anything 8 I think the explanation for why place-​centering is not important is that place is not important in the way that time and persons are. Of course, it is instrumentally, derivatively important. Everything is. 9 I owe this way of thinking of things to Broome, Weighing Goods. 10 In an early version of “Consequentialize This,” but the passage didn’t survive into the published version of the paper. 11 Portmore, Commonsense Consequentialism,  90–​91.



World-Centered Value    35

as exotic as world-​centered value. For instance, which kind of weather is better depends on whether we head to the beach, go for a sail, or play soccer. We sometimes put it like this: a windy day is better if we’re going for a sail, but not if we’re going to play soccer. Weather is not ordinarily called an “outcome,” but that doesn’t really matter—​states of affairs, sets of worlds, are the sorts of things that get ordered. This is not what Portmore means. The dependence in the ordinary examples can be cashed out in a nice way, familiar to decision theorists. We say that for A to be better than B conditional on C is just for A&C to be better than B&C. For instance, sailing and a windy day is better than sailing and a calm day.12 That sort of dependence, plainly, is not what Portmore has in mind. He means, if I perform A2, then O1 is better than O2, but if I perform A1, then O2 is better than O1. But he doesn’t mean that A1& O2 is better than A1& O1. To see this, consider an example of a prohibition dilemma. Dance Dilemma Suppose I promised Maurice that I’d mambo and I promised Chuck that I’d cha-​cha; but I can’t do both. Whichever dance I do, I will break my promise. It is impermissible to break either promise, so whichever I do I will do something impermissible. So A1 is that I mambo, and O1 is that . . . I mambo; things I do are trivially outcomes of their own. And A2 and O2 are that I cha-​cha. Portmore does not mean to say that I mambo and I cha-​cha is better than I mambo and I mambo whereas I cha-​cha and I mambo is better than I cha-​cha and I cha-​cha.13 Suppose someone tried to paraphrase Portmore: Oh, so you mean cha-​chaing and mamboing is better than both cha-​chaing and cha-​chaing and also mamboing and mamboing? No, he doesn’t think the impossible dance move is better than both possible ones.14 Still, what Portmore is saying does seem to be the sort of thing we can say using conditionals. If I cha-​cha, then doing the mambo will have been better than doing

Note the parallel with conditional probability. If I say a Democratic victory is more likely than a Republican one given good weather, I’m just saying that the probability of a Democratic victory and good weather is higher than the probability of a Republican victory and good weather. 13 For one thing, the mixed conjunctions are impossible. Mamboing excludes cha-​chaing. 14 Actually, this example may be bad since in a way it does seem like doing the two things is better—​that would fulfill my promises—​but unfortunately impossible. But this is an accidental feature of the example. If I break my promise to Maurice, then keeping that promise is better than keeping the promise to Chuck; and if I break the promise to Chuck, then keeping that promise is better than keeping the promise to Maurice. That can’t be to say that keeping and breaking the promise to Maurice is better than keeping it and keeping it. 12

36    Jamie Dreier

the cha-​cha (would have been); and if I do the mambo, then doing the cha-​cha will have been better than mamboing (would have been). This is perfectly comprehensible; the syntax is difficult but kosher. It sounds like it’s perversely stating something that just couldn’t be true  .  .  .  but that, Portmore can reasonably say, is because it exposes the strangeness of Prohibition Dilemmas. One odd thing, though, about the above future perfect way of putting it is that it sounds like we mean something else. Suppose I have uncanny bad judgment about which is the best dance to do to the music. You might also have no idea which is best for this song, but you might say, “If Jamie does the cha-​cha, then the mambo will have been better, and if he does the mambo then the cha-​cha will have been better.” And I might agree with you, resigned to my own bad taste. But that is not what Portmore and Brown have in mind either. They aren’t talking about one thing being better than another given the evidence of the condition. I would like to be able to say what they mean by giving you a model that is both clear, in that it uses formal objects in the good way (which to me means it shows that I’m not cheating, not using some word ambiguously . . . rather than the obscurantist way), and intuitively graspable, which to my mind means that we know how to take the model and do things with it that we want to do with normative theories. 2.2. Getting a Grip on World-​Centering

When I  tell you that an agent-​centered view orders properties rather than states of affairs, and that I will model these with sets of centered worlds rather than sets of worlds, I know this is not very intuitive to some. I can make it more intuitive like this. The (proper, correct, fitting . . .) response to the discovery that a state of affairs is good is: to try to bring it about. The response to the discovery that a prop­ erty is really good is: to try to have it. Thus, wealth, being wealthy, is a property, and a good one, which means it’s sensible of you to try to bear that property. But what about the property of being pleasant?15 Surely that is a good property, but it is a property of experiences. A centered consequentialism takes the good thing to be:  having pleasant experiences. It ranks properties that the agent can have, not ways the world could be. And consider the property: being first in line for the new iGadget. That is a good property to have, but obviously it can’t be good that the property be instantiated by somebody or other. It would be nice to have a similar explanation of what value centered on worlds amounts to. Unfortunately, there is no

Thanks to Doug Portmore and Christian Seidel for asking me about this example. 15



World-Centered Value    37

analogous way of understanding world-​centered value. Maybe we can make sense of centering on worlds in some other way. Suppose we are behind a veil of ignorance, and two situations are described to us.16 In both situations each of two professors agreed to referee a paper; one of them completed the task on time, and the other is thirty days late. The difference between the two situations is that in the first, J is the tardy one, and in the second, K is the tardy one. Now, which situation do I  prefer—​which do I  hope is actual? I  don’t know, behind the veil of ignorance. I will know as soon as you tell me whether I am J or K. Tardiness in completing refereeing tasks is bad: it is an agent-​centered disvalue. (Obviously, it also has agent-​neutral disvalue; the point of the example is that each of the two situations is worse from one person’s point of view than the other.) Now suppose two other situations are set out for us.17 This time we know who we are, but not when it is. In the first, we undergo a very painful experience on Tuesday. In the second, we undergo a somewhat less painful experience on (the following) Friday. Which do we prefer? We don’t know. It depends on which day it is. If today is Wednesday we prefer the first situation, but if it’s Monday we prefer the second. The disvalue of dated experiences is time-​centered. Can we find a pair of situations, A and B, such that we cannot answer the question of which we morally prefer until we know which world is actual? If we knew exactly what the world is like, we would know which of A or B obtained, so if there are world-​centered values, this is a hint that they are not going to show up in our deliberations about what to do in the way that agent-​or time-​centered values do. Still, it will help if we can find such a pair. I think we can. For a first example, we can think about a relatively low-​key variation on the Non-​Existence Problem. Population Expansion Policy A policy we could adopt would bring some new people into existence, with very good lives (better than ours, perhaps); that consequence is situation A. But the resources required for these new people will come from us, so our lives will be slightly worse, on the whole. Plausibly, we need not adopt the policy, in which case the outcome is situation B. We may think (I am inclined to), it’s true that all those possible people would have led very good lives, but they do not exist, so nobody is worse off and some of us are better off for their absence, so the outcome we chose is better. I think Frank Jackson suggested this way of characterizing agent-​centered value in discussion. 17 After Derek Parfit’s “Painful Surgery” example, obviously. See Parfit’s Reasons and Persons, 165. 16

38    Jamie Dreier

We might worry that we would have to have judged differently had we made the opposite choice. For then we would have all those thriving people in our world, and it would be hard (and, I am going to assume, hard because incorrect) for us to judge that the world would have been a better place without all of those thrivers just because we ourselves would have been just slightly better off. Should this counterfactual (had we made the other choice, we would have correctly judged that the world was better for it) make us rethink our earlier conclusion, that since we did not make the person-​generating choice, we have nothing to regret?18 I say: it should not. We can maintain both at once: that judged from the perspective of the counterfactual situation, the more populous world with more happy people is better than the one we chose, while judged from the perspective of the actual world, the less populous world in which we are a little bit better off is better. That we can do because the value we are judging is world-​centered.19 Here are two interesting features of the example that are going to arise often if we center value on worlds. First, suppose we don’t really want to create the new people, since we are a bit selfish (or maybe not exactly selfish: we care about each other and know we’d have to sacrifice, collectively, to create new people). So, we think we are not going to create them. So, we think we are in the world relative to which it is better that they not exist. So, we can enkratically follow our selfish preference. This is a bit suspicious, but perhaps not really a problem. We somehow save ourselves from wrongdoing by doing the thing that we would have correctly counted wrong had we not done it. Second: suppose you notice that a lot of your worldmates seem to think, because of what seems to you to be bad reasoning, that they are under some obligation to create the new people. So, you think it is pretty likely that you are going to do it. So, you think it’s pretty likely that it is the best thing to do, so your conscience makes you vote in favor of creating the new people. Something seems fishy about this train of reasoning.20 There is another way to reason about which choice is best when we haven’t decided which to make. Maybe we could take each ranking to be represented by its own utility function, say UA (for the ranking centered on the world with the extra new people) and UB. Then we could use our credence for A and our credence for B to weight the utilities delivered by those functions, and reason that the option with the higher expectation is the better one to choose—​“subjectively” better, as some Compare Harman, “ ‘I’ll Be Glad I Did It.’ ” 19 There are more compelling examples that are agent-​centered as well as world-​centered. They are better in being more compelling, but worse in that the second kind of centering might be a distraction. 20 For something like this reason, Erik Carlson says, “An action’s moral status does not depend on whether or not it is performed” (Consequentialism Reconsidered, 100–​101). Carlson only rules our world-​centering of particular sorts. Similar worries guide Hedden and Hare, “Self-​Reinforcing and Self-​Frustrating Decisions.” 18



World-Centered Value    39

say, or better in light of our information. So we would compute UA(A)×cr(A) and UB(B)×cr(B), and choose the option with the higher product.21 But I don’t think this is good reasoning. We do not have these utility functions at hand. What we have is a better ordering centered on each world. We can construct utility functions for each, if the structure of the orderings fits the usual axioms of decision theory, but then it doesn’t make straightforward sense to compare the different utilities with one another. We’re in the same situation that we find ourselves in when, doing decision theory, we construct a function that represents your preferences expectationally, and a function that represents mine expectationally, and then ask whose utility is greater for some particular situation. There is no meaningful answer to this question. The problem is that when one utility number is greater than another, this is supposed to mean that it is preferred, in decision theory; but there is nobody whose preferences this comparison can represent, if one of the utility numbers is yours and the other is mine. Similarly, when the pair of utility numbers are the values of the two different functions, UA and UB, a comparison of them should say which outcome is better; but there is no world whose ranking is the relevant one for the comparison. Comparing world-​centered utilities, as far as I can see, is analogous to interpersonal comparison of decision theory’s utilities. The second interesting feature is that whichever course we adopt, we will properly be glad and correctly judge that we did the best. Can we add: whichever we do we will have done the best? Imagine we are in the situation of choice, trying to decide which to do. We know that if we adopt the person-​generating choice and bring about A, then our world is the A-​world and our ranking is the A-​world ranking, so indeed A is better than B and we made the right choice. We also know that if we adopt the selfish choice and bring about B, then our world is the B-​world and our ranking is the B-​world ranking, so B is better than A and we made the right choice. So, we will make the right choice.22 But there is a mistake lurking nearby. We did not, in fact, choose the person-​generating option, so in fact B is best. It is not true that had we chosen the complacent option the outcome would have been better. It’s true that that outcome would have been better according to the ranking we would have

21 I think this is the application to a simple case of a formalization like Hare and Hedden’s formalization of SICK, in which the ‘U’ is a variable ranging over utility functions. 22 Argument by cases can go wrong when conclusions are about what things are better than which, so I am not as confident about this conclusion as I might be. See my “Practical Conditionals,” esp. the King Henry and Lowlands Livestock examples; I  though the examples cast doubt on modus ponens, but proof by cases is equally implicated, and the most plausible diagnosis, I now think, is that the failure of the inference is due to the interaction between the two rules.

40    Jamie Dreier

had . . . but we do not have that ranking.23 So, we would not have performed the right act. Thus, these two come apart, surprisingly: (Before) Whichever act we perform, we will perform the right act. [True] (After) Had we performed the other act, we would have performed the right act. [False] Still, in the example the choice we are faced with looks to be the opposite of a dilemma. So let’s turn to a dilemmatic example. Here is Campbell Brown’s: A couple would like to have a child; if they were to have a child, this would increase their preference satisfaction. However, the child would prefer that she didn’t exist. Moreover, the child’s preference would be sufficiently strong or intense as to outweigh the preferences of the parents in the method of preference-​ aggregation that we endorse. Suppose first that they do have the child. In this case the child is an actual person, and so her preferences count in determining what the parents ought to do. And, since the child’s preference outweighs the parents’ preferences, our theory implies that the parents acted wrongly; they ought not to have had the child. Suppose next that the parents don’t have the child. In this case, the child is a merely possible person, and so her “preferences” don’t count. Thus, again, our theory implies that the parents acted wrongly.24 The example is structurally just like the person-​generation example, except that the welfare of the extra person counts against the situation in which she exists (though only in the world in which she exists—​it does not count at all in the world in which she doesn’t exist). So the rankings relative to each world, and thus the deontic verdicts, have flipped. Neither act will bring about the outcome that is (being careful to state it properly!) best according to that outcome’s world’s ranking. Portmore suggests that this means both acts are prohibited, and we have a prohibition dilemma. One could reject this interpretation. Either the couple is going to have the child, or they are not. Suppose they do. Then it is not the case that both acts are prohibited, since refraining from having the child is permissible, since (as a matter of fact, according to our world’s ranking) its outcome is best. Suppose they don’t. Then it is not the case that both acts are prohibited (since having the child is

23 If you counted a tail as a leg, how many legs would a horse have? It would have five legs according to the way you counted legs . . . but you don’t count legs that way. 24 Brown, “Consequentialize This” (2004), 33–​34.



World-Centered Value    41

permissible, because its outcome is best). But one of the suppositions must be true, so it is not the case that both acts are prohibited. So again, oddly, these two verdicts come apart: (V1)  Whatever the parents do, they will have acted wrongly. (V2)  Each of the parents’ options is prohibited. V1 is the conclusion of Campbell Brown’s argument, and I don’t see what could be wrong with that argument. But I just argued that V2 is false. 2.2.1. What  to Do

Answering questions about which option is better is supposed to answer questions about what to do. It’s supposed to help us figure out what to do, and indeed in standard cases it is supposed to settle the question of what to do. Suppose the better than ranking is agent-​centered: then each agent gets substantial advice, since each may consult its own ranking. Suppose it is time-​centered:  then on each occasion we get substantial advice, since on any occasion we may consult the ranking for the time of that occasion. We might hope, then, that we can get substantial advice from world-​centered ranking by consulting the particular ranking for our world—​the actual world, that is to say. But there is a crucial difference between the more familiar centerings on persons and times, on the one hand, and centering on worlds. It is that the choices we make determine which world we’re in; more colloquially and less modal-​metaphysicsy, our choices determine the way the world is, what the world is like. They never determine what time it is, and they never determine which agents we are. At least, not in the relevant sense. Our choices do determine what sort of person we are, but the answer to the question “Which person in the situation I know about am I?” is never “That depends on what you choose.” When the question of which world is actual is answered partly by the choice whose evaluation depends on which world is actual, world-​centered value is impractical. We have seen two kinds of case so far: in one, each of the options leads to a world whose own ranking puts it above the alternative, and in the other, the dilemma case, each option leads to a world whose own ranking puts it below the alternative. Neither of these seems to ground any kind of useful advice, and in neither case do we get substantial help in deciding what to do. Of course, that is the way prohibition dilemmas work. When I face one and I ask, “What must I do?,” the answer is the disappointing and frustrating “You must do the one you are not going to do.”

42    Jamie Dreier

That’s no use. But it’s not the fault of the model, or a theory, or an analysis. That’s just how it is with prohibition dilemmas. And conversely in the example of Population Expansion Policy, the answer to my practical question is “Whichever you will do, you must do that one,” which is a cheerier thought but of no practical help whatsoever. So in these cases, the facts about which option is better cannot play the role in deliberation that we expect them to play. There might be other cases in which world-​centered value is useful in deliberation, or there might be other, indirect ways in which it can be practical. As to the first: it might turn out that my choice between A and B will not itself affect the relative ranking of the outcomes. It might turn out that in the ranking centered on the world as it will be if I choose A, the A-​world outranks the B-​world, and in the ranking centered on the world as it will be if I choose B, the A-​world still outranks the B-​world. Then, it seems, I  must choose A.  But that’s to say that when the rankings agree, I get practical advice. And that means I get practical advice when the centeredness of value on worlds is not relevant. Here is a way that world-​centered value has some impact on us, although it is not directly practical: it can tell us how to feel and what to prefer. For example, after we act (wrongly) in a prohibition dilemma, it can tell us that we chose the worse option, and (plausibly) that means we ought to feel bad about what we did. In Population Expansion Policy, world-​centered value judgments can make us feel pleased with what we did. We’ll know to prefer our actual choice to the alternative. But unlike most preferences, this preference of ours is blocked from ever having any motivational efficacy, even in principle.25 Still, this shows that world-​centered value is not without any impact on our lives. One last note before I turn from Portmore’s method for consequentialist modeling of dilemmas. Even though the model I  suggested as an interpretation of Portmore’s remarks delivers the correct verdict (that is, the verdict that makes the choice into a prohibition dilemma), it has the wrong feel. For when I have promised Charles that I will cha-​cha and Maurice that I will mambo, it does feel like whichever I do I will have done something wrong, but it does not feel like whichever I do the other will turn out to have been better. Having broken my promise to Maurice, I shall rebuke myself, but not with the remonstrance “It would have been better to have done the cha-​cha.” I think this particular sort of prohibition dilemma—​to my mind the most plausible sort—​will need a different treatment by consequentializers. The Portmore & Brown model can help with other, more exotic dilemmas, but it doesn’t seem to fit this most ordinary kind. I think we are stuck with these Impractical Preferences anyway. See my “Rational Preference” for details. 25



World-Centered Value    43

A

A+

B–

B

Figure 2.1  Parfit’s Population Paradox

2.3. Apparent Failures of Transitivity of “Better Than”

I think centering value on worlds is more helpful in clarifying what’s going on in some well-​known examples in which the relation “better than” appears not to be transitive. Larry Temkin has some infamous examples that he claims display this transitivity failure. Of particular interest to me is his idea that the proper response to Derek Parfit’s Population Paradox sequence is that the sequence gets better at every step but ends up worse.26 To remind you briefly (see Figure 2.1): in A, the people are very well off and their welfare is equal. A+ is like A but with extra people, spatially separated from the original people, with lives worth living; B–​equalizes A+ without gaining or losing total welfare; B is the same as B–​but without the spatial separation. The common intuitions are that A+ is at least as good as A, B–​is at least as good as A+, and B is at least as good as B–​. But most people do not think B is as good as A. And to press home the intuition, we could string together a chain of similar comparisons to reach Z, a world with a gigantic population of persons whose lives are barely worth living—​and the proposition that Z is at least as good as A is the Repugnant Conclusion. Temkin’s own view is that at each step (the small steps between A and B) the new situation is indeed at least as good, but the last situation is not as good as A. At least as good isn’t transitive.27 Many of us find the idea that better isn’t transitive puzzling, or absurd. “Better” is a comparative. Comparatives are always transitive.28 Temkin’s book Rethinking 26 Parfit, Reasons and Persons, §131. 27 I won’t go into the possibility that better than is transitive but at least as good as isn’t; this isn’t a promising option. 28 I don’t think an argument is really needed for this, but here is one anyway. For a gradable adjective G, it is part of the meaning of the comparative construction or inflection, and arguably all of the meaning, that for x to be G-​er than y is for x to count as G in every context in which y does but not conversely (that is, there are contexts in which y counts as G and x does not). But this relation is transitive.

44    Jamie Dreier

the Good tries to explain the intuition of transitivity failure. It is not clear to me whether the idea is to explain the appearance of transitivity failure or to explain why transitivity really does fail; to me the interesting project is the former, since I am sure the latter is impossible, but Temkin may feel differently.29 He is not trying to find a psychological explanation for our intuitions, but a feature of normative reality that would explain why the normative landscape is properly described by apparent violations of the transitivity of better. One centerpiece of his explanation involves the distinction between what he calls an Internal Aspect view of good and an Essentially Comparative view. Very roughly, the idea is that there are certain values in or of states of affairs that show up only in comparison to some, but not other, states of affairs. Or perhaps better:  there are features that are always present, but their value does not show up except when the state is compared to certain other ones. Here is how he describes the two views: The Internal Aspects View of Outcome Goodness Roughly, for each outcome, O, how good that outcome is all things considered depends solely on how good it is with respect to each moral ideal that is relevant for assessing the goodness of outcomes, and on how much all of the relevant ideals matter vis-​à-​vis each other, where these depend solely on O’s internal features. Moreover, for any two outcomes, O1 and O2, O1 will be better than O2, all things considered if and only if the extent to which O1 is good all things considered, as determined solely on the basis of O1’s internal features, is greater than the extent to which O2 is good all things considered, as determined solely on the basis of O2’s internal features. In addition, if O1 is better than O2, all things considered, the extent to which this is so will depend solely on the extent to which O1 is good, all things considered, is greater than the extent to which O2 is good, all things considered.30 The Essentially Comparative View of Outcome Goodness Roughly, there is at least one outcome, O, such that there is no answer to the question of how good O is all things considered based solely on O’s internal features; or, even if, for each outcome, O, there is an answer to the question of how good O is all things considered based solely on O’s internal features, there are at least two outcomes, O1 and O2, such that how O1 compares with O2 all things considered is not simply a function of the extent to which O1 is good,

29 In conversation he reports that he does, in fact, feel differently. But I don’t have space to say why I don’t think his reasons are good ones. 30 Temkin, Rethinking the Good, 370.



World-Centered Value    45

all things considered, based solely on O1’s internal features and the extent to which O2 is good, all things considered, based solely on O2’s internal features.31 (I don’t think Temkin ever reformulates to remove the “roughly”; I take these to be the official formulations.) For me, Temkin’s statements do not manage to bring the central notion into sharp focus. There are two main ideas. One is that the goodness of an outcome is determined fully by its internal features. The other is that whether one outcome is better than another is determined by how good each of them is. How exactly these two features are related is not fully clear to me. A first thought might be that Internal Aspects articulates the view that better is an internal relation. An internal relation is one that supervenes on the intrinsic natures of its relata.32 For example, being taller than, or being two centimeters taller than, is intrinsic, since when x is two centimeters taller than y, anything with the same intrinsic properties as x (in particular, height) will be two centimeters taller than anything with the same intrinsic properties as y. On the other hand, being three centimeters away from is an extrinsic relation, since (or assuming) positions of objects in space are not intrinsic properties, so even if two pennies are three centimeters apart there may be a pair of perfect duplicate pennies that are not two centimeters apart. The Internal Aspects view could be that when one outcome is better than another this is determined by the goodness of each, in turn determined by the internal features of each separately. Maybe the Internal Aspect view is the view that better is an intrinsic relation. If that’s what the Internal Aspect view says, then the Essentially Comparative view is rather difficult to understand; but perhaps this is a virtue of the interpretation, since after all the Essentially Comparative view is difficult to understand! (Not just because of the language used to state it, I mean, but because of the strangeness of the idea itself.) To get a good sense of what it amounts to, we have to think a bit about intrinsic properties of outcomes. What is an outcome, and what sorts of properties of outcomes are intrinsic to them? An outcome is a state of affairs, or if you prefer, an event. (I’ll stick with states of affairs.) A state of affairs is picked out by a that clause. States of affairs may be more specific or less: there is a state of affairs consisting entirely of its raining tomorrow in Providence. Possible worlds are also states of affairs—​utterly specific ones. An example of Temkin’s at first seems to confirm that he means the Essentially Comparative view says that better is not an intrinsic relation. At one point he cites W. D. Ross’s view about proportional justice to illustrate the abstract idea that how Temkin, Rethinking the Good, 371. 32 Lewis, On the Plurality of Worlds, 62. 31

46    Jamie Dreier

good a state of affairs is may depend on more than the bare facts of distribution of goods to persons:  there may be two states of affairs with the same number of persons and the same distribution of welfare overall, but in the first state of affairs the happier people are also more deserving because of their past activities than the unhappier people, and conversely in the other situation. So here is an extrinsic prop­ erty of a state of affairs that is (on plausible views, anyway) a determiner of which other states it is better than: its history with respect to the activities of its denizens. Perhaps an Internal Aspects theorist would be forced to deny this. But no: Temkin is using this example to show that Internal Aspects theorists may include the histories of states of affairs as intrinsic properties of them, or internal properties. (Perhaps states of affairs are always temporally extended ad infinitum, so that they always include all of their own histories.) He explains, “On an Internal Aspects view, it can be relevant to the goodness of an outcome how people fare in absolute terms, or relative to what they deserve, or relative to others in that outcome.”33 There is a second passage that again suggests the Internal/​External Relations distinction. Temkin quotes Parfit: “Whether inequality makes . . . the outcome worse depends on how it comes about. . . . When inequality is produced by Mere Addition, it does not make the outcome worse. . . . Since the inequality in A+ is produced by Mere Addition, this inequality does not make A+ worse than A.”34 As Temkin puts it, Parfit’s point is that “A+’s inequality is not regrettable if the alternative is A.”35 It appears that Parfit thinks a feature of an outcome that matters to its valuation is how it comes about (or perhaps different, how one of its features, here inequality, comes about), what produced it, its causal ancestry. Temkin’s gloss is in line with this interpretation: what the alternative was when somebody chose this outcome is surely part of its causal history, and that seems to be what Temkin is pointing to. Here is a very clear example of Amartya Sen’s making a very similar point:36 Apples to Oranges Suppose you are at someone’s house at a moderately formal departmental function and she offers you your choice from a bowl of fruit. In Scenario One, the only fruits in the bowl are a large apple and an orange. Feeling hungry, you take the large apple. In Scenario Two, the only fruits in the bowl are a small apple and an orange. Feeling hungry, you take the orange. In Scenario Three, the only fruits in the bowl are a small apple and a large apple. Feeling the eyes Temkin, Rethinking the Good, 375. 34 Quoting from Parfit, Reasons and Persons, 425. 35 Temkin, Rethinking the Good, 368. 36 See Sen, “Internal Consistency of Choice,” 501–​502. 33



World-Centered Value    47

of the hostess and the other guests upon you, and sensing that the larger apple is obviously superior to the small one, you know that it would be rude to pick the larger apple, so you take the smaller one. Your preferences fail transitivity! But no, of course, they do not. The “outcomes” are not merely: You have a small apple; you have a large apple; you have an orange. They are more like: You have a small apple; you have an orange; you have a large apple obtained by rudeness; you have a large apple obtained without rudeness. The causal history of the “same outcome” makes a difference, in this case to whether the state of affairs contains rudeness. And, in particular, the piece of the causal history that contains the situation of your choice is quite relevant, since what makes the choice of the larger apple rude is that the alternative was the smaller apple. In favor of this interpretation is that it can explain and vindicate apparent failures of transitivity of preference, and it could just as well explain and vindicate apparent failures of transitivity of better. The explanation is, to my mind, very satisfying, since it shows that there is a genuine ordering of outcomes, only the outcomes were not well characterized in the way we might have thought (not by features intrinsic to them, on one way of drawing boundaries around states of affairs). If one of Temkin’s favorite examples of transitivity failure could be explained along these lines, a good deal of progress would be made toward a fuller understanding of what is going on in those examples. Temkin does not mean this very clear idea. He means something else. He sometimes describes the Essentially Comparative view in a way that makes it sound like the view that temporally extrinsic, causal history features of a state of affairs are relevant to its goodness. But that is not the view.37 What he means is not the view that inequality can be bad or not bad depending on how it is brought about, but rather the view that whether the inequality lurking in a situation contributes disvalue to it depends on what other situation it is compared to. And this doesn’t mean that by drawing the comparison, we ourselves bring about a new disvalue or aggravate the evil of inequality—​that would be a bizarre metaphysical view, not least because typically the inequalities under consideration are fictional and thus causally impervious (one would expect) to what we real and actual people do and think. For another interpretation of the Essentially Comparative view, I  can return (finally!) to world-​centered value. To simplify a bit, we can leave out the issue of the value of equality and focus instead on the problem of the value of nonexistence. Many people (myself included) are inclined to judge that adding persons to a See Dancy, “Essentially Comparative Concepts,” for an attempt to straighten out a similar ambiguity in Temkin’s essay, “Rethinking the Good.” 37

48    Jamie Dreier

population typically makes it neither better nor worse.38 That’s why it does not seem to make things worse if we move from A to A+ in Parfit’s series. The new people who exist in A+ and not A have, by stipulation, “lives worth living,” so it is very hard to judge that they (would) make the world a worse place. On the other hand, in Population Expansion Policy it does not seem that the world would have been a better place if it contained the extra people we could have created but didn’t, and that judgment is insensitive to smallish changes in the story to how good the lives of those people would have been. Now, when a change, like adding persons to the population, does not make things better and does not make things worse, that could be because it leaves things exactly as good as they were. But this cannot be right, since exactly as good as is transitive. Suppose we enhance the lives of the new people in A+, so that instead of having lives just barely worth living they have quite comfortable and pleasant lives; call this A++. It cannot be that A is exactly as good as A+ and also exactly as good as A++, since A++ is manifestly better than A+. We can vindicate the intuition, perhaps in the spirit of Temkin’s picture, in two steps.39 First, we can center our better than (and our exactly as good as) on worlds. At each world w, let any pair w1 and w2 be ranked according to the aggregate of well-​being of w denizens. So w1 is better than w2 just in case its aggregate is higher, and they are equally good if their aggregates are equal. The ranking, clearly, will depend on w (as well as on w1 and w2), since w will determine which persons count in the aggregate. Let me not spell out the second step yet. Just notice that when we compare A to A+ and to A++, it seems natural to center ourselves on A. Then we can judge that A+ is no better or worse than A, since it is no better or worse for anyone who counts (namely, the A persons); and similarly when we compare A to A++. But when we compare A++ to A+, we do not naturally center ourselves on A40 but on either A+ or A++, and according to our recipe it won’t matter which since they have the same residents. Then A++ is ranked above A+, since all of their denizens count in their aggregates and some do better in A++ while none do worse. As I mentioned at the outset, I am not trying to defend the intuitive judgment many people make. I am trying to show how it could make sense. The centering idea

Proper accounting of this thought is one main theme of Broome, Weighing Lives. 39 The main idea here is due to Dasgupta, “Lives and Well-​Being.” Dasgupta makes different use of the idea, and in particular his method for choosing between worlds involves a kind of “filter” at the first stage. If I were actually trying to solve the Population Problems rather than to model views about them, I think I would take Dasgupta’s suggestion very seriously, since it (a) is entirely intelligible and (b) yields pretty intuitive verdicts. 40 I think it might be different if the story were a very realistic and practically pressing one in which instead of some imaginary A we had the actual world; then it might be hard to resist centering on that one even in our thought experiments. 38



World-Centered Value    49

seems to explain and make sense of some intuitions. And it also provides a model for Temkin’s Essentially Comparative notions of good. For in comparing X to Y, we naturally take up the perspective of X (or Y), and thus use X’s (or Y’s) ranking to make the comparison. When we compare A to A+, the well-​being of the extra persons in A+ has no impact on our judgment, but when we compare A++ to A+ it does. The world-​centered model shows how this can make sense. Furthermore, it shows how transitivity is saved despite its apparent failure in our comparisons, for when we think that A++ might be better than A+ even though we’ve already judged that A and A+ are equally good and so are A and A++, we are switching from one better than relation to another. Holding fixed a world of centering, better is transitive, but when our judgments slide from one center to another the relation we seem to be talking about is not transitive at all. To make this last point explicit, consider the interesting dyadic relation, relating a world W to a world X: W is betterW than X In English (more or less): W is better from the perspective of W than X. This relation is not transitive. Nor is: W is equally goodW as X Parfit’s A bears this relation to A+ and to A++, but A+ and A++ do not bear it either to the other. Nor is it even symmetric, nor is the better than version asymmetric. Despite the way we might try to pronounce them or otherwise name them, the first relation is not a comparative (what would it be the comparative of?) and the second is not an equivalence relation. 2.4. Conclusion

As I explained at the end of section 2.2., we do not use world-​centered value in our choices—​or at least, even if value is centered on worlds the fact that it is does not show up in our decision-​making. In this way, world-​centered value is not as important as agent-​centered value or time-​centered value. But centering on worlds can play a theoretical role, modeling the confusing axiological ordering displayed in the consequentialized version of a theory that has room for prohibition dilemmas, and centered rankings can explain the feeling of regret (or absence thereof ) we naturally and properly feel when we think back on our actual choices, and the preferences we

50    Jamie Dreier

find in ourselves when we think about and compare scenarios in which different persons exist. References Broome, John. Weighing Goods: Equality, Uncertainty and Time. Oxford: Wiley-​Blackwell,  1995. Broome, John. Weighing Lives. New York: Clarendon Press, 2004. Brown, Campbell. “Consequentialize This.” Unpublished manuscript, September 10, 2004. Brown, Campbell. “Consequentialize This.” Ethics 121 (2011): 749–​771. Carlson, Erik. Consequentialism Reconsidered. Dordrecht: Springer, 1995. Dancy, Jonathan. “Essentially Comparative Concepts.” Journal of Ethics & Social Philosophy 1 (2005): 1–​15. Dasgupta, Partha. “Lives and Well-​Being.” Social Choice and Welfare 5 (1988): 103–​126. Dougherty, Tom. “On Whether to Prefer Pain to Pass.” Ethics 121 (2011): 521–​537. Dreier, James. “Structures of Normative Theories.” The Monist 76 (1993): 22–​40. Dreier, James. “Rational Preference.” Theory and Decision 40 (1996): 249–​276. Dreier, James. “Practical Conditionals.” In Reasons for Action, edited by David Sobel and Steven Wall, 116–​133. Cambridge, UK: Cambridge University Press, 2009. Dreier, Jamie. “In Defense of Consequentializing.” In Oxford Studies in Normative Ethics, vol. 1, edited by Mark Timmons, 97–​119. Oxford: Oxford University Press, 2011. Harman, Elizabeth. “ʻI’ll Be Glad I Did It’ Reasoning and the Significance of Future Desires.” Philosophical Perspectives 23 (2009): 177–​199. Hedden, Brian, and Caspar Hare. “Self-​Reinforcing and Self-​Frustrating Decisions.” Noûs 50 (2015): 604–​628. Kagan, Shelly. The Limits of Morality. Revised ed. Oxford: Clarendon Press, 1991. Lewis, David K. On the Plurality of Worlds. Malden, MA: Wiley-​Blackwell, 2001. Parfit, Derek. Reasons and Persons. Oxford: Clarendon Press, 1987. Portmore, Douglas W. Commonsense Consequentialism:  Wherein Morality Meets Rationality. New York: Oxford University Press, 2014. Prior, Arthur N. “Thank Goodness That’s Over.” Philosophy 34 (1959): 12–​17. Scheffler, Samuel. Consequentialism and Its Critics. New York: Oxford University Press, 1988. Schroeder, Mark. “Teleology, Agent-​Relative Value, and ʻGood.’” Ethics 117 (2007): 265–​295. Sen, Amartya. “Internal Consistency of Choice.” Econometrica 61 (1993): 495–​521. Temkin, Larry. “Rethinking the Good: Moral Ideals and the Nature of Practical Reasoning.” In Reading Parfit, edited by Jonathan Dancy, 290–​345. Oxford: Blackwell, 1977. Temkin, Larry. Rethinking the Good:  Moral Ideals and the Nature of Practical Reasoning. Oxford: Oxford University Press, 2012.

3 MULT I D I MENSIONA L CONSE QUE NTIA L I S M A N D P OPUL AT I O N  E THICS

Martin Peterson

Is it better to bring about a world in which a small number of people enjoy high levels of well-​being, or would it be better to bring about a world with a much larger population in which everyone has a life barely worth living? Much work in population ethics has focused on a series of negative results. Derek Parfit argues in Reasons and Persons that none of the theories he discusses is morally acceptable. Yew-​Kwang Ng1 and Gustaf Arrhenius2 generalize Parfit’s conclusion by showing that no theory of population ethics is consistent with a number of intuitively reasonable normative desiderata. They do this by proving a set of formal impossibility theorems. In this chapter, I argue that multidimensional consequentialists have reason to reject some of the key premises of Parfit’s Mere Addition Paradox, as well as Arrhenius’s sixth impossibility theorem. The latter is the most general and far-​reaching impossibility theorem in the literature on population ethics. My main point is that multidimensional consequentialists can reasonably maintain that the mere addition of people who have lives worth living is not always entirely right. To add what Parfit calls “extra people” is right with respect to one moral aspect (the number of lives worth living) but wrong with respect to another (the average quality of life).

1 Ng, “What Should We Do about Future Generations?” 2 Arrhenius, “Future Generations” and “Population Ethics.”

51

52    Martin Peterson

Multidimensional consequentialists believe that an act’s rightness or wrongness depends on several irreducible moral aspects.3 If no act is optimal with respect to all aspects, then no act is entirely right. On the multidimensional view, moral rightness and wrongness literally come in degrees. An act is right to the highest degree if and only if it is optimal with respect to all applicable aspects, meaning that if two or more aspects clash, even the best alternative is somewhat wrong. The claim that an act’s deontic status depends on several irreducible aspects can be clarified by an analogy with geometry.4 The area of a triangle depends on two variables (its height and the length of its base), but the area of a circle depends on only one variable (its radius). In a similar vein, multidimensional consequentialists believe that an act’s deontic status depends on at least two irreducible aspects or variables (for instance, equality and well-​being), whereas one-​ dimensional consequentialists believe that an act’s deontic status depends on a single aspect or variable. For a straightforward example of a one-​dimensional view, consider total hedonistic act-​utilitarianism. According to this theory, the only aspect that determines an act’s deontic status is whether it maximizes happiness. The fact that both the intensity and the duration of a pleasurable experience matter (as well as the intensity and duration of painful ones) does not make the theory multidimensional. These factors can be reduced to a single aspect:  the total amount of happiness brought about by the act. What matters for the classification of a theory as multidimensional is whether it is possible to characterize an act’s deontic status as a function of a single aspect. By definition, theorists who believe that we should assign weights to different aspects and then maximize the weighted sum, or aggregate them in some other way, defend one-​dimensional views. If it is possible to aggregate two or more aspects into a new, composite aspect, then those aspects could obviously be reduced to a single composite aspect. This chapter has five further sections. Section 3.1 recapitulates the Repugnant Conclusion and the Mere Addition Paradox. Section 3.2 explains the basic tenets of multidimensional consequentialism. Section 3.3 examines multidimensional consequentialism in greater detail and responds to a couple of objections. Thereafter, in sections 3.4 and 3.5, the multidimensional analysis of what I take to be the core issues in population ethics are discussed at length.

3 See Peterson, The Dimensions of Consequentialism, for a defense of multidimensional consequentialism. 4 By “deontic status” I mean an act’s (degree of ) rightness or wrongness. The analogy with geometry is discussed in ­chapter 1 in Peterson, Dimensions of Consequentialism. See also Peterson, “Reply to Ralf Bader.”



Multidimensional Population Ethics    53

A

B

C

...

Z

Figure 3.1  The Repugnant Conclusion

3.1. The Repugnant Conclusion and the Mere Addition Paradox

The Repugnant Conclusion is one of the key results in Parfit’s discussion of population ethics. It is derived by comparing a series of future worlds A, B, C . . . , Z. The width of each rectangle in Figure 3.1 is proportional to the number of people in that world. The height represents their level of well-​being. In all worlds, all people have lives worth living. Many traditional consequentialists believe that B is better than A, given that there are sufficiently many more B-​people than A-​people, and given that the quality of life for the B-​people is just a little bit lower than that for the A-​people.5 (Note that there is no inequality in either of these worlds.) For the same reason, C is better than B, and the same applies, mutatis mutandis, to every intermediary world between C and Z. However, as famously pointed out by Parfit, this line of reasoning leads to the following Repugnant Conclusion: “For any possible population of at least ten billion people, all with a very high quality of life, there must be some much larger imaginable population [Z]‌whose existence, if other things are equal, would be better, even though its members have lives that are barely worth living.”6 If we accept the premise that “better than” is transitive, then Z is better than A. However, if we compare Z and A, we may find this conclusion hard to accept. Intuitively, A is better than Z. But it cannot be true that Z is better than A and that A is better than Z. The Repugnant Conclusion is an important element of the Mere Addition Paradox. In order to state the Mere Addition Paradox, imagine a world A+ in which there is one group of people of the same size as the only group in A, with the same high quality of life. There is also another group of people in A+ that Parfit calls “extra people.” They all have lives worth living and affect no one else (see Figure 3.2). Now compare Divided B to A+. Because the average level is higher in Divided 5 Average utilitarianism is a well-​known exception. However, average utilitarianism entails that a world inhabited by a single very happy person is better than the world we are currently living in. 6 Parfit, Reasons and Persons, 388.

54    Martin Peterson

the average level in A+

A

B

A+

Divided B

Figure 3.2  The Mere Addition Paradox

B, this world seems to be better than A+. The extra people in Divided B benefit more than what is lost by the best-​off people in A+. Moreover, Divided B is exactly as good as B. However, if we put all these moral judgments together, we arrive at a problematic conclusion: A+ is not worse than A, Divided B is better than A+, and B is equally as good as Divided B. Therefore, B is not worse than A, and by repeating this argument sufficiently many times it follows that Z is not worse than A. But we also believe that Z is worse than A because we reject the Repugnant Conclusion. We have now contradicted ourselves. A possible response to the Mere Addition Paradox is to give up the premise that “better than” and “not worse than” are transitive relations.7 Parfit has recently expressed sympathy for this solution.8 However, Arrhenius shows that analogous contradictions arise if the premises are formulated as normative adequacy conditions about the rightness and wrongness of acts, rather than as evaluative claims about better or worse worlds.9 This indicates that Parfit’s solution is not as general as one might have hoped. 3.2. Multidimensional Consequentialism with Fixed Populations

Before the multidimensional theory is applied to cases with variable population size, it is helpful to consider how it deals with cases in which the identity of all individuals is fixed. Consider the example in Table 3.1. The numbers denote each individual’s well-​ being. Act 1 brings about the largest sum total of well-​being (100 units). Utilitarians believe that Act 1 is right, while the other two alternatives are wrong. Strict 7 See, e.g., Temkin, Rethinking the Good. 8 See Arrhenius, “Population Ethics,” for a discussion of this proposal. 9 See Arrhenius, “Future Generations” and “Population Ethics.”



Multidimensional Population Ethics    55

Table 3.1.  One-​versus multidimensional consequentialism Act Alice’s well-​being Bob’s well-​being Act 1 Act 2 Act 3

98 50 1

2 49 1

egalitarians, who care about nothing but equality, believe that Act 3 is the only right alternative. The prioritarian analysis depends on which priority function is applied. A reasonable and frequently mentioned priority function is the root function.10 By calculating the square root of all the numbers in Table 3.1, it can be easily verified that Act 2 is the only right alternative according to this and many other priority functions. Multidimensional consequentialists analyze the alternatives in Table 3.1 differently. For expository purposes it is helpful to imagine that equality and total well-​being are two irreducible moral aspects, as suggested in the introduction. I will not try to defend this assumption here. My goal is just to explain how the multidimensional analysis differs from one-​dimensional ones. In sections 3.4 and 3.5, I  will adopt a slightly more complex version of the multidimensional view, according to which each individual person counts as a separate moral aspect. None of the alternatives in Table 3.1 is optimal with respect to both total well-​ being and equality. Act 1 is optimal with respect to total well-​being but scores poorly with respect to equality. Act 3, on the other hand, is optimal with respect to equality but scores poorly with respect to total well-​being. Act 2 is not optimal with respect to either of the two aspects, although it scores well with respect to both of them. Multidimensional consequentialists, therefore, believe that because none of the available alternatives is optimal with respect to all applicable aspects, no alternative is entirely right. All three alternatives are somewhat right and somewhat wrong, although some of the somewhat right acts are right to a higher degree than the others.11 Because Act 2 is almost right with respect to each of the two aspects, Only positive numbers are permitted by the root function, so this version of prioritarianism is not compatible with all measures of well-​being. Economists have proposed alternative functions that permit for negative levels of well-​being. 11 Alastair Norcross defends a somewhat similar view he calls “scalar consequentialism,” according to which acts are better or worse than others, but never right or wrong. That is, goodness and badness vary in degrees, but scalar consequentialists reject the notions of rightness and wrongness altogether. For a comparison of scalar and multidimensional consequentialism, see section 2.5 of Peterson, Dimensions of Consequentialism, and see Norcross, “Good and Bad Actions,” for a general discussion of scalar consequentialism and some related ideas. 10

56    Martin Peterson

and thus also a little bit wrong with respect to each of them, it is all things considered much closer to being entirely right than to being entirely wrong. Act 1 is, all things considered, right to a much lower degree because it scores so poorly with respect to equality. The same holds mutatis mutandis for Act 3. The nonbinary notions of moral rightness and wrongness adopted by multidimensional consequentialists are technical concepts, which may or may not reflect the ordinary meaning of those terms. There is little reason to think that our existing vocabulary is always the most appropriate one. The scientific meaning of “heat” and “one meter” differ significantly from the original, ordinary meanings of those terms. Just like in the sciences, our willingness to accept the technical vocabulary proposed by multidimensional consequentialists should depend on whether this is useful for describing phenomena the moral theorist has reason to care about, rather than on what most people ordinarily mean when they use those terms.12 An act’s degree of rightness can be represented by a real number between 0 and 1. From a technical point of view, this is unproblematic. As long as the number of degrees is not larger than the unaccountable infinity of real numbers in the unit interval, this representation will not restrict the multidimensional consequentialist. However, the technical details of a numerical account can be spelled out in different ways.13 In what follows I will state what I believe to be one of the most attractive numerical versions of multidimensional consequentialism for fixed populations. According to this version of the theory, an act’s degree of rightness with respect to total well-​being is determined by calculating the ratio between the total amount of well-​being brought about by the act (on a normalized positive scale) and the maximum amount obtainable by any alternative act.14 Equality can be measured in different ways. One of the most widely used measures is the Gini index. Because the Gini index of a perfectly equal distribution is 0, and 1 represents maximal inequality, an act’s degree of rightness with respect to equality equals 1 minus its Gini index.15 By calculating the Gini index for each distribution of well-​being in Table 3.1 we obtain the numbers in the rightmost column of Table 3.2.

12 For a discussion of this point, see section 2.5 in Peterson, Dimensions of Consequentialism; Crisp, “Rightness, Parsimony, and Consequentialism.” 13 For a detailed discussion of three different versions of multidimensional consequentialism, see the appendix to Peterson, Dimensions of Consequentialism. The version of the theory summarized here is equivalent to one of the three versions, but note that I remain neutral about which of the three versions is best. 14 This measure clearly presupposes that well-​being can be measured on a ratio scale. If only interval measurement is possible some other measure has to be chosen. 1 n n 15 By letting w be the mean of w1 ,..., wn the Gini index g can be written as g ( w1 ,..., wn ) = 2 ∑ ∑ wi − w j 2n w i =1 j =1



Multidimensional Population Ethics    57

Table 3.2.  Degrees of moral rightness with respect to (w.r.t.) total well-​being and equality Act Right w.r.t. well-​being Right w.r.t. equality Act 1 Act 2 Act 3

1 .99 .02

.52 .99 1

Table 3.2 summarizes the three acts’ degree of rightness with respect to each of the two aspects:  total well-​being and equality. Each act’s degree of rightness, all things considered, is a function of these variables. But how should we aggregate rightness with-​ respect-​to-​a-​moral-​aspect to rightness all-​things-​considered? One possibility could be to maintain that the overall degree of moral rightness is determined by the moral aspect with which the act scores least well, such that the act’s overall degree of rightness equals the degree to which the act is right with respect to that aspect. This is the maximin account of all-​things-​considered rightness. It holds that Act 2 is right to degree .99, while Act 1 is right to degree .52, and Act 3 is right to degree .02. This is the view I myself find most plausible. That said, one could also assign weights to each aspect and then calculate the weighted average degree of rightness and wrongness of each alternative. This is the weighted average account. Its main weakness is that it is hard to understand where the weights would come from. A third possibility, which I believe is inferior to both the first and second, is to maintain that the overall degree of moral rightness is determined by the moral aspect with respect to which the act scores best. This is the maximax account. The chief objection to the maximax account is that it makes it “too easy” for an act to become entirely right. For instance, according to the maximax rule, both Act 1 and Act 3 in Table 3.2 would be entirely right, but not Act 2. This is not very plausible. 3.3. Moral Aspects and Degrees of Rightness

In the previous sections I  offered two examples of moral aspects:  equality and total well-​being. These examples were chosen for expository purposes. According to the version of multidimensional consequentialism I believe to be most plausible, each person’s individual well-​being level, not the total well-​being in the population, counts as a moral aspect.16 I  shall return to my preferred version 16 See Peterson, Dimensions of Consequentialism.

58    Martin Peterson

of the multidimensional theory in section 3.4. The aim of the present section is to spell out the distinction between multidimensional and one-​dimensional theories in more general terms, without referring to any particular examples of moral aspects. Let me first say a few words about the relation between aspects and dimensions. Generally speaking, a dimension is a conceptual space in which an aspect can vary. For instance, the height and width of a rectangle are two different aspects that belong to the same physical dimension (length), but the mass and velocity of a falling object are examples of different aspects belonging to different dimensions. In previous work, I have argued that the well-​being enjoyed by each person, rather than the total well-​ being enjoyed by the group, forms a set of different moral aspects belonging to the same moral dimension (persons).17 However, even if one thinks that it is the well-​being experienced by each person that counts, not the well-​being enjoyed by the group, it does not follow that one is a multidimensional consequentialist. This is because utilitarians, for instance, acknowledge that there is a sense in which one could say that an act’s deontic status depends on the well-​being enjoyed by each person: individual well-​being affects the total well-​being. So in that sense utilitarians could say that an act’s deontic status depends on more than one aspect. However, utilitarians also believe that we can reduce all those aspects (the well-​being enjoyed by each person) to a single composite aspect: the total well-​being enjoyed by all individuals together. It was in order to steer clear of this type of maneuver I pointed out in the introduction that multidimensional consequentialists believe that an act’s deontic status depends on several irreducible aspects. It is worth keeping in mind that multidimensional consequentialism, technically speaking, allows for views in which all the different moral aspects belong to the same dimension. Some multidimensional views may, in fact, turn out to be merely “multiaspectual.” However, because this term is a bit clumsy I prefer the term “multidimensional,” even when I discuss views holding that all aspects belong to the same dimension. Let me elaborate a bit on how I use the term “moral aspect.” Consider the following definition: Something counts as a moral aspect if and only if it directly influences an act’s deontic status, irrespective of how other aspects are altered. That something directly influences the deontic status of an act should be understood as a claim

See ­chapter 3 in Peterson, Dimensions of Consequentialism. 17



Multidimensional Population Ethics    59

about functional relationships:  an aspect, a, directly influences the deontic status, d, of an act if and only if d is a function of a.18 A potential problem with this definition is that the term “direct influence” can be interpreted in more than one way, even when we add the condition that it must be possible to capture this influence by some mathematical function. For instance, a possible but not very plausible interpretation of direct influence is to read it as a causal notion. Utilitarians do not believe that an act’s rightness is caused by the fact that it brings about the greatest amount of well-​being. Many utilitarians instead believe that the act’s property of being right supervenes on its property of bringing about the largest amount of well-​being. However, note that if we read “direct influence” as “supervenes on,” then we open a new set of metaphysical worries. This is because our understanding of supervenience is almost as unclear as our understating of causation. Many technical and slightly different notions of supervenience have been proposed in the literature, but it does not help much to say that direct influence means (some particular form of ) supervenience until we have been able to explain what exactly supervenience is.19 So how should we understand the term “direct influence”? I believe the best we can do, without giving up the claim about functions, is to appeal to the analogy with geometry mentioned in the introduction. The area of a triangle is a function of, and is thus directly influenced by, two aspects (its height and the length of its base), but the area of a circle is a function of, and thus directly influenced by, just one aspect (its radius). How do we know this? Well, if we increase the length of the base while holding the height of the triangle constant, its area increases. The same applies mutatis mutandis to circles. As far as I am aware, there is no philosophically uncontroversial way of explaining in what sense the variables of a function directly influence the output of a function that goes beyond this. In response to this, Ralf Bader has objected that the notion of direct influence captured by the notion of a function is implausible: “Understanding what it is for something to directly influence something else in terms of the latter being a function of the former is not particularly plausible. This is because functional relationships can be chained since functions can be composed, whereas (any intuitive understanding of ) direct influence does not allow for chaining. If x is a function of y, and

This definition is from Peterson, Dimensions of Consequentialism, 3. In a footnote on the same page I point out that the identity function has to be excluded, because an act’s deontic status is not an aspect that determines its deontic status. 19 For an overview, see McLaughlin and Bennett, “Supervenience.” 18

60    Martin Peterson

y is a function of z, then x will be a function of z. . . . This, however, does not mean that z directly influences x.”20 Bader is right that there is some way of understanding “direct influence” that would make the notion of a function misplaced, for the reason he mentions. For instance, if you directly influence my political opinions by having a long conversation with me (perhaps you point out that ninety-​one prisoners were executed in Texas between January 1, 2010, and November 15, 2016, which you think is ninety-​one people too many) and I then directly influence the political opinions of my neighbor by telling her what you told me, it seems to be a mistake to conclude that you have directly influenced the political opinions of my neighbor. It appears that your influence on my neighbor’s political opinion was indirect. However, it does not follow from this that there is no intuitive understanding of direct influence that allows for chaining. The best response to Bader’s objection is thus to point out that we should understand the notion of direct influence such that it allows for chaining. Consider, for instance, average utilitarianism. Is this a multi-​or one-​dimensional view? Well, average utilitarianism can be described in more than one way. We could say that average utilitarianism is the view that an act’s deontic status is a function of a single moral aspect: its average utility. However, we could also say that average utilitarianism is the view that an act’s deontic status is a composite function in which each individual’s well-​being is a separate moral aspect. The second view can then be characterized by a composite function in which all individual well-​being levels are first transformed to an average value by the first function, and the second function assigns a deontic predicate such as “right” or “wrong” to the act depending on how high this average value is. This suggests that there is a sense in which the average utility directly influences the act’s deontic status, and because it is possible to reduce all aspects to a single aspect, average utilitarianism is a one-​dimensional view, not a multidimensional one. So the fact that a function can be chained with other functions is not a threat to the distinction between one-​and multidimensional consequentialism. Doug Portmore has proposed another type of objection to multidimensional consequentialism. He asks us to consider a situation in which the multidimensional consequentialist is faced with a choice among three alternative distributions of well-​being: “My considered moral intuition is that A2 [Act 2 in Table 3.3] is entirely right. And this is what utilitarianism implies. MDC [multidimensional consequentialism], by contrast, implies that A2 is somewhat wrong, for [it] is suboptimal both in terms of equality and in terms of Alice’s wellbeing, each of which are moral

Bader, “Review of Martin Peterson’s The Dimensions of Consequentialism,” 620. 20



Multidimensional Population Ethics    61

Table 3.3.  Portmore’s objection Act Alice’s well-​being Bob’s well-​being Act 1 Act 2 Act 3

1,000,000 999,999 1

–​100 999,997 1

aspects on MDC. Indeed, MDC implies that all three alternatives are somewhat wrong. So utilitarianism seems to do better in this case.”21 Portmore correctly points out that utilitarians and multidimensional consequentialists analyze the situation in Table 3.3 differently. Utilitarians believe that Act 2 is entirely right in the binary sense. Multidimensional consequentialists, on the other hand, believe that Act 2 is right to a high degree, although not entirely right. This is because another alternative, Act 3, brings about more equality. The gist of Portmore’s objection is that it is absurd to maintain that Act 2 is somewhat wrong. On his view, Act 2 (but not Act 1 or Act 3)  is entirely right. Before I discuss what multidimensional consequentialists have to say in response to Portmore’s objection, it is worth calculating exactly how right and wrong the three alternatives in his example actually are according to this theory. A reason for doing this is that the calculation may affect our considered intuition about the case.22 Let us suppose, once again, that equality and well-​being are two separate moral aspects. Now, if an act’s all-​things-​considered rightness is calculated according to the maximin rule, it can be easily verified that Act 1 is, all things considered, right to degree .499950, while Act 2 is right to degree .999999, and Act 3 is right to degree .000001 (see Table 3.4). The numbers in Table 3.4 indicate that the disagreement between utilitarians and multidimensional consequentialists about Act 2 is, from a practical point of view, quite minuscule. Utilitarians believe that this act is entirely right in the binary sense (that is, right to degree 1 in the multidimensional vocabulary), while multidimensional consequentialists believe that Act 2 is “merely” right to degree .999999. So what is at stake in this discussion? I believe the numbers in Table 3.4 reveal that utilitarians and multidimensional consequentialists have roughly the same, but not exactly the same, intuitions about Portmore, “Review of Martin Peterson’s The Dimensions of Consequentialism,” 748. 22 Here I focus on the version of multidimensional consequentialism in which equality and total well-​being are the relevant moral aspects. However, everything I say here holds true mutatis mutandis for the version of the theory (discussed below) in which everyone’s individual well-​being levels count as separate aspects. 21

62    Martin Peterson

Table 3.4.  Implications of Portmore’s objection Act Right w.r.t. well-​being Right w.r.t. equality Act 1 Act 2 Act 3

.499950 1 .000001

.500100 .999999 1

the three alternatives in Portmore’s example. The key difference is that multidimensional consequentialists, unlike utilitarians, can capture the fact that Act 2 brings about a tiny bit of inequality in their final moral verdict. So by concluding that Act 2 is right to degree .999999 instead of 1, multidimensional consequentialists can articulate a more nuanced verdict than utilitarians. Act 2 is almost entirely right, but because the act produces some inequality it is also a tiny bit wrong. Let me now explain where I believe Portmore’s reasoning goes wrong. As I understand his objection, he believes that the morally best alternative is always entirely right, even if the best alternative is not optimal with respect to all moral aspects. Otherwise, he cannot explain why he thinks Act 2 is entirely right. However, it is a mistake to believe that the best alternative is always entirely right, for at least two reasons. First, Portmore’s assumption prevents us from expressing the type of nonbinary moral nuances that feature in the multidimensional analysis described above. If we believe that these moral nuances are important moral phenomena that a plausible moral theory should be able to account for, then we ought to reject Portmore’s assumption. Second, there is an important practical difference between the verdicts reached by utilitarians, who accept Portmore’s assumption, and multidimensional consequentialists, who reject it. In another paper, Portmore formulates the following principle, which he thinks, “is, perhaps, the most fundamental and least controversial normative principle concerning action”:23 “[W]‌e ought to perform the alternative that we have most reason (all things considered) to perform. Of course, there could be more than one alternative that is tied for first place. So, to be a bit more careful, I should say that we ought to perform one of the alternatives that we have optimal reason to perform. More formally, the principle that I have in mind holds:

W. Portmore, “Perform Your Best Option,” 437. 23



Multidimensional Population Ethics    63

PYBO1 A subject, S, is permitted to perform an action, j, if and only if (and because) j-​ing is an alternative for S and there is no other action that is an alternative for S that S has more reason to perform.24 Multidimensional consequentialists believe PYBO1 is false. In their view, this is not “the most fundamental and least controversial normative principle concerning action.” Consider, for instance, a case in which some alternative (Act 1) is morally right to degree .8, and some other alternative (Act 2) is right to degree .9. There are no other relevant differences. So, all things considered, subject S has more reason to perform Act 2 than Act 1. However, it does not follow that S is therefore permitted to perform only Act 2. On the contrary, multidimensional consequentialists claim that the agent is permitted to carry out each alternative with a probability that reflects its degree of rightness (and possibly some other relevant considerations). The higher the degree to which the act is right, the higher should the probability be that the act is performed, everything else being equal. Alternatives that are right to degree 1 should be carried out with probability 1 (unless more than one alternative is right to degree 1, as explained above), and alternatives that are right to degree 0 should be performed with probability 0.25 Why should we accept this view? Why not always perform the best alternative, which we have most reason to perform? I  admit that my argument for preferring the view I propose is far from conclusive, but I nevertheless think it is the best argument one can give. Here it is: The fact that Act 1 is morally right to some positive degree should be properly reflected in the action performed by a morally motivated and fully informed and rational decision-​maker, even if Act 2 is right to a somewhat higher (but nonmaximal) degree. We can think of this as a type of fittingness relation between actions and reasons. My objection to Portmore is thus that it is more fitting to randomize between two alternatives that are somewhat right and somewhat wrong, because if you perform the alternative you have most reason to perform with probability 1, you thereby neglect the complexity of the normative landscape. Other means of doing justice to the complexity of the normative landscape, such as expressing regret about the shortcomings of all available options, would not be “visible” in the option eventually selected by the agent. Such attempts to do justice to the complexity of the normative landscape are merely cosmetic. If you perform Act 2 with probability 1 because this option is right to a higher degree than Act 1, your Portmore, “Perform Your Best Option,” 436. 25 If it is deemed implausible that acts that are right to a very low degree are permitted to be performed with some nonzero probability, then one could add that only acts that are right to a degree above some threshold should be performed with some nonzero probability. In Peterson, Dimensions of Consequentialism, I did not discuss this possibility, but I now think this might be an attractive response to some potential counterexamples. 24

64    Martin Peterson

action would not accurately reflect the fact that Act 1 was also right to a fairly high degree. Your behavior would have been the same if Act 1 had been right to degree 0. 3.4. The Repugnant Conclusion from a Multidimensional Perspective

Let us now discuss how multidimensional consequentialists analyze the Repugnant Conclusion. The verdict reached by the multidimensional consequentialist will, of course, be sensitive to what she takes the relevant moral aspects to be. As mentioned earlier, I believe that each and every person counts as a separate moral aspect, rather than the total amount of aggregated well-​being. I also believe that equality counts as a separate moral aspect.26 The claim that each person counts as a separate moral aspect is, however, easy to misunderstand. It is therefore appropriate to say a bit more about what I mean by this. Strictly speaking, it is not the person herself that counts as a moral aspect, but rather the changes we bring about to each person’s well-​being. The following example helps to explain what I have in mind: Imagine that you can either preserve the status quo in society or increase the well-​being of the worst-​off person by one unit by reducing the well-​being of the best-​off person by one unit. On the multidimensional view I propose, there are three moral aspects to consider: (i) equality, (ii) changes to Mr. Worst-​Off ’s well-​being, and (iii) changes to Mr. Best-​ Off ’s well-​being. With respect to two of these aspects, (i) and (ii), it is right to redistribute well-​being from Mr. Best-​Off to Mr. Worst-​Off. That act would lead to more equality, and to more well-​being for Mr. Worst-​Off. However, with respect to aspect (iii), Mr. Best-​Off, the act is wrong. This is because Mr. Best-​Off will end up with less well-​being. Needless to say, it is possible that this redistribution of well-​being is right to a much higher degree than it is wrong, but it is not entirely right.27 Acts that lead to suboptimal outcomes for some agents are always wrong to some (perhaps very low) degree. Only acts that are optimal for everyone are entirely right.28 See ­chapter 3 in Peterson, Dimensions of Consequentialism, for a detailed defense of the claim that persons count as separate aspects. 27 For a discussion of the implications of this view for the Pareto principle, see Peterson, Dimensions of Consequentialism,  24–​25. 28 A possible objection to the suggestion that equality as well as individual well-​being count as separate moral aspects, proposed by Bader, “Review of Martin Peterson’s The Dimensions of Consequentialism,” is that every plausible measure of equality (such as the Gini index) seems to depend on individual well-​being levels. It thus seems difficult to hold all well-​being levels fixed in the population but vary its equality. However, as I explain in The Dimensions of Consequentialism (56–​65) the claim that it is individual well-​being levels that matter is, strictly speaking, more complex. My view is that the best measure of how well things go for an individual at time t should be based on what I call “individual moral value.” This, in turn, is a measure that depends on two variables: (i) on how high or low the individual’s well-​being is at t, and (ii) on whether the individual’s 26



Multidimensional Population Ethics    65

Let us apply this version of multidimensional consequentialism to the Repugnant Conclusion. According to the view I  propose, it is right to some degree to bring about the A-​world instead of the Z-​world, but doing so is also wrong to some degree. All things considered, it is neither entirely right nor entirely wrong to bring about the A-​world instead of the Z-​world. If each person counts as a separate moral aspect, then there will be a large number of people in the Z-​world who benefit if we bring about that world. It is better to be alive and lead a life worth living than to not exist at all (with respect to that moral aspect), even if one’s well-​being is low. However, there is also a small group of people who would have been better off in the A-​world but do not exist in the Z-​world. If we believe that each individual is a separate moral aspect, then we should count the well-​being of each person not just in the possible worlds in which she is alive, but also in the worlds in which she could have existed. This does not amount to saying that being alive with a well-​being of zero units (or some other neutral level, depending on how the scale is constructed) is morally equivalent to not existing at all. The claim at stake here is just the following:  With respect to one moral aspect, individual i, it is morally right to see to it that i exists and has a life worth living, compared to not existing at all. This is a claim about rightness with respect to a single moral aspect, not with respect to rightness all-​things-​considered. Some traditional one-​dimensional theories entail that the large quantity of aggregated well-​being in the Z-​world outweighs the much smaller amount of aggregated well-​being in the A-​world, but multidimensional consequentialists can avoid this Repugnant Conclusion by insisting that each person counts as a separate moral aspect. The multidimensional analysis also has consequences for the Mere Addition Paradox. This holds true no matter whether we believe each person is a separate moral aspect or insist that it is the total aggregated well-​being that counts. Imagine, to start with, that equality and total well-​being are the only moral aspects. To some degree, it will then be right to choose the A+-​world instead of the A-​world because there is more well-​being in A+. However, it will also be wrong to some degree to choose the A-​world instead of the A+-​world, because there is some inequality in A+. Even if neither of the groups is aware that some other group is not equally well off, the mere fact that not everyone is equally well off makes it wrong to some degree with respect to equality to prefer A+ over A. It can be easily verified that the conclusion will be the same if we believe that each and every person counts as a separate moral aspect.

well-​being was higher or lower in the past. The flexibility built into this measure enables us to hold all individual moral values fixed at the same time as we vary the equality in the population.

