Oxford Studies in Normative Ethics [9] 9780198846253, 0198846258

Citation preview

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

Oxford Studies in Normative Ethics

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

Oxford Studies in Normative Ethics Volume 9 Edited by

MARK TIMMONS

1

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom 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 in certain other countries © the several contributors 2019 The moral rights of the authors have been asserted First Edition published in 2019 Impression: 1 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 licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2019945427 ISBN 978–0–19–884625–3 (hbk.) ISBN 978–0–19–884626–0 (pbk.) DOI: 10.1093/oso/9780198846253.003.0001 Printed and bound in Great Britain by Clays Ltd, Elcograf S.p.A. Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

Contents Acknowledgments List of Contributors

Introduction Mark Timmons 1. Being Prepared: From Duties to Motives Barbara Herman

vii ix

1 9

2. The Normative Structure of Request Micha Gläser

29

3. Discretionary Moral Duties R. Jay Wallace

50

4. Taking it Personally: Third-Party Forgiveness, Close Relationships, and the Standing to Forgive Rosalind Chaplin 5. Persons as Things Mark Schroeder 6. Fitting Love and Reasons for Loving Christopher Howard 7. “But I Voted for Him for Other Reasons!”: Moral Permissibility and the Doctrine of Double Endorsement Alida Liberman

73 95 116

138

8. On the Virtue of Taking Oneself Lightly Macalester Bell

161

9. From Duty for the Right Reasons Teemu Toppinen

183

10. Deontological Decision Theory and the Grounds of Subjective Permissibility Seth Lazar

204

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

vi



11. Agent-Relative Prerogatives and Suboptimal Beneficence Ralf M. Bader

223

12. Well-Being, the Self, and Radical Change Jennifer Hawkins

251

13. From Teleosemantics to Normative Ethics Jacob Ross

271

Index

295

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

Acknowledgments All but one chapter in this collection were presented at the ninth annual Arizona Workshop in Normative Ethics that took place in Tucson, Arizona, January 11–13, 2018. I thank the Center for the Philosophy of Freedom and the Department of Philosophy at the University of Arizona for their generous financial support of the workshop. Of course, the views expressed in the volume’s chapters do not necessarily reflect the views of the Center or the Department. Thanks to Robert Wallace for his help in making the workshop run smoothly, to Karolina Wisniewska for preparing the index, and to Betsy Timmons for her expert organizational skills and support. Thanks also to the 2018 de facto program referees: Marcia Baron, Adam Cureton, Michael Gill, Elinor Mason, Doug Portmore, and Theron Pummer. Two anonymous referees for Oxford University Press offered excellent comments on the workshop papers and constructive advice to our authors. Thanks, finally, to Peter Momtchiloff, my OUP editor, for his guidance and support. Mark Timmons Tucson, AZ

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

List of Contributors Ralf M. Bader is Fellow of Merton College and an Associate Professor in the Philosophy Department at the University of Oxford. Macalester Bell is Associate Professor of Philosophy at Bryn Mawr College. Rosalind Chaplin is a PhD student at the University of California, San Diego. Micha Gläser is a postdoctoral fellow in political philosophy at the University of Zürich. Jennifer Hawkins is Associate Research Professor of Philosophy at Duke University. Barbara Herman is the Griffin Professor of Philosophy and Professor of Law at the University of California, Los Angeles. Christopher Howard is Assistant Professor of Philosophy at McGill University. Seth Lazar is Associate Professor in the Research School of Social Sciences at the Australian National University. Alida Liberman is Assistant Professor of Philosophy at Southern Methodist University. Jacob Ross is Associate Professor Philosophy at the University of Southern California. Mark Schroeder is Professor of Philosophy at the University of Southern California. Teemu Toppinen is an Academy of Finland Research Fellow at the University of Helsinki, Department of Practical Philosophy. R. Jay Wallace is Professor of Philosophy at the University of California, Berkeley, where he holds the Judy Chandler Webb Distinguished Chair.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

Introduction Mark Timmons

Oxford Studies in Normative Ethics features new work in the field of normative ethical theory. The chapters in this ninth volume collectively address the following topics: the relation between duty and motive, the structure of requests, discretionary moral duties, third-party forgiveness, persons as things, fitting love, the doctrine of double endorsement, taking oneself lightly, duty and right reasons, deontological decision theory, suboptimal beneficence, the self and radical change, and teleosemantics and normative ethics. Is motive ever morally relevant to whether an action is wrong? On what seems to be the majority view the answer is no, and this view is bolstered by the dictum that since motives are not within one’s direct voluntary control, the rightness or wrongness of an action cannot depend on them. However, intuitively there are cases in which an otherwise not impermissible action, such as turning down an offer on one’s house, is wrong when it is racially motivated. Such cases support the minority view. In “Being Prepared: From Duties to Motives,” Barbara Herman defends the minority view by first considering duties of due care that put pressure on both the majority view and the dictum about motives. A duty of due care arises as a result of having other, more primary duties whose value (or point) determine what one is to doing in exercising due care in complying with them. For instance, if the value of gratitude (its point) is to protect and promote particular social bonds, then due care must be taken in how one demonstrates their thankfulness by, for example, being attentive to the size of the benefit, how much time, effort, or money it cost the benefactor, and so on. Being attentive to such details and taking due care is to be properly motivated. Duties of due care favor the minority view because they arguably “draw on a broad practical ability that [according to Herman] has to be rooted in motive.” In defending this view, Herman provides an account of motive that respects

Mark Timmons, Introduction In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0014

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

2

 

the dictum yet preserves the connection between duties of due care and motivation. Micha Gläser in “The Normative Structure of Request,” explains that the phenomenon of request seems to involve a number of facets that are in tension with one another. For instance, on one hand it seems perfectly fine (can’t hurt) to make a request, or turn one down, though, on the other, one often feels that making a request is out of place, and that it is not always proper to refuse one. It also seems that often there is good reason to make a request and that the fact that a request was made is a good reason for compliance. However, in some cases, we feel that although one has good reason to make a request, one should not have to do so (having to do so reflects poorly on the requestee), while in others we feel that being asked without good reason on the part of the requester is out of place. “How then,” asks Gläser, “could a request as such ever be a good reason to do anything?” In responding to this question, Gläser’s aim in the chapter is two-fold. Working with formal constraints on an adequate account of request, he first critically assesses what he calls the “mere addition” account according to which requesting creates a kind of reason for the addressee whose normative force differs from that of a command. In making a request the requester attempts to add a pro tanto reason to the requestee’s reasons that she has independently of the request. Against this view (that he attributes to Raz) Gläser objects that it fails to adequately account for obligations that can arise for both parties in a requester–response relationship. Gläser then proceeds to advance a “reasonable harmony” account of request, according to which the constitutive aim of requests is to realize a harmony of ends between the parties in question. Abuse of this aim by one or another party is what explains how requests can be out of place, as well as how one sometimes has an obligation to comply with the request. In “Discretionary Moral Duties,” R. Jay Wallace considers the category of duty, recognized by common sense, which allow wide latitude or discretion in how one is to comply with them. Such discretion is characteristic of duties of gratitude and mutual aid. What seems theoretically puzzling about such duties, as Wallace explains, is “not the idea that morality grants us some discretion as to how to live up to the requirements is lays down on us, but that there are cases in which we seem to have especially wide discretion in this matter.” To address this puzzle, Wallace proceeds by employing a “relational” conception of moral obligation, according to which obligations are directed to another party who has a claim to the person’s compliance. On this view, disregarding an obligation changes the relationship between the

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi



3

two parties in that the person under obligation has wronged the other party which in turn grounds the standing of the wronged party to object to how they have been treated. Crucially, in Wallace’s version of the model, obligations need not always correlate with strict rights on the part of those to whom an obligation is owed, yet, for example, a benefactor does have a moral claim on the beneficiary that she use her discretion to respond appropriately to the kindness expressed toward her. This and other features of Wallace’s model help make sense of the wide discretion associated with gratitude and mutual aid. On one conception of forgiveness, only the victims of wrongdoing are in a position to forgive their offenders. This, of course, rules out third-party forgiveness. Rosalind Chaplin in “Taking it Personally: Third-Party Forgiveness, Close Relationships, and the Standing to Forgive” defends thirdparty forgiveness. On her “victim-relative” account, it can be fitting for one to forgive (or refuse to forgive) wrongs that are done to victims to whom one stands in special relationships, including, for example, friends and loved ones. Chaplin’s defense involves providing a case-based support for the view, articulating norms that govern third-party forgiveness, rebutting various objections to her victim-relative account, and explaining how having the standing to take wrongs done to victims “personally” grounds these cases of forgiveness. She concludes by suggesting that her account provides some reason to embrace attitude-based accounts of the nature of forgiveness— accounts according to which paradigmatic cases of forgiveness involve foreswearing negative attitudes such as resentment. What is involved in the Kantian idea of treating others as persons, never merely as means—i.e., merely as things? One way of addressing this question is to appeal to P. F. Strawson’s distinction between two stances from which one can relate to persons: a causal, so-called objective stance, and a participant stance that correspond respectively to treating someone merely as a means and treating someone as a person (as an “end-in-themselves” as Kant would say). One might then suppose that interpreting others’ behavior from within an objective stance always precludes treating them from within a participant stance. In “Persons as Things,” Mark Schroeder refers to this supposition as the “Kantian Gambit,” and proceeds to argue that it is “not quite true.” Schroeder describes multiple cases in which to charitably interpret others, thus treating them as persons, seems to require interpreting their behavior in causal terms, analogous to how one ought to charitably interpret the philosophical work of others. An upshot of his chapter is that it can be a vice to always interpret the behavior of others from a strictly participant

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

4

 

perspective; sometimes one needs to view what others do as part of a causal nexus, if one is to sustain a healthy relationship with them. What makes love fitting? In “Fitting Love and Reasons for Loving,” Christopher Howard defends the “quality view,” according to which it is the qualities of the beloved that make loving that person fitting. In order to defend this view, Howard responds to common objections and argues that the main competing view—the “relationship view”—is false. The common objections include, for example the problems of promiscuity and trading up. If it is fitting to love some particular person in virtue of their lovable qualities, it seems to follow that it is fitting to love anyone with those same qualities (promiscuity). And again, if someone is more lovable than one’s current beloved, it seems to follow that it is fitting to love that person even more (trading up). In response to these (and related objections) Howard argues that the quality theorist should embrace the seeming implications of her view, and then argue that from the fact that it would be fitting, for example, to trade up, it does not follow that one thereby has sufficient all-things-considered reason to do so. That is, the quality theorist should embrace the idea that the loveable qualities of someone provide “fit-related” reasons to love, yet recognize that they are not the only reasons for love. There are, in addition, “value-related” reasons provided by the fact that one is in a valuable loving relationship that give one sufficient reason to maintain that relationship and not, for instance, trade up. Against the relationship view (and details aside), Howard argues that the fact that one is in a valuable relationship isn’t among the fit-related reasons for love; not only is it not necessary for fitting love, it is not sufficient. Consider the Trump supporter who opposes many of the sexist, racist, and xenophobic comments made by Trump when campaigning for U.S. president, but strongly favors Trump’s promise to “drain the swamp” and remove corrupt Washington insiders and so votes for him. Under what conditions is it morally permissible to endorse morally problematic people, projects, artworks, political symbols, or other morally controversial things when those things have good qualities one endorses but also have bad qualities one opposes? This is the question Alida Liberman addresses in “ ‘But I Voted for Him for Other Reasons!’: Moral Permissibility and the Doctrine of Double Endorsement.” The doctrine in question includes three constraints on the moral permissibility of endorsement of morally controversial things: (i) the good qualities of the thing must be separable from bad qualities of that thing, (ii) the bad qualities cannot significantly outweigh the

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi



5

good ones, and (iii) there are no alternatives (at all, or that are equally easy to effectively support) that allow one to endorse the good qualities in question. Liberman begins with an account of endorsement, and proceeds to defend DDE on the basis of a range of examples taken from recent events. As Macalester Bell notes, in everyday life, people sometimes refer to the importance of not taking oneself too seriously, that doing so is a kind of failing; often a moral failing. This suggests that not taking oneself too seriously requires taking oneself lightly. But then, what is it to take oneself lightly? Of what value is it? Furthermore, isn’t taking oneself lightly in tension with Kantian self-respect? Doesn’t it open the door to a kind of self-alienation? Is taking oneself lightly inappropriate in “dark times” characterized by widespread oppression and hardship? In “On the Virtue of Taking Oneself Lightly,” Bell addresses these questions. She develops a conception of taking oneself lightly, according to which it centrally involves a disposition to take an attitude of amusement toward oneself and one’s projects. After then arguing that this disposition should be recognized as a virtue, she proceeds to explain the ways in which such a trait is valuable. Its value, then, helps explain why it is not inconsistent with Kantian self-respect understood as recognition self-respect. Finally, Bell explains why this trait is an important virtue, especially in dark times. The generic idea of an action’s moral worth is that of an action that is not only right but whose performance is not accidental. According to one conception of moral worth that can be traced back to Kant, the performance of an action has moral worth (and the agent in so acting is morally praiseworthy) only if the moral thought that the action is right (one’s duty) plays a role in explaining one’s action. Call this the “moral motive view.”¹ According to the “right reason view,” the performance of an action has moral worth (and the agent in so acting is morally praiseworthy) just in case it is performed for the reasons that in fact make the action right. Both views have attractive features, yet they also have their share of challenges. The right reasons view is criticized for failing to properly capture the nonaccidental rightness of morally worthy action. Even if acting for what are the right reasons is necessary for moral worth, it does not seem sufficient. The moral motive view, by contrast, does not seem to provide a necessary condition on moral worth. There are cases (e.g., Huck Finn’s helping Jim escape in Mark Twain’s novel) where it does not seem that the moral ¹ Some contemporary advocates add further conditions so that the moral motive view also states a sufficient condition of moral worth. See, for example, Sliwa 2016.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