66    Martin Peterson

Parfit explicitly denies that the inequality in A+ makes that world worse than A. He writes, “We cannot plausibly claim that the extra people should never have existed, merely because, unknown to them, there are other people who are even better off.”29 Multidimensional consequentialists will, of course, agree with this. However, the multidimensional consequentialist believes that this type of one-​dimensional analysis is too blunt. Rather than just concluding that it would not have been better (or right) all things considered if the extra people in A+ had not existed, multidimensional consequentialists offer a more nuanced analysis. On their view, it would be right to bring about the A+-​world instead of the A-​world because there is more (individual) well-​being in A+. But it would also be wrong to some degree to bring about the A-​world instead of the A+-​world, because there is some inequality in A+. The all-​things-​considered verdict is nonbinary. This is a possibility that is overlooked in Parfit’s analysis. 3.5. Arrhenius’s Sixth Impossibility Theorem

We are now in a position to explain how multidimensional consequentialists analyze Arrhenius’s impossibility results. Because multidimensional consequentialism is a claim about the rightness and wrongness of acts, and not about the goodness or badness of possible worlds, I will focus on Arrhenius’s sixth impossibility theorem.30 In my view, this is the most general and interesting formal result in the literature on population ethics. Arrhenius’s sixth impossibility theorem is derived from six normative conditions. The theorem simply says that no moral theory can satisfy all of them. The challenge for the multidimensional consequentialist is, thus, to show that we have reason to reject at least one of the six conditions. The first condition is the Condition of Separate Satisfiability: For any agent and any situation, it is logically possible for her not to act in a morally wrong way.31 This condition excludes the existence of moral dilemmas of the traditional type in which all alternatives are entirely wrong. It is thus a rather strong condition that not

Parfit, Reasons and Persons, 425. Italics in original. 30 Arrhenius, “Future Generations.” Note that the multidimensional response could be applied, mutatis mutandis, to all of Arrhenius’s theorems. 31 Arrhenius, “Future Generations,” 191. 29



Multidimensional Population Ethics    67

all one-​dimensional consequentialists accept; for instance, authors who believe that cyclical orderings are possible reject this condition.32 However, from a multidimensional point of view this condition is less controversial. Arrhenius distinguishes between “wrong” and “not wrong” (and never uses the term “right” in his formal treatment). So multidimensional consequentialists can read “wrong” as “entirely wrong” but still believe that acts that are not entirely wrong may be wrong to a very high degree.33 Hence multidimensional consequentialists can account for intuitions about moral dilemmas (of the broad type in which we face a dilemma whenever every alternative is somewhat wrong) without rejecting the Condition of Separate Satisfiability. Multidimensional consequentialists thus have no reason to reject Arrhenius’s first condition. There seems to be no minimally plausible version of the theory according to which all alternatives in some situation are entirely wrong. Arrhenius’s second condition is the Normative Weak Quality Addition Condition: For any population X, there is at least one perfectly equal population with very high welfare such that if it is wrong in a certain situation to add this population to X, then it is also wrong in the same situation to add any population with very low positive welfare to X, other things being equal.34 This condition entails that it could be wrong to add a population with very low well-​being even if this would bring about more equality. Multidimensional consequentialists, therefore, reject the Normative Weak Quality Addition Condition. To see why, imagine a case in which the mere addition of a population with very low well-​being brings about more equality. In such a case, it would be right with respect to equality to add the population with very low positive well-​being, but wrong with respect to equality to add the perfectly equal population with very high well-​being mentioned in Arrhenius’s formulation of the condition.35 32 Temkin, Rethinking the Good, defends a conditional version of this claim. 33 For a multidimensional consequentialist, a moral dilemma is a situation in which no available alternative is entirely right, meaning that every alternative is somewhat (but perhaps not entirely) wrong. A consequence of this account of moral dilemmas is that the total number of dilemmas will be much larger than previously thought. Readers who find this worrisome can think of moral dilemmas in a narrow sense as cases in which every alternative is entirely wrong. 34 Arrhenius, “Future Generations,” 194. 35 In this discussion I do not wish to take a stand on which aggregation mechanism for all-​things-​considered rightness we have most reason to accept. Note, however, that if we opt for the maximin account discussed in section 3.2, the counterexample to Arrhenius’s condition has to be constructed such that the change in equality affects the minimum level of rightness that determines the acts all-​things-​considered degree of rightness.

68    Martin Peterson

Arrhenius’s third condition is the Normative Inequality Aversion Condition: For any triplet of welfare levels A, B, and C, A higher than B and B higher than C, and for any population A with welfare A [where A is a real number], there is a larger population C with welfare C such that if it is wrong in a certain situation to choose a perfectly equal population B of the same size as A∪C and with welfare B, then it is also wrong in the same situation to choose A∪C, other things being equal.36 This condition resembles, but is not identical to, the Pigou-​Dalton condition, which is frequently discussed in the literature on economic inequality. The Pigou-​ Dalton condition holds, roughly, that if we make the income distribution in the world more equal by redistributing income from the rich to the poor, without reducing the overall amount of economic resources, then the new world is at least as good as the initial one. The Normative Inequality Aversion Condition can be plausibly rejected by the version of multidimensional consequentialism that holds that every person counts as a separate moral aspect. On this view, any extra well-​being we give to a person, or any extra life we create that is worth living, contributes toward making the act right to a higher degree.37 So even if it is entirely wrong in some situation to choose a perfectly equal population B, it does not follow that it is entirely wrong in the same situation to choose A∪C, other things being equal. This is because each person in A is better off than the corresponding (but nonidentical) people in B. Therefore, A∪C scores better with respect to some moral aspects: the members of A. As explained earlier, it is better with respect to the A-​aspects to be alive and have a life worth living than to not exist at all. Choosing A∪C may thus be right to some degree at the same time as choosing B is entirely wrong. Arrhenius’s fourth condition is the Normative Non-​Extreme Priority Condition: There is a number n of lives such that for any population X, if it is wrong in a certain situation to choose a population consisting of the X-​lives, n lives with very high welfare, and a single life with slightly negative welfare, then it is also wrong in the same situation to choose a population consisting of the X-​lives and n+1 lives with very low positive welfare, other things being equal.38 Arrhenius, “Future Generations,” 192. 37 As noted earlier, I do not take any stand in this essay on how moral rightness is to be aggregated across different aspects. It is, of course, true that one can construct aggregation principles that preserve the conditions of Arrhenius’s theorems, but such maneuvers seem to be ad hoc and overly complex. 38 Arrhenius, “Future Generations,” 192. 36



Multidimensional Population Ethics    69

This condition can also be rejected by the sophisticated multidimensional theory mentioned above, according to which the well-​being experienced by each individual counts as a separate moral aspect. This holds true no matter how large n is. In order to see this, note that if we choose the population consisting of the X-​lives and n+1 lives with very low positive welfare we change the situation to the better for one person: the “single life with slightly negative welfare.” Therefore, this act is not entirely wrong. To some positive degree, which is likely to be tiny if n is large enough, the act is right. From a multidimensional perspective, an act is entirely wrong only if it is suboptimal with respect to every aspect, so no act that leads to an improvement for at least one person will be entirely wrong according to the sophisticated multidimensional theorist who holds that every person counts as a separate aspect. The upshot of all this is that multidimensional consequentialists can reject at least three of Arrhenius’s conditions. The remaining two conditions will not be discussed here. I  see no reason to reject them, but I  invite the reader to consider them carefully.39 To avoid any misunderstanding, I am not claiming that every version of multidimensional consequentialism is incompatible with the three conditions I  have discussed. All I have shown here is that some plausible versions of multidimensional consequentialism are incompatible with some of Arrhenius’s conditions. This invites the moral theorist to choose between two options. The first is to accept some such version of multidimensional consequentialism and reject Arrhenius’s impossibility theorem as irrelevant. The second is to insist that all of Arrhenius’s conditions are true and thus accept that no theory of population ethics is satisfactory. I leave it to the reader to adjudicate which option is best. References Arrhenius, Gustaf. “Future Generations:  A Challenge for Moral Theory.” F.D. dissertation, Uppsala University, 2000. Arrhenius, Gustaf. “Population Ethics and Different-​Number-​Based Imprecision.” Theoria 82 (2016): 166–​181.

Arrhenius’s fifth condition is the Normative Non-​Sadism Condition: “If it is wrong in a certain situation to add any number of lives with positive welfare, then it is also wrong in the same situation to add any number of lives with negative welfare, other things being equal” (Arrhenius, “Future Generations,” 194). The sixth of Arrhenius’s conditions is the Normative Egalitarian Dominance Condition, which is also a very reasonable condition that all minimally plausible one-​and multidimensional theories satisfy: “If population A is a perfectly equal population of the same size as population B, and every person in A has higher welfare than every person in B, then, in any situation involving a choice between A and B, it is wrong to choose B, other things being equal” (191). 39

70    Martin Peterson Bader, Ralf M. “Review of Martin Peterson’s The Dimensions of Consequentialism.” Dialectica 68 (2014): 620–​625. Crisp, Roger. “Rightness, Parsimony, and Consequentialism: A Response to Peterson.” Ethical Theory and Moral Practice 19 (2016): 39–​47. McLaughlin, Brian, and Karen Bennett. “Supervenience.” In Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Spring 2018 edition. https://​plato.stanford.edu/​archives/​spr2018/​ entries/​supervenience/​. Ng, Yew-​Kwang. “What Should We Do about Future Generations? The Impossibility of Parfit’s Theory X.” Economics and Philosophy 5 (1989): 235–​253. Norcross, Alastair. “Good and Bad Actions.” Philosophical Review 106 (1997): 1–​34. Parfit, Derek. Reasons and Persons. Oxford: Oxford University Press, 1984. Peterson, Martin. “Multi-​Dimensional Consequentialism.” Ratio 25 (2012): 177–​194. Peterson, Martin. The Dimensions of Consequentialism. Cambridge, UK: Cambridge University Press, 2013. Peterson, Martin. “Reply to Ralf Bader.” Dialectica 68 (2015): 625–​629. Portmore, Douglas W. “Perform Your Best Option.” Journal of Philosophy 110 (2013): 436–​459. Portmore, Douglas W. “Review of Martin Peterson’s The Dimensions of Consequentialism.” Journal of Moral Philosophy 13 (2016): 747–​750. Temkin, Larry S. Rethinking the Good:  Moral Ideals and the Nature of Practical Reasoning. Oxford: Oxford University Press, 2012.

4 CONS EQ UENTIA LISM A ND COORDINAT I O N

How Traditional Consequentialism Has an Attitude Problem

Douglas W. Portmore

According to act-​c onsequentialism (hereafter “consequentialism”), whether you should perform an act depends on the value of its consequences. But the value of its consequences depends on what you and others will formerly, simultaneously, and subsequently be doing. Imagine, for instance, that you’re driving behind a slow-​moving truck. Would it be good for you to accelerate? Well, it depends on whether you would simultaneously change lanes. For if you were to accelerate without changing lanes, you would crash into the back of the truck.1 Here, we have a coordination problem: the best available consequences will obtain only if you can coordinate your using your foot to press down on the gas pedal to accelerate with your using your hands to steer to the left to change lanes. More generally, the sort of coordination problem that I am interested in has the following structure: (1) S1 is deliberating at t0 about whether or not to ϕ at tx and (2) although S1’s ϕ-​ing at tx would not itself have good consequences, good consequences would ensue if both S1 ϕs at tx and S2 ψs at ty, where S1 and S2 are each time-​slices of an agent, but not necessarily of the same agent, and where x and y are both greater than 0 but where x can be less than, greater than, or equal to y.2 In the above case, which 1 This is based on an example from Goldman, “Doing the Best One Can,” 186. 2 These time-​slices are not instantaneous. S1 endures throughout t0–​tx. And S2 endures throughout ty, which lasts for however long it takes for her to ψ. One possibility, then, is that S1 and S2 are just two distinct time-​slices of

71

72    Douglas W. Portmore

I’ll call The Car, ϕ-​ing refers to accelerating, ψ-​ing refers to changing lanes, tx and ty both refer to t1, and S1 and S2 each refer to you at t1. But we can imagine several other variations on this general structure, depending on (1) what ϕ-​ing and ψ-​ing refer to, (2) whether S1 and S2 are time-​slices of the same agent, and (3) whether tx is earlier than, later than, or identical to ty. In this chapter, I will consider what consequentialists should say about how S1 should treat S2 and the possibility that she will ψ at ty. At one end of the spectrum, consequentialists could hold that, in deciding whether or not to ϕ at tx, S1 should always treat S2 as a force of nature over which she has no control and, thus, treat the possibility that S2 will ψ at ty as she would the possibility that a hurricane will take a certain path. On this view, S1 is to determine whether S2 will ψ at ty and should ϕ at tx only if S2 is going to ψ at ty. At the other end of the spectrum, consequentialists could hold that S1 should always treat S2 as someone available for mutual cooperation and, thus, treat the possibility of S2’s ψ-​ing at ty as something to be relied upon. On this view, S1 is to rely on S2’s cooperation and so play her part in the best cooperative scheme by ϕ-​ing at tx. A third and intermediate position would be to hold that whether S1 should, as of t0, treat S2 as a force of nature or as someone available for mutual cooperation depends on whether S1 can, as of t0, see to it that S2 will ψ at ty. On this view, what matters is not whether S2 is, as a matter of fact, going to ψ at ty, but whether S1 can, as of t0, see to it that S2 will ψ at ty. I’ll argue for this third position, and I’ll argue that sometimes S1 can, as of t0, see to it that S2 will ψ at ty by nonvoluntarily forming certain reasons-​responsive attitudes (e.g., certain beliefs, desires, and intentions).3 Thus, an important implication of my view is that consequentialists should be concerned not only with the same agent, with S1 being the t0–​tx time-​slice and S2 being the ty time-​slice. But another possibility is that S1 is the t0–​tx time-​slice of one agent and S2 is the ty time-​slice of some other (numerically distinct) agent. 3 The notion of a reasons-​responsive attitude is, perhaps, the same as Scanlon’s notion of a judgment-​sensitive attitude, an attitude that is sensitive to the subject’s judgments about reasons (What We Owe to Each Other, 20). But Scanlon’s notion is, if not distinct, misleading, for we can respond to reasons without having any judgments about what our reasons are. “We respond to reasons when we are aware of facts that give us these reasons, and this awareness leads us to believe, or want, or do what these facts give us reasons to believe, or want, or do” (Parfit, On What Matters, 493). Thus, we can respond to reasons while neither knowing that this is what we are doing nor having any judgments about our reasons (461). Reasons-​responsive attitudes include all and only those mental states that a rational subject will tend to have, or tend not to have, in response to reasons (or apparent reasons)—​facts (or what are taken to be facts) that count for or against the attitudes in question. So beliefs are clearly reasons-​responsive attitudes, for a rational subject will, for instance, tend to believe that it will rain in response to her awareness of facts that constitute decisive reasons for her believing this. Although reasons-​responsive attitudes include many mental states, they exclude feelings of hunger, nausea, tiredness, and dizziness, which are not responsive to reasons. Suppose, for instance, that I have too quickly consumed a good-​size meal and am still feeling hungry, as there has not yet been sufficient time for my brain to receive the relevant physiological signals from my stomach. Even if I am aware that I’ve eaten more than enough to be satiated, my hunger is not responsive to this awareness. Instead,



Consequentialism and Coordination    73

an agent’s voluntary actions, but also with her nonvoluntary attitude formations. Indeed, I will argue that consequentialists should hold that whether an agent ought to perform some future act depends on whether she would perform that future act if she were now to form and perform all the acts and attitudes that she is required to form and perform. Thus, I believe that traditional consequentialists have an attitude problem. What an agent ought to do depends on what attitudes she ought to form. But traditional consequentialists have no way of accommodating this fact. So I will propose a new type of consequentialism—​one that can accommodate the fact that what we ought to do depends on what attitudes we ought to form. 4.1. Two Extreme Positions

As I noted, the position that I will be arguing for lies between two extremes. At one end of the spectrum, we have actualist consequentialism (AC), which holds that an agent should ϕ if and only if the prospect of her ϕ-​ing is better than that of any alternative. More precisely, the view holds: AC  For any subject S and any available act ϕ, S is, as of t0, morally required to ϕ if and only if the prospect of S’s ϕ-​ing is better than that of any alternative.4 I’ve formulated AC in terms of the prospect of S’s ϕ-​ing as opposed to how the world would be if S were to ϕ, because there may be no fact of the matter as to how the world would be if S were to ϕ. For one, the laws of nature may be indeterministic. And if they are, there may be several different ways the world could turn out rather than just one way that it would turn out if S were to ϕ. For another, ϕ may be underspecified, and S may lack control over how ϕ is to be further specified. And in that case, there would be no fact of the matter as to how the world would have been had S ϕ-​ed. To illustrate, imagine that Sam is playing craps and passes on being the shooter. How would the world have been had Sam rolled the dice? Would it have been a world in which he rolled boxcars, snake eyes, a hard eight, or what? There may be no determinate answer to this question.5 For what he rolls depends on minute it is responsive only to the physiological signals that supposedly take about twenty minutes to travel from the stomach to the brain. 4 A subject’s properties can change over time. For instance, a subject may in the past have had the property of being required to ϕ even though she doesn’t now have this property due to her present inability to ϕ. And, on my parlance, the phrase “S is, as of t, obligated to ϕ” is equivalent to “S has at t the property of being obligated to ϕ.” And, likewise, the phrase “ϕ-​ing is, as of t, an option for S” is equivalent to “S has at t the property of having ϕ as an option.” 5 For a defense of this claim, see Hare, “Obligation and Regret.”

74    Douglas W. Portmore

differences in the force and trajectory of his throw, and he lacks the dexterity to determine the precise force and trajectory of his throw. In any case, the prospect of S’s ϕ-​ing is just a probability distribution over the various possible ways the world could turn out if S were to ϕ. And I’ll call one prospect better than another if and only if it has more expected value than the other. To understand AC better, recall the sort of coordination problem that I’m interested in: (1) S1 is deliberating at t0 about whether or not to ϕ at tx and (2) although S1’s ϕ-​ing at tx would not itself have good consequences, good consequences would ensue if both S1 ϕs at tx and S2 ψs at ty. On AC, S1 is to treat the possibility that S2 will ψ at ty as she would the possibility that a hurricane will take a certain path. That is, she is to determine its course and act accordingly. This is true, on AC, even if S1 and S2 are time-​slices of the same agent, even if tx is identical to ty (suppose that they’re both t1), and even if S1 has, as of t0, the same control over whether she (S2) ψs at t1 as she has over whether she (S1) ϕs at t1. To illustrate, consider again The Car and imagine that you are so frustrated with the slow-​moving truck in front of you that you have formed the intention to accelerate while steering straight ahead so that you’ll crash into the back of the truck. Given this intention, you would not change lanes if you were to accelerate. So the prospect of your accelerating is much worse than the prospect of any alternative. Thus, AC implies that you should not accelerate. But AC has this implication even if your job depends on your accelerating while changing lanes. For suppose that the only way that you’ll get to work on time—​and let’s assume that your job depends on your getting to work on time—​is by accelerating while changing lanes so as to get around this slow-​moving truck. If you accelerate without changing lanes, you will rear-​end the truck, which will result in your losing both your job and your driver’s license. If you change lanes without accelerating, you will be rear-​ended by the Ferrari that’s coming up fast from behind in the only adjacent lane, and this too will result in your losing both your job and your license. And if you just stay behind the slow-​moving truck at the same rate of speed, you’ll be late for work, lose your job, but keep your license. So the only way to keep both your job and your license is to accelerate while changing lanes. Assume for the sake of simplicity that the relevant objective probabilities are all either 0 or 1. Thus, the probability that you’ll be rear-​ended by the Ferrari if you change lanes without accelerating is 1, and the probability that you’ll keep your job if you are late for work is 0. Given this, the expected values of the four options are as depicted in Table 4.1. AC holds that when deciding what you should do with your foot (such as whether to use it to press down on the gas pedal to accelerate), you should treat what you would simultaneously do with your hands (such as use them to steer straight ahead so as not to change lanes) as an immutable force of nature. You are to determine its



Consequentialism and Coordination    75

Table 4.1.  The Car Foot: Accelerate at t2 Foot: Accelerate at t2

Hands: Change lanes at t2

Hands: Don’t change lanes at t2

0 −10

−10 −5

course and act accordingly. And since your intention is to rear-​end the slow-​moving truck, you correctly determine that you would not steer the wheel so as to change lanes even if you were to accelerate. Thus, AC implies that you should not accelerate even though you must do so in order to keep your job.6 The problem, then, with AC is that it has you deliberate as follows: “(A1) I have a conditional obligation to refrain from accelerating if I’m not going to change lanes; (A2) as a matter of fact, I’m not going to change lanes (for I intend to rear-​ end the truck); and, therefore, (A3) I have an unconditional obligation to refrain from accelerating.” The problem with deliberating in this way is that it treats the truth of A2 as fixed for the purposes of deliberation when its truth depends on the course of your deliberations. For suppose that you were instead to deliberate as follows: “(B1) I have an unconditional obligation to act so as to keep both my job and my driver’s license; (B2) in order to keep both my job and my driver’s license, I  must accelerate while changing lanes; and, therefore, (B3) I  have an unconditional obligation to accelerate while changing lanes.” If you were to deliberate in this way and, as a result, form the intention to accelerate while changing lanes, then this is precisely what you would do. And if this is what you would do, then A2 would be false. So it seems that we should treat your hands’ steering straight ahead not as some immutable force of nature but as an event over which you have control. For, indeed, you have as much control over what your hands will do as you have over what your foot will do. Both your hands and your foot will behave in accordance with your intentions, or so we’re assuming. Thus, in deciding what to do with your foot (e.g., whether to press down on the pedal to accelerate), you should look not to what

6 Although AC implies that it is impermissible for you to press down on the gas pedal with your foot to accelerate given that there is an alternative (such as not pressing down on the gas pedal and just continuing to follow slowly behind the truck) whose prospect is better than its prospect, AC implies that you are required to perform the conjunctive act consisting in your pressing down on the gas pedal with your foot to accelerate while steering with your hands to change lanes, for the prospect of your performing this conjunctive act is better than that of any alternative option that you have. Thus, AC conflicts with the thought that obligation is closed under logical entailment. Strangely, proponents of AC are content with this—​see, for instance, Jackson and Pargetter, “Oughts, Actions, and Actualism.” And see Portmore, Commonsense Consequentialism, 181–​183, for further discussion.

76    Douglas W. Portmore

your hands would do given your present intention, but to what your hands could do—​that is, to what your hands would do if, as a result of deliberating properly, you formed the intention both to press down on the gas pedal with your foot and to steer to the left with your hands to change lanes. So, AC seems implausible. And this brings us to the other end of the spectrum: possibilist consequentialism (PC), which holds that whether an agent should ϕ just depends on whether it is part of the best possible pattern of action involving her and others. More precisely, the view holds: PC For any subject S and any available act ϕ, S is, as of t0, morally required to ϕ if and only if S’s ϕ-​ing is part of every optimal maximal pattern of action that is possible as of t0. A pattern of action is a set of actions that spans across times and/​or individuals. So my picking up takeout, bringing it home, and then my wife and I sitting down to eat it together is one pattern of action, and the two of us going out for dinner is another. And some patterns of action are more specific than others, where one pattern is more specific than another if and only if the performance of the one logically necessitates the performance of the other but not vice versa. Thus, my picking up takeout, bringing it home, and then my wife and I sitting down to eat it together is more specific than my picking up takeout. And my picking up takeout from the Chinese restaurant across the street is more specific than my picking up takeout. A maximal pattern of action, then, is one that is maximally specific, where a possible pattern of action is maximally specific if and only if there is no other possible pattern of action that is more specific than it. And a pattern of action is possible if and only if all the acts in which that pattern consists would be performed provided that each of the agents involved in that pattern had the right intentions at the right times.7 More precisely, a pattern of action is possible as of t0 if and only if there is some schedule of intentions, I, extending over both a set of agents, S, and a time-​interval, T, beginning at t1, such that the following are all true: (a) if the intentions of those in S were to follow I, then all the acts in which that pattern consists would be performed; (b) each of the agents in S has at t0 the capacity to continue, or to come, to have the intentions that I specifies for her at t1; and (c) for any time tx in T after t1 (1 < x), if the intentions of those in S followed I up until tx, then each agent in S would, just before tx, have the capacity to continue, or to come, to have the intentions that I specifies

7 Here, by ‘the right intentions’, I  don’t mean ‘the correct or appropriate intentions’, but instead ‘the specific intentions that are required to realize the pattern’.



Consequentialism and Coordination    77

Table 4.2.  The Buttons Coop: Pushes at t3 Coop: Not-​pushes at t3

Uncoop: Pushes at t3

Uncoop: Not-​pushes at t3

10 0

0 6

for her at tx.8 Lastly, a maximal pattern of action that’s possible as of t0 is an optimal one if and only if there is no other maximal pattern of action that’s possible as of t0 whose prospect is better than its prospect. The upshot of this definition of ‘possible’ is that a pattern of action can be possible even if that pattern will not be realized, and even if no agent can see to it that that pattern will be realized. Thus, the pattern of action consisting in my picking up takeout, bringing it home, and then my wife and I sitting down to eat it together counts as a possible pattern of action even if we are so angry with each other at the moment that, no matter what anyone says or does at present, we’re not going to sit down together. This pattern of action is, nevertheless, possible at present, because there is a schedule of intentions that meets criteria (a)–​(c) above. That is, if, after I  returned with the food, we were each to form the intention to sit down with the other, then that’s what we would do. There may be nothing that I, she, or anyone else could do or say at present to get each of us to form this intention, but so long as we each have the capacity to form this intention and would sit down together if we did, then the pattern counts as possible in the sense defined above. Unlike actualist consequentialism, possibilist consequentialism has plausible implications in The Car. Possibilist consequentialism implies that you’re required to accelerate.9 After all, all the optimal maximal patterns of action that are possible as of t0 include your accelerating while changing lanes so that you keep both your job and your driver’s license. Unfortunately, though, possibilist consequentialism has implausible implications in the following case. I call it The Buttons, because it involves two individuals, Coop and Uncoop, each with a button in front of him.10 Depending on whether each pushes his button at t3, the consequences will be as depicted in Table 4.2. 8 This definition of what it is for a pattern of action to be possible is modeled after Ross’s definition of a performable option; see Ross, “Actualism, Possibilism, and Beyond,” 81. 9 If you’re having trouble seeing how you could be morally required to accelerate, then just imagine that you have a moral obligation to take care of your family and that to do so you must hold onto both your job and your driver’s license. 10 This is based on a case from Regan, Utilitarianism and Co-​operation.

78    Douglas W. Portmore

Assume that Coop (the cooperative one) desires that they both push at t3. But, unfortunately, Uncoop (the uncooperative one) lacks this desire. And Coop is no dupe; he will push his button at t3 if and only if he finds that Uncoop desires at t2 that they both push. And let’s assume that they will each be able to read the other’s mind at t2 to see what the other desires. So Uncoop will not-​push at t3, because he has no desire to push. And Coop will not-​push at t3 because he reads Uncoop’s mind at t2 and finds that he lacks the desire that they both push. Although Coop will not-​push at t3, we can still ask:  Should he push? And, according to PC, the answer is that Coop is required to push his button at t3. For Coop’s pushing his button at t3 is part of every optimal maximal pattern of action that is possible as of t0. All the optimal maximal patterns of action that are possible as of t0 include both Coop’s pushing his button at t3 and Uncoop’s pushing his button at t3, for only if they both push at t3 will the optimum ten utiles be produced. If neither pushes, a mere six utiles will be produced. And, if only one pushes, zero utiles will be produced. So if Coop abides by PC, he’ll push his button at t3, and this will result in zero as opposed to six utiles being produced. But requiring Coop to push in this situation seems absurd. If Coop knows that he is unable to influence Uncoop’s behavior, then Coop should treat Uncoop’s behavior as a force of nature, like a hurricane. That is, he should determine its course and act so as to secure the best possible consequences given its determined course. The lesson in all this, I believe, is that we need to find some consequentialist view that lies between these two extremes. At one extreme, actualist consequentialism (AC) holds that, in every instance of the coordination problem that I’ve been describing, S1 is to treat the possibility that S2 will ψ at ty as she would the possibility that a hurricane will take a certain path. We should reject this view, for, in some instances (such as in The Car), S1 has, as of t0, as much control over whether or not S2 will ψ at ty as she does over whether or not she will ϕ at tx. S1 should not treat S2 as a force of nature when deciding whether or not she will ϕ at tx if she has the same control over whether or not S2 will ψ at ty as she does over whether or not she will ϕ at tx. At the other extreme, possibilist consequentialism (PC) holds that, in every instance, S1 is to treat the possibility that S2 will ψ at ty as she would the possibility that she will change lanes in The Car. We should reject this view as well, for, in some instances, S1 has no more control over whether or not S2 will ψ at ty than she does over what path a hurricane will take. So what we need is a version of consequentialism that is sensitive to whether or not S1 can see to it that S2 will ψ at ty. It is to these sorts of views that I now turn.



Consequentialism and Coordination    79

4.2. Cooperative Consequentialism

Donald Regan’s cooperative consequentialism (CC) is one view that lies between these two extremes, but, as we’ll see, it’s untenable.11 Even so, it is worth considering, for it teaches us some important lessons, which will, in the end, help us find a better theory. The view holds: CC For any subject S and any available act ϕ: S is, as of t0, required to correctly apply the CC decision procedure. And S is, as of t0, required to ϕ if and only if S would ϕ if S were to correctly apply the CC decision procedure. Cooperative consequentialism is unlike most moral theories (including AC and PC) in that it requires agents to do more than just perform certain voluntary actions. It requires that they also correctly apply a certain decision procedure: viz., the CC decision procedure. Now, the CC decision procedure is quite complex, but the basic idea, which is sufficient for our purposes, is simple enough: each agent should (1) hold “himself ready to do his part in the best pattern of behavior for the group of cooperators,” (2) “identify the other agents who are willing and able to co-​operate in the production of the best possible consequences,” and (3) “do his part in the best plan of behaviour for the group consisting of himself and the others so identified, in view of the behaviour of non-​members of that group.”12 So, on CC, “correct behaviour is required, but certain attitudes and beliefs are required as well.”13 Agents must have a desire or willingness to cooperate with each other and have true beliefs about which others are willing to cooperate. Thus, as Regan puts it, CC is not exclusively act-​oriented.14 If an agent performs the act that she would have been directed to perform had she correctly applied CC’s decision procedure but does so without correctly applying CC’s decision procedure, then although she has performed the right act, she has not done all that the theory requires of her. To illustrate this unique and important feature of CC, consider a variant on The Buttons in which Coop does not hold himself ready to cooperate. Assume that he Regan calls his theory cooperative utilitarianism as opposed to cooperative consequentialism, but given both that he is not concerned to advance any particular view (let alone a utilitarian one) about what consequences are good and that he explicitly says that he chose the former term only for the sake of familiarity, it seems that the latter term is actually the more appropriate one. See Regan, Utilitarianism and Co-​operation, 1. 12 Regan, Utilitarianism and Co-​operation, x, 135. 13 Regan, Utilitarianism and Co-​operation, 124. 14 A theory T is not exclusively act-​oriented if and only if T requires of agents something more than just the performance of certain voluntary acts. This is not Regan’s definition, for he provides no definition; see Utilitarianism and Co-​operation, 109. But I believe that this definition captures (at least, roughly and sufficiently well for our purposes) the notion that he has in mind. 11

80    Douglas W. Portmore

doesn’t even bother to read Uncoop’s mind, forms no belief as to whether Uncoop is willing to cooperate, and not-​pushes simply out of inertia. If so, Coop has performed the right act, for given Uncoop’s unwillingness to cooperate, the best plan of behavior for the group consisting of himself and all others willing to cooperate (the empty set) is the one where he not-​pushes. But he has not done all that CC requires of him. For, on CC, Coop was required not only to not-​push but also to identify Uncoop as a noncooperator and to believe him to be such, which is something that he failed to do.15 Requiring agents to “do” more than just perform certain voluntary acts may seem unusual for a moral theory, but it is essential to CC’s ability to avoid the sorts of counterintuitive implications that plague actualist consequentialism (AC).16 Consider, again, The Buttons. AC fails to require Uncoop to “do” all that he must do to secure the best possible outcome. To secure this outcome, Uncoop must not only voluntarily push but also nonvoluntarily form the desire that they both push. For forming this desire by t2 is the only way to ensure that Coop pushes. Despite this, AC doesn’t require Uncoop to form this desire. Nor does it require Uncoop to push. Strangely, AC requires Uncoop to not-​push, because given his lack of a desire that they both push, Coop is not going to push. And given that Coop is not going to push, the prospect of Uncoop’s not-​pushing is better than the prospect of Uncoop’s pushing. Thus, Uncoop should, according to AC, not-​push. This problem arises for AC because it focuses exclusively on acts when the best possible outcome will be realized only if Uncoop desires at t2 that they both push. But being exclusively act-​oriented, AC is unable to require Uncoop to nonvoluntarily form this desire or any other attitude. Of course, AC could require Uncoop to perform whatever voluntary act might cause him to desire that they both push—​ perhaps, the act of going to a talented hypnotist who can implant the desire in him. But let’s assume that there is no such available act and, thus, that the only way that Uncoop can come to desire that they both push is by recognizing and responding appropriately to the decisive reason that he has to form this desire—​viz., the fact 15 Forming the belief that Uncoop is a noncooperator is not something Coop would do voluntarily—​that is, as a result of his intending or willing himself to form this belief. For people don’t form beliefs at will—​at least, not typically. So, if Coop were to form the belief that Uncoop is a noncooperator, this would be done nonvoluntarily as a result of his reading Uncoop’s mind at t2 and finding him lacking in any desire to cooperate. 16 Perhaps it does not seem terribly unusual. There are, after all, moral theories (such as global consequentialism; see Pettit and Smith, “Global Consequentialism”) that require agents to have certain motives, intentions, and character traits, and not just to perform whatever acts might result in their acquiring them. Nevertheless, what makes CC stand out as being particularly unusual is that it makes an agent’s obligation to perform an act derivative of her obligation to do that something more. Thus, on CC, an agent’s fundamental obligation is to correctly apply the CC decision procedure, and she has a derivative obligation to perform a certain act if and only if she would perform that act if she were to fulfill this fundamental obligation.



Consequentialism and Coordination    81

that their both pushing would be optimal.17 This, I take it, is the typical way in which agents come to have attitudes such as beliefs, desires, and intentions. Typically, they form these attitudes not by willing or intending to form them but by responding to the reasons they have for forming them—​that is, in the way that one comes to believe that the floor is wet when one walks on a wet floor with one’s bare feet. The lesson, here, is that if we’re going to come up with a theory that avoids AC’s troubles with The Buttons, it will need to be one that’s not exclusively act-​oriented. CC avoids being exclusively act-​oriented by requiring agents not only to perform certain actions but also to correctly apply the CC decision procedure. Unfortunately, though, requiring agents to correctly apply the CC decision procedure is also its downfall. For there are two devastating objections to this. The first is that CC runs afoul of the ‘ought’-​implies-​‘can’ principle. To require agents to correctly apply the CC decision procedure is to require them to form certain beliefs and attitudes whether they have the capacity to form them or not.18 For instance, CC requires agents to form true beliefs about who is and isn’t willing to cooperate, about what all the noncooperators are going to do, and about what the best pattern of behavior for the group of cooperators is given what all the noncooperators are going to do.19 But an agent may not have the mental capacity to form such beliefs. Suppose, for instance, that the relevant agent is an infant boy who has the capacity to form the intention to push a button, but not the capacity to form a belief about what the best pattern of behavior is. Perhaps, he lacks the relevant concepts to even conceptualize such a belief. Or, perhaps, a mad neuroscientist has inserted electrodes into his brain so as to make it impossible for him to form such a belief.20 Regardless, CC requires the boy to form a true belief about what the best pattern of behavior is. And this is absurd. The boy can’t be required to form beliefs that he lacks the capacity to form and so can’t form. The second problem with CC’s requiring agents to apply the CC decision procedure is that sometimes the costs associated with applying the CC decision procedure far outweigh its benefits. And, in some instances, applying the CC decision procedure may even be disastrous. No consequentialist theory should require an agent to ϕ (e.g., to apply the CC decision procedure) if whether or not she ϕs is a voluntary

I’m assuming that the fact that their cooperating would produce the most good constitutes a decisive reason for Uncoop to desire their mutual cooperation. Thus, it is a fact about the object of the desire, not any fact about the having of the desire, that constitutes a decisive reason for Uncoop to desire that they both push. So I  avoid any worries associated with the wrong-​kind-​of-​reasons problem. See Rabinowicz and Rønnow-​ Rasmussen, “The Strike of the Demon,” for a discussion of the problem. 18 This objection originates with Barley, “Review,” 158–​159. 19 Regan, Utilitarianism and Co-​operation, 151–​158. 20 Barley, “Review,” 158. 17

82    Douglas W. Portmore

choice and choosing to ϕ would have disastrous consequences.21 To illustrate, suppose that there is a mad telepath who will monitor Coop’s decision-​making process in The Buttons and will blow up the whole world if Coop applies the CC decision procedure. And assume that whether or not Coop applies the CC decision procedure is a voluntary choice—​that is, although fully and correctly applying the CC decision procedure involves more than just performing certain voluntary acts (it will, for instance, involve nonvoluntarily forming certain beliefs and other attitudes), choosing to proceed via this decision procedure is a voluntary choice. In any case, let’s call this revised version of The Buttons: The Mad Telepath. Even in cases such as The Mad Telepath, CC requires agents to apply the CC decision procedure.22 This is problematic for CC, for besides its being counterintuitive to think that Coop is required to voluntarily choose to apply the CC decision procedure when doing so would be disastrous, this implication of CC seems anticonsequentialist in spirit. Thus, we might wonder not only whether CC is a plausible consequentialist theory but also whether it is even a consequentialist theory at all. Regan is well aware that CC has this sort of counterintuitive implication. Indeed, I borrow the mad telepath example (slightly revised) from him.23 But Regan is willing to countenance such implications, because he thinks both that (C1) no theory that is exclusively act-​oriented can avoid counterintuitive implications in cases such as The Buttons and that (C2) no theory that is not exclusively act-​oriented can avoid counterintuitive implications in cases such as The Mad Telepath. Now, Regan would seem to be correct about C1. No exclusively act-​oriented theory can avoid counterintuitive implications in cases such as The Buttons. In such cases, the relevant agent can see to it that the best possible outcome is realized only by nonvoluntarily forming certain attitudes. Performing certain voluntary actions is not enough. To illustrate, consider yet another variant on The Buttons. Assume that, in this variant, Uncoop pushes his button but without desiring that they both push. Assume that he pushes only because his commanding officer orders him to do so. Given that Uncoop lacks a desire that they both push, Coop reads his mind, finds the desire lacking, and not-​pushes. Thus, the best possible outcome is left unrealized despite the fact that Regan does talk of there being a choice between applying the CC decision procedure and applying some other decision procedure; see Utilitarianism and Co-​operation, 179. Now, although I think that no consequentialist theory should require an agent to ϕ if whether or not she ϕs is a voluntary choice and her choosing to ϕ would have disastrous consequences, I think that consequentialists must allow that there is nothing wrong with, say, an agent’s nonvoluntarily forming a belief in response to her reasons even if her forming this belief would have disastrous consequences. Suppose, for instance, that an evil demon threatens to kill millions if I form the belief that it is raining and then she places me outside in the middle of a rainstorm. Here, I have no choice but to form the belief that it is raining despite the disastrous consequences of my doing so. 22 Regan, Utilitarianism and Co-​operation, 177. 23 See Regan, Utilitarianism and Co-​operation, 177. 21



Consequentialism and Coordination    83

Uncoop pushes. Clearly, then, it is not always sufficient to perform certain voluntary acts. Sometimes, the only way to ensure that the best possible outcome is realized is to form certain attitudes. But even if Regan is correct about C1, he is, I believe, incorrect about C2. Regan defends C2 by arguing that since any theory that’s not exclusively act-​oriented will have to require “something more” of agents than the mere performance of certain voluntary acts, “there is always the possibility that there will be a mad telepath in some situation who will blow up Macy’s [or the whole world] in response to that ‘something more.’ ”24 But, from this, it doesn’t follow that every theory that is not exclusively act-​oriented must require agents to voluntarily choose to do that “something more” even when the bad consequences of that something more far outweigh whatever good consequences it may have, which is exactly what gets CC in trouble. For instance, a theory that required Coop to decide to not-​push by way of choosing to apply a certain decision procedure (say, CC’s decision procedure) but only when the good effects of so choosing outweighed the bad effects would not be exclusively act-​oriented and, yet, unlike CC, would avoid counterintuitive implications in The Mad Telepath.25 So, as I  said at the beginning of this section, CC is untenable. It violates the ‘ought’-​implies-​‘can’ principle and has some very counterintuitive implications. But it was, as I promised, worth considering, for, in doing so, we have learned some important lessons—​three lessons, in fact: (L1)  a moral theory should not be exclusively act-​oriented—​more specifically, it should require agents to nonvoluntarily form certain attitudes and not just to voluntarily perform certain actions; (L2)  a consequentialist theory should never require an agent to voluntarily choose to ϕ when ϕ-​ing would have disastrous consequences; and (L3)  a moral theory should avoid running afoul of the ‘ought’-​implies-​‘can’ principle and so it should require agents to form and perform only those acts and attitudes that they can form and perform. In addition to these three lessons, we have, from section 4.1, learned a fourth: (L4)  a moral theory should lie somewhere between the extremes of actualist consequentialism and possibilist consequentialism and, thus, should be sensitive to whether or not the agent can, in the sorts of coordination cases that Regan, Utilitarianism and Co-​operation, 181. 25 I’m not claiming that this revised version of CC doesn’t have its own problems. 24

84    Douglas W. Portmore

I’ve been discussing, see to it that relevant coordination takes place—​that is, a moral theory should treat other acts26 (whether they be performed by the agent herself or by others) as she does the path of a hurricane only when the agent has no more control over whether these other acts will be performed than she does over whether a hurricane will take a certain path. In the next section, I make use of these lessons in formulating a more plausible alternative to cooperative consequentialism. 4.3. Rationalist Consequentialism

I’ve argued that an adequate moral theory will, in the sorts of coordination cases that I’ve been discussing, be sensitive to whether S1 has control over whether S2 ψs at ty—​that is, to whether S1 can see to it that S2 will ψ at ty. But I would deny that S1 must have voluntary control over whether S2 ψs at ty. Voluntary control, I take it, is the sort of control that we exert directly over our voluntary actions (such as raising one’s arm) and indirectly over those things that we manipulate via such actions (such as the movement of the car that one’s driving). Thus, S1 has, as of t0, voluntary control over whether S2 ψs at ty if and only if there is some act ϕ (which may or may not be identical to ψ) such that (1) whether S2 ψs at ty depends on whether S1 ϕs and (2) S can, as of t0, ϕ at will in the way that I can raise my arm at will. I deny that voluntary control is required, because, in many instances, S1 can see to it that S2 will ψ at ty even though S1 does not have voluntary control over whether S2 ψs at ty. To illustrate, consider again The Buttons. Uncoop lacks voluntary control over whether Coop pushes at t3, for there is no voluntary act that he could perform that would result in Coop’s pushing at t3. Yet Uncoop can see to it that Coop pushes at t3. He can see to it that Coop pushes at t3 by forming at t2 the desire that they both push. And he has control over whether he forms this desire. For he has, we’ll presume, the capacity both to recognize the fact that their both pushing would be desirable and to respond to this fact by forming the appropriate attitude—​namely, the desire that they both push. Now, some may question whether Uncoop is exercising any sort of control in forming the desire that they both push in response to his reasons. For control must be active, and forming an attitude in response to reasons seems less than fully active. But even if forming an attitude in response to reasons is not active in the way that performing a voluntary action is, it is, nevertheless, active. For it involves

If we’re assessing the deontic status of S’s ϕ-​ing at t and/​or S is deliberating about whether or not to ϕ at t, then the other acts are all and only those that are not S’s ϕ-​ing at t. 26



Consequentialism and Coordination    85

thinking, and thinking is active. Compare, then, the formation of an attitude with the feeling of a sensation (such as hunger or dizziness). The feeling of a sensation is completely passive. We simply suffer our sensations. By contrast, the formation of an attitude is active in that we shape our attitudes by attending to, reflecting upon, and responding to our reasons. So we do exert a kind of control over our reasons-​ responsive attitudes: it’s just not voluntary control. We may call the sort of control that we exert directly over our reasons-​responsive attitudes (e.g., beliefs, desires, and intentions) and indirectly over those things that we influence via such attitudes (e.g., our voluntary acts) rational control. As I see it, S1 has, as of t0, rational control over whether S2 ψs at ty if and only if (1) S1 has, as of t0, the capacity to respond appropriately at t1 to her reasons and (2) whether S2 ψs at ty depends (in the right way) on whether S1 responds appropriately at t1 to her reasons. Now, clearly, Uncoop has, as of t0, rational control over whether Coop pushes at t3. For he has, as of t0, the capacity to respond appropriately to his reasons by forming the desire that they both push, and Coop will push at t3 if he forms this desire. I think, then, that whether S1 ought, as of t0, to ϕ at tx depends on whether she has, as of t0, rational as opposed to voluntary control over whether S2 ψs at ty. But it may seem strange to tie an agent’s obligations to what she has rational as opposed to voluntary control over. But there are several reasons for thinking that it is right to do so. First, many of our obligations concern reasons-​responsive attitudes. For instance, parents ought to want what’s best for their children, agents ought to intend to take what they believe to be the necessary means to their ends, and people acquainted with the relevant scientific evidence ought to believe that the Earth is far more than a few thousand years old. Yet, we lack voluntary control over such attitudes. To see this, consider that if we had voluntary control over such attitudes, we could form them at will and, thus, do so to win a bet. But we can’t. I can’t, for instance, form the belief that Aristotle went for a swim on his thirtieth birthday just because you bet me $20 that I couldn’t.27 But we do have rational control over these attitudes. Thus, I would form this belief if I were to become aware of good reasons for thinking that this is so—​or, at least, I would insofar as I’m rational. Second, we often have obligations to perform mixed acts—​acts that have both a voluntary component and a nonvoluntary component. Such acts include acting in good faith, offering a sincere apology, and expressing one’s gratitude. To perform such an act, one must have certain reasons-​responsive attitudes. One can’t act in good faith without having the intention to follow through with one’s part of the bargain. One can’t offer a sincere apology without feeling contrite. And one can’t express one’s gratitude without feeling grateful. So whereas one may have voluntary See McHugh, “Epistemic Deontology and Voluntariness.” 27

86    Douglas W. Portmore

control over whether one says the words “I’m sorry” or “Thank you,” one doesn’t have voluntary control over whether one offers a sincere apology or expresses one’s gratitude.28 For we don’t have voluntary control over whether we feel contrite or grateful. Thus, we don’t have voluntary control over whether we perform such mixed acts. But we do have rational control over such mixed acts. For instance, I have rational control over both whether I utter the words “I’m sorry” and over whether I feel contrite. Third, a lack of rational control over the attitudes that led one to perform a certain voluntary act seems sufficient to undercut responsibility for that act. To illustrate, suppose that I have both the ability and the opportunity to kill my neighbor’s dog. And assume that, at present, I have no desire to kill or otherwise harm this dog and, indeed, have the same sorts of values with respect to nonhuman animals that other animal liberationists have. But, now, suppose that an evil neuroscientist abducts me and performs brain surgery on me, manipulating my brain in such a way that I no longer have the capacity to recognize and respond appropriately to the moral reasons there are for not harming nonhuman animals. Moreover, he implants in me a device that stimulates my neurons so as to cause me to form a very strong desire to harm dogs. Consequently, when I wake up from the surgery, I kill the dog. In this case, I did exercise my voluntary control in killing the dog. Yet I don’t seem responsible for having killed the dog, precisely because I did not exercise rational control in killing the dog. For these three reasons, it seems that what matters is rational, not voluntary, control. So, we should interpret the fourth lesson in terms of rational control. This lesson, you will recall, was that a moral theory should lie somewhere between the extremes of actualist consequentialism and possibilist consequentialism and, thus, should be sensitive to whether or not the agent has, in the sorts of coordination cases that I’ve been discussing, control over whether the relevant coordination takes place. More specifically, a moral theory should be sensitive to whether S1 has, as of t0, rational control over whether S2 ψs at ty. Now, if the consequentialist is going to take this (L4) and the other three lessons (L1–​L3) to heart, she’ll need to adopt what I call rationalist consequentialism (RC). On this view, whether you’re required to perform some future act depends on whether you would perform that act if you were, at present, to respond appropriately to your reasons. Admittedly, this by itself is not consequentialist. But what makes the view consequentialist is that it holds, additionally, that your responding appropriately at present to all your reasons entails your presently ϕ-​ing if and only if

See King, “Actions That We Ought, but Can’t.” 28



Consequentialism and Coordination    87

the consequences of your presently ϕ-​ing while otherwise responding appropriately to your reasons are optimific. More precisely, the view is: RC For any subject S and any available act ϕ: S is, as of t0, morally required to ϕ at tx if and only if responding appropriately at t1 to all her reasons entails her ϕ-​ing at tx (1 ≤ x). Responding appropriately at t1 to all her reasons entails her ϕ-​ing at t1 if and only if there is no ψ such that the prospect of her ψ-​ing at t1 while otherwise responding appropriately at t1 to her reasons is better than the prospect of her ϕ-​ing at t1 while otherwise responding appropriately at t1 to her reasons.29 The idea, then, is that in determining whether you are, as of t0, obliged to ϕ at tx, we must first look at what acts and attitudes you must form and perform at t1 in order for you to respond appropriately at t1 to all your reasons. And we conclude that you are, as of t0, obliged to ϕ at tx if and only if you would ϕ at tx if you were to form and perform at t1 all these acts and attitudes that you’re required to form and perform at t1. To better understand this view, it will be helpful to work through its implications in some of the above cases. Let’s start with The Car, and let’s again assume that you are so frustrated with the slow-​moving truck in front of you that your present intention (that is, your intention at t0) is to accelerate while steering straight ahead so that you’ll crash into the back of the truck. However, if you were to respond appropriately to your reasons, you would revise your intention and form at t1 the intention to accelerate and change lanes. And if you were to form this intention at t1, you would at t2 accelerate and change lanes. Thus, unlike AC, RC entails that you should accelerate at t2. And, given that you must accelerate at t2 in order to keep your job, that seems like the right answer. Next, consider The Buttons. Unlike PC, RC rightly implies that Coop is obligated to not-​push at t3. Since Uncoop is going to not-​push at t3 and since Coop is powerless to change this, Coop is obligated to form at t1 the intention to not-​push at t3, which is what would have the best consequences given that Uncoop is going not-​ push at t3. And given that Coop will not-​push at t3 if he forms at t1 the intention to not-​push at t3, RC entails that he is obligated to not-​push at t3. But things are different for Uncoop. If Uncoop were to respond appropriately to his reasons, he would form at t1 the desire that they both push. For he has decisive reasons to form this desire given that their both pushing is what would be best. Moreover, if Uncoop were to form at t1 the desire that they both push, Coop would then push at t3. Thus, I further develop a view like this in Portmore, “Maximalism and Moral Harmony.” 29

88    Douglas W. Portmore

the prospect of Uncoop’s responding appropriately at t1 to his reasons by forming at t1 the desire that they both push and then pushing at t3 is better than the prospect of his responding appropriately at t1 to his reasons by forming at t1 the desire that they both push and then not-​pushing at t3. For the prospect of the former is 10 utiles, whereas the prospect of the latter is 0 utiles. And given that Uncoop would push at t3 if he were to form at t1 both the intention to push at t3 and the desire that they both push, RC entails that Uncoop is obligated to push at t3. Lastly, RC implies, in The Mad Telepath, that you should not choose to proceed via the CC decision procedure, for the prospect of your proceeding via the CC decision procedure and otherwise responding appropriately to your reasons is much worse than the prospect of your proceeding via some other decision procedure and otherwise responding appropriately to your reasons. So, unlike CC, RC never requires you to voluntarily choose to do what would lead to a disaster. 4.4. Conclusion

As we’ve just seen RC takes into account all of lessons L1–​L4 and gets intuitive verdicts in all the above cases. Now, I  haven’t argued that RC is the only theory that both accounts for lessons L1–​L4 and gets intuitive verdicts in all these cases, let alone that it is the best of those that do. But I don’t yet see any better way of taking account of these lessons. And I’m unaware of a consequentialist theory that does a better job in these or other cases. So it seems to me that RC is, at least, a promising consequentialist theory. The hope, then, is that even if it doesn’t live up to its promise, I will have at least specified some important lessons that must be taken into account when seeking a replacement.30 References Barley, B. “Review of Utilitarianism and Cooperation by Donald Regan.” Noûs 18 (1984): 152–​159. Goldman, Holly S. [now Holly M. Smith]. “Doing the Best One Can.” In Values and Morals, edited by Alvin I. Goldman and Jaegwon Kim, 185–​214. Dordrecht: D. Reidel, 1978. Hare, Caspar. “Obligation and Regret When There Is No Fact of the Matter about What Would Have Happened If You Had Not Done What You Did.” Noûs 45 (2011): 190–​206.

For helpful discussions and comments, I thank Chrisoula Andreou, Richard Yetter Chappell, Dale Dorsey, Christian Seidel, Travis Timmerman, students in my fall 2013 seminar on individual and collective action, commentators on some of my posts at PEA Soup (http://​peasoup.typepad.com/​peasoup/​), and participants of the workshop “Consequentialism: New Directions, New Problems?” (held at Friedrich-​Alexander-​Universität Erlangen-​Nürnberg). 30



Consequentialism and Coordination    89

Jackson, Frank, and Robert Pargetter. “Oughts, Actions, and Actualism.” Philosophical Review 95 (1986): 233–​255. King, Alex. “Actions That We Ought, but Can’t.” Ratio 27 (2014): 316–​327. McHugh, Conor. “Epistemic Deontology and Voluntariness.” Erkenntnis 77 (2012): 65–​94. Parfit, Derek. On What Matters. Vol. 2. Oxford: Oxford University Press, 2011. Pettit, Philip, and Michael Smith. “Global Consequentialism.” In Morality, Rules and Consequences: A Critical Reader, edited by Brad Hooker, Elinor Mason, and Dale E. Miller, 121–​133. Edinburgh: Edinburgh University Press, 2000. Portmore, Douglas W. “Maximalism and Moral Harmony.” Philosophy and Phenomenological Research 96 (2018): 318–​341. Portmore, Douglas W. Commonsense Consequentialism:  Wherein Morality Meets Rationality. New York: Oxford University Press, 2011. Rabinowicz, Wlodek. and Toni Rønnow-​Rasmussen. “The Strike of the Demon: On Fitting Pro-​ Attitudes and Value.” Ethics 114 (2004): 391–​423. Regan, Donald. Utilitarianism and Co-​operation. New York: Oxford University Press, 1980. Ross, Jacob. “Actualism, Possibilism, and Beyond.” In Oxford Studies in Normative Ethics, vol. 2, edited by Mark Timmons, 74–​96. Oxford: Oxford University Press, 2012. Scanlon, Tim M. What We Owe to Each Other. Cambridge, MA: Belknap Press, 1998.

5 F ITT I NGNES S OB JE CT IONS TO CONSE QU EN T I A L I S M

Richard Yetter Chappell

Traditional critics of consequentialism, from Bernard Williams to Michael Stocker, have objected to the apparent implications of (maximizing) consequentialism for moral agency. The consequentialist agent—​an agent who has fully internalized the truth of consequentialism and has the attitudes and dispositions (if any) that would be appropriate given the truth of the theory—​may not seem a plausible contender for being a morally appealing or virtuous agent. According to the familiar caricature, the consequentialist agent would be “cold and calculating,” have “one thought too many” before acting, would regard others in an objectionably instrumental fashion—​as mere “receptacles of value”—​and be incapable of genuine friendship.1 Consequentialist moral theorists, for their part, have largely dismissed such character-​based objections as irrelevant to the truth of consequentialism properly understood as a criterion of rightness rather than a proposed decision procedure.2 New work in the foundations of ethics—​extending the fitting attitudes analysis of value to yield a broader notion of normative fittingness as a (or perhaps even the) fundamental normative concept—​provides us with the resources to clarify and renew the force of traditional character-​based objections to consequentialism. 1 See, e.g., Williams, “A Critique of Utilitarianism”; Williams, “Persons, Character and Morality,” 18; Regan, The Case for Animal Rights, 208–​211; Kapur, “Why It Is Wrong.” 2 Bales, “Act-​Utilitarianism.”

90



Fittingness Objections to Consequentialism    91

According to these revamped fittingness objections, consequentialism is incompatible with plausible claims about which attitudes are truly fitting. If a theory’s implications regarding the fittingness facts are implausible, then this can be taken to cast doubt on the truth of the theory. Section 5.1 explicates how traditional character-​based objections can be understood as challenging the consequentialist conception of a morally fitting agent, and why such “fittingness objections” are a challenge to consequentialism itself. Section 5.2 explains why I take consequentialism to have fittingness implications, and why standard consequentialist responses to character-​based objections are inadequate. Section 5.3 explores Railton’s “sophisticated” consequentialist psychology and argues that it, too, fails to address the problem. Section 5.4 introduces the notion of “well-​ calibrated” dispositions, by investigating the question whether it’s always rational to act on a disposition that it’s rational to acquire. Finally, in section 5.5, I draw on this conception of “well-​calibrated” dispositions to show how I  think the consequentialist can successfully respond to a range of paradigmatic fittingness objections. 5.1. The Fittingness Objection

The consequentialist—​and especially, the utilitarian—​agent is sometimes presented, in caricature, as one who calculates expected utilities before each decision, who finds the needs of those before his eyes to be no more salient than those inaccessible and far away, and who is ready and willing to commit atrocities in the name of efficiency, without hesitation or regret.3 Such an agent seems morally perverse, far from exemplifying the kind of ideal moral character one would expect to find in an agent who has internalized the true moral theory and has the kinds of attitudes and dispositions that are morally appropriate or fitting. You may initially doubt that consequentialism has any implications, deleterious or otherwise, for fitting attitudes. Full discussion of this concern must wait until section 5.2.1. To get clear on the basic idea in the meantime, the relevant concept may be grasped via the following pattern: It’s fitting to desire that which is good or desirable, to admire the admirable, believe what is (genuinely) credible, and so on. So, for example, if utilitarianism holds that what’s good is just the welfare of sentient beings, then the fitting utilitarian agent is one who desires just the welfare of sentient beings. If such desires are shown to be not actually fitting, then that’s just to say that utilitarianism is false: it makes mistaken claims about which things are desirable.

3 See note 1, especially Williams, “A Critique of Utilitarianism.”

92    Richard Yetter Chappell

Of course, what’s desirable (fitting to desire) may come apart from what it would be optimal or desirable to desire, just as what it’s fitting to believe (based on the evidence) may differ from the beliefs that are optimal (given various practical incentives).4 When these two kinds of assessment diverge, consequentialists will insist that what matters, practically speaking, is the promotion of value. So it may be that we should, in such cases, try to bring it about that we have optimal-​but-​unfitting attitudes. (Such an outcome is, after all, itself fitting to desire and to pursue.5 It may be that we ought to try to acquire a belief that p even if we ought not to believe that p. It may be rational to act so as to bring about an irrational belief or other attitude. What should I believe? and What should I bring it about that I believe? are very different questions—​one answered by norms of belief, and the other by norms of action.) But the practical primacy of value should not be taken to imply that questions of fittingness lack theoretical import. As we’ll see, the consequentialist needs to answer such questions if she is to offer an adequate response to character-​ based objections to her view. Fittingness objections work as follows. We begin with a sketch of an agent’s psychology that seems to accurately represent a consequentialist outlook or perspective, and yet also seems intrinsically defective or unfitting from a moral point of view. If it’s true both that (i) the described agent accurately represents a “fitting consequentialist agent”—​an agent that has fully internalized the truth of consequentialism and has the attitudes and dispositions (if any) that would be appropriate (fitting to their objects) given the truth of the theory—​and yet (ii) the described agent is morally unfitting, then these premises together cast doubt on the truth of consequentialism. Why is this? Analytically, if an agent that qualifies as “fitting” according to candidate moral theory X is actually morally unfitting, then X is not the true moral theory. If X were the true complete moral theory, then the X-​fitting agent would ipso facto be the morally fitting agent. Further, if an incomplete theory X cannot be coherently supplemented in such a way as to yield plausible verdicts about fittingness, then that casts doubt on X’s claim to even be part of the complete moral picture. If it is shown to be incompatible with plausible verdicts about fittingness, that would be grounds for thinking the theory simply false.

4 Practical reasons which render their target attitudes useful but not fitting are often called the “wrong kind of reasons.” Cf. Rabinowicz and Rønnow-​Rasmussen, “The Strike of the Demon”; Chappell, “Fittingness.” 5 What if an evil demon will punish you for desiring that you have the optimal first-​order desire, or for acting so as to bring it about? Then it’s desirable (fitting to desire) that you somehow simply acquire the optimal first-​ order desire (by brute chance, say) without in any way aiming or desiring to do so. Since a desire for the optimal first-​order desire is itself undesirable (due to the threat of punishment) despite being fitting, this means that it is also fitting to desire that you not desire that you have the optimal first-​order desire.



Fittingness Objections to Consequentialism    93

This is, I  believe, a powerful (and underappreciated) line of argument. Just as we have intuitions about what the morally right action would be in various cases—​ intuitions which must be brought into reflective equilibrium with any moral theory we can ultimately accept—​so too we have intuitions about fitting attitudes and character traits, or what a virtuous person would look like. If we can reasonably assign some default trust to these intuitions, then it’s a strike against a moral theory if it violates them. And if the violation is severe enough, such considerations could well provide decisive grounds for rejection. That is how we should understand character-​based objections to moral theories. Next I argue that the standard consequentialist dismissal of these objections is unwarranted. The form of the objection is one that needs to be taken seriously. 5.2. Why Two Standard Responses Fail

5.2.1. Must Consequentialism Have Fittingness Implications?

One might question whether there’s any such thing as “the fitting consequentialist agent.” I sometimes introduce fittingness talk in terms of what’s rationally warranted from the “point of view” of a moral theory, but you might well wonder whether moral theories are really the sorts of things that can have perspectives. If they’re not, or if consequentialism in particular needn’t have any implications regarding fittingness, then aren’t fittingness objections to consequentialism unable to get off the ground? I have two broad replies to this line of concern. My first (and more ambitious) response is to try to convince you that these ideas do all make sense. But I also have a more conciliatory backup option in case this fails. What is the “perspective” of a moral theory? It’s just an abstraction from the perspective of a rational agent who has fully internalized the theory, and whose psychology thus reflects, in isomorphism, the dictums of the theory—​being attuned to just the considerations that the theory identifies as morally significant. I find it plausible that any normative claim has some corresponding specification in the psychology of the fitting agent.6 It may even be (though I don’t need this, and won’t argue for it, here) that these implications for the fitting psychology are what ultimately give content to our various normative concepts and the claims that we make 6 See Chappell, “Fittingness.” For a rough intuitive motivation of this idea, note that a connection to warranted agential responses establishes the relevance of normative claims to us as agents. What would be the point in calling something “good,” for example, if goodness had no implications for what we are warranted in caring about? Free-​floating normative claims, that lacked any fittingness implications, would seem not to speak to us. But who else is there for normative claims to speak to? It seems they would be left silent, inert, and empty of normative significance.

94    Richard Yetter Chappell

with them. For example, given the conceptual link between goodness and desirability, to claim that the happiness of sentient beings is good straightforwardly implies that it is fitting to desire that sentient beings be happy. Anyone who failed to have such a desire would clearly not count as having properly internalized the thesis that happiness is good. More controversially: it may be that the appropriateness of desiring the general happiness is what gives content to the claim that it’s good. This link to fitting agential responses is what makes the normative claim significant to us as agents: it has implications for how we should be. Insofar as any particular consequentialist view presupposes a theory of the good, it has implications for the fittingness of the corresponding desires, at least. But if other sorts of normative claims are practically significant at all, they too must presumably have some sort of agential (fittingness) implications. Presumably right actions, for example, are those that it’s fitting for us to choose (or to intend). If a theory marks certain kinds of acts—​lying, say—​as inherently and absolutely wrong, this might be reflected in a fitting psychology by refusing to even entertain such acts as options.7 Any genuine moral considerations should presumably find some traction in the psychology of the fitting agent, whereas bad reasons (or nonreasons) shouldn’t. If I’m right about all that, then any moral theory will ultimately have fittingness implications, even if they aren’t explicit in canonical statements of the theory. So long as we can identify which attitudes and habits of thought inevitably follow from the proper internalization of a moral view, then we have a grasp on what the “fitting agent,” according to that theory, looks like. And we can then assess whether this agent plausibly is morally fitting. But suppose I’m wrong about all that. Suppose we can’t, strictly speaking, infer any fittingness claims (besides perhaps the value–​desirability link) from the core tenets of a moral theory. And suppose one were to endorse a conceptually “sparse” form of consequentialism which made no explicit claims about value, either, but just directly specified that agents morally ought to maximize happiness (or whatever). Would such a sparse view still be subject to fittingness objections? I think it would. This is because we could still raise questions about whether these basic deontic claims are compatible with plausible claims about fittingness. Even if the theory has no positive fittingness implications, it surely has negative implications, as there are constraints on what fittingness claims can coherently be combined with various deontic claims. For example, the deontic claim that we ought always to maximize happiness is clearly in tension with claims that it’s fitting to value or desire

7 Compare Anscombe’s worries about the consequentialist’s willingness to consider anything as betraying a “corrupt mind” (“Modern Moral Philosophy,” 17).



Fittingness Objections to Consequentialism    95

things other than happiness, or that a fitting agent would never even think of lying, or that it’s fitting to prefer to save your child’s life over that of two strangers. So there remains a real challenge here. Even the most conceptually sparse consequentialist must either argue that such fittingness claims are incorrect—​that they misdescribe what attitudes and patterns of thought are truly warranted or fitting—​ or else argue that her sparse consequentialism can be supplemented or developed in such a way as to coherently combine plausible fittingness claims with her original deontic claim. (Which of these two is the most promising strategy in any given case will, of course, depend on the details. For example, I don’t think it’s true that lying should always be unthinkable. But it seems right to me that certain kinds of actions should not generally be “on our radar,” and so I  go on—​later in the chapter—​to show how consequentialists might accommodate this.) For ease of exposition, I  will continue to speak of the “fitting consequentialist agent.” But if you are not convinced that consequentialism has positive fittingness implications, feel free to read this as shorthand for “the most plausible supplementary view of the ‘fitting agent’ that is compatible with consequentialism.” 5.2.2. Criteria of Rightness versus Decision Procedures

In response to character-​based objections, consequentialists standardly distinguish between criteria of rightness and decision procedures.8 Just because utilitarians hold that an act is right when it maximizes expected utility (say), it doesn’t follow that they recommend actually trying to calculate utilities in your everyday life. Indeed, given that such constant calculation would be predictably counterproductive (due to lack of time, misleading evidence, cognitive bias, setting bad precedents, etc.), utilitarians would strongly recommend against it!9 All this is true enough, but beside the point. I agree that it’s an open empirical question whether being morally fitting (whatever that turns out to involve) would bring about good results in our actual circumstances (just as it’s an open empirical question whether being rational more generally has positive instrumental value). If it wouldn’t have good results, then consequentialism may recommend against its own internalization—​the possibility of such “self-​effacingness” is a familiar (and unproblematic) feature of the view. Nothing I’ve said denies any of this: I haven’t claimed that we necessarily ought to try to become fitting agents, come what may. Any consequentialist should agree that there are more important things than the quality of our characters, after all. 8 Bales, “Act-​Utilitarianism.” 9 See, e.g., Mackie, “Rights, Utility, and Universalization.”

96    Richard Yetter Chappell

The objection is not to consequentialism’s recommendations but to its implications. The standard consequentialist response assumes that the only way that consequentialists can assess decision procedures (and psychological elements more generally) is in terms of their instrumental value, or whether they’re worth inculcating. In the previous section, I argued that this is not so. There’s also a fact of the matter as to what the “fitting” consequentialist psychology would be, quite independently of what psychology consequentialism recommends (on grounds of utility) that we try to inculcate. But if the fitting consequentialist psychology can be shown to be not actually morally fitting, that would—​as previously explained—​pose a serious problem for the view. So we can’t just ignore decision procedures and other psychological elements. Nor can we merely settle for identifying those which best promote value, and are thus recommended by consequentialism. As normative theorists, interested in whether or not consequentialism is a true moral theory, we must also investigate what kind of psychology would be a fitting psychology to possess, were consequentialism true. We can then assess whether this fitting consequentialist psychology is plausibly morally fitting, and hence whether consequentialism itself remains an eligible moral theory. In the following sections, I  explore two very different strategies for constructing a nondefective consequentialist psychology in answer to this challenge. First I consider the Railtonian “sophisticated” psychology, with nonconsequentialist desires. Then I explain and defend my preferred account, according to which critics are mistaken to assume that an agent with fitting utilitarian motivations would thereby conform to their caricature. 5.3. Sophisticated Consequentialism

5.3.1. Explication

Railton contrasts two kinds of hedonistic (or, more broadly, consequentialist) psychologies, which we may consider as candidate views of what’s fitting: “subjective” and “sophisticated” psychologies.10 The subjective hedonist is solely motivated by concern for his own happiness. However, the “paradox of hedonism” suggests that such a person is likely to end up quite unhappy. Happiness may be better achieved by those who are motivated by other concerns. Railton thus introduces the Railton, “Alienation, Consequentialism, and the Demands of Morality,” 142–​143. Below, I follow Railton in focusing on egoistic hedonism for simplicity. But the basic distinction between “subjective” and “sophisticated” psychologies should carry over, in obvious fashion, to other (including impartial) forms of consequentialism. Basically, the “subjective” consequentialist agent directly desires the things her theory specifies as good, whereas the “sophisticated” consequentialist agent instead possesses a desire-​regulating faculty that will tend to generate in her whatever desires can be expected to best serve to promote the good. 10



Fittingness Objections to Consequentialism    97

sophisticated hedonist—​let’s call her “Sophie”—​who “aims to lead an objectively hedonistic life (that is, the happiest life available to [her] in the circumstances) and yet is not committed to subjective hedonism.”11 Sophie may thus possess and act on distinctively nonhedonistic motives—​e.g., concern for others—​if such desires are conducive to her living a happier life overall. Once she has moved beyond subjective hedonism and acquired a happy collection of nonhedonistic motivations, we may begin to wonder in what sense Sophie is still a “hedonist” at all, rather than a whole-​hearted pluralist. What sets Sophie apart, according to Railton, is that her psychology continues to be regulated by a counterfactual condition according to which, despite her various desires, she “would not act as [s]‌he does if it were not compatible with [her] leading an objectively hedonistic life.”12 Whereas the subjective hedonist regulates her individual actions according to hedonistic norms, Sophie’s hedonism instead regulates her desires and dispositions. So, for example, her pro-​friendship disposition may lead Sophie to perform individual acts that reduce her happiness—​e.g., answering her friend’s distraught 3 a.m. call—​ but her “hedonic monitor” is not triggered to intervene unless it becomes clear that the relationship as a whole is detrimental to her happiness, such that she would be better off in the long run with different desires and dispositions. One may question how this regulative mechanism is supposed to work. In particular, we may wonder whether Sophie has an overriding desire to possess hedonically fortunate dispositions, that she will act upon (overriding her other, nonhedonistic desires) whenever she’s in a position to do so. But such an agent may be better described as a simple maximizer of happiness-​promoting dispositions, rather than a sophisticated maximizer of happiness! Sophie’s hedonism is better understood as manifested not in a desire at all, but rather a higher-​order mechanism that serves to regulate her desires through some subpersonal causal process. The key difference is that this hedonistic mechanism, unlike a desire, never directly manifests itself in action. It is not itself a motivation that she may act on (though it may cause her to acquire some independently motivating hedonistic desires, insofar as these would cause her to live a happier life). Its control over her actions is instead wholly indirect: the hedonic monitor shapes Sophie’s desires in hedonically fortunate ways, and then she acts on those desires, whatever they may be. The “sophisticated” psychology may thus be described in two parts: First, there is the agent’s overarching “primary goal,” which she may identify with during reflective Railton, “Alienation, Consequentialism, and the Demands of Morality,” 143. 12 Railton, “Alienation, Consequentialism, and the Demands of Morality,” 145. 11

98    Richard Yetter Chappell

moments, but which does not tend to directly motivate her actions. Instead, she is moved by the “secondary” desires and dispositions that are produced and regulated by a mechanism that is responsive to her primary goal. 5.3.2. Evaluation

Supposing that the psychology described in section 5.3.1 is coherent, it’s an interesting question how exactly we should evaluate it. Is it a plausible account of the fitting consequentialist psychology? According to (egoistic) hedonism, one’s own pleasure is the only end that’s truly desirable or worth pursuing. Sophie then seems irrational, by hedonistic lights, in that her desires are not necessarily directed at what is (according to this theory) desirable, and her actions likewise fail to be sensitive to hedonistic reasons: she often benefits others at her own expense. On the other hand, she is not completely insensitive to hedonistic reasons: her desire-​regulating faculty ensures that she maintains the desires that (the evidence suggests) it is hedonically best for her to have—​and if circumstances change, so will her dispositions. This suggests an important sense in which Sophie’s reflective hedonism is ultimately “in control,” even if it is not what moves her. We may thus need to draw a distinction between (local) act and (global) agent-​rationality, allowing us to say that Sophie is rationally fitting or responsive to reasons, even if her particular actions are not. It’s worth noting that even this vestige of rational sensitivity may, in special circumstances, make her worse off. Consider Parfit’s example of the society of perfectly rational egoists, some of whom come to realize that it will advance their interests to become irrational in a specific respect: namely, if they become transparently disposed to follow through on their threats regardless of the costs to themselves.13 Such a “threat-​fulfiller” can then strap a bomb to his chest and threaten an egoist that he will detonate it (killing them both) unless the egoist complies with his whims. He can safely make such threats, because he knows the egoist would sooner comply than die. As Parfit further shows, the rational response for the remaining egoists is to turn themselves into transparent “threat ignorers” who are stably disposed to (irrationally) ignore threats no matter the costs to themselves. A threat-​ fulfiller will leave the ignorers alone, because he knows that if he were to threaten them, they would ignore him, and he would then detonate the bomb, killing them both. (Note that the threat-​fulfiller will not issue threats that he expects will make him worse off. It is merely fulfilling threats that he does blindly.) In comparison to the pure threat-​ignorers, Sophie is more apt to have her instrumental rationality exploited. Given transparency, the threat-​fulfiller will know that Parfit, Reasons and Persons,  20–​21. 13



Fittingness Objections to Consequentialism    99

if he threatens Sophie, she will comply. For Sophie’s regulating mechanisms will not allow her to maintain a disposition once it becomes clear that it is disastrous for her long-​term happiness. And a threat-​ignoring disposition becomes clearly disastrous as soon as one is actually issued with a credible apocalyptic threat. So, a threat-​fulfiller will know that he can safely threaten Sophie, and she will (if necessary change her dispositions and) comply rather than die. To avoid such exploitation, Sophie would have to alter her psychology so that she would become a pure (unregulated, insensitive) threat-​ignorer—​at which point she would no longer be a sophisticated hedonist. She would just be (however fortunately) irrational, by hedonistic lights. We thus find that a Railtonian sophisticated psychology is by no means guaranteed to endorse itself as the most fortunate psychology to possess in every possible situation. But it offers a suggestive alternative to the standard conception of an instrumentally rational psychology. Insofar as we are drawn to the idea that rationality should not normally be a curse (even if it may be in certain special circumstances), we may see Sophie’s two-​level psychology—​with its capacity for her primary goal to control and regulate her secondary, action-​g uiding motivations—​as an improvement over the subjective hedonist’s unitary motivational structure. While acknowledging that Sophie’s actions are often locally irrational (by hedonistic lights), we may be more concerned to evaluate her global rationality as an agent. In this respect, at least, she may at first glance seem more reasonable. I think there are important grounds for doubting this conclusion, however. Let’s return our attention from hedonism to utilitarianism. The sophisticated utilitarian—​call her “Sophu”—​will have whatever motivations are most conducive to promoting the general welfare. So, in particular, if an evil demon threatens to torture an innocent population unless Sophu comes to intrinsically want them to suffer,14 then Sophu will be led to acquire this fortunate but malicious motivation.15 This is a good outcome, in the circumstances, as it prevents a lot of suffering. But if any desire is unfitting by utilitarian lights (or common intuition, for that matter), One might wonder if Sophu’s resulting pro-​suffering desire will be merely instrumental, since it is produced (by her regulating mechanism) as a means to promoting welfare. But we must take care to distinguish conditions on the desire’s existence from conditions on its content (or, in other words, to distinguish a desire’s persistence conditions from its fulfillment conditions). Sophu’s regulating mechanism ensures that the desire’s existence is contingent upon its promoting utility, but that doesn’t mean that considerations of utility enter into the content of the desire. Sophu is motivated to pursue suffering for its own sake. It’s just that her motivations will change when they cease to promote (expected) utility. 15 At least, this is so on my interpretation of the “sophisticated” psychology. Mason, “Do Consequentialists Have One Thought Too Many?,” 256, suggests an alternative view on which “[w]‌e can develop new motives from old motives, but only when they are consistent.” This would rule out a sophisticated utilitarian acquiring malicious motivations, however beneficial they might be. However, it’s not clear to me whether this is meant to be a conceptual constraint on rational agency, or just a contingent empirical hypothesis about how new motivations actually develop in people. 14

100    Richard Yetter Chappell

it is surely an intrinsic desire that others suffer. Sophu has, quite virtuously, made herself vicious. And note that it is not just her actions, but her desires—​her very self, we might think—​that is impugned here. She (noninstrumentally) desires what is blatantly undesirable, thus violating the most basic criteria for qualifying as a fitting agent. Yet she is still a sophisticated utilitarian. So this “sophisticated” consequentialist psychology is not adequate as an account of the fitting consequentialist psychology. The advocate of sophisticated utilitarianism might at this point defend Sophu’s utilitarian credentials by pointing out that her deepest commitments remain pure and altruistic, even as they respond to the unfortunate circumstances by shaping her motivations in this malicious-​but-​instrumentally-​valuable direction. So there at least remains something fitting about Sophu’s psychology. But it nonetheless contains potential moral defects of a sort that cast doubt on claims that she qualifies as a fitting utilitarian agent overall. So it is worth investigating whether we can do better with a more direct approach. 5.4. Rational Transmission and Well-​Calibrated Dispositions

We can identify where the “sophisticated” psychology goes wrong (or fails to accurately represent a fitting consequentialist perspective) by considering the relation between (i) the rationality of acquiring and maintaining a desire or disposition and (ii) the rationality of “acting on” the disposition, i.e., performing an action that the disposition characteristically disposes you toward. Consider the following simple principle of rational transmission: (RT-​past)  For any disposition D and act A that is characteristic of D: If it was rational to acquire D then it is rational to perform A. Parfit’s above-​described case of the threat-​fulfillers casts doubt on this principle. It may well be rational for a self-​interested agent to acquire the threat-​fulfilling disposition, but if (through some irrational quirk) a threatened target unexpectedly ignores the agent’s apocalyptic threat, it is surely not rational for the agent to follow through and blow herself up. Such disastrous stubbornness would seem, on the contrary, quite crazy. Gauthier is not wholly convinced by this counterexample to RT-​past, but suggests and endorses a weaker transmission principle, which we may formulate as follows for any disposition D and act A that is characteristic of D:16 Gauthier, “Rationality and the Rational Aim.” 16



Fittingness Objections to Consequentialism    101

(RT-​present)  If it was rational to acquire D and is rational to maintain it presently, then it is rational to perform A. This principle, if true, could potentially vindicate the rationality of the sophisticated consequentialist’s actions (and hence, arguably, the fittingness of her motivating desires). However, Parfit points out that even this weakened transmission principle is susceptible to counterexamples, such as: Schelling’s Case. A robber threatens that, unless I unlock my safe and give him all my money, he will start to kill my children. It would be irrational for me to ignore this robber’s threat. But even if I gave in to his threat, there is a risk that he will kill us all, to reduce his chance of being caught. . . . [I]‌t would be rational for me to take a drug that would make me [transparently] very irrational. The robber would then see that it was pointless to threaten me; and since he could not commit his crime, and I  would not be capable of calling the police, he would also be less likely to kill either me or my children. . . . But while I am in my drug-​induced state, and before the robber leaves, I act in damaging and self-​ defeating ways. I beat my children because I love them. I burn my manuscripts because I want to preserve them.17 Parfit stipulates that these destructive acts are not necessary to convince the robber that you are irrational. So they have no good effects, though they stem from a disposition (namely, the disposition to act irrationally) that it is worthwhile, for extrinsic reasons, to acquire and maintain. Are these acts rational? I share Parfit’s sense that they are not. So the transmission principles considered thus far fail, suggesting a robust disconnect between the rationality of acquiring and maintaining a disposition versus the rationality of acting upon it. The fundamental explanation for this disconnect is that an agent’s dispositions can have other consequences besides producing downstream acts in the agent herself. In particular, you might be harmed or rewarded directly on the basis of whether you possess some disposition, independently of whether you act on it. This suggests that we can distinguish (i)  dispositions that have high expected value, all things considered, and (ii) dispositions that have high expected value in respect of the downstream actions they’ll tend to produce. We can call the former class of dispositions “desirable” and the latter “well-​calibrated.” Dispositions that are desirable but not well-​calibrated we may call “extrinsically desirable.” It is these extrinsically desirable

Parfit, On What Matters, 437–​439. 17

102    Richard Yetter Chappell

dispositions that feature in Parfit’s cases of “rational irrationality,” i.e., whereby it is rational to acquire and maintain such a disposition, but irrational to act upon it.18 While acknowledging this possibility, we may still think that there must be some transmission principles according to which the rational status of a general rule or disposition can be inherited by the particular acts it prescribes. And, indeed, the distinction I’ve just highlighted suggests an obvious candidate principle: we just need to restrict the dispositions in question to those that are “well-​calibrated,” i.e., desirable for their (expected) impact on your downstream actions, rather than for other reasons. Consider the following transmission principle: (RT-​Calibrated)  For any dispositional set D and act A that is characteristic of D: If D is well-​calibrated, i.e., expectably good to possess in virtue of the downstream actions it tends to produce, then it is rational to perform A. This seems much more promising, though I discuss a residual concern in a note.19 In sum:  while I  am uncertain that any such transmission principle is ultimately vindicated, formulations that focus on the subset of dispositions that are well-​calibrated, in my described sense, would seem to have the best shot. And, as we will see, these are just the dispositions that may be possessed by the “subjective” act-​consequentialist agent, in contrast to the unfitting but (extrinsically) desirable dispositions that we saw could be part of the “sophisticated” consequentialist psychology. We are now in a position to spell out what I take the fitting act-​ consequentialist agent to look like. 5.5. The Act-​Consequentialist  Agent

Suppose we accept my earlier suspicion that “sophisticated” psychologies, with their extrinsically desirable dispositions, are not fitting consequentialist psychologies. The remaining option for defending consequentialism against fittingness objections is to spell out a nondefective “subjective” consequentialist psychology. In attempting this task, I will especially make use of the idea that our account of the fitting consequentialist agent, while restricted to consequentialist motivations, may at least appeal to 18 This parallels the familiar distinction between “object-​given” and “state-​given” reasons. The dispositional state is a useful state to be in, but the (metaphorical) “object” of the state—​the set of actions it disposes you toward—​does not merit such a disposition. 19 A  disposition might improve one’s actions overall while being predictably detrimental in some specific subdomain (e.g., a disposition that helps a previously hostile person become more attentive to the emotional well-​being of those she interacts with, but less likely to help the distant needy). In such a case, we presumably do not want to endorse the clearly detrimental (in)actions just because one’s other actions have improved.



Fittingness Objections to Consequentialism    103

well-​calibrated, if not merely extrinsically desirable, guiding dispositions. This restriction is one of the main features that sets apart my straightforward account of the fitting consequentialist psychology from the “sophisticated” view explored in section 5.3. 5.5.1. Motivating versus Guiding Dispositions

Let’s begin by distinguishing what I’ll call “guiding” and “motivating” dispositions.20 Our noninstrumental desires or motivations are our driving concerns, or what move us to action. They represent the goals we hope to realize through acting. On the other hand, this motivational “oomph” can be steered or guided by strategies and heuristic dispositions that shape our behavioral responses in pursuit of those goals. We may think of our guiding dispositions as, roughly, the psychological manifestation of instrumental rationality. They take our desires as inputs, and output a suitable action or intention.21 The standard caricature of the consequentialist agent assumes that we can “read off ” both kinds of dispositions from the moral theory. From its theory of the good—​ say, the utilitarian view that what matters is just the welfare of sentient beings—​we get the fitting utilitarian motivations. That much I agree with: the fitting utilitarian will desire the welfare of sentient beings.22 But the standard caricature also takes the “maximizing” aspect of consequentialism to settle the guiding dispositions of the fitting consequentialist agent: she will (allegedly) decide how to act by, in each instance, conducting an expected-​value calculation, and then perform whatever action she judges to have the highest expected value. It is this feature of the imagined consequentialist agent that is responsible for so much of her apparent defectiveness (as we will see in section 5.5.3). And it is this feature that I deny we should attribute to the fitting consequentialist agent. Can we coherently reject the critic’s assumption that the fitting consequentialist must have these defective guiding dispositions? I think we can. Note, first of all, that our primary example of a theory’s fittingness implications (namely, the goodness–​ fitting desire link) concerns motivating dispositions. There’s no such clear link between our moral theories and any putative implications for the fittingness of our guiding dispositions. So it seems prima facie open to us to dispute this assumption.

Thanks to Philip Pettit for his assistance in formulating this distinction. 21 I remain neutral on the question of whether practical reasoning is best understood as concluding in action or intention. 22 Though it’s an important question whether we interpret this as a single monolithic desire for aggregate welfare, or—​as I prefer—​a plurality of desires, one for each sentient being’s welfare. See Chappell, “Value Receptacles.” 20

104    Richard Yetter Chappell

Moreover, I think there are strong theoretical grounds for expecting the assumption to be false. Namely: (standard maximizing) consequentialism is naturally understood as a view that simply combines ordinary instrumental rationality with a specification of the moral ends to be pursued.23 This theoretical understanding strongly suggests that the fitting guiding dispositions should be determined by our independent account of instrumental rationality (as I will go on to explore). The distinctive positive work of consequentialism as a moral theory is just to add the theory of the good, with its associated implications for fitting motivating dispositions. The remaining distinctive content of consequentialism is negative: it simply denies that various other normative elements, e.g., deontological side-​constraints, play any fundamental role in determining what’s right (or what fitting moral reasoning is sensitive to). Finally, I hope to offer an “existence proof ” of the coherence of a fitting consequentialist psychology that lacks defective guiding dispositions. In particular, I’ll show that utilitarian motivating dispositions can be coherently combined with a plausible account of the instrumentally rational guiding dispositions. If there’s some reason the resulting agent fails to qualify as a “fitting utilitarian,” I think the onus is on the critic to explain why this is so. To fulfill this task, I begin with a brief sketch of some “well-​calibrated” guiding dispositions which I take to be (a) prerequisites for competent human agency, and hence (b) constitutive of instrumental rationality, at least for agents with human-​ size minds. I  will then show how an agent with fitting utilitarian motivations could also possess these well-​calibrated guiding dispositions. Since these guiding dispositions appear to be compatible with utilitarian (or other consequentialist) motivating dispositions, I conclude that there is no barrier to the consequentialist supplementing her theory in just this way to secure a plausible vision of the “fitting consequentialist” agent. I  will wrap up by illustrating how my well-​calibrated fitting consequentialist can be used to address prominent character-​based objections to consequentialism. 5.5.2. Defective Deliberation and the Well-​C alibrated Agent

Let’s consider four central features of the fitting agent’s guiding dispositions. Firstly—​as perhaps the most obvious prerequisite for competent agency—​we have epistemic rationality: that is, the agent must have well-​calibrated expectations about her environment, or a basic understanding of what counts as evidence for what. She cannot take the roar of a dangerous predator as evidence that a cute puppy awaits Scheffler, “Agent-​Centred Restrictions.” 23



Fittingness Objections to Consequentialism    105

her outside. She needs to have generally reasonable beliefs about her environment, and about what would be effective means for realizing her ends (whatever those might be). Next, at the borderline of the epistemic and the practical, we will find constraints on how the agent is disposed to allocate her limited attentional resources. She must be generally attentive to possible threats and opportunities in her immediate environment, while also—​in a calm moment, when appropriate—​considering more abstract mental models of past and possible future scenarios (for the sake of planning, self-​evaluation, etc.). The details aren’t too crucial for my purposes, but as we’ll see, it’s important that the fitting agent not dwell excessively on the past. Third, the competent agent requires well-​ calibrated habits, instincts, or subpersonal “predispositions”24—​an “auto-​pilot” set, e.g., to avoid pain, be cooperative, and help others in need—​to secure effective automatic behavior in normal circumstances. One reason for this is that in time-​critical situations, the agent cannot afford to pause to reflect on her situation at all. Often, a competent agent will be moved immediately (without conscious deliberation) to act, upon registering pertinent information about her environment. This is no mere behavioral reflex, as the agent is genuinely acting for reasons. But the rational processing goes on “below the surface.” Once equipped with such well-​calibrated predispositions, the fitting agent may act on them without need for excessive self-​monitoring or executive control, and—​ in so doing—​she may trust that she is acting for the best. Our fitting agent may, in this way, reap the practical benefits of “satisficing” without the theoretical baggage.25 The fourth and final component that I’ll discuss here is the possession of well-​ calibrated triggers for executive oversight. On pain of regress, we cannot always deliberate about whether to start deliberating. So, as previously noted, the agent’s default guidance must be from nondeliberative “predispositions.” But when these are not up to the task—​when, say, the agent is faced with novel or complex circumstances for which her predispositions aren’t so well calibrated to deal with—​the agent’s subpersonal mechanisms must recognize this and respond by triggering explicit deliberation on the part of the agent. In summary: the fitting human-​like agent—​if she is to be capable of acting competently in a wide range of “normal” circumstances—​will rely heavily on well-​calibrated predispositions, rather than explicit deliberation or calculation, to guide her actions

Pettit and Brennan, “Restrictive Consequentialism.” See also Kahneman, Thinking, Fast and Slow. 25 Cf. Slote and Pettit, “Satisficing Consequentialism.” Slote’s satisficing consequentialist merely aims to achieve “good enough” consequences, which makes sense as a practical strategy but seems rather more puzzling as an account of the rationally warranted ultimate aims. 24

106    Richard Yetter Chappell

in pursuit of whatever her goals may be. That’s just what it is for agents with human-​ size minds to be instrumentally rational or have fitting guiding dispositions. And this will be so even if her goal is to promote the well-​being of sentient creatures as much as she is able. This, I propose, is how we should understand the fitting consequentialist agent. She may have straightforwardly utilitarian (or whatever) desires, which are then translated into action via the above-​described “well-​calibrated” guiding dispositions. 5.5.3. Addressing the Objections

We are now in a position to assess how my conception of the fitting consequentialist agent stands up to various anticonsequentialist objections. We can first note that my “well-​calibrated” fitting consequentialist will not be “constantly calculating.” Absent any triggering of her executive faculty, the fitting utilitarian (for example) will respond directly to the salient needs of others—​a child drowning in a pond, say—​without mediation by explicit deliberation, let alone abstract judgments of “permissibility.” In this way, the fitting consequentialist will not exhibit what Williams famously called “one thought too many.”26 The fitting consequentialist’s reliance on generally reliable predispositions also undermines the objection that she would engage in “marginally-​beneficial rule-​ breaking,” such as breaking a promise whenever the benefits from doing so seem to even slightly outweigh the costs.27 Because overt calculation often goes awry, the competent consequentialist will—​ as we’ve seen—​rely heavily on her generally reliable predispositions in everyday life, only pausing to reflect when her well-​calibrated subpersonal mechanisms alert her to the need (say, due to complex novel circumstances that her “auto-​pilot” wasn’t designed to deal with). Everyday promise-​keeping is not exactly novel, so for the fitting agent the question whether to keep a promise shouldn’t even arise, unless there’s something special about the situation that calls for her executive oversight. That’s enough to defeat the claim that the fitting consequentialist would commonly engage in marginally beneficial rule-​breaking. But we may draw an even stronger conclusion. For suppose that our agent’s executive oversight happens to be triggered. In a typical case, what should she conclude? We can stipulate that in fact the outcome would be marginally better if she broke her promise, but presumably the

Williams, “Persons, Character and Morality,” 18. 27 Hooker, Ideal Code, Real World, raises the objection in terms of objective rightness. I don’t have particularly strong intuitions about objective rightness, as opposed to what a morally conscientious agent would do, so I restate the objection here in the latter terms. 26



Fittingness Objections to Consequentialism    107

agent herself will not have any easy way of knowing this. (Among other things, she’d need to first consider the possibility of self-​serving bias corrupting her judgment, and also to weigh the apparent benefits of rule-​breaking in this instance against the long-​run value of retaining a reputation for trustworthiness.) Maybe if she heard the booming voice of God reassuring her of this fact, then she could rationally go ahead and break her promise without further worry. Such behavior no longer seems intuitively troubling to me. But in ordinary circumstances, when rule-​breaking might seem more worrying, it’s almost never going to be clear that rule-​breaking is beneficial unless it is significantly (not merely marginally) so.28 So our agent is faced with an immediate choice: she can (i) break the rule even though it’s not yet clear to her whether this would have good results on net; (ii) sink further cognitive resources into investigating a question that she probably shouldn’t have bothered to ask in the first place; or (iii) simply keep her promise and turn her attention to more important matters. It seems pretty clear that, in this sort of case, option (iii) is the way to go.29 In sum: breaking a rule will generally only be obviously worthwhile in cases where it is also of significant benefit (in which case many would approve of rule-​breaking anyway). If it’s only of marginal benefit, this fact typically won’t be sufficiently clear for a reasonably self-​doubting, fallible agent to immediately act upon. And the low potential payoff means that it isn’t really worth inquiring further: better just to stick with the generally reliable rule of thumb. So a fitting consequentialist generally won’t be found engaging in marginally beneficial rule-​breaking after all. (She’d even share our intuition that there’s something awfully dubious about any agent who would act that way.) This gives her the kind of stable predictability needed for others to regard her as an eligible and (more or less) trustworthy partner for social cooperation. This discussion brings out the fact that the standard caricature of a consequentialist agent assumes that she will be unreasonably overconfident in her ability to calculate expected values accurately. But even if a consequentialist initially judges (just based on the first-​order evidence) that she would do best to break some generally beneficial rule, she may also realize that most people who make such judgments in similar situations are mistaken. Since she has no particular reason to think that she is one of the lucky few who makes this judgment correctly, the general fact serves as a kind of higher-​order evidence that her initial judgment was mistaken. All things Moore, Principia Ethica, claimed that agents will never be in an epistemic position to justifiably break such rules, but we needn’t be quite so pessimistic. 29 In special circumstances, option (ii) may be truly costless, and so there’s a possibility that the agent could reasonably undertake such an investigation, and responsibly reach the true conclusion that breaking the rule really is justified in this case. But this won’t be typical, and the crucial point for my purposes is just that one’s prima facie utility judgments won’t provide sufficient justification for breaking generally reliable rules. 28

108    Richard Yetter Chappell

considered, then, a reasonable expected value judgment should, in this sort of circumstance, end up reinforcing the general rule rather than licensing typically misguided unilateral rule-​breaking.30 The objections considered thus far—​that the consequentialist would have “one thought too many” and that she would engage in “marginally-​beneficial rule-​ breaking”—​suggest the need to distinguish (i) the appropriate answer to a question, and (ii) whether a well-​functioning agent would ask that question in the first place. The need for this distinction becomes especially apparent when we consider the following objection from Michael Stocker: Maximizers hold that the absence of any attainable good is, as such, bad, and that a life that lacks such a good is therefore lacking. I disagree. One central reason for my disagreement stems from the moral psychological import of regretting the absence or lack of any and every attainable good. This regret is a central characterizing feature of narcissistic, grandiose, and other defective selves. It is also characteristic of those who are too hard on themselves, who are too driven and too perfectionistic.31 This objection strikes me as deeply misguided. It may be unfortunate, and indeed even inappropriate (“defective”), to actively regret every little regrettable thing. But those things may be regrettable all the same. Crucially, this is not to say that a rational agent must regret them. It is more like a hypothetical imperative: if you closely attend to the features in question, this should induce feelings of regret. But it may be a kind of rational defect to attend to the wrong things, if there are more pressing matters to attend to. As we saw in section 5.5.2, the fitting agent would allocate her attentional resources in a way that avoids excessive dwelling on hypotheticals. So we can agree with Stocker that the agents he describes are defective, without thinking that the maximizing consequentialist would exhibit any such trait. On my picture, the consequentialist will have only a conditional disposition to regret the lack of a good insofar as she attends to this lack. But she’ll usually have more important things to attend to, so she shouldn’t actually end up actively regretting things very often

This observation may also aid the consequentialist in addressing alleged “counterexamples” to consequentialism, involving organ harvesting, pushing people in front of trolleys, etc. Insofar as we can accommodate the intuition that there’s something wrong with any person who would act this way (in any remotely realistic circumstances such behavior would be far too reckless for a morally conscientious agent aware of her own fallibility to rationally perform), the consequentialist may doubt whether we have such a clear grasp of any further intuitions about the “objective wrongness” of the irrationally reckless act. (Consequentialists certainly won’t feel any inclination to regret the occurrence of the utility-​maximizing act, for example.) 31 Stocker, Plural and Conflicting Values, 321. 30



Fittingness Objections to Consequentialism    109

at all. She is, in this sense, appropriately responsive to reasons for regret, without having to be constantly responding to them. I’ve now shown how the well-​calibrated fitting consequentialist avoids three of the “character-​based objections” extant in the literature. Equipping the consequentialist agent with well-​calibrated guiding dispositions helps to undermine claims that the fitting consequentialist psychology is inherently defective. 5.5.4. Act-​ versus Rule-​C onsequentialist  Agents

In light of my appeal to rules and dispositions, some readers may be puzzled by my labeling the resulting agent a fitting “act-​consequentialist” agent. To avoid any confusion on this front, let me wrap up by briefly characterizing what I take to be the two main differences between (fitting) act-​and rule-​consequentialist psychologies. First, while both make use of rules, they do so in very different ways. The act consequentialist adopts “rules of thumb” for instrumental purposes, but her fundamental aim (reflected in my account of the fitting motivating dispositions) makes no essential reference to rules: she just wants to bring about the best possible outcome, and refraining from deliberation is one (guiding) strategy she might employ, at appropriate times, as a means to this end. Rule-​consequentialism, by contrast, builds reference to rules into its criterion of right action, and hence the corresponding “fitting psychology” must likewise accord some fundamental, noninstrumental significance to rules—​e.g., in the agent’s fundamental desires or motivating dispositions. (This then opens her up to distinctively characterological objections of “rule-​worship.”) A second, more straightforward difference is that she may employ rules with very different contents. I’ve suggested that a fitting act-​consequentialist could (while retaining her fitting character) only make use of “well-​calibrated” dispositions—​ dispositions whose value stems from the improved quality of the actions they dispose the agent toward. But insofar as rule-​consequentialism appeals to rules that it would be good to internalize for whatever reason, she may well end up calling “fitting” even dispositions that are merely extrinsically desirable. In other words, the fitting rule-​consequentialist agent would look much more like the kind of “sophisticated” agent described in section 5.3. For example, suppose that suffering is always bad, but that widespread adoption of retributive attitudes toward punishment would form part of the optimal moral “code” in a certain society. Since rule-​consequentialism assigns direct, noninstrumental significance to the rules of the optimal code, the fitting rule-​consequentialist agent would presumably have to have corresponding noninstrumental desires. That is, rule-​consequentialism, in the imagined circumstances, has the implication that retributive punishment is fitting to desire. But we’ve supposed that in fact it isn’t

110    Richard Yetter Chappell

desirable: suffering is always bad. (The fitting act-​consequentialist, by contrast, may only desire to promote the good. If better results would be obtained with different attitudes, then this may lead her to try to transform her character so that she no longer qualifies as a fitting act-​consequentialist at all. But while her view might thus recommend retributive or other attitudes, it maintains the distinction between the practically recommended attitudes and the ones that are fitting to their objects as a matter of principle.) This then provides the basis for a simple new argument against rule-​ consequentialism. Rule-​consequentialism implies that literally anything (from retributive punishment to gratuitous torture) could be rendered fitting to desire, just by tweaking the incentives surrounding the creation and maintenance of the public’s moral code. But these things are not, in such circumstances, fitting to desire. Such desires are supported by the “wrong kind of reasons”—​it is not the objects of these desires that are desirable, but rather the state of possessing the desire itself (or something to do with promulgating the moral rule, depending on what exactly is responsible for generating the good consequences). Rule-​consequentialism thus has false implications about what’s fitting, and is thereby shown to be false itself.32 References Anscombe, G. E. M. “Modern Moral Philosophy.” Philosophy 33 (1958): 1–​19. Bales, R. E. “Act-​Utilitarianism: Account of Right-​Making Characteristics or Decision-​Making Procedures?” American Philosophical Quarterly 8 (1971): 257–​265. Chappell, Richard Yetter. “Fittingness: The Sole Normative Primitive.” Philosophical Quarterly 62 (2012): 684–​704. Chappell, Richard Yetter. “Value Receptacles.” Noûs 49 (2015): 322–​332. Gauthier, David. “Rationality and the Rational Aim.” In Reading Parfit, edited by Jonathan Dancy, 24–​41. Oxford: Blackwell, 1997. Hooker, Brad. Ideal Code, Real World: A Rule-​Consequentialist Theory of Morality. Oxford: Oxford University Press, 2000. Kahneman, Daniel. Thinking, Fast and Slow. New York: Farrar, Straus and Giroux, 2011. Kapur, Neera Badhwar. “Why It Is Wrong to Be Always Guided by the Best: Consequentialism and Friendship.” Ethics 101 (1991): 483–​504. Mackie, John Leslie. “Rights, Utility, and Universalization.” In Utility and Rights, edited by R. G. Frey, 86–​105. Oxford: Basil Blackwell, 1985.

Thanks to Eden Lin, Errol Lord, Sarah McGrath, Tim Mulgan, Martin Peterson, Philip Pettit, Doug Portmore, Christian Seidel, Derek Shiller, Peter Singer, Michael Smith, Helen Yetter-​Chappell, the New York Corridor reading group, the University of York’s Mind and Reason group, and the University of Duisburg-Essen’s Meta Essen workshop on fittingness for helpful comments and discussion. 32



Fittingness Objections to Consequentialism    111

Mason, Elinor. “Do Consequentialists Have One Thought Too Many?” Ethical Theory and Moral Practice 2 (1999): 243–​261. Moore, G. E. Principia Ethica. Cambridge, UK: Cambridge University Press, 1903. Parfit, Derek. Reasons and Persons. Oxford: Clarendon Press, 1987. Parfit, Derek. On What Matters, vol. 1. Oxford: Oxford University Press, 2011. Pettit, Philip, and Geoffrey Brennan. “Restrictive Consequentialism.” Australasian Journal of Philosophy 64 (1986): 438–​455. Rabinowicz, Wlodek, and Toni Rønnow-​Rasmussen. “The Strike of the Demon: On Fitting Pro-​ Attitudes and Value.” Ethics 114 (2004): 391–​423. Railton, Peter. “Alienation, Consequentialism, and the Demands of Morality.” Philosophy and Public Affairs 13 (1984): 134–​171. Regan, Tom. The Case for Animal Rights. Berkeley: University of California Press, 1994. Scheffler, Samuel. “Agent-​ Centred Restrictions, Rationality, and the Virtues.” Mind 94 (1985): 409–​419. Slote, Michael, and Philip Pettit. “Satisficing Consequentialism.” Proceedings of the Aristotelian Society, Supplementary Volumes 58 (1984): 139–​176. Stocker, Michael. Plural and Conflicting Values. Oxford: Oxford University Press, 1989. Williams, Bernard. “A Critique of Utilitarianism.” In Utilitarianism: For and Against, by J. J. C. Smart and Bernard Williams, 75–​150. Cambridge, UK: Cambridge University Press, 1973. Williams, Bernard. “Persons, Character and Morality.” In Moral Luck: Philosophical Papers, 1973–​ 1980, 1–​19. Cambridge, UK: Cambridge University Press, 1981.

II

Problems SCRUTINIZING THE NEW WAVE’S THEORETICAL BASIS

6 T H E G O O D O F CONSE QUE NTIA LIZ E D D EO N TO LO GY

Monika Betzler and Jörg Schroth

6.1. The New Promise: The Compelling Idea without Counterintuitive Implications

Consequentialism is one of the most prominent normative ethical theories, and both defenses and objections to it are legion. The strongest argument in its favor is its so-​ called compelling idea (CI), which Samuel Scheffler captures as follows: “One thing [consequentialist theories] all share . . . is a very simple and seductive idea: namely, that so far as morality is concerned, what people ought to do is to minimize evil and maximize good, to try, in other words, to make the world as good a place as possible. On the face of it, this idea, which lies at the heart of consequentialism, seems hard to resist.”1 Even opponents of consequentialism, such as Philippa Foot, admit that this idea is particularly persuasive: “What is it, let us now ask, that is so compelling about consequentialism? It is, I think, the rather simple thought that it can never be right to prefer a worse state of affairs to a better. It is this thought that haunts us.”2 Accepting that it can never be right to prefer a worse state of affairs to a better one implies a moral requirement to produce the best state of affairs.3 1 Samuel Scheffler, Introduction to Consequentialism and Its Critics, 1. 2 Foot, “Utilitarianism and the Virtues,” 62. 3 Douglas Portmore appears to find it even more compelling if—​allowing for a satisficing consequentialism—​ CI is formulated so as to commit us not to the requirement but to the permission to produce the best state of

115

116    Monika Betzler and Jörg Schroth

Yet, although consequentialism is based on an intuitively compelling idea, it also seems to have some highly counterintuitive implications. Most strikingly, (i) consequentialism allows and even prescribes the most horrendous actions even if they have only slightly better consequences than alternative actions, and (ii) it allows and even demands the most severe sacrifices even if they yield only slightly better consequences than alternative actions.4 While to many philosophers these implications are reason enough to discard consequentialism altogether, others have tried to avoid them by refining and revising standard consequentialism into indirect versions of consequentialism that distinguish between a criterion of rightness and a decision procedure. Along these lines it has been argued that, with this distinction in place, we will quickly see that horrendous acts and severe sacrifices will rarely if ever be required, and they will certainly not be required in order to obtain only slightly better consequences. But this strategy does not dismiss the counterintuitive implications, because even if the decision procedure never demands that such actions be performed, they are still morally right according to the criterion of rightness. At this point in the debate it seems that there is nothing left but to decide whether or not we mind that consequentialism in theory has morally counterintuitive implications if in practice no one is ever required to act accordingly, or whether we find the mere theoretical possibility of horrendous acts being right a sufficient reason for rejecting consequentialism. But safeguarding our moral intuitions by discarding consequentialism in favor of deontology5 has its price too. Deontology is often considered an unattractive moral theory because it comprises a heap of unconnected moral intuitions without an underlying rationale and with­ out a method for justifying or deciding between conflicting intuitions.6 What is

affairs: it can never be wrong to prefer a better state of affairs to a worse one. For the arguments presented in this chapter it is irrelevant which version of CI one adopts. For reasons of simplicity we will discuss the maximizing variant, according to which it can never be right to prefer a worse state of affairs to a better one. See Portmore, “Can an Act-​Consequentialist Theory Be Agent-​Relative?,” 377n44; Portmore, “Combining Teleological Ethics,” 110n15. 4 Cf. Hooker, Ideal Code, Real World, 145f.; Portmore, Commonsense Consequentialism, 3f. 5 In what follows we restrict ourselves to deontology as a particularly prominent nonconsequentialist theory. 6 While this criticism might be thought to be primarily called for by Ross’s intuitionistic variant of deontology, it proves to be equally called for in other variants of deontology. In the case of contractualism, for example, what underlies the idea that one must never do what others can reasonably reject just is a heap of intuitions, such as that one must never use others merely as a means, that one must never harm the innocent, and so on. Similarly, Kantian deontology seems to presuppose intuitions, such as intuitions underlying the golden rule, even if it is a theory that provides many more resources for an underlying rationale and a method for justifying these intuitions. This might be contested, to be sure. Our aim is not to criticize intuitionistic variants of deontology in particular, but to show that they all share an equally problematic feature. Many thanks to Paul Hurley for pressing us to expand on this.



The Good of Consequentialized Deontology    117

more, these intuitions seem to presuppose the priority of the right over the good,7 which for consequentialists means missing the point of morality, namely to further the good: “The idea that the unifying and justifying function of all of our ethical categories is ultimately to make our lives go better, or to make the world a better place, is one that I find utterly compelling. If that is not the point of the whole business of moral thinking, then I find it difficult to imagine what the point might be. What else could morality be for?”8 And to make things worse, if not to say unacceptable, deontological moral intuitions seem to lead to some kind of paradox: many of these intuitions take the form of agent-​relative constraints such as, e.g., “Act so as to ensure that you yourself kill no one.” Such a constraint must not be violated even if a single violation of it would minimize the total number of violations of the same constraint. But it seems paradoxical to claim that an action is morally objectionable and yet allow that many of these objectionable actions might be performed, although one could easily prevent this by performing only one such morally objectionable action oneself.9 Although agent-​relative constraints are charged with being paradoxical, they are at the heart of deontology. They guarantee what is intuitively plausible about deontology by forbidding performing horrendous acts in order to maximize the good. It is precisely standard consequentialism’s agent-​neutral requirements (“Act so as to ensure that as few killings as possible occur”) that account for consequentialism’s counterintuitive implications, because they allow and even require one to perform horrendous acts if they are the only means to maximize the good and make the world a better place. Since agent-​relative constraints sometimes forbid the maximization of the good, they are incompatible with the compelling idea of consequentialism. This leaves us, it seems, with two equally unattractive options:  a theory with a compelling idea but counterintuitive implications, or a theory that fits with our moral intuitions but cannot accommodate consequentialism’s compelling idea and has paradoxical implications. In the past few years a sophisticated new solution to this impasse has emerged. This solution consequentializes deontology, that is, it devises a new moral theory that has a consequentialist structure and yet manages to accommodate deontology’s agent-​relative constraints.10 Due to its consequentialist 7 To be sure, not all deontological theories adhere to this priority claim. See, e.g., Herman, “Leaving Deontology Behind.” But we believe that it is still at the core of most deontological theories. 8 Sumner, “Two Theories of the Good,” 1f. 9 For a first formulation of the paradox see Nozick, Anarchy, State and Utopia, 30. See also Scheffler, The Rejection of Consequentialism, 133f. 10 See Dreier, “Structure of Normative Theories”; Dreier, “In Defense of Consequentializing”; Louise, “Relativity of Value”; Smith, “Neutral and Relative Value after Moore”; Smith, “Two Kinds of Consequentialism”; Portmore, “Combining Teleological Ethics”; Portmore, “Consequentializing Moral Theories”; Portmore, “Consequentializing”; Portmore, Commonsense Consequentialism. Some earlier attempts at consequentializing

118    Monika Betzler and Jörg Schroth

structure, the new theory claims to retain the compelling idea of consequentialism and pretends to avoid the paradox of deontology while at the same time to stand in perfect accord with our moral intuitions. But is this too good to be true? This chapter attempts to answer this question. After having laid out the motivation behind the consequentializing project,11 we will, in section 6.2, offer a diagnosis of what this new direction of consequentialism precisely entails. After these preliminaries, in sections 6.3 to 6.5, we will present three main objections to this new direction of consequentialism. Our aim is to evaluate whether it is able to meet its own declared goals, namely to cater to our deontological intuitions while providing a consequentialist rationale, to retain the compelling idea and to avoid the paradox of deontology. We will defend the view that it fails in all three respects. 6.2. Consequentialized Deontology

Let us begin with our analysis of the consequentializing project. How does the resulting new moral theory—​ which we refer to as “consequentialized deontology”12—​differ from standard consequentialism? Which conditions does this new theory have to meet to ultimately prove successful? Recall that consequentialized deontology has to combine the following three criteria in one moral theory: (i) to retain standard consequentialism’s compelling idea, (ii) to avoid the paradox of deontology, and (iii) to avoid counterintuitive implications while retaining all plausible deontological intuitions. Defenders of consequentialized deontology suggest a simple explanation and an equally simple solution for the alleged incompatibility between the compelling idea and deontological constraints: it results from the tacit assumption that both belong to different and incompatible theories of the right. Standard consequentialists take the right to consist in maximizing the (agent-​neutral) good, whereas deontologists

are Broome, Weighing Goods; Sen, “Evaluator Relativity and Consequential Evaluation”; Kordig, “Structural Similarities.” 11 Friends of consequentializing diverge in their theoretical aims. Portmore, for example, starts out defending an intuitively plausible consequentialism (cf. “Can an Act-​Consequentialist Theory Be Agent-​Relative?”; “Combining Teleological Ethics”). Dreier and Louise, by contrast, are more interested in carving out the border between deontology and consequentialism and relocating it. But they all share the assumption that they can build deontological intuitions into a new moral theory with a consequentialist structure. 12 As will become clear, there are different versions of consequentialized deontology, just as there are different versions of consequentialism and deontology. Strictly speaking, these terms refer to classes of theories—​ consequentialist theories, deontological theories, and consequentialized deontological theories—​rather than to a particular theory.



The Good of Consequentialized Deontology    119

take it to consist in following deontological constraints, which, sometimes at least, regard the maximization of an agent-​neutral good as wrong. With this assumption, the compelling idea and deontological constraints are indeed irreconcilable. What seems to have been overlooked, according to the defenders of consequentialized deontology, is that a consequentialist theory of the right (that is, the demand to maximize) can be linked not only to an agent-​neutral theory of the good but also to an agent-​relative theory of the good, and that agent-​relative deontological constraints can be reinterpreted as agent-​relative values, which, as values, no longer belong to a theory of the right but to a theory of the good. If it proves possible to reconceptualize agent-​relative deontological constraints as agent-​relative values,13 the incompatibility between different theories of the right vanishes: the compelling idea can be retained as part of a consequentialist theory of the right, while deontological intuitions become part of an agent-​relative theory of the good. In a further step, the agent-​relative values can be ranked in such a way that maximizing them yields the same verdict as standard deontological constraints. For example, the best outcome should be identified as the one in which the agent-​relative value “that I do not kill” is realized. According to the “recipe of consequentializing,” this is achieved in the following way.14 Take whatever considerations deontologists consider relevant for determining the deontic status of an action and regard those considerations as relevant for determining the ranking of outcomes, such that the following holds true for any action: if a deontological theory permits an action, deny that there is any alternative action with better consequences; if it forbids an action, maintain that there is an alternative action with better consequences; if it demands or requires an action, maintain that the consequences of that action are better than the consequences of any possible alternative action. To illustrate the consequentializing procedure, let us consider an example. Take a standard deontological constraint that forbids us from killing a person so as to prevent two other people from being killed. Accordingly, it is wrong for Cora to kill Carl to prevent Dora’s killing Daniel and David. In a first step, we translate this deontological constraint into a comparative agent-​relative value judgment: the outcome in which Cora kills Carl to prevent Dora from killing Daniel and David is worse from Cora’s standpoint15 than the outcome in which she does not kill Carl See Jan Gertken’s chapter in this volume. Gertken is skeptical about whether such a reconceptualization is possible. 14 See Dreier, “Structure of Normative Theories,” 23; Portmore, “Consequentializing Moral Theories,” 39; Portmore, “Consequentializing,” 330. See also Dreier, “In Defense of Consequentializing,” 98f. 15 Note that “worse from Cora’s standpoint” must not be taken to mean “worse for Cora,” but rather something like “worse relative-​to-​Cora” or “worse relative-​to-​Cora’s-​position.” We will not discuss here what exactly 13

120    Monika Betzler and Jörg Schroth

(but in which Dora kills Daniel and David). In a second step, we apply the consequentialist criterion of rightness, according to which an action is right if and only if it leads to the best outcome from Cora’s standpoint. By applying this criterion, it follows that Cora must not kill Carl. Thus, to construct an intuitively plausible consequentialism in the guise of consequentialized deontology we merely have to reinterpret the deontological intuitions—​which within standard deontology belong to a theory of the right—​as agent-​relative values. Then the agent-​neutral theory of the good of standard consequentialism has to be replaced by an agent-​relative theory of the good and a suitable ranking of outcomes. We now have a consequentialist moral theory with the same moral verdicts as those of a plausible deontological moral theory. Such a “deontic equivalence” between consequentialized deontology and standard deontology is, according to Portmore, possible for any plausible deontological theory.16 This new consequentialist counterpart theory—​“consequentialized deontology,” in our terminology—​is thought to correspond to our commonsense intuitions and avoids the paradox of deontology because the agent-​relative values (which correspond to, because they are derived from, our commonsense deontic intuitions) now seem conducive to maximization. Consequentialized deontology thereby claims to retain the compelling idea, according to which it can never be right to prefer a worse outcome to a better one. The better outcome, however, is to be interpreted in an agent-​relative fashion.17 All that deontologists cherish can now be formulated as a consequentialist demand or constraint, or so it seems. Consequentialized deontology keeps only two central features of standard consequentialism:  first, the rightness of acts is determined by their (constitutive and

this is supposed to mean. A satisfying answer to this question is, of course, of the utmost importance for the consequentializing project. 16 Cf. Portmore, “Consequentializing Moral Theories,” 40. Cf. Dreier, “Structure of Normative Theories”; Dreier, “In Defense of Consequentializing,” 104f. 17 The consequentializing project must not be confused with another way of putting morality under the consequentialist umbrella: consequentialism can also expand its agent-​neutral theory of the good and incorporate all the values that deontologists cherish. For example, consequentialists can build dignity as a value into their theory of the good and then demand the minimization of violations of dignity. This demand implies that a person has to disregard the dignity of a particular person if this is the only possibility for preventing more violations of dignity from happening. This is what has been called “the consequentialist vacuum cleaner” (McNaughton and Rawling, “Agent-​Relativity and the Doing-​Happening Distinction,” 168f; McNaughton and Rawling, “On Defending Deontology,” 41). Since deontologists reject such minimizing violations (and therefore, according to consequentialists, run into the paradox of deontology), the consequentialist vacuum cleaner does not imply deontic equivalence. Standard consequentialists could import deontological values, but they still come to moral verdicts that substantially differ from those of deontology.



The Good of Consequentialized Deontology    121

causal) consequences only; second, better outcomes are to be preferred to less good outcomes (the maximization principle).18 It parts company, however, with other familiar features of standard consequentialism that have typically been regarded as very attractive: it does not define the good independently of the right and therefore does not justify morality with an empirically accessible nonmoral good.19 According to standard consequentialism, in applying the criterion of rightness one only has to answer the (largely) empirical question of what the (expected) outcome of each available act is. One does not, in order to determine the right action, need to invoke various (often contestable) deontic intuitions that then (again intuitively) have to be weighed. As a result, it gives up the priority of the good over the right.20 Therefore, consequentialized deontology cannot score against deontologists with the claim that they rely on one intuition only for justifying the consequentialist criterion of rightness (namely the intuition that the right consists in the maximization of the nonmoral good), whereas deontologists must invoke many moral intuitions in their criterion of rightness. Consequentializers invoke the same moral intuitions as deontologists. Finally, consequentializers have to abandon the idea that morality gives all agents a common goal,21 namely to promote the agent-​neutral good and to make the world a better place in that sense. This is to be distinguished from the idea that every person has the same goal, i.e., to pursue a standpoint-​relative good. This is only a formal sameness, while the idea that all persons pursue a common goal is a substantive idea: it consists of what is good agent-​neutrally conceived. In order to make room for deontological intuitions, consequentializers sacrifice the parsimonious assumptions of standard consequentialism:  one intuition, one common goal for everyone, straightforward application. One might think that these implications already lessen the attractiveness of consequentialized deontology. But they might be put up with if, in return, the project were successful in achieving its declared aim to devise a moral theory that retains the compelling idea and yet has no counterintuitive implications. To achieve this aim, however, at least the following conditions have to be met:

Cf., e.g., Dreier, “In Defense of Consequentializing”; Portmore, Commonsense Consequentialism, 5. 19 To be sure, consequentialized deontology is not the only, though it is the most recent, variant of consequentialism that abandons appeal to a nonmoral good. See, e.g., Nozick, Anarchy, State and Utopia; Scheffler, The Rejection of Consequentialism; Kagan, The Limits of Morality. But to the extent that this leads to a more problematic theory, our criticism is just broader in scope. 20 Cf. Portmore, “Consequentializing Moral Theories,” 63. Cf. also Dreier, “In Defense of Consequentializing,” 115f., who considers the priority of the right or the good unimportant; and Brown, “Consequentialize This,” 753, who speaks of a co-​instantiation of the right and the good without committing himself to any priority. 21 See Parfit, Reasons and Persons, 27. 18

122    Monika Betzler and Jörg Schroth





(i) Deontological intuitions have to allow for an interpretation in terms of consequences that are interpreted in an agent-​relative fashion. (ii) Consequences that are interpreted in an agent-​relative fashion need to be conducive to maximization, so that paradoxical verdicts are precluded. (iii) Consequentialized deontology must therefore provide a justified (non–​ad hoc) ranking of those consequences such that applying the maximization principle on that ranking in principle yields the same moral verdict as the original (i.e., not yet consequentialized) deontological moral theory, yet provides informative and justified guidance. (iv) The intuitive plausibility of the compelling idea of standard consequentialism, i.e., that it can never be right to prefer a worse outcome to a better one, has to be retained by consequentialized deontology.

We follow defenders of consequentialized deontology in assuming that deontological intuitions allow for an interpretation in terms of consequences. Our strategy is to investigate whether the resulting theory is able to achieve its declared aims. As it turns out, however, consequentialized deontology is unable to meet the conditions necessary to do so. By introducing an agent-​relative ranking of consequences, the intuitive plausibility of the compelling idea cannot be sustained. Even further modifications do not help to bolster CI. In addition, a justified ranking of consequences that provides more moral orientation than deontology, yet gives the same moral verdicts, cannot be provided. This is not only because an agent-​relative theory of the good is not in sight (as Portmore readily concedes22), but also because further attempts to provide a justified ranking prove to be unsuccessful.23 Finally, consequentialized deontology does not make the paradox of deontology go away. 6.3. The Costs of Consequentializing: Lack of Moral Orientation

To consequentialize a given deontological theory we must apply the recipe for consequentializing and translate its deontic judgments (about permitted, required, and forbidden actions) into judgments of agent-​relative value.24 But how is this Cf. Portmore, Commonsense Consequentialism, 11. 23 We develop some of these criticisms in Betzler and Schroth, “Konsequentialisierung.” In what follows we build on and substantially expand on these criticisms. 24 We grant this possibility for the sake of argument. Stephen F.  Emet doubts that our intuitions regarding constraints can be retained by a translation into agent-​relative values. It rather seems “self-​centered” to judge a state of affairs as worse only because we ourselves caused it (“Agent-​Relative Restrictions,” 5ff.). There are other objections to the translation of deontological constraints into agent-​relative values that we cannot discuss in this chapter. 22



The Good of Consequentialized Deontology    123

ranking of agent-​relative values to be achieved and justified? Three procedures for ranking consequences have been proposed.25 The first procedure begins with a nonmoral theory of the good (hedonism, for example) or, respectively, with a definition of the good that is independent of the right and proceeds from there, without presupposing any deontic intuitions, i.e., judgments about rightness of acts, to a ranking of consequences. Since this ranking of consequences is independent of deontic intuitions, it only accidentally grounds moral verdicts that suit these intuitions. It is therefore no help to the consequentializer who needs a procedure that guarantees a perfect fit with our deontic intuitions.26 If such a fit is to be provided, we seem to be forced to rely on our deontic intuitions in coming up with a ranking. In that case, the ranking is no longer independent of our judgments about the rightness of acts, but explicitly presupposes such judgments and therefore does not give the good priority over the right. One such procedure that (trivially) guarantees a perfect fit with our deontic intuitions is the following: Independently of our judgments about the ranking of outcomes, we start with our deontic intuitions. We then determine that an outcome ranks higher than an alternative if the corresponding act is deontically better than that corresponding to the alternative outcome. For example, if we think that torture is forbidden, a world in which I do not torture someone (to save lives) is better than one in which I  torture. The ranking of outcomes is thus fully determined by our deontic intuitions.27 This second procedure seems to be the ideal method for consequentializing deontology: it exactly maps our deontic intuitions onto a consequentialist theory. But if consequentializers accept this procedure, they would merely resort to deontological intuitions and translate deontology into consequentialized deontology. This would add neither any further substantial moral content nor a consequentialist justification of our deontological intuitions; the moral content and justification would have simply been brought over from the deontological theory. That would amount to a deontological justification of the agent-​ relative ranking rather than to a consequentialist justification of our deontological intuitions. The second procedure, then, is the ideal method for consequentializing deontology only if consequentializers confine themselves to translating any

See Portmore, “Consequentializing Moral Theories,” 62ff., and Commonsense Consequentialism, 112ff., who explicitly discusses these three procedures. 26 This first procedure is that of standard utilitarianism and is a clear case of the priority of the good over the right. Portmore first referred to this as the “fundamentalist procedure” (“Consequentializing Moral Theories,” 62) and later as the “Anti-​Footean procedure” (Commonsense Consequentialism, 112). 27 This second procedure is the opposite of the first (utilitarian) procedure and gives the right priority over the good. Portmore refers to it as the “Footean procedure” (“Consequentializing Moral Theories,” 62; Commonsense Consequentialism, 112). 25

124    Monika Betzler and Jörg Schroth

plausible deontological theory into a consequentialized deontological theory. But, as Portmore insists in his more recent work,28 consequentializing should go beyond mere translation. Otherwise it will not go beyond the original, largely unjustified deontological theory,29 and the moral advice of consequentialized deontology will not go beyond our pretheoretical moral judgments (which are the input to the procedure); more importantly, it will not provide moral orientation in cases in which we disagree or lack confident moral intuitions.30 Note that Portmore, in explicitly referring to moral orientation, adds a new goal to the consequentializing project that goes beyond merely devising a consequentialist moral theory that avoids counterintuitive moral implications. Since the second procedure cannot provide moral orientation, a third, coherentist procedure is suggested31 in which pretheoretical deontic or evaluative judgments are not regarded as fixed points that determine the ranking of outcomes. Rather, we may revise our pretheoretical deontic judgments as well as the evaluation of outcomes until we reach a reflective equilibrium in the ranking of outcomes that corresponds to our moral intuitions.32 The ranking of outcomes is then not fully but—​in order to guarantee the intuitive plausibility of the resulting theory—​to a large extent determined by our deontic intuitions.33 In this way our deontic intuitions can be harmonized with consequentialist considerations, even about agent-​neutral assessments of outcomes in cases where our pretheoretical deontic intuitions are incomplete. Does this procedure provide more moral orientation than the second procedure or standard (i.e., the original, not-​yet-​consequentialized) deontology? To answer this question, let us present three outcomes and consider how a deontological theory would rank them: See Portmore, Commonsense Consequentialism, 113n74. 29 The reproach of being unjustified might hold far less for Kantian theories. But to the extent that even Kantian moral verdicts depend on intuitions, their justification might not be as grounded as Kant purports his theory to be. 30 This objection implies that the deontological theory cannot provide any moral orientation either. This claim, however, presupposes that the deontological theory consists of an unsystematic heap of moral judgments that are not embedded in a theoretical structure with general moral principles. Portmore, however, does not argue for this claim. 31 See Portmore, “Consequentializing Moral Theories,” 63f., and Commonsense Consequentialism, 113ff. 32 On the pages following the introduction of the coherentist procedure, Portmore provides no example applying the coherentist procedure to a case that involves the revision of our deontic or evaluative judgments. He presents only examples where our ranking of outcomes corresponds to our moral intuitions about the permissibility of actions (cf. Commonsense Consequentialism, 114f.). 33 If one rejects the fundamental procedure and instead applies the Footean or the coherentist procedure to arrive at a ranking, one must rely on moral intuitions. This invites the objection that the procedure is circular: the rightness of actions is determined by the goodness of their consequences, and the goodness of the consequences is determined by the rightness of actions (cf. Frankena, Ethics, 14f.). Although Portmore (“Consequentializing Moral Theories,” 41) promises to rebut this objection, he fails to offer an argument against it. 28







The Good of Consequentialized Deontology    125

(A) A terrorist was persuaded to reveal where he had hidden the bomb, and all endangered lives were saved. (No one tortured, no lives lost.) (B) A terrorist could not be persuaded to reveal where the bomb was. The bomb could not be found in time, and it detonated, thereby causing many deaths. (No one tortured, many lives lost.) (C) Only after being tortured did the terrorist reveal where the bomb was. The bomb was found, and all endangered lives were saved. (One person tortured, no lives lost.)

A moral absolutist would stick to this ranking no matter how large the number of casualties, and thus rank the outcomes (A)  > (B)  > (C). Although deontologists also prefer outcomes with fewer casualties to outcomes with more casualties, this is not true of absolutists if the outcome with fewer casualties can be obtained only by performing a wrong action (i.e., by violating an absolute constraint). Moderate (nonabsolutist) deontologists, however, would with an increasing number of casualties arrive at a point (say, one thousand deaths) where they no longer regard (C) as the worst outcome, but (B*): (B*) A terrorist could not be persuaded to reveal where the bomb was. The bomb could not be found in time, and it detonated, thereby causing one thousand deaths. Moderate deontologists thus arrive at the following ranking:  (A) > (C)  > (B*). However, they are confronted with the notorious problem of locating the threshold in the number of casualties, which changes the ranking such that the outcome arrived at through torturing the terrorist is no longer the worst outcome.34 This problem is inherited by consequentialized deontology despite the method of reflective equilibrium. Neither moderate deontologists nor consequentializers have a solution for determining where the threshold for weighing deontic intuitions against the outcomes for the affected parties lies. Both have to rely on the intuitions of the individuals who have to decide such cases.35 Besides, there is nothing that prevents moderate This leads to the problem of how one can compare the intrinsic wrongness of an action with the bad consequences of refraining from performing the action. 35 For this reason, some critics of consequentializing argue that the advantages of a new consequentialist theory are also arguments for deontologizing. Cf. Sachs, “Consequentialism’s Double-​Edged Sword,” 266f.; Hurley, “Consequentializing and Deontologizing.” Others, like Peterson (“A Royal Road to Consequentialism?,” 154), believe that consequentialists can take deontological intuitions into account, whereas deontologists cannot take consequences into account. Yet this “asymmetry thesis” can hardly be upheld if one keeps in mind that most deontologists are not absolutists who completely disregard consequences. 34

126    Monika Betzler and Jörg Schroth

deontologists from seeking a reflective equilibrium. Whichever considerations consequentializers propagating reflective equilibrium bring in to arrive at a ranking could also be brought in by moderate deontologists. Both just engage in intuitive weighing of considerations about outcomes and the intrinsic badness of actions; i.e., both resort to intuitions, and it is not only consequentializers who can bring in intuitions about consequences. Consequentializers cannot resort to any considerations that are unavailable to moderate deontologists, but are left with the same intuitions that moderate deontologists have without providing any further consequentialist rationale.36 This means that the consequentializers’ coherentist procedure does not provide a more informative and thus better moral orientation than moderate deontology, but, on the contrary, the moral content of consequentialized deontology is borrowed in full from deontological theory. On the one hand this is not surprising, since consequentializers explicitly affirm that every remotely plausible deontological theory can be consequentialized. On the other hand, it shows that consequentializing is nothing but a translation procedure and consequentialized deontology is nothing but a notational variant of moderate deontology. After all, consequentializers do not provide any criteria for deciding which of the remotely plausible deontological theories should be consequentialized and adopted as the new intuitive, nonparadoxical, maximizing moral theory. Despite consequentializers’ allegations to the contrary, consequentialized deontology cannot yield the additional—​compared to standard moderate deontology—​ moral orientation that consequentializers are looking for. Consequentializing does not establish a new moral theory, but only a schematic moral theory or the structure of a moral theory that can be filled with moral content in various ways. As long as we have not decided on a particular moral content—​and consequentializers offer no help in this respect—​we have not yet achieved a complete moral theory and therefore have no alternative to standard consequentialist and deontological theories. Consequentialized deontology therefore has no advantages over moderate deontology with regard to the justification of moral verdicts and moral orientation in cases of conflict, but instead has to swallow the disadvantages that consequentialists like to use as arguments against deontology; the objection, often advanced against Ross’s theory, that deontology is merely an unconnected heap of duties,37 then applies to consequentialized deontology, which is equally confronted with unconnected and potentially conflicting intuitions. The method of reflective equilibrium neither allows a compelling justification of these different intuitions, 36 For a similar observation, see Tenenbaum, “The Perils of Earnest Consequentializing,” 233. 37 See, e.g., Alexander and Moore, “Deontological Ethics,” section 4. See also McNaughton, “An Unconnected Heap of Duties.”