6

 

thought that one’s action is right figures in morally worthy actions.² One might propose conjoining the two views so that acting from the right reasons and for the reason that the action is right yield an adequate conception of moral worth. However, Teemu Toppinen in “From Duty for the Right Reasons,” argues that as long as one accepts a cognitivist account of moral judgment, these views don’t admit of combination. According to Toppinen, if one rejects cognitivism in favor of an expressivist understanding of moral judgment, then, it is possible to reconcile the two views, combining them into a single conception of moral worth. It is common in normative ethical theory to distinguish objective from subjective permissibility (and impermissibility). Objective permissibility refers to the deontic status of actions in light of all the facts, while its subjective counterpart is permissibility in light of an agent’s limited epistemic perspective. A guiding aim of Seth Lazar’s “Deontological Decision Theory and the Grounds of Subjective Permissibility,” is to defend a criterion of subjective permissibility that can figure in a kind of decision theory for deontological theories (that is, theories that take the right to be prior to the good). Lazar works with “a compelling idea” in decision theory that there is a systematic way of combining an agent’s epistemic states with objective inputs to yield the sought-after criterion. The question Lazar addresses in his chapter is, “What are the [objective] inputs that together with the agent’s epistemic states, determine subjectively permissible action?” One answer is “all-out subjectivism” that denies that there can be a criterion of subjective permissibility that features an objective input that systematically combines with subjective epistemic states. Another proposed answer, the “objective verdicts” account, holds that when one is ignorant of the objective permissibility of an action, one is to try to perform the objectively permissible act. Lazar explains why neither of these views provides an adequate account of subjective permissibility. His positive proposal, the “objective reasons” view, grounds subjective permissibility in an agent’s limited epistemic state together with objective reasons—facts that would obtain were one to perform some action; facts that count for or against that action. Is suboptimal beneficence morally permissible? To bring this question into focus, Ralf M. Bader in “Agent-Relative Prerogatives and Suboptimal Beneficence” has readers consider the following choice situations: I (a) one can do nothing or (b) one can save y’s left arm at some cost to himself;

² For this example, see Arpaly 2002.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi



7

II (a) one can do nothing or (b) save both of y’s arm at the same cost as before; III (a) one can do nothing or (b) save y’s left arm at some cost to himself or (c) save both of y’s arms at the same cost to himself. Suppose that in case I the cost the agent must undergo in order to save y’s arm is sufficiently large compared to the benefit to y and so choosing b is good but not morally required, either choice is permissible. Again, suppose in case II helping is not required even though the benefit to y is greater than in case I because the cost is sufficiently high. Then, as in case I, both (a) and (b) are permissible. However, in choice situation III, only (a) and (c) are permissible: to choose (b) given that the cost of choosing (c) would be the same is impermissible. After arguing that suboptimal beneficence is impermissible, Bader proposes to explain this in terms of a theory that incorporates agent-relative prerogatives that ensure that one is not required to do what is impartially best, yet makes suboptimal beneficence impermissible. Bader then proceeds to consider how this kind of prohibition, in order to avoid dynamic inconsistency, requires developing global (as opposed to local) choice principles. Could undergoing a radical change ever be prudentially good for an individual? Perhaps upon reflection most would give an affirmative answer, but then one may ask: What distinguishes radical changes that are prudentially good from those that are prudentially bad? This is the question Jennifer Hawkins addresses in “Well-Being, the Self, and Radical Change.” Radical change is here defined as cases where either several of a person’s core values change, or where some deep feature of her psychology changes (which will typically bring along with it changes in core values). Hawkins frames her question within a particular theory of prudential choice—the “future-based reasons view”—according to which whether a choice is prudentially good for someone depends on the good or bad things that various possible futures hold for that individual. As Hawkins explains, this kind of view allows for radical change to be prudentially good (perhaps best), yet there is the worry that the view is too lax in allowing too many cases of radical change to be counted as prudentially good. In considering how to rule out some radical change as bad, Hawkins considers but rejects various proposals that feature considerations of either numerical identity or of character identity. She also rejects the idea that radical self-change is simply a brute type of prudential bad. Hawkins concludes by dwelling on the possibility that the high transition costs of radical change may explain why some kinds of radical self-change would not be prudentially good. Given that morally wrong actions are those that are worthy of moral disapproval, one approach to normative ethics is to ask: What kinds of

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

8

 

actions are deserving of moral disapproval? In pursuing an answer to this question, Jacob Ross in “From Teleosemantics to Normative Ethics,” develops a particular version of a teleosemantic account of cognitive content that he calls “adaptation semantics” as a basis for developing an account of the cognitive content of the attitude of moral disapproval. Roughly speaking, according to adaptation semantics, in order for some property (or combination of properties) to be the cognitive content of the attitude of moral disapproval, the property in question (or the combination) must play the right role in an adaptationist explanation of the behavioral dispositions associated with this particular attitude. After spelling out the content in question, Ross proceeds to draw some implications for normative ethics. One striking implication of the account is that all non-relativistic normative ethical theories are to be rejected.

Bibliography Arpaly, Nomy. 2002. “Moral Worth,” Journal of Philosophy 99: 223–45. Sliwa, Paulina. 2016. “Moral Worth and Moral Knowledge,” Philosophy and Phenomenological Research 93: 393–418.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

1 Being Prepared From Duties to Motives Barbara Herman

There is disagreement about whether the moral wrongness of an action can depend at all on motive. I’m on the minority side that thinks it can, and in the minority of that minority who is some sort of deontologist. Of course, as is usual with such things, once we get into what’s at stake, the disagreement that starts the discussion is the least of it. As we’ll see, even the terms of the disagreement—about wrongness and about motive—don’t survive unchallenged. Chief among pillars supporting the majority view is a normative claim that I believe conflates impermissibility and moral wrongness. It begins with something true: impermissibility is motive-insensitive. It is a verdict forbidding a kind of doing: killing the innocent to produce even a large benefit is a paradigm example. Adding a good motive makes no difference. Like the legal notion it mirrors, its negation is also said to be motive insensitive: if an action is not impermissible, it is permissible, not wrong, regardless of motive, morally available to any agent. The moral phenomena often call for a second-order category of “permissible but . . . ”: permissible but mean, permissible but insulting, permissible but not the sort of thing a good person would do. This puts motive in the picture, just later on: not at the point of saying of an action that it should not be done because the doing is wrong, and wrong at least partly because of motive. But what if the basic moral terms of action evaluation do not mirror the pair illegal and legal but are instead a triad, with impermissible and permissible not contradictory notions but instead true contraries? That would create space for an action being “not impermissible” and yet wrong. We would hold onto the idea that an action is impermissible if it is of an act-type whose moral valence is context insensitive—so not available in the balancing of reasons. But an action that is “not impermissible” might still be morally

Barbara Herman, Being Prepared: From Duties to Motives In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0001

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

10

 

disallowed in some circumstances, not in others.¹ We could then say an action is morally wrong either if it is of an act-type that is impermissible, or, if not impermissible, ruled out in particular circumstances or conditions. And such conditions might involve motive. There would be no ex ante barrier from the logic of impermissibility. There is a rationale for insisting on motive insensitivity in a regime of positive law that imposes zones of policeable action. What makes it true that I may be interfered with or afterwards penalized is that what I’m doing is taking what is not legally mine, whether my motive is good or bad. However, the interpersonal moral sphere is not in this way the private shadow of public law; there is no comparable regime of moral policing. We have at most imperfect duties to prevent others’ wrongdoing and are not typically in a moral position to impose penalties for wrongs done. Other data is mixed. Looking at negative duties, motive tends to enter as a mitigating factor for culpability. But then many, if not most, negative duties are prohibitions. Avoiding a prohibition is a kind of success, regardless of motive. For a positive duty like gratitude, however, motive is constitutive of satisfying the duty: grudging or manipulative return of benefit isn’t the real thing (feeling grateful is not required). There are also motives that seem to infect the actions they support, such as discriminatory and predatory motives. Where racism influences how one exercises a right to rent or sell or an abusive sub-text underlies an invitation to lunch, the actions so-motivated ought not be done. I think there’s enough here to keep the door open for an account of a connection between motive and moral wrongness. There is, in addition, a non-normative issue that will require attention. I think of it as a kind of dictum: that because motives are not under our direct control, we cannot have a duty, or a moral requirement on action, that depends on motive. An ought implies can kind of thing. How this plays out is not as straightforward as it initially might seem. In defense of the minority view, I will proceed by first laying out a kind of duty that puts instructive pressure on the majority view and on the dictum. These are the duties of due care or non-negligence.² If, as I argue, they can

¹ This can happen when the conditions of wrongness are adjacent to those of permissibility— an action that’s morally okay if done to a stranger but not to a friend. We will say it’s wrong to treat a friend that way, but not that the action as such is impermissible. With the triad we also get a natural way of treating exceptions—as belonging to a class of actions that are typically but not necessarily wrong. ² I first explored this fruitful subject in a seminar co-taught with Seana Shiffrin in Fall 2013. For some of her views on the topic, see “The Moral Neglect of Negligence,”

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

11

depend on motive, we have a choice: accept the moral phenomena and question the idea of motive that resists it, or stick with an account of motive we might do well to reject and play tag with the moral phenomena. In the second part of the paper I make some large claims about revising the common idea of what a motive is that supports the first choice. At the end, I’ll try to explain how it is that the wrong motive can (sometimes) affect an action’s moral valence.

1 Let’s begin with a brief survey of the duty of due care. When you and I interact, on the street or in my front room or even over the internet, lots of things can happen. We pass and make room for one another, we exchange ideas or gifts, we make plans to meet: in all, we intend things to go a certain way. But we also bump into one another, fail to notice a salient change, make an offending remark, miss a meeting—without intending these outcomes. Suppose we could have done better. We were negligent. Of course, even if we take the care we could and should take, failure happens. There are hazards that we can’t anticipate; others’ actions may alter the circumstances in ways that defeat our cautions. So it matters how things come about. Failure isn’t proof of negligence; success is consistent with it (ordinary morality doesn’t require harm to find negligence). Suppose one makes a promise to A to do X, where X is complicated enough to require some planning and care to execute. I promise to bring bread; I forget the bakery closes at 6:00 and arrive at 6:30; it turns out the bakery has changed its hours and is open until 7:00. I bring the bread, as promised. I also acted negligently. Have I succeeded in one thing and failed in another? One can’t succeed negligently, only despite one’s negligence. Had I forgotten the promise and brought the bread for some other reason, we would not say I kept my promise. The same is true when the happy result is owed to luck. Knowing that negligence lies in failure to direct the right kind of critical attention to our acting doesn’t tell us what we are required to do. Even so, it’s not hard to identify carelessness: the right turn made without looking; loudly voicing a criticism without attending to who can hear it; closing a valve or a in Vol. 3 Oxford Studies in Political Philosophy (eds. David Sobel, Peter Vallentyne, and Steve Wall, Oxford University Press, 2017).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

12

 

latch, but not all the way; relying on memory rather than writing down a complex schedule. If these acts all seem to involve taking for granted something that shouldn’t be, we don’t, in acting with due care, need to pay attention to everything, just to those areas where absence or misdirection of attention is a fault (one could be attending very carefully to the identity of the birds in the branches while making the right turn). When we say, “Pay attention!” we assume that it’s obvious where attention isn’t being directed—that the target already knows where the spot is in what she is doing that might warrant the warning. It’s as if we are primed for failure to take due care. It’s important to be clear that the duty of due care is not a special case of the requirement to take necessary means, though it often does its work in tandem with the instrumental requirement. Having a duty to repay, I must gather and then convey funds in a timely manner. But servicing the duty instrumentally is not all that is required.³ I should not, for the pleasure of it, act in ways that lead you to doubt that I will repay. That’s a kind of negligence. Likewise, if I have a duty to be truthful about something, I should try to make the truth accessible (not deliver medical advice in an overly technical manner). I put out the fire but in a way that needlessly damages your property. Failures of this sort are not instrumental failures; they rather undermine the value of the dutiful performance. So why not a duty to help and a duty not to cause damage; a duty to speak truthfully and a duty not to obfuscate; a duty to pay debts and a duty to appear to intend to pay one’s debts. There are of course circumstances in which one has and must act on multiple independent duties.⁴ But in the cases at issue, the supposed second duty somehow belongs to the first. We do not have straight-up duties to be transparent or not to obfuscate or to act without incurring costs. We also have no standing duty to remember. We have a duty to remember (or to take steps so that we remember) when forgetting is a foreseeable impediment to fulfilling another duty. We have a duty not to obfuscate when lack of clarity undermines the point of the truthful communication. And so on. When our not taking steps to remember, or not making the effort to speak clearly, incapacitates us with respect to ³ Instrumental failures are typically regarded as a species of irrationality, not negligence. In some cases, failure to take the best or better available means is negligent: the doctor who won’t change his entrenched practices need not think his ways are best. ⁴ The literature tends to focus on cases where duties conflict, but, as we’ll see in a moment, there is interesting work to be done on the effects duties have on each other when they can be co-instanced.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