The Good of Consequentialized Deontology    127

nor does it show how exactly they can be weighed, particularly in cases of conflict.38 The lack of a criterion for weighing different requirements (prima facie duties) in deontology corresponds to the lack of a criterion for weighing and evaluating different intuitions about deontic constraints and outcomes in order to determine the best outcome in consequentialized deontology. In spite of these disadvantages, one might argue in favor of consequentialized deontology that we may now at least construct a moral theory that caters to our commonsense intuitions (even if these intuitions might conflict and are no better justified than in standard deontology), retains the compelling idea of consequentialism, and is nonparadoxical. Let us therefore investigate whether consequentialized deontology retains the compelling idea and is indeed nonparadoxical. 6.4. Abandoning Consequentialism’s Compelling Idea

Recall again that what is thought to be so compelling about consequentialism is CI: It can never be right to prefer a worse state of affairs to a better one. Accordingly, in standard consequentialism it is always permitted (and even required) that we maximize the good or, respectively, produce the best outcome—​the good and the best outcome being characterized agent-​neutrally, and thus from the observer’s point of view. In consequentialized deontology, however, we no longer subscribe to the good and the best outcome but only to the good relative to the agent’s point of view, the good relative to the patient’s point of view, and the good relative to the observer’s point of view.39 The last corresponds to the agent-​neutral good of standard consequentialism, but in consequentialized deontology it is not—​in contrast to standard consequentialism—​the best outcome an agent ought to produce. On the contrary! To avoid the counterintuitive implications of standard consequentialism, it must be guaranteed in all cases in which producing the best outcome relative to the observer’s point of view contradicts our moral intuitions (tested by a reflective equilibrium) that, first, the best outcome relative to the observer’s point of view is not the best outcome relative to the agent’s point of view, and, second, the agent is not permitted to bring about the outcome that is best relative to the observer’s agent-​neutral point of view. Only if the agent is not permitted to bring about the best outcome relative to the observer’s point of view can it be guaranteed (in accordance with our deontological intuitions) that, for example, the agent must not torture in order to save human lives. 38 Hurley (“Consequentializing and Deontologizing,”) therefore concludes that all of the “normative guidance” in such a consequentialist theory can be deontological, not consequentialist. 39 Portmore, “Can an Act-​Consequentialist Theory Be Agent-​Relative?,” 376n35.

128    Monika Betzler and Jörg Schroth

Since the agent is forbidden to produce the best outcome relative to the observer’s point of view if this outcome is not the best from the agent’s point of view, the compelling idea that it is always permitted (or required) to maximize the good relative to the observer’s point of view must be abandoned. One might think that this already suggests the failure of consequentialized deontology, since, after all, its declared aim is to retain the compelling idea of consequentialism. But consequentializers do not think that they have given up consequentialism’s compelling idea. Instead, and somewhat surprisingly, it is proposed that the compelling idea, thus far, has been misunderstood.40 The compelling idea is not that it can never be right to prefer a worse state of affairs to a better one where this implies an observer-​relative (i.e., agent-​neutral) theory of the good. Rather, the as-​yet-​ unrecognized compelling idea is (what Scanlon41 calls) the “teleological conception of reasons” (TCR): TCR  An agent’s reasons for performing an action are determined by her reasons for preferring its outcome to those of the available alternatives, such that, if S has most reason to perform x, then, of all the outcomes that S could bring about, S has most reason to desire that x’s outcome obtains.42 On an agent-​relative interpretation of TCR an agent has more reason to prefer the outcome in which she does not kill anyone but allows that some people die to the outcome in which she kills one person and thereby saves the lives of some other people. Combining the teleological conception of reasons, however, with an agent-​relative reading commits us to abandoning the idea that Scheffler and other consequentialists find so compelling:  to “make the world as good a place as possible,” which, on a natural reading, implies that the world is to be made as good as possible from an agent-​neutral point of view. Consequentializers could retort that the world is not a better place if—​as Scheffler and others have thought—​the good from the standpoint of the observer is maximized. Instead, is a better place only if the good from the agent’s standpoint—​now in the guise of her strongest reason for desiring a particular outcome—​is maximized. But why would the world be a better place if every agent does what she herself has (strongest) reason to desire? Why would the world be a better place if more killings occurred, but everyone maximized the outcome in This is suggested by Portmore, Commonsense Consequentialism, 33n18. Portmore resorts to a different compelling idea in response to criticisms by Mark Schroeder (“Teleology, Agent-​Relative Value, and ʻGood’ ” ) leveled against the viability of any agent-​relative conception of the good. 41 Scanlon, What We Owe to Each Other, 84. 42 Portmore, Commonsense Consequentialism, 33f. 40



The Good of Consequentialized Deontology    129

which they do not kill anyone? In what sense would this world be a better place? Why would it be better than a world in which fewer killings occur because agents sometimes kill in order to minimize killings? There seems to be no explanation except the deontological (or Footean) view that a world in which a (seemingly) good outcome is achieved by an intrinsically wrong action (violating a constraint) cannot be a good world (or a better world than the world in which one did not do the intrinsically wrong action and allowed more killings to occur). Apart from the fact that these are just two different intuitions about what makes the world a good place without further rationale, it remains dubious how the genuinely deontological intuition could be sold as the compelling idea of consequentialism. Consequentializers could admit that what an agent has strongest reason to desire does not make the world as such a better place. But then, why should we do it? There is nothing compelling about maximizing the good from the agent’s point of view if no further rationale can be provided that is more compelling than making the world a better place. The new compelling idea in the guise of TCR is less compelling than CI for at least two reasons. First, TCR lacks a compelling justificatory rationale. In consequentialized deontology there is no common goal that we all ought to work toward in order to make the world as good a place as possible. Instead, each agent ought only to produce the best state of affairs that she has reason to desire, i.e., from her point of view. This is less compelling because there is nothing that could explain why “making the world as good a place as possible from my (i.e., from the agent’s) perspective” is a goal that we all should strive for and that we all should accept. Even if we could subscribe to the view that this is what we would ideally want from our respective perspectives and thus individually, we cannot make sense of it as something we could collectively want to pursue as a common goal. Moreover, without further moral qualification of the reasons we have for preferring one state of affairs over another, almost anything can give an individual reasons for desiring a state of affairs over another, and almost anything can give an individual reasons for desiring a state of affairs the realization of which, however, could be detrimental to many others. After all, neither the victims nor bystanders of a particular action would endorse an action that harmed more people than an alternative action that was available to the agent. To the extent that reasons have to be intelligible and potentially justifiable considerations to qualify as reasons at all, it is hard to believe that what an agent has reason to do relative to her standpoint—​e.g., to refrain from killing even if she will prevent more killings by doing so—​is a reason the victims of her action would have reason to accept. It is not an action (omission) that will be intelligible and justifiable to them.

130    Monika Betzler and Jörg Schroth

Hence the rationale of TCR can only be grounded in the individual perspective of the agent, and is thus far less compelling for those affected by that action. Second, TCR is incomplete as an explanation about what an agent morally ought to do. We can come up with many reasons for desiring or preferring certain states of affairs, some of which do not account for what we morally ought to do. One might think that these are merely pro tanto reasons for desiring, but as pro tanto reasons they are ill-​suited to account for moral reasons. To make sense of the idea that an agent has all-​things-​considered reasons to prefer a particular state of affairs over another, we need to again resort to a theory of the good that explains those reasons in terms of their desirability. This is independent of what a particular agent finds herself desiring or preferring. Such a theory, however, is lacking in TCR. Instead, it not only leaves it largely open which reasons we might have for desiring a particular state of affairs. It also leaves it open whether those reasons have any moral pedigree. To sum up, TCR as the new compelling idea of consequentialized deontology, agent-​ relatively conceived, is far less compelling than CI as proposed by standard consequentialism. Consequentialized deontology not only does not retain the original compelling idea; its newly introduced TCR lacks both a justificatory rationale and a compelling explanation of what an agent morally ought to do. One might still maintain that even though consequentialized deontology has certain costs, there is something important to be said for it; after all, these costs might be worth the price if we finally manage to come up with a nonparadoxical moral theory. The question, therefore, is whether this is really the case.

6.5. Is Consequentialized Deontology Nonparadoxical?

One of the advantages of consequentialized deontology might still be that it avoids the so-​called paradox of deontology. After all, one of the main reasons to turn away from a plausible deontological theory is that it yields paradoxical verdicts: on the one hand, it forbids actions because of their intrinsic badness; on the other hand, it allows more of those intrinsically bad actions to happen. But with an explanation lacking for why doing wrong is worse than allowing more wrong to happen (or why doing wrong is just incomparably bad), it simply seems incoherent to hold on to constraints if this opens the door to more badness. To avoid this alleged paradox, one needs to understand what precisely brings it about. According to a common explanation, deontological constraints are not conducive to maximizing rationality, and it seems difficult to imagine a more attractive way of



The Good of Consequentialized Deontology    131

thinking about rationality.43 After all, this is what underlies the original CI. Maximizing rationality requires bringing about the best state of affairs compared to any alternative. It thus presupposes a comparative ranking of outcomes: If one of two options achieves an agent’s goal better than the other, a rational agent should choose the better one.44 In addition, TCR holds on to maximizing rationality. Reasons for preferring a particular state of affairs over another are grounded in what is better than an alternative. Any way of avoiding the paradox constraints would have to be conducive to maximization and, at the same time, would still have to require the agent to fulfill them, potential reasons to the contrary notwithstanding. They would thus have to be stringent; i.e., they would have to have more weight (albeit not absolute weight) than reasons about other considerations, such as agent-​neutral consequences, and would have to ensure that it is the agent herself who fulfills them: Do what is best (maximization) and stick to the constraints applying to you (stringency), even if this does not maximize the good (from the observer’s point of view). If constraints are just taken to be what is best from the agent’s point of view or ground the strongest reasons to prefer states of affairs in which they are fulfilled, constraints no longer seem at odds with maximization. The solution of consequentialized deontology seemingly lies in its agent-​relative axiology or, later, in its teleological conception of reasons. What constraints require is thus translated into valuable outcomes for the respective agent (or outcomes the agent has reason to prefer) compared to other alternatives. From the agent’s perspective, there is simply no reason to prefer violating another person’s rights, even if this would prevent more violations of rights by others. What the agent is taken to maximize is the fulfillment of what a particular constraint requires of her. With an agent-​relative axiology or with a teleological conception of reasons in place, both maximization and a noncomparative commitment to what a constraint requires seems possible. All the agent can (and should) do is act as much as is possible toward an agent-​relative valuable outcome or the preferred reasons to be realized. And that always entails—​in virtue of an agent-​relative theory of the good or a teleological conception of reasons—​that it is the agent herself who responds to every instance of that value or to the reasons she prefers.45 Hence we avoid the agent’s liability to disregard the stringency of constraints in cases of countervailing reasons about consequences. But does this maneuver really avoid the paradox? There are two concerns to be raised, one internal to this maneuver, the other external to it, and thus pertaining Scheffler, The Rejection of Consequentialism, 143. 44 Scheffler, The Rejection of Consequentialism, 143. 45 See Pettit, “The Consequentialist Perspective.” 43

132    Monika Betzler and Jörg Schroth

more to its theoretical reach. As to the first, internal concern, it has been pointed out that to ensure that it is always the agent herself who responds to constraints and yet is able to maximize these, a further relativization is needed. The reason for this is the following: Maximizing the agent-​relative value (or acting as much as I can on the reason I have to prefer) of my not killing, for example, involves that I do not kill even if I thereby allow more killings to happen. But this still leaves the possibility that I prefer to kill now so as to prevent myself from killing more later. To ensure that such trade-​offs within the person are precluded, a further temporal relativization needs to be added.46 Whatever value is relative to an agent is also relative to the agent at the present time. As a result, if constraints are to be conducive to maximization without falling prey to comparative assessments, the agent needs to ensure that she does not kill now. It is this amendment from temporal relativity, however, that we ultimately take to undergird the second, external concern. There is a more general problem with such a relativized account:  it avoids the paradox in theory, but not in spirit. While the theory simply precludes comparative assessments of outcomes and prescribes the maximization of whatever constraints require, it does not undo the harm that can occur if an agent does what a constraint requires, and this results in others violating their constraints. To simply ban the consideration of bad or harmful outcomes from theoretical treatment is ad hoc and does nothing to tackle the problem underlying it. So even if the theory solely allows for the maximization of agent-​relative values or teleological reasons by fiat, the question remains why an agent is required to maximize doing what a constraint requires of her if this entails that more violations of constraints will happen. What remains paradoxical is that the theory as such does not allow for these questions, but they nevertheless reappear in practice. And there are no theoretical resources for answering them. The theory is nonparadoxical, but it does not cohere with our deeply held intuitions about what is at issue in thinking about constraints: it simply does not explain why we should do what a constraint requires even if this entails that more violations of constraints happen. To be sure, there are various explanations at hand, e.g., that constraints prevent harming others by violating their rights, and that this is what can be brought about only by the particular agent (and not by others). But we find no such explanation in consequentialized deontology. Hence it is difficult to maintain that the consequentializing project avoids the paradox of deontology. The paradox reappears as soon as we turn away from the remodeled theory to practice and wonder whether and why we should not kill others even if this prevents more killings from happening.

See Louise, “Relativity of Value,” 534. Parfit had already noted in Reasons and Persons, 137–​148, that there is an important symmetry between claims that refer to oneself and claims that refer to the present time. 46



The Good of Consequentialized Deontology    133

6.6. What Is Consequentializing Good For?

According to our arguments so far, the declared aims of consequentialized deontology have not been achieved, and the prospects for achieving them seem bleak: it proves to be trivial and does not provide moral orientation. The compelling idea of standard consequentialism cannot be retained, and the new compelling idea in the guise of TCR proves to be less compelling. In addition, and even if consequentialized deontology is nonparadoxical in theory, consequentializers are unable to explain why we continue to feel confronted with a paradox in practice. All of this is due to the nature of the consequentializing project and cannot be remedied. As a consequence, consequentialized deontology is not a plausible alternative to standard consequentialism or moderate deontology. One might still wonder who could find the aims of the consequentializing project attractive. Standard consequentialists surely won’t be willing to give up their compelling idea and their moral intuitions in favor of deontological intuitions. Could deontologists be more inclined to welcome the consequentializing proj­ ect? They could argue that their rationale has finally proved to be superior, and that consequentialized deontology just proves that no more can be had than intuitions. To the extent, however, that the consequentializing project cannot really avoid the paradox, it does not score better than any plausible deontological theory either. The good of consequentializing thus seems to be rather limited. It does not meet its own aims, and it does not prove to be an interesting alternative to either standard consequentialism or a plausible deontological theory. Instead, it seems that the consequentializing project, albeit a theoretical possibility, is good for nothing.47 References Alexander, Larry, and Michael Moore. “Deontological Ethics.” In Stanford Encyclopedia of Philosophy, edited by Edward Zalta. October 17, 2016. http://​plato.stanford.edu/​entries/​ ethics-​deontological/​. Betzler, Monika, and Jörg Schroth. “Konsequentialisierung—​Königsweg oder Sackgasse für den Konsequentialismus?” Zeitschrift für philosophische Forschung 68 (2014): 279–​304. Broome, John. Weighing Goods:  Equality, Uncertainty, and Time. Oxford:  Oxford University Press, 1991. Brown, Campbell. “Consequentialize This.” Ethics 121 (2011): 749–​771. Dreier, James. “Structure of Normative Theories.” The Monist 76 (1993): 22–​40. 47 Previous versions of this chapter benefited from written comments by Marius Baumann, Paul Hurley, and Christian Seidel. We are also grateful to the audience at the conference “Consequentialism: New Directions, New Problems?” (held at Friedrich-​Alexander-​Universität Erlangen-​Nürnberg) for helpful feedback.

134    Monika Betzler and Jörg Schroth Dreier, Jamie. “In Defense of Consequentializing.” In Oxford Studies in Normative Ethics, vol. 1, edited by Mark Timmons, 97–​119. Oxford: Oxford University Press, 2011. Emet, Stephen F. “Agent-​Relative Restrictions and Agent-​Relative Value.” Journal of Ethics & Social Philosophy 4 (2010): 1–​13. Foot, Philippa. “Utilitarianism and the Virtues.” 1985. In Moral Dilemmas and Other Topics in Moral Philosophy, 59–​77. Oxford: Oxford University Press, 2002. Frankena, William K. Ethics. 2nd ed. Englewood Cliffs, NJ: Prentice-​Hall, 1973. Herman, Barbara. “Leaving Deontology Behind.” In The Practice of Moral Judgment, 208–​240. Cambridge, MA: Harvard University Press, 1993. Hooker, Brad. Ideal Code, Real World: A Rule-​Consequentialist Theory of Morality. Oxford: Oxford University Press, 2000. Hurley, Paul. “Consequentializing and Deontologizing:  Clogging the Consequentialist Vacuum.” In Oxford Studies in Normative Ethics, vol. 3, edited by Mark Timmons, 123–​153. Oxford: Oxford University Press, 2013. Kagan, Shelly. The Limits of Morality. Oxford: Oxford University Press, 1991. Kordig, Carl R. “Structural Similarities between Utilitarianism and Deontology.” Journal of Value Inquiry 8 (1973): 52–​56. Louise, Jennie. “Relativity of Value and the Consequentialist Umbrella.” Philosophical Quarterly 54 (2004): 518–​536. McNaughton, David. “An Unconnected Heap of Duties.” Philosophical Quarterly 46 (1996): 433–​447. McNaughton, David, and Piers Rawling. “Agent-​ Relativity and the Doing-​ Happening Distinction.” Philosophical Studies 63 (1991): 167–​185. McNaughton, David, and Piers Rawling. “On Defending Deontology.” Ratio 11 (1998): 37–​54. Nozick, Robert. Anarchy, State and Utopia, New York: Basic Books, 1974. Parfit, Derek. Reasons and Persons. Oxford: Oxford University Press, 1984. Peterson, Martin. “A Royal Road to Consequentialism?” Ethical Theory and Moral Practice 13 (2010): 153–​169. Pettit, Philip. “The Consequentialist Perspective.” In Three Methods of Ethics, by Marcia Baron, Philip Pettit, and Michael Slote, 92–​174. Oxford: Oxford University Press, 1997. Portmore, Douglas. “Can an Act-​ Consequentialist Theory Be Agent-​ Relative?” American Philosophical Quarterly 38 (2001): 363–​377. Portmore, Douglas. “Combining Teleological Ethics with Evaluator Relativism:  A Promising Result.” Pacific Philosophical Quarterly 86 (2005): 95–​113. Portmore, Douglas. “Consequentializing Moral Theories.” Pacific Philosophical Quarterly 88 (2007): 39–​73. Portmore, Douglas. “Consequentializing.” Philosophy Compass 4 (2009): 329–​347. Portmore, Douglas. Commonsense Consequentialism:  Wherein Morality Meets Rationality. Oxford: Oxford University Press, 2011. Sachs, Benjamin. “Consequentialism’s Double-​Edged Sword.” Utilitas 22 (2009): 258–​271. Scanlon, Thomas M. What We Owe to Each Other. Cambridge, MA:  Harvard University Press, 1998. Scheffler, Samuel. The Rejection of Consequentialism. New York: Oxford University Press, 1982. Scheffler, Samuel. Introduction to Consequentialism and Its Critics, edited by Samuel Scheffler, 1–​13. Oxford: Oxford University Press, 1988.



The Good of Consequentialized Deontology    135

Schroeder, Mark. “Teleology, Agent-​Relative Value, and ʻGood.’” Ethics 117 (2007): 265–​300. Sen, Amartya. “Evaluator Relativity and Consequential Evaluation.” Philosophy and Public Affairs 12 (1983): 113–​132. Smith, Michael. “Neutral and Relative Value after Moore.” Ethics 113 (2003): 576–​598. Smith, Michael. “Two Kinds of Consequentialism.” Philosophical Issues 19 (2009): 257–​272. Sumner, L.W. “Two Theories of the Good.” Social Philosophy and Policy 9 (1992): 1–​14. Tenenbaum, Sergio. “The Perils of Earnest Consequentializing.” Philosophy and Phenomenological Research 88 (2014): 233–​240.

7 T HE RES T R I C T I ONS OF CONSE QUE NTIA L I S M

Jan Gertken

Both commonsense morality and many ethical theories are committed to the view that there are deontic restrictions, which prohibit certain morally problematic actions even in cases where such acts minimize the occurrence of actions of the very same type. However, many find it hard to reject what is sometimes called consequentialism’s compelling idea, according to which it can never be wrong to bring about the best possible outcome.1 At least at first glance, the compelling idea seems to conflict with the idea that there are deontic restrictions. Some have taken this as a reason to reject the view that there are restrictions, or to argue that these are paradoxical.2 More recently, however, some consequentialists have also tried to show that restrictions can be accommodated within a consequentialist framework, and it is widely believed that in order to achieve this, one has to make suitable assumptions about so-​called agent-​relative value.3

1 On the compelling idea, see Portmore, “Combining Teleological Ethics”; Schroeder, “Teleology”; Smith, “Two Kinds.” For the purpose of this paper, I assume that the compelling idea deserves its name, although what seems most compelling to me is rather the view that it is always right to do what is best. That the best act is always the one leading to the best possible outcome is far from trivial (cf. Hurley, Beyond Consequentialism, ch. 5). 2 See, e.g., Scheffler, “Agent-​Centred Restrictions.” On the “paradox of deontology,” see also Heuer, “The Paradox.” 3 For variations of this proposal, see Dreier, “Structures of Normative Theories”; Amartya Sen, “Evaluator Relativity”; Portmore, “Combining Teleological Ethics”; Smith, “Two Kinds.”

136



The Restrictions of Consequentialism    137

In this chapter, I  argue that the importance of agent-​relative value has been overrated for the purpose of integrating restrictions into a consequentialist outlook. It is not our understanding of value but rather our understanding of outcomes that is crucial for a consequentialist account of restrictions. After introducing the concepts needed for the following discussion (section 7.1), I show that it is possible to integrate restrictions into a consequentialist framework without appealing to agent-​relative value, namely, by employing an agent-​centered conception of outcomes (section 7.2). I next argue that even consequentialist accounts of restrictions that rely on ascriptions of agent-​relative value fail to be satisfying unless they are complemented by an agent-​centered conception of outcomes (section 7.3). I offer some thoughts on how such a conception is best understood and on how it can be supported by further considerations (section 7.4). 7.1. Conceptual Groundwork

Consequentialism, as I  will understand it here in this chapter, is the view that an act is morally right if and only if (and because) its outcome realizes at least as much value (or at least as much value of a certain sort) as all alternative outcomes that the agent can bring about.4 By an outcome of an action, I mean a state of affairs that is brought about by that action.5 I use “the outcome of an action” to refer to the collection of all states of affairs brought about by that action, and I use “action” in a wide sense that includes everything that is a possible object of an intention. It has become common to label the type of value that is ascribed by judgments about what is good, valuable, or desirable simpliciter (and their comparative forms) agent-​neutral value, and to distinguish this sort of value from agent-​relative value, which is ascribed by judgments to the effect that certain things, events, or states of affairs are good, valuable, etc. relative to some particular agent (and their comparative forms). Correspondingly, we can distinguish two kinds of consequentialism: Agent-​ neutral consequentialism comprises all consequentialist theories that merely rely on ascriptions of agent-​neutral value. In contrast, forms of agent-​relative consequentialism are consequentialist theories that involve an axiology that is at least partly agent-​relative.

4 My discussion focuses on value-​based forms of objective act-​consequentialism in its maximizing variant as a theory of moral rightness. This facilitates the presentation of the main points, and the central results generalize to other forms of consequentialism that incorporate the compelling idea. 5 Cf. Sosa, “Consequences of Consequentialism,” 102.

138    Jan Gertken

As I use the term “restriction,” a restriction against ϕ-​ing rules out ϕ-​ing even in situations where doing so minimizes instances of ϕ-​ing.6 This formulation is open to different interpretations, which concern the scope of the prohibition against ϕ-​ing. If ϕ-​ing is never permissible, a restriction against ϕ-​ing is absolute. However, one might accept a restriction against ϕ-​ing while allowing that it is permissible to ϕ in some situations, e.g., if very bad consequences or catastrophic events would otherwise occur.7 Let us call a potential restriction case any situation wherein someone else will perform a morally problematic action ϕ unless the agent ϕ-​s herself. Let us furthermore say that a potential restriction case is covered by a restriction against ϕ-​ing if it is wrong to ϕ for the agent in that situation. Which potential restriction cases are covered by a restriction is a substantial normative question, and different ethical views that accept restrictions can disagree on this issue. However, there is one class of situations that need to be covered for there to be any restriction against ϕ-​ing at all, namely, those potential restriction cases that do not involve any further morally relevant aspects beyond the fact that someone else will perform a morally problematic act ϕ unless the agent ϕ-​s herself. I will call such cases minimal restriction cases. If there is to be any restriction against ϕ-​ing at all, there must be a minimal restriction against ϕ-​ing, i.e., a restriction that covers minimal restriction cases. In order to illustrate these ideas, consider the restriction against killing innocents, which I take to be a fairly uncontroversial example for a restriction, and which I will use from now on to illustrate and discuss the main points.8 For there to be any restriction against killing innocents, it must at least be the case that killing innocents is wrong in situations where someone else will kill an innocent person unless the agent does so and where no further relevant facts beyond this are to be taken into consideration. Every attempt to make sense of some kind of restriction against killing

6 According to this definition, the existence of a restriction against ϕ-​ing does not imply that it is sometimes wrong to bring about the best outcome. This is in contrast to a common definition of restrictions according to which a restriction against ϕ-​ing obtains if it is wrong (other things being equal) to ϕ in order to maximize the good, even if by ϕ-​ing one minimizes actions of the very same type (cf. Portmore, Commonsense Consequentialism, 97–​98). Thus defined, it is trivial that agent-​neutral consequentialism cannot accommodate restrictions, whereas it is a substantial and philosophically interesting question whether it can accommodate restrictions as defined in the main text. Some authors use “constraint” for what I call “restrictions” (cf. Kamm, Morality, Mortality, 207). Others use “constraint” either for the idea that some actions may not be done in order to maximize the good (Kagan, Limits, 4), or for the more complex idea that some actions may not be done to maximize the good, even if by performing them one minimizes actions of the very same type (cf. Nozick, Anarchy, State, and Utopia, 28–​35; Schroeder, “Teleology”). 7 Cf. Kagan, Limits, 189–​191. 8 To simplify formulations, I will mostly use “kill” in what follows and take it to be understood that this specifically refers to acts of killing innocents.



The Restrictions of Consequentialism    139

innocents must therefore at least make sense of the minimal restriction against killing innocents (=MRK): MRK  For all agents A: absent further relevant facts, it is wrong for A to kill an innocent person, even in situations where someone else will kill an innocent person unless A does so. Minimal restrictions such as MRK suffice to formulate the main challenge that deontic restrictions pose for consequentialist views, which is why my discussion will mostly concern them. This focus allows us to address the issue of whether accepting restrictions conflicts with a theory’s commitment to consequentialism’s compelling idea in the easiest possible way. So far, I have said that some consequentialists try to make sense of restrictions or want to integrate restrictions into their theories. This needs to be made more precise. Let us say that a theory T consequentializes MRK if and only if T is a consistent consequentialist theory that implies MRK. In order to do so, T needs to comprise a suitable axiology that, together with the consequentialist criterion of rightness, implies MRK. Say that a consequentialist theory T accommodates MRK if and only if T consequentializes MRK without incurring any controversial commitments over and above those that are involved in the very idea of MRK itself. A theory T might consequentialize MRK without accommodating it, since in order to consequentialize MRK, T might have to make problematic postulates that one is not committed to by accepting MRK as such. The distinction between consequentialization and consequentialist accommodation is important, since making sense of a normative phenomenon is incompatible with explaining it away. If a consequentialist theory makes the view that there are restrictions look more problematic than it does on its own, this theory does not provide an account of restrictions, but rather undermines the conviction that there are any restrictions to be given an account of. In what follows, I will therefore assume that consequentialist accounts of restrictions do not aim at mere consequentialization, but rather at accommodation. 7.2. Two Approaches to Consequentializing Restrictions

There is a well-​known line of thought that suggests that no version of agent-​neutral consequentialism can consequentialize a restriction such as MRK.9 Consider the two minimal restriction cases and the partial description of the relevant outcomes 9 My presentation of the relevant points follows Schroeder, “Teleology,” and Emet, “Agent-​Relative Restrictions.”

140    Jan Gertken

Table 7.1.  Two minimal restriction cases with agent-​indifferent partial description of the relevant outcomes Case #1: B will kill D unless A kills C Case #2: A will kill D unless B kills C Outcome of A’s killing (=o1): A kills C, B does not kill D, C is dead . . .

Outcome of B’s killing (=o3): B kills C, A does not kill D, C is dead . . . 

Outcome of A’s refusal to kill (=o2): B kills D, A does not kill C, D is dead, A lets D die . . .

Outcome of B’s refusal to kill (=o4): A kills D, B does not kill C, D is dead, B lets D die . . .

shown in Table 7.1 (“A” and “B” stand for the same agents in both cases, and “C” and “D” stand for the same innocent persons). Assume that there is an agent-​neutral consequentialist theory T that implies MRK. In order to do so, T needs to imply both that it is wrong for A to kill in case #1 and that it is wrong for B to kill in case #2. Hence T needs to imply both that o2 is better than o1 and that o4 is better than o3. Abstracting from all possible contingent differences, the relevant evaluative asymmetry between these outcomes needs to somehow result from the negative contribution made by A’s killing and B’s killing, respectively.10 However, this different contribution cannot be due to the negative agent-​neutral value of acts of killing innocents. For no matter how much negative value we ascribe to acts of this type, this will not result in any evaluative difference between o1 and o2, or between o3 and o4, since there is no difference between o1 and o2, or between o3 and o4, as far as the occurrence of acts of killing innocents is concerned. If there is a relevant difference between these outcomes, it cannot concern what is done, but only who does it. What needs to be the case for o2 to be better than o1 and for o4 to be better than o3 is thus this: A’s killing must be worse than B’s killing, and B’s killing must be worse than A’s killing. These assumptions, however, are jointly inconsistent, assuming that the better than relation is asymmetric. Hence there is no agent-​neutral consequentialist theory that consequentializes MRK. It is plausible to assume that this inconsistency does not arise for judgments about agent-​relative value. Assuming both that A’s killings are worse relative to A than B’s killings and that B’s killings are worse relative to B than A’s killings seems consistent. If there are no loopholes in the argument outlined above, the conclusion to be I take the relevant value contribution to be made by the fact that A and B each kill an innocent person, and not by something that necessarily results from this (such as the fact that A and B each violate a right, are each responsible for an innocent person’s death, each fail to respect a person’s dignity, etc.). The relevant points discussed in what follows are independent of this assumption. 10



The Restrictions of Consequentialism    141

Table 7.2.  The two minimal restriction cases with agent-​centered partial description of the relevant outcomes Case#1: B will kill D unless A kills C. Case #2: A will kill D unless B kills C. Outcome of A’s killing (=o1): A kills C, B does not kill D, C is dead . . .

Outcome of B’s killing (=o3): B kills C, A does not kill D, C is dead . . .

Outcome of A’s refusal to kill (=o2): A does not kill C, D is dead, A lets D die . . .

Outcome of B’s refusal to kill (=o4): B does not kill C, D is dead, B lets D die . . .

drawn is that one needs to accept the framework of agent-​relative consequentialism in order to consequentialize restrictions.11 However, there is such a loophole, which friends of agent-​neutral consequentialism might try to exploit, since the discussion has so far relied on a hidden assumption that could be questioned. Let us have another look at the description of the relevant outcomes in Table 7.1, and let us suppose that a correct description of the outcomes of A’s killing and of A’s refusal to kill in case #1 mentions less than what is listed in the first column. More specifically, assume that if A kills in a minimal restriction case such as case #1, the outcome of A’s action includes A’s killing, whereas if A does not kill, the outcome does not include B’s killing, but merely D’s death and A’s act of letting D die. Assume that similar modifications are made with regard to the outcomes of B’s actions in case #2. The resulting partial description of the relevant outcomes is shown in Table 7.2.12 Relying on this description, any axiology that takes killings of innocents to be agent-​neutrally worse than mere deaths of innocents and agent-​neutrally worse than acts of letting innocents die will imply both that o2 is better than o1 and that o4 is better than o3 without involving any inconsistency. Rejecting the assumption that other agents’ killings are part of the outcome of an agent’s refusal to kill in minimal restriction cases hence allows one to give a straightforwardly agent-​neutral consequentialist account of MRK. Let us call a conception of outcomes agent-​centered if and only if it implies the following: for all agents A, there is a type of action ϕ such Cf. Emet, “Agent-​Relative Restrictions,” 2–​3; Schroeder, “Teleology,” 267. 12 Given how “potential restriction case” and “minimal restriction case” have been defined above, it is not part of the definition that other agents’ acts of ϕ-​ing would be an outcome of A’s refusal to ϕ in these cases. It is necessary and sufficient for A’s situation to count as a potential restriction case that someone else will perform a problematic act of ϕ-​ing unless A ϕ-​s. Inferring from this that A would bring about someone else’s ϕ-​ing by refusing to ϕ requires further assumptions. 11

142    Jan Gertken

that only A can bring about outcomes including instances of ϕ-​ing that are done by A. Say that a conception of outcomes that is not agent-​centered is agent-​indifferent. This additional terminology allows us to summarize the main upshot of the previous discussion as follows: One way to consequentialize restrictions consists in developing a suitable agent-​centered conception of outcomes, which at least deals with those cases that are covered by restrictions.13 Given a suitable agent-​centered conception of outcomes, both agent-​neutral and agent-​relative axiologies can be used for the purpose of such a consequentialization.14 So there is at least one way to consequentialize restrictions that is open to advocates of agent-​neutral consequentialism. I will argue that there is an important sense in which this is also the most promising approach. More specifically, I  will argue that unless consequentialist theories accept an agent-​centered conception of outcomes, even an agent-​relative axiology does not allow them to accommodate deontic restrictions such as MRK.15 7.3. How an Agent-​Indifferent Conception of Outcomes Endangers Consequentialist Accounts of Restrictions

In section 7.2, we have seen that, given an agent-​indifferent conception of outcomes, agent-​relative versions of consequentialism need to postulate a difference between the agent-relative value of one’s own killings and the agent-relative value of acts of killing done by other agents in order to consequentialize MRK. I will call this postulate the evaluative asymmetry hypothesis (= EAH): EAH  For all agents A  and B: other things being equal, A’s acts of killing innocents are worse relative to A than B’s acts of killing innocents. 13 Instead of restricting what counts as an outcome of an action, one might also restrict the range of outcomes that matter to assessing the moral rightness and wrongness of actions, and rule out those outcomes that involve (certain types of ) actions performed by other agents. (Elinor Mason suggests a proposal along these lines in her contribution to this volume.) This approach, however, seems to conflict with the compelling idea, at least on its most straightforward understanding. It is thus not an option for the sort of consequentialism I am considering in this paper. 14 Whether a consequentialist account of restrictions is then best formulated in terms of agent-​neutral or agent-​ relative value partly depends on whether agent-​relative or agent-​neutral consequentialism provides a better interpretation of the compelling idea (cf. note 1). For more on this question, see the discussion in Schroeder, “Teleology”; Portmore, Commonsense Consequentialism; Sachs, “Consequentialism’s Double-​Edged Sword”; and the contribution by Monika Betzler and Jörg Schroth in this volume. 15 This claim needs to be qualified in one important respect. Throughout my discussion, I follow the somewhat established view that considers outcomes of actions to consist in states of affairs. Revising our conception of what type of entities count as outcomes, in the way suggested by Jamie Dreier in his contribution to this volume, seems to open up still further options.



The Restrictions of Consequentialism    143

I will now argue that given a commitment to an agent-​indifferent conception of outcomes, agent-​relative consequentialist accounts of MRK that rely on EAH face a dilemma:  either they postulate ungrounded evaluative asymmetries, in which case their accounts involve implausible commitments that are not involved in the very idea of MRK itself, or they try to account for the evaluative asymmetry postulated by EAH, in which case they have to accept an additional assumption that sabotages the intended consequentialization of MRK. In either case, they do not accommodate MRK.16 EAH cannot be a basic assumption. Consequentialists who accept EAH should be able to say something informative about why the asymmetry stated by EAH holds. Nothing is valuable without there being something that makes it valuable, and some entity x cannot be more valuable than some other entity y without there being something that makes x more valuable than y—​no matter whether the value in question is agent-​neutral or agent-​relative. A consequentialist account of MRK that postulates ungrounded differences in value does not accommodate MRK, since a commitment to ungrounded differences in value is both implausible and not involved in the very idea of MRK itself. Consequentialist accounts of MRK that rely on EAH must therefore be able to point to some factor that makes A’s killings worse relative to A than B’s killings, for all agents A and B (at least all other things being equal). There are two ways to do this. For one thing, it could be argued that whereas A’s killings have negative value relative to A in virtue of some fact, B’s killings do not have negative value relative to A. For another thing, it could be argued that although both A’s and B’s killings have negative value relative to A, there is something about either A’s or B’s killing that makes it the case that A’s killings are worse relative to A than B’s killings. We can set aside the first option: it is not only plausible to assume that A’s killings are bad relative to A; it is also plausible to assume that B’s killings are bad relative to A (and, of course, bad relative to B as well). It is not only fitting for A to desire that she does not kill, or to prefer her not-​killing to her killing; it is also fitting for A to desire that B does not kill, and to prefer B’s not-​killing to B’s killing. Given the

Mark Schroeder has criticized agent-​relative consequentialist theories on semantic grounds and objected that judgments about so-​called agent-​relative value cannot be interpreted as value judgments in a way that serves the consequentializer’s purposes. Although Schroeder accepts fitting attitude analyses of relative value judgments as intelligible, he argues that it is hard to see how consequentialists could consistently consequentialize constraints on this basis (“Teleology,” 292–​294; on fitting attitude analyses see also Rabinowicz and Rønnow-​Rasmussen, “The Strike of the Demon”). His objections do not extend to suggestions to consequentialize restrictions by appeal to agent-​relative value, as the notion is used here (cf. note 6). For criticism of Schroeder’s arguments, see Smith, “Two Kinds”; Suikkanen, “Consequentialism”; Dreier, “In Defense of Consequentializing.” 16

144    Jan Gertken

plausible assumption that the fittingness of attitudes such as desires and preferences provides evidence for their objects’ value,17 this supports the view that not only an agent’s own killings but also killings done by others have negative value relative to that agent.18 7.3.1. The Importance of Being Me (Relative to Me)

This means that consequentialists who accept EAH in order to consequentialize MRK need to give us an account of why A’s killings are worse relative to A than B’s killings, even though both acts have negative value relative to A. Given the very setup of this challenge, the only difference between A’s killing and B’s killing to which consequentialists could point in order to account for the postulated difference in agent-​relative value must have something to do with the fact that the killing in each case is performed by a different agent. The idea must be that if my agency is involved in an act that has negative value relative to me, this somehow makes this act worse relative to me than it would be if my agency were not involved. On this view, the fact that an agent A is identical to the agent performing some act ϕ with negative value relative to A affects ϕ’s negative value relative to A. Since this influence of identity is best thought of as a factor that intensifies the negative value that ϕ has independently,19 let us hence call the view under consideration the Intensifying Identity Hypothesis (= IIH): IIH If ϕ has negative value relative to A, and if ϕ is performed by A, then the fact that A is identical to the agent performing ϕ intensifies the negative value that ϕ has relative to A (making ϕ worse relative to A than it would be if someone else performed that act). As far as I can see, IIH is the consequentialist’s most promising option for giving an account of the evaluative asymmetry postulated by EAH.20 Although there is a lot to 17 This does not imply a commitment to a fitting-​attitude analysis of agent-​relative value (cf. note 16), but can also be accepted by advocates of rival conceptions of what agent-​relative value consists in. 18 That something is bad relative to A does not imply that it cannot be bad relative to someone else as well. We could, of course, reserve the expression “agent-​relative value” for those objects or states of affairs that have value only relative to some particular agent. In this case, however, the above considerations would call into question that A’s killings have agent-​relative value in this more restricted sense. 19 On the notion of an intensifier, see Dancy, Ethics without Principles, 41–​42, 55. 20 The resulting account belongs to the class of agent-​focused accounts of restrictions, which try to explain restrictions in terms of normatively relevant properties of agents or their actions. Agent-​focused accounts are commonly contrasted with victim-​focused accounts, which focus on normatively relevant properties of potential victims (cf. Kamm, Morality, Mortality, 237–​258; Portmore, Commonsense Consequentialism, 100–​102).



The Restrictions of Consequentialism    145

be said about the pros and cons of this principle, I will grant for the following discussion that IIH is a defensible assumption.21 Instead, I will argue that IIH solves the problem of accounting for EAH only at the cost of sabotaging the intended agent-​ relative consequentialization of MRK, given the framework of an agent-​indifferent conception of outcomes that makes postulating EAH attractive in the first place. My argument relies on three premises.

(1) In order to consequentialize restrictions, one needs a notion of outcome according to which it is true that, if p is part of the outcome of A’s ϕ-​ing, then A’s bringing about p is also part of that outcome.

It is widely acknowledged that in order to consequentialize restrictions, one needs a conception of outcomes that is comprehensive enough to include both causal and noncausal consequences of an act and, most important, even the act itself among its own outcome.22 As David Sosa has pointed out in his defense of such a broad understanding of outcomes, this means that if p is an outcome of A’s ϕ-​ing (and hence part of what A brings about by ϕ-​ing), so is A’s bringing about p.23

(2) Not only acts of killing have negative value (agent-​relative or agent-​ neutral); acts of bringing about acts of killing have negative value as well (agent-​relative or agent-​neutral).

To support this idea, note that just as it seems fitting for me to desire that you do not kill, it is also fitting for me to desire that I do not bring about your killing, and fitting for me to desire that you do not bring about your killing.24

(3) For all agents A: A’s act of killing cannot differ in (agent-​neutral or agent-​relative) value from A’s act of bringing about A’s act of killing, at least in those cases where A brings about A’s act of killing by killing.

See Kamm, Morality, Mortality, 240–​253, and Emet, “Agent-​Relative Restrictions,” for objections to the agent-​ focused approach. 22 Cf. Portmore, “Consequentializing Moral Theories,” 39. 23 Cf. Sosa, “Consequences of Consequentialism,” 112. 24 Of course, your killing and my bringing about your killing together cannot be worse (agent-​neutrally or agent-​ relatively) than your killing and worse (agent-​neutrally or agent-​relatively) than my bringing about your killing. Assuming otherwise would be a problematic instance of double counting, given that the negative value of bringing about your killing is not independent from the negative value of your killing. Premise (2) is an assumption about what has negative value, not an assumption about the aggregation of the relevant type of negative value. 21

146    Jan Gertken

Premise (3) seems trivially true to me. I will now argue that premises (1)–​(3), together with IIH, call into question the viability of a consequentialization of MRK that relies on an agent-​indifferent conception of outcomes. Assuming an agent-​indifferent conception of outcomes, (1), together with the description of the relevant minimal restriction cases, implies that:

(4) The outcomes of A’s killing and of A’s refusal to kill in a case covered by MRK will at least involve the following: Outcome of A’s killing (=o1): A kills, B does not kill, A brings it about that A kills . . . Outcome of A’s refusal to kill (=o2): A does not kill, B kills, A brings it about that B kills . . .

Next, IIH, together with premise (2), implies:

(5) A’s bringing about B’s killing is worse relative to A  than B’s bringing about B’s killing.

From (3), it follows that:

(6) B’s killing has the same negative value relative to A as B’s bringing about B’s killing.

From (5) and (6), it follows that:

(7) A’s bringing about B’s killing is worse relative to A than B’s killing.

As I have pointed out before, IIH implies that:

(8) A’s killing is worse relative to A than B’s killing.

Hence, we can conclude from (4), (7) and (8) that:

(9) Outcome o1 contains A’s killing, which is worse relative to A than B’s killing (which is part of o2). Outcome o2 contains A’s act of bringing about B’s killing, which is also worse relative to A than B’s killing.

At this point, it is important to remember that in order to consequentialize MRK, an agent-​relative consequentialist theory not only needs to imply that A’s killing is worse relative to A than B’s killing; it must also imply that the whole outcome of A’s killing is worse relative to A  than the whole outcome of A’s refusal to kill in those



The Restrictions of Consequentialism    147

cases that are covered by MRK. Given the results of the argument so far that are summarized in (9), it is very hard to see how the desired implication could follow. For o1 to be overall worse relative to A  than o2, it is not enough that A’s killing (which is part of o1) is worse relative to A than B’s killing (which is part of o2). It also needs to be shown that A’s killing (which is part of o1) is worse relative to A than A’s bringing about B’s killing (which is part of o2). It is hard to see which resources agent-​ relative consequentialist views could use to that effect. First, IIH does not allow us to conclude that A’s killing is worse relative to A than A’s bringing about B’s killing. This is because A’s killing and A’s bringing about B’s killing are not actions of the same type performed by different persons, but rather different types of actions performed by the same person. This means that there remains a gap in the consequentialist account of the minimal restriction against killing, even if IIH is accepted. Some other principle needs to be brought into play to close this gap, but I find it very hard to see what such a principle could look like.25 What is more, according to IIH, what makes A’s killing worse relative to A than B’s killing is the very same intensifying factor that also makes A’s bringing about B’s killing worse relative to A than B’s killing. According to IIH, both A’s bringing about B’s killing and A’s killing are acts that are worse relative to A than B’s killing for the very same reason, namely due to the fact that they involve A’s agency. This means that IIH not only leaves a gap in the consequentialist account of MRK; it also offers strong support for the view that the outcome of A’s killing (=o1) is not worse relative to A than the outcome of A’s refusal to kill (=o2). 7.3.2. Upshot

Given the framework of an agent-​indifferent conception of outcomes, agent-​relative consequentialist theories that aim to consequentialize MRK face a dilemma: either they postulate an ungrounded evaluative asymmetry between one’s own killings and One might object that A’s bringing about A’s killing must be worse relative to A than A’s bringing about B’s killing, since the first act involves A’s killing, whereas the second involves B’s killing. Given that A’s bringing about A’s killing and A’s killing are equally bad relative to A, this would allow one to derive the conclusion that A’s killing is worse relative to A than A’s bringing about B’s killing. (I owe this suggestion to Christian Seidel.) However, in the current context, it cannot be taken for granted that A’s bringing about A’s killing is made overall worse relative to A than A’s bringing about B’s killing by the fact that the former act involves A’s killing and the latter act involves B’s killing. This assumption neglects the idea that A’s being involved in A’s bringing about B’s killing crucially affects the A-​relative value of that action, but not the A-​relative value of B’s killing. In other words, the suggestion ignores the very idea that IIH expresses. What is more, if, as I argue in the main text, IIH furthermore provides us with strong reasons to conclude that A’s killing is not worse relative to A than A’s bringing about B’s killing, the aforementioned argument could at most call into question the consistency of the account under consideration, which would also undermine the intended consequentialization of MRK. 25

148    Jan Gertken

the killings done by others, which means that their account involves implausible commitments that are not involved in the very idea of MRK itself, or they accept IIH, in which case their account fails to yield the desired consequentialization of MRK. In either case, they fail to accommodate MRK. Consequentialists who want to accommodate restrictions such as MRK should hence reject agent-​indifferent conceptions of outcomes, no matter whether they prefer an agent-​neutral or partially agent-​relative axiology. Next, I will sketch how such an approach could be developed and supported by further considerations, and how it could help the consequentialist account for both MRK and certain extensions of this minimal restriction. My aim is not to present a finished account, however, but to present a perspective for further examination. 7.4. Restricting Outcomes

Many consequentialists take p to be an outcome of an action ϕ if p counterfactually depends on ϕ.26 If we accept this as a sufficient condition, then it is hard to see how actions of other agents could fail to be part of the outcome of one’s own actions in cases that are covered by restrictions. However, there are independent reasons to conceive of an action’s outcome more narrowly, if we stick to our pretheoretical understanding of what it means to bring about a state of affairs or an event by acting, and if we identify the outcome of an agent’s action with everything that the agent brings about in this sense.27 This notion of outcome is broad enough to include noncausal consequences of an action and, as a limiting case, maybe even the action itself.28 It nevertheless does not cover every action of other agents that counterfactually depends on one’s own, not even in those cases where one knows about this dependence. For example, my neighbor would not drink a coffee right now if I had choked him to death two minutes ago, but I did not bring about my neighbor’s act of drinking by not choking him to death (not even given that I could foresee that he would drink his coffee in this case). I did not in any meaningful way contribute to his drinking. He does that all by himself, since the action originated in him and since he also controls both his attempt to drink the coffee and the success of Some accounts go even further and understand the outcome of ϕ to include the whole way the world is if ϕ is performed, or identify the outcome of ϕ with the possible world that is actual if ϕ is performed (cf. Portmore, Commonsense Consequentialism, 56). 27 It is this notion of bringing about that seems most directly connected to consequentialism’s compelling idea, which is usually formulated by relying on an intuitive and pretheoretical understanding of bringing about something by acting. Fully developed consequentialist accounts often employ notions of “outcome” or “bringing about” that function as technical terms and seem far removed from this ordinary understanding. 28 For a dissenting view, see Paul Hurley’s contribution to this volume. 26



The Restrictions of Consequentialism    149

his attempt. I do not play a part in any of this. The fact that act ψ counterfactually depends on act ϕ does not seem sufficient for ψ to count as something that is brought about by ϕ in the ordinary sense of “bringing about.” This idea can be given a further rationale. Our ordinary notion of bringing about is closely connected to assumptions about responsibility. The range of what I bring about in this sense is determined by how far my agency extends in shaping the world, and hence encompasses those aspects of the world for which I am responsible—​at least in those cases where I know what I am doing in acting that way. Typically, however, the responsibility for one’s own action lies solely with the agent who performs the act. In many cases, I am not responsible for what others do, even if their actions counterfactually depend on what I do.29 These remarks suggest that our ordinary notion of bringing about does not apply to all actions that counterfactually depend on what we do. Consequentializing restrictions would be easy if it could furthermore be shown that we never bring about other persons’ actions in this sense. Unfortunately, matters are not quite so simple, since there are three types of situations where it does seem to make sense to conceive of other persons’ actions as something that we bring about:

(i) situations where A’s action enables B to succeed in what B does; (ii) situations where A causes B to act by means that bypass B’s capacity for rational choice; (iii) situations where A gives B decisive reason to act in a certain way.

In the remainder of this section, I will consider whether the impression that other persons’ actions are brought about in these kinds of cases is correct, and I will examine to what extent a consequentialist account of the restriction against killing that relies on the ordinary notion of bringing about can admit that we sometimes bring about other agents’ acts of killing by refusing to kill.30 7.4.1. Enabling an Action’s Success

Let us start by considering cases wherein the success of B’s attempt to kill depends on A’s refusal to kill. Imagine, e.g., that B has set loose a trolley that will kill a person unless A throws on the tracks another person, who will stop the trolley but die in the process. In such cases, it might be argued, A’s refusal to kill enables the success of The idea that an agent’s responsibility is potentially limited with regard to those aspects of the world that are brought about by other agents is also discussed in Elinor Mason’s contribution to this volume. 30 For more on how an action’s influence on other actions could matter from a consequentialist perspective, see Douglas Portmore’s chapter in this volume. 29

150    Jan Gertken

B’s attempt to kill and hence contributes to what B does, since without A’s refusal, B would merely try to kill. Even if we do not question this inference, such cases are best understood as situations wherein A and B together bring about that particular part of B’s action that is constitutive of the act’s success. In the case under consideration, this is the death of the person who gets killed by B. What is brought about by both agents in such cases is hence merely a part of B’s action, and there are other parts of what B does that are not brought about in this way, such as, e.g., B’s intention or choice. This suggestion receives further support by the fact that in many such cases, the act of the person that enables the success of an attempt to kill will occur later than the decision to kill. Here, the person enabling another person’s act of killing cannot bring about the complete act of killing, since one cannot bring about events that started before one began to act. As far as an agent’s contribution to another person’s action goes, such cases seem furthermore relevantly similar to situations where an agent’s help precedes the act the success of which it enables. Situations where B will kill unless A kills, due to the fact that A’s refusal to kill will enable the success of B’s attempt to kill, are therefore most plausibly understood as follows: First, A does not bring about B’s killing in such cases, but only a certain part of B’s killing. Second, given the plausible assumption that complete killings are worse than mere deaths, and also worse than mere parts of killings, what A brings about by killing in such cases is worse (and worse relative to A) than what A brings about by refusing to kill. Consequentialists thus have significant resources to account for the wrongness of killing in those situations that involve enabling relations between actions of different agents and that are covered by a restriction against killing. 7.4.2. Bypassing the Capacity for Rational Choice

In this and the next paragraph, I will consider potential restriction cases where it makes sense to say that one person brings about another person’s complete action. One such type of case concerns situations where A causes B to behave in a certain way by means that bypass B’s capacity for discerning, evaluating, and responding to reasons, i.e., B’s capacity for rational and autonomous choice. Paradigm examples of this kind include, e.g., hypnosis, brainwashing, cases where a person is caused to have an extreme emotion that makes her lose control over what she does, and situations wherein an agent gets intoxicated to the same effect. It seems plausible that one can bring about another person’s actions in such ways, and with some creativity, potential restriction cases of this kind can surely be described.



The Restrictions of Consequentialism    151

This does not rule out a consequentialist account of restrictions against killing. Arguably, if A gets B to ϕ by affecting B in a way that bypasses B’s capacity for rational and autonomous choice, then B does not act voluntarily. This implies, first, that there are no minimal restriction cases where A  brings about B’s involuntary killing by refusing to kill, for the voluntariness of an act is a morally relevant prop­ erty. What is more, insofar as involuntary killings can be taken to have less negative value than voluntary killings, consequentialists can also account for the wrongness of killing innocents in those potential restriction cases where B’s involuntary killing is brought about by A’s voluntary refusal to kill (and where A’s killing would also be a voluntary act). In such cases, consequentialists could argue, there is an important evaluative difference between A’s voluntary killing and B’s involuntary killing, which A would bring about by refusing to kill. Cases where A is caused to kill involuntarily, but where A’s refusal to kill would have brought about B’s involuntary killing, are more difficult to handle. Here, consequentialists seem best advised to reject the assumption that such cases are covered by any restriction against killing at all. For one thing, it is far from clear whether involuntary acts are a proper subject for moral judgments of rightness and wrongness. And even if they are, it seems clear that if A is made to kill in the way described, then A could not have acted otherwise in an important sense. Assuming that an action can be wrong only if the agent performing it ought to have acted otherwise, consequentialists could appeal to the widely accepted ought implies can principle to argue that A does not act wrongly in cases where A is caused to kill involuntarily and where her refusal to kill would bring about an involuntary act of killing performed by B. Rejecting the view that potential restriction cases of this kind are indeed covered by a restriction against killing might not be a hard bullet to bite. 7.4.3. Leaving No Reasonable Alternative

The previous discussion indicates that in order to bring about another agent’s voluntary action, one needs to bring about the action in a way that appeals to this person’s capacity for rational choice. This is surely possible. Assume, for example, that I make you an offer that you “cannot refuse” by offering you a sizable amount of money for doing me a small favor and that you, in the light of this consideration, do what I ask of you. Or consider a case where a superior commands a soldier to perform a certain action and where the soldier performs the action in response to the command. These seem to be good examples of cases where it makes sense to say that B’s action is brought about by A in a rational way. This seems due to the fact that in such cases, B is given decisive reasons to act in a certain way, which leave B no reasonable alternative to acting in that way. In such cases, it seems appropriate to regard A as

152    Jan Gertken

responsible for what B does in response to these reasons. After all, A made it the case that B had no reasonable alternative way of acting available to him. Despite that, B’s act remains voluntary and autonomous in such cases, since it was the upshot of an exercise of B’s ability to discern and respond to reasons. Making this concession is compatible with a consequentialist account of restrictions against killing innocents. First, there cannot be minimal restriction cases where A’s refusal to kill an innocent person makes it the case that B has decisive reason to kill an innocent person, since potential restriction cases of this kind always involve further morally relevant factors. There are not many candidates for situations where an agent could have decisive reasons to kill an innocent person. The most promising seem to be cases where the person who gets killed nonculpably threatens the life of the agent (or of someone else who is innocent) in a way that can be averted only by killing, situations involving voluntary euthanasia, and situations where killing an innocent person is the only way to prevent a catastrophic event. If A’s refusal to kill an innocent person puts B in a situation where B has decisive reasons to kill an innocent person, then A’s situation cannot be a minimal restriction case, due to the presence of further relevant factors over and above the fact that B will kill an innocent person unless A kills an innocent person. Second, consequentialists also have significant resources for handling those potential restriction cases where A’s refusal to kill an innocent person would give B decisive reason to kill an innocent person. For if B kills an innocent person and has decisive reason for doing so, then B either kills in self-​or other-​defense, or performs an act of voluntary euthanasia, or prevents a catastrophe by her act of killing. If A’s act of killing does not also have one of these properties in the potential restriction case under consideration, then B’s act of killing has a morally relevant property that A’s act lacks. It is plausible to assume that in this case, B’s act of killing is not as bad as the act of killing that A needs to perform in order to not bring about B’s killing, precisely in virtue of the fact that only B kills in defense, performs an act of voluntary euthanasia, or does what is necessary to prevent a catastrophe. Consequentialists can thus explain why it would be wrong of A to kill in such cases. However, if A’s and B’s acts of killing share one of the aforementioned relevant properties, then, at least other things being equal, it is plausible to assume that A’s killing is not wrong, due to reasons that are independent of the fact that B will kill unless A kills. Consequentialists can accommodate this view as well. Hence, a consequentialist account of restrictions can be extended to those situations which are such that A’s refusal to kill would bring about B’s killing by giving B decisive reasons to kill, and which are also covered by a nonminimal restriction against killing innocents.



The Restrictions of Consequentialism    153

7.4.4. Taking  Stock

It seems to me that if we identify the outcome of an action with everything that the agent brings about in the ordinary sense, then our conception of outcomes will qualify as agent-​centered. According to the resulting view, the only person who can bring about those voluntary acts of A for which A has reasonable alternatives is A herself. As the preceding paragraphs indicate, this conception of outcomes allows consequentialists to accommodate the minimal restriction against killing innocents, and also some important extensions of it, due to the following facts: First, there are no minimal restriction cases where A brings about B’s act of killing by not killing someone herself. What A can bring about by refusing to kill in a minimal restriction case is her refusal to kill, an innocent person’s death, and her act of letting an innocent person die, but not B’s killing. Second, regarding those potential restriction cases where A does bring about B’s killing by not killing someone herself, consequentialists have resources to either account for the view that A’s killing is wrong in such cases, or to argue that these cases are in fact not covered by a restriction. 7.4.5. Excluding Too Much?

In light of these results, one might now worry that the agent-​centered understanding of outcomes excludes too much. For even if one does not want to accept those far-​ reaching assumptions about negative responsibility that consequentialist views are often taken to imply, it seems that the results of other persons’ actions can matter for what one ought to do even in those cases that the agent-​centered understanding of outcomes suggested here does not address.31 For example, if I refuse to meet a blackmailer’s unimportant request and the blackmailer kills someone in response to my refusal, then it seems that even if by refusing I did not give the blackmailer decisive reason to act in that way, the victim’s death is still relevant to the moral assessment of my refusal, and furthermore makes it wrong for me to refuse.32 Can consequentialists make sense of this even if they deny that I bring about the blackmailer’s act of killing in such a case?

31 More generally, it might seem that the approach suggested here implies that outcomes of B’s ϕ-​ing cannot also be outcomes of what A does in those cases where (i) A does not bring about B’s ϕ-​ing, (ii) A’s act does not enable ϕ’s success, and (iii) ϕ is not the upshot of A and B acting together in a way that involves a shared intention. Such a view would closely resemble Alan Gewirth’s contentious principle of intervening action (Human Rights, 183, 229–​230) or the proximate cause criterion for ascriptions of causal responsibility in the context of law (cf. Wright, “Causation”). 32 I borrow this example from Driver, Consequentialism,  62–​63.

154    Jan Gertken

I see two possible ways to do this. First, consequentialists might argue that although I do not bring about the death of the victim by my refusal in the blackmail case, I nevertheless let that person die (and bring about that act). We can meaningfully ascribe negative value to such acts, and we need not assume that I also bring about the death of the person I let die for this to make sense. Since this view is furthermore compatible with assuming that the victim’s death counterfactually depends on my refusal, it does not commit us to the implausible assumption that I can let someone die even in those cases where my decision plays no role whatsoever in his death. Second, just because the blackmailer’s act of killing is excluded from the things I bring about by refusing, it does not follow that everything the blackmailer brings about by killing is excluded as well. Everything that is brought about by my ϕ-​ing is brought about given some set of background conditions that are not brought about by my ϕ-​ing (and that in many cases cannot even be brought about by any action). Thus, even when another agent’s action is not an outcome of what I do, it can make sense to treat that agent’s action as part of the background conditions against which it is determined what I  bring about. One could employ a different standard for actions, i.e., for intentional and potentially autonomous and rational reactions, and for everything else when it comes to determining outcomes. Accordingly, the conception of outcomes suggested here allows one to merely cut out the middle man and to count the death of the victim in the blackmail case as an outcome of my refusal without having to treat the blackmailer’s act of killing as something that I bring about as well.

7.5. Upshot

In this chapter, I have argued that it is not an agent-​relative theory of value, but rather an agent-​centered conception of outcomes that provides the most promising option for accommodating deontic restrictions into a consequentialist outlook. I have furthermore sketched how this might be achieved regarding one specific type of restriction, relying on certain normative background assumptions and on certain intuitions about our everyday notion of bringing something about. Much more needs to be done to determine whether a fully general, sufficiently precise and satisfying consequentialist account of restrictions can be developed in this way. Personally, I am skeptical that at the end of the day, such an account will be as appealing as the most important nonconsequentialist alternatives. However, I  hope to have shown that an agent-​centered conception of outcomes provides consequentialists at least with important new leeway for accommodating restrictions. This might not lead to a very



The Restrictions of Consequentialism    155

good account of deontic restrictions, but it might well lead to the best account that consequentialists can give.33 References Dancy, Jonathan. Ethics without Principles. Oxford: Clarendon Press, 2004. Dreier, James. “Structures of Normative Theories.” The Monist 76 (1993): 22–​40. Dreier, Jamie. “In Defense of Consequentializing.” In Oxford Studies in Normative Ethics, vol. 1, edited by Mark Timmons, 97–​119. Oxford: Oxford University Press, 2011. Driver, Julia. Consequentialism. London: Routledge, 2012. Emet, Stephen. “Agent-​Relative Restrictions and Agent-​Relative Value.” Journal of Ethics and Social Philosophy 4 (2010): 1–​13. Gewirth, Alan. Human Rights: Essays on Justification and Applications. Chicago: University of Chicago Press, 1982. Heuer, Ulrike. “The Paradox of Deontology, Revisited.” In Oxford Studies in Normative Ethics, vol. 1, edited by Mark Timmons, 236–​267. Oxford: Oxford University Press, 2011. Hurley, Paul. Beyond Consequentialism. Oxford: Oxford University Press, 2009. Kagan, Shelly. The Limits of Morality. Oxford: Clarendon Press, 1989. Kamm, Frances. Morality, Mortality. Vol. 2. Oxford: Oxford University Press, 1996. Nozick, Robert. Anarchy, State, and Utopia. New York: Basic Books, 1974. Portmore, Douglas W. “Combining Teleological Ethics with Evaluator Relativism: A Promising Result.” Pacific Philosophical Quarterly 86 (2005): 95–​113. Portmore, Douglas W. “Consequentializing Moral Theories.” Pacific Philosophical Quarterly 88 (2007): 39–​73. Portmore, Douglas W. Commonsense Consequentialism. Oxford: Oxford University Press, 2011. Rabinowicz, Wlodek, and Toni Rønnow-​Rasmussen. “The Strike of the Demon: On Fitting Pro-​ Attitudes and Value.” Ethics 114 (2004): 391–​423. Sachs, Benjamin. “Consequentialism’s Double-​Edged Sword.” Utilitas 22 (2010): 258–​271. Scheffler, Samuel. “Agent-​ Centred Restrictions, Rationality, and the Virtues.” Mind 94 (1985): 409–​419. Schroeder, Mark. “Teleology, Agent-​Relative Value, and ‘Good.’” Ethics 117 (2007): 265–​295. Sen, Amartya. “Evaluator Relativity and Consequential Evaluation.” Philosophy and Public Affairs 12 (1983): 113–​132. Smith, Michael. “Two Kinds of Consequentialism.” Philosophical Issues 19 (2009): 257–​272. Sosa, David. “Consequences of Consequentialism.” Mind 102 (1993): 101–​122. Suikkanen, Jussi. “Consequentialism, Constraints and the Good-​Relative-​To: A Reply to Mark Schroeder.” Journal of Ethics and Social Philosophy 3 (2008): 1–​8. Wright, Richard W. “Causation in Tort Law.” California Law Review 73 (1985): 1737–​1828.

33 Many thanks to Monika Betzler, Benjamin Kiesewetter, Thomas Schmidt, Jörg Schroth, Christian Seidel, and the participants of the workshop “Consequentialism: New Directions, New Problems?” (held at Friedrich-​ Alexander-​Universität Erlangen-​Nürnberg) for very helpful comments on earlier versions of this paper.