13

the primary duties we have, it makes sense to describe these situations as failures to take due care. Suppose that you do not fail in doing what your duty to repay requires, but the manner of conveyance introduces other kinds of difficulty. You repay the debt with a check, but after the banks have closed, and I needed use of the funds today. We know each other well enough so that you knew that. If it was easy, or easy enough, for you to have avoided the difficulty, then you should have. While I have no claim on you to do that based in the duty to repay, there surely seems to be a failure of due care here. But if so, with respect to what? Plausibly, the standing duty you have to be mindful of a friend’s welfare. So not a conflict of duties; it was possible to attend to both duties with the same action. And although there is only one claim on you— were you able to attend only to the debt you would violate no duty—there is some kind of requirement that in the servicing of the debt you not do less than you easily could to avoid serious inconvenience to your friend. If you do less than you could, it would seem to be a failure of due care in the exercise of one duty with respect to another duty. I think the data from the cases tells us that as a moral requirement the duty of due care has no content of its own, but is a secondary duty whose application and content depend on independent or primary duties. There’s a nice term for it that Kant uses (though doesn’t invent): an accessoric duty—a duty that comes into play when another duty is to be acted on.⁵ The duty of due care implies a standard for action but does not provide one. Failure in its sphere means that something that required attention because one had a primary duty was not managed effectively. This is what divides negligence from pure accident. Due care demands that we attend to and avert the effects of a wide range of foreseeable sources of failure (both generic and personal) that impinge on dutiful action, and that we manage the intersections of action on a primary duty with other duties (possibly but not necessarily conflicting duties). The duty calls for no specific action or action-kind; it is not owed as a ⁵ See his Metaphysics of Morals 6:448. There’s a question whether due care is a separate secondary duty or a secondary part of what is required of us by a duty (any duty). One of the things we see in the examples is that in acting in the space of even a quite narrow duty like debtpaying we may touch various things. There are lingering effects of the way we chose to act, ways that may interact with other duties—things that the narrow description of what we have a duty to do won’t capture. That the moral fault in negligence is distinctive across cases suggests a category of duty. The lack of any content of its own suggests due care might be a formal or structural element of what a duty is. We will be in a better position to say more about this in the later sections of the paper.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

14

 

matter of right. Indeed, due care can sometimes be satisfied without our doing anything, though not in the manner of negative perfect duties, e.g., not to murder, which we can satisfy simply by not considering it; due requires our judging correctly that what needs to be done is being done. And since not everyone faces the same demands of due care in similar circumstances, we are often managing things on our own. Now here is the controversial claim. In the way that duty of due care is something like a managerial requirement, it concerns not just our actions but also our motives. Think of it this way. There’s a volunteer weekend and some dozen of us want to spend it refurbishing an inner city playground. A site has to be identified, materials purchased and transported, assignments made, and so on. One of us agrees to act as the group’s manager. She knows what the group is aiming at as well as the values and resources of the members. She knows that it matters to the group that the task be finished in the one weekend. Several people really care about repurposing old materials. There is a budget; they all want the playground to be attractive and safe; it would help get the work done if there were decent food available. The manager has a list of things to do. She also has a different kind of list—of things to attend to, to balance, to anticipate. Quirks and rules and relationships, missing tools and disagreements. She will need to be flexible and creative, focused on and able to focus others on what matters to the project and to the group. She must, in effect, organize her agency around their shared values. If we set aside worries sourced in the dictum, I think the natural thing to say about that second list, the one that organizes her agency, is that the manager must take on a motivational burden in accepting her role. If her decisions are made and expressed in ways that ignore the group’s values, she will be negligent because she failed to be motivated as she should have been. Of course the manager is a volunteer, so we could say that the negligent manager’s failure was in assessing her motives when taking on the job. Perhaps she should have declined it. That would resolve the issue about duty and motive. The problem is that this response is not available for the moral agent’s obligation of due care: she is not a volunteer. We can’t decline the job of effectively managing ourselves as we act for the duties and obligations we have.⁶ ⁶ At this point, an ordinary thought about motive may surface: surely when we ask about a person’s motive we’re asking about the psychological back story (what was her motive in taking the job?), not about the organization of her agency. There is no reason to resist that thought.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

15

So how should we resolve the conflict between this data and the motive-excluding dictum that comes with due care? We know three things. Non-negligence is not a reflection of the primary duty’s action-directive; negligence is not a fault in instrumental reasoning; there is no rule that fixes how one should act across varied circumstances. Both the key fact and the puzzle of due care is in the range of concerns an agent might have to negotiate in order not to be morally negligent. It has to be something related to primary duties that gives the agent direction. So let’s take a step back and focus on the kind of guidance a duty provides. A primary duty is rarely only a rule of action.⁷ Duties provide guidance in virtue of both their content and their value. Sometimes all that is required of us is that we avoid doing what a duty forbids (by its rule or content)—not to take more than our share or violate a right or fail to show up as promised. At other times we need to attend to what makes the duty matter (its value). It’s the duty’s value that sets the agenda for due care and that can make a demand on motive.⁸ Consider the promissory duty. It covers sub-duties about promisemaking, promise-keeping, and promise-breaking. They are the content conditions (or rules) of promising. There is also what promising is for—its object or value. Our grasp of that is what orients due care. If we take the point of promising to be about securing cooperation without depending on altruism, that puts the focus of due care on avoiding promise-breaking since defection defeats promising’s point. But if we thought the point of promising is to enable us to give another limited authority over what we do, our attention would be on trust as well as security of performance: that both parties act in good faith; that defection plus compensation is not an equivalent; that the promisee will not take inappropriate advantage. If it’s the first, the focus of due care is on preparation and performance, and perhaps in anticipation, on possible repair. If it’s the second, the concern is about maintaining the relationship that the promise brings about, or its fit with other relationships the parties may have.

However, as I’ll argue, the truth we see in the ordinary thought can mask the work motive does as a cause of action. ⁷ I don’t think it ever is, but I won’t argue for that. ⁸ This is why moral negligence can’t be regulated by an analogue of strict liability. The rationale is not about risk responsibility—who should bear it—but about norms of correct action.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

16

 

Or take the right to privacy and different accounts of what it, or the correlative duty, protects. There need be no disagreement about the kinds of actions that invade privacy, but different accounts of what the right of privacy is for (its value) can yield different contents for a related duty of due care. If the right protects a valued interest, compensation for violation (for the sake of other interests) is within the horizon of the right, and that would affect how careful one needs to be. But if privacy matters as an integrity value—because it partly constitutes the social boundary of who we are, of our identity—then due care will call for a high threshold of alertness so that privacy concerns are kept separate from other interests. As with promising, without an account of the value of the right or the duty we do not fully grasp the terms of compliance. Because it reflects our being beholden to the interpreted value of a duty, due care makes demands on an agent’s ability to draw the value of the primary duty into her practical deliberation and proactive attitudes. It is that feature, a demand on a broad practical ability, that I think has to be rooted in motive. A friend of the dictum might respond: Why motive? Can’t someone just be schooled to act in all the ways a duty’s value directs? Well, how? The point is that there is no finite list of actions to be taught, no lessons to practice: due care can call for an indeterminable variety of duty-relative responses that outstrip any fixed curriculum. Since this is a significant point of resistance, let’s walk through one more extended example. George knows that one should be on time to meetings. Rushing in at the last minute counts as being on time but is also disruptive. If he made a little extra effort with bus schedules, he could arrive a few minutes early and the meeting would start smoothly. We tell him that. We haven’t told him he had a second duty he was unaware of—that there’s a duty of being on time and one of being considerate. What we’ve alerted him to is that the point of being on time is being considerate. But it isn’t always inconsiderate to arrive in the nick of time. One could imagine a group indifferent to hurried arrivals so long as everyone is seated when the gavel is struck or the movie starts. But if the point and value of being on time in the meeting context is to facilitate a multi-person activity, it’s a disrespectful lack of due care not to make the extra effort. That’s what the just-on-time negligent George fails to see. I want to say that George does make a mistake about what the beingon-time duty calls for. He misses something. But since the duty is not a set of performance requirements with George missing one of them (it’s not like missing a step or an ingredient in a recipe), the failure is not in the province

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

17

of belief. If George had a grasp of the value of the duty he would be alert to how he should adjust his actions to meet the contingent needs of different settings. Correction does not add due care to the being-on-time duty as an extra rule of orientation; what is wanted is George, as an agent of the duty, motivated to exercise due care in realizing the value of the duty in his acting. In light of this, it might be tempting to take a page from negligence as the law manages it and assess George using a reasonable person standard. After all, the point of such a standard is, in a way, to evade the problem at issue: how to hold someone responsible for something he doesn’t see or recognize. But the reasonable person standard is part of a theory of juridical accountability for harms, and that doesn’t fit George’s very ordinary case. I think we make better sense of George’s having done something wrong if we can allow that his failure involves a motive he needed to have had for him to track the moral values in play. To make this a real possibility, we will need both a fuller account of what a motive is such that it can play the value-tracking role in an account of wrongdoing and a way of talking about wrongness and motive that remains sensitive to the dictum’s challenge that we maintain a connection between wrongness and what we can will.

2 We seem to be faced with two conflicting positions. On the one hand, I am to accept that we can’t love or feel pity or be outraged by injustice at will. Knowing we should be otherwise, we can take steps to change, but we can’t get there by bare choice. So no duty directly to have a feeling or an attitude or a motive that we might lack. On the other hand, I believe, and want to argue, that as moral agents, as bearers of duties and obligations, we have a duty of due care even though satisfying due care can require our having some motive. I take the first position to be undeniable. I find the second supported by the moral data. So what to do? In the back of my mind is this thought. If we are not troubled by Strawsonian reactive attitudes as a necessary element of our moral agency (attitudes that are beholden to norms of appropriateness), why be troubled by a connection between moral agency and motives (think proactive attitudes) that organize the agent, connecting her actions and attitudes to what matters in having a duty?

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

18

 

This suggests a strategy and a crucial distinction. Strawson’s point is that as moral subjects we have reactive attitudes: those we regard as moral agents we judge according to attitudes we impute to them about how they regard us (and, reasonably, themselves). We don’t demand of human beings that they have attitudes they may lack; however, in taking another to be a normal moral person “we demand some degree of goodwill or regard.” Analogously, we might say, due care doesn’t require of someone that she have a motive she lacks; nonetheless, a motive can be a condition of our doing what due care requires.⁹ We might think of it this way. You couldn’t have a duty to have hands or sight, although having them, you can have duties that depend on their use, even duties that require you to right now do more with them than you have done before.¹⁰ Now, you can’t have any duties unless you have certain practical capacities. If having the right kind of motive is among them, the argument is on its way. Now, not just anything we might call “motive” will do. Whatever its neuro-physical or rational nature, whether it involves a responsiveness to reasons or is something independent, motive, as I want to use the term, designates an action-oriented organization of the whole agent.¹¹ Instinct, not desire, is the better model. When we think of a desire, we tend to think of an affective state with an object (or object-type) that moves the agent to activity when an object-relative trigger is encountered or recognized. When we think about instinct, we think of a system state involving arousal, heightened attention, quickened activity, focused on an object as fits the instinct, moving the subject towards its object in a characteristic way. It’s this global aspect of the instinct model that I would have “motive” inherit. It puts the emphasis on activity, on the causality of agency, and on value.¹² ⁹ If there could be a kind of agent that negotiates the terrain of due care by means of a different order of cognitive processing, I doubt its primary duties would have the same form as ours. ¹⁰ That I’ve never used my hands to perform a water rescue or to staunch a flow of blood is no reason to think I can’t be called on to use them for that purpose if the need presents. I may be surprised by what I can do. ¹¹ I hear in ‘motive’ something from the French motif: the subject of a painting or a melodic phrase which provides structure to a musical work and acts as a unifying element. ¹² I speak of models here because motive and desire certainly and instinct to some extent are terms of art. We say that Hume has a desire theory (the complexity of which is often not reflected in modern desire theories), and Kant has a motive theory, though when he speaks of motives, especially the moral motive, it tends to be mistakenly read as, functionally, an odd sort of desire. Neither Hume nor Kant need to be read as hostile to motive in the system sense. Treating motive as a kind of disposition is a more plausible move from the Humean side. It’s not a counter-direction for this discussion because what counts as a disposition can range from primitive and isolated response features of the organism to tendencies to recite romantic poetry.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