8 CO M M O NS ENS E CONSE QUE NT IA LISM, M O R A L R AT I O N A L I S M , AN D O PT I O NS

Dale Dorsey

Classical consequentialism has a lot going for it. It seems to answer to deep-​seated beliefs about the impartiality of morality, the importance of benefit, harm, and happiness, and to the commonsense thought that an act couldn’t be wrong if it resulted in the best possible state of affairs. Despite these points in favor, however, most people have been tempted to reject classical consequentialism for its failure to capture two commonsense features of the moral landscape, viz., options and constraints: options to refrain from promoting the best possible state of affairs, and moral constraints not to perform various act-​types, even if to do so would result in a better state of affairs.1 However, or so many think, the death of classical consequentialism need not yield the death of consequentialism. In recent years a new variety of consequentialist views have sprung up—​what I shall refer to, after Portmore, as “commonsense consequentialism”—​that seek to embrace the importance of options and constraints while retaining the consequentialist focus on the outcomes of our actions. However, commonsense consequentialism seems, at least on its face, a bit puzzling. First, why, if we’re going to be consequentialists, should we accept the principal features of commonsense moral theory, and in so doing give up the traditional and 1 For the loci classici of the discussion of options and constraints, see Scheffler, The Rejection of Consequentialism; Kagan, The Limits of Morality.

156



Consequentialism, Moral Rationalism, and Options    157

plausible rationales for the classical view? Second, if we’re going to be commonsense moral theorists, why should we accept a consequentialist version of constraints and options, rather than simply holding that not all of our actions are evaluated given the quality of their consequences? Now, that’s not to say that there are no answers to these questions. But I’m going to explore one broad answer here, viz., a commitment to moral rationalism, or the suggestion that if a person is morally required to perform some action ϕ, he or she is thereby normatively, or as a matter of practical rationality, required to ϕ. In this paper, I argue that moral rationalism cannot provide a rationale for commonsense consequentialism. This is because moral rationalism is probably false. My critique of commonsense consequentialism is modest, for two reasons. First, I’m going to ignore the rationale for a commonsense morality that appeals strictly to agent-​centered constraints. Though I have criticized these elsewhere,2 I’m simply going to focus on options here. And hence there may be room for a commonsense consequentialism that is merely constraint-​accepting rather than option-​accepting, given everything I  say. Also, as I’ve intimated, because I  argue that moral rationalism—​upon which commonsense consequentialism rests—​is only probably false, I can conclude only, at best, that the rationales for commonsense consequentialism probably fail, nothing stronger. 8.1. The “Commonsense” Bit

My critical examination of commonsense consequentialism begins with the commonsense bit. To this end, let’s focus our discussion by saying a little more about classical consequentialism, against which commonsense moralities are generally contrasted. According to classical consequentialism, at least as I  understand it, moral agents are morally required to act in such a way that no other action would have produced a better state of affairs, where “betterness,” here, is construed impartially or impersonally. Stated in this way, classical consequentialism includes utilitarianism, telic egalitarianism,3 prioritarianism,4 and so forth. Perhaps the most notable feature of classical consequentialism is its insistence that individuals act in a way that would produce the best results impartially construed. However, many have felt that this requirement is too strong. Certainly we would like to allow individuals some latitude to pursue their own projects, aims, and to promote the interests of their closest loved ones, etc., in a way that would be typically 2 Dorsey, “A Challenge for Constraints”; see also Jan Gertken’s contribution to this volume. 3 Parfit, “Equality or Priority?” 4 Weirich, “Utility Tempered with Equality.”

158    Dale Dorsey

disallowed by an injunction to always promote the impartially best state of affairs. And thus we arrive at the commonsense commitment to moral options. Generally speaking, or so I hereby stipulate, options are moral permissions not to promote the impartially best state of affairs—​they provide a justification for refusing to act as traditional consequentialists would require but instead as dictated by, for instance, one’s own interests or the interests of one’s loved ones in particular. Of course, that we should accept such options is not obvious. Many people think that moral reasons ought to be impartial in content in the way dictated by classical consequentialism. According to Samuel Scheffler, for instance, the thought that “[t]‌he moral point of view, according to this strand of thought, is a standpoint that one attains by renouncing any distinctive attachment to oneself, and by acting instead from a thoroughly selfless concern for all,” is an “important strand in our substantive thinking about morality.”5 David Brink says something similar: “It is, of course, a substantive moral claim that it is this kind of impartiality which is characteristic of the moral point of view, but it is a plausible claim which cannot be rejected lightly.”6 But given that options not to conform to the impartially best state of affairs are, naturally enough, moral permissions to take up not an impartial point of view but a partial point of view (one more heavily weighing one’s own interests and the interests of the near and dear) it would appear that to justify options we must have a plausible rationale to do so, at least as strong as the considered judgment in favor of moral impartiality, captured by classical consequentialism. And insofar as commonsense consequentialism relies on the acceptance of options, this rationale will be crucial to justifying the commonsense bit. To begin, take the following suggestion by Peter Railton: “[W]‌e must recognize that loving relationships, friendships, group loyalties, and spontaneous actions are among the most important contributors to whatever it is that makes life worthwhile; any moral theory deserving serious consideration must itself give them serious consideration. As William K. Frankena has written, ‘Morality is made for man, not man for morality.’ ”7 Scheffler says something very similar: [Consequentialism] singles out the impersonal calculus as identifying the right course of action for the individual, no matter how his own projects and plans may have fared at the hands of that calculus, and despite the fact that from the impersonal standpoint his own deepest concerns and commitments have no

5 Scheffler, Human Morality, 120. 6 Brink, “Utilitarian Morality,” 432. 7 Railton, “Alienation, Consequentialism, and the Demands of Morality,” 155. The Frankena passage is taken from Ethics, 116.



Consequentialism, Moral Rationalism, and Options    159

distinctive claim to attention. In other words, it requires that he devote energy to his projects in strict proportion to the weight from the impersonal standpoint of his doing so. . . . [I]‌t is . . . a decisive objection to consequentialism that it includes the requirement. For if, as Rawls has said, “the correct regulative principle for anything depends on the nature of that thing,” we must surely reject any regulative principle for persons which ignores the independence of the personal point of view.8 Railton and Scheffler, here, appear to be picking up on very similar thoughts from Frankena and Rawls. These four thinkers seem to agree that morality as an institution ought to be “made for” actual men and women, or ought to be specifically responsive to the way people really are—​the “nature” of human beings as we find them. For Scheffler, this means that morality must grant significance not just to the impersonal, but also to the personal, point of view. For Railton, this means that morality “must recognize” the significance of individuals’ personal points of view.9 Of course, the natural question is: What does it mean to say that morality should be made for human beings like us? And why does respecting the thought that morality should be made for folks like us lead to an acceptance of moral options? After all, classical consequentialism accepts “ ‘ought’ implies ‘can’ ”: no action would be morally required of an agent that that agent could not perform. Given this, there is a perfectly good sense in which classical consequentialism respects the nature of real human agents: it doesn’t require any action of any real person that is beyond the capabilities of that person.10 And hence if the thought in question is to motivate moral options in particular, it would have to be interpreted in a stronger way than the simple insistence upon “ ‘ought’ implies ‘can.’ ” But what is this sense? Scheffler and Railton (via Rawls and Frankena), I think, are picking up on the thought that while no morality should ask of someone what he or she cannot do, it is also the case that no morality should ask of someone that which it is unreasonable to ask of him or her. In taking human beings as they are, in respecting their nature, we are not just respecting their basic limitations, but we 8 Scheffler, The Rejection of Consequentialism, 57–​58. The Rawls citation is from A Theory of Justice, 29. 9 Of course, Railton argues that the significance of an individual’s personal point of view can be adequately accommodated by a classical consequentialist moral theory, but I’ll leave that point aside here. For the sake of argument, I’m granting the commonsense moral theorist (including the commonsense consequentialist) the suggestion that the significance of one’s personal point of view necessitates options. 10 Of course, there is a further literature on this claim that seeks to show that consequentialists may have problems given the existence of humdrum cognitive limitations. See, for instance, Miller, “Actual Consequence Act-​ Utilitarianism”; Wiland, “Monkeys, Typewriters, and Objective Consequentialism”; Howard-​Snyder, “The Rejection of Objective Consequentialism.”; Dorsey, “Consequentialism, Cognitive Limitations, and Moral Theory.” I’ll leave aside this important debate here.

160    Dale Dorsey

are also respecting what they reasonably care about. But why should this be? Why should we think that morality in particular should not ask of people what is unreasonable to ask of them? The answer, or so it seems, is that morality purports, as noted by Scheffler, to be the “regulative principle” of human beings. As Railton says, “Moral considerations are often supposed to be overriding in practical reasoning.”11 Thus it would appear that this line of investigation has led us to the following commitment: Moral Rationalism: If x is morally required to ϕ at t, x is normatively required to ϕ at t. A word of conceptual clarification:  To be normatively required to do something means, more or less, that one is all-​things-​considered required to do it, one ought to do it, practical reasons tell decisively in favor of it, it forms the full account of how one ought to live, and so forth. Moral rationalism then claims that moral requirements in particular are those to which agents ought to conform—​if morality tells me I should perform some action, I ought to perform that action, whether or not it is in my self-​ interest, in the interests of the near and dear, and so forth. If moral rationalism is true, or so the thought goes, it’s hard to see how a classical consequentialism (or indeed any options-​denying moral theory) could survive. After all, it is surely the case that a person’s deepest concerns and commitments, whatever their moral significance, are extremely normatively significant. They provide reasons of substantial weight and importance, reasons for action that we all recognize. If this is true, then without options to allow individuals to conform to their deepest concerns and commitments, moral requirements don’t have a prayer of yielding decisive normative requirements, at least not in all cases.12 Thus it would seem that a commitment to moral rationalism is sufficient to defend moral options. But there’s good reason to believe it’s also necessary. Of course, there’s no conceptual incoherence in saying that one could accept moral options without moral rationalism.13 However, while there may be nothing incoherent about accepting options and denying moral rationalism, there’s good reason to believe that without moral rationalism moral options should be rejected. Elsewhere, I have offered a number of arguments for this conclusion; I’ll summarize three of them here.14 First, take the following example. Imagine that you crash a dinner party attended by admissions personnel at top universities; your purpose Railton, “Alienation, Consequentialism, and the Demands of Morality,” 155. 12 See, for instance, Portmore, Commonsense Consequentialism,  27–​28. 13 See, for instance, Scheffler, “Morality’s Demands and Their Limits,” 536. 14 The first and second of these arguments have appeared in somewhat different forms in Dorsey, “Weak Anti-​ Rationalism and the Demands of Morality” and “How Not to Argue against Consequentialism.” 11



Consequentialism, Moral Rationalism, and Options    161

in so doing is to persuade these individuals that your daughter is a strong student worthy of admission, but has been unfairly overlooked because she attended a rural high school that failed to develop the same advanced curriculum as that of her competitors at schools in wealthier and more urban areas. You spend the entire dinner pestering person after person, interrupting conversations, moving around the table, and generally behaving like a boor, never mind the fact that you weren’t invited to begin with. Nevertheless, as you knew it would, your behavior results in your daughter getting a second look, and being admitted—​rightly—​into a top university. In this case, it seems to me, there is every reason to believe that in promoting your daughter’s interests you behaved rudely, out of congruence with the demands of etiquette. But would we object to this first-​order theory of etiquette, simply on the grounds that it fails to allow etiquette-​compatible options to promote the interests of the near and dear? Certainly not! We seem to feel no pressure to accept a theory of “etiquette options”—​options to more heavily weigh one’s own interests and the interests of the near and dear in comparison to other paradigmatically significant concerns of etiquette. The same goes for prudence. It would appear quite true to say, for instance, that sacrificing one’s own well-​being for the sake of the interests of one’s child is imprudent—​there is no prudential option to more heavily weigh the interests of the near and dear against paradigmatically significant prudential factors. But why is this? The answer, or so it seems to me, is that it is not the case that one fails normatively whenever one fails to act in one’s own interests or in conformity with the demands of etiquette. And hence the explanation of the existence of moral options, rather than, e.g., etiquette or prudential options, seems to rely on this distinctive feature of morality, not shared by etiquette and prudence: one ought always to conform to moral demands. Second, without a commitment to moral rationalism, the allowance of specifically moral options is antiaccommodationist. To see this, let’s assume that we have a very strong considered judgment that it is morally important for individuals to have the option (at least) to follow through on their personal commitments from the perspective of the moral domain (call this considered judgment Options). Assuming that we treat the truth of moral rationalism as negotiable, we can interpret Options as a judgment not about morality but instead strictly about normativity or the content of practical reasons. And in so doing, we can accommodate not only Options (albeit with a slightly shifted emphasis), but also the general impartiality of the moral point of view, which seems, as discussed already, a central considered judgment that stands behind classical, rather than commonsense, consequentialism. This would seem preferable to simply giving up the, e.g., impartiality of the moral domain. The only way to avoid this accommodationist reinterpretation is to hold that such a maneuver gives up something important: moral rationalism.

162    Dale Dorsey

Third, and finally, consider that options can come in different shapes and sizes. Surely, for instance, we would not accept a view according to which an individual has the option to massacre an entire civilization for the sake of a mild chuckle for her daughter (call this the “massacre is OK view”). This account of options is surely too strong. But also consider the suggestion that one is justified in granting moral priority to one’s children, but only by a factor of one additional utile. Call this the “one utile view.” This account of options is surely too weak. But why? Why are we interested in allowing a stronger account of options than the one utile view? The answer, I  think, is this. Although it allows greater partiality than classical consequentialism, the one utile view continues to deliver the result that the vast amount of attention one spends on oneself and one’s loved ones is morally unjustified. But if this is correct, the benefit in terms of allowable partiality granted by the one utile view is not enough to outweigh the traditional rationale for classical consequentialism. If we’re only going to accept the one utile view, we may as well just opt for classical consequentialism. Thus we need an account of moral options weaker than the massacre is OK view, and stronger than the one utile view. We need, in other words, some threshold at which moral options become strong enough to justify a view that cannot accommodate the general plausibility of an impartial moral theory. Plausibly, the threshold can and should be set at the point at which moral demands become unreasonable. If we set the threshold here not only do we do so in a principled way, but we now have the power to accept moral rationalism, which seems to make up for the loss of impartiality. If this line of reasoning is correct, and I think it is, this once again shows that options rely on an assumption of moral rationalism. After all, the only version of moral options that would plausibly justify a rejection of classical consequentialism is one that explicitly guarantees that morality demands only what is reasonable to demand. But if moral rationalism is not true, i.e., if there’s no reason to expect that every moral requirement will issue demands that are “reasonable,” the point of setting the threshold at the “reasonable” seems to lose its rationale. Given everything that’s been argued here, I think we’re licensed to conclude that commonsense consequentialism, to defend its “commonsense” bit, not only does rely on moral rationalism, but must. 8.2. The “Consequentialism” Bit

Let’s say for a minute that we should accept moral options of the kind endorsed by commonsense consequentialism. What should the resulting options-​accepting moral theory look like? Here’s one possibility. Generally speaking, individuals face a very strong moral reason, or set of moral reasons, to perform whatever action that



Consequentialism, Moral Rationalism, and Options    163

brings about the best resulting state(s) of affairs. But individuals also sometimes have sufficient moral reason to, say, conform to prudential norms, or to care for their loved ones in a way that does not advance the quality of the resulting outcome (i.e., to exercise moral options). In other words, people have a “background” moral reason that conforms to the classical consequentialist picture. But this reason is not always decisive. Sometimes people have sufficient moral reason—​options—​not to promote the best state of affairs. Call this picture “commonsense nonconsequentialism.” On the other hand, commonsense consequentialism holds that the significance of one’s own interests and the interests of the near and dear must be given a consequentialist interpretation—​it must be, strictly speaking, the outcome of one’s action that justifies one’s refusal to conform to the demands of a more traditional consequentialist view. But, and here’s the question, why—​if we’re going to be commonsense moral theorists—​should we accept a consequentialist interpretation, rather than the more, dare I say, traditional nonconsequentialist interpretation? One possible answer is that in defending options, one is trivially evaluating actions on consequentialist grounds. To more heavily weigh one’s own welfare or the interests of the near and dear and so forth seems to be a straightforwardly consequentialist criterion of evaluation, viz., the extent to which a particular action, which one has an option to perform, does or does not end up benefiting the interests of the near and dear, etc. However, it’s unclear that this rationale can succeed in all cases. Surely there are some actions one has the option to perform that cannot so easily be fit into a consequentialist analysis. Perhaps one has the option simply to, for instance, tell the truth to one’s spouse when guilty of infidelity because doing so is important in interpersonal norms, in a way that isn’t justified by the outcome of one’s so doing (that is, outcome for one’s welfare or one’s spouse’s welfare, and so on).15 Furthermore, though the following argument is to some degree impressionistic, this question becomes all the more important because commonsense consequentialism cannot straightforwardly latch onto the traditional considered judgment(s) that seem to stand behind a consequentialist moral view. And here I’m making specific reference to the following passage from Foot: “What is it, let us now ask, that is so compelling about consequentialism? It is, I think, the rather simple thought that it can never be right to prefer a worse state of affairs to a better.”16 Why is it that commonsense consequentialism cannot make use of this rationale? Two reasons. First, given the acceptance of options, each agent must have a different rank-​ ordering of states of affairs. But then the notion of “bestness” or “goodness” that 15 This needn’t yield a moral constraint, but rather an option to conform to the specific norms of one’s marriage which are important to you for various reasons. 16 Foot, “Utilitarianism and the Virtues,” 198.

164    Dale Dorsey

underlies Foot’s compelling thought would seem to be interpreted as something like “good-​relative-​to” the agent in question. But as Mark Schroeder has argued,17 it is quite doubtful that this notion expresses anything like the standard “good” relation that motivates the compelling thought stated by Foot.18 Of course, this might be contested, but even if one can overcome Schroeder’s skepticism, it is surely plausible to say that Foot’s compelling idea can be trivially restated by the claim that the “peculiar attraction” of consequentialism is that it can never be right to prefer an impartially worse state of affairs to a better. Second, even if we ignore this point, the acceptance of options seems to straightforwardly imply that whatever “betterness” relation is being used by commonsense consequentialism cannot be the straightforward one that seems to stand behind classical views. This is because, options being options, one will typically also have the moral option to perform the action that is impersonally best. Call this rank-​ordering “betternessa.” But, options being options, one also will have the option to perform an action that is highly ranked, not in terms of betternessa, but in terms of some other ranking, “betternessb,” which will more heavily weigh the interests of the near and dear, and so forth. Here it is clear that even if we evaluate outcomes in terms of their betterness, according to commonsense consequentialism, one must have moral permission not only to perform the action that is besta, but also that is, at least, bestb. And hence deontic status cannot simply be a product of a straightforward impersonal betterness relation.19 But even if we ignore this point, it remains worth wondering what rationale commonsense consequentialism can offer for the “consequentialist” bit. For Portmore, we begin with the adoption of a particular theory of practical reasons, viz., the teleological theory:  “On this view, an agent’s reasons for performing an action are determined by her reasons for preferring its outcome to those of the available alternatives, such that, if S has most reason to perform x, then, of all the outcomes that S could bring about, S has most reason to desire that x’s outcome obtains.”20 Assume this holds. But the teleological theory of practical reasons does not yet yield a consequentialist moral theory. The missing component is moral

Schroeder, “Teleology, Agent-​Relative Value, and ‘Good.’ ” 18 For a detailed argument to this effect, see Schroeder, “Teleology, Agent-​Relative Value, and ‘Good,’ ” 279–​ 287; Monika Betzler and Jörg Schroth’s contribution to this volume (section 6.4). Portmore seems to accept this point in holding that commonsense consequentialism should evaluate states of affairs not on grounds of betterness or their desirability, but instead on the extent to which an agent has reasons to desire that particular outcome. Cf. Portmore, Commonsense Consequentialism, 6. See also 34–​38. 19 Indeed, Portmore agrees. Portmore defends a “dual ranking act-​consequentialism,” according to which one has moral permission to ϕ if and only if there is no other action that is both morally better and all-​things-​ considered better for one to perform (Commonsense Consequentialism, ch. 5). 20 Portmore, Commonsense Consequentialism, 33. 17



Consequentialism, Moral Rationalism, and Options    165

rationalism.21 The teleological theory of practical reasons would require a consequentialist moral theory on the assumption that we have decisive practical reason to conform to moral requirements. Were morality not teleological in structure, there could be some moral requirements that required an action that did not produce a favored set of consequences, and hence, given the teleological theory of practical reasons, we would lack decisive reason to act in that way. Indeed, I think it is no accident that commonsense consequentialism makes use of moral rationalism in defending its consequentialism bit. If, for instance, we accept Schroeder’s skepticism, it’s hard to see what sort of purely internal rationale—​ internal, that is, to morality—​the commonsense consequentialist might offer for its consequentialism once we give up on the generally plausible thought that, morally speaking, it is right to act for the (impersonally) best. Indeed, this is especially pressing insofar as commonsense nonconsequentialism can pay substantial lip serv­ ice to this intuition. For commonsense nonconsequentialism states of affairs are rank-​ordered agent-​neutrally—​according to their (impersonal) betterness, and we have very strong reasons to promote the (impersonally) best state of affairs. It just so happens that we have options, on occasion, not to produce the best state of affairs (given our prudential interests, the interests of the near and dear, etc.). But even if Schroeder’s skepticism can be overcome, it remains the case that commonsense consequentialism seems to rely in fact on moral rationalism in two distinct and important ways. First, it relies upon it in defense of its commonsense, as opposed to classical consequentialism. In defending the importance of moral options, moral rationalism is a key assumption. Without moral rationalism, it is difficult to see why we should accept moral options to conform to partial reasons. Second, commonsense consequentialism relies on moral rationalism in defense of its commonsense consequentialism, as opposed to commonsense nonconsequentialism. Thus if moral rationalism is false, commonsense consequentialism is in real trouble. And I think it’s probably false. 8.3. Why Moral Rationalism Is Probably False

The argument goes, roughly, like this. First, moral rationalism is not a priori: in other words, moral rationalism cannot be established as true or false independently of substantive first-​order investigation into the content of moral reasons, requirements, and so forth. To show that moral rationalism is true, one must show, as a result of substantive inquiry (at least) into the content of the moral domain, that moral requirements Portmore, Commonsense Consequentialism,  33–​34. 21

166    Dale Dorsey

are those that we have decisive reason to perform. Thus whether moral rationalism is true depends—​at least—​on the existence of independent arguments for moral options. After all, as we saw in section 8.1, not to countenance options is to offer a moral theory incompatible with moral rationalism. But because arguments for moral options depend on the presumption of moral rationalism, these arguments are not forthcoming. And hence, or so I claim, moral rationalism is probably false: we have no rationale to accept specifically moral options as required to substantiate moral rationalism. A big caveat:  I am not here discussing the interesting and important argument offered by Portmore for moral rationalism (indeed, a priori moral rationalism),22 though I have done so elsewhere.23 Suffice it to say, however, that I consider a priori rationalism to result in a very substantial problem; one can, then, take the problem I note here as an attempt at a reductio of any a priori argument for moral rationalism. 8.3.1. Moral Rationalism Is Not A Priori 24

The first stage in my argument seeks to show that moral rationalism cannot be established independently of first-​order inquiry into the content of moral and rational demands. As I shall state it here, moral rationalism is not a priori. Broadly speaking, I argue that a priori rationalism must grant the wrong epistemic status to important, and controversial, moral theses. To show this, I present three cases in which it seems right to say that a certain course of action is normatively justified not on the basis of specifically moral concerns, but rather on the basis of extramoral concerns, that is, facts that are paradigmatically relevant from nonmoral domains, including protocol, prudence, and aesthetics.25 If my take on these cases is right, the moral rationalist must insist that extramoral concerns play a role in the determination of the acts one is morally justified in performing (given that, according to moral rationalism, moral requirement entails rational requirement it must also hold that being rationally justified entails not failing to live up to one’s moral requirements, i.e., being morally justified). But this, in itself, is not the source of my objection. Instead, I hold that to allow that moral rationalism is a priori requires one to hold that the fact that extramoral concerns play a role in moral inquiry is itself a priori, which is an unacceptable account of the epistemic status of this claim. Most important, see Portmore, Commonsense Consequentialism,  43–​44. 23 See Dorsey, “How Not to Argue against Consequentialism.” 24 This section was the basis for pp. 60–65 of my The Limits of Moral Authority (which beat this paper to the press). I like to thank OUP for permission to include this here. 25 Note: to say that a particular practical reason is extramoral is not to say that this concern couldn’t play a role in the moral domain. It is just to say that this concern plays a paradigmatic role from nonmoral domains. My self-​interest, for instance, is an extramoral concern. This isn’t to say that it’s nonmoral, but just that it plays a paradigmatic role in the determination of the prudential valence of actions. 22



Consequentialism, Moral Rationalism, and Options    167

Take, first: Sarah: Sarah stands before the Queen. A few weeks ago, she promised a friend that, as a political statement, she would speak to the Queen before the Queen speaks to her and, furthermore, would address the Queen by saying “Hey there, Queensie!” This is a genuine promise that her friend regards as an important pact, and a feature of their joint political strategy. Furthermore, if Sarah does not break protocol, this will generate slightly less happiness—​just slightly—​ than were she to break protocol (imagine that though no one will be pained if she breaks protocol, her friend will be angry if she does not). As Sarah stands before the Queen, she is faced with a question:  Do I  keep my promise, or do I conform to protocol? I find it absurd to suggest that in this case Sarah behaves in a normatively unjustified manner in conforming to the demands of protocol. There appear to be a range of practical considerations that protocol, in this case, reflects: basic politeness, a due deference to VIPs, an interest in “fitting in” or “doing what’s done,” living up to the expectations of those around you, and perhaps other considerations. But these factors are extramoral, i.e., are paradigmatically considerations relevant from the domain of etiquette or protocol.26 Perhaps you don’t feel the force of Sarah’s case. Take another: Andrea: Andrea is deciding whether to attend Eastern Private College or Local Big State University. To attend Eastern Private, Andrea would have to travel halfway across the country and would get to see her family only rarely. However, Andrea’s family has undergone a series of tremendous hardships, including the death of Andrea’s younger sibling, which devastated her parents. If she were to attend LBSU, Andrea could live at home and successfully tend to her parents’ emotional needs, which is clearly essential for their well-​being, at very little additional cost in time or energy. Nevertheless, it is important to Andrea, simply for her own sake, to go to Eastern Private. (Assume Andrea’s future prospects would not be hampered in any significant way by staying at home.) Imagine that Andrea chooses to go to Eastern Private rather than LBSU. Given the description of the case, I find it very plausible to say that Andrea is justified in moving away to attend Eastern Private. She is justified in doing so, it seems to me, because it is important to Andrea to move away; she has a prudential interest in doing so. It is

Thanks to David Sobel for challenging comments on this score. 26

168    Dale Dorsey

important to her, for her own sake, to go to Eastern Private. This prudential interest, or so it would seem, is itself sufficient to justify doing so. Finally, take: Fred Astaire:  In Blue Skies (1946), an otherwise forgettable Bing Crosby vehicle, Fred Astaire recorded what was to become his most iconic dance performance (“Puttin’ on the Ritz”). The product, according to Astaire, of “five weeks of backbreaking physical work,” it was intended by Astaire to be his final, and crowning, achievement.27 It is a truly astounding demonstration of Astaire’s considerable talent.28 This case is underdescribed, as I’m not privy to the various circumstances surrounding Astaire’s performance that would or would not deliver a verdict that it is morally justified. But we can certainly imagine: perhaps Astaire had made a promise, long ago, to his wife Phyllis to retire from film to spend more time with his family. But this promise had been broken again and again—​including to participate in Blue Skies. The sequence of broken promises took a toll on his marriage and on his wife and children, in part but certainly not only because his broken promises caused a gradual erosion of the feeling of security and trust they had in him.29 But even if we’re willing to say this, it seems wrong to say that Astaire wasn’t justified in putting this brilliant performance on film. And it is justified, so far as I can tell, because the performance, of itself, has tremendous value qua performance. What do these cases show? Here’s one thought. It would appear that we are perfectly willing to commit to the normative justification of action on the basis of concerns that are extramoral, i.e., paradigmatically relevant from the perspective of nonmoral domains. This is especially clear given that moral factors themselves tell at least at first glance against actions that Fred, Sarah, and Andrea are rationally justified in performing, given the cases as described. But by itself, merely pointing this out doesn’t threaten moral rationalism. Here’s why: as noted by Scheffler and 27 Astaire intended to retire from film after Blue Skies, but returned only two years later with Judy Garland in Easter Parade (1948). 28 Bosley Crowther, not known for his effusiveness, wrote of the dance performances in Blue Skies, that “[b]‌est of the lot . . . is Mr. Astaire’s electrifying dance to that ancient and honorable folk-​song, ‘Puttin’ on the Ritz.’ Turned out in striped pants and top hat, Mr. A. makes his educated feet talk a persuasive language that is thrilling to conjugate. The number ends with some process-​screen trickery in which a dozen or so midget Astaires back up the tapping soloist in a beautiful surge of clickety-​clicks. If this film is Mr. A.’s swan song, as he has heartlessly announced it will be, then he has climaxed his many years of hoofing with a properly superlative must-​see” (Crowther, “Blue Skies”). 29 Honestly, this is all made up. From all accounts I can find, Astaire’s first marriage, which ended tragically in 1955 when his wife died of lung cancer, was blissful.



Consequentialism, Moral Rationalism, and Options    169

others, to use the above cases in an argument against moral rationalism seems to presuppose a certain vision of the moral point of view, i.e., that moral obligations cannot take on board other, explicitly extramoral, considerations.30 After all, some might hold that Andrea’s prudential interest in attending Eastern Private renders her decision to do so morally, not just rationally, justified. Here’s another way to put this point. In Sarah’s, Andrea’s, and Fred’s cases, the actions in question are sufficiently normatively justified by the practical significance of extramoral concerns, such as concerns of prudence, etiquette, and aesthetics. But it is perfectly possible to say that the extent to which an action is morally justified is a product not simply of the per se moral considerations—​which seem, in all three cases, to clearly tell against the rationally justified action—​but also of the extramoral considerations, such as the ones that seem sufficient to rationally justify the actions of Sarah, Fred, and Andrea. Hence moral rationalism survives. Simply because some very important moral considerations tell against the actions in question doesn’t mean they’re not morally justified by the extent to which concerns of aesthetics, prudence, and etiquette tell in their favor. To put this another way, moral rationalism is not threatened by the above cases if the following principle is true: Extramoral: Extramoral considerations, including those of prudence, etiquette, and aesthetics, have at least a justificatory role to play in determining the extent to which acts favored by those extramoral considerations are morally justified. One could accommodate Extramoral by one of two mechanisms. One might say, first, that the deontic categories internal to the moral point of view (i.e., moral requirement, moral permissibility, etc.), will necessarily be a function of moral reasons in favor of and against acts so evaluated. For instance, Shelly Kagan writes, “[S]‌ince we are concerned with what is required by morality, the relevant reasons—​whether decisive or not—​must be moral ones.”31 If one accepts this proposal, one could accommodate Extramoral by treating facts that seem paradigmatically extramoral, such as facts about prudential or aesthetic value, say, as genuine moral reasons. Morality cares about such facts. Of course, this might be thought implausible by some. For instance, it is difficult to see that there could be any moral reason to, say, refrain from speaking to the Queen prior to having been spoken to. Furthermore, though there may be some moral value in Fred Astaire’s performance, the facts one points to to defend this claim (i.e., that it causes tremendous happiness, say) are typically distinct from the facts that are most salient in justifying it (i.e., that it is a demonstration of Scheffler, Human Morality, 58–​60. Thanks also to Connie Rosati for pressing this point. 31 Kagan, The Limits of Morality, 66. 30

170    Dale Dorsey

tight coordination, rhythm, synchronization, and syncopation; or, more generally, that it is simply sublime). Of course, this is just a bare intuition, so I leave open this proposal as an option by which to accommodate Extramoral. However, if one wishes to reject the claim that considerations, say, of etiquette will also be genuine moral reasons, there is another mechanism by which to accommodate Extramoral. According to Portmore, nonmoral reasons can possess a form of weight from the moral point of view, i.e., moral justifying strength (which is distinct from strength that counts in favor of a requirement). Nonmoral reasons, therefore, can count in favor of moral justification to ψ rather than ϕ, despite the fact that ϕ-​ing is morally better (i.e., supported by stronger per se moral reasons, reasons with per se moral requiring strength) than ψ-​ing.32 Given this, if actions not supported by moral reasons seem rationally justified, this does not entail that immorality is rationally justified. Rather, nonmoral reasons such as the reason to conform to protocol have the power to morally justify particular actions, even in the face of reasons that have per se moral strength, such as the reason to keep one’s promises, etc. This would seem to defuse any problem about the above cases: given the strength of, say, Fred’s justifying reasons (derived from specifically aesthetic concerns), his performance is morally justified given the weighty nonmoral considerations (and their accompanying moral justifying strength) in its favor.33 Given the dialectic so far, we should refuse to be confident about moral rationalism unless we are confident about Extramoral—​whatever the mechanism we adopt. Because Sarah’s, Andrea’s, and Fred’s actions seem to be sufficiently justified by these considerations (even when confronted with compelling moral considerations against), if Extramoral is false, this is very good evidence against moral rationalism. But, and this is the crucial claim, Extramoral cannot be established a priori. Whether, for instance, the fact that an action conforms to protocol is itself a moral reason to perform it, or whether it is a nonmoral, but nevertheless morally relevant reason to perform it, seems obviously a question to be answered by substantively considering Portmore, Commonsense Consequentialism, 372, 375n12. 33 There is some reason to believe that the first and second mechanisms are not as different as one might at first believe. If, as I do, one wishes to use the term “moral reason” to refer to any consideration that plays a role in the determination of the moral valence of action, then Portmore’s proposal should just be understood to be a version of the claim that extramoral considerations can also be moral reasons (i.e., those that possess only justificatory power). But Portmore understands the notion of a “moral reason” to refer to any consideration that has the power to morally require action or render action morally supererogatory (Commonsense Consequentialism, 123n7). And hence, to suggest that extramoral considerations have merely justifying strength is, for Portmore, not to commit to the claim that such considerations are also moral reasons. But insofar as I consider a fact a moral reason so long as it has any influence on moral evaluation, Portmore’s proposal may be compatible with the first mechanism, simply expressed in different language. Of course, these mechanisms could come apart if the partisan of the first mechanism also believed that extramoral considerations maintain power to render action morally required or supererogatory. 32



Consequentialism, Moral Rationalism, and Options    171

the content of the moral point of view and the extent to which moral requirements are influenced by such considerations. And if this is correct, then to show that moral rationalism is true one cannot simply be neutral concerning the content of the moral point of view. One must establish that considerations of aesthetics, prudence, and protocol play their distinct role from within the moral domain (whether as providing genuine moral reasons, or nonmoral reasons that also possess moral justifying strength). A response is worth considering. The a priori rationalist might claim that we know that Extramoral is correct because we know that moral rationalism is true a priori. In other words, because we’ve established via an a priori argument that moral rationalism is correct (whatever the argument is), we can be confident that Extramoral is correct because, as I say, Extramoral is required for the truth of moral rationalism. After all, if a priori rationalism is true, moral rationalism is a limiting condition on any such first-​order inquiry. And hence we can simply deduce from this fact that if, e.g., aesthetic considerations provide sufficient rational justification in the face of contrary moral considerations, they must also provide sufficient moral justification in the face of contrary moral considerations. But this is precisely the epistemological problem I  seek to expose here. The a priori rationalist appears willing to say that we can know independently of any first-​ order moral inquiry a substantive, controversial, first-​order moral claim, viz., that considerations of aesthetics, etiquette, and prudence play a role in determining what is morally justified for moral agents at a particular time. This can be seen even more clearly if we understand precisely what it means for a particular first-​order moral claim to be a priori. Though, perhaps, we needn’t say that Extramoral is totally unrevisable given subsequent inquiry, to hold that it is a priori we must have a very high degree of confidence.34 In fact, we should at least be willing to say that it is plausible to hold that Extramoral is a limiting condition on first-​order moral inquiry—​that judgments that conflict with Extramoral should simply be ruled out as participants in a genuinely moral reflective equilibrium.35 Put another way, it should be plausible to say that Extramoral is a “gatekeeper” when it comes to first-​ order moral inquiry:  contrary considered judgments should simply be ruled out of hand without a substantive hearing.36 But this is absurd, whatever claims about 34 Thanks to Richard Yetter Chappell for helpful comments here. I discuss a similar point in more detail in my “Moral Distinctiveness.” 35 Of course, this doesn’t mean that they will, in the end, be rejected. All I mean to say is that if Extramoral is a priori it should be at least plausible to say they should be simply ruled out of hand. 36 Notice that to claim that Extramoral is a priori is not equivalent to saying that Extramoral should constrain all normative inquiry. Rather, it’s equivalent to saying that, once we’ve established, e.g., Fred’s normative permission to record “Puttin’ on the Ritz” on the basis of nonmoral concerns, we are committed to accepting

172    Dale Dorsey

Reason-​as-​such we’re already committed to. (Notice that nothing I say here rules out the possibility of a priori claims about the moral domain. But it merely notes the implausibility of holding that Extramoral is such a claim.) And hence we should reject the claim that moral rationalism is a priori.37 8.3.2. Why, If Moral Rationalism Isn’t A Priori, It’s Probably False

Moral rationalism is not a priori. This means that whether moral rationalism is true must await the correct theory of the moral point of view (along with, incidentally, the correct theory of normativity or practical reasons). And hence if moral rationalism is true, it had better be the case that the first-​order moral theory we end up accepting doesn’t issue moral requirements that we lack decisive reason to perform. However, recall the general justification of moral options. A morality “made for” regular folks is one that allows people to conform to their deepest commitments and concerns. Any other morality would fail to issue requirements that are normatively decisive in all cases. And hence to defend moral rationalism, one must defend the existence of moral options. But here’s the problem. Recall that the most plausible rationale for moral options specifically appealed to moral rationalism itself. But because moral rationalism is not a priori, this assumption—​moral rationalism—​must be defended by reference to the content of moral requirements. But whatever moral theory one accepts, this theory, if it is to help defend moral rationalism, must include options. But, as already argued, to defend options, we must again assume moral rationalism. And you see the problem here. We are not allowed to assume that moral rationalism is true prior to the vindication of options. But this entails that one cannot use moral rationalism in a defense of options. Once we recognize that moral rationalism isn’t a priori, so goes the most important argument in favor of moral options. And with it goes a thesis crucially important for the defense of moral rationalism. But actually it gets worse. Let’s assume that the appeal to strict impartiality does, in fact, provide at least some rationale for a classical consequentialism. If this rationale is plausible, then one need only combine that rationale with a commitment to the (also plausible) claim that we have sufficient practical reason to be partial to

Extramoral as a “gatekeeping” feature of the distinct moral domain whatever else we find plausible about that domain. This claim is worth rejecting. 37 A big caveat here. The moral epistemic status of Extramoral may not seem so implausible if moral rationalism is analytically true, or if moral requirements just refer to normative requirements. But this proposal, or so I’ve argued elsewhere, cannot make sense of our moral experience. See my “How Not to Argue.”



Consequentialism, Moral Rationalism, and Options    173

the near and dear. Once we have these two claims in place, we are licensed to hold that moral rationalism is false. Any partisan of moral rationalism must either reject classical consequentialism and its attendant rationale or reject a commitment to the claim that we have sufficient reason to be partial to ourselves and/​or the near and dear. But the first seems outlawed: the first possibility requires a prior assumption of moral rationalism, which is illegitimate given that moral rationalism is not a priori. But the second is also clearly outlawed—​at least insofar as moral rationalism is to defend commonsense consequentialism. And hence not only is it the case that we have very little reason to believe that moral rationalism is or should be true, but there is actually positive reason to believe it should be false: the plausibility of the strict impartiality embodied by classical consequentialism, and the insistence that no reasonable requirements could fail to take into consideration a person’s deepest concerns and commitments. Thus I  think we are licensed to claim that moral rationalism is probably false. Why the “probably”? I admit that one remaining option is to show that we should accept moral options without the presumption of moral rationalism, that is, without assuming that we ought to conform to our moral requirements. However, this is a very difficult line of argument to state plausibly. Among other challenges, one must explain why morality permits such options while other domains do not, without explaining the difference in terms of the comparative normative significance of morality. But I leave open the possibility that such a rationale could be found. Hence the “probably.”

8.4. No Commonsense, No Consequentialism

If moral rationalism is probably false, this causes very serious problems for commonsense consequentialism. In particular, it would seem that there is clearly no rationale for the commonsense bit. (Again, note that I’m ignoring here the possibility of a constraint-​accepting consequentialism; perhaps one could defend a “half-​ commonsense” consequentialism that accepts constraints without options.) Given that the commonsense bit of commonsense consequentialism seems to rely on moral rationalism, and given that moral rationalism is probably false, we should probably reject the commonsense bit of commonsense consequentialism. We lack a specific rationale to accept an options-​accepting consequentialism against the classical version of consequentialism, which is independently motivated by plausible principles, e.g., the strict impartiality of the moral point of view. So there appears to be little rationale for a specifically options-​accepting consequentialist moral theory. But let’s leave this aside. Even if we are somehow convinced that

174    Dale Dorsey

options permit an adequate rationale independently of moral rationalism, the failure of moral rationalism shows that we also have little rationale to accept commonsense versions of consequentialism, rather than nonconsequentialist commonsense views. This is true even if we accept the teleological theory of practical reasons. After all, the most compelling defense of commonsense consequentialism is via a commitment to the teleological theory of reasons combined with a commitment to moral rationalism. But if we reject moral rationalism, we needn’t accept consequentialism even if we accept the teleological theory of reasons. And hence the argument for commonsense consequentialism simply fails: we can accept that many moral reasons are nonteleological in character, but admit that when they are they simply won’t be rationally overriding or provide decisive reasons. Insofar as moral rationalism is probably false, this possibility is fully open depending on, specifically, the extent to which one has moral permission to conform to the strongest teleological reason. Thus whether we should accept a consequentialist or nonconsequentialist options-​ accepting moral theory must be decided specifically on grounds of independent moral inquiry—​independent, that is, of any requirement that moral obligations fit or match up with a person’s practical reasons, whatever their character. References Brink, David. “Utilitarian Morality and the Personal Point of View.” Journal of Philosophy 83 (1986): 417–​438. Crowther, Bosley. “Blue Skies.” New York Times, October 17, 1946. Dorsey, Dale. “Weak Anti-​Rationalism and the Demands of Morality.” Noûs 46 (2012): 1–​23. Dorsey, Dale. “Consequentialism, Cognitive Limitations, and Moral Theory.” In Oxford Studies in Normative Ethics, vol. 3, edited by Mark Timmons, 179–​202. Oxford: Oxford University Press, 2013. Dorsey, Dale. “How Not to Argue against Consequentialism.” Philosophy and Phenomenological Research 90 (2015): 20–​48. Dorsey, Dale. “Moral Distinctiveness and Moral Inquiry.” Ethics 126 (2016): 747–​773. Dorsey, Dale. The Limits of Moral Authority. Oxford: Oxford University Press, 2016. Dorsey, Dale. “A Challenge for Constraints.” Unpublished manuscript. Foot, Philippa. “Utilitarianism and the Virtues.” Mind 94 (1985): 196–​209. Frankena, William K. Ethics. 2nd ed. Englewood Cliffs, NJ: Prentice-​Hall, 1973. Howard-​ Snyder, Frances. “The Rejection of Objective Consequentialism.” Utilitas 9 (1997): 241–​248. Kagan, Shelly. The Limits of Morality. Oxford: Oxford University Press, 1989. Miller, Dale. “Actual Consequence Act-​Utilitarianism and the Best Possible Humans.” Ratio 16 (2005): 49–​62. Parfit, Derek. “Equality or Priority?” Lindley lecture, University of Kansas, 1991.



Consequentialism, Moral Rationalism, and Options    175

Portmore, Douglas W. Commonsense Consequentialism:  Wherein Morality Meets Rationality. Oxford: Oxford University Press, 2011. Railton, Peter. “Alienation, Consequentialism, and the Demands of Morality.” In Facts, Values, and Norms, 151–​186. Cambridge, UK: Cambridge University Press, 2003. Rawls, John. A Theory of Justice. Cambridge, MA: Harvard University Press, 1971. Scheffler, Samuel. The Rejection of Consequentialism. Oxford: Oxford University Press, 1982. Scheffler, Samuel. “Morality’s Demands and Their Limits.” Journal of Philosophy 83 (1986): 531–​537. Scheffler, Samuel. Human Morality. Oxford: Oxford University Press, 1992. Schroeder, Mark. “Teleology, Agent-​Relative Value, and ‘Good.’” Ethics 117 (2007): 265–​295. Weirich, Paul. “Utility Tempered with Equality.” Noûs 17 (1983): 423–​439. Wiland, Eric. “Monkeys, Typewriters, and Objective Consequentialism.” Ratio 18 (1997): 352–​360.

9 N E W CO NS EQ UENTIA LISM A ND THE  NE W D O I N G-​A L LOWI N G DIS T I NC T I O N

Paul Hurley

Samuel Scheffler characterizes traditional consequentialism as a moral theory holding that “the right act . . . is the one that will produce the best overall outcome, as judged from an impersonal standpoint.”1 On such an account if action B is the promoting of the impersonally best overall state of affairs, it will never be right not to perform B. Yet it often does seem right not to perform such actions. Thus, to take one paradigmatic example, although it seems clear that I should save five rather than one, it seems equally clear to most people that I should not save five by driving over one.2 Standard examples of deontic constraints, moral constraints upon acting to bring about the best overall outcome, are also cases of this sort. Thus, I ought not (and morally ought not) to break a promise that I have made to another even if such a promise breaking on my part will somehow result in two others keeping promises that they would otherwise break. The following claim captures this commonsense distinction between doing and allowing:

1 Scheffler, Introduction to Consequentialism and Its Critics, 1. 2 This example is introduced by Philippa Foot in “The Problem of Abortion,” and discussed by Warren Quinn in “Actions, Intentions, and Consequences.”

176



The New Doing-Allowing Distinction    177

Doing-​Allowing/​Agent-​Neutral: Agents sometimes have decisive moral reasons not to do B even though B promotes an agent-​neutrally better outcome.3 By not doing B I  will be allowing an agent-​neutrally worse outcome to happen. Nonetheless, I morally ought not to do B. There have been myriad attempts to make something like this distinction more precise, but a common element among virtually all such attempts is the commitment, captured in Doing-​Allowing/​Agent-​Neutral, to there being decisive moral reasons not to perform some action B even though performing B will bring about an agent-​neutrally better outcome. Although traditional consequentialist moral theories challenge the legitimacy of the commonsense distinction captured by Doing-​Allowing/​Agent-​Neutral, a new wave of consequentialists have broken with their traditional forbearers and joined Kantians, virtue ethicists, and many Humeans in arguing for the acceptance of Doing-​Allowing/​Agent-​Neutral. Like Scheffler’s traditional consequentialists, such new consequentialists put forward a ranking of outcomes from best to worst, but theirs is an evaluator-​relative rather than an agent-​neutral ranking of outcomes.4 Like the agent-​neutral, impersonal ranking, such an evaluator-​relative ranking takes into account only reasons to promote or bring about outcomes; it differs in taking into account agent-​relative as well as agent-​neutral reasons to promote outcomes, e.g., reasons to minimize my rights violations as well as reasons to minimize rights violations overall. The result is a ranking of outcomes that is relative to the agent in question rather than neutral among agents—​of outcomes that are better or worse relative to her. Like Scheffler’s traditional consequentialists, these new consequentialists focus on the act that will produce the highest ranked state of affairs, but unlike them the initial focus is upon acts that are rationally rather than morally required; not the right action, the action that an agent is morally required to perform, but the action that the agent ought to perform in the decisive reasons implying sense of ought.5 Such new consequentialists argue that in some cases not doing B promotes the evaluator-​relatively best (highest ranked) outcome, although 3 An agent-​neutrally or impersonally better outcome is an outcome that is higher ranked from a standpoint that takes into account only relevant reasons to promote outcomes that are neutral among persons, e.g., reasons to promote overall well-​being or to minimize overall rights violations. An evaluator-​relatively better outcome, by contrast, is understood as an outcome ranked higher from a standpoint that takes into account both such agent-​neutral reasons to promote outcomes and relevant agent-​relative reasons to promote outcomes, e.g., the agent’s reasons to promote her own well-​being or to minimize her own rights violations. 4 See, for example, Smith, “Neutral and Relative Value after Moore”; Dreier, “Structures of Normative Theories”; Louise, “Relativity of Value”; Portmore, Commonsense Consequentialism, ch. 4. 5 For accounts that explicitly articulate and defend this relationship between what agents ought to do in the decisive reasons implying sense and the evaluator-​relative ranking of outcomes, see Smith, “Neutral and Relative Value after Moore”; Portmore, Commonsense Consequentialism.

178    Paul Hurley

doing B promotes the agent-​neutrally best outcome. In such cases agents have decisive reasons, and in some cases distinctively moral decisive reasons, to refrain from doing B. Thus, such new consequentialists join other moral theorists in accepting Doing-​Allowing/​Agent-​Neutral. But many such new wave consequentialists shift their target for rejection to a new doing-​allowing distinction: Doing-​Allowing/​Evaluator-​Relative: Agents sometimes have decisive reasons not to do B even though B promotes an evaluator-​relatively better outcome. Doing-​Allowing/​Agent-​Neutral is a claim about decisive moral reasons not to do some action B that is better in virtue of promoting agent-​neutrally better outcomes; Doing-​ Allowing/​Evaluator-​Relative is a claim about decisive reasons simpliciter not to do some action B that is better in virtue of promoting evaluator-​relatively better outcomes. Like Doing-​Allowing/​Agent-​Neutral, Doing-​Allowing/​Evaluator-​Relative accords with commonsense. Moreover, it too is endorsed by adherents to alternative moral theories. But although new consequentialists accept the traditional doing-​allowing distinction, they reject Doing-​Allowing/​Evaluator-​Relative. Their grounds for such a rejection, the import of such a rejection, and the defensibility of their grounds, are our topic going forward. It is noteworthy that each of these Doing-​Allowing claims has the same general form: Doing-​Allowing/​Form: Agents sometimes have decisive reasons of some sort not to do B even though B is better in some respect. I will have occasion to introduce other claims with this general form in the course of my argument.6 My strategy in what follows is to clarify the nature of the debate concerning this new doing-​allowing distinction, and its importance for normative ethics, by way of a contrast with its more traditional counterpart. In section 9.1 I will take up traditional challenges to Doing-​Allowing/​Agent-​Neutral and highlight certain replies that are frequently endorsed in some form by new wave consequentialists themselves. I will demonstrate in section 9.2 that it is the appeal by many such evaluator-​ relative consequentialists to a teleological conception of reasons (henceforward TCR), upon which all reasons, both agent-​relative and agent-​neutral, are reasons

6 I am indebted to Elinor Mason for identifying both the general form of these claims and the nature of various particular instantiations of this general form.



The New Doing-Allowing Distinction    179

to promote outcomes,7 that commits them to the rejection of Doing-​Allowing/​ Evaluator-​Relative. In section 9.3 I  will show that, like Doing-​Allowing/​Agent-​ Neutral, Doing-​Allowing/​Evaluator-​Relative finds ample support in the appeal to common sense, but I will demonstrate in section 9.4 that unlike the traditional arguments against Doing-​ Allowing/​ Agent-​ Neutral, new wave consequentialist arguments against Doing-​Allowing/​Evaluator-​Relative are grounded in widely held accounts of values, attitudes, and actions. It is this rejection of Doing-​Allowing/​ Evaluator-​Relative, however, that establishes the cornerstone of the consequentialist evaluative framework, that the deontic evaluation of actions is properly determined through appeal to the telic evaluation of outcomes. Moreover, the rejection of Doing-​Allowing/​Evaluator-​Relative, coupled with certain commitments that are widely recognized as independently compelling, provides deep structural support for some form of consequentialist moral theory. TCR is the crucial premise in the new consequentialist argument against Doing-​ Allowing/​Evaluator-​Relative. Critics of TCR, hence of the TCR-​based argument against this new doing-​allowing distinction, challenge the accounts of value, attitudes, and actions upon which the defense of TCR is based. In section 9.5 I will outline alternative accounts of value, attitudes, and actions that do not support TCR. Adjudication between these competing accounts takes us far beyond normative ethics and even metaethics into central questions in the philosophy of mind, action theory, and the nature of normativity. In section 9.6 I will take up one such question with direct bearing on this debate, the question of the nature of the distinct “directions of fit” characteristic of the theoretical and practical spheres. 9.1. New Consequentialism and the Old Doing-​Allowing Distinction

Consider first another doing-​allowing claim that instantiates Doing-​Allowing/​ Form. Like Doing-​Allowing/​Agent-​Neutral, this instantiation appeals to actions that promote an agent-​ neutrally better outcome. But like Doing-​ Allowing/​ Evaluator-​Relative, it appeals to decisive reasons simpliciter, not specifically moral reasons, not to perform such actions: Doing-​Allowing/​Agent-​Neutral/​Reasons:  Agents sometimes have decisive reasons not to do B even though doing B is promoting an agent-​neutrally better outcome.

7 For two articulations of such a teleological conception see Scanlon, What We Owe to Each Other, 79–​87; Portmore, Commonsense Consequentialism,  56–​83.

180    Paul Hurley

Just as Doing-​Allowing/​Agent-​Neutral would be false if the only moral reasons were reasons to promote the agent-​neutrally best overall state of affairs, so too this instantiation, Doing-​Allowing/​Agent-​Neutral/​Reasons, would be false if the only reasons were reasons to promote the agent-​neutrally best overall state of affairs. Although some theoretical currents may seem to pull in the direction of denying Doing-​Allowing/​Agent-​Neutral/​Reasons, such a position attracts few adherents.8 It is taken to be obvious that we have agent-​relative as well as agent-​neutral reasons, and that taking into account such agent-​relative reasons we sometimes have decisive reasons not to perform some action B even though B would promote an agent-​neutrally better outcome. Although there is no agreement about what these other agent-​relative reasons are, or when they are sufficient or decisive, recognition that there are some such reasons is typically taken to provide sufficient grounds for accepting this doing-​allowing distinction. Although few argue against Doing-​Allowing/​Agent-​Neutral/​Reason, many traditional consequentialists have argued against Doing-​Allowing/​Agent-​Neutral.9 Often such arguments emphasize that that there is no clear, independent way of drawing such a distinction that can straightforwardly resolve controversial cases.10 But we have seen that no such clear, independent way is taken to be necessary to vindicate Doing-​Allowing/​Agent-​Neutral/​Reasons. Rather, it is taken to be enough that common sense tells us there are agent-​relative reasons, and that they are sometimes decisive reasons not to promote the agent-​neutrally best outcome. If, by analogy, common sense tells us there are agent-​relative moral reasons, and that in at least some cases they provide decisive reasons not to promote the agent-​neutrally best outcome, our evidence for Doing-​Allowing/​Agent-​Neutral would appear to be on equal footing. Moreover, commonsense morality certainly recognizes such reasons. For example, we have already seen that few deny the commonsense appeal of deontic constraints, moral requirements not to lie, cheat, kill, steal, or break promises. Yet deontic constraints seem most naturally to be understood as reflecting the decisiveness of just such impartial but agent-​relative moral reasons that agents have not to do some action B even though B prevents more such equally wrongful actions from happening. Thus, if what is needed to endorse Doing-​Allowing/​Agent-​ Neutral is what suffices in the case of Doing-​Allowing/​Agent-​Neutral/​Reasons, we

8 But see recent arguments by Singer and De Lazari-​Radek in The Point of View of the Universe for a version of this position. 9 See, for example, Rachels, “Active and Passive Euthanasia”; Bennett, “Morality and Consequences” and “Negation and Abstention”; Tooley, “An Irrelevant Consideration.” 10 Scheffler discusses such arguments in “Doing and Allowing,” 216–​217.



The New Doing-Allowing Distinction    181

seem to have what we need. Many new wave consequentialists recognize the force of this parallel.11 What of the traditional consequentialist’s “Compelling Idea,” an idea that, on one standard interpretation, rules out any such cases in which I can have decisive moral reasons not to do B if B brings about the agent-​neutrally best outcome? Traditional consequentialists take the Compelling Idea to be that it is always morally permissible to promote the agent-​neutrally best outcome.12 Our new consequentialists counter that what is independently compelling is the idea that it is always morally permissible to promote the best outcome, but not the agent-​neutrally best outcome. The latter, they suggest, is a problematic specification of the Compelling Idea that is not independently compelling, and it is only this problematic specification of the Compelling Idea, not the Idea itself, that raises problems for a consequentialist upholding Doing-​Allowing/​Agent-​Neutral. The foregoing are not offered as decisive arguments against the traditional consequentialist challenge to Doing-​Allowing/​Agent-​Neutral. But they do reflect a line of thought that a growing number of evaluator-​relative consequentialists accept in some form—​that what agents have decisive moral reasons to do is not always what promotes the agent-​neutrally best state of affairs. Such consequentialists allow that I have not only agent-​neutral reasons to promote the care of children, but also, for example, agent-​relative reasons to promote the care of my child, agent-​relative reasons to promote my caring for my child, and agent-​relative reasons to promote my keeping my promises.13 But recognition of such agent-​relative reasons supports Doing-​Allowing/​Agent-​Neutral. An agent with such a moral reason to promote her promise-​keeping, for example, can have decisive reasons not to break her promise even when doing so allows more promises overall to be broken. 9.2. New Consequentialism and the New Doing-​Allowing Distinction

New wave consequentialists break with their more traditional forebears in taking into account agent-​relative as well as agent-​neutral reasons to promote, generating a ranking of outcomes relativized to each agent. But many also retain an element Such distinctively moral agent-​ relative reasons to promote outcomes are recognized by Portmore, Commonsense Consequentialism, ch. 4; Dreier, “Structures of Normative Theories”; Smith, “Deontological Moral Obligation.” 12 For versions of this formulation of the Compelling Idea, see Scheffler, The Rejection of Consequentialism, 4; Schroeder, “Teleology, Agent-​Relative Value, and ‘Good,’ ” 281; Dreier, “In Defense of Consequentializing,” 100; Sachs, “Consequentialism’s Double-​Edged Sword,” 264. 13 These examples of reasons draw upon David McNaughton and Piers Rawlings’s important discussion of agent-​ relative and agent-​neutral reasons in “Agent-​Relativity and the Doing-​Happening Distinction.” 11

182    Paul Hurley

of traditional consequentialism—​the teleological conception of reasons (TCR)—​ according to which all practical reasons are such reasons to promote or bring about outcomes.14 TCR thus holds that when such reasons to promote decisively favor some outcome, promoting/​bringing about this highest ranked outcome will be the action that the agent has decisive reasons simpliciter to perform: what the agent “has most reason to do is to bring about the possible world . . . that she has most reason to want to be actual,” the highest ranked outcome relative-​to-​her.15 If we follow the convention of identifying the evaluator-​relatively highest ranked outcome as the evaluator-​relatively best outcome, and of identifying outcomes that are ranked higher than others as evaluator-​relatively better,16 the action that an agent has decisive reasons to perform is the action that promotes the highest ranked outcome, and the highest ranked outcome is the evaluator-​relatively best outcome, the best outcome relative-​to-​the-​agent. When such reasons decisively favor some course of action, we ought, in Parfit’s decisive reasons implying sense,17 to promote the outcome favored by such reasons. Doing-​Allowing/​Evaluator-​Relative holds, by contrast, that agents sometimes ought not to do B even though it promotes an evaluator-​ relatively better outcome: Doing-​Allowing/​Evaluator-​Relative:  Agents sometimes have decisive reasons not to do B even though B promotes an evaluator-​relatively better outcome. To advocate TCR is to reject Doing-​Allowing/​Evaluator-​Relative. It is to accept that there is never decisive reason not to do B when the relevant reasons to promote states of affairs favor B, hence when B promotes an evaluator-​relatively better outcome. To endorse TCR, as many new wave consequentialists do, is thus to target a new doing-​allowing distinction for rejection, Doing-​Allowing/​Evaluator-​Relative. 9.3. The Commonsense Case against TCR and for Doing-​Allowing/​ Evaluator-​Relative

An obvious response to such an argument against Doing-​Allowing/​Evaluator-​ Relative is to reject TCR, the conception of reasons upon which all reasons are reasons

In Scanlon’s terminology “to aim at some result” (What We Owe to Each Other, 84). 15 Portmore, Commonsense Consequentialism, 56. 16 I am setting aside, for the purposes of this essay, challenges to the very coherence of evaluations of outcomes as better and worse relative to an agent. See, for example, Schroeder, “Teleology, Agent-​Relative Value, and ‘Good’ ”; Dreier, “In Defense of Consequentializing.” 17 Parfit, On What Matters, 33. 14



The New Doing-Allowing Distinction    183

to bring about states of affairs. If TCR is false, in particular, if some reasons to act are not reasons to promote or bring about states of affairs, and are only mischaracterized as such, then even if the relevant reasons to promote decisively favor some course of action B, other reasons to act can provide decisive reason simpliciter not to do B. Such an agent will have decisive reasons not to do B even though B promotes the evaluator-​relatively best outcome, vindicating Doing-​Allowing/​Evaluator-​Relative. Our ordinary practices of practical reasoning seem to fuel skepticism about TCR. For example, because TCR holds that all reasons are reasons to produce/​bring about outcomes, it maintains that what an agent has decisive reasons to do is always to promote some result, to bring about some outcome. TCR thus suggests that all reasons to act are reasons to perform actions of this one particular type, bringings about or producings of results/​outcomes. My reasons to run, or argue, or lie, or build a house are ultimately reasons to perform actions of one type, bringings about of outcomes. I thus might run, and choose intentionally to run, but my reasons to run are reasons to promote/​bring about some outcome, if only to bring about the outcome that I run/​have  run. Yet such actions of bringing about or producing outcomes would seem to be only one type of action among others, and Annette Baier is surely right in pointing out that we take ourselves not merely to have reasons to perform actions of this one type, to bring about or promote outcomes (in Baier’s terminology, “to contribute causally to a variety of states of affairs”), but to perform actions of myriad sorts other than bringings about, e.g., “to write, to walk, to argue, to announce.”18 We certainly do seem to have reasons to bring about outcomes. But we also seem to perform hosts of other actions that are not bringings about, e.g., keeping promises, going for walks, or contemplating paintings, and to have reasons to perform such actions that are not reasons to bring it about that we do. Reasons to perform all types of actions appear in our ordinary practices to be on all fours. Whether the action is a bringing about or some other type, e.g., going for a walk or keeping a promise, I can ask why you are performing such an action, and expect you to respond by providing reasons. Nor are the reasons offered always reasons to bring about outcomes, or grounded in such reasons. I can of course perform some action, e.g., keeping my promise, in order to bring something about, e.g., that you trust me. But I can also seemingly bring something about, e.g., that a tree is planted in your name, in order to keep a promise. I may have decisive reasons to keep a promise or to bring about some outcome. Why treat such reasons, as TCR mandates, in categorically different ways? In particular, why treat reasons to perform actions of all types as reasons to perform actions of only one type, bringings about of outcomes? Baier, “Act and Intent,” 653. 18

184    Paul Hurley

TCR raises particularly distinctive problems concerning our commonsense understanding of our relationship to our own actions. Consider first an agent’s relationship to the actions of a third person. I can have reasons to bring it about that Sue acts. Typically, I do this by bringing it about that she has reasons to act. For example, I can bring it about that she goes to the meeting by calling in a favor, ordering her, or paying her, thereby giving her reason to go. TCR treats the first personal case, bringing about my own actions, in a structurally similar way. Just as I have reasons to bring it about that Sue goes, my reasons to go are understood as reasons to bring it about that I go.19 But it is hard to make sense of this first-​personal account. As an agent attempting to bring it about that I act, I seem to have the wrong kind of relationship to my action; it is an outcome that I am attempting to bring about, not an action that I am doing for reasons.20 It is my reasons for going, not my reasons for bringing it about that I go, that are directly relevant to the action of going, and there would appear to be ample grounds for doubting that all of my reasons to act are, or are typically grounded in, reasons to bring it about that I act. Don’t I often just have reasons to keep a promise or go for a walk? The suggestion by some advocates of TCR is that these are best understood as cases in which the outcome I have reason to bring about just is the act, i.e., in which one runs, for example, or keeps a promise, “merely for the sake of bringing it about that one runs,” or that one keeps one’s promise.21 I run and keep my promise to bring it about that I run and keep my promise. Once again, however, this seems difficult to square with commonsense practices. My aim in such cases would appear to be to go for a run, not to bring it about that it is true that I do; it is the truthmaker of the proposition—​ running—​that is my object, not bringing about the proposition’s truth—​that I run. Indeed, my reasons for bringing it about that I run often seem to be importantly distinct from my reasons to run.22 The most effective strategy to bring it about that I run may in some circumstances be to put myself in a situation that coerces me to

Portmore suggests that the outcome at which we aim “need not be anything having to do with the causal consequences of the act,” because it “could be nothing more than to bring it about that one performs the act” (“The Teleological Conception of Reasons,” 117). But the aim of bringing it about that one performs the act is to bring about a causal consequence—​that one performs the act. Moreover, such a consequence can be brought about in myriad ways, e.g., via hypnosis, or self-​coercion, or simply by performing the act in order to bring it about that one has. 20 As Richard Moran points out, “the stance from which a person speaks with any special authority about his belief or his action is not a stance of causal explanation but the stance of rational agency,” and a “person’s own relation to his . . . intentional actions must express the priority of justifying reasons over purely explanatory ones” (Authority and Estrangement, 127–​128). 21 Portmore, Commonsense Consequentialism, 56. 22 This is a variation upon a point made by Stephen Darwall, Welfare and Rational Care, 93, and Talbot Brewer, The Retrieval of Ethics,  20–​21. 19



The New Doing-Allowing Distinction    185

run, or to undergo hypnosis, but such steps to bring it about that I go for a run often seem to undermine or bypass my reasons to go for a run. In many cases I appear to have reasons not to perform some action, but reasons to bring it about that I do perform it. For example, I may have reasons of mutual respect not to lie, but prudential reasons to bring it about that I lie, e.g., to take steps to lower my inhibitions against telling such a lie. Reasons to bring it about that I act seem to be at cross purposes, in such cases, with my reasons not to perform such an action. In other cases, it is my reasons to act that seem to provide me with reasons to bring it about that I act, not vice versa. Thus, if I have reasons to keep my promise, but know that I am akratic with respect to promise keeping, I have reason to bring it about that I keep my promise, e.g., by taking various steps to guard against such akratic tendencies.23 I seem to have reasons to bring it about that I keep my promise in such cases precisely because I  have reasons to keep my promise coupled with grounds for doubting that I will act for such reasons. Reasons to bring it about that I act come into view in such cases precisely because they are atypical, aberrant cases in which my rational authority is compromised, leading me to suspect that I will not act for the reasons that I have to act. Pace TCR, our ordinary, commonsense practices recognize reasons to bring about outcomes as only some among the myriad reasons that we have to act, and recognize actions that we have reasons to perform that are neither bringings about of outcomes, nor properly rationalized as outcomes that we have reasons to bring about. Kantians, virtue ethicists, and many Humeans align themselves with common sense in recognizing such reasons and such actions, hence in rejecting TCR and endorsing Doing-​Allowing/​Evaluator-​Relative. Moreover, such commonsense grounds for accepting Doing-​Allowing/​Evaluator-​Relative would appear to be analogous to the commonsense grounds endorsed by new evaluator-​relative consequentialists for accepting Doing-​Allowing/​Agent-​Neutral. Why do such consequentialists then accept such grounds in the one case and reject them in the other? 9.4. Arguments in Support of TCR and against Doing-​Allowing/​ Evaluator-​Relative

The answer is that new wave consequentialists take there to be powerful arguments for discounting common sense in the case of Doing-​Allowing/​Evaluator-​Relative, arguments of precisely the sort that they take to be absent in the case of Doing-​ Allowing/​Agent-​Neutral. The real case for TCR is to be found in distinctive—​and Such cases are discussed by Moran, Authority and Estrangement, 128. 23

186    Paul Hurley

distinctively state of affairs centered—​accounts of values, attitudes, and actions, accounts that lead us away from our ordinary, commonsense practices of practical reason giving toward TCR and the denial of Doing-​Allowing/​Evaluator-​Relative. It will be useful to characterize briefly the support that the adoption of such state of affairs centric accounts of value, attitudes, and/​or actions provides for the state of affairs centric account of reasons, TCR. I will take up each in turn. In the next section I will demonstrate that other, non-​state of affairs centric accounts of value, attitudes, and actions provide no support for TCR or for the rejection of Doing-​ Allowing/​Evaluator-​Relative. 9.4.1. Value

One such source of support for TCR, hence for rejection of Doing-​Allowing/​ Evaluator-​Relative, is provided by commitment to a prevalent theory of value, the theory upon which the primary bearers of value are states of affairs. Following Douglas Portmore, I will characterize such a view as state of affairs value monism.24 If it is states of affairs that are the primary bearers of value, then it is natural to take all fundamental reasons to be reasons to promote such valuable states of affairs, and natural to hold that “our relation to states of affairs lies in being able to realize them, to prevent them from occurring, or to make their occurrence more or less likely.”25 On such a state of affairs centered account, the proper response by a rational agent to value will always be promotion.26 All ultimate reasons will be reasons to promote valuable outcomes, and what I ought to do will always be to promote the highest ranked state of affairs. TCR finds strong support in such a theory of value. Although all reasons do not seem to be reasons to promote outcomes, acceptance of such an account of value provides grounds for concluding that they are after all. 9.4.2. Desire

One prevalent account of practical attitudes such as desires, wants, and intentions also appears to countermand common sense, providing grounds for adopting TCR and rejecting Doing-​Allowing/​Evaluator-​Relative. Such an account takes desires to be the fundamental practical attitudes, as beliefs are the fundamental theoretical attitudes. Moreover, although in everyday discourse many desires take things as their Portmore, Commonsense Consequentialism, 66ff. It is noteworthy that Portmore himself does not endorse state of affairs value monism. See also Jamie Dreier’s arguments in this volume for recognizing the fundamental value of properties, hence for rejecting state of affairs value monism. 25 Scanlon, What We Owe to Each Other, 80. 26 See, for example, Pettit’s argument for this result in “Non-​consequentialism and Universalizability.” 24



The New Doing-Allowing Distinction    187

objects (“I want a Harley Davidson”), and many more take actions as their objects (“I want to go to that movie”), such an account takes all desires fundamentally to be propositional attitudes, attitudes which, like beliefs, take states of affairs in the world (captured by “that clauses”) as their objects (“I desire that you come”). On such a view, a desire is a “state of being motivated, or of wanting something to happen and being to some degree disposed to make it happen, if we can.”27 Consider, within the context of this account, the apparently stark contrast between wanting to act and desiring that some outcome obtains. To desire that an outcome obtains is to be disposed to bring about the truth of the state of affairs that is the content of the desire—​to bring it about or make it happen. But to want to act, on such an account, is also to be disposed to bring about the truth of a state of affairs, the state of affairs in which I act in the relevant way. The only difference is that in one case the state of affairs that is brought about is (or is in part constituted by) my acting, while in the other case the state of affairs that I am disposed to make happen is not some action on my part. Wanting to act turns out to be a special case of desiring that some state of affairs happens. Such an account of desire dovetails seamlessly with TCR. Consider first subjectivist, desire-​based accounts of reason, which take desires themselves to provide reasons.28 The reasons provided by desires, thus understood as attitudes of being motivated to bring about states of affairs, will all be reasons to bring about the states of affairs that are their objects—​TCR follows naturally. Now consider objectivist, value-​based accounts of reasons. Such accounts typically take desires to be attitudes that are sensitive to considerations that count in favor of their objects. If desires are attitudes of being motivated to bring about the truth of the states of affairs that are their propositional contents, the reasons to which such attitudes are sensitive seem naturally to be provided by considerations that count in favor of the states of affairs that are their objects. Again, such an account of the relevant attitudes suggests that practical reasons are considerations that count in favor of propositional contents as to be promoted, i.e., that reasons are fundamentally reasons to promote. The reasonable agent will desire that ϕ if there are reasons that count in favor of bringing it about that ϕ is the case, and will cease to desire that ϕ upon coming to realize that relevant considerations count against its object as to be promoted. On either subjectivist or objectivist accounts of reasons, acceptance of such a state of affairs centered account of practical attitudes invites TCR, a state of affairs centered account Parfit, On What Matters, 43. 28 I am here appealing to Parfit’s distinction between accounts that are subjectivist about reasons, upon which our reasons for acting “are all provided by . . . certain facts about what would fulfill or achieve our present desires or aims,” and accounts that are objectivist about reasons, upon which our “reasons are given by facts about the objects of these desires” (On What Matters, 45). 27

188    Paul Hurley

of practical reasons as reasons to promote, thereby undermining Doing-​Allowing/​ Evaluator-​Relative. 9.4.3. Action

Buoyed by such accounts of value and/​or attitudes, it can seem plausible to follow Portmore in interpreting the platitude that it is through our actions that “we affect the way the world goes” as the claim that “our intentional actions necessarily aim at making the world go a certain way,”29 i.e., that we always act to bring it about that some state of affairs obtains. Such state of affairs centered accounts of values and/​or attitudes, that is, can make it seem far less problematic to take all reasons for action to be reasons to bring about outcomes and to take all actions to be such bringings about. We saw in the previous section that such an account is hard to reconcile with common sense and ordinary practices of reason giving. But within the context of commitments to such state of affairs centered accounts of value and attitudes we can seem to be drawn ineluctably toward its acceptance. If it is states of affairs that are the primary bearers of value, or desires are attitudes toward states of affairs, then it can seem natural to understand reasons to act to be reasons to promote states of affairs, and to take ourselves always to have most reason to bring about the highest ranked state of affairs. The puzzle for someone with such commitments becomes not why the state of affairs centered theory of reasons (TCR) should be adopted, but what possible grounds there could be for avoiding it. Similarly, the puzzle is not whether Doing-​Allowing/​Evaluator-​Relative should be rejected, but why it initially comes across as plausible. These state of affairs centered accounts of desire, action, value, and reason (TCR), and the apparent support that they provide for the rejection of Doing-​Allowing/​ Evaluator-​Relative, are all in evidence in the work of Michael Smith, a leading advocate of new wave consequentialism. Smith defends an analysis of what agents “ought to do” that proposes a “reduction of one moral concept (the concept of what we ought to do) to another pair of moral concepts (the concepts of goodness and badness),”30 such that an agent ought to do ϕ if and only if it is the available action “that produces the most good and the least bad”31 on an evaluator-​relative ranking of the outcomes produced. The action that an agent ought to perform, Smith argues, is the action of producing or bringing about the evaluator-​relatively best outcome. Moreover, his state of affairs centered account of value is articulated through appeal 29 Portmore, Commonsense Consequentialism, 56. 30 Smith, “Neutral and Relative Value after Moore,” 576. 31 Smith, “Neutral and Relative Value after Moore,” 576.