19

Because motive in the system or agency sense is about how things go from the point of view of the responsive and active agent, it’s a model that suits the complexity of moral action. In the face of a duty, say repaying a debt, a suitably motivated moral agent is active, focused on a task that means something, aware of how she should act and ready to do so. The debt is on my mind; I’m already on my way to taking care of it even if it’s not due until next week; I am, if the debt is personal, grateful; and if I know you are uncomfortable about such things, already seeking a way to show my gratitude without embarrassing you. If on the day I come to pay my debt you turn out to be out of town and I have to leave what I owe in your mailbox, with no way to express my gratitude, it’s not merely one among the many things I planned for that I now cannot do, my debt-paying activity is now incomplete (leaving a sense of loss and regret). Of course if the debt is to my bank, my attention is fixed on making the payment on time, period. The different responses don’t imply a different duty or a different motive. It’s awkward to capture this with a desire model.¹³ Since the role of desire is to orient the agent to bringing about its object, the work of a desiremodeled moral motive is to identify and secure the performance of a required action (paying the debt), as if an object of desire; there is no natural connection with the value content of the duty.¹⁴ I find it helpful in thinking about the system idea of motive to look at the person from the perspective of the active human body. Consider the difference between thinking of hunger as a motive (or as motive-like) and hunger as a desire. I think of the desire as an activity-inducing physical module that turns on and off in response to calorically significant bodily changes. The prompted activity that began as an infant’s rooting develops soon enough into seeking breakfast and then even interesting or desirable breakfast. The structure remains the same: in a healthy person, some physical change or state triggers some kind of food-seeking activity. That’s not wrong, but there’s something odd about the modular picture. We do have responsive modules that relate to hunger and its satisfaction. The pancreas is one. Its function is to help with the digestion of food and to regulate blood-sugar levels (that in turn affect feelings of hunger). The gastro-intestinal tract is another: it secretes ghrelin that arouses appetite ¹³ Awkward, not impossible. Ptolemeic effort can align the results of very different models. One may, however, wind up with something like Hume’s double relation of impressions and ideas. ¹⁴ It’s not just a problem about desire. Dispositions set to track the balance of reasons are usually expected to latch onto the action or effect the balance favors, not the value.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

20

 

and affects when a person reaches satiety. But hunger is neither just a response to the presence of ghrelin nor just a response to the presence (or absence) of food. Hunger involves multiple systems and organs, triggers, arousal, attentional focus, perception and seeking, engagement with memory, and a characteristic (though not necessarily reliable) feeling. In priming the agent for an object, it may depend on past experience (infants suckle at all sorts of knob-like objects until they are organized enough to put smell and feel and familiarity together to actively seek the breast). Given its complexity, hunger may also be interfered with at different points in its emergence and by the demands of other systems. Associations are formed that cancel attraction to earlier desired objects; development and learning change and enlarge the available array of responses, leading in the higher mammals to the possible exercise of active choice. Hunger, so described, involves the organization of an active being’s agency. That’s hunger as a motive.¹⁵ When we tell this kind of story about motive we are at once naturalist and psychologist. Suppose we were interested in competitiveness as a motive: it affects perception, judgment, and physiology in certain settings. Not a desire for winning or a desire for self-advancement. It involves a complex social relation to an opponent that one seeks, though need not expect, to best. Competitiveness doesn’t make sense in all creatures. Some are natural cooperators; some go it alone; some will struggle with others if a resource is scarce but don’t seek or need the struggle. An agent moved by competitiveness looks at the struggle in a positive way. We should say that something in the creature’s nature, its make-up, tends this way or can be shaped that way. A naturalist will tell a story about how the tendency serves the species. A psychologist will locate the drive in the economy of the individual, marking the difference in the competitiveness that belongs to mate-seeking and that of playing chess. It’s of interest where the sources of pleasure are, where danger is sought or tolerated. Both instincts and motives have vicissitudes. In humans, the work of mechanisms like repression and sublimation can mask the deep structure of a motive: we trace out how mating drives lead to shoe fetishes, or competition to a monastic life.

¹⁵ And what of desire? It will still make sense to talk about desires in the way we talk about urges—that middle of the night desire for chocolate. And we might want to see desire as a kind resultant—an action-prompting state caused by the motive system honing in on an object. What is abandoned is the belief-desire model of explaining the generation of action.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

21

This seems to me a natural way to talk about motives. We say that self-interest leads Mary to focus on a goal, ignore a loved-one, lose sleep. Patriotism makes George susceptible to fear-mongering by politicians and suspicion of his neighbor, to wear funny clothes, even to give up his life. In each case motive does not point to an object, even a vague one, that instrumentally shapes subsequent response and activity, but describes a potentiality of the agential system as it responds to some value. If we add “rational” or “moral” to the naturalist’s account, we are not merely adding something to a response set or object-palette. (As if to be rational is to be an animal +.) Both norms introduce distinctive cognitive and affective organization of agency, involving their own triggers and forms of arousal, attentional focus, and a suite of responses. Rational concepts and forms of reasoning have a role in the organization of agency: in coming to judgment, in the way we act, and in the mix of appropriate responses to all stages and perturbations of the practical situation. To be in the grip of a rational motive is not just to be under pressure to do something or realize some outcome or even to instantiate the directive of some principle. It involves an organized response of the person to a principle-involving value.¹⁶ So, likewise, when we compare the agent moved by sympathy with another moved by the moral motive it’s not just that the former attaches to pain and distress and the latter to the principle of beneficence; the two agents read and respond to need differently, each of their motives carrying a different organizing value. A moral response to the pain of a person calls on us to give her some say about how and even if her need is to be met. “How can I help you?” is not just a façon de parler. This gives us an idea of why in terms of content and structure, the candidate moral motive as a system responsiveness to moral value better fits the data than something modular that provides the motive drive for moral action. That still leaves a possibility question: is our nature such that it supports a distinctive kind of agential responsiveness to a principleinformed judgment that articulates a rational or moral value? There’s some mystery about how even to approach such a question. I take it to be about the existence conditions for a kind of being. Here’s what I mean. For a living being there must be a significant match between its needs, its capacities, and its environment so that, actively or passively, it is able to sustain itself through generation and change. If it’s an animal, it has basic ¹⁶ I see shortness of breath; the EMT sees the early stages of a heart attack. I am moved to calm the agent; the EMT acts aggressively to initiate life-saving measures.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

22

 

systems that manage pain, hunger, and thirst, as well as response to danger, each able to trigger life-supporting activity. They are, in the language of this paper, system motives (or their precursors). Some animals have strong social motives that lead them to patterns of feeling and behavior that are not always optimal for the individual’s material well-being. They will engage in dangerous mating practices, forego eating for the sake of their young. If, as with African elephants, herd attachments leave them mourning their dead, the fact that mourning depresses life does not show that there are conflicting systems of desire, one for intimacy, one for life, that vie for dominance. It makes good or better sense to posit a complex principle of affective organization or a system of motivation that we would expect to manifest in a variety of context-appropriate ways across individual and social behaviors.¹⁷ I think it is not implausible to regard the moral motive as in this sense an element of the existence conditions of a kind of rational being (as reflected in its perceptual, cognitive, and affective systems, as well as in its habitat). So if the existence conditions for us as a rational life form involved morality, understood as respectful relations that involve recognition of the value of persons as such, then human beings would have to have, at least in potential, a distinctive agential organization unlike other natural agents, one that can be responsive to the normative demands of duties and the values that support them.¹⁸ Although I think we have evidence that supports this (despite the painful present), I regard the argument as transcendental, not empirical. If you thought we were nothing so fancy, just smart animals with complex welfare-functions, that might speak against our having a moral motive, though not against the idea of system motive as such. I believe the moral data points to a richer conception of our nature. It is no counter-argument that it can take experience, relatively benign circumstances, and a morality-supportive social environment to get the motivational system of the moral agent working right. After all, we don’t ¹⁷ An implication of the system view that I’ll note and then set aside for future discussion is that you don’t have to be a natural agent to have motives in the system sense—i.e., you don’t have to have a psychology. What matters is whether an entity’s causality is organized by principles that express and are responsive to values. ¹⁸ This is in fact Kant’s view. The moral motive is a system motive, an organizing principle of the human will as the capacity for good willing (Kant speaks of “determining ground”). It is a source of our pleasure in respectful relations and our sense of offense at intentional cruelty. It is constituted by, not attached to, moral principle. This sort of view is, happily, confirmed in integrative behavioral and neuroscientific studies that track the emergence of moral behaviors in system-developing stages. See, for example, “The Development of Human Fairness,” in Nature Human Behavior 1, 42 (2017).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

23

think that health is self-generating either, though we assume a desire for it. Perturbations and distortions are normal; a motivational system is typically robust enough to allow for self-correction, though not always or easily or in all circumstances. We expect mature persons to take a fair amount of responsibility for their cognitive and their motivational systems functioning well. Suppose we find that racialized anxiety affects someone’s practical judgment and action. We don’t infer an absence of a moral capacity; we instead recognize the role of this kind of fear in distorting a person’s practical organization. Even when the region of failure is not first-person accessible or even first-person correctible, we are not excused. Think of the discussion around implicit bias—the idea that persons can sincerely avow a standard of impartiality that they not only fail to meet, but they also cannot identify the source of failure in themselves. Correction may have to be drawn from a non-agential source: we can use procedural devices to approximate a morally defensible outcome, or seek a change in contributing social conditions. It matters to response and correction what the nature of the defect is. It could be that an agent’s motivational system is deformed or corrupted. This might show in a sense of satisfaction at the confirmation of a biasdetermined outcome (consistent with an official rueful attitude). It’s also possible that a person can be moved but not motivated by her bias, so a more directly causal rather than an agential explanation is warranted (as is the case with our tendency to vote for the taller candidate, or buy the item to the right in our visual field). Noticing this can cause surprise and distress, though also greater openness to change—we have less at stake in some of the things that merely move us. We cannot always tell whether we are moved or motivated. We sometimes discover after the fact that what moved us was other than a motive: we unwittingly got caught in a cultural stereotype or were swept away by social pressure. It is on us to be alert to these effects, and where tools of resistance or correction are available, to make use of them.¹⁹ Sometimes what merely moves us is something of value. Someone finds he has a fascination with form and color that intrudes on his career as an X. The fascination can develop into a principle of his activity, a motive, if he becomes a Y. Managing the tension between being motivated and being moved is an ongoing project of acceptance and sublimation. It helps us make sense of our vulnerability to

¹⁹ We can ask a parallel question about the surrounding society that encodes bias in its institutions and culture—how much is passive legacy, how much living value.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

24

 

akrasia and phenomena like adverse preference formation. This sort of complexity is not an unwelcome cost of adopting the system view. To get us back on track, recall that the problem that set us on the path of a system motive was the awkward finding that the duty of due care had regions of requirement where the duty’s satisfaction depended on a motive—viz., to self-organize in order to manage an open-ended, valuedirected, moral task. The thought was that a different understanding of the sort of thing a motive could be might eliminate the reason for the blanket prohibition on motive-dependent duties. The target for the inquiry was then a motive that does not need to be summoned at will, tracks relevant value, and is sufficient for its object, dutiful action. If the moral motive is a system motive, it is also a species of rational motive and thereby inherits much of its structure, though not its content or its connection with moral necessity. Normal human agents have a capacity to carry intentional action into circumstances they neither control nor fully grasp (circumstances that include opaque features of their own agency), while adjusting their actions to both the goal and the point of their acting. This calls for more than instrumental resiliency. Persons also need the ability to be responsive to the values that make their ends and actions worthy of choice. Both abilities direct an agent’s causality to an intended effect; both are embedded in practical skills that involve attention to working around unanticipated obstacles; but it is the value sensitivity that keeps the agent’s activity anchored in her rationale for setting her end in the first place.²⁰ Responsiveness to value in acting is not something extra, not something about our character or the meaning of our actions in some larger sense; it’s

²⁰ As I was writing this, there were fires burning in the Los Angeles hills a few miles northwest of my office. I thought: It’s the end of term; I have scheduled office hours for nervous students with a paper due. The campus has been declared safe; I’ve checked my traffic app and see that I can get to my office without too much delay. My instrumental reasoning was done. But I had other concerns, of a different sort. Some were about contributing to the traffic nightmare unfolding; quite a lot about what my students needed and how my decisions would affect them, about the contours of my responsibility and what else I might do that could make a difference. Coughing, I was reminded that there were health issues for me (I was just getting over the flu). I want to say that what set this mix of concerns in motion was a motivational engagement I have with my work as a teacher. It both prompted my projected end—getting to my office by 1:30— and injected it into a lattice of values that not only affected my decision about what to do in these circumstances, but also about what to tell my students, what new options I needed to make available if the emergency continued (did I really need them to bring in hard copies of their final paper), and so on. It was an extreme situation, but formally entirely ordinary. Failure to give adequate attention to the values in play (I think of them as saturating deliberation), not only risks our reaching the wrong conclusion about what to do, but we may then mismanage, because we fail to notice, options and collateral needs.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

25

about getting things right. I may play tennis because it matters to my health or I just enjoy it or it’s a venue for maintaining a friendship. Whichever is the case affects how (and sometimes whether) I proceed. If it’s health, I might do something else equivalent—or, if I have a tendency to flake, take extra vigilance; if it’s enjoyment, I might decide “not today”; the connection with friendship introduces a host of things I need to be attentive to, including, though now for a different reason, being on time. Negligence, carelessness, and the like, are threats to end-pursuit generally. Part of our maturation as a rational agent is to develop both instrumental discipline and non-instrumental managerial skills to stay connected to what matters. It is one’s motives, be they a concern for health or pleasure or friendship (singly or nested), that provide the connection—a normative connection—between end, value, and action. Of course I have significant latitude in the face of the value of my discretionary ends. There may be cases where the costs of staying with what’s of value makes a project or pursuit not worth maintaining. We lack this freedom with ends or actions that fall under moral duty. The point is that when we turn to the moral case, the foundation of due care is already part of the fabric of rational action. That there is a duty of due care of the kind I have described reflects the nature of the values that give rise to the different duties and the kinds of practical problems they pose for us. We can have such a motive-dependent duty because we already have in place the capacity to be responsive to value in a motive-involving way. The agent who fails at the duty of due care does not err because she lacks a motive; she had the motive as a capacity, a potential that she failed to deploy. In focusing an agent’s attention on a primary duty’s value, the duty of due care sets terms for reasonable response to the situational complexity that comes with our acting as duty requires. We may need to see, feel, anticipate, plan, leave ourselves vulnerable, take our time, engage others even when, if we were just thinking instrumentally, we could realize the bare object of a duty directly and by ourselves. A morally competent agent comes prepared. To be sure, not every moral agent is motivationally alive to the values that belong to her duties, and even if she is, she may not pick up on all that matters. We have structural limits and we are not always in peak form. The question is whether, given the presence of a capacity sufficient to engage with the relevant elements, failure that has its source in motive is any more problematic than failure of cognitive capacities that lead to moral mistakes. If we can be responsible for failing to put two and two together, we can be responsible for inattention to privacy concerns as our beneficent actions blunder into information that should not be shared.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

26

 

3 None of this is to deny that there is a third- or second-personal perspective from which we may be indifferent to the concerns of motive. An insensitive, underprepared, or hasty agent may tell the truth, return what’s owed, violate no rights, indeed meet all the demands of a moral situation. We might think of this as the view of moral requirement from the perspective of acceptable complaint. Things look different from the perspective of the duty-responding agent. She borrows a book; intends to and does return it. The episode of borrow-and-return has a middle period—an arena of agent responsibility. It doesn’t govern the order of chapters read, or who you talk to about what you read, but it does have something to say about wear and tear and about at whose convenience the return is made. A specific duty gives us a target and a forward looking sphere of responsibility much of which is managed by the duty of due care. It’s the task of the moral motive to prime the agent to meet these demands. In fact, much more than due care is involved in a motive-involved moral competence. Think about what happens when we acquire a region of duty— a domain of moral concern that centrally includes requirements on action or ends. The concept that something belongs to someone else is not, when we grasp it, just another bit of sortal moral information. Understanding that some thing could be “not mine” and instead “yours” changes a person’s orientation towards the physical world of things she encounters. Some things can be used but must be returned (it’s Mary’s truck); other things can be used but then must be left in place when you’re done (those trucks stay in the playground for whoever wants to play with them); still others can come to be yours (through gifts or purchases or chance finding). The objects may all be the same, but their place in the scheme of moral relations is quite different. For the person acquiring the moral concept, it’s not just about whether something can be kept or has to be returned. There are matters of delight, the enjoyment of possession, and the different possibilities of sharing. The issues around using public common objects together are not the same as when we share privately owned objects (e.g., playing under the implicit threat of Jane taking her ball home). The spaces objects occupy look and feel different to us; we move among them in distinctive ways. Love and aggression are directed at persons via their owned objects. We act to preserve what a loved one values; we give gifts. We may need to manage envy. Since we are talking about a system motive whose elements need to be actualized, we should expect the same motive to manifest somewhat differently

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

 :    

27

in different agents. Moved to help, how readily one responds to some need can depend on past experiences. One person has vivid memories of earlier successes at rescue; another carries the painful imprint of a complex failure. The one reinforces practical confidence; the other creates space for hesitation. These are not counter-motives but factors, like one’s degree of physical strength or dexterity, that the motivated agent works with. She may seek a tool or help to take up the slack, or knowing her own anxieties, just like knowing her physical limitations, may give her reason to defer to a different actor. Barring something unusual, difficulty in keeping track of one’s money is not an excuse for failure to pay debts. Morality has its own reasonable person standard. Where one is unable to meet the standard, there may be spheres of activity that are off limits. With all of this in place, we can take up the question whether and how motive can affect the moral valence of an action, making an otherwise notimpermissible action wrong. A positive answer turns on whether a wrongful performance can be a result of a failure in value-tracking. Consider two cases. The first case is of a landlord who uses his discretion in selecting tenants to avoid renting to someone because of his race.²¹ Even if he is acting “within his rights” the action is clearly wrong. I think what makes the action wrong is a motive that draws an “externally rightful” exercise of discretion into a deliberative schema that tracks contrary-to-duty value. It subverts discretion. In the second case motive makes an action wrong because the controlling value it introduces in the acting puts someone at unjustified risk. We can borrow Derek Parfit’s example of a Mafioso entering a coffee shop poised indifferently to pay for his coffee or shoot the barista as fits his mood and the moment. As it happens, he pays. Parfit concluded: bad motive, but nothing wrong done. I think it’s more like driving without using brakes: wrong to do even if no one is injured. So not a case of ‘no harm no foul.’ Generalizing, the typical instance of motive-involved wrongdoing occurs when an action of a type not in itself impermissible has its source in a motive that causes a mismatch between the value of a governing moral requirement and the value the agent’s motive in fact tracks. Think, for example, about keeping a promise to create a false sense of trust (like a long con), or acts of competitive gratitude. By subverting the action’s apparent value, the motive can in effect make the action a lie. What looks all right isn’t what it seems to

²¹ The example is drawn from T. M. Scanlon, Moral Dimensions (Harvard University Press, 2008), 71–3.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

28

 

be; as what it really is, it is wrong to do. If that’s right, motive in such cases is a proper part of action description. Not surprisingly, many cases of motive-involved wrongful action involve value-subverting exercises of the discretion that comes with imperfect duties. These are regions of duty-governed activity where what we do or how we act is, to a considerable extent, up to us. A morally bad motive is one of the limits to discretion—as when one’s motive in helping is to foster dependence or one’s motive in giving a gift is to belittle the recipient. The reason such a motive is wrong-making is not that it’s nasty; rather, as with other cases, the value the motive brings to the doing makes the action inconsistent with the value the duty directs us to realize in our acting. * * * I don’t expect this account to put the majority view about wrongness and motive to rest. It is too entrenched, dependent on underexamined views about what motives are and an unnecessarily juridical account of wrongful action. My hope is that the alternative I’ve described is sufficient to motivate some further thinking.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

2 The Normative Structure of Request Micha Gläser

1. Introduction Request is curious. A moment’s reflection reveals as much. For example, we often say such things as “Can’t hurt to ask” and “It’s okay to say no,” yet we also frequently vex over whether to ask and agonize over saying no, and when we do ask, we often don’t actually think that it’s okay to say no if we’re honest with ourselves. How do the former, more easy-going qualities go together with the latter, more high-strung ones? For another example, we sometimes say things like “So-and-so asked me to do such-and-such, so I did it.” Normally our audience will be inclined to accept statements of this sort as perfectly good answers to the question why we did what we did, not least so-and-so, who will likely find the answer especially gratifying. A request thus seems to at least potentially constitute a perfectly good reason to do the thing requested. At the same time, we don’t generally go around making requests of others without ourselves having good reason to do so, and were we to find out that this is what so-and-so was doing, we may well not take kindly to it. However, if we shouldn’t be asking without being “backed” by good reason of our own, then doesn’t whatever normative force a request of ours possesses simply derive from the thing that gave us reason to ask in the first place, and if so, how could a request as such ever provide reason to do anything? And yet it seems to. While these various curiosities make request a low-hanging fruit for philosophical investigation, there is little literature on the subject, and less still that does them even approximate justice. The present chapter attempts to deliver progress on this front.

2. Normative Effect and Discretion We might think of request as an interpersonal normative performative. Like an Austinian performative (see Austin 1975), a request purports to be doing Micha Gläser, The Normative Structure of Request In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0002

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

30

 ̈

something, rather than merely reflecting some state of affairs that holds independently of the act of address itself. More specifically, a request purports to be doing something normative. It purports to have a certain normative effect on the requestee. Finally, the normative effect in question must be conceived as somehow internal to the request’s being addressed at the requestee. There is a categorical difference between the normative position of the requestee on the one hand and that of a third party merely overhearing the request on the other, say. What is the purported normative effect of request? As noted above, a requestee’s “So-and-so asked me to do such-and-such, so I did it” indicates that she took the requester’s request as her reason for performing the act requested. A statement of this sort seems to reflect success on the part of the requester qua requester. The requester attempted to get the requestee to act “in conformity with the request from the request,” and this is what the requestee did. This suggests that a request presents itself as a reason for action for its addressee qua addressee of a request.¹ An account of request needs to account for its reason-constituting nature so understood. Call this the normative effect constraint. Command, consent, and promising are other species of the genus interpersonal normative performative. The first of these resembles request most closely. Consider a statement parallel to the one above: “So-and-so told me to do such-and-such, so I did it.” What the commandee is saying here is that she took the commander’s command as her reason for doing as commanded. Again, such a statement reflects success on the part of the addressor, which suggests that commands also purport to constitute reasons for their addressees. Where request and command seem to differ is in the kind of reason they purport to constitute. Request seems to amount to a normatively “feebler” form of address than command, to use a term of Elizabeth Anscombe’s (see 1981, p. 18). In particular, a requester in making her request grants discretion to the requestee as to whether to do the thing requested, whereas no such thing is true of commander and commandee. An account of request needs to capture the discretion-granting character of request. Call this the discretion constraint. Everyday life seems to confirm not just the existence but also the interpersonal significance of the difference in “normative grade” between request ¹ The reason in question thus possesses all the trappings of a “second-personal reason” in Stephen Darwall’s sense. See his The Second-Person Standpoint (2006), especially pp. 5–10.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

31

and command. People sometimes make a point in emphasizing that they weren’t trying to command their addressee but instead were merely making a request of them: “I wasn’t telling you to do such-and-such. I was just asking. You don’t have to do it. Feel free to say no.” Statements of this sort are meant to highlight the discretion the addressor grants the addressee in deploying the form of address of request rather than command. People sometimes also make a point in emphasizing that they were doing the opposite: “You misunderstood me. I wasn’t asking you to do such-andsuch. I was telling you to do it. You don’t have a choice here. You have to do it.” Or, shorter: “That’s an order!” Similarly, the addressees of putative acts of command sometimes protest their treatment by saying things like: “You don’t get to tell me anything.” Note that what the addressee is taking issue with here is the form of address of command as such, rather than the substance of the command in question, which for all the would-be commandee says might be perfectly unobjectionable were it to be put forward as a request. In addressing the addressee through the form of address of command, the addressor is guilty of “normative overreach” of a specifically interpersonal kind, or so the addressee’s rejection of the act of address seems to imply. The familiarity of these sorts of statements suggests that the normative space between request and command is not only genuine but also of genuine concern to us.

3. The Mere Addition Account The contrast between request and command is the subject of a brief but suggestive passage in Joseph Raz’s The Authority of Law: Requesting and ordering . . . entail intending that the act of requesting or ordering be taken as a reason to perform the act ordered or requested. What then is the difference between them? One such difference is relevant to our purpose. Suppose that a man makes a request and is told in reply that his request was considered, but on balance it was found that the reasons against the action requested overrode those for it including the request itself. He will no doubt be disappointed, but he will not feel that his request was disregarded. He has nothing to complain about. He must concede that whatever his hopes, he intended no more than that the action be taken on the balance of reasons, his request being one of them . . . This is

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

32

 ̈ not so if he gave an order. A man who orders someone else does not regard his order as merely another reason to be added to the balance by which the addressee will determine what to do. He intends the addressee to take his order as a reason on which to act regardless of whatever other conflicting reasons exist (short usually of an emergency or other extreme circumstances). (2009, pp. 14–15)

Raz’s suggestion here is that requests purport to constitute mere pro tanto reasons for their addressees whereas commands claim to impose (what he elsewhere calls) “protected reasons” on them, where a protected reason for some act is a normative compound consisting of a first-order reason in favor of the act and a second-order “exclusionary” reason against acting for other reasons for or against the act (see his Practical Reasons and Norms (1999), p. 191). Since on Raz’s view protected reasons just are obligations, it follows that requests purport to add mere (i.e., pro tanto) reasons to their addressees’ balance of reasons whereas commands purport to impose obligations on them. Call this the mere addition account of request.² The mere addition account provides for a straightforward interpretation of the two constraints from above. The purported normative effect of request consists in its adding itself as a pro tanto reason to the requestee’s balance of reasons. The discretionary element of request in turn resides in the pro tanto and so non-obligatory nature of the reason a request purports to constitute. By the requester’s own lights, a noncompliant requestee merely fails to comply with a particular pro tanto reason applying to her (namely the request). Hence, the requestee enjoys discretion in that her noncompliance as such isn’t a failure to do something she is obligated—and in that sense has—to do.