The New Doing-Allowing Distinction    189

to state of affairs centered accounts of desire and reason. Judgments of value are about what an agent “would desire to be the case in C if she had a psychology that eludes all forms of rational criticism,” i.e., were her desires “maximally informed and coherent and unified.”32 Desires are fundamentally attitudes toward states of affairs as to be made the case, actions are the means of making such desired states of affairs the case, and the evaluator-​relatively best state of affairs, the state of affairs that an agent ought to produce in any circumstance, is always the state of affairs that she would desire over all others were she rational. Doing-​Allowing/​Evaluator-​Relative must be rejected on such an account because what the agent ought to do is to produce the evaluator-​relatively best state of affairs (when such a determinate state of affairs is available), the state of affairs that she would desire to be the case were her desires maximally informed and coherent. The argument for TCR and against Doing-​Allowing/​Evaluator-​Relative makes the case for a state of affairs centered evaluative framework, upon which the deontic evaluation of actions as what ought and ought not to be done is determined entirely through appeal to reasons to promote states of affairs and the evaluator-​ relative ranking of states of affairs that results: agents rationally ought to do what promotes the evaluator-​relatively best state of affairs. Although Doing-​Allowing/​ Evaluator-​Relative is articulated in terms of decisive reasons simpliciter rather than decisive moral reason in particular, its acceptance generates a powerful presumption in favor of consequentialist moral theories, upon which the moral deontic evaluation of actions is determined through appeal to the telic evaluation of states of affairs. Consider, for example, the conjunction of TCR with an idea that is widely recognized as independently plausible, that agents are at least morally permitted to perform any action that they have decisively good reasons to perform. If, as TCR suggests, agents always ought to promote the evaluator-​relatively best state of affairs, then the Compelling Idea of consequentialist moral theory, that I am always at least morally permitted to promote the best outcome, follows straightforwardly from TCR coupled with this independently plausible idea. It is thus not surprising that philosophers who appeal to TCR to reject Doing-​Allowing/​Evaluator-​Relative also tend to endorse some form of moral consequentialism as well,33 and that many philosophers who reject moral consequentialism take themselves to be committed to rejecting TCR and endorsing Doing-​Allowing/​Evaluator-​Relative.

Smith, “Neutral and Relative Value after Moore,” 592, 596. To be rational, for Smith, is to have and exercise “a capacity to have the psychological states that coherence demands” (596). Thus, to judge some state of affairs to be in the relevant sense good, for Smith, is to judge that she would desire it insofar as she was rational. 33 See, for example, Smith, “Two Kinds of Consequentialism”; Portmore, Commonsense Consequentialism, chs.  5–​7. 32

190    Paul Hurley

I have argued that common sense provides grounds for both a rejection of TCR and an endorsement of Doing-​Allowing/​Evaluator-​Relative. But we have also seen that widely held state of affairs centered accounts of values, attitudes, and actions provide support for TCR, hence for the rejection of Doing-​Allowing/​Evaluator-​ Relative. Defenders of Doing-​Allowing/​Evaluator-​Relative thus must provide, and provide support for, alternatives to these state of affairs centered accounts of value, attitudes, and action. I  will suggest in the next section that this is precisely what many defenders of Doing-​Allowing/​Evaluator-​Relative attempt to do. 9.5. Arguments against TCR and for Doing-​Allowing/​Evaluator-​Relative

9.5.1. Value

I have argued that state of affairs value monists, holding that states of affairs are the primary bearers of value, have grounds to adopt TCR, hence to reject Doing-​ Allowing/​Evaluator-​Relative. Some opponents of consequentialism and Doing-​ Allowing/​Evaluator-​Relative are sympathetic to such a theory of value, but such sympathies raise notorious problems for their normative ethics.34 Others leave such an account of value unchallenged, arguing for an account of right action that is prior to and independent of an account of value (the good). But we have seen that left unchallenged, such state of affairs value monism supports an account of reasons that threatens to undermine any such independent account of the right.35 Many philosophers, however, reject such state of affairs value monism and adopt a theory of value that is not state of affairs centered. Some, including Kant and many Kantians, can be read as retaining a form of value monism, but arguing that wills rather than states of affairs, for example, are the primary bearers of value.36 Others are value pluralists,37 maintaining that other things in addition to states of affairs, e.g., persons and objects, are primary bearers of values, and that the reasons to which they give rise are often not reasons to promote. Nico Kolodny, for example, simply takes it to be a “mistake . . . to think that things of value are sources of reasons only in the sense that, when we are able to bring about something of value, we have reason to do so.”38 Others follow Scanlon in adopting a buck-​passing theory of value, upon Thomas Nagel, for example, is a deontologist who evinces such sympathies. For discussions of this aspect of Nagel’s account, and the difficulties that it creates for his deontological normative ethics, see Scanlon, What We Owe to Each Other, 81–​84, and my Beyond Consequentialism, 194–​208. 35 Michael Smith makes this point in “Neutral and Relative Value after Moore,” 587, as does Barbara Herman in “Leaving Deontology Behind,” in her The Practice of Moral Judgment, 208–​210. 36 See, for example, Herman, The Practice of Moral Judgment, 214ff. 37 See, for example, Scheffler, “Valuing,” 23–​42, particularly his emphasis on “the wide range of things that we value” (33); Kolodny, “Aims as Reasons,” 66–​69. 38 Kolodny, “Aims as Reasons,” 68. 34



The New Doing-Allowing Distinction    191

which being valuable is not itself a property that provides us with reasons; rather, to call something valuable is to “say that it has other properties that provide reasons for behaving in certain ways with regard to it.”39 Reasons to promote or bring about are typically recognized on such buck-​passing accounts as only a subset of such reasons that are invoked by calling something valuable.40 Such alternatives to state of affairs value monism invite conceptions of reasons upon which recognition of value is reflected in reasons to act that are not all reasons to promote outcomes, just as common sense suggests. 9.5.2. Desire

I have suggested that commitments to the claims that desires are the fundamental practical attitudes and that they are fundamentally propositional attitudes provide presumptive support for TCR and the rejection of Doing-​Allowing/​Evaluator-​ Relative. By contrast, commitment to accounts upon which intentions to ϕ rather than desires that ϕ are the fundamental practical attitudes, or upon which not all desires are fundamentally propositional attitudes, provides no such support for TCR and the rejection of Doing-​Allowing/​Evaluator-​Relative. We have already seen that ordinary desire discourse appears to support such an alternative view of desire. We cite desires that ϕ, but also intentions and desires to ϕ and desires for ϕ. Advocates of the position that all desires are fundamentally propositional attitudes sometimes cite as a virtue of their account that all desires can be translated into propositional attitude form, but, as Talbot Brewer points out, such translation can also “be run in the opposite direction,” into “desire to X form,” and “intuitively these latter representations seem to yield a more perspicuous representation of the true objects of desires.”41 In the next section we will take up deeper arguments, grounded in the direction of fit contrast, for taking desires understood as propositional attitudes to be the fundamental practical attitudes. But it seems clear that absent such additional arguments there is a presumption in favor of taking at least some practical attitudes to be fundamentally intentions, wants, and desires to ϕ that provide, or are sensitive to, reasons to ϕ. On a subjectivist account of reasons, desires thus understood would provide reasons to perform the actions that are their objects, and would provide no grounds for assuming, contrary to common sense, that all such reasons to act are reasons

Scanlon, What We Owe to Each Other, 96. 40 See, for example, Scanlon’s discussion of reasons of friendship and valuing friendship in What We Owe to Each Other,  88–​90. 41 Brewer, The Retrieval of Ethics, 21. See also Thompson, Life and Action, 121–​134. 39

192    Paul Hurley

to promote outcomes. On an objectivist account of reasons, the reasons to which such desires and intentions are sensitive would be considerations that count in favor of their objects—​actions rather than states of affairs—​as to be performed. Again, such an account would provide no grounds for adopting TCR or challenging the commonsense appeal of Doing-​Allowing/​Evaluator-​Relative. 9.5.3. Action

Such philosophers who do not accept state of affairs value monism, and who do not accept that desires are fundamentally attitudes toward states of affairs as to be promoted, will not find it at all natural to parse the platitude that actions are the means by which we affect the world as the claim that all actions aim at making some state of affairs obtain in the world. Indeed, this will strike them as an attempt to interpret all actions as bringings about, an interpretation that conflicts with ordinary practice and finds no support in what they recognize to be the most plausible accounts of value, attitudes, and action. With the distinction clearly in view between the platitude and the prima facie implausible interpretation of it, they will embrace the platitude, which provides no support for TCR, and reject the prima facie implausible interpretation as unsupported by either theory or our commonsense practices. 9.6. Deeper Grounds

I have sketched accounts of value, desire, and action that provide support for TCR, and alternative accounts that align more with common sense while providing no such support. Such a mapping of the philosophical terrain does little to address the deeper arguments that might be marshalled in support of one or the other of these accounts. Such supporting arguments, I suggest, typically draw upon claims and commitments that lead well beyond normative ethics and even metaethics into deep questions in the philosophy of mind and the theory of action. For example, support for the account of desires as fundamentally propositional attitudes, and as the fundamental practical attitudes, is sometimes taken to be provided by appeal to accounts of the contrasting directions of fit characteristic of the theoretical and practical spheres. On one prevalent view of this contrast, the fundamental attitudes characteristic of the theoretical and practical spheres are distinguished not by their objects, which are in each case states of affairs, but by the contrasting directions of fit that such propositional attitudes take toward the states of affairs that are their contents. The contrast, on this view, is between beliefs that aim to fit the world, and desires that aim to make the world



The New Doing-Allowing Distinction    193

fit them.42 This contrast takes desires that ϕ to play an analogous and equally fundamental role in the practical sphere to the role played by beliefs that ϕ in the theoretical sphere, and to have such a practical attitude toward a proposition is to be “disposed to act on the world in ways calculated to make the proposition true.”43 Such an account of contrasting directions of fit can seem to manifest the virtues of both simplicity and symmetry. There are two fundamental attitudes. They have the same contents—​states of affairs. They are distinguished entirely by the contrasting directions of fit characteristic of these different fundamental attitudes toward states of affairs. Once we accept the role played by desires on this prevailing account of directions of fit, upon which they are understood as propositional attitudes that play a fundamental role in the practical sphere analogous to the role of beliefs in the theoretical sphere, it is natural to take decisive reasons for acting to be fundamentally reasons for promoting the states of affairs that agents have the most reason to desire, and to hold that the best action promotes the best (highest ranked) state of affairs. Acceptance of such an account of directions of fit thus provides grounds for accepting state of affairs centered accounts of desire, action, and reason, hence for rejecting Doing-​Allowing/​Evaluator-​Relative. But although such an account of contrasting directions of fit provides support for desires understood as propositional attitudes, hence for TCR, alternative accounts provide no such support. For example, although G. E. M. Anscombe’s arguments in Intention are frequently cited as the locus classicus of this account of contrasting directions of fit, Richard Moran and Martin Stone have demonstrated that such an interpretation of Anscombe is simply mistaken.44 To the extent that Anscombe is making a case for contrasting directions of fit at all, they show that she is best understood as identifying a normative distinction between the paradigmatic forms of mistake and the correlative standards of correctness that are characteristic of the theoretical and practical spheres. The fundamental contrast is that whereas in the theoretical sphere it is expressions of beliefs that can be mistaken, failing to fit the world, in the practical sphere it is actions performed that can be mistaken, failing to fit the relevant expressions of intention.45 For presentations of this prevailing form of the contrast, see Platts, Ways of Meaning, 256–​257; Smith, The Moral Problem, 111–​119; Audi, “Moral Judgments and Reasons for Action,” 129; Boyle and Lavin, “Goodness and Desire,” 171ff. 43 Brewer, The Retrieval of Ethics,  16–​17. 44 See the interpretations of Anscombe’s arguments in Intention that are developed by Moran and Stone, “Anscombe on Expression,” and Frost, “On the Very Idea of Direction of Fit,” 464ff. See also Scanlon, Being Realistic about Reasons, 65–​66, for a very different argument against such appeals to the direction of fit contrast. 45 Moran and Stone, “Anscombe on Expression,” 67ff; Frost, “On the Very Idea of Direction of Fit,” 464. Frost is dubious that this distinction identified by Anscombe is any longer plausibly characterized as one of contrasting directions of fit. 42

194    Paul Hurley

On the theoretical side of Anscombe’s actual distinction the relevant attitudes are propositional attitudes, beliefs that ϕ, and the failure is a failure of the belief expressed to fit with the world. But on the practical side the relevant attitudes are intentions to ϕ, performative (rather than propositional) attitudes with actions (rather than states of affairs) as their objects, and the failure is a failure of performance—​a failure to act in a way fitting with the intention expressed.46 Expressions of beliefs (“He bought asparagus”) and expressions of intentions (“I am going to buy asparagus”) manifest the distinctive, contrasting commitments characteristic of the theoretical and practical spheres—​ commitments to the world fitting my beliefs and to my actions fitting my intentions. Note that desires that ϕ do not capture the practical side of this contrast at all. To believe that ϕ is to be committed to ϕ being the case; to desire that ϕ is not to be committed to bringing it about that ϕ is the case. If I have a stronger countervailing desire that ϕ not be the case, I may well have a desire that ϕ while at the same time having an abiding commitment that not ϕ. To intend to ϕ, by contrast, is to be committed to ϕ-​ing, as to believe that ϕ is to be committed to ϕ being the case.47 Such an alternative account of directions of fit raises challenges to the status of desires that ϕ as the fundamental practical attitudes. Both parties to such disputes concerning the proper understanding of the direction of fit contrast agree that actions are done for reasons, and that the relevant practical attitudes play a central role in rationalizing explanations of actions. Advocates of the standard direction of fit account maintain that desires that ϕ, propositional attitudes, are the practical attitudes that play this central role in the rationalizing explanation of actions.48 Anscombe and those who follow her, by contrast, take actions themselves, and intendings, plannings, and wantings to act—​performances and performative attitudes—​to play this central role in the rationalizing explanation of actions. Why am I buying eggs? Because I am making an omelet, or am planning to make an omelet, or intend to make an omelet, or want to do so later in the week. Indeed, desires that ϕ, they argue, are particularly ill-​suited to play such a role in rationalizing explanations.49 Adoption of such an alternative account of contrasting directions of fit, and of the attitudes that play a central role in rationalizing explanations, shifts our focus from attitudes with states of affairs as their objects to attitudes with actions as their objects. The

Arguments for understanding certain practical attitudes as performative rather than propositional attitudes, attitudes with actions to be done rather than propositions to be made true as their objects, can be found in Thompson, Life and Action, ch. 8; Moran and Stone, “Anscombe on Expression”; Boyle and Lavin, “Goodness and Desire”; Brewer, The Retrieval of Ethics, ch. 1; Frost, “On the Very Idea of Direction of Fit,” e.g., 462–​463. 47 On this point see Moran and Stone, “Anscombe on Expression,” 46, 64–​66. 48 The classic account of this role for desires understood as propositional attitudes in rationalizing explanations of actions is provided by Davidson, “Actions, Reasons, and Causes.” 49 Thompson, Life and Action, 120–​134; Brewer, The Retrieval of Ethics,  19–​24. 46



The New Doing-Allowing Distinction    195

judgments to which such attitudes will be sensitive will be considerations that count in favor of performing the actions that are their objects. Such an account, however, provides no support whatsoever for TCR. A thorough defense of this interpretation of Anscombe’s account, much less of the account itself, would take us far beyond the scope of this essay. The point of sketching the account here is rather to demonstrate that appeals to practical attitudes, to the direction of fit contrast, and to the nature of rationalizing explanations of actions do not resolve questions in favor of TCR. Instead, they push the debate for and against TCR into deep questions in the philosophy of mind and the philosophy of action. Accounts of value that reject state of affairs value monism provide no support for TCR. Accounts that take performative attitudes such as intending and wanting to ϕ to play the relevant contrasting role to beliefs in accounts of contrasting directions of fit, and that take such performative attitudes to play a central role in the rationalizing explanation of actions, provide no support for TCR. Indeed, they align with commonsense grounds for rejecting TCR and for upholding the new doing-​allowing distinction, Doing-​ Allowing/​Evaluator-​Relative, along with the old one, Doing-​Allowing/​Agent-​Neutral. 9.7. Conclusion

Traditional consequentialism confronted normative ethics with the denial of the traditional doing-​allowing distinction; new forms of evaluator-​relative consequentialism confront normative ethics with the denial of a new doing-​allowing distinction. Only once the new doing-​allowing distinction is revealed for what it is, its relationship to the teleological conception of reasons (TCR) is made clear, the import of its adoption is clarified, and the transethical nature of the grounds for and against it are made apparent, can the debate between advocates and opponents of such a distinction be fully and effectively joined. My project in this paper has been to perform this philosophical ground-​clearing.50 References Anscombe, G. E. M. Intention. Cambridge, MA: Harvard University Press, 2000. Audi, Robert. “Moral Judgments and Reasons for Action.” In Ethics and Practical Reason, edited by Garrett Cullity and Berys Gaut, 125–​159. Oxford: Oxford University Press, 1997. I  am grateful to Christian Seidel and Elinor Mason for insightful written comments on earlier drafts, and to audiences at the Friedrich-​Alexander-​Universität Erlangen-​Nürnberg and the Claremont Colleges. Thanks also to Doug Portmore, Julia Driver, Dale Dorsey, Rivka Weinberg, Amy Kind, Andrew Schroeder, and Alex Rajczi for helpful discussions. 50

196    Paul Hurley Baier, Annette. “Act and Intent.” Journal of Philosophy 67 (1970): 648–​658. Bennett, Jonathan. “Morality and Consequences.” In The Tanner Lectures on Human Values, vol. 2, edited by Sterling M. McMurrin, 45–​116. Salt Lake City: University of Utah Press, 1981. Bennett, Jonathan. “Negation and Abstention.” Ethics 104 (1993): 75–​96. Boyle, Matthes, and Douglas Lavin. “Goodness and Desire.” In Desire, Practical Reason, and the Good, edited by Sergio Tenenbaum, 161–​201. New York: Oxford University Press, 2010. Brewer, Talbot. The Retrieval of Ethics. Cambridge, MA: Harvard University Press, 2009. Darwall, Stephen. Welfare and Rational Care. Princeton, NJ: Princeton University Press, 2002. Davidson, Donald. “Actions, Reasons, and Causes.” In Essays on Actions and Events, 3–​19. Oxford: Clarendon Press, 1980. Dreier, James. “Structures of Normative Theories.” The Monist 76 (1993): 22–​40. Dreier, Jamie. “In Defense of Consequentializing.” In Oxford Studies in Normative Ethics, vol. 1, edited by Mark Timmons, 97–​119. Oxford: Oxford University Press, 2011. Foot, Philippa. “The Problem of Abortion and the Doctrine of Double Effect.” In Virtues and Vices, 19–​32. Berkeley: University of California Press, 1978. Frost, Kim. “On the Very Idea of Direction of Fit.” Philosophical Review 123 (2014): 429–​482. Herman, Barbara. The Practice of Moral Judgment. Cambridge, MA:  Harvard University Press, 1993. Hurley, Paul. Beyond Consequentialism. Oxford: Oxford University Press, 2009. Kolodny, Niko. “Aims as Reasons.” In Reasons and Recognition, edited by R. Jay Wallace, Rahul Kumar, and Samuel Freeman, 43–​78. New York: Oxford University Press, 2011. Louise, Jennie. “Relativity of Value and the Consequentialist Umbrella.” Philosophical Quarterly 54 (2004): 518–​536. McNaughton, David, and Piers Rawlings. “Agent-​ Relativity and the Doing-​ Happening Distinction.” Philosophical Studies 63 (1991): 167–​185. Moran, Richard. Authority and Estrangement. Princeton, NJ: Princeton University Press, 2001. Moran, Richard, and Martin Stone. “Anscombe on Expression of Intention:  An Exegesis.” In Essays on Anscombe’s Intention, edited by Anton Ford, Jennifer Hornsby, and Frederick Stoutland, 33–​75. Cambridge, MA: Harvard University Press, 2011. Parfit, Derek. On What Matters. Vol. 1. Oxford: Oxford University Press, 2011. Pettit, Philip. “Non-​ consequentialism and Universalizability.” Philosophical Quarterly 50 (2000): 175–​190. Platts, Mark. Ways of Meaning. London: Routledge, 1979. Portmore, Douglas. Commonsense Consequentialism:  Wherein Morality Meets Rationality. New York: Oxford University Press, 2011. Portmore, Douglas. “The Teleological Conception of Reasons.” Mind 120 (2011): 117–​153. Quinn, Warren. “Actions, Intentions, and Consequences: The Doctrine of Doing and Allowing.” In Morality and Action, 149–​174. Cambridge, UK: Cambridge University Press, 1993. Rachels, James. “Active and Passive Euthanasia.” New England Journal of Medicine 292 (1975): 78–​86. Sachs, Benjamin. “Consequentialism’s Double-​Edged Sword.” Utilitas 22 (2010): 258–​271. Scanlon, Thomas M. What We Owe to Each Other. Cambridge, MA:  Harvard University Press, 1998. Scanlon, Thomas M. Being Realistic about Reasons. Oxford: Oxford University Press, 2014.



The New Doing-Allowing Distinction    197

Scheffler, Samuel. Introduction to Consequentialism and Its Critics, edited by Samuel Scheffler, 1–​13. Oxford: Oxford University Press, 1988. Scheffler, Samuel. The Rejection of Consequentialism. Oxford: Oxford University Press, 1994. Scheffler, Samuel. “Doing and Allowing.” Ethics 114 (2004): 215–​239. Scheffler, Samuel. “Valuing.” In Reasons and Recognition, edited by R. Jay Wallace, Rahul Kumar, and Samuel Freeman, 23–​42. New York: Oxford University Press, 2011. Schroeder, Mark. “Teleology, Agent-​Relative Value, and ‘Good.’” Ethics 117 (2007): 265–​295. Singer, Peter, and Katarzyna De Lazari-​Radek. The Point of View of the Universe. Oxford: Oxford University Press, 2014. Smith, Michael. The Moral Problem. Oxford: Blackwell, 1994. Smith, Michael. “Neutral and Relative Value after Moore.” Ethics 113 (2003): 576–​598. Smith, Michael. “Two Kinds of Consequentialism.” Philosophical Issues 19 (2009): 254–​272. Smith, Michael. “Deontological Moral Obligation and Non-​Welfarist Agent-​Relative Values.” Ratio 13 (2011): 351–​363. Thompson, Michael. Life and Action. Cambridge, MA: Harvard University Press, 2008. Tooley, Michael. “An Irrelevant Consideration:  Killing versus Letting Die.” In Killing and Letting Die, edited by Bonnie Steinbock and Alastair Norcross, 103–​111. Englewood Cliffs, NJ: Prentice Hall, 1980.

10 CO N S EQ UENT I ALISM, RATIONA LIT Y, A N D KA N T I A N R ES P EC T

Tim Henning

It is a natural idea that agents’ preferences play a part in determining what it is rational for them to do. A second natural idea is that this aspect of practical rationality has a consequentialist or teleological structure. Where there’s a preference, there is a value to be promoted, namely, the satisfaction of the preference—​or so the thought goes. This teleological aspect is often assumed to be what explains the rational norm: rational agents ought, ceteris paribus, act so as to bring about the outcomes they prefer because that way the aim of preference-​satisfaction will be served. This view of preferences and rationality matters greatly for discussions of consequentialism in ethics, as moral consequentialists often base defenses of their view on it. As I will describe more fully below, classical consequentialist authors as well as “new wave” consequentialists argue that consequentialism fits best with our convictions about rationality in nonmoral contexts, especially in contexts where agents deliberate in the light of their preferences. To give a prominent example right away, here is how Peter Singer defends “preference utilitarianism”: Now, imagine that I  am trying to decide between two possible courses of action—​perhaps whether to eat all the fruits I  have collected myself, or to share them with others. Imagine, too, that I am deciding in a complete ethical vacuum, that I know nothing of any ethical considerations. . . . How would 198



Consequentialism, Rationality, and Kantian Respect    199

I make up my mind? One thing that would be still relevant would be how the possible courses of action will affect my interests.  .  .  . Suppose I  then begin to think ethically, to the extent of recognising that my own interests cannot count for more, simply because they are my own, than the interests of others. In place of my own interests, I now have to take into account the interests of all those affected by my decision. This requires me to weigh up all these interests and adopt the course of action most likely to maximise the interests of those affected.1 Commenting on this passage at a later point in his book, Singer adds that “interests” here are to be identified with preferences.2 Thus the argument for “preference utilitarianism” is very simple:  Take the natural idea that nonmoral reasoning is concerned with how an action will affect the satisfaction of one’s own preferences; add the idea that moral judgments do not grant any special importance to the satisfaction of one’s own preferences vis-​à-​vis the satisfaction of others’ preferences; derive the result that just as nonmoral reasoning is a matter of maximizing egoistic preference satisfaction, ethics is a matter of maximizing universal preference satisfaction. This chapter will not criticize the idea that preferences play a part in determining what it is rational to do. (In fact, I will sketch an attempt to explain why this is so.) However, I will criticize the second natural idea, i.e., the one that is exploited by consequentialists in ethics. I will argue that the rational import of preferences is not always explained in terms of features of outcomes. I will sketch an alternative account that is Kantian in spirit. It explains the import of preferences not in terms of consequences but in terms of respect for persons. But my main point is negative:  Moral consequentialism is not a natural extension of an uncontroversial view of rational decision-​making. Even basic features of rational decision-​making need more resources than consequentialism offers. It is tempting to reverse the argument: If basic features of rationality cannot be explained in consequentialist terms, there is no reason to expect morality to be explicable in such terms either. But this further argument is beyond the scope of this chapter. Its main point is the rejection of the consequentialist view of the rational import of preferences. Section 10.1 discusses the alleged structural parallel between certain norms that connect preferences and rationality, on the one hand, and consequentialist ethics, on the other hand. Section 10.2 then describes decision situations of a particular type, situations that involve what I will call predictable preference accommodation. Here it seems intuitively rational for agents to act on their present preferences, but 1 Singer, Practical Ethics, 13. 2 See Singer, Practical Ethics, 94.

200    Tim Henning

it also seems clear that this verdict cannot be explained in terms of (rankings of ) outcomes. Section 10.3 discusses attempts to deal with these cases in consequentialist frameworks and rejects them. Section 10.4 discusses an appeal to self-​evidence. Finally, section 10.5 outlines the Kantian account. 10.1. Consequentialism and Rationality

I will discuss a view about the relation between rationality and preferences. The view is weak; it is the view that among two or more alternative actions open to an agent, this agent’s preference for the outcome of one of these actions makes it rational to choose this action if there are no other, stronger reasons to perform one of the others. This view is weak in that it assigns to preferences only the status of tie-​ breakers. Where other reasons leave ties, preferences will tip the balance.3 Many authors endorse stronger views, claiming that preferences are more than tie-​ breakers, or even that preferences exclusively determine what it is rational for us to do. Some identify the requirement to pursue one’s preferences with a certain aspect or form of rationality, most notably, perhaps, with prudence. Thus Phillip Bricker writes, “I take the mark of a prudent person to be roughly this: that he acts so as to get what he wants, has wanted, or will want.”4 But no matter whether these stronger views are correct, most people accept at least the weak view above. It will be the focus of this chapter. (In what follows, I will sometimes speak of “rationality” in order to pick out the aspect expressed by the weak view. This is just a matter of convenience.) It is well known that this aspect of rationality lends itself to a particular type of representation. If an agent’s preferences for outcomes satisfy some basic conditions of coherence, then these preferences define an ordering over these outcomes. In fact, if these preferences are coherent in the specific sense used by von Neumann and Morgenstern,5 they define an ordering that can be represented by a set of utility functions, all of which are positive linear transformations of one another. Thus an agent who acts in accord with these coherent preferences can be represented as someone who strives for outcomes with the most “expected utility.” This structural fact naturally invites the idea that when it comes to the rational import of 3 This view is often expressed by saying that in these circumstances, preferences give us reasons. I will avail myself of this way of speaking, but I want to note that it is open to serious objections—​see, e.g., Broome, “Normative Requirements”; Kolodny, “Why Be Rational?”; Raz, “The Myth of Instrumental Rationality.” It is worth noting, however, that even most of these critics grant that preferences bear on rationality; they doubt that rationality is associated with reasons. In section 10.5, I sketch an account that explains why preferences can be reasons in a way that, I think, avoids the relevant charge of “bootstrapping.” 4 Bricker, “Prudence,” 382. 5 Von Neumann and Morgenstern, Theory of Games and Economic Behavior.



Consequentialism, Rationality, and Kantian Respect    201

preferences, consequences come first in the order of explanation. If an agent has a preference for a particular outcome, this outcome has, relative to this agent, a certain feature that explains why he or she has reason to (or is rational to) strive to realize it. (The natural candidate for this feature would probably be the satisfaction of the agent’s preference in the outcome. We do consider it a good thing if preferences are satisfied.) If the reasons given by our preferences are explained in some such way, I will call them outcome-​based. Of course, from the mere fact that preferences can be represented by rankings of outcomes it does not follow that outcomes or their rankings come first in the order of explanation. This observation is familiar from analogous debates about “consequentializing” in ethics. If consequentializers are right, any normative theory will have a consequentialist counterpart. But, as Portmore notes,6 this does not make them genuinely consequentialist. For a genuinely consequentialist view, the order of explanation is crucial. Such a theory first specifies a ranking of outcomes in a way that is independent of a moral assessment of acts and only afterward goes on to say that an act is right not just if and only if but because it promotes the highest-​ranking outcome. The same point holds for the rational import of preferences. Just because an agent who rationally follows his preferences can be represented as maximizing some value in the outcome, that does not mean that this latter representation captures the fact that is fundamental, normatively speaking. Still, as I  said, such a consequentialist explanation of the rational import of preferences seems natural. And once adopted, it can be put to use for consequentialist arguments in ethics. In fact, even critics of moral consequentialism often accept that the relevant aspect of rationality will serve this aim. Rawls says that in utilitarianism, “the principle of choice for an association of men is interpreted as an extension of the principle of choice for one man.”7 Rawls thinks this extension is harmful in that it leads utilitarians to ignore the separateness of persons. But moral consequentialists derive support from the alleged consequentialist structure of rationality. There are many ways in which this is undertaken. Ambitious versions try to derive forms of moral consequentialism from norms of rationality by means of additional premises. I  gave one prominent example, Singer’s defense of “preference utilitarianism.” The general model (and the more rigorous elaboration) of this line of argument was presented by J. Harsanyi.8 He proves that the preferences of a rational agent, when combined with a weak Pareto condition and an anonymity condition, 6 Portmore, “Consequentializing Moral Theories.” 7 Rawls, A Theory of Justice, 21. 8 Harsanyi, Essays on Ethics, Social Behavior, and Scientific Explanation.

202    Tim Henning

yield preferences that can be represented by social welfare functions that are utilitarian. The consequentialist structure is seen as built into the starting point, i.e., rationality. Less ambitious versions of this line of argument, often put forward by so-​called new wave consequentialists,9 appeal to considerations of theoretical economy or of coherence. At any rate, all of these arguments presuppose that reasons given by our preferences are outcome-​based. It is only if we make this assumption that we can follow the above arguments to the conclusion that in ethics, likewise, rankings of outcomes fundamentally explain our obligations. This is precisely the assumption that I want to criticize in what follows. 10.2. Rational Choice and Predictable Preference Accommodation

It is well known that changing preferences pose interesting challenges in economics and decision theory.10 I want to focus on a particular form of preference change.11 Predictable Preference Accommodation (PPA) An agent faces predictable preference accommodation if and only if she contemplates performing an action ϕ such that she can predict that performing ϕ will change her preferences in such a way that afterward, she will prefer the outcome of her ϕ-​ing to the outcomes she could have brought about by performing alternative actions. In cases of PPA, we face actions that will predictably make us like their own consequences best. By way of example, consider Peter, who tries to decide what to study. Should he enroll in philosophy or go to law school instead? Peter prefers philosophy, all things 9 Just a few recent examples: At the heart of Douglas W. Portmore’s defense of consequentialism, Commonsense Consequentialism, we find a theory of reasons for action. And at the heart of the defense of that view (78f.), we find an appeal to prudential reasons as uncontroversial paradigms of outcome-​based reasons. Julia Driver, on the first page of her introductory book on consequentialism, says, “The view has a prima facie plausibility, at least, due to the fact that very many people consider the consequences of alternative courses of action before making a decision about what to do” (Consequentialism, 1). Similarly, Tim Mulgan argues for consequentialism by appeal to “an analogy between moral rationality and individual rationality. On many views of the latter, it is rational for an agent to seek to maximize her own expected utility. By analogy, a moral theory should tell us how a rational agent would behave if she attached equal weight to the well-​being of all agents” (The Demands of Consequentialism, 14). 10 See, e.g., Hammond, “Changing Tastes and Coherent Dynamic Choice.” 11 Cases of this type have received discussion in Bricker, “Prudence,” 397–​400; Gibbard, “Interpersonal Comparisons,” 176; Bykvist, “Prudence for Changing Selves.” I will discuss some of their views below.



Consequentialism, Rationality, and Kantian Respect    203

considered. But he is also certain that if he were to go to law school, that would have far-​reaching effects on what he reads, whom he gets to know, the problems he gets to discuss, etc., and eventually also on his preferences. Peter is certain that he would develop an interest in the law, so that after a while he would be glad he chose that path. So when he considers going to law school, Peter faces PPA. I will focus on a pair of different cases, though. Here is Case 1. Consider Suzy, age twenty-​four. She and her husband are a happy couple. They have good jobs; things are going well for them. They are thinking about having a baby. Though they regard this as a decision that they should make together, they have agreed to approach the question by first thinking about what each of them would decide if the decision was his or hers alone. Reflecting on the matter, Suzy quickly recognizes that she prefers not to have a child right now. Although the circumstances would be suitable, and although she does want children later in her life, right now her priorities are to pursue her career some more and to travel and see the world with her husband. “After all,” she thinks, “we can still have many babies when I am thirty and older.” Suzy prefers the outcome in which she does not have a baby now over the one in which she does. Call the first A, the second B. But here is something Suzy knows about herself: if she were to have a baby now, that would change everything, including her basic preferences. To put it in her own voice: “Suppose that I become a mother now. I am fairly sure that once my baby exists, I will love him or her more than anything else—​and I will be glad I did it. My priorities will have changed, and I will no longer wish that things had gone differently, given that this would mean that my child would not exist. In fact, I think I will strictly prefer things as they are.” Although Suzy prefers outcome A now, she faces PPA in outcome B. Hence she faces two different outcomes in which her preferences will predictably support the respective outcomes. For what it’s worth, I find it rather easy to put myself in Suzy’s shoes, and I see no good grounds for suspecting that her predictions must be false or unreliable.12 Now suppose that Suzy starts to wonder what the significance of PPA for her prudentially rational choice might be. Suzy tells us, “The bottom line is: If I conceive now, I will be glad I did it. And, of course, if I do not conceive now—​which I prefer—​I will later be glad I did not do it. But doesn’t that mean that my options are equally good? Whatever I do, I will later be glad I did it. So I might as well conceive now! Or, if these options are incommensurable or on a par,13 maybe I should simply toss a coin.” True, recent studies show that a considerable number of mothers regret their choice (cf. Donath, “Regretting Motherhood”). And Paul, Transformative Experience, argues that a preference like Suzy’s is not rational if based on ideas of “what it will be like” to become a mother. I do not think these points license general skepticism about predictions like Suzy’s. 13 In the sense of Chang, “The Possibility of Parity.” 12

204    Tim Henning

My impression is that something has gone wrong here. It would be irrational for Suzy to conceive now, or even to toss a coin and run a risk of conceiving now. After all, she does not want to have a child now. She has other plans and she will not miss out on having babies by pursuing them. Maybe the explanation is that we implicitly think of Suzy as someone who would feel happier in outcome A? No. It is natural to think that not only Suzy’s preferences but also her feelings of happiness are likely to change. We may suppose that Suzy herself predicts she will feel very happy as a mother—​in fact, no less happy than in outcome A. But this prediction is perfectly consistent with her present preferences for outcome A. The reason, we may suppose, is this: what matters to Suzy is not just how happy she will be in an outcome, but also whether she wants to be the person who is this happy in this outcome. Here is what she says when she imagines herself in B: “Sure, this life would be happy. But at the end of the day, it is not the happy life I want to live. This happiness would come at the cost of things that matter greatly to me now. Imagining myself, I see a rather pathetic figure, who has never seen anything else than diapers and a suburban home.” Even with this detail added, I find nothing incoherent in Suzy’s preferences. They seem reasonable, and I still think that she would not be rational in choosing to become a mother now. Among the few authors who discuss cases of this type, some tend to disagree.14 On their view, what matters is only the level of satisfaction that an agent would achieve in the alternative outcomes. However, I find it very hard to believe that from the rational point of view, agents may simply ignore what they want their life to be like at present. And in fact, Bricker himself15 expresses doubts in cases where the divergence between present preferences and accommodated preferences is great, as it arguably is in our example. To further confirm the impression that present preferences do matter, consider next Case 2. Here, Suzy faces the same two options with the same two outcomes. She can conceive now and predictably be glad she did it. Alternatively, she can pursue her career and travel with her husband for the next few years, and also predictably be glad she did it. So far, everything is equal. But in Case 2, PPA works in the other direction. What Suzy actually prefers now is to have a baby right away. She appreciates the value that others find in traveling and success, but as things stand, Suzy does not want these things for herself. She

E.g., Bricker, “Prudence.” Bykvist, “Prudence for Changing Selves,” seems to disagree as well, although he mainly focuses on the different question of whether the fact that Suzy would not have preferred outcome B in outcome A counts against choosing outcome B. 15 Bricker, “Prudence,” 399ff. 14



Consequentialism, Rationality, and Kantian Respect    205

wholeheartedly prefers outcome B. But suppose that, again, Suzy knows something important about her future preferences: “If I were to pursue my career and to travel first, I would soon stop thinking about having a baby. I have always found it easy to feel very committed and even enthusiastic about the projects I have chosen. And I am sure that once I started, I would enjoy the traveling and the professional success. In fact, knowing myself, I am certain that I would soon be so absorbed in these projects that I would not want to trade that course of life, per impossibile, for the alternative course of life in which I conceived now. I would prefer to pursue these other projects first and have babies later.” Crucially, Cases 1 and 2 seem similar with regard to the outcomes our protagonist faces, and that includes the preferences that she expects to have in these outcomes. In both cases, Suzy can choose among the same outcomes, and it is part of each outcome that she will like it best. (The reader may be worried about the claim that the outcomes in the two cases are really the same. I will address this and other worries in due course.) Still, I think that our judgments about what it is rational for Suzy to choose in Case 1 and in Case 2 differ. I do not think it is clear that Suzy should not conceive now in Case 2. On the contrary, I think it would be perfectly reasonable for her to do so, more reasonable indeed than choosing the alternative. If having a baby now is what she wants most, and if she does not care for traveling and success, why choose the latter, or toss a coin and run a risk of choosing the latter? By contrast, the account proposed by Bricker (at least for cases in which there is a less dramatic change) recommends otherwise. We are free to make the same assumption in Case 2 as we made in Case 1, i.e., that the degrees of preference satisfaction (and happiness) are the same in both outcomes. Given this, Bricker’s account would entail that in both of our cases, either choice would be equally rational. But that would mean that we may as well ignore the question of whether we are in Case 1 or Case 2. We could, that is, neglect the fact that Suzy in Case 1 and Suzy in Case 2 differ fundamentally in their plans and aspirations. Again, I cannot recognize this as a recommendation of rationality. After all, Bricker himself says that matters of prudence “must be determined from the standpoint of [the agent’s] own concept of the good.”16 (As I said, Bricker uses “prudence” to pick out the aspect of rationality to which preferences matter.) Now, what is the significance of these two cases of PPA? Intuitively, Cases 1 and 2 seem to confirm the general view about rationality and preferences that was presented in section 10.1. First, the intuition that Suzy’s present preferences in both Case 1 and Case 2 are reasonable suggests that in both of these cases, there are no stronger, preference-​independent reasons that require that she chooses one outcome Bricker, “Prudence,” 382. 16

206    Tim Henning

rather than the other. Second, I have claimed that the intuitive judgment about these cases is still that there is a difference in what it is rational for Suzy to choose. So our cases seem to fit the above view, i.e., that preferences constitute (at least) rational tie-​breakers. However, these cases pose a problem for consequentialists. Again, Cases 1 and 2 seem to differ in what it is rational for Suzy to choose. But it seems hard to vindicate this judgment in a consequentialist framework. While the cases differ with regard to Suzy’s present preferences, there is, due to PPA, no such difference in Suzy’s preferences, or their satisfaction, in the corresponding outcomes in the two cases. The outcomes seem to be alike in all relevant respects, including Suzy’s preferences and their level of satisfaction. So we seem to face a difference in what is rational for Suzy to do without a corresponding difference in the outcomes. This is an obvious problem for the consequentialist account of the rational import of preferences (see section 10.1). For instance, the import of Suzy’s present preference for outcome A in Case 1 cannot lie in the fact that she will end up with a satisfied preference in outcome A. For the same holds in Case 2. Thus, these cases suggest that not all of the reasons given by our preferences are outcome-​based.

10.3. Some Replies on Behalf of Consequentialism about Rationality

How can consequentialists about rationality deal with our cases? Clearly, there are two general strategies. They can either deny that Cases 1 and 2 differ in what Suzy rationally ought to choose, or they can deny that the corresponding outcomes in these two cases really are relevantly similar. I have already discussed the first strategy. It does not seem correct to say that from the point of view of rationality, we may as well ignore the fact that in Case 1 and Case 2, Suzy differs fundamentally in what she wants her life to be like. I have nothing to add at this point. The second strategy is more promising. One version of it should be considered first, both because it will seem tempting (even obvious) to many consequentialists and because its problems point us toward an important condition of adequacy for further proposals. The idea is to claim that the above presentation relies on an overly narrow conception of an outcome. Generally, consequentialists are free to include in their conception of an outcome whatever they consider relevant. So why not say that being preferred (or not) at present just is a feature of a future outcome—​not an intrinsic one, presumably, but a feature nonetheless? On this conception of an outcome, the corresponding outcomes in our Cases 1 and 2 do differ, and the problem vanishes.



Consequentialism, Rationality, and Kantian Respect    207

I do not object to generous conceptions of outcomes per se. My complaint about the present move is that it is not explanatory. An adequate account must be able to explain why Suzy’s present preferences should matter in the way they seem to matter. This is by no means obvious. Why, we might ask, should an agent like Suzy judge outcomes in the light of preferences that will no longer be there once the outcome is there? The question is serious. As I said, some authors deny that present preferences matter in this way. Take consequentialists who claim that all that matters about an outcome is the happiness that results in the outcome, or the satisfaction of those preferences that one will have in this outcome. As I said, this seems to me to be the wrong verdict about our cases. But there is a genuine challenge here. If these authors ask why Suzy’s present preferences should matter, it is no help at all to tell them that being preferred by Suzy at the time of choice just is a relevant feature of outcomes. Any account of the differences between the corresponding outcomes will have to indicate what makes these differences relevant. In other words, the claim that present preferences, unlike accommodated future preferences, fix what it is rational to do is controversial, not least among consequentialist authors. So we need a substantive account of the relevant differences between corresponding outcomes. The same goes for Jamie Dreier’s proposal (in his contribution to this volume, note 7) that we conceive of value as time-​centered, and that this will suffice to take care of my problem case. On this proposal, we need a way to allow that the relevant evaluations of outcomes generate different orderings at different times. The idea is then that “the time relevant to the deontic status of an act (including, perhaps, its rationality) is the time of the act itself.”17 But like the previous proposal, this by itself does not explain why Suzy’s present preferences call the shots and settle the relevant ranking. Suppose Suzy takes her own reasoning very seriously, and asks us, “Seriously, if preference satisfaction is the aim to be promoted, then why should I judge a future outcome in terms of preferences that will no longer be there once the outcome is there?” If we tell her that value just is time-​centered in this way, she is unlikely to think that her doubts have been answered. One strategy would be to search for a reason why accommodated preferences should not be taken into account. If this works, we arrive at a picture that resembles more ordinary cases. In Case 1 Suzy prefers not to conceive now, and if she acts accordingly, that will satisfy this relevant preference. If, by contrast, she conceives now, that will satisfy a preference of an irrelevant, accommodated kind. Likewise in Case 2: if Suzy conceives now, this satisfies a relevant preference; if she does not but travels and pursues her career, this will only satisfy an irrelevant accommodated preference. See Dreier’s contribution to this volume, note 7. 17

208    Tim Henning

Thus we would be able to identify a difference between the cases after all, i.e., a difference in which outcome produces the relevant kind of preference satisfaction. Of course, we need to be told why accommodated preferences may be neglected in this way. Here is a proposal that I  think will not work. The accommodated preferences in our cases cannot be rejected as irrational or unreasonable or insufficiently responsive to reasons, etc. If Suzy were to become a mother in Case 1, her accommodated preference for the resulting state of affairs would be perfectly reasonable. And I do not think that it would be reasonable only in a “pragmatic” sense, i.e., in the sense that it would be good for her and her child that she has this preference, although this preference may in fact be lacking in other rational credentials (such as reasons “of the right kind,” as it is often put). On the contrary, the existence of an actual child is of much value, and it is a source of excellent reasons, and there is no reason to suppose that Suzy’s preference isn’t responsive to this value or these reasons. Similar things go for Case 2. Maybe we can rule out accommodated preferences on other grounds, though. Note that as I have presented Case 1, Suzy not only has a preference for outcome A; she also has a higher-​order preference for these preferences, and hence for keeping them. If we wish, we may also use more emphatic Frankfurtian idioms18 and say that Suzy is identified with her present preferences, or that they are wholehearted. Similar things might be said about Case 2 and the preference for outcome B. Now, might not these higher-​order preferences count against choosing an outcome in which Suzy’s first-​order preferences will change? They might indeed. But as far as a consequentialist explanation is concerned, they will raise exactly the same problem. For PPA affects these higher-​order attitudes, too. If, in Case 1, Suzy were to become a mother now, and to develop a preference for being a mother, it is unlikely that she would then still want her old preferences back. It is very natural to assume that she would then identify wholeheartedly with her preference for being a mother. Since second-​order attitudes would predictably change along with first-​order attitudes, they would not mark a relevant difference in the two outcomes. However, Elizabeth Harman gives a more sophisticated account of why accommodated future preferences, even if they would be reasonable, should not be treated as relevant now.19 Here is one of her examples:  Suppose I  have a twenty-​ year-​old daughter who is deaf. I may agree that things would have been better in all relevant respects (impersonally, for my daughter, and for myself ) if she had been cured as a baby. Still, Harman claims, it may be reasonable for me to prefer things as they are because I love my daughter as the person she is now. However, this does not See Frankfurt, The Importance of What We Care About. 19 Harman, “ ‘I’ll Be Glad I Did It.’ ” 18



Consequentialism, Rationality, and Kantian Respect    209

mean that it would have been reasonable to treat this preference as a guide earlier. Clearly, it would be wrong for parents to decide not to cure their baby daughter of deafness on the grounds that they will later love the person she will have become then. Harman’s explanation is this: The outcome in which the child is deaf is worse (impersonally, and for the baby, and for the parents). But love for a particular person can override such considerations of betterness; hence the future accommodated preference would be reasonable in the future. But one cannot now love one’s baby daughter as the deaf adult she will become if one does not cure her. So now there is nothing to override questions of betterness. So the accommodated preference is not now a reasonable guide. Can consequentialists about rationality use these considerations in order to explain why we may legitimately ignore the accommodated preferences in our two cases? No. There is a crucial difference between our cases and Harman’s. In her example, the outcome in which the daughter is deaf is worse, so the relation of love is required to make reasonable a preference for this outcome. But, as I said, it is simply implausible to say that outcome B in Case 1 is worse (impersonally, or for Suzy, or for the baby), so that love for a particular individual is likewise required to render reasonable a preference for this outcome. For in Case 2 Suzy has a present preference for this very outcome, and as I said, it is implausible to say that this is a preference for an outcome that is worse. Here is a further attempt to use the consequentialist’s second strategy, i.e., a further attempt to argue that the corresponding outcomes in Cases 1 and 2 differ after all. Consequentialists can point to the fact that preference accommodation will not be frictionless, so to speak. True, in Case 1 Suzy will come to like outcome B if she chooses it. But it will probably take some time to get to like it. There will be a transition period between the time when she notices the pregnancy and the time when she has fully developed a mother’s preferences. During this period, she will retain her original preference for success and traveling, and she will know that she misses out on these things. So there will be a period in which she will find strong preferences unsatisfied. This is surely an important difference between outcome B in Case 1 and outcome B in Case 2, for in this latter case there will probably be no such period of regret and unsatisfied preferences. Similar things, of course, will apply in the comparison of outcome A in Case 1 and outcome A in Case 2. However, this reply is not convincing either. First of all, we are free to suppose that the friction in our example will be rather minimal. Second, this proposal does not seem to identify the reason that is intuitively most relevant. It seems odd to suggest that the only reason Suzy ought not to conceive now in Case 1 is that otherwise there will be brief episodes of frustration in the first months. Presumably, the reasons Suzy herself would cite have to do with the final outcomes. Finally, this

210    Tim Henning

proposal fails to make sense of the episodes of regret and frustration themselves. The only thing that counts against choosing outcome B, on the present proposal, is that there will be such episodes of regret and frustration. But then what are the reasons for these episodes? If outcome B is otherwise perfectly fine, what is there to regret and be frustrated about? Surely, Suzy cannot regret her choice on the grounds that she experiences this very regret itself. Another possibility would be to claim that psychological change is itself a cost or a disadvantage. Maybe agents have reasons to seek a high degree of diachronic stability and constancy in their goals and characters. Accordingly, there would be a relevant difference between corresponding outcomes in the two cases. In Case 1, choosing A  would ensure diachronic continuity of Suzy’s preferences, whereas in Case 2, choosing B would ensure this. How can we explain that discontinuity itself is a cost? Famously, Michael Bratman has claimed that stable diachronic structures of plans are important for us. They make available the benefits of various kinds of intertemporal and social coordination, and also a certain kind of self-​government.20 Maybe these are reasons why Suzy should choose the outcome in which her preferences stay as they are? No. Extensive as the changes would be, I see no reason to think that these changes of plans will actually impair Suzy’s ability to plan or her self-​government. I conclude that it is just not clear how consequentialists should respond to the problem raised by Cases 1 and 2. They either have to deny (what I take to be) strong intuitions about rational choice, and I see no convincing (and non-​question-​ begging) grounds for doing so. Or they must point to differences between the corresponding outcomes. But so far we found none. 10.4. A Self-​Evident  Truth?

These cases of PPA make it seem doubtful that the reasons given by our preferences are all outcome-​based. It seems that an agent’s preferences matter even when they are not associated with any relevant differences in the outcomes. This result, however, leaves us with the question of why present preferences matter in this way. As I have stressed, it is by no means obvious why this should be so. Unless we come up with a good answer to this question, consequentialists would be free to stick to their guns and reject the intuition that I voiced. Some might think that this demand for an explanation is unwarranted. Maybe it is just a basic, self-​evident truth that present decisions must go by present preferences? See Bratman, Intention, Plans, and Practical Reasoning and “Reflections on Law, Normativity, and Plans.” 20



Consequentialism, Rationality, and Kantian Respect    211

No. In many cases, we do anticipate changing preferences. (This is why many people shy away from getting tattoos with the name of their lover.) In philosophy, many people subscribe to so-​called reflection principles, which require us to adjust present attitudes in accord with predicted future changes in attitudes.21 In decision theory, Richard Jeffrey has devised the famous maxim “Choose for the person you expect to be when you have chosen.”22 While Jeffrey has used his slogan to illustrate his particular account, R. A. Briggs has argued that all major decision-​theoretic approaches are usefully interpreted as attempts to identify those acts that are, as she puts it, “justifiable to your future self—​the person you will be immediately after acting.”23 So it is surely not evident that present choices must go by present preferences. What we need, therefore, is an explanation of why Suzy ought to go by her present preferences in Cases 1 and 2, an explanation that is compatible with the view that there are other cases in which it is rational to adjust one’s present choice to predicted future changes in attitude. I will now argue that we find such an explanation in an unexpected place, namely, in Kant’s ethics. 10.5. A Kantian Account

The core idea of my account is this: Sometimes we ought to give weight to people’s preferences (including our own), not because the resulting satisfaction of preferences is such a great thing but because doing so is a way to respect their authority. How so? If you deliberate about a certain practical matter and you give some weight to the question of what (say) Peter wants, then you make the outcome of the deliberation depend, to some degree, on Peter’s verdict. You thereby let him play a part in the decision and treat him as someone whose verdict counts for something. Let me explain:  If we face a decision, there is not only one but two normative questions that arise. The first one is, of course, Which is the correct decision, i.e., the one that is best supported by the relevant reasons? The second question is often overlooked, but it is no less important. It is the question of who should make the decision, i.e., who is the one with the authority to decide. The significance of the distinction can easily be grasped if we take an example from Stephen Darwall.24 Suppose a middle-​aged woman has dinner with her parents, and they are trying to get her to eat “her” broccoli. Of course, the parents may well be right that it would be best if she ate it. So they may be right that this would be the Van Fraassen, “Belief and Will” and “Belief and the Problem of Ulysses and the Sirens.” 22 Jeffrey, The Logic of Decision, 16. 23 Briggs, “Decision-​Theoretic Paradoxes as Voting Paradoxes,” 17. 24 Darwall, “Desires, Reasons, and Causes.” 21

212    Tim Henning

best decision. But it is not theirs. What their daughter eats is not for them to decide. Their attempt to determine what she eats is a failure of respect, and, as Darwall notes, if she falters and eats it, this is a failure of self-​respect. So just because the parents may be right about what the daughter’s choice should have been, that does not mean that they have a right to interfere with it in this way. The question of whether a person’s decision must be respected is simply different from the question of whether it is correct. I think the difference between these questions is captured very well in Kant’s Metaphysics of Morals, particularly the Doctrine of Right. In a nutshell, Kant there argues that Right must protect people’s equal ability to exercise their Willkür to the greatest extent possible, and that it must do so precisely independently of whether the maxims, i.e., the reasons behind these exercises of Willkür, meet the criterion of the Categorical Imperative. Right protects our freedom to make even unreasonable and immoral choices (including lying and suicide, which Kant notoriously did not take lightly). So Kant is very serious about the distinction between two normative statuses that a choice can have: first, its being correct (supported by reason) and, second, its being properly left to the Willkür of the subject. For reasons of space, I cannot go into exegetical questions. But even if my account does not deserve the title “Kantian,” it still seems promising systematically. The important point is that the parents in Darwall’s example should respect that their daughter’s decision not to eat broccoli is her decision. This requires that they react to her choice in the right way. They should not treat her preference as a reason that bears on whether it is better or worse if she eats broccoli. For in that case, they would still engage in the business of asking whether she should do it. They would still act as if they were in charge of deciding what to do. But they should not treat this as a matter for deliberation on their part. Instead, they should grant their daughter the authority to decide. That means that they should treat her preference as what H. L. A. Hart calls a peremptory reason. Hart illustrates this notion with the example of a command: If a hearer accepts a command, he “take[s]‌the commander’s will instead of his own as a guide to action and so [takes] it in place of any deliberation or reasoning of his own: the expression of a commander’s will is intended to preclude or cut off any independent deliberation by the hearer of the merits pro and con of doing the act.”25 If we go by another person’s preference in order to defer to her authority to decide, this preference is treated as a peremptory reason in this sense. Consequently, like a command, it does not enter into a competition with other reasons, “not even as the strongest or dominant reason, for that

Hart, “Commands and Authoritative Reasons,” 253. 25



Consequentialism, Rationality, and Kantian Respect    213

would presuppose that independent deliberation was to go on, whereas the commander intends to cut off or exclude it.”26 As Darwall’s example suggests, this can apply even in the first-​person case. In the first instance, of course, the daughter should try to determine what is best. She should weigh reasons for and against eating broccoli, and form a preference on that basis. After this is done, however, there is still the second question. Given that she now has a preference, what follows for the question of what to do? Should her verdict count, or should others have a say? Now the question is no longer whether eating broccoli is best. The question is: Whose view on this matter should count? If the daughter concludes, correctly, that her verdict should count, she has a reason not to eat broccoli. This is not because she adds her preference to the stock of reasons she weighed before, in the first step. So she does not engage in an illicit kind of bootstrapping. Her preference for not eating broccoli is a reason not because it somehow renders it better not to eat it, but because it means that not eating it is the way for her to make her verdict the one that counted. Now, how does this view help with our problem? One first important point is this: in a Kantian framework, the relevance of preferences is not always explained in terms of outcomes. As I said, the daughter does not treat her own preference for not eating broccoli as a consideration that shows why her eating broccoli would be worse, all things considered, than her eating it. More important, her parents should not treat the preference in this way. Instead, they should treat the daughter’s preference as a constraint that excludes deliberation on their part. At this point, it should be noted that it would be a distortion to propose that the daughter’s having decided for herself is a valuable feature of an outcome that explains why she should be allowed to have her way. The parents would still fail to show proper respect for her authority if they approached the problem first by asking what they think their daughter ought to do, and then deciding that they think that it would be for the best to allow her to decide. That way, they would still consider themselves to be the final authorities, in charge of what their daughter eats unless they themselves find it preferable to delegate the decision to their daughter. (“Look, Mommy and Daddy have decided that things will be better if you are allowed to decide this for yourself.”) This is why it is important to treat the daughter’s preference as a peremptory reason in Hart’s sense. These points go some way toward solving our problem. Recall what was puzzling about cases of PPA like Cases 1 and 2: Why should Suzy judge outcomes in the light of preferences that will not be around once these outcomes are there? Why, e.g., should Suzy even ask whether she wants to become pregnant now, if she is convinced Hart, “Commands and Authoritative Reasons,” 253. 26

214    Tim Henning

that she will want to be pregnant once it is done? The answer is that respect for what someone wants is not only a matter of promoting outcomes. In a case like this, Suzy’s preference should not be seen as only one consideration among many that bear on which decision is correct. Otherwise, her parents (or even her neighbors) could feel entitled to make the decision just as much as she does, as long as they treat her preference as one of the reasons they have to weigh. That is absurd, of course. When all is said and done, Suzy should be the one to decide. (Or rather, she and her partner should make the decision. But, as I said, they have decided to approach the question first by deliberating as if each were the only relevant authority.) So she should go by her preference not because this preference adds to the other reasons she took into account when forming that very preference, but rather simply because her verdict is the one that should guide the final decision. And everybody else should support her decision not because they think it correct but because it is hers. However, this does not solve our problem completely. We still need to explain why Suzy’s accommodated preferences do not deserve respect in the same way. As I said, we often do take our future preferences into account. So the Kantian had better not say that, in general, only present preferences engender requirements of respect. But then we have to ask why Suzy should be required to respect only her present preferences, and not the preferences she will come to have. This explanation should point to facts specific to cases of PPA, and especially to Cases 1 and 2. It is here that the real strength of the Kantian account emerges. Accommodated future preferences have one feature that makes them special precisely by Kantian lights. What is special about cases of PPA is the dependence of the future preference on one’s present choice. In both Case 1 and Case 2, whether Suzy will later have a preference for A or B depends on what she chooses now. This dependence makes it impossible to regard these future preferences as sources of independent constraints. Whether a particular future preference comes into existence is settled only when Suzy has chosen. So, e.g., in Case 1, Suzy cannot decide to conceive now out of respect for the fact that her future self will prefer it this way, simply because whether her future self will prefer it this way depends on what she decides now. Clearly, a “constraint” that automatically changes in such a way as to fit one’s choice is not a constraint on choice at all. So deferring to one’s future self ’s preference, i.e., giving that future self some authority, is impossible. This looks like the asymmetry we are looking for. First, since respect for persons is not a matter of promoting certain outcomes, it is, after all, correct for Suzy to look at outcomes in the light of preferences that will no longer be around once the outcomes are there. Second, Suzy’s future preferences cannot be similarly respected, since they are dependent on the choice at hand.