² Other for the most part cursory treatments of request occur in David Enoch, “Giving Practical Reasons” (2011) and “Authority and Reason-Giving” (2014); David Owens, Shaping the Normative Landscape (2012, see especially pp. 86–7, 98–100, and 228–9); Mark Lance and Rebecca Kukla, “Leave the Gun: Take the Cannoli! The Pragmatic Topography of SecondPerson Calls” (2013, see especially pp. 460–3); Margaret Little, “In Defence of Non-Deontic Reasons” (2013, see pp. 131–5); and James Lewis, “The Discretionary Normativity of Requests” (2018). While there are significant differences between these various discussions, they all subscribe to the mere addition account in the point that matters for my purposes, namely that requests create non-obligatory reasons. Since despite its brevity the passage just quoted offers an instructive treatment of the implications of the mere addition account, I will use it as my point of reference throughout. The one exception to the general partiality to the mere addition account of which I am aware is Geoffrey Cupit’s “How Requests (and Promises) Create Obligations” (1994). I will discuss Cupit’s account—to which I am extremely sympathetic—in slightly more detail in footnote 16 below.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

33

4. Compliance The violation of an obligation is usually taken to constitute a distinctive kind of normative defect. Specifically, it is generally thought that an agent who acts contrary to some obligation of hers is ipso facto guilty of having committed a wrong. Following P. F. Strawson’s well-known discussion in “Freedom and Resentment” (2008), a person who committed a wrong is usually taken to be the appropriate object of blame understood as a sui generis interpersonal attitude, with its subspecies of (“personal”) resentment (on the part of the “wrongee”), (“impersonal”) indignation (on the part of third-party members of “the moral community”), and guilt (understood as self-blame on the part of the wrongdoer). An agent’s failure to comply with one of her pro tanto reasons on the other hand doesn’t carry the idea of a wrong in its train, nor accordingly the appropriateness of blame. Crucially, this is often taken to be so not merely when the balance of her reasons speaks against whatever the particular pro tanto reason in question favors, but also when the two coincide. While failing to act on the balance of her reasons does render an agent liable to rational criticism, such criticism is different in kind from full-blown blame qua appropriate response to wrongdoing.³ Given the conceptual connection between obligation and the reactive attitudes, and given that requests purport to constitute mere pro tanto reasons, it seems to follow on the mere addition account that it couldn’t be appropriate for a requester to resent a requestee for not doing as requested. Yet it is easy to imagine at least apparent counterexamples here. Consider the following case: Movie Night Rick and Erica are friends. They live across town from each other. Erica is really good with computers. Rick is not. Rick has a critical deadline coming up the next day at noon. He is having trouble with his laptop. He has already tried fixing the problem himself, to no avail. He is reaching out to Erica to come over tonight to help him. She tells him that, while she feels for him, she would rather stay in and watch a movie. The appropriate reaction on Rick’s part here seems to be something in the register of resentment towards Erica. Rick does have “something to complain

³ Darwall states this point in a number of places, most explicitly in “Moral Obligation: Form and Substance” (2013), p. 43.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

34

 ̈

about,” despite the fact that Erica did consider his request (or perhaps precisely in virtue of how she considered his request). However, if resentment is the proper response to the suffering of a wrong, then the intuitive appropriateness of Rick’s resentment towards Erica implies that in turning down his request she in some way wronged him. Yet this is precisely what the mere addition account seems to have to deny. Movie Night-style cases thus seem to make trouble for the mere addition account. The passage from Raz quoted above may seem to provide the materials for a defense of the mere addition account here. Raz there says that the person whose request was turned down has “nothing to complain about” if the action was “taken on the balance of reasons, her request being one of them.” This suggests that, if Erica acted against the balance of her reasons, then perhaps Rick does have something to complain about. Let’s assume that the balance of reasons applying to Erica did speak in favor of helping out Rick rather than watching her movie. Her doing the latter therefore qualifies as acting against the balance of her reasons, the proposal goes, which in turn renders appropriate Rick’s resentment towards her. One bit of trouble with the proposal is that it puts pressure on the supposed conceptual connection between obligation and the reactive attitudes. What the proposal says is that persons are susceptible to blame not just for failing their obligations but also for failing to do what they have allthings-considered reason to do. The mere addition account so extended thus seems to collapse (mere) rational criticism into (full-blown) blame. Moreover, it is difficult to see what remains of the distinction between (personal) resentment and (impersonal) indignation, given that the appropriate occasion for blame on this picture—i.e., a person’s failing to do what she has allthings considered reason to do—as such doesn’t provide for the distinction between sufferers and mere third parties. But perhaps the distinction here could be understood in terms of the substance of the relevant reasons.⁴ Moreover, someone might reject the distinction between what the balance of a person’s reasons calls for on the one hand and what she is obligated to do on the other altogether and instead hold that to act wrongly—and so contrary to obligation—just is to act contrary to the balance of one’s (moral) reasons. Let me therefore set these prima facie concerns aside and instead turn to a deeper problem. ⁴ Raz’s own “interest theory” of rights, which grounds rights and their correlative duties in the substantive interests of the right-holder (see 1986, ch. 7), in fact provides the model for such an account.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

35

The problem is that, according to the mere addition account, the ground for Erica’s blameworthiness would have to be her failure to do what she has all-things-considered reason to do, rather than her failure to comply with his request, which—qua mere pro tanto reason—may or may not coincide with her balance of reasons. The mere addition account would have to say that, insofar as Rick blames Erica for her failure to comply with his request, he would be making a mistake similar to that of blaming a shopkeeper whose dishonest dealings lose him business for his imprudence. The mistake here is that, while it is true that the shopkeeper is being imprudent in being dishonest, it is his dishonesty that merits blame, not his imprudence. Likewise—the mere addition account would have to maintain—it is true that Erica failed to comply with Rick’s request in failing to act on the balance of her reasons, yet blame is due on account of the latter rather than the former, so Rick is making a mistake in blaming Erica for her failure to comply with his request as such. Yet this seems clearly false. Rick doesn’t seem to be making a mistake in blaming Erica for her failure to comply with his request here, pace the mere addition account. Note that the problem carries over to a second possible reply to Movie Night-style cases on behalf of the mere addition account, one that leaves intact the presumed internal relation between obligation and the reactive attitudes. According to this reply, what warrants Rick’s complaint against Erica is that she has (not merely all-things-considered reason but) an obligation to help him out, where this obligation applies to her independently of his request. Once again, the problem is that, since Erica’s failure to comply with Rick’s request is merely accidentally related to her failing to do what she is obligated to do on this picture, Rick’s complaint that Erica didn’t comply with his request comes out as misguided, in the sense that it picks up on something that is merely accidental to the thing for which Erica is actually blameworthy, i.e., her failure to comply with her prior obligation. Again, however, this seems false. There doesn’t seem to be anything misguided about Rick’s blaming Erica for her failure to comply with his request as such here. What I therefore want to propose is a third formal constraint on the concept of request. An account of request needs to explain how it is that compliance with a request as such might be obligatory for a requestee, rather than something that merely happens to coincide with what the requestee is obligated to do. I will call this the compliance constraint. The mere addition account is unable to account for this constraint. At best it is able explain how it might be obligatory for a requestee to act in conformity with a

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

36

 ̈

request—namely whenever the thing she has all-things-considered reason to do or is obligated to do anyway also happens to be the thing that is being requested of her—but not how it might be obligatory for a requestee to act from, or on, a request.

5. Good Reason The trouble for the mere addition account doesn’t end there, however. Note that request seems to carry obligation with it not only on the part of the requestee—as in Movie Night—but also on the part of the requester. Consider the following variant of the case: Easy Fix Rick and Erica are still friends and live across town from each other; she is still good with computers and he isn’t; he still has his critical deadline coming up tomorrow and is reaching out to her for help. However, now Erica agrees to come over, skipping her much-anticipated movie night, only to find that Rick could have disposed of the problem himself with minimal effort. He just couldn’t be bothered. It seems that now Erica would appropriately feel resentment towards Rick, which in turn suggests that, in putting his request to Erica, Rick somehow wronged her. He shouldn’t have asked her to come over under these circumstances, yet he did. Easy Fix-style cases thus pose a challenge to the mere addition account. It isn’t obvious how adding a pro tanto reason to a person’s balance of reasons might be the sort of thing that constitutes a wrong against that person. In reply, a defender of the mere addition account might argue that Easy Fix differs from Movie Night in that here the appeal of the movie night does outweigh Rick’s request in Erica’s balance of reasons, so that Erica does have all-things-considered reason to decline to help him and stay at home instead. The defender of the mere addition account might then invoke a principle to the effect that requesters owe it to their addressees to issue only requests which end up belonging to what might be called the “winning coalition” of reasons applying to the requestee,⁵ or at least only requests which aren’t going to be ⁵ A request’s figuring in the winning coalition of reasons need not involve its tipping the balance of reasons in favor of the act requested. The winning coalition may have won even in the absence of the request.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

37

clearly outweighed, or some such. Applied to Easy Fix, it follows that Rick owes it to Erica not to ask her to come help him and so wrongs her in doing so. There is something suggestive about this proposal. Cases such as Easy Fix illustrate that we owe it to others not to request them to do things which we believe they lack (what I will call) “good reason” to do, where a good reason provides a good answer to what following Anscombe we might call the “Why?” question, that is, an answer that renders the thing done rational, in a perfectly respectable sense of the word. When a person has all-thingsconsidered reason not to do something, then presumably she also lacks good reason to do it and in fact has good reason not to do it; this is why the proposal is suggestive. What we have here is therefore a fourth formal constraint on the concept of request. It is incumbent on an account of request to explain why requesters owe it to their requestees only to ask them for things the latter have good reason to do. I will call this the good reason constraint. The trouble with the proposal is that the principle to which it appeals— i.e., to make only those requests that would figure in the requestee’s winning coalition of reasons—is inadequate to the task of explaining the kind of wrong Rick is committing against Erica in Easy Fix, and so ultimately the kind of wrong to which the good reason constraint makes reference. The point might be brought out by reflecting on the ground of Erica’s resentment towards Rick here. I suspect that Erica will feel as though Rick led her on. In his very act of address, he acted as though things were one way, but it turns out that they were really another way. Specifically, in making his request Rick had acted as though Erica had good reason to come over and help him, when in reality she didn’t. The complaint underlying Erica’s sense of anger against Rick is that he is guilty of a form of insincerity towards her. However, if Erica justifiably feels led on by Rick, then it seems that she was entitled to presume that she did have good reason to come over and help him simply in virtue of his making his request of her. What this brings out is that the good reason constraint must be understood as internal to the form of address of request. To request something of someone knowing that the person doesn’t have good reason to do the thing requested isn’t to commit a wrong against the requestee that is merely accidentally related to the nature of one’s act of request qua act of request. Rather—and to use yet another term of Austin’s—it is to abuse the form of address of request. It is to act in bad faith.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

38

 ̈

Note that insincerity is only one of two possible forms of bad faith here. The other is what might be called “brazenness.”⁶ Suppose that Rick is unapologetic when Erica takes him to task over his request, or suppose that he already tells her on the phone that he hadn’t put in even five minutes of his own time because, “Why bother if I can get you to do it instead?” Presumably Erica will feel the same kind of resentment here as she would have had he instead led her on in the way described above, even if here she didn’t actually make the trip. This suggests that the cost she incurs in traveling across town in the original description of the case is merely incidental to the ground of her resentment towards him. Rather, it is the bad faith he is exhibiting in both versions of the case. It is the internal character of the wrong at play here which the present proposal on behalf of the mere addition account fails to capture. Whatever the nature of the wrong captured by the obligation I imagined the mere addition account as proposing in reply to Easy Fix, it isn’t the wrong of abusing the form of address of request. The wrong of making a “losing” request as the mere addition account has to conceive of it is extrinsic to the normative structure of request. According to the mere addition account, in making her request the requester creates a mere pro tanto reason for the requestee in favor of the act requested. The obligation on which the proposal turns on the other hand concerns the requestee’s balance of reasons, with which the request qua mere pro tanto reason ipso facto stands in a merely accidental relation. Perhaps the point might be clarified by analogy to how utilitarianism has to conceive of the notion of moral wrongness. Suppose that, instead of aiming at maximal net overall utility, I am tending to some personal, nonoptimific project. I am thereby acting wrongly according to at least certain standard versions of utilitarianism. However, the wrong of which I am guilty here isn’t that of bad faith. I am not in the very act of pursuing my personal project insincerely or brazenly presenting myself as in the business of maximizing utility. What this shows is that the principle of utility constitutes an external rather than internal standard of action. The mere addition account is in a position parallel to that of utilitarianism here. The only way in which it could account for the wrong of which Rick is guilty in Easy Fix is as one external to the form of request as such. It therefore cannot account for the specific nature of the wrong of which ⁶ The distinction here is, I think, similar to the distinction between insincerity and “shamelessness” Philippa Foot draws in her Natural Goodness (2001, see p. 19).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

39

Rick seems to be guilty. Nor does it help the mere addition account to appeal to some independent obligation we owe to other people not to lead them on or the like, of which Rick runs afoul here and which therefore accounts for the propriety of Erica’s sense of resentment towards him. The trouble is that the mere addition account lacks the resources to explain why Rick’s request counts as leading Erica on and as such qualifies as a violation of the obligation in question. To invoke an obligation of this sort in reply to the objection here is therefore to effectively concede the point.

6. Transparency The problem is fundamental. Recall that the mere addition account has it that the normative effect a request claims for itself is that of constituting a pro tanto reason next to and on a par with all the other pro tanto reasons applying to the requestee. As such, it purports to constitute a pro tanto reason for doing the thing requested independently of those other reasons.⁷ It follows that there couldn’t be an internal standard governing what one person might ask of another. Any such standard would need to take the form of an external principle, such as the principle entertained in the section right above to only make requests that would belong to the winning coalition of reasons applying to the requestee. In other words, the mere addition account portrays the requester as a kind of practical King Midas, whose speech has the power to turn anything it “touches” into a reason for action. Whatever moral principles the mere addition account might devise in order to “capture our intuitions” about the cases adduced above are bound to serve as merely external constraints on the requester’s exercise of her Midas-like power. Any such constraint would have to be akin to a moral principle telling the actual (i.e., mythical) Midas not to touch other people, say: if he were to touch them, he would be turning them into gold just like anything else he might touch; it’s just that he’s not supposed to (or rather, it’s just that he’s not supposed to). But this misdescribes how a good-faithed requester conceives of her own normative power qua requester. A good-faithed requester instead takes the normative force of

⁷ This doesn’t mean that the mere addition account couldn’t allow for the latter to outweigh the request. The point is just that its being outweighed wouldn’t derive from anything internal to the request as such but instead would be a function of its (merely external) relation to its normative “environment.”

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

40

 ̈

her request to be transparent to the normative force attaching to conditions that hold independently of her act of request.⁸ The idea of transparency so conceived provides for further specification of the good reason constraint. For her request to qualify as good-faithed, the requester has to believe that it reflects something that independently gives the requestee good reason to do as requested. If her request fails to meet this condition, it fails on its own formal terms.⁹ To make the point more vivid, consider what Rick might say to Erica in making his (good-faithed) request of her in Movie Night: “I know it’s a lot to ask, but would you mind coming over to help me fix my laptop tonight? I have this crucial deadline coming up tomorrow, so you would be helping me out here big time here. I’ve already tried to fix the problem myself but wasn’t able to, and I know you’re really good with this stuff, so it’ll probably take you fifteen minutes tops. So: would you mind?” As the mere addition account would seem to have to analyze matters, Rick must be understood as shifting normative gears between the first and second sentence of his little speech and then back again between the second-to-last and last. In, or through, the first and last sentence he is creating a pro tanto reason for Erica, whereas in between he is reporting on all the independently existing reasons that apply to her alongside his request. However, this seems to be the wrong picture of what is going on here. What Rick is doing in those middle sentences is explaining the rationale behind his request—and in doing so establishing its bona-fide nature—rather than pointing to reasons that exist independently of the reason he is creating by uttering the sentences before and after.¹⁰ ⁸ For what I take to be related uses of the notion of transparency, see Richard Moran, Authority and Estrangement (2001), pp. 60–5, and T. M. Scanlon, What We Owe to Each Other (1998), p. 133. The same goes, I think, for A. J. Julius’s principle of “Independence” in Reconstruction (2013, see p. 8), absent the language of transparency. ⁹ There is some similarity between my argument here and Raz’s “dependence thesis,” which forms part of his “service conception” of authority and reads as follows: “All authoritative directives should be based on reasons which already independently apply to the subjects of the directives and are relevant to their action in the circumstances covered by the directive” (1986, p. 47, emphasis added). Somewhat ironically then, the service conception qua account of legitimate command comes closer to capturing the normative structure of request than does Raz’s discussion of request in the passage quoted at the beginning of Section 3 above. ¹⁰ By asking twice, is Rick creating two numerically distinct reasons for Erica to help him out? It at least isn’t obvious by what resources the mere addition account might avert this— obviously absurd—conclusion.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

41

Or, to make the same point from a different angle, consider a final variation on the Rick-and-Erica situation: Babysitter Everything is as in Movie Night, except that Erica turns down Rick’s request not because she would rather be watching her movie but because she is babysitting her infant niece at her sister’s tonight. Here the appropriate reaction on Rick’s part would seem to be something like: “Oh, I didn’t know. Never mind then.” What this response indicates is that Rick is retracting his request in light of what he now knows and that, accordingly, he wouldn’t have asked in the first place had he known about her babysitting duties before. Unlike in Easy Fix, Erica isn’t going to hold his request against him here, but only because he is excused by his excusable ignorance. That Rick is retracting his request reflects his recognition that Erica doesn’t in fact have good reason to comply with his request. The normative presupposition on which his request was meant to be based thus turns out to fail, from which it in turn follows that by Rick’s own lights the request itself is null and void normatively speaking. Suppose that, instead of retracting his request, Rick were to say to her: “Oh, I knew about your babysitting duties. Still, would you mind coming over?” Once Erica were to come to realize that he wasn’t somehow confused or speaking in jest, she would presumably be more than a little irritated with him and might in fact find the exchange difficult to look past for the purpose of carrying on her friendship with him, since only someone brazen or— to use an expression of Anscombe’s to which I will return below—a “megalomaniac” would be saying this sort of thing with a straight face. Yet all of this is difficult to make sense of on the mere addition account. As the mere addition account has it, Rick may well admit that Erica’s babysitting obligations outweigh his request here, but it isn’t clear why retraction would be in order. A defender of the mere addition account might be tempted to retrieve the proposal from the preceding section and argue that the requester should withdraw her request on account of her duty only to make requests that would figure in the winning coalition of reasons applying to Erica. However, once again the proposal provides the wrong kind of ground for Rick’s retraction. It explains the appropriateness of Rick’s retracting his request in terms of some merely external principle and so cannot account for the request’s collapsing or falling through. In fact, it already tells against the mere addition account that we have a hard time

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

42

 ̈

envisioning how Rick might even articulate the normative status the theory ascribes to his request with Erica’s reply in place. Rick would have to say something like: ‘Understood. No worries. The request still stands though.” But this would be a disorienting thing for him to say here. It is the sort of formulation properly used in the face of an invitation turned down, not as a reply to a request declined.¹¹

7. The Reasonable Harmony Account I will now turn to my own proposal. My point of departure is a remark of Kant’s in the Doctrine of Virtue. Kant there says that “All moral relations of rational beings, which involve a principle of the harmony of the will of one with that of another, can be reduced to love and respect . . . In the case of love the basis for determining one’s will can be reduced to another’s end, and in the case of respect, to another’s right” (6:488 (1999, p. 600), emphases in original). What I want to suggest is that the normativity of request needs to be understood in terms of the idea of a “harmony of ends” or—adding language from Rawls—a “reasonable harmony of ends” between requester and requestee. The idea of a reasonable harmony of ends structures the normative positions of both requester and requestee in a request–response transaction. More specifically, it is an internal standard governing requester and requestee qua requester and requestee that they engage in the joint attempt to realize a reasonable harmony of ends between them. I will call this proposal the reasonable harmony account of request. Let me apply the proposal to the first two examples from above, Movie Night and Easy Fix (leaving the third example, Babysitter, for the next section). Erica’s refusal to come over and help Rick in Movie Night abuses the form of address of request (here as respondent) because in it she grants her own comfort unreasonable precedence over his need for help and so flouts the “regulative idea” of a reasonable harmony of ends. The same applies the other way around in Easy Fix. In asking Erica to sacrifice her evening in order to save himself a minor amount of trouble, he is assigning himself unreasonable priority over her. In his deployment of the form of address of request he is implicitly portraying their relationship as one in

¹¹ Invitation thus seems to amount to an even “feebler” form of address than request.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

43

which her time counts for nothing compared to his. This is what her resentment against him picks up on. What counts as a reasonable harmony of ends is itself a relational matter. It isn’t merely a question of balancing the parties’ interests and concerns understood independently of the relationship in which they stand. Rather, which interests and concerns are the proper subject of the relevant balancing is a question that can only be answered by reference to the relation between requester and requestee. To see the point, return once more to Movie Night. The reasonableness of Rick’s request that Erica help her with her laptop isn’t (or at least isn’t merely) premised on the independently greater weight attaching to important deadlines vis-à-vis movie nights. Rather, it is only because Erica and Rick are friends—that is, are up in each other’s lives in a particular fashion—that the question of comparing their needs and concerns for the purpose of determining what Erica has good reason to do arises in the first place. In other words, the relational is prior to the comparative. Rick’s request that Erica help her with her laptop thus presupposes for its “normative felicity” his enjoying a particular standing with her, that of being her friend. Where the requisite standing is absent, the act of request “misfires,” to use one more bit of Austinian terminology. Were Rick to make his request of Erica outside of a standing-generating context of this sort, the latter would be entitled to reject Rick’s request, not (or again at least not merely) in virtue of its comparative excessiveness, but because of its mistaken presupposition that there even is a question of comparison here in the first place. In a case like that, were Rick to respond to Erica’s refusal to help him with “But don’t you agree that, objectively speaking, my deadline matters more than your night of binge-watching?”, the proper answer on Erica’s part wouldn’t be, “You’re wrong” but, “You’re missing the point.” Of course not all requests depend for their felicity on the existence of a special relationship between requester and requestee. Suppose that Rick and Erica aren’t friends but strangers standing in the security line at the airport, and suppose that Rick asks Erica for permission to cut in front of her because he would miss his plane otherwise. The situation here is normatively analogous to that in Movie Night: were she to turn him down, resentment on his part would be appropriate. Alternatively, suppose that he asks her for permission to cut in front of her because he simply likes sitting in the airline lounge better than standing in line at security. Here the situation corresponds to that in Easy Fix normatively speaking: the very request warrants resentment, on account of either its insincerity or its brazenness.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

44

 ̈

Hence, while the substance of the relationship at play here is different from friendship, its normative structure is the same.¹² We rarely notice these sorts of complications because we generally have a keen sense of when it is appropriate to ask what of whom. Moreover, where we are less confident, we have instruments at our disposal for expressing our uncertainty, such as qualifying our requests with “ . . . but feel free to say no if it’s too much to ask” or the like. “Do you think that you could do suchand-such?” is often at least in part a request for the sort of information that would fill in what a reasonable harmony of ends looks like between the parties. In fact, much of the seeming “easygoingness” of request that made the mere addition account seem appealing in the first place in fact springs from uncertainty of this sort, rather than from the supposed pro tanto character of request on which the mere addition account turns. Oversimplifying things slightly, we might say that requests constitute not pro tanto reasons but prima facie obligations.

8. The Four Constraints Revisited Exploiting the wider context of Kant’s idea of a harmony of ends helps explain how the reasonable harmony account meets both the discretion constraint and the compliance constraint. Kant’s Metaphysics of Morals revolves around the distinction between two kinds of duties, duties of virtue and duties of right. The distinction has two interrelated aspects. First, duties of virtue are “wide” whereas duties of right are “narrow.” The former dictate the adoption of ends, which may possibly be achieved by more than one course of action, whereas the latter require specific actions. Second, duties of virtue are by their very nature unenforceable. The adoption of ends cannot be compelled, not because we lack the technology or the like, but because the adoption of an end by its very nature has to be done freely. The concept of right on the other hand is analytically “connected with an authorization to coerce” (6:231 (1999, p. 388)) according to Kant. This means that duties of right are at least in principle susceptible to enforcement, even if not every particular duty of right is enforceable in a straightforward fashion. My suggestion is to conceive of request as purporting to impose duties of virtue and command as purporting to impose duties of right and to

¹² I owe the example to Arthur Ripstein.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

45

interpret the discretion constraint in terms of the difference in in-principle enforceability between them. A commander in issuing her command presupposes that it isn’t up to the addressee whether to perform the act commanded. Rather, there is a sense in which a commander in her act of command purports to thereby have determined the commandee’s choice. By contrast, in issuing a request a requester leaves it up to the requestee whether to do perform the act requested. This is the—literal—sense in which a requester grants the requestee discretion: in her request she defers to the requestee’s own choice. It is also what the disclaimer “You don’t have to do it” is meant to express. It isn’t meant to suggest that the requestee wouldn’t be doing anything wrong—in the sense of violating a duty of virtue—were she to refuse compliance, which is how the mere addition account had to interpret the discretion constraint. Understanding the distinction between request and command in terms of the idea of enforceability makes it easy to see why people put so much stock in whether they are being asked or told to do something. What is at stake is not whether noncompliance would be morally wrong or the like but rather what sort of relationship the addressee shares with the addressor, and in particular whether she is and is mutually acknowledged to be, “her own master,” to use an expression of Kant’s. The idea that requests impose duties of virtue explains the sense in which compliance with a good-faithed request is obligatory, and so how the reasonable harmony account captures the compliance constraint. The internal nature of the wrong of failing to act from the end of realizing a reasonable harmony of ends in turn derives from the latter’s status as an internal standard of request. To participate in a request–response exchange while failing to adopt the end of thereby jointly bringing about a reasonable harmony of ends is not merely to do something that is wrong by some standard that is external to the making of and responding to requests. Instead, it is to engage in the exchange of request-giving and -receiving in a way that abuses the very form of address it instantiates. Note that the internal relation between virtue and the adoption of ends also provides for a more complete picture of what it is for a requestee to comply with a request. Consider Erica’s negative response to Rick on account of her babysitting duties in Babysitter. While she isn’t complying with his request in the narrow sense of performing the act he had asked her to perform, she is complying with his request in the wide sense of adopting the end of bringing about a reasonable harmony of ends; or, more cautiously, nothing in what she says gives him ground for thinking

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

46

 ̈

otherwise. Note here how natural it would be for her to respond with a counter-proposal along the lines of: “I can’t tonight, but how about I come over first thing in the morning?” She would thereby be expressing her bonafide cooperation in constructing a reasonable harmony of ends between them. Even if tomorrow will be too late for Rick, it would be odd to describe Erica as having refused Rick’s request here, not because there isn’t a sense in which we might say she did, but because the term suggests the sort of rupture in their relationship that precisely didn’t occur thanks to the good will she showed in the way in which she engaged with his request.¹³ Let me turn to the question of how the reasonable harmony account satisfies both the normative effect constraint and the good reason constraint. The context surrounding Anscombe’s talk of “megalomania” to which I had referred in Section 6 is instructive here. She uses the expression in her “What Is It to Believe Someone?”, where she says that “It would be a megalomaniac who complained of not being believed, when he agrees that the thing that was not believed was, anyway, not true” (1979, p. 150). Anscombe’s argument brings out the structural analogy between testimony and request. Just as a request at least in principle constitutes a good reason for doing the thing requested, testimony at least in principle constitutes a good reason for believing the fact testified to. And yet a person giving testimony has to conceive of herself as reporting on something that holds independently of her act of testimony. If the testifier herself believes that the facts don’t bear out her testimony, then she is guilty of either insincerity or megalomania, and specifically of the megalomania of conceiving of herself as the epistemic counterpart to our practical King Midas from Section 6 above. My proposal is that the same thing goes for request. The analogy to testimony provides a model not only for the transparency of request but also for the nature of its normative effect. Testimony purports to provide its addressees with a reason for believing the thing to which it testifies. However, it isn’t meant to serve as a piece of evidence alongside whatever other evidence there is concerning the proposition at hand. Rather, it is supposed to provide its addressee with a new form of rationally believing the proposition. With the act of testimony in place it becomes possible for her to believe the testifier that p, rather than believing that p “merely

¹³ This doesn’t mean that it is always possible to display good will of the requisite sort without actually performing the act requested. Sometimes the only way to comply in the wide sense is to comply narrowly.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

47

monadically.” In other words, the normative effect of testimony consists in making available trust as a rational mode of believing something.¹⁴ Similarly, acts of request don’t constitute reasons for action alongside all the other reasons applying to the requestee, as the mere addition account has it. Rather, through her request the requester makes it possible for the requestee to perform the act requested on the requester’s request, and so to act not merely “monadically” but “relationally.” The normative effect of request is therefore not material but formal. Rather than bringing a new normative “object” into existence, it provides the requestee with the possibility of a distinct mode of acting rationally.¹⁵ Statements like “So-and-so asked me to do such-and-such, so I did it” might therefore be understood as expressions of a form of practical trust. Just as in believing the testifier the recipient of testimony exhibits trust in the testifier that there is indeed (independent) good reason to believe the thing to which the testifier testified, in acting on the requester’s request the requestee shows trust in the requester that there is indeed (independent) good reason for her to do the thing requested.¹⁶ It is the implicit extension of trust that explains why so-and-so finds a statement of this sort so gratifying,

¹⁴ My understanding of testimony is indebted to Richard Moran’s writings on the subject— see his “Problems of Sincerity” (2005a), “Getting Told and Being Believed” (2005b), and “Testimony, Illocution and the Second Person” (2013)—Berislav Marušić’s work on evidence and agency—see more or less everything he has written, but especially Evidence and Agency (2015)—and A. J. Julius’s “Mutual Recognition” (2016). ¹⁵ I am not claiming that a request could never itself act as a substantive reason for doing the thing requested. For instance, perhaps we could imagine a case where the requester derives pleasure from the fact of the requestee’s acting as requested as such, independently of what it is that is being requested. Note that even here it seems to be the requester’s pleasure that is reasongiving for the requestee, for which the request merely serves as the occasion, so I’m not sure whether even this example serves as a proper case in point. Even if it does, however, we could only understand it as parasitic on the primary case, in which request is normatively transparent in the sense described above. Yet the mere addition account would have to view this sort of example not as parasitic but instead as a pure case of request. ¹⁶ This is the place to acknowledge Geoffrey Cupit’s subtle and insightful discussion in “How Requests (and Promises) Create Obligations” more properly. Very briefly, Cupit argues that requests create obligations because in failing to comply with a request the requestee “makes a fool of ” the requester, who in making her request had invested trust in the requestee’s being the sort of person who is capable of acting on another person’s request, and so in her being a person capable of a good will. It should be clear that its general direction is extremely congenial to my own account of request, especially in its strategy of analyzing what is normatively presupposed in request qua form of address, as well as in its appeal to the notion of trust. Interestingly, Cupit focuses on the trust the requester invests in the requestee, whereas in my case it is the other way around. However, I take these two appeals to the notion of trust to be mutually consistent and indeed complementary.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

48

 ̈

not (or in any case not merely) because of the material benefit to her that will ordinarily attend it.¹⁷, ¹⁸

References Anscombe, G. E. M. (1979). “What Is It to Believe Someone?” In: Rationality and Religious Belief. Ed. by C. F. Delaney. University of Notre Dame Press, pp. 141–51. Anscombe, G. E. M. (1981). “On Promising and its Justice, and Whether it Need be Respected in Foro Interno.” In: Ethics, Religion, and Politics: Collected Philosophical Papers Vol. III. Blackwell, pp. 10–21. Austin, J. L. (1975). How to Do Things with Words. Ed. by J. O. Urmson and Marina Sbisà. Harvard University Press. Cupit, Geoffrey (1994). “How Requests (And Promises) Create Obligations.” In: The Philosophical Quarterly, vol. 44, no. 177, pp. 439–55. Darwall, Stephen (2006). The Second-Person Standpoint: Morality, Respect, and Accountability. Harvard University Press. Darwall, Stephen (2013). “Moral Obligation: Form and Substance.” In: Morality, Authority, and Law: Essays in Second-Personal Ethics I. Oxford University Press, pp. 40–51. Enoch, David (2011). “Giving Practical Reasons.” In: Philosophers’ Imprint, vol. 11, no. 4, pp. 1–22. Enoch, David (2014). “Authority and Reason-Giving.” In: Philosophy and Phenomenological Research, vol. 89, no. 2, pp. 296–332.

¹⁷ While I argue for a far-reaching structural analogy between request and testimony here, I do not mean to deny that the parallel between them breaks down eventually, but I lack the space to explore the issue here. For related discussion see Darwall, The Second-Person Standpoint (2006), pp. 56–7. ¹⁸ I had the opportunity to present earlier versions of this chapter at the Konstanz–Zürich Kolloquium, the University of Fribourg, the Early Career Conference in Moral and Political Philosophy at the Humboldt University of Berlin, and the Arizona Workshop in Normative Ethics. I am grateful to my audiences at these venues for their constructive criticisms. I have benefited from conversations about these materials with (at least) the following people: Ralf Bader, Jacob Barrett, Sophie Cote, Kyla Ebels-Duggan, Angelika and Eberhard Gläser, Jasmin Johnson Gläser, A. J. Julius, Felix Koch, James Lewis, David Owens, Japa Pallikkathayil, Al Prescott-Couch, Sebastian Rödl, Peter Schaber, Bettina Schöne-Seifert, Mark Schroeder, Houston Smit, Holly Smith, Felix Timmermann, Mark Timmons, Teemu Toppinen, Daniel Viehoff, Juri Viehoff, and Jakub Wrzesniewski. I thank Dan Khokhar, Elodie Malbois, Masakazu Ogami, and Arthur Ripstein, as well as two anonymous referees for Oxford University Press, for their extremely helpful written feedback. The chapter developed out of a portion of my dissertation, so I owe a particular debt of gratitude to Selim Berker, Tim Scanlon, and especially Christine Korsgaard.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

49

Foot, Philippa (2001). Natural Goodness. Oxford University Press. Julius, A. J. (2013). Reconstruction. Unpublished manuscript. Julius, A. J. (2016). “Mutual Recognition.” In: Jurisprudence, vol. 7, no. 2, pp. 193–209. Kant, Immanuel (1999). Practical Philosophy. Ed. and trans. by Mary J. Gregor. Cambridge University Press. Lance, Mark and Rebecca Kukla (2013). “Leave the Gun: Take the Cannoli! The Pragmatic Topography of Second-Person Calls.” In: Ethics, vol. 123, no. 3, pp. 456–78. Lewis, James H. P. (2018). “The Discretionary Normativity of Requests.” In: Philosophers’ Imprint, vol. 18, no. 20, pp. 1–16. Little, Margaret Olivia (2013). “In Defence of Non-Deontic Reasons.” In: Thinking about Reasons: Themes from the Philosophy of Jonathan Dancy. Ed. by David Bakhurst, Brad Hooker, and Margaret Olivia Little. Oxford University Press, pp. 112–36. Marušić, Berislav (2015). Evidence and Agency: Norms of Belief for Promising and Resolving. Oxford University Press. Moran, Richard (2001). Authority and Estrangement: An Essay on Self-Knowledge. Princeton University Press. Moran, Richard (2005a). “Getting Told and Being Believed.” In: Philosophers’ Imprint, vol. 5, no. 5, pp. 1–29. Moran, Richard (2005b). “Problems of Sincerity.” In: Proceedings of the Aristotelian Society, vol. 105, no. 1, pp. 325–45. Moran, Richard (2013). “Testimony, Illocution and the Second Person.” In: Aristotelian Society Supplementary Volume, vol. 87, no. 1, pp. 115–35. Owens, David (2012). Shaping the Normative Landscape. Oxford University Press. Raz, Joseph (1986). The Morality of Freedom. Clarendon Press. Raz, Joseph (1999). Practical Reason and Norms. Rev. edn. Oxford University Press. Raz, Joseph (2009). The Authority of Law: Essays on Law and Morality. 2nd edn. Oxford University Press. Scanlon, T. M. (1998). What We Owe to Each Other. Harvard University Press. Strawson, P. F. (2008). “Freedom and Resentment.” In: Freedom and Resentment and Other Essays. Routledge, pp. 1–28.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

3 Discretionary Moral Duties R. Jay Wallace

My topic in this chapter will be the idea that there are discretionary moral duties. By this I mean moral obligations that cede to the agent who stands under them wide latitude in determining the actions that will count as satisfying them. Moral considerations of this kind are puzzling along several different dimensions, and they have only a marginal place within many influential ways of conceptualizing the domain of interpersonal morality, such as consequentialism.¹ But the idea that some moral duties are discretionary in this way is part of moral common sense, and also something that seems to me to be plausible and appealing. So I want to consider some of the theoretical considerations that seem to put pressure on this idea, and see how they might be parried. A particular crux will be the question of whether an attractive conception of moral obligation can coherently be combined with the idea that the obligations it defines cede to the agent wide discretion as to how they are to be satisfied. My topic overlaps, to some extent, with the territory traditionally covered by the doctrine of imperfect moral duties. But I shall not frame my discussion in terms of the distinction between perfect and imperfect duties, if only because I find that distinction elusive and confusing.² Instead, I want to single out one of the features that have often been attributed to imperfect duties, that of agential discretion, and to consider how it might be made sense of on its own, abstracting from questions raised by different accounts ¹ Classical consequentialism allows discretion only in cases of ties, in which more than one prospective action would be equally optimific. The class of such cases could be expanded somewhat by emphasizing, for instance, incommensurability in the values to be maximized and imprecision in the agent’s credences about likely outcomes. But the agential discretion that results from these moves will still be more constrained than the wide discretion we take ourselves to have in many intuitive cases of moral obligation, which includes discretion to act in ways that are suboptimal. ² For a survey of different ways of understanding the idea of an imperfect duty, see, for example, George Rainbolt, “Perfect and Imperfect Obligations,” Philosophical Studies 98 (2000), pp. 233–56. R. Jay Wallace, Discretionary Moral Duties In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0003

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

51

of imperfect duties in the history of modern ethics. If we can succeed