Consequentialism, Rationality, and Kantian Respect    215

This is sketchy, for sure. Much more would have to be added to see whether this type of view can actually be made to work. But my main aim here is not to convince the reader of this account. It is to convince her or him that there might be attractive nonconsequentialist accounts of Suzy’s reasons. Without a rough sketch of this kind, one might be left with the worry that if the outcomes do not explain Suzy’s reasons, nothing does. It would then be tempting to deny the intuition that there is a difference in what it is rational to choose in Cases 1 and 2 after all, not because one can refute or explain away the intuition to the contrary, but because of lack of an alternative account. The above sketch is supposed to show that this reaction would be too quick. 10.6. Conclusion

I have argued that consequentialists, both traditional and new wave ones, cannot base defenses of their theories on an uncontroversial account of the import of preferences in rational prudential decision-​making. This is because, as I have argued, this rational import of preferences does not lend itself in all cases to an explanation in terms of the value of the satisfaction of preferences in the preferred outcomes. If you take seriously the rational import of preferences, you may instead find yourself driven to a nonconsequentialist view of core aspects of rational decision-​making.27 References Bratman, Michael E. Intention, Plans, and Practical Reasoning. Cambridge, MA:  Harvard University Press, 1987. Bratman, Michael E. “Reflections on Law, Normativity, and Plans.” In New Essays on the Normativity of Law, edited by Stefano Bertea and George Pavlakos, 73–​ 85. Portland, OR: Hart, 2011. Bricker, Phillip. “Prudence.” Journal of Philosophy 77 (1980): 381–​401. Briggs, R. A. “Decision-​Theoretic Paradoxes as Voting Paradoxes.” Philosophical Review 119 (2010): 1–​30. Broome, John. “Normative Requirements.” Ratio 12 (1999): 398–​419. Bykvist, Krister. “Prudence for Changing Selves.” Utilitas 18 (2006): 264–​283. Chang, Ruth. “The Possibility of Parity.” Ethics 112 (2002): 659–​688. Darwall, Stephen. “Desires, Reasons, and Causes.” Philosophy and Phenomenological Research 67 (2003): 436–​443.

For helpful discussions, I  thank audiences in Konstanz, Erlangen, Zürich, and Frankfurt am Main. Many people provided valuable input. I want to thank them all, especially C. Seidel, J. Gertken, and M. Willaschek. 27

216    Tim Henning Donath, Orna. “Regretting Motherhood: A Sociopolitical Analysis.” Signs: Journal of Women in Culture and Society 40 (2015): 343–​367. Driver, Julia. Consequentialism. London: Routledge, 2012. Frankfurt, Harry G. The Importance of What We Care About. Cambridge, UK:  Cambridge University Press, 1988. Gibbard, Allan. “Interpersonal Comparisons: Preference, Good, and the Intrinsic Reward of a Life.” In Foundations of Social Choice Theory, edited by Jon Elster and Aanund Hylland, 165–​ 193. Cambridge, UK: Cambridge University Press, 1992. Hammond, Peter J. “Changing Tastes and Coherent Dynamic Choice.” Review of Economic Studies 43 (1976): 159–​173. Harman, Elizabeth. “‘I’ll Be Glad I Did It’ Reasoning and the Significance of Future Desires.” Philosophical Perspectives 23 (2009): 177–​199. Harsanyi, John. Essays on Ethics, Social Behavior, and Scientific Explanation. Dordrecht: Reidel, 1976. Hart, H. L.  A. “Commands and Authoritative Reasons.” In Essays on Bentham, 243–​268. Oxford: Oxford University Press, 1982. Jeffrey, Richard. The Logic of Decision. 2nd ed. Chicago: University of Chicago Press, 1983. Kant, Immanuel. Gesammelte Schriften, edited by the Preußische Akademie der Wissenschaften. Berlin: Walter de Gruyter and predecessors, 1902–​. Kolodny, Niko. “Why Be Rational?” Mind 114 (2005): 509–​563. Mulgan, Tim. The Demands of Consequentialism. Oxford: Oxford University Press, 2001. Paul, Laurie A. Transformative Experience. Oxford: Oxford University Press, 2014. Portmore, Douglas. “Consequentializing Moral Theories.” Pacific Philosophical Quarterly 88 (2007): 39–​73. Portmore, Douglas. Commonsense Consequentialism:  Wherein Morality Meets Rationality. Oxford: Oxford University Press, 2011. Rawls, John. A Theory of Justice. Cambridge, MA: Harvard University Press, 1971. Raz, Joseph. Practical Reason and Norms. 1975. Oxford: Clarendon, 1999. Raz, Joseph. Engaging Reason. On the Theory of Value and Action. Oxford:  Oxford University Press, 1999. Raz, Joseph. “The Myth of Instrumental Rationality.” Journal of Ethics and Social Philosophy 1 (2005): 1–​28. Shaw, William. “The Consequentialist Perspective.” In Contemporary Debates in Moral Theory, edited by Julia Driver, 5–​20. Oxford: Blackwell, 2006. Singer, Peter. Practical Ethics. 2nd ed. Cambridge, UK: Cambridge University Press, 1991. Van Fraassen, Bas. “Belief and the Will.” Journal of Philosophy 81 (1984): 235–​256. Van Fraassen, Bas. “Belief and the Problem of Ulysses and the Sirens.” Philosophical Studies 77 (1995): 7–​37. Von Neumann, John, and Oskar Morgenstern. Theory of Games and Economic Behavior. Princeton, NJ: Princeton University Press, 1944.

III

Scope ASSESSING NEW WAVE CONSEQUENTIALISM IN CONTEXT

11 CONS EQ UENTIA LISM A ND MORA L RE SP O N S I B I L I T Y

Elinor Mason

One family of objections to consequentialism focuses on the very wide-​and far-​reaching nature of the consequences of our actions. If rightness depends on the consequences of our actions, then in order to act rightly we have to be able to take into account the tiny ripple effects that happen laterally and reaching far into the future. There are various different objections in this family. One version of the objection is just that consequentialism is very demanding. A more subtle version of the demandingness objection is due to Bernard Williams, who argues not simply that consequentialism asks a lot but that it asks the wrong sort of thing: it requires us to take on responsibility for the actions of others. If someone else’s actions depend on mine, even through her own choices, I have to take into account what she will do as much as any other consequence of my action.1 Another objection in this family, often called the epistemic objection, points out that we cannot foresee the distant consequences of our acts.2 And this leads to a related worry, that if we restrict ourselves to the consequences that we can reasonably foresee, we are moving away from the raison d’être of consequentialism.3

1 See Williams, “A Critique of Utilitarianism.” 2 This problem is addressed by many consequentialists, including Smart, “An Outline of a System of Utilitarian Ethics”; Norcross, “Consequentialism and the Unforeseeable Future”; Kagan, Normative Ethics. 3 See Lenman, “Consequentialism and Cluelessness.” I return to his argument at the end of this chapter.

219

220    Elinor Mason

In this chapter I suggest a new way to answer these objections. The new wave of consequentialism, as it has been termed, has attempted to deal with these objections with a series of abstract and technical maneuvers, such as introducing agent-​relative elements into the theory of value,4 consequentializing,5 and changing modes and focal points of assessment.6 These approaches all attempt to solve the problem by adjusting consequentialism in some way. I propose an alternative approach. I argue that we can deal with these common objections to consequentialism by attending to a (neglected) constraint on normative theories, that the scope of normative theories is restricted to those actions which we can reasonably be held responsible for. This involves showing three things: (1) that there is an independent, systematic rationale for this responsibility constraint on normative theories; (2) that common objections to consequentialism can be understood/​couched in terms of appeals to what we can reasonably be held responsible for; (3) that consequentialism can embrace the responsibility constraint and thus avoid the objections. 11.1. The Responsibility Constraint

Consequentialist theories have, it is commonly said, two parts, a theory of the good and a theory of the right. It is plausible that theories of the good have nothing to do with responsibility. What’s good is good independently of who or what is responsible for it. Sunny days are good. But an account of rightness seems to have some necessary connection to notions of moral responsibility, though it is not clear precisely what the connection is. One way to put this point is to say that there is a conceptual constraint on normative theories. But the relevant concept is not a well-​defined one. There is room for argument about how exactly to determine and defend the limits on what can count as right or wrong action. However, we do have various platitudes that support the connections between normative notions like rightness and wrongness and responsibility. For example, it is platitudinous to say that if something is impossible for an agent, morality cannot tell her to do that. A theory cannot tell an agent to teletransport, even if that is the only way to save a drowning person.

4 See, e.g., Broome, Weighing Goods; Feldman, Utilitarianism, Hedonism, and Desert; Portmore, Commonsense Consequentialism. 5 See, e.g., Portmore, Commonsense Consequentialism. 6 For example, Kagan, “Evaluative Focal Points”; Driver, Uneasy Virtue; Peterson, The Dimensions of Consequentialism.



Consequentialism and Moral Responsibility    221

This thought has been called the “ought implies can” principle: ​an “ought” only applies if the agent “can” do what is being asked. Unfortunately, the sense of “can” in “ought implies can” is not at all clear. Consider another, familiar, example: the only way to save someone is to beat Karpov at chess.7 I can beat Karpov, in the sense that I can physically make the moves, but I cannot in a more important sense—​I don’t know how. Although “ought implies can” is not a useful way to capture the thought, there is surely a cogent thought here. A normative theory is limited in what it can tell people to do. It is not straightforwardly limited by what is possible for an agent—​the notion of possibility can be spelled out in various ways, as the Karpov example illustrates. Thus the constraint needs to be defined more carefully. I shall be moving toward some clarification in this chapter, but it is worth a rough and noncontroversial formulation for now: The Responsibility Constraint: A normative theory must give an account of right action such that an agent could reasonably be deemed responsible for acting rightly or wrongly. This obviously rules out accounts of consequentialism that instruct the agent to do things that are not possible at all for the agent. But it is not immediately obvious what else is ruled out. The responsibility constraint may also rule out accounts of consequentialism that allow all causal consequences to be relevant to rightness—​rightness cannot depend on consequences I (nonculpably) do not know about or cannot control, because I cannot be responsible for things I do not know about or cannot control. Many consequentialists have embraced this conclusion, arguing that we should focus on the consequences that we can foresee.8 Alastair Norcross, for example, takes it as obvious that there is a responsibility constraint on normative theories: “[T]‌here may be a temptation to abandon consequentialist methods of reasoning on the grounds that what one is able to bring about, or able to be certain to bring about, is somehow dwarfed by the scale of what else is beyond one’s control or beyond one’s epistemic reach. In each case the cure is to focus on what is within one’s control.”9 7 The example is from Howard-​Snyder, “The Rejection of Objective Consequentialism.” I  discuss it in “Consequentialism and the ‘Ought Implies Can’ Principle.” See also Miller, “Actual-​Consequence Act Utilitarianism”; Wiland, “Monkeys, Typewriters, and Objective Consequentialism.” 8 Bentham, An Introduction to the Principles of Morals and Legislation; Mill, Utilitarianism (though Mill’s view is not completely clear); Smart “An Outline of a System of Utilitarian Ethics”; Gruzalski, “Foreseeable Consequence Utilitarianism”; Norcross, “Consequentialism and the Unforeseeable Future”; Jackson, “Decision-​Theoretic Consequentialism”; Oddie and Menzies, “An Objectivist’s Guide to Subjective Value”; Timmons, Moral Theory; Howard-​Snyder, “The Rejection of Objective Consequentialism”; Zimmerman, Living with Uncertainty; and elsewhere. 9 Norcross, “Consequentialism and the Unforeseeable Future,” 256.

222    Elinor Mason

Frank Jackson uses the responsibility constraint to argue for a decision-​theoretic form of consequentialism, or what is now more usually called prospectivism. He says, “[T]he fact that a course of action would have the best results is not in itself a guide to action, for a guide to action must in some appropriate sense be present to the agent’s mind. We need, if you like, a story from the inside of an agent to be part of any theory which is properly a theory in ethics. . . . [T]he passage to action is the very business of ethics.”10 The responsibility constraint, at least in a minimal form, is widely embraced by consequentialists. I shall argue that this is not an ad hoc move, but is based on a gen­ eral constraint on normative theories. I shall also argue that consequentialism can make use of a less minimal form of the responsibility constraint. Before I do that, I shall try to show that the family of objections to consequentialism that I mentioned above has a common thread. In one way or another, they are all complaining that consequentialism makes us responsible for too much. In other words, the complaint is that consequentialism does not respect the responsibility constraint. 11.2. Objections to Consequentialism

Williams complains that consequentialism is committed to making no distinction between what I do and what others do or what just happens. Williams calls this the doctrine of negative responsibility, and this is closely related to the doing/​ allowing distinction. As Williams puts it, “From the moral point of view [for the consequentialist], there is no comprehensible difference which consists just in my bringing about a certain outcome rather than someone else’s producing it.”11 Williams’s thought here is that consequentialism says that we are responsible for what happens, or what others do, but that is too much responsibility—​we are only responsible for what we do. Relatedly, it is clear that part of what is going on in the background of discussions about alienation, or doing and allowing, is a question about agency, about what sort of thing we can be responsible for. Here is Williams on integrity: It is absurd to demand of such a man, when the sums come in from the utility network which the projects of others have in part determined, that he should just step aside from his own project and decision and acknowledge the decision which utilitarian calculation requires. It is to alienate him in a real sense from

Jackson, “Decision-​Theoretic Consequentialism,” 466–​467. 11 Williams, “A Critique of Utilitarianism,” 96, my emphasis. 10



Consequentialism and Moral Responsibility    223

his actions and the source of his action in his own convictions. It is to make him into a channel between the input of everyone’s projects, including his own, and an output of optimific decision; but this is to neglect the extent to which his projects and his decisions have to be seen as the actions and decisions which flow from the projects and attitudes with which he is most closely identified. It is thus, in the most literal sense, an attack on his integrity.12 One way to read this passage is to read Williams as saying that we shouldn’t demand a utilitarian calculation because it would be bad for the agent—​it would ask him to do something that would be damaging to him. Thus we might understand the objection as normative—​an instance of the usual sort of intuitive counterexample to normative theories: strict deontology is counterintuitive because it would tell you to lie to the inquiring murderer; consequentialism is counterintuitive because it tells you to lynch an innocent person to pacify a mob; and so on. We could understand Williams as saying consequentialism is counterintuitive because it tells us to neglect our own projects. On the other hand, we might understand Williams’s objection as conceptual. The demand is absurd because it violates some conceptual requirement—​it is a bit like asking someone to teletransport—this is just not the sort of thing that counts as an exercise of agency. In other words, the utilitarian normative theory violates the responsibility constraint. Here is Williams again, this time on the example where Pedro (at gunpoint) offers Jim a choice between killing one person himself and “allowing” Pedro to kill all twenty: “That may be enough for us to speak, in some sense, of Jim’s responsibility for that outcome, if it occurs; but it is certainly not enough . . . for us to speak of Jim’s making those things happen.”13 Pedro is the person who makes the effects. The point here, again, is that in the right sense of responsibility, which is (of course) not causal responsibility, Jim is not responsible for the deaths. I am not endorsing Williams’s view of what the responsibility constraint says. I think it is controversial—​we need to assess it in the context of our best accounts of responsibility. My point here is to shift the argument away from a purely normative argument to an argument about responsibility. I think the traditional picture neglects the role of accounts of responsibility in thinking about the shape of normative theories. So it is often said, as if it is obvious, that consequentialists cannot countenance a distinction between doing and allowing.14 And it does seem obvious Williams, “A Critique of Utilitarianism,” 116–​117. 13 Williams, “A Critique of Utilitarianism,” 109. 14 See Paul Hurley’s chapter in this volume for more discussion of the doing/​allowing distinction. Hurley accepts that the best consequentialist strategy for accommodating the doing/​allowing distinction is to admit agent-​ relative elements. Neither Hurley nor I are concerned with the issue of how to draw the line between doings and allowings. For discussion of that, see Howard-​Snyder, “Doing vs. Allowing Harm.” 12

224    Elinor Mason

if we think that the issue is a purely normative one, i.e., a question of what sort of moral principle seems most intuitively attractive in the context of familiar thought experiments. Should you kill one to save five? Consequentialists have looked at the consequences and assumed that they must come down on the side that results in fewer deaths. And then, to justify this normative conclusion, they have criticized the distinction between doing and allowing on the grounds that it is basically selfish: to think that it is permissible to allow five to die is squeamish—​it must be a fear of dirty hands, or excessive concern with one’s own acts, or something.15 More recently, consequentialists have argued that we can endorse a doing/​ allowing distinction if we allow some agent-​centered constraints.16 This sort of argument accepts the basic framework as involving a purely normative issue, i.e., as a question about whether it is morally permissible to allow some bad effects. Thus impartiality is abandoned or modified with the justification that, on reflection, it is not the case that morality requires full impartiality after all. Deontologists have similarly accepted that these questions about doing and allowing, impartiality, and demandingness are normative questions, and have argued that the doing/​allowing distinction, or limits on demandingness, can be defended by thinking about some aspect of our normative theory. For example, Philippa Foot and Judith Thomson have both argued that a distinction between positive and negative rights can be used to show that doing harm is worse than allowing harm.17 Thinking about these challenges as being rooted in the responsibility constraint changes the landscape. Consequentialists usually accept that there is such a constraint, and are happy with the restrictions imposed by it when a very thin conception of responsibility is in play. It is open to consequentialists to accept a thicker conception of responsibility, and thus allow, without having to defend agent relativity for its own sake, restrictions on what can be asked of an agent. As with any normative theory, the consequentialist account of right action must limit itself to the realm of what we can reasonably be deemed responsible for. If I cannot be responsible for unforeseeable consequences, they cannot figure in an account of right action. By extension, if I cannot be responsible for what others do, or what happens, or the impersonal good, then those things cannot be part of what makes actions right. Of course this raises two difficult questions: First, what does our best account of responsibility say? Do we have any reason at all to think that we can’t be responsible This is what Williams himself imagines the consequentialist response to be. Arguably this is (an exaggerated version of ) Philip Pettit’s argument in “The Consequentialist Perspective.” 16 See Portmore’s Commonsense Consequentialism for a thorough account of how consequentialism can allow agent relativity. 17 Foot, “Killing and Letting Die”; Thomson, “Killing, Letting Die and the Trolley Problem” and The Realm of Rights. 15



Consequentialism and Moral Responsibility    225

in these more controversial cases? And second, if we limited the consequentialist normative theory more dramatically in response to the responsibility constraint, do we still have a consequentialist theory? I cannot answer these questions thoroughly in this chapter, but I will discuss both. I will argue that there is at least one case, coercion, where the responsibility constraint applies to consequentialism. I will go on to argue that even if we accept a fairly thick reading of responsibility in the responsibility constraint we still have a consequentialist theory. 11.3. Accounts of Responsibility

We all agree with the responsibility constraint understood in a very weak sense. As I said, we all agree that consequentialism cannot tell us to teletransport to save the drowning person. Yet there is disagreement about the doctrine of negative responsibility (the doing/​allowing distinction), impartiality, and the demandingness of consequentialism. We have to think about our conception of responsibility—​how much substance do we get from the responsibility constraint? And the methodology for thinking about it is not obvious. What does it mean to say that it is reasonable to deem an agent responsible for a particular outcome? First we need to distinguish between approaches to thinking about issues of doing/​allowing, or negative responsibility, that do not actually use an account of responsibility at all, and those that do. For an example of the former I will use Frances Kamm’s discussion of Williams on negative responsibility. For an example of the latter I shall use Samuel Scheffler on the doing/​allowing distinction.18 Kamm inverts the order of things in her discussion of Williams:  whereas Williams argues that Jim would not necessarily be responsible for the death of nineteen people if he refuses Pedro’s offer, Kamm argues that Jim is not necessarily responsible for the death of the one Indian if he does shoot.19 Kamm and Williams are both very persuaded by the thought that the responsibility is Pedro’s whatever happens. Kamm claims that Pedro “keeps responsibility” when he is the instigator of the offer. Thus Pedro, rather than Jim is “responsible” for the death of the one. Kamm explicitly makes a distinction between three aspects to responsibility: voluntariness, accountability, and moral responsibility. Voluntariness is the primary condition of responsibility—​libertarians claim that we are responsible in that sense only when we have genuine alternate possibilities and we freely choose between them.

Kamm, “Responsibility and Collaboration” and Intricate Ethics; Scheffler, “Doing and Allowing.” 19 This discussion comes from my “Coercion and Integrity,” where I discuss Kamm’s argument in more detail. 18

226    Elinor Mason

Compatibilists claim that we are free in that sense when we have the right sort of quality of will, and that alternate possibilities are not required.20 Accountability is a quite different notion: it is about when an agent should be held responsible (or accountable), when it is appropriate to punish or reward the agent. Voluntariness is a precondition for accountability, but beyond that, variations in accountability can be explained by something other than variations in voluntariness. We might not be accountable for merely self-​regarding acts, for example, or for certain sorts of harm to others—​not because our acts are not voluntary, but because of the moral rules surrounding those sorts of acts. This leaves the third element, “moral responsibility.” Kamm is not at all clear what she means by that. In fact, in Kamm’s descriptions of the various cases, moral responsibility and accountability never come apart, so in Kamm’s discussion the notion of moral responsibility is redundant. As a consequence, Kamm’s story is all about accountability, which we could understand as depending purely on the moral rules that are in play in the situation. Kamm is providing an account of how instigation has an effect on the moral situation, in just the way that promises have an effect on the moral situation. If I make you a promise I am bound to you in a new way—​I have new duties. Similarly, one might think, Pedro’s instigation of the offer excuses Jim from duties of reparation that he would otherwise have had when he kills someone. Thus Kamm’s account is not really about responsibility; rather it is about the normative rules surrounding accountability.21 By contrast, Scheffler argues that an account of responsibility can provide independent support for a doing/​allowing distinction.22 Scheffler argues that in seeing oneself as an agent at all, one has to accept that doing—​a primary exercise of agency—​is more important than allowing, which is just a secondary exercise of agency. Thus Scheffler thinks that being a responsible agent has a conceptual link to accepting the doing/​allowing distinction. Scheffler’s argument is complex, and I will not go further into the details. I mention it only briefly to draw attention to the form. Scheffler, like Williams, thinks that denying a doing/​allowing distinction is problematic because it violates the responsibility constraint. The idea that quality of will is a descriptive and natural property that can be used to make sense of moral responsibility is the foundational compatibilist idea that we find in Hobbes, “Of Liberty and Necessity,” and Hume, “Of Liberty and Necessity.” The basic idea is developed by both Strawson (“Freedom and Resentment”) and Frankfurt (“Freedom of the Will and the Concept of a Person”) in articles that have been very influential on contemporary compatibilists. In what follows I rely chiefly on Frankfurt’s account. Frankfurt focuses on quality of will: the identification of the agent with her act. 21 To be fair to Kamm, some of what she says suggests that she does think there is more to Jim’s lack of responsibility than lack of accountability. What I say below could be seen as expanding on some of Kamm’s remarks about Jim’s being the Captain’s agent. 22 For criticism of Scheffler’s view, see Bradley and Stocker “ʻDoing and Allowing’ and Doing and Allowing.” 20



Consequentialism and Moral Responsibility    227

My own suggestion is somewhat simpler than Scheffler’s. My argument draws on Frankfurt’s account of responsibility, according to which an agent is responsible when she identifies with her act in the right way.23 The idea, in brief, is that we simply cannot identify with our acts under certain descriptions. I start with coercion and argue that we cannot identify with acts that are coerced. This opens the door to arguments that we cannot identify with acts described as productions of very distant consequences, or as allowings rather than doings. Returning to Williams’s example: we can see what happens to Jim as coercion. When an agent’s will is dominated by another agent, as Jim is by Pedro, we can say that the coercee is not responsible because he does not bear the right relationship to his action. The agent cannot identify with his action in the right way. This is so even though he acts (in some sense) voluntarily and, we can say for the sake of argument, for good reasons. The problem with Jim is that when Pedro has dominated his will, his action is not his own in the right way for him to be responsible for it. This is not a precise account of the problem: obviously there is more to be said. I am just trying to illustrate the fact that there are accounts of responsibility that lend themselves readily to this approach. So even a consequentialist can say that Jim is not responsible when he kills one person in response to Pedro’s threat. The next step is to go back to right action. If Jim is not responsible when he is coerced, what does that say about whether an account of right action can tell him what to do in that situation? The responsibility constraint links right action to responsibility—​an account of right action must be such that the agent could reasonably be deemed responsible for that action. So if the action is not one that the agent could reasonably be deemed responsible for, it cannot be right or wrong. Does that mean the moral theory has nothing at all to say? The situation with coercion is rather complicated, as in cases like Jim’s the coercer is manipulating Jim’s reasons: i.e., the coercer is to some extent taking Jim’s point of view in order to know what the reasons that Jim will respond to are. Jim thinks (as most people think) that it would be better for one person to die and nineteen people to live than for twenty to die. Pedro has, without Jim’s consent, given Jim a causal role in the outcome. At this point, Jim’s agency has been overridden, and so if the responsibility constraint applies here, the normative theory cannot tell him what to do. Yet, in some important sense it seems that Jim should shoot, just that when he does, he is not responsible. How can this be? One thing we might mean when we say that Jim is not responsible is simply that it is permissible for him to shoot. This was how I interpreted Kamm’s use of the notion of responsibility here: Kamm is not See Frankfurt, “Freedom of the Will and the Concept of a Person” for Frankfurt’s original statement of this idea, which he develops in some of the essays in The Importance of What We Care About. 23

228    Elinor Mason

really saying that Jim has ceased functioning as an autonomous agent; she is saying that the normative situation is very complex. However, I want to say more, so I face this puzzle—​on the one hand Jim should shoot; on the other, he is not functioning as a fully autonomous agent when he does. Coercion, as I am understanding it in this case, involves one person setting up the reasons so that it is overwhelmingly reasonable for the coercee to do what the coercer wants to get her to do.24 There may be other sorts of coercion, where the coercer manipulates the coercee using nonrational means, but in this sort of case, the coercion depends on the coercee’s responsiveness to reasons. Additionally, in order for the coercer’s offer to be coercive, it must make the reasons very powerful. A small balance of reasons in favor of the coercer’s preferred option is not coercion. It must be that the coercee would be being unreasonable to refuse the offer—​would be doing something that is obviously crazy, to put it more colloquially. So part of my claim, in claiming that the coercee’s agency is lost here, is that the agent can be responsive to reasons and yet not fully functioning as an agent.25 The background account of responsibility goes beyond reasons-​responsiveness and involves something like identification as well. Jim is responding to reasons when he accepts Pedro’s offer. So what sort of reasons are in play here? The puzzle is that if there is no responsibility, there cannot be a directive from the normative theory—​just as the normative theory cannot tell an agent to produce the best consequences in cases where the consequences are unforeseeable. This case is similar, in that the agent’s actions will affect a state of affairs that clearly has value. Unforeseeable consequences have value; even the consequentialist normative theory, limited by the responsibility constraint, must remain silent about what the agent should do (it can’t say, “Produce the best consequences,” because that violates the responsibility constraint). But this case is trickier, because Jim can see the reasons, and in some sense he can act in accordance with them or against them, and so it seems that consequentialism tells him to do what will have the best consequences here just as it does elsewhere. My view is that consequentialism does not tell Jim what to do in response to Pedro’s offer. Jim is put in a position where his agency is compromised, and so the normative theory must bow out, so to speak. But the theory of value remains untouched—​it is still the case that it is better for nineteen people to live than for none to live. And this is how the coercion works in this case: Pedro puts Jim in the causal chain. Jim can see that complying with Pedro’s request results in more value, Note that the situation in coercion is not necessarily that the coercer wants the coercee to do it—​Pedro might not care either way; Pedro’s aim with respect to controlling Jim is not necessarily his overall aim—​he wants to get Jim to kill someone. The aim is domination. There is another thorny issue that I do not discuss here, which is how to distinguish between threats and offers (and whether offers can be coercive). 25 Thus my account is not compatible with reasons-​responsiveness accounts of moral responsibility, like that presented by Wolf in Freedom within Reason or by Fischer and Ravizza in Responsibility and Control. 24



Consequentialism and Moral Responsibility    229

so it would be unreasonable of him to opt out. But he does not bear the right relationship to his action for it to count as right or wrong. Before I go further I should be explicit about my methodology here. My argument is that there is a distinction between acting wrongly and acting unreasonably, and in the case of Jim, it would be unreasonable but not wrong if he refuses Pedro’s offer. This is not supposed to be an argument that makes sense of all our pretheoretical intuitions about the case. It is a philosophical fix. But I am not perturbed by the somewhat awkward fit with our pretheoretical intuitions about the case and about the available vocabulary. First, the area of responsibility is one where we are particularly unreliable in our intuitions, as many commentators have pointed out.26 Second, the terms ‘rightness’ and ‘wrongness’ are used in many different ways, and some of those ways are very broad, such that there is no relationship between rightness and wrongness and responsibility—​the terms are used as if they are coextensive with ‘badness’ and ‘goodness’. Inevitably, then, a narrower usage of those terms will conflict with some linguistic intuitions. I am proposing a narrow and in some ways odd revision of our ordinary use, but (I hope) this is a more fruitful use for those terms.27 My claim again, then, is that Jim would be acting unreasonably but not wrongly if he refuses Pedro’s offer. He would be acting unreasonably because there are good reasons to accept the offer, namely the fact that nineteen people will live. But Jim cannot identify with his action, he cannot own it in the right way, as Pedro has dominated his will. So the reasons are not rightness-​and wrongness-​making reasons. Jim does not bear the right relation to the reasons for them to have that status. But they are nonetheless reasons, and Jim would be unreasonable if he ignored them. This is somewhat analogous to what we might say about reasons of prudence. If we have a view that leaves self-​interested action out of the realm of right and wrong, it is clear that someone can act against her reasons without rendering her action wrong. We might further think that reasons of prudence are reasons of value: my welfare matters morally; it provides me with reasons, but it does not provide me with rightness-​and wrongness-​making reasons. That’s because of the relation I bear to those reasons: according to a fairly plausible view, when the welfare is my own I have a special relationship to the reason-​producing power of that welfare. Similarly

See particularly Vargas, Building Better Beings. 27 The idea is that the theory of the good determines what is reasonable and unreasonable across the whole range of actions, while the concepts of right and wrong apply only within that subdomain of actions for which we can reasonably be held responsible—​they are inapplicable outside that domain, so it is not true that it would be right for Jim to accept Pedro’s offer, but it would not be wrong either. One implication of this view, as Christian Seidel pointed out to me, is that a common principle fails: it is not true that for all actions, either that action is permissible or impermissible. Rather the deontic categories apply only to a subclass of actions (and so the principle is true only within that subclass). I accept this implication. 26

230    Elinor Mason

here, the lives of the nineteen people matter, and produce reasons, but in Williams’s example Jim does not have the right sort of relationship to the reason-​producing power of those nineteen lives for them to be rightness-​and wrongness-​making. I have argued that Jim would be unreasonable but not acting wrongly in refusing Pedro’s offer. He might also be morally bad, even though he would not have acted wrongly. One can be morally bad without acting wrongly. One obvious way is through having bad motivations that never result in action. The situation where Jim could justly be accused of moral badness because he refuses the offer is similar in that it depends on Jim’s motivations, not his action. If Jim refuses the offer through cowardice, he is somewhat bad. If he refuses through a desire for revenge on the local community for some past slight that Pedro knows nothing about, then he is very bad indeed. However, if he refuses because he finds himself constitutionally unable to kill a person, he is not bad at all. I will make one final remark about this argument. It illustrates one clear superiority of consequentialism over pure deontology. By “pure deontology” I mean the view that actions are right or wrong simply as actions, and rightness is not based on consequences at all. Imagine a pure deontologist in Jim’s situation. Her theory tells her not to kill. It doesn’t tell her anything else. A pure deontology (by stipulation) has only a theory of right action, and no theory of the good. So it has no resources for saying anything at all about other ethically relevant features of a situation. What the consequentialist says is, on my picture, very complex. But in this sort of case the simplicity of pure deontology seems like naiveté. Deontic categories are very limited: we need a theory of the good and an account of the nondeontic ways that a theory of the good can recommend action that goes beyond deontic categories. In sum, in cases where the agent cannot be responsible for an action, that is not the sort of action that can be right or wrong. That’s just what the responsibility constraint says, and it is not controversial. I  have argued that the responsibility constraint may take us into territory that is usually thought of as purely normative, and so may give consequentialism resources to, for example, admit a doing/​allowing distinction. Of course, coercion is a very complex case, and my argument here depends essentially on the presence of an intervening agent. What should we say about cases where some see a doing/​allowing distinction and there is no intervening agent? Or the cases I mentioned earlier, where the more general issue of whether we can be deemed responsible for the impersonal good, demanding as that is? This is work still to be done; as I said, here I am only hoping to open the door. 11.4. Cluelessness and Swamping

All this brings us to the worry that the responsibility constraint is actually fatal to consequentialism as a normative theory. Consequentialism has two parts, a theory



Consequentialism and Moral Responsibility    231

of the good and a theory of the right. The theory of the good has something to say about everything—​the consequences we cannot control, the ones we cannot foresee, the ones we are not properly agents with respect to, and so on. But the theory of the right is much more limited; it is limited to those consequences that we can reasonably be deemed responsible for. But now we have a tension. Those other consequences still matter, and we are causally related to them, so our normative theory, in directing us only regarding consequences we can be responsible for, has lost sight of its raison d’être. This is James Lenman’s worry in his discussion of the epistemic objection. As Lenman points out, it seems that consequentialists, at least “pure consequentialists,” should be concerned with all consequences. And our acts have massive casual ramifications, ramifications, that, given our theory of value, matter as much as anything we can predict. On the one hand consequentialists say that what matters is all these distant states of affairs; on the other hand they say that what we should do is guided by a tiny corner of the vast plain of what matters. This is the swamping problem. It is worth distinguishing between the original epistemic objection and Lenman’s “swamping” objection. The original objection is that we don’t know what the consequences of our actions will be. And the answer to that is straightforward—​ consequentialists agree that there is a responsibility constraint, and so consequentialism concerns only what an agent can be responsible for; in this context let’s use Lenman’s terminology, and call that “visible consequences.” But Lenman has a further worry: A consequentialist might seek to agree to this limiting of moral focus by again taking the line that, in matters of assessing the moral or rational merits of actions, of assigning praise and blame, we should concern ourselves with subjective rightness, for subjective rightness is precisely concerned only with visible consequences. The trouble is that, as I have noted, a consequentialist must understand this concern as motivated by the belief that maximizing value with respect to visible consequences is a reliable means to maximizing value with respect to overall consequences. And this belief does not appear at all secure. Given this, we might prefer a theory that tells a different story about what the point is of our concern with visible consequences. And such a story would precisely not be consequentialist.28 It is important to get the deep point clear here. Lenman grants the indifference postulate—​the idea that it is rational to assume that the positive and negative distant Lenman, “Consequentialism and Cluelessness,” 365. 28

232    Elinor Mason

consequences cancel each other out—​for the sake of argument. If we are “entitled” to the indifference postulate, then all we have to count are visible consequences. But we need to be clear about what we mean when we say that we are entitled to an indifference postulate: if we mean that we are entitled to it because the probabilities really do suggest that the value of the unforeseeable future is neutral, then in granting us an indifference postulate Lenman is granting us the solution to his problem.29 If, however, we mean that we will use it because we don’t have anything else to go on, then we have solved the epistemic problem, but not the swamping problem.30 Those consequences are out there, and by consequentialist lights they should be guiding us. They matter, and if consequentialism is a normative theory that makes rightness depend on consequences, it is woefully short of living up to its name. Lenman says that a consequentialist must think that maximizing value with respect to visible consequences is a reliable means to maximizing value with respect to overall consequences. Obviously the consequentialist should not say this. The reason to maximize the value of visible consequences is not that that will entail maximizing overall; clearly that entailment does not hold. So what is it? What we don’t want to say is that we maximize the value of visible consequences because, given the responsibility constraint, that is all we can do. That’s exactly the view that makes consequentialism seem lame—​it is like saying, “I want to go to Spain, but it’s really too far, so I will just go as far as the railway station.” What we ended up doing bears very little relation to what we started out wanting to do. Here is another way to put the problem: as Lenman points out, all plausible moral theories are concerned with visible consequences. But it seems that the consequentialist rationale for being concerned with only the visible consequences is unsatisfactory. We might as well be deontologists, or virtue ethicists, and admit that our concerns are rather short range, that what is close to us is more morally relevant, and so on. Lenman points out a connection between the epistemic objection and Williams’s integrity objection: he argues that in allowing too much into the moral picture, we lose track of what a human life is really about—​the individual is lost in the massively expanded ethical sphere.31 Smart in “An Outline of a System of Utilitarian Ethics,” Kagan in Normative Ethics, and Norcross in “Consequentialism and the Unforeseeable Future” all seem to be answering the epistemic objection in this way. 30 Lenman, “Consequentialism and Cluelessness,” 357. See also Burch-​Brown, “Clues for Consequentialists,” 110, for a very clear statement of this distinction. In my earlier paper on this topic, “Consequentialism and the Principle of Indifference,” I argue for an indifference postulate that admits ignorance, but I try to show that there are too many companions in guilt to worry about it. I now see that more needs to be said in response to the swamping objection. One option for consequentialists would be to insist that the right action really is the one that has the best consequences and argue that the Responsibility Constraint is not important. I do not think that this is a promising strategy. 31 Lenman, “Consequentialism and Cluelessness,” 370. 29



Consequentialism and Moral Responsibility    233

But this looks like an appeal to the responsibility constraint, and if it is, then consequentialism and other ethical theories are on very similar ground. The following passage from Lenman makes it look very much as though he thinks that any theory would be concerned with all consequences if that were consistent with the responsibility constraint: What is common ground to all plausible ethical theories is the moral significance of visible consequences. When we can foresee harm to others in the outcome of our actions, we owe them the respect of taking this properly into account. And we owe it to others also to be adequately conscientious in foreseeing such harm. Of course, the invisible consequences of action very plausibly matter too, but there is no clear reason to suppose this mattering to be a matter of moral significance any more than the consequences, visible or otherwise, of earthquakes or meteor impacts (although they may certainly matter enormously) need be matters of, in particular, moral concern.32 Notice that this is exactly what I have argued a consequentialist can say: consequences we are not responsible for still matter—​the theory of the good has something to say about them, but they cannot be a part of normative theory because they are not the sort of thing we can be responsible for. So it seems that both Lenman and the responsibility-​aware consequentialist want to say the same thing. All consequences matter, but not all can be the subject of normative theory. So it looks as though consequentialism and nonconsequentialism are in the same boat with respect to the responsibility constraint. Of course the nonconsequentialist could say that consequences we are not responsible for don’t matter. But that’s a very odd view.33 For both consequentialism and nonconsequentialism, unforeseeable consequences matter. The contrast between the consequentialist and the nonconsequentialist lies in the other sorts of thing that matter. According to nonconsequentialists, consequences matter, but other things matter too—​some actions are right or wrong just for their own sake, or because of their relationship to character and virtue rather than to consequences. Thus understood, the problem for the consequentialist rationale in relation to the responsibility constraint must be that whereas for nonconsequentialists, the responsibility constraint just limits one part of the theory, for consequentialism, the responsibility constraint limits the whole theory. 32 Lenman, “Consequentialism and Cluelessness,” 365. 33 Pure deontologists don’t have to claim that distant consequences don’t matter, in that there is no contrast with the consequences that do matter—​for pure deontology, morality is just about action.

234    Elinor Mason

But put like this, the swamping problem does not seem like a very serious worry. Since it is still only consequences that matter, there are no grounds for thinking that consequentialism is undermined by the responsibility constraint. It is simply limited by it. But we always knew that—​all theories are limited by it. A limit is not necessarily a bad thing, and not necessarily in tension with the thing it limits: a party that must end by midnight is still a party. Consequentialism is a normative theory like any other, and comes with limits. 11.5. Conclusion

There is a responsibility constraint on consequentialism, and that might shape the theory more radically then we have hitherto supposed. I  have not attempted to say how much, or in what direction the responsibility constraint can limit consequentialism; that is work to be done. I have argued that there is nothing sinister or undermining about the responsibility constraint; we don’t have to worry that the rationale for consequentialism is swamped if we accept the constraint. We may end up with a theory that has more in common with nonconsequentialist views than traditional versions of consequentialism, because in accepting the responsibility constraint we allow fewer consequences into the normative theory. However, such a view is still a distinctively consequentialist view insofar as consequentialism allows only consequences into the normative theory.34 References Bentham, Jeremy. An Introduction to the Principles of Morals and Legislation (1789). Edited by James H. Burns and H. L. A Hart. Oxford: Oxford University Press, 1996. Bradley, Ben, and Michael Stocker. “‘Doing and Allowing’ and Doing and Allowing.” Ethics 115 (2005): 799–​808. Broome, John. Weighing Goods: Equality, Uncertainty and Time. Oxford: Wiley-​Blackwell, 1991. Burch-​Brown, Joanna M. “Clues for Consequentialists.” Utilitas 26 (2014): 105–​119. Driver, Julia. Uneasy Virtue. Cambridge, UK: Cambridge University Press, 2001. Feldman, Fred. Utilitarianism, Hedonism, and Desert:  Essays in Moral Philosophy. Cambridge, UK: Cambridge University Press, 1997. Fischer, John Martin, and Mark Ravizza. Responsibility and Control:  A Theory of Moral Responsibility. Cambridge, UK: Cambridge University Press, 1998. Thanks to Luc Bovens, Christian Seidel, Andreas Schmidt, and Steve Sverdlik for comments. I am also grateful to the audience at the University of Pennsylvania Law School, where an earlier draft of this paper was presented, and to the University Center for Human Values at Princeton University for a Laurence S. Rockefeller Visiting Fellowship. 34



Consequentialism and Moral Responsibility    235

Foot, Philippa. “Killing and Letting Die.” In Killing and Letting Die, edited by Bonnie Steinbock and Alastair Norcross, 280–​289. New York: Fordham University Press, 1994. Frankfurt, Harry G. “Freedom of the Will and the Concept of a Person.” Journal of Philosophy 68 (1971): 5–​20. Frankfurt, Harry G. The Importance of What We Care About: Philosophical Essays. Cambridge, UK: Cambridge University Press, 1988. Gruzalski, Bart. “Foreseeable Consequence Utilitarianism.” Australasian Journal of Philosophy 59 (1981): 163–​176. Hobbes, Thomas. “Of Liberty and Necessity” (1646). In Hobbes and Bramhall on Liberty and Necessity, edited by Vere Chappell, 15–​42. Cambridge, UK: Cambridge University Press, 1999. Howard-​ Snyder, Frances. “The Rejection of Objective Consequentialism.” Utilitas 9 (1997): 241–​248. Howard-​Snyder, Frances. “Doing vs. Allowing Harm.” In Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Winter 2011 edition. http://​plato.stanford.edu/​archives/​win2011/​ entries/​doing-​allowing/​. Hume, David. “Of Liberty and Necessity” (1748). In An Enquiry concerning Human Understan­ ding and Other Writings, 73–​91. Cambridge, UK: Cambridge University Press, 2007. Jackson, Frank. “Decision-​Theoretic Consequentialism and the Nearest and Dearest Objection.” Ethics 101 (1991): 461–​482. Kagan, Shelly. Normative Ethics. Boulder, CO: Westview Press, 1998. Kagan, Shelly. “Evaluative Focal Points.” In Morality, Rules and Consequences: A Critical Reader, edited by Brad Hooker, Elinor Mason, and Dale E. Miller, 134–​155. Edinburgh: Edinburgh University Press, 2000. Kamm, F. M. “Responsibility and Collaboration.” Philosophy and Public Affairs 28 (1999): 169–​204. Kamm, F. M. Intricate Ethics: Rights, Responsibilities, and Permissible Harm. New York: Oxford University Press, 2007. Lenman, James. “Consequentialism and Cluelessness.” Philosophy and Public Affairs 29 (2000): 342–​370. Mason, Elinor. “Consequentialism and the ‘Ought Implies Can’ Principle.” American Philosophical Quarterly 40 (2003): 319–​331. Mason, Elinor. “Consequentialism and the Principle of Indifference.” Utilitas 16 (2004): 316–​321. Mason, Elinor. “Coercion and Integrity.” In Oxford Studies in Normative Ethics, vol. 2, edited by Mark Timmons, 180–​205. Oxford: Oxford University Press, 2012. Mill, John Stuart. Utilitarianism (1861). In The Collected Works of John Stuart Mill, vol. 10: Essays on Ethics, Religion and Society, edited by John M. Robson, 203–​260. Toronto: University of Toronto Press, Routledge & Kegan Paul, 1985. Miller, Dale E. “Actual-​Consequence Act Utilitarianism and the Best Possible Humans.” Ratio 16 (2003): 49–​62. Norcross, Alastair. “Consequentialism and the Unforeseeable Future.” Analysis 50 (1990): 253–​256. Oddie, Graham, and Peter Menzies. “An Objectivist’s Guide to Subjective Value.” Ethics 102 (1992): 512–​533. Peterson, Martin. The Dimensions of Consequentialism. Cambridge, UK: Cambridge University Press, 2013. Pettit, Philip. “The Consequentialist Perspective.” In Three Methods of Ethics: A Debate, edited by Marcia Baron, Philip Pettit, and Michael A. Slote, 92–​174. Oxford: Blackwell, 1997.

236    Elinor Mason Portmore, Douglas W. Commonsense Consequentialism:  Wherein Morality Meets Rationality. Oxford: Oxford University Press, 2011. Scheffler, Samuel. “Doing and Allowing.” Ethics 114 (2004): 215–​239. Smart, J. J. C. “An Outline of a System of Utilitarian Ethics.” In Utilitarianism: For and Against, edited by J. J. C. Smart and Bernard Williams, 3–​74. Cambridge, UK: Cambridge University Press, 1973. Steinbock, Bonnie, and Alastair Norcross, eds. Killing and Letting Die. 2nd edition. New York: Fordham University Press, 1994. Strawson, Peter F. “Freedom and Resentment.” Proceedings of the British Academy 48 (1962): 1–​25. Thomson, Judith Jarvis. “Killing, Letting Die and the Trolley Problem.” In Rights, Restitution, and Risk:  Essays in Moral Theory, edited by W. Parent, 78–​93. Cambridge, MA:  Harvard University Press, 1986. Thomson, Judith Jarvis. The Realm of Rights. Cambridge, MA: Harvard University Press, 1990. Timmons, Mark. Moral Theory: An Introduction. Lanham, MD: Rowman & Littlefield, 2002. Vargas, Manuel. Building Better Beings:  A Theory of Moral Responsibility. Oxford:  Oxford University Press, 2013. Wiland, Eric. “Monkeys, Typewriters, and Objective Consequentialism.” Ratio 18 (2005): 352–​360. Williams, Bernard. “A Critique of Utilitarianism.” In Utilitarianism: For and Against, by J. J. C. Smart and B. Williams, 77–​155. Cambridge, UK: Cambridge University Press, 1973. Wolf, Susan. Freedom within Reason. Oxford: Oxford University Press, 1990. Zimmerman, Michael J. Living with Uncertainty. Cambridge, UK:  Cambridge University Press, 2008.

12 KA NT I A NI S M , CONSE QUE NTIA LISM, AN D D ET ER R EN C E

Steven Sverdlik

Comprehensive moral theories like consequentialism are evaluated in part by their application to specific moral practices and institutions. Consequentialism began in the eighteenth century in the form of utilitarianism, and the early utilitarianism of Beccaria and Bentham had a marked focus on legal punishment and its reform.1 In the same era Kant was developing his moral and political theory, and it was in the context of his treatment of legal punishment that he made some of his sharpest attacks on utilitarianism.2 These attacks have echoed down to the present and have had a deep influence on the contemporary philosophy of punishment. Certainly consequentialism has had an influence in this field, especially in the work of H. L. A. Hart.3 But it is fair to say that retributivism, the theoretical approach largely deriving from Kant, has been dominant since the 1970s. This chapter is designed to help restore consequentialism to a leading role in our thinking about the moral foundations of the practice of legal punishment.

1 Beccaria’s On Crimes and Punishments is one of the sources of the phrase “the greatest happiness of the greatest number.” It had a significant impact on Bentham. Bentham spent much time and effort in his earlier career thinking about the criminal law and the design of prisons. Semple, Bentham’s Prison. 2 Kant, Metaphysics of Morals, 140–​144 (= 331–​335 of the Prussian Academy edition). Kant criticizes Beccaria by name on 143–​144 (334–​335), but he makes other arguments against utilitarianism on 140–​141 (331–​333). 3 Hart, Punishment and Responsibility. See also Braithwaite and Pettit, Not Just Deserts.

237

238    Steven Sverdlik

The particular aspect of legal punishment that we will focus on is its use to deter potential criminals from committing crimes. Kant and his followers have argued that the use of punishment to deter people from committing crimes is morally objectionable in principle. Consequentialism is often equated with the claim that the only legitimate goal of punishment is deterrence. This is false, because the theory supports the use of punishment to achieve, among other things, the moral reform of the offender, incapacitation, and the expression of the community’s moral condemnation of certain actions.4 But consequentialism certainly does support the claim that deterrence is an important and legitimate goal of punishment. We will explore the degree to which the most plausible versions of a Kantian approach to punishment differ from consequentialism with regard to the moral legitimacy of deterrence. I argue that they do not differ. After defending my interpretation, I go on to consider a recent theory of punishment developed by Victor Tadros. Tadros appeals to a moral principle that deliberately utilizes Kantian phraseology: the Means Principle. Tadros uses this principle to criticize consequentialism. I show that the Means Principle is not a plausible moral principle by using arguments that derive from my interpretation of Kant. The Means Principle is therefore unsuited to be the way to frame a discussion of whether consequentialism is acceptable. I discuss in section 12.4 a version of Kantian moral theory that explicitly recognizes that the pursuit of deterrence is morally legitimate. This helps us to understand the moral significance of deterrence in what Rawls called “the circumstances of justice.” In the final section I note one reason for thinking that further features of the consequentialist approach to deterrence make it superior to the Kantian-​R awlsian. We have reason to think that consequentialism should resume its former position at the center of the philosophy of punishment. 12.1. Objecting to Deterrence

The idea that punishment may permissibly be used to deter potential criminals from committing crimes is accepted by moral common sense, I  believe. It is true that moral common sense accepts that there are some limits on the pursuit of deterrence. For one thing, at some point an increase in the severity of punishment, even if it deterred more potential criminals, would be rejected as morally unacceptable. So, for example, most people would object to imposing thirty years of prison on an offender who was convicted of stealing one laptop, even if it could be shown 4 Sverdlik, “Punishment and Reform.”



Kantianism, Consequentialism, and Deterrence    239

that doing this resulted in fewer incidents of this crime. However, we will mainly be examining the claim that the pursuit of deterrence is objectionable in principle. There is a second argument related to deterrence that we will also not examine. This is the well-​known objection to consequentialism that asserts that this theory must grant that it is sometimes morally permissible to punish someone known to be innocent if that would deter potential offenders.5 The “in principle” objection to deterrence applies even to the punishment of the guilty. The objection in principle is often attributed to Kant. This stems from the way that he sometimes formulates the supreme principle of morality, the Categorical Imperative. Here is his statement of the relevant formulation, the Formula of Humanity: FH: Act in such a way that you always treat humanity, whether in your own person or in the person of any other, never simply as a means, but always at the same time as an end.6 This asserts, in part, that it is morally wrong to treat people simply as a means. The objection would be that in punishing a person to achieve deterrence we are doing precisely that. Consider “general deterrence,” that is, punishing an offender who has committed a crime in order to induce other people to refrain from doing so. If this is our aim in punishing the offender—​the objection goes—​then we are inflicting pain or a loss on her simply as a means of influencing the behavior of other people. She is being used like a scarecrow, a mere thing, and this, it is claimed, is always morally wrong. Did Kant himself actually make this objection? Scholars have disagreed about the answer to that question. The most explicit relevant passage in his published works says this: “Punishment . . . can never be inflicted merely as a means to promote some other good for the criminal himself or for civil society. . . . For a man can never be treated merely as a means for the purposes of another or be put among the objects of rights to things: His innate personality protects him from this.”7 This certainly seems to be stating that we may not punish a person simply to further the ends of others. But, as A. C. Ewing noted, it is also stating that we may not inflict punishment on an offender in order to promote her own good.8 So Kant seems to be objecting to more than the pursuit of general deterrence or, indeed, any form of deterrence.9 5 In Sverdlik, “The Origins of the ‘Objection,’ ” I show that this objection to utilitarianism does not occur in Kant, and was developed much later. 6 Kant, Groundwork, 96 (= 429 in the Prussian Academy Edition). 7 Kant, The Metaphysics of Morals, 140–​141 (331). 8 Ewing, The Morality of Punishment, 17n1. 9 For an attempt to find some place for deterrence in Kant’s own texts, see Byrd, “Kant’s Theory of Punishment.”

240    Steven Sverdlik

The argument that Kant seems to make has appeared often in the literature on punishment since the 1970s and helped to give it a markedly retributivist orientation. A  number of important contemporary writers, such as Antony Duff, Andrew von Hirsch, and (at one time) Jean Hampton, endorse the argument, or something closely related.10 What is more, it seems to be implicitly accepted by Robert Nozick and Michael Moore, among others, who try to develop entirely retributive theories of punishment where deterrence has no role.11 Finally, something like the Formula of Humanity appears in the most substantial recent work in the philosophy of punishment, Victor Tadros’s The Ends of Harm. In this work Tadros develops a theory of punishment, the “duty view,” that he claims is neither retributivist nor consequentialist. Tadros asserts that a certain amount of general deterrence is morally permissible, so he does not object to it in principle. But it is not permissible on consequentialist grounds. In other words, he accepts that punishment may be designed to achieve the goal of deterrence, but this permission derives from a nonconsequentialist moral theory. Tadros accepts what he calls the Means Principle, and he uses this to object to consequentialism. This principle can be formulated as follows. (As I explain in the note we need to add the bracketed word to Tadros’s own formulations.) MP: If an action harms a person [simply] as a means of producing good for other persons then it is not permissible.12 MP has some similarities to the Formula of Humanity (FH), and Tadros reinforces our sense of the connection by arguing that MP can be supported by appealing to the Kantian idea that persons have a special moral status. This status, he says, demands that we not be regarded as available for the use of others. Echoing Kant, Tadros says that we are not so available because “as humans, we are capable of determining what ends to set for ourselves.”13 Duff, Trials and Punishments, 178–​186; Duff, Punishment, Communication, and Community, 82–​85; von Hirsch, Censure and Sanctions, 11–​13; Hampton, “The Moral Education Theory of Punishment,” 211f. Duff, von Hirsch, and Hampton appeal to Hegel’s claim that deterrent threats treat rational agents “like a dog.” Hegel, Elements of the Philosophy of Right, 125. The focus of Kant’s remarks in The Metaphysics of Morals is the act of imposing punishment; Hegel’s is on the threat of its imposition. But both are asserting that features of deterrence constitute a deep affront to an agent’s status as a rational being. 11 Nozick, Philosophical Explanations, 363–​397, esp. 372; Moore, Placing Blame, 27–​30, 153–​188. 12 Cp. Tadros, The Ends of Harm, 13, 114, 115, where the principle is stated without “simply.” It is clear that Tadros accepts part of the idea that it is usually understood to convey. See 135–​136, where he grants that it is permissible to harm a person who consents to sacrifice her life for the sake of others. Taking Tadros to endorse MP as I have formulated is charitable, since it makes it less vulnerable to counterexample. 13 Tadros, The Ends of Harm, 127. On Tadros’s debt to Kant, see Tadros, The Ends of Harm, 2. For Kant’s emphasis on the ability to set goals, see Groundwork, 105 (437); The Metaphysics of Morals, 195 (392). On the other hand, 10



Kantianism, Consequentialism, and Deterrence    241

Tadros does not believe that MP is absolute; he thinks that there are cases where a person may permissibly be harmed in order to benefit others. And he argues that in some cases deterrent punishments can be shown to be permissible because they fall under such an exception to MP. But MP puts the practice of punishment under a serious cloud of suspicion, and it is said to be based on Kantian ideas about the permissible treatment of rational agents. I will consider whether Tadros can use MP to rule out the pursuit of general deterrence in the design of a system of punishment—​unless we can find an exception to it that a nonconsequentialist moral theory allows for. 12.2. FH and Deterrence

In this section we examine Kant’s Formula of Humanity, with an eye on its implications for the use of punishment to achieve deterrence. The interpretation of this principle is controversial. I will first consider a natural interpretation of it which entails that punishment aimed at deterrence is wrong. I will then present a second and, I believe, more plausible interpretation that can be used to show that punishment aimed at deterrence is permissible.14 We can begin by recalling that Kant famously distinguished between acting “in conformity with duty” and acting “from duty.” And he went on to assert that a merchant, who returns the correct change to someone from self-​interest, acts in conformity with duty.15 We can take him to mean that the merchant carries out her moral obligation, although she does not act from a sense of duty. This suggests that Kant accepts this statement, which we can call the Objectivity of Obligation: OO: An action which would otherwise be obligatory does not lose that status because of the motivational characteristics of the agent performing it.16 Tadros sometimes implies that his Means Principle is a variant of the Doctrine of Double Effect, which has a rather different intellectual provenance and basis. Tadros, The Ends of Harm, 116, 150. The Doctrine of Double Effect clearly distinguishes between foreseen and intended effects, but FH is not plausibly seen as doing this—​ or so I will argue in section 12.2. Hence, the rule about harming that would most straightforwardly be derivable from FH would not be MP. See also note 38. 14 I previously investigated the interpretation of FH. Sverdlik, Motive and Rightness, 103–​116. The best discussions of FH by Kantians are Hill, Dignity and Practical Reason, ch. 2; Wood, Kant’s Ethical Thought, ch. 4. 15 Kant, Groundwork, 65 (397). 16 Kant generally accepted the objectivity of the deontic statuses of wrongness and permissibility (both narrow and wide) also. Sverdlik, Motive and Rightness, especially chs. 1, 5, and 6. Consequentialism also accepts that deontic status in general is largely objective. Sverdlik, Motive and Rightness, chs. 3–​4; Sverdlik, “Consequentialism, Moral Motivation, and the Deontic Relevance of Motives.” See also Scanlon, Moral Dimensions, chs. 1–​3; Tadros, The Ends of Harm, ch. 7.

242    Steven Sverdlik

By “motivational characteristics” I mean the ends or goals that the agent has in acting as well as the means that she is employing to reach them. In the case of the merchant her end is to advance her self-​interest, and the means she is employing is giving her customer the correct change. We can now ask whether the Formula of Humanity is consistent with the objectivity of obligation. The natural interpretation of FH takes it to entail this: FH1.  If an action treats a person simply as a means then it is not permissible. This is the sort of entailment of FH that we are interested in; if FH1 is true it seems to entail that punishment intended to achieve deterrence is wrong. It is widely understood that the modifier “simply” in FH1 is critical. Using people as means as such is not sufficient for wrongdoing, since we permissibly use people as means in many ways. We permissibly determine what time it is by having others tell us; we are permissibly transported by bus drivers and airline pilots; and so on. FH1 states that it is using people simply as a means that is sufficient for acting wrongly. What is it to treat someone simply as a means? A plausible suggestion is this: Treating Simply as a Means: S treats T simply as a means if and only if S uses T as a means and S has an attitude of indifference to T’s interests that explains why she uses T as a means.17 This statement articulates the idea that wrongdoing involves treating rational agents as mere things who have no capacity to set goals for themselves. The wrongdoing agent, we might say, completely subordinates another person or persons to her own aims, and disregards the aims of the other person or persons except insofar as they serve her own. Despite its appeal, using this statement to interpret FH1 (and thus FH) cannot be correct. It is inconsistent with the objectivity of obligation. To see this, we can amplify Kant’s own example of the self-​interested merchant. Suppose that she is a complete egoist who always chooses to do what she believes is in her self-​interest. She regards the interests of others as insignificant, and she will only act in a way that benefits them if she believes that this will benefit her. She believes that giving the correct change to a customer furthers her self-​interest and she does so only in order

Cp. Parfit, On What Matters, 213–​215. I have modified Parfit’s proposal in the light of a helpful comment by Tim Henning. Hill and Wood hardly consider the issue of the meaning of ‘simply’, presumably because they believe that FH is best interpreted in a way that bypasses its interpretation. Hill, Dignity and Practical Reason, 41–​42; Wood, Kant’s Ethical Thought, 142–​143. 17



Kantianism, Consequentialism, and Deterrence    243

to achieve that end. According to the natural interpretation of FH the merchant acts wrongly. But if obligation is objective then she does not act wrongly; she acts in conformity with duty. That is, she performs an action that she is obligated to perform.18 FH, as we are interpreting it, cannot be correct. These reflections suggest that the central idea in FH is really the requirement always to treat a person as an end (or, to treat the humanity in a person as an end).19 This does not rule out the possibility that some actions are wrong that involve treating people (or the humanity in them) simply as a means. This could be true when treating people simply as a means also fails to treat them (or the humanity in them) as ends. An interpretation along these lines, we will see, has the great advantage of yielding an objective conception of obligation.20 What it is to treat humanity in a person as an end? Commentators on Kant generally agree that the reference to “humanity” in FH signifies rational nature.21 Furthermore, the commentators generally agree that the essential feature of rational nature for Kant is the capacity to set ends and judge the desirability of goals for one’s self.22 Tadros, we saw, appeals to this idea too. In trying to forestall a natural misunderstanding of his thinking, Kant introduces some special terminology. We think of an “end” as a state that an agent seeks to produce by her action. Kant sometimes speaks of this as a “subjective end.”23 Kant makes a crucial distinction: he says “end” has a different meaning in FH: “Persons . . . are not merely subjective ends, whose existence as an object of our actions has a value for us: they are objective ends—​that is, things whose existence is in itself an end.”24 Kant thus claims that rational agents are a second sort of “end” that morality properly recognizes. This is a striking and unusual use of the term “end.”25 We might say that it articulates the idea that moral requirements are the proper responses to the distinctive value that objective ends have. For a given agent S, another agent T constitutes such an objective end. Indeed, FH tells us that S is herself an objective end with regard to her own activity.

Cp. Parfit, On What Matters, 216. 19 Wood, Kant’s Ethical Thought, 143. Cp. Hill, Dignity and Practical Reason,  41–​42. 20 It is true that Kant veers away from objectivism in The Metaphysics of Morals. See Sverdlik, Motive and Rightness, 117–​125. 21 See Kant, Groundwork, 57 (389), 92 (425). 22 Kant, Groundwork, 105 (437); Kant, The Metaphysics of Morals, 195 (392). See Hill, Dignity and Practical Reason, 38–​41; Wood, Kant’s Ethical Thought, 118–​122. For dissent: Dean, The Value of Humanity in Kant’s Moral Theory, chs. 2–​4. 23 For example, Kant, Groundwork, 96 (428). 24 Kant, Groundwork, 96 (428), emphasis deleted. “Objective end” also occurs at 98 (431). Cp. 96 (429): “an objective principle”, 105 (437). 25 Henry Sidgwick calls it “perplexing” (The Methods of Ethics, 390). 18

244    Steven Sverdlik

Kant gives no general characterization in the Groundwork of the value that objective ends have. Commentators who try to expound Kant’s conception of this value therefore use a variety of materials to piece it together.26 Speaking broadly and roughly, we can say this:  treating rational beings as ends (which I  will take to be treating the humanity in them as an end) would consist in not killing or maiming them; not destroying or impairing their rational capacities; not coercing them; and giving them assistance in achieving the (“subjective”) ends they rationally set for themselves.27 Let us grant that the conception of rational agents as ends can be used to generate a system of moral obligations and prohibitions. Treating people as ends can be said to consist in acting in ways that suitably relate to the moral status of persons as rational goal-​setters. The concepts involved in stating what we are morally obligated and prohibited from doing—​deception, coercion, killing, and assistance, for example—​will be suitably objective in our sense: one can do these things (and refrain from doing them) to achieve an indefinitely large number of (subjective) ends or goals. S can carry out her obligation to assist T in furthering her ends, for example, “from duty,” but she can also do this from self-​interest, and to achieve many other goals. This second interpretation of FH thus preserves the objectivity of obligation. If an action is required in order to treat a person as an end this action will remain obligatory whatever exactly the agent is seeking to do by so acting. Let us formulate this second interpretation of FH as follows: FH-​E: Always treat rational beings as ends. We can now return to the issue of using punishment to achieve deterrence. Kant obviously believes that state officials are sometimes morally obligated to impose punishment on some people. Using FH-​E we can say that it entails that these officials must punish the criminals if they are to treat them as ends. Let us suppose that Kant is truly asserting in the passage quoted above that punishing a person in order to achieve deterrence is using her simply as a means to reduce the incidence of crime. If the obligation to punish is objective it follows that an official who imposes an obligatory punishment in order to achieve deterrence is acting in conformity with duty, but not from duty. It would not be wrong of her to punish another person to achieve this goal.

See, e.g., Hill, Dignity and Practical Reason,  43–​50. 27 Sverdlik, Motive and Rightness, 111–​114. 26



Kantianism, Consequentialism, and Deterrence    245

We can conclude that the best interpretation of FH, FH-​E, tells us this: if imposing punishment is ever obligatory, it will remain obligatory even if it is imposed only to achieve deterrence. 12.3. Tadros’s Means Principle

Let us now consider a moral principle that plays a central role in Tadros’s important work in the philosophy of punishment, The Ends of Harm. The previous discussion of FH will help us to see that it is problematic. Tadros accepts the Means Principle, which I suggested can be formulated thus: MP: If an action harms a person simply as a means of producing good for other persons then it is not permissible.28 This principle has some similarities to FH, and Tadros supports MP by appealing to the Kantian idea that “as humans, we are capable of determining what ends to set for ourselves.”29 Tadros rejects consequentialism. He asserts that it does not recognize the moral status we have in virtue of our nature as rational goal-​setters. Consequentialism therefore permits harming some persons merely as means of producing good for other persons. He illustrates the problem by contrasting the well-​known examples that he calls “Bridge” and “Trolley Driver.” In Bridge a bystander can push a large person off a bridge, which will stop a runaway trolley and save the lives of five people on the track ahead. A consequentialist will say that pushing the large person into the path of the trolley is permissible. Tadros believes that doing this is wrong. However, he believes this about the Trolley Driver case: if a bystander pulls a switch that sends a runaway trolley off a track with five people on it and onto a track with one person on it she acts permissibly. Tadros believes that MP explains why pushing the large person is wrong, while redirecting the trolley onto one person is permissible. In the latter case the single person that the trolley will hit is not harmed as a means, let alone simply as a means. Her death is simply a foreseeable effect of turning the trolley onto the track that she is on. As I said, Tadros does not believe that MP is absolute; he thinks that there are cases where a person may permissibly be harmed simply in order to benefit others. For example, if S harms T and refuses to compensate T for her loss, S may be compelled

Cp. Tadros, The Ends of Harm, 114. For this formulation, see note 12. 29 Tadros, The Ends of Harm, 127. Cp. 124, 153. 28

246    Steven Sverdlik

to compensate T. In this case S may permissibly be used simply as a means to benefit T. It is via this sort of exception to MP that Tadros tries to establish, through an extended set of arguments, a right of the state to punish offenders in order to deter potential offenders. The right of the state to do this derives from the right of people to defend themselves, and the duty of wrongdoers to compensate their victims. If wrongdoers refuse to do this, victims have the right to compel them to do so. Such compelled compensation may involve protecting victims from future harm by other wrongdoers via deterrent threats and punishments. Finally, the state in some circumstances can be regarded as having had this right to deter wrongdoers transferred to it. Tadros, then, does not object in principle to punishing people in order to achieve deterrence, and he grants that it is permissible in certain circumstances. But Tadros uses MP to object to consequentialism in general, and to its approach to punishment in particular. I  will consider whether Tadros can use MP to object to consequentialism, and the pursuit of general deterrence in the design of a system of punishment—​unless we can find an exception to it that somehow allows for deterrence. I believe that our examination of FH can help us to see why it cannot be so used. Let us begin by considering how MP is related to FH and FH-​E. MP cannot be the fundamental principle of morality, since it gives only a sufficient condition of moral impermissibility. Tadros understands this point, since he claims that MP is one of a number of specific moral duties or principles. Furthermore, MP mentions harm; FH and FH-​E do not.30 This makes sense because the fundamental principle of morality must be able to establish that certain kinds of actions which do not harm people—​like some types of deception—​are wrong. MP need not do this. Finally, FH and FH-​E do not include the phrase “producing good for others.” This seems to allow for the possibility that it could be wrong for S to use T simply as a means to benefit T. Indeed, recall that Kant stated that “punishment . . . can never be inflicted merely as a means to promote some other good for the criminal himself.”31 Since MP is supposed to state a sufficient condition of wrongness it could establish that punishing to achieve deterrence is wrong—​unless there is an exception to it. We will now consider whether MP is at least consistent with FH-​E, the best interpretation of FH. And to avoid the problem of the exceptions to that statement that Tadros eventually grants, let us consider a fairly ordinary situation that MP does cover. Suppose that R is a sheriff who evicts tenants from their homes in accordance with legal procedures only in order to support herself and her family. She For other principles prohibiting harming people simply as a means see Parfit, On What Matters, 228–​232. 31 Kant, The Metaphysics of Morals, 140 (= 331). 30



Kantianism, Consequentialism, and Deterrence    247

is completely indifferent to the interests of the tenants. We can also suppose that the sheriff is morally required—​say, by her oath of office—​to evict the tenants. This example is exactly parallel to Kant’s example of the self-​interested merchant. Kant would presumably say that the sheriff acts “in conformity with [moral] duty.” To deny that the sheriff is carrying out her moral obligation is to deny the objectivity of obligation. MP tells us that the sheriff acts wrongly. There are two ways to think about the problem here. One is as an internal issue for Kantianism. Tadros does not fully commit to deriving MP from a Kantian principle like FH or FH-​E, although we saw that he suggests a Kantian foundation for it. But, we have seen, FH is best interpreted as FH-​E, which preserves the objectivity of obligation. And Tadros’s MP entails that obligation is not objective. Thus, insofar as Tadros is utilizing Kantian moral theory in his philosophy of punishment, he generates a contradiction by affirming MP. But there is not merely an internal contradiction here. Commonsense moral thought accepts that obligation is at least generally an objective concept. We say that someone “did the right thing for the wrong reason,” which seems to presuppose the truth of OO.32 Therefore we can also criticize MP for its intuitively implausible implications:  MP entails that legal officials act wrongly when self-​ interest alone leads them to carry out duties requiring them to impose harms on other people. We can conclude this: MP is not a plausible moral principle and it is not a plausible way of specifying how Kant’s idea of respecting rational goal-​setters applies to the sorts of activities that harm them. It is therefore doubly unsuited to be the way to frame a discussion of whether deterrent punishments are permissible.33 We now encounter an important problem, however. FH-​E might be thought to entail a moral principle similar to MP that could be formulated as follows: MP1:  If an action harms a rational agent it is morally impermissible. After all, any harming of a rational agent impairs to some degree her ability to set goals for herself and to pursue them. And here is the corresponding statement of the moral obligation: MP2:  It is morally obligatory not to harm rational agents.

I have argued that obligation and the other basic deontic statuses (of permissibility and wrongness) are not entirely objective. Sverdlik, Motive and Rightness. But the sort of exception embodied in MP is much too broad, as the example of the sheriff shows. 33 This criticism applies a fortiori to the actual formulation that Tadros uses (The Ends of Harm, 114). 32

248    Steven Sverdlik

MP2 is objective: one can carry it out in a specific case while pursuing an indefinitely large number of goals. However, MP2 is highly problematic. It seems to rule out the more severe forms of punishment for any purpose whatsoever. It also entails that the actions of the sheriff are wrong. It might be said that reformulating MP as MP1 poses no problem for Tadros. If he has presented a set of arguments that carves an exception out of MP, then they could also carve an exception out of MP1. So he could proceed to defend the permissibility of deterrence in his nonconsequentialist way. But accepting MP1 would create a serious dialectical problem for Tadros. If he accepts MP1 he loses his basis for objecting to consequentialism. MP can be used to show that pushing the large person is wrong but turning the trolley is not. MP1, however, entails that both actions are wrong. We now seem to have reached this further conclusion: the moral reasoning that is embodied in the best interpretation of FH, that is, FH-​E, entails that harming rational agents is always wrong. But this means that any punishment that harms the offender it is imposed on is morally wrong. So FH-​E tell us that imposing severe punishments to achieve deterrence is wrong after all! But, by the same token, it tells us that imposing severe punishments in order to give wrongdoers what they deserve is wrong. FH-​E seems to reject the use of punishment to achieve either retributivist or consequentialist aims. 12.4. Contractarianism and Deterrence

In this section we will further pursue the question of how the idea that morality requires us to treat rational agents as ends applies to the issue of punishment. Put in this form, the question has not been much discussed.34 We will see that, properly understood, Kantian theory accepts that punishment may be used to achieve deterrence. So, on this fundamental point Kantianism and consequentialism agree. In the next section I will argue that consequentialism gives us a better account of one important limit on the severity of punishment than the best version we now have of the Kantian philosophy of punishment. The problem we uncovered in the previous section is quite general, and arises for activities other than punishment. If FH-​E requires us never to harm rational agents, then many acts of self-​defense are wrong; so are many acts of warfare. It might be argued that conscription is wrong for the same reason.35 The general solution in Kantianism, I  believe, involves taking FH-​E in a contractarian direction.36 If we But see Hill, “Treating Criminals as Ends in Themselves,” esp. 310–​311. 35 Oliver Wendell Holmes rejected FH because of these last two apparent implications of it (Collected Legal Papers, 304). 36 This supplements FH with the Formula of the Kingdom of Ends. Kant, Groundwork, 439 (106). 34



Kantianism, Consequentialism, and Deterrence    249

grant that each rational being is an end we then try to find a way to represent the morally significant interests of each of them. We are led to the idea of a sort of moral legislature, where representatives seek to advance the morally significant interests of the relevant group of rational beings.37 Treating people as ends would consist in behaving in ways that conform to the rules that the legislature adopts. These rules can have exceptions. But the rules would be largely—​and perhaps entirely—​ objective, in the sense discussed above. For example, suppose there is a rule that prohibits harming other persons and that this rule has an exception that requires certain officials to enforce just court orders. Then these officials may carry out the orders simply as a means of supporting themselves or their families, and for various other reasons. The contractarian approach does not rule out the possibility that an agent’s motivation can affect whether an action of hers is obligatory, permissible, or wrong, but this is unlikely to occur often.38 We will see that deterrent punishment is not such a case. John Rawls’s “original position” is the most well-​known philosophical device for conceiving of this legislature, but there are others. Indeed, Rawls himself described a number of variants of his own favored description of the legislature. I will examine Rawls’s device partly because it is familiar, and partly because we finally have an excellent study of how Rawls’s approach to justice would apply to the issue of legal punishment, thanks to Sharon Dolovich.39 Dolovich’s paper is the most systematic attempt I know of that applies Kantian-​contractarian theory to the philosophy of punishment. The application of Rawlsian contractarianism to legal punishment is not straightforward. Rawls’s own work after 1955 does not address it, even though he focused on the justice of the basic structure of society’s political and legal systems. This curious feature of his work stems from a paradox in A Theory of Justice (and in Political Liberalism): on the one hand, he stipulates that the moral legislators are in, and know Cp. Hill, “Treating Criminals as Ends in Themselves,” 304f. 38 See Parfit’s discussion of “Tunnel” (i.e., Tadros’s “Trolley Driver”) and “Bridge” (On What Matters, 218–​221). One of his points is this: a rational agent would have no preference for being killed as a side effect of the action of turning the trolley in Tunnel as compared to being killed as a means in Bridge. This fact helps to establish that the quasi-​contractarian Consent Principle that Parfit is discussing would not condemn the act of turning the trolley or the act of pushing the person on the bridge in front of the trolley. Kantian reasoning, Parfit implies, is not generally going to endorse moral principles like the Doctrine of Double Effect or, we can add, Tadros’s MP. Putting the issues here in terms of FH-​E, however, we might well say that being pushed in front of a trolley destroys a person’s rational capacities, as does being hit by a trolley that was directed away from other people. This suggests that Kantian reasoning condemns both actions, rather than neither. (Cp. MP1.) But note that each form of Kantian reasoning treats the two actions as having the same deontic status, whereas MP treats one as permissible and one as wrong. Both forms of Kantian reasoning preserve the objectivity of obligation; MP does not. 39 Dolovich, “Legitimate Punishment.” 37

250    Steven Sverdlik

that they are in, what he calls the “circumstances of justice.”40 These circumstances make complete conformity to the principles of justice, whatever they turn out to be, difficult. On the other hand, Rawlsian moral legislators are to select rules that will apply to a situation of “strict compliance.” That is, the legislators are to assume that the agents they represent will comply fully with whatever substantive principles of distributive justice they choose.41 In this sense, the principles chosen are part of “ideal theory.”42 In characterizing the circumstances of justice Rawls draws on the work of Hart, Hume, and (indirectly) Hobbes.43 These conditions include moderate scarcity of resources, physical vulnerability, limited altruism, and “various shortcomings of knowledge, thought and judgment.”44 Hart, Hume, and Hobbes all envision punishment as necessary for mitigating some of the most problematic consequences of that set of conditions.45 Yet the principles of justice in Rawls’s theory are to be chosen on the assumption that they will be very largely obeyed with little need or no need for legal punishment.46 As Dolovich realizes we therefore need to change Rawls’s assumptions and goals in order to apply his contractarianism to the problem of legal punishment. The moral legislators have presumably already adopted the primary substantive moral rules prohibiting acts like murder and theft. We can also suppose that rules have already been adopted that establish, at least in outline, a system of political institutions involving legislatures, courts, and the executive branch, and the basic protections afforded by the rule of law. The moral legislators know that some people will not conform to the primary substantive moral rules. In considering whether to use legal punishment they will assume that doing so may have an effect on how much compliance with the primary rules comes about. That is, they assume that punishment might affect the rates at which crimes occur. One feature of the circumstances of justice needs emphasis: imperfections in rationality. As we saw, Rawls takes note of this. It has long been contended that many Rawls, A Theory of Justice, 109–​112. Cp. the briefer treatment in Rawls, Political Liberalism, 66. 41 Rawls, Theory of Justice, 125–​126. 42 Rawls, Theory of Justice, 7–​8, 215–​216. 43 For the Hume and Hart references, see Rawls, Theory of Justice, 109n3. But Hart (who frames his discussion as “the minimum content of natural law”) in turn cites Hobbes. See Hart, The Concept of Law, 189–​195, 303. 44 Rawls, Theory of Justice, 110. 45 Hobbes, Leviathan, 215; Hume, A Treatise of Human Nature, 535f.; Hart, Concept of Law, 193. Rawls’s neglect of the role of punishment in his treatment of justice was not due to his misinterpreting his sources, however, but to his adaptation in moral theory of a method used in economics, viz., developing an idealized theoretical model as a point of departure for treating more realistic cases. 46 Rawls, Theory of Justice, 211–​212. Rawls’s assumption can be compared to an instance of a traditional form of ideal theory, Plato’s Republic. In the Republic there are no institutions of punishment, apparently because no one commits crimes. Mackenzie, Plato on Punishment, 41n39. 40



Kantianism, Consequentialism, and Deterrence    251

criminals are imperfectly rational. The imperfections in their rationality could be of various sorts. It is often said to consist in defects in formal rationality, as we might call it, that is, in calculating probabilities, or in discounting the future. But it might also consist in defective “weighing” of moral considerations as against self-​interest, or in moral cognition, or in the strength of the agent’s will. None of this means that all criminals or potential criminals are irrational to such an extent that they should be excused from punishment. But it does mean that the legislators who represent criminals or potential criminals should be more rational than the people they are representing.47 We are thus considering this question: Would these moral legislators, when designing rules for legal institutions, use criminal sanctions to deter potential offenders? Dolovich concludes that they would. The relevant part of her argument can be summarized as follows.48 The two relevant positions in this question are potential crime victim and potential punished offender.49 Wrongful acts that are properly criminalized often impose losses on victims; Kant might say that such crimes impair or destroy the rational capacities of such victims. Punishments impose such losses on criminals. Behind Rawls’s veil of ignorance the legislators would not know whether they are representing a potential victim or a potential offender, or indeed, both: one person could be a victim and a punished offender. These legislators also do not know what the probability is that the person they represent will be a victim or a punished offender. But if punishments act as deterrents and reduce the amount of crime, every moral legislator would conclude that there would sometimes be a net expected gain in the ability of the agent she represents to exercise her rational capacities. For example, if it is possible that lenient punishments have enormous deterrent effects with regard to serious crimes like murder, it would be irrational to forgo the added safety that the punishments achieve merely to avoid the possible minor losses brought about by the sanctions. This conclusion gives us all that we need for now. Dolovich’s argument gives us good reason to say that Kant’s Formula of Humanity, when applied to state punishment, permits the pursuit of deterrence in some circumstances. In the next section we will consider Dolovich’s position on what these circumstances are. But we can

In Theory of Justice Rawls tends to take the parties to be actual citizens, required to reason in a certain way (109–​ 112). In Political Liberalism they are described as “representatives” of citizens (24–​25, 306–​307). Dolovich describes most potential criminals as participating in the deliberations, but not all (“Legitimate Punishment,” 358–​363). 48 Dolovich, “Legitimate Punishment,” 379–​385. Dolovich thinks that her argument would justify specific deterrence, general deterrence, and incapacitation (383). 49 Dolovich, “Legitimate Punishment,” 357. Later Dolovich recognizes a third position, a falsely convicted person (378). 47

252    Steven Sverdlik

pause here and draw some important conclusions. Let us assume that Dolovich’s application of FH conforms to FH-​E, and is correct in its main features. 1. The Kantian contractarian reasoning leading to whatever principles would be chosen to guide (actual) legislators and other legal officials cannot be seen as disregarding the interests of punished offenders. It is true that each agent’s representative is supposed to be concerned only to advance that agent’s interests, so the representatives have no concern for anyone else’s interests in themselves. But the veil of ignorance and the requirement that principles must be accepted by all the representatives ensure that everyone’s interests are considered.50 Contractarian reasoning therefore cannot be seen as treating anyone, including agents who will be punished in accordance with its principles, simply as a means. 2. Any official who imposed a punishment in conformity with such a principle could therefore be described by a Kantian as treating the offender as an end. 3. There is every reason to think that the principle mandating punishment chosen by the moral legislators will be objective in my sense since it would be comparable to the principle governing the sheriff described above. This would allow us to say, for example, that a jailer could permissibly impose a punishment simply as a means of supporting herself. This jailer could be described by a Kantian as acting in conformity with duty but not from duty.51 4. The use of legal punishment to achieve deterrence is not necessarily morally problematic or exploitative. When used to decrease the occurrence of morally wrong actions, it should not be compared to an individual’s pursuit of her own self-​interest—​even though the objectivity of obligation ensures that an individual official who is imposing a punishment with either sort of motivation may be acting in conformity with duty. Deterrent punishment to reduce the incidence of crime is a rational strategy for dealing with the circumstances of justice, especially the limited rationality and moral motivation of agents subject to moral and legal requirements. The rational representatives of every person subject to a system of moral prohibitions that are themselves acceptable would also agree to utilize punishment to reduce the incidence of the prohibited actions in certain circumstances.52

Sverdlik, Motive and Rightness, 131–​133. 51 Cp. Sverdlik, Motive and Rightness, 138–​140. 52 This point also tells against those who object to punishment as such, whether it seeks to achieve deterrence, retribution, or some other aim. See Boonin, The Problem of Punishment, 28–​29 and passim. 50



Kantianism, Consequentialism, and Deterrence    253

Consequentialists will of course agree that the pursuit of deterrence is a legitimate goal of punishment. Although Kant himself suggested that the two moral theories differed on this issue, and many later philosophers concurred, we can see that this is false. Consequentialists can also happily accept the argument that deterrence is a rationally and morally acceptable strategy for dealing with certain features of the circumstances of justice. 12.5. Contractarianism and Consequentialism

What, then, does separate consequentialism and Kantianism as they apply to the philosophy of punishment? In this section we can begin to see part of the answer to that question. I take Dolovich’s paper to give us some sense of how a plausible Kantian philosophy of punishment would work. I  argue that consequentialist theory can be seen to work better in one important way. Dolovich not only develops a Kantian argument to establish that punishment may be used to achieve deterrence. She uses it to set a limit on the severity of punishments. Here is the argument. Rawls’s guiding idea is that behind the veil of ignorance legislators would adopt a cautious decision-​rule, maximin. This tells them to make the worst outcome that can happen to the individuals they represent as good as it can be. Consider how it applies to deterrence. If punishment achieves deterrence, then some people who would be the victims of criminal activity if no punishment occurs will not be victimized. Suppose that one instance of a certain type of crime causes on average a loss of magnitude M to its victim. Suppose further that a punishment of a severity less than M for one instance of this type of crime will ensure that at least two different people are not victimized. Now consider these two alternatives:



(a) There is no punishment for this type of crime. Criminals avoid suffering losses and may gain from their crimes, but some people are victimized. On average each victim suffers a loss of magnitude M. (b) Each person convicted of this crime undergoes a punishment imposing a loss of a magnitude less than M. For each instance of such punishment at least two fewer victims suffer on average a loss of magnitude M.

Even when the moral legislators do not know the probability that the person they represent will commit this particular type of crime, or will be a victim of it, the maximin strategy (or an extension of it) would tell them all to favor (b).53 The worst that Dolovich says that generally deterrence would be permitted via an extension to maximin, namely, “leximin.” 53

254    Steven Sverdlik

happens to one of the two categories of people in (a) is a loss of magnitude M. In (b) one person suffers a loss of a magnitude less than M, and fewer people suffer a loss of an average magnitude of M. Minimization of the greatest losses requires choosing (b). While the maximin decision-​rule favors (b), it will also set a limit on the severity of punishments: the loss imposed on an offender by the punishment for one instance of a crime may not exceed the loss suffered by a victim of it.54 Sanctions cannot be so severe that a punished criminal suffers a greater loss from it than a potential victim would suffer from her criminal act. This limitation on severity resembles the one that W. D. Ross famously argued for.55 Ross asserts that in an act of wrongdoing the wrongdoer “extinguishes” her own moral rights to the extent that she violates the rights of her victim.56 (The more common expression now for this process is the “forfeiture” of rights.) Someone who only violates a property right of another person extinguishes fewer of her own moral rights than someone who kills another person, which violates the victim’s right to life. Punishment is permissibly imposed on someone who extinguishes her rights, but it is limited in its severity by the extent of the rights that the wrongdoer extinguishes. Ross thinks of the moral limit on the severity of punishment as resulting from the forfeiture of rights by people who violate the rights of others. Dolovich’s Kantian argument focuses on the potential loss suffered by a victim to set the ceiling. But they set the same sort of ceiling. Now consider how consequentialism addresses this sort of issue. Consequentialism is often supported by arguments that involve appealing to the impartial consideration of interests with full information. If this is done when the severity of punishment is the issue, we would not be limited to considering only the magnitude of the loss a punished offender would undergo for one offense and the magnitude of the loss that one potential victim of that sort of crime undergoes. We also would have information about the probability that an offender would be punished for a certain type of crime, and the probability that a person would be victimized in that way.

Leximin tells its user to maximize the well-​being of the worst-​off, and then to maximize the level of the next worst-​off, and so on. Dolovich claims that such a rule permits minimizing the number of people undergoing the greater loss (i.e., victims), even if there is no acceptable way to bring the number of victims to zero. Dolovich, “Legitimate Punishment,” 379–​385. Cp. Rawls, Theory of Justice, 72. 54 Dolovich, “Legitimate Punishment,” 386–​400. Dolovich makes her argument using simplified categories of offenses (“serious” and “nonserious”), and punishments (humane incarceration and less burdensome impositions) (386–​387). She grants some exceptions to this principle (399–​400). They do not bear on the criticism I present below. 55 Ross, The Right and the Good, 60–​63. This approach is developed by Wellman, “Rights and State Punishment” and “The Rights Forfeiture Theory of Punishment.” 56 Ross, The Right and the Good, 60.



Kantianism, Consequentialism, and Deterrence    255

There is a well-​known problem with approaches to punishment that fail to take this sort of information into account. Certain crimes like burglary are rarely punished in modern societies because the chances of being apprehended are low. Dolovich’s Kantian approach would limit the punishment for one act of burglary to the loss suffered by the average victim of this crime. The same ceiling exists whether a burglar has a 1 in 10 or a 1 in 1,000 chance of being punished. But burglars who believe that they have only a 1 in 1,000 chance of undergoing the limited punishment that Dolovich allows will have little to fear from it.57 If, say, someone engages in burglary forty times a year and succeeds in stealing property in thirty of them, her net gain in doing so may be great, even if she is punished as much as once a year in the amount Dolovich would allow. Such a burglar will not be deterred by this punishment. A consequentialism that allowed probabilities to be reckoned at the fundamental level in generating principles limiting the severity of punishments would not be so constrained. It would therefore allow one punishment of a burglar convicted of one act of burglary to be more severe than the loss suffered by a typical victim. This loosening of the severity limit is a point in favor of consequentialism. 12.6. Conclusion

Kant’s insistence that we may not use people simply as a means has suggested to many philosophers that the pursuit of deterrence is morally objectionable in principle. And the conviction that we may not harm a person simply as a means to benefit others led Victor Tadros to look for a justification of punishment that does not rest on a consequentialist basis. I argued that the best interpretation of the Formula of Humanity, FH-​E, permits the pursuit of deterrence; it also helps us to see that the moral principle Tadros invokes in his criticism of consequentialism is formulated implausibly. Kantian moral theory does not conflict with consequentialism on a fundamental point in the philosophy of punishment. However, I gave one reason for thinking that consequentialism handles a further issue about the permissible severity of deterrent punishments better than the best formulation of a Kantian philosophy of punishment that I know of. There are many other fundamental issues in the philosophy of punishment where we can investigate the relative merits of the two theories. On the other hand, we now have the important arguments of Derek Parfit claiming that the best interpretation Goldman, “The Paradox.” Goldman takes the issue to be a dilemma for a Rossian approach, which he regards as the best available. I take it as an objection to it and to the similar limitation in Dolovich. Bentham in effect recognizes the issue and concedes that utilitarianism is here “severe,” but necessarily so. Bentham, An Introduction to the Principles of Morals and Legislation, 170. 57

256    Steven Sverdlik

of Kantianism is extensionally equivalent to rule consequentialism and Thomas Scanlon’s contractarianism.58 If so, there is a better form of Kantianism than the Rawlsian version that Dolovich develops, since this clearly differs from consequentialism. How Parfit’s Triple Theory and its version of Kantianism would apply to the philosophy of punishment has yet to be explained. I have argued that the consequentialist component in a convincing philosophy of punishment needs considerable emphasis.59 References Beccaria, Cesare. On Crimes and Punishments. Translated by David Young. Indianapolis: Hackett, 1986. (Originally published in 1764.) Bentham, Jeremy. An Introduction to the Principles of Morals and Legislation. Edited by James H. Burns and H. L. A. Hart. London: Methuen, 1982. Boonin, David. The Problem of Punishment. Cambridge, UK: Cambridge University Press, 2008. Braithwaite, John, and Philip Pettit. Not Just Deserts. Oxford: Oxford University Press, 1990. Byrd, Sharon. “Kant’s Theory of Punishment.” Law and Philosophy 8 (1989): 151–​200. Dean, Richard. The Value of Humanity in Kant’s Moral Theory. Oxford:  Oxford University Press, 2006. Dolovich, Sharon. “Legitimate Punishment in Liberal Democracy.” Buffalo Criminal Law Review 7 (2004): 307–​442. Duff, R. Antony. Punishment, Communication, and Community. Oxford:  Oxford University Press, 1993. Duff, R. Antony. Trials and Punishments. Cambridge, UK: Cambridge University Press, 2001. Ewing, Alfred, C. The Morality of Punishment. London: Kegan Paul, 1929. Goldman, Alan. “The Paradox of Punishment.” Philosophy and Public Affairs 9 (1972): 42–​58. Hampton, Jean. “The Moral Education Theory of Punishment.” Philosophy and Public Affairs 13 (1984): 208–​238. Hart, H. L. A. Punishment and Responsibility. Oxford: Oxford University Press, 1968. Hart, H. L. A. The Concept of Law. 2nd ed. Oxford: Oxford University Press, 1994. Hegel, Georg W. F. Elements of the Philosophy of Right, translated by Hugh B. Nisbet. Cambridge, UK: Cambridge University Press, 1991. Hill, Thomas. Dignity and Practical Reason. Ithaca, NY: Cornell University Press, 1992. Hill, Thomas. “Treating Criminals as Ends in Themselves.” In Virtue, Rules and Justice, 296–​319. Oxford: Oxford University Press, 2012. Hobbes, Thomas. Leviathan, edited by Crawford B. Macpherson. Harmondsworth, UK: Penguin, 1968. Holmes, Oliver W. Collected Legal Papers. New York: Harcourt, Brace, 1920.

58 Parfit, On What Matters, part III. 59 I  am indebted to John Deigh, Robert Howell, Luke Robinson, Justin Fisher, and Tim Henning for their comments. Christian Seidel has been especially helpful and encouraging.



Kantianism, Consequentialism, and Deterrence    257

Hume, David. A Treatise of Human Nature, edited by Peter H. Nidditch. Oxford:  Oxford University Press, 1978. Kant, Immanuel. Groundwork of the Metaphysic of Morals, translated by Herbert J. Paton. New York: Harper and Row, 1964. Kant, Immanuel. The Metaphysics of Morals, translated by Mary Gregor. Cambridge, UK: Cambridge University Press, 1991. Mackenzie, Mary M. Plato on Punishment. Berkeley: University of California Press, 1981. Moore, Michael. Placing Blame. Oxford: Oxford University Press, 1997. Nozick, Robert. Philosophical Explanations. Cambridge, MA: Harvard University Press, 1981. Parfit, Derek. On What Matters. Vol. 1. Oxford: Oxford University Press, 2011. Rawls, John. Political Liberalism. New York: Columbia University Press, 1996. Rawls, John. A Theory of Justice. 2nd ed. Cambridge, MA: Harvard University Press, 1999. Ross, William, D. The Right and the Good. Oxford. Oxford University Press, 1930. Scanlon, Thomas. Moral Dimensions. Cambridge, MA: Harvard University Press, 2008. Semple, Janet. Bentham’s Prison. Oxford: Oxford University Press, 1993. Sidgwick, Henry. The Methods of Ethics. 7th ed. Indianapolis: Hackett, 1981. Sverdlik, Steven. Motive and Rightness. Oxford: Oxford University Press, 2011. Sverdlik, Steven. “The Origins of the ‘Objection.’” History of Philosophy Quarterly 29 (2012): 79–​101. Sverdlik, Steven. “Punishment and Reform.” Criminal Law and Philosophy 8 (2014): 619–​633. Sverdlik, Steven. “Consequentialism, Moral Motivation, and the Deontic Relevance of Motives.” In Moral Motivation, edited by Iakovos Vasiliou, 259–​ 283. Oxford:  Oxford University Press, 2016. Tadros, Victor. The Ends of Harm. Oxford: Oxford University Press, 2011. von Hirsch, Andrew. Censure and Sanctions. Oxford: Oxford University Press, 1993. Wellman, Christopher. “Rights and State Punishment.” Journal of Philosophy 106 (2009): 419–​439. Wellman, Christopher. “The Rights Forfeiture Theory of Punishment.” Ethics 122 (2012): 317–​393. Wood, Allen. Kant’s Ethical Thought. Cambridge, UK: Cambridge University Press, 1999.

Index

absolutism, 3n8, 94, 125 abstraction, resort to, 10–​13, 15, 21, 24 accountability, 225–​226, 226n21 action theory, 179, 188–​190, 192 aesthetics, 166, 169–​171 agency agent-​relativity and, 144, 147 as an end, 239, 243–​244, 248 attitudes and, 79, 90–​91, 110 dispositions and, 104, 109 doing-​allowing distinction and, 188–​190, 192, 222–​223, 223n14, 226–​228 fittingness and, 17, 91–​95, 100, 103–​110 goodness and, 96n10 outcomes and, 119–​124, 137 punishment and, 239–​241 rationality and, 98–​99, 99n15, 184n20, 243–​245,  248 reasons and, 105, 184, 202n9, 228, 231–​234 responsibility and, 149, 219, 222–​223, 226–​227 restrictions and, 144, 144n20, 148–​150 rightness and, 151, 242–​244 utilitarianism and, 91–​92, 100 value and, 144 agent-​centeredness. See also agent-​relativity consequences and, 32–​33 constraints and, 157, 224

259

in new wave consequentialism, 9–​10, 12–​13, 18–​19,  24 outcomes and, 137, 141–​142, 153–​154 rankings of outcomes and, 41 restrictions and, 15 value and, 23, 31, 36–​37, 49 agent-​indifferent. See agent-​neutrality agent-​neutrality consequences and, 10, 131, 137 doing-​allowing distinction and, 177–​185 goodness and, 18, 119–​120, 120n17, 128 outcomes and, 5–​6, 8n36, 9, 127, 142–​148, 177–​181,  177n3 reasons and, 177, 177n3, 178–​182, 181n13 restrictions and, 142, 142n14 rightness and, 8n36 utilitarianism and, 31 value and, 6, 120n17, 137, 140–​145 agent-​relativity. See also agent-centeredness; evaluator-relativity agency and, 144, 147 consequences and, 10, 122–​123 constraints and, 9, 117–​119, 122n24, 132, 157 doing-​allowing distinction and, 177–​191, 223n14, 224 goodness and, 119, 122, 131 outcomes and, 5, 10, 15–​21, 32–​35, 127–​132, 154–​155, 178–​183

260    Index agent-​relativity (cont.) rationality and, 10, 12, 23 reasons and, 4, 13, 15, 18, 177–​182, 177n3, 181n11, 181n13 restrictions and, 15, 18–​19, 136–​137, 142, 142n14 value and, 4, 15–​19, 31–​33, 119–​122, 131–​137, 142–​146, 142n14, 220 world-​centered value and, 38n19 alienation objection, 3, 5, 222 Andreou, Chrisoula, 88n30 Anscombe, Elizabeth, 1–​3, 2nn4–​5, 2n7, 3n8, 94n7, 193–​195, 193n45 Arrhenius, Gustaf, 51, 54, 66–​69, 67n35, 68n37, 69n39 assessments asymmetry in, 144–​145 attitudes and, 92 desires and, 92 fittingness and, 92 focal points of, 11, 17, 220 goodness and, 8 mode of, 11, 16–​17, 220 outcomes and, 124, 153, 201 preferences and, 33n7 rightness and, 4, 8, 15, 201 value and, 92, 132 Astaire, Fred, 168–​170, 168n26, 168n28 attitudes agency and, 79, 90–​91, 110 assessments and, 92 control of, 17, 73, 84–​86 desires and, 79–​88, 99n14, 186–​189, 191–​194, 194n46, 194n48 fittingness and, 17, 90–​93, 95, 143–​144, 143n16, 144n17 formation of, 80–​85 outcomes and, 82–​83 practical, 186–​187, 191–​195, 194n46 propositional, 187, 191–​194, 194n46 psychology and, 91 rationality and, 17, 194–​195, 194n48 reasons and, 72, 72–​73n3, 81, 84–​88, 92n4, 188 turn to, 17 utilitarianism and, 91 value and, 90 authority, 184n20, 185, 211–​214 axiology, 7, 10, 18, 33–​34, 131, 137, 141–​142, 148. See also value Bader, Ralf, 59–​60 Baier, Annette, 183

Beccaria, Cesare, 237, 237n1 Bentham, Jeremy, 1, 3n7, 22n72, 237, 237n1, 255n57 better-​than relation, 43–​47, 43n27, 53–​54, 140, 163–​164,  209 Betzler, Monika, 18–​19, 115, 155n33 Bovens, Luc, 234n34 Bratman, Michael, 210 Brewer, Talbot, 191 Bricker, Phillip, 200, 204–​205 Briggs, Rachael, 211 Brink, David, 158 Broome, John, 6n26 Brown, Campbell, 32, 34, 36, 40–​42 Carlson, Erik, 38n20 Categorical Imperative, 212, 239. See also Formula of Humanity causation, 59 Chappell, Richard Yetter, 17, 23, 88n30, 90, 171n33 character-​based objections, 17, 90–​91, 93, 95–​96, 104, 109 codes, 8, 14, 109–​110 coercion, 225, 227–​230, 228n24, 244 commands, 82, 151, 212–​213 common sense consequences and, 7, 133, 157 constraints and, 117, 122n24, 127, 132, 156–​157, 173, 180, 220 deontology and, 18–​19, 116–​127, 116n6, 118n11, 125n35, 129, 223 desires and, 186, 191 doing-​allowing distinction and, 20, 176–​177, 179–​180, 185–​186,  190 goodness and, 124n33 nonconsequentialism, 21, 163, 165, 174 outcomes and, 163–​164, 185, 191–​192 psychology and, 44, 160 punishment and, 238 rationality and, 9–​10, 20, 23, 210, 215 reasons and, 188, 192 responsibility and, 229 restrictions and, 154 rightness and, 93, 108n30, 124n33 utilitarianism and, 31, 116n6 world-​centered value and, 33–​35 consequences. See also outcomes agent-​neutrality and, 10, 131, 137 agent-​relativity and, 10, 32, 122–​123 common sense and, 7, 133, 157 consequentializing and, 126

Index    261 definition of, 2, 3n7 deontology and, 122, 125n35, 230, 233n33 dispositions and, 101 goodness and, 124n33, 231, 233 preferences and, 199–​201 punishment and, 250 rationality and, 33n7 reasons and, 131, 165 responsibility and, 219–​221, 224, 227–​228, 231–​234,  232n30 restrictions and, 145, 148 rightness and, 124n33, 125n34, 219, 232 value and, 17, 71, 157, 228, 232–​233 visible, 231–​233 consequentialism actualist, 8, 73–​78, 75n6, 80–​81, 83–​84, 86 commonsense, 156–​158, 161–​165, 164n19, 173–​174 compelling idea of, 9, 12, 18, 115–​122, 127–​133, 136–​139, 181, 189 cooperative, 79–​84, 80n16, 88 definition of, 1–​2, 2–​3n7, 6nn26–​27, 11–​15, 14n63, 31–​32, 137, 176 emancipation of, 3–​6, 10–​13 first wave, 3–​6 formalizations of, 6–​8, 11, 13–​14 multidimensional (see multidimensional consequentialism) new wave (see new wave consequentialism) objections to (see objections to consequentialism) possibilist, 76–​78, 83–​84, 86 rationalist, 17, 84, 86–​88 research program of, 6–​9 rule, 8, 11–​12, 14, 18, 109–​110, 256 satisficing, 7–​8, 105n25, 115–​116n3 scalar, 55n11 second wave, 6–​12, 15 as term of art, 2, 24 consequentializing agent-​neutrality and, 120n17, 147–​148 agent-​relativity and, 12–​13, 16, 23, 127–​128, 147–​148 common sense and, 15, 18–​19, 23, 120–​124, 122n24, 126–​127, 133 consequences and, 122, 126 constraints and, 118, 120n17 definition of, 10, 10n45, 117, 139 deontology and, 18, 117–​133, 118n12, 121n19, 125n35 desires and, 129 goodness and, 119–​121, 127

objections to, 12, 15, 18–​19, 23–​24, 118, 122–​127, 133, 201 outcomes and, 121, 145, 201 procedures for, 10–​11, 119, 122–​124, 124n33, 126 rationality and, 15 reasons and, 13, 128–​130, 133, 228–​230 restrictions and, 19, 139–​145, 143n16, 147–​149, 153–​155 rightness and, 119–​121, 229 value and, 16, 119 world-​centered value and, 31, 49 constraints. See also restrictions agency and, 105 agent-​relativity and, 9, 117–​119, 122n24, 132, 157 common sense and, 117, 122n24, 127, 132, 156–​157, 173, 180, 220 definition of, 138n6 deontology and, 104, 118–​120, 120n17, 122n24, 130–​131 fittingness and, 94 goodness and, 130 outcomes and, 131, 176 preferences and, 213–​214 rationality and, 130–​131 respect and, 214 responsibility and, 21, 220–​234, 232n30 restrictions and, 138n6 value and, 119, 122n24, 143n16 contractarianism, 248–​250, 253–​256 coordination problems, 17, 23, 71, 74, 78, 83–​84 Crowther, Bosley, 168n27 Darwall, Stephen, 212–​213 Dasgupta, Partha, 48n39 decision procedures, 11, 17, 79–​83, 82n21, 88, 90, 95–​96,  116 decision theory, 35, 202, 211 demandingness objection, 7, 9, 21, 219, 224–​225 deontic status, 4, 11n50, 13, 52, 53n4, 58–​60, 164, 241n16 deontology common sense and, 18–​19, 116–​127, 116n6, 118n11, 125n35, 129, 223 consequences and, 122, 125n35, 230, 233n33 consequentialized, 118–​133, 118n12, 121n19 constraints and, 104, 118–​120, 120n17, 122n24, 130–​131 doing-​allowing distinction and, 224 goodness and, 230 moderate, 13, 125–​126

262    Index deontology (cont.) objections to, 116–​117, 126, 230 outcomes and, 122–​127, 129 paradox of, 117–​118, 120, 120n17, 122, 130–​132 pure, 230, 233n33 responsibility and, 233n33 restrictions and, 119 rightness and, 118, 121, 230 strict, 223 value and, 120n17 variations of, 116n6 desires. See also preferences assessments and, 92 attitudes and, 79–​88, 99n14, 186–​189, 191–​194, 194n46, 194n48 common sense and, 186, 191 consequentializing and, 129 definition of, 186–​187, 191–​194 dispositions and, 97, 101–​103, 106 doing-​allowing distinction and, 186–​188, 191–​192 fittingness and, 91–​96, 92n5, 99–​101, 109–​110,  145 goodness and, 94, 103, 130 outcomes and, 164n17, 187 psychology and, 96n10, 97–​98 rationality and, 101 reasons and, 20, 98, 99n14, 164n18, 186–​187, 191–​194 utilitarianism and, 99n14 value and, 189 deterrence, 22, 237–​255, 252n52 dignity, 120n17 direction of fit, 179, 191–​195, 193nn44–​45 dispositions agency and, 104, 109 consequences and, 101 desires and, 97, 101–​103, 106 fittingness and, 90–​92, 102–​106, 102nn18–​19 guiding, 17, 103–​104, 106, 109 motivating, 17, 91, 103–​104, 109 predispositions, 105–​106 rationality and, 101, 103–​104 value and, 101, 109 well-​calibrated, 17–​18, 91, 100–​106, 109 doing-​allowing distinction agency and, 188–​190, 192, 222–​223, 223n14, 226–​228 agent-​neutrality and, 177–​185 agent-​relativity and, 177–​191, 223n14, 224

common sense and, 20, 176–​177, 179–​180, 185–​186,  190 definition of, 176–​178 deontology and, 224 desires and, 186–​188, 191–​192 form of, 178 new, 181–​182, 195 as objection to consequentialism, 223–​224 old, 179–​181 outcomes and, 176–​192 reasons and, 177–​180 responsibility and, 225–​226 restrictions and, 20 value and, 177, 186, 190 Dolovich, Sharon, 249–​256, 251nn48–​49, 253–​254nn53–​54 Dorsey, Dale, 19–​20, 88, 88n30, 156, 160n14, 195n50 double effect, 241n13, 249n38 Dreier, Jamie, 16, 19n66, 21–​23, 31, 118n11, 142n15, 207 Driver, Julia, 195n50, 202n9 Duff, Antony, 240 duty, acting from, 240–​244, 247 egalitarianism, 55, 157 Emet, Stephen F., 122n24 epistemic objection, 219, 231–​232, 232n29 equality, 46–​47, 53–​57, 60–​62, 64–​68, 64n28, 67n35 etiquette, 161, 167, 169–​170 evaluator-​relativity. See also agent-​relativity agency and, 188–​190, 192 agent-​neutrality and, 177–​185 common sense and, 20, 176–​177, 179–​180, 185–​186,  190 desires and, 186–​188, 191–​192 in new wave consequentialism, 20 outcomes and, 176–​192 reasons and, 177–​180 restrictions and, 20 value and, 177, 186, 190 Ewing, A. C., 239 extramoral concerns, 166, 166n24, 168–​172, 170n32, 171–​172nn34–​36 fittingness agency and, 17, 91–​95, 100, 103–​110 assessments and, 92 attitudes and, 17, 90–​93, 95, 143–​144, 143n16, 144n17

Index    263 constraints and, 94 desires and, 91–​96, 92n5, 99–​101, 109–​110, 145 dispositions and, 90–​92, 102–​106, 102nn18–​19 goodness and, 93n6, 94, 103 implications, 17–​18, 91, 93–​95, 93n6, 103–​104,  110 objections, 17–​18, 90–​95, 102 psychology and, 15, 17, 92–​94, 96–​104, 109 rationality and, 98, 104 reasons and, 63, 104–​105 restrictions and, 143–​146 rightness and, 94 utilitarianism and, 96, 99–​100, 103–​104,  103n22 value and, 17, 90, 92, 94–​96, 143–​144, 143n16 Foot, Philippa, 1, 6, 9, 9n37, 115, 129, 163–​164,  224 Formula of Humanity, 22, 239–​248, 241n13, 248n35, 249n38, 251–​252, 255 Frankena, William K., 158–​159 Frankfurt, Harry, 208, 226n20, 227 Frost, Kim, 193n45 functions chaining of, 59–​60 maximizing, 8 outcomes and, 32, 34 priority, 55 rightness and, 8, 57 root, 55, 55n10 utility, 38–​39, 200 value and, 8 well-​being, 55, 202 Gauthier, David, 100 Gertken, Jan, 19, 19n66, 22–​23, 33n5, 119n13, 136, 215n27 Gewirth, Alan, 153n31 Gini index, 56–​57, 64n28 Goldman, Alan, 255n57 goodness. See also value agency and, 96n10 agent-​neutrality and, 18, 119–​120, 120n17, 128 agent-​relativity and, 119, 122, 131 assessments and, 8 common sense and, 124n33 comparative view of, 44–​49 consequences and, 124n33, 231, 233 consequentializing and, 127 constraints and, 130

degrees of, 55n11 deontology and, 230 desires and, 94, 103, 130 fittingness and, 93n6, 94, 103 internal aspect view of, 44–​47 outcomes and, 6n26, 8, 44–​47 psychology and, 96n10 reasons and, 12 respect and, 231 responsibility and, 220, 229n27 rightness and, 6n26, 117, 121, 123n26, 124n33, 190, 229n27 theory of, 94, 103–​104, 119–​120, 122–​123, 128–​131, 220, 229n27, 230–​233 utilitarianism and, 91 world-​centered value and, 44–​49 Hampton, Jean, 240, 240n10 happiness, 2, 3n7, 52, 94–​97, 99, 204–​205, 207, 231n1 Hare, R. M., 1, 4n16, 39n21 Harman, Elizabeth, 208–​209 Harsanyi, John, 201–​202 Hart, H. L. A., 212–​213, 237, 250 Hedden, Brian, 39n21 hedonism, 4, 32–​33, 96–​99, 96n10, 123 Hegel, G. W. F., 240n10 Henning, Tim, 20–​23, 21n71, 33n7, 198, 242n17, 256n59 Hill, Thomas, 242n17 Hobbes, Thomas, 226n20, 250 Holmes, Oliver Wendell, 248n35 Hume, David, 226n20, 250 Humeanism, 177, 185 Hurley, Paul, 20–​22, 116n6, 127n38, 133, 176, 223n14 impartiality, 8, 156–​158, 161–​162, 172–​173, 224–​225,  254 impossibility theorems in population ethics, 51, 66–​69,  66n30 inequality, 46–​47, 53–​57, 60–​62, 64–​68, 64n28, 67n35 instrumental rationality, 98, 103–​104 integrity objection, 3, 21, 222–​223, 232 internalization of moral theory, 8, 14, 94–​95 intuitions. See common sense Jackson, Frank, 37n16, 222 Jeffrey, Richard, 211 justice, 3, 45–​46, 238, 249–​253

264    Index Kagan, Shelly, 169 Kamm, Frances, 225–​228, 226n21 Kantianism, 22, 124n29, 190, 211–​212, 237–​241, 237n2, 241n16, 243–​248, 249n38, 251–​253, 256. See also Categorical Imperative; punishment, Kantian view of Kiesewetter, Benjamin, 155n33 Kind, Amy, 195n50 Kolodny, Nico, 190 Lenman, James, 231–​233 Lin, Eden, 110n32 Lord, Errol, 110n32 Louise, Jennie, 118n11 Mason, Elinor, 21–​24, 99n15, 142n13, 149n29, 178n6, 195n50, 219 maximizing, 2, 4, 8, 108, 115, 117–​122, 127–​128, 131–​132 McGrath, Sarah, 110n32 McNaughton, David, 181n13 mere addition paradox, 16, 46, 51–​54, 65–​67 metaethics, 12n56, 179, 192 Mill, John Stuart, 1–​3, 1n1, 3n7, 22n72 Moore, G. E., 1, 3n7, 107n28 Moore, Michael, 240 moral aspects, 11, 13n62, 16, 51–​52, 55, 57–​62, 64–​65,  68–​69 moral character, 17, 90–​91, 93, 95–​96, 104, 109 moral constraints. See constraints moral legislators, 249–​251, 253–​254 moral orientation, 18, 122, 124, 126, 133 moral point of view, 92, 158, 161, 169–​173, 222 moral psychology, 15 moral rationalism, 12n58, 19–​20, 157, 160–​162, 164–​173 moral responsibility. See responsibility Moran, Richard, 184n20, 193 Morgenstern, Oskar, 200 Mulgan, Tim, 110n32, 202n9 multidimensional consequentialism assessments in, 11 common sense in, 16, 56, 67 definition of, 11, 52, 58 dilemmas in, 67n33 goodness in, 66 objections to, 60–​63 ranking in, 11 rightness in, 11, 51–​52, 55–​58, 61–​63, 61n22, 66–​67, 69, 69n39 utilitarianism, contrasted with, 58–​62

value and, 64–​65n28 versions of, 56n13, 61n22 Nagel, Thomas, 1, 190n34 negative responsibility, 3, 153, 222, 225 neutrality. See agent-​neutrality new wave consequentialism agent-​neutrality in,  23–​24 agent-​relativity and, 13, 23–​24 assessments in, 11, 15 attitudes in, 23 definition of, 13, 177 doing-​allowing distinction in, 20, 176–​182, 185, 195 objections to, 21, 143n16, 220 outcomes in, 20, 23, 177, 181, 202 preferences in, 202 rationality in, 19–​20, 20n70, 215 reasons in, 20, 177, 181, 188, 202 responsibility in, 23, 220 value in, 12, 17 Ng, Yew-​Kwang, 51 nonconsequentialism, commonsense, 21, 163, 165, 174 non-​existence problem, 37–​40,  47–​48 Norcross, Alastair, 55n11, 221 normative ethics, 1, 20, 24, 178–​179, 190, 192, 195 normativity, theory of, 12n56, 15, 161, 172, 179 Nozick, Robert, 1, 240 objections to consequentialism alienation, 3, 5, 222 character-​based, 17, 90–​91, 93, 95–​96, 104, 109 demandingness, 7, 9, 21, 219, 224–​225 doing-​allowing distinction, 223–​224 epistemic, 219, 231–​232, 232n29 fittingness, 17–​18, 90–​95, 102 integrity, 3, 21, 222–​223, 232 overview of, 3–​9, 9n37, 219 responsibility, 222, 224, 232n30 separateness of persons, 3, 65, 201 swamping, 231–​232, 232n30, 234 obligations, moral, 75n6, 85, 169, 174, 247 options, moral, 19–​20, 156–​166, 159n9, 172–​174 ought-​implies-​can principle, 81, 83, 151, 159, 221 outcomes agency and, 119–​124, 137 agent-​neutrality and, 6, 8n36, 9, 127, 142–​148, 177–​181,  177n3 agent-​relativity and, 5, 10, 15–​21, 32–​35, 127–​132, 154–​155, 178–​183

Index    265 assessments and, 124, 153, 201 attitudes and, 82–​83 common sense and, 163–​164, 185, 191–​192 consequences and, 123 constraints and, 131, 176 definition of, 7, 10–​11, 19n66, 45, 137, 148, 148n27, 206 deontology and, 122–​127, 129 desires and, 164n17, 187 doing-​allowing distinction and, 176–​192 functions and, 32, 34 goodness and, 6n26, 8, 44–​47 preferences and, 33n7, 199–​215 rationality and, 21, 198–​200 reasons and, 13, 21, 128–​129, 164n18, 177–​192, 202–​207 respect and, 211–​215 responsibility and, 222–​223, 225 restrictions and, 137–​149, 141n12, 142n13, 153–​154 rightness and, 5, 10, 142n13 value and, 8, 19, 33n5, 34, 39–​42, 137, 186, 201 world-​centered value and, 33–​34, 38–​46 Parfit, Derek, 1, 5, 46–​54, 98–​102, 132n46, 187n28, 249n38, 255–​256 pattern of action, 76–​79, 77n8, 81 permission, moral, 115n3, 158, 164, 164n19, 171–​172n35, 174, 246 Peterson, Martin, 16, 51, 110n32 Pettit, Philip, 103n20 philosophy of mind, 179, 192, 195 Pigou-​Dalton condition, 68 population ethics, 16, 37–​40, 47–​49, 51–​54,  64–​69 Portmore, Douglas W., 17, 23, 32, 71, 122, 124n30, 124n33, 184n19, 186, 186n24, 188 on the compelling idea, 12n58, 115–​116n3 on consequentializing, 120, 123–​124, 123n26, 201 on multidimensional consequentialism, 60–​63,  71 on prohibition dilemmas, 34–​36, 40, 42 on reasons, 164, 164nn18–​19, 170, 170n32, 202n9 predispositions, 105–​106 preferences. See also desires accommodated, 20–​21, 23, 33n7, 199, 202, 204, 207–​209,  214 assessments and, 33n7 coherence among, 202–​205

consequences and, 199–​201 constraints and, 213–​214 outcomes and, 33n7, 199–​215 rationality and, 20–​21, 33n7, 198–​201, 200n3, 205–​207, 209, 215 reasons and, 198–​210, 200n3, 215 respect and, 199, 211–​215 satisfaction of, 207 utilitarianism, 198–​199, 201–​202 value and, 42, 198, 215 prioritarianism, 55, 55n10, 157 prohibition dilemmas, 16, 32–​36, 40–​42, 49 prospectivism, 222 prudence, 161, 166, 169, 171, 200, 203, 205, 229 psychology attitudes and, 91 common sense and, 44, 160 desires and, 96n10, 97–​98 fittingness and, 15, 17, 92–​94, 96–​104, 109 goodness and, 96n10 rationality and, 99, 100, 189, 189n32 sophisticated, 91, 96–​97, 96n10, 99–​103, 99n15, 109 subjective, 96, 96n10, 102 utilitarianism and, 99–​100 value and, 96 punishment. See also retributivism agency and, 239–​241 common sense and, 238 consequences and, 250 deterrence through, 22, 237–​255, 252n52 Kantian view of, 22, 237–​240, 240n10, 246–​248, 251–​253 objection in principle to, 22, 238–​240, 246, 252n52, 255 rationality and, 251 rightness and, 246 severity of, 22, 238, 248, 253–​255 utilitarianism and, 237, 255, 255n57 Railton, Peter, 5, 91, 96–​99, 96n10, 158–​160, 159n9 Rajczi, Alex, 195n50 rankings (of outcomes), 5, 10, 15–​16, 201, 202 rational control, 17, 85–​86, 151 rational irrationality, 102 rationality agency and, 98–​99, 99n15, 184n20, 243–​245,  248 agent-​relativity and, 10, 12, 23 attitudes and, 17, 194–​195, 194n48 common sense and, 9–​10, 20, 23, 210, 215

266    Index rationality (cont.) consequences and, 33n7 consequentializing and, 15 constraints and, 130–​131 desires and, 101 dispositions and, 101, 103–​104 epistemic, 104 fittingness and, 98, 104 instrumental, 98, 103–​104 outcomes and, 21, 198–​200 practical, 9–​10, 12, 103n21, 157, 198 preferences and, 20–​21, 33n7, 198–​201, 200n3, 205–​207, 209, 215 psychology and, 99, 100, 189, 189n32 punishment and, 251 reasons and, 21 respect and, 21 responsibility and, 86 utilitarianism and, 201 value and, 33n7 rational transmission, 100–​102 Rawlings, Piers, 181n13 Rawls, John, 1, 159, 201, 238, 249–​250, 250nn45–​46,  251n47 reasons agency and, 105, 184, 202n9, 228, 231–​234 agent-​neutrality and, 177, 177n3, 178–​182,  181n13 agent-​relativity and, 4, 13, 15, 18, 177–​182, 177n3, 181n11, 181n13 attitudes and, 72, 72–​73n3, 81, 84–​88, 92n4, 188 common sense and, 188, 192 consequences and, 131, 165 desires and, 20, 98, 99n14, 164n18, 186–​187, 191–​194 doing-​allowing distinction and, 177–​180 fittingness and, 63, 104–​105 goodness and, 12, 119 objectivist account of, 192 outcomes and, 13, 21, 128–​129, 164n18, 177–​192, 202–​207 practical, 12, 92n4, 160–​161, 164–​165, 183, 187, 198, 202n9 preferences and, 198–​210, 200n3, 215 prudential, 185, 202n9, 229 rationality and, 21 respect and, 211–​215 responsibility and, 228, 228n25 responsiveness to, 17, 21n71, 72, 72–​73n3, 85, 109, 228–​229 subjectivist account of, 191

teleological theory of, 12, 18–​20, 21n71, 128–​133, 164–​165, 174, 178–​195, 198 turn to, 10, 12–​13, 12n56, 15, 21, 24 types of, 102n18 value and, 186–​189, 195, 229 reflective equilibrium, 93, 124–​127, 171 Regan, Donald, 79, 79n11, 79n14, 82, 82n21 regret, 63, 108–​109, 108n30, 209–​210 relativity. See agent-​relativity; evaluator-​relativity Republic (Plato), 250n46 repugnant conclusion, 43, 52–​54, 64–​65 respect constraints and, 214 definition of, 211 goodness and, 231 outcomes and, 211–​215 preferences and, 199, 211–​215 rationality and, 21 reasons and, 211–​215 rightness and, 212 value and, 243–​244 responsibility accountability and, 225–​226, 226n21 accounts of, 223, 225–​230 agency and, 149, 219, 226–​227 common sense and, 229 consequences and, 219–​221, 224, 227–​228, 231–​233,  232n30 constraints and, 21, 220–​234, 232n30 definition of, 21, 220–​222 deontology and, 233n33 doing-​allowing distinction and, 225–​226 goodness and, 220, 229n27 moral, 225–​226, 226n20 negative, 3, 153, 222, 225 as objection to consequentialism, 222, 224, 232n30 outcomes and, 222–​223, 225 rationality and, 86 reasons and, 228, 228n25 rightness and, 220–​221, 224, 227, 229–​230,  229n27 voluntariness and, 84–​86, 151, 153, 225–​226 restrictions. See also constraints accounts of, 144n20 agency and, 144, 144n20, 148–​150 agent-​neutrality and, 142, 142n14 agent-​relativity and, 9, 15, 18–​19, 136–​137, 142, 142n14 common sense and, 154 consequences and, 145, 148

Index    267 consequentializing and, 19, 139–​145, 143n16, 147–​149, 153–​155 constraints and, 138n6 definition of, 138, 138n6 deontic, 136, 139, 142, 154–​155 deontology and, 119 doing-​allowing distinction and, 20 fittingness and, 143–​146 minimal, 140–​148, 141n12, 151–​153 outcomes and, 137–​149, 141n12, 142n13, 153–​154 potential, 138, 141n12, 150–​153 value and, 19, 137, 142n14, 143n16 retributivism, 22, 109–​110, 237, 240, 248 rightness agency and, 151, 242–​244 agent-​neutrality and, 8n36 aggregation of, 67n35, 68n37 assessments and, 4, 8, 15, 201 common sense and, 93, 108n30, 124n33 consequences and, 124n33, 125n34, 219, 232 criterion of, 5, 56, 56n14, 65, 90, 95–​96, 116, 120–​121 degrees of, 52, 56–​57 deontology and, 118, 121, 230 fittingness and, 94 functions and, 8, 57 goodness and, 6n26, 117, 121, 123n26, 124n33, 190, 229n27 outcomes and, 5, 10, 142n13 punishment and, 246 respect and, 212 responsibility and, 220–​221, 224, 227, 229–​230,  229n27 theory of, 119–​120, 220, 231 utilitarianism and, 59 value and, 6n27, 7–​8 rights, 3, 5, 131–​132, 177, 177n3, 224, 254 Rosati, Connie, 169n29 Ross, Jacob, 77n8 Ross, W. D., 1–​3, 45–​46, 116n6, 126, 254, 255n57 rules, moral, 4–​5, 8, 14, 106–​109, 108n30, 226, 249–​250 satisficing, 7–​8, 14, 105, 105n25, 115–​116n3 Scanlon, T. M., 20n68, 72n3, 128, 190–​191, 256 Scheffler, Samuel, 1, 8n36, 9, 115, 158–​160, 168–​169, 176–​177, 225–​227 Schmidt, Andreas, 234n34 Schmidt, Thomas, 155n33 Schroeder, Andrew, 195n50 Schroeder, Mark, 128n40, 143n16, 164–​165

Schroth, Jörg, 18–​19, 115, 155n33 second wave consequentialism, 6–​12, 15 Seidel, Christian, 1, 36n15, 88n30, 110n32, 133n47, 155n33, 195n50, 215n27 self-​interest, 160, 166n24, 229, 241–​242, 247, 251 Sen, Amartya, 5, 5n22, 8n36, 46 separateness of persons objection, 3, 65, 201 Shiller, Derek, 110n32 Sidgwick, Henry, 1–​3, 3n7, 22n72, 23 Singer, Peter, 110n32, 198–​199, 201 Slote, Michael, 105n25 Smith, Michael, 110, 188–​189, 189n32 Sobel, David, 167n25 Sosa, David, 145 state of affairs value monism, 186, 188–​192, 195 Stocker, Michael, 90, 108 Stone, Martin, 193 Strawson, Peter F., 226n20 supervenience, 45, 59 Sverdlik, Steven, 22, 24, 234n34, 237 swamping objection, 231–​232, 232n30, 234 Tadros, Victor, 238, 240–​241, 240n12, 241n13, 243, 245–​248, 249n38, 255 teleological conception of reasons (TCR), 12, 18–​20, 21n71, 128–​133, 164–​165, 174, 178–​195,  198 Temkin, Larry, 32, 43–​49 Thomson, Judith, 224 time-​centered value, 16, 23, 31–​33, 33n7, 37, 41, 49, 132, 207 Timmerman, Travis, 88n30 transitivity, 16, 43–​44, 43nn27–​28, 47–​49, 53–​54 trolley problems, 149, 245, 249n38 utilitarianism agency and, 91–​92, 100 agent-​neutrality and, 31 attitudes and, 91 common sense and, 31, 116n6 definition of, 2–​3n7, 4, 4n15 desires and, 99n14 distinguished from consequentialism, 3–​6 fittingness and, 96, 99–​100, 103–​104, 103n22 goodness and, 91 objections to, 3–​4, 4n16, 6–​7, 239 preferences, 198–​199, 201–​202 psychology and, 99–​100 punishment and, 237, 255, 255n57 rationality and, 201 rightness and, 59 utility functions, 38–​39, 39n21, 200 utility monster, 4

268    Index value. See also goodness; world-​centered value agency and, 144 agent-centeredness and, 23, 31, 36–​37, 49 agent-​neutrality and, 6, 120n17, 137, 140–​145 agent-​relativity and, 4, 15–​19, 31–​33, 119–​122, 131–​137, 142–​146,  220 assessments and, 92, 132 attitudes and, 90 bearers of, 22, 186, 188, 190 buck-​passing account of, 190–​191 consequences and, 17, 71, 157, 228, 232–​233 constraints and, 119, 122n24, 143n16 definition of, 189 deontology and, 120n17 desires and, 189 dispositions and, 101, 109 doing-​allowing distinction and, 177, 186, 190 fittingness and, 17, 90, 92, 94–​96, 143–​144,  143n16 functions and, 8 goodness and, 44 outcomes and, 8, 19, 33n5, 34, 39–​42, 137, 186, 201 preferences and, 42, 198, 215 psychology and, 96 rationality and, 33n7 reasons and, 186–​189, 195, 229 respect and, 243–​244 restrictions and, 19, 137, 142n14, 143n16

rightness and, 6n27, 7–​8 state of affairs monism, 186, 188–​192, 195 time-​relative, 16, 23, 31–​33, 33n7, 37, 41, 49, 132, 207 voluntariness, 84–​86, 151, 153, 225–​226 von Hirsch, Andrew, 240 von Neumann, John, 200 Weinberg, Rivka, 195n50 well-​being, 51–​69, 55n10, 56n14, 64n28 well-​calibrated dispositions, 17–​18, 91, 100–​101,  109 Willaschek, M., 215n27 Williams, Bernard, 3n8, 6, 6n24, 90, 107, 219, 222–​227,  224n15 Wood, Allen, 242n17 world-​centered  value agent-​relativity and, 38n19 common sense and, 33–​35 consequentializing and, 31, 49 definition of, 16, 31–​32 goodness and, 44–​49 models of, 16, 32 outcomes and, 33–​34, 38–​46 usefulness of, 33–​34, 42–​43, 48–​50 Yetter-​Chappell, Helen, 110n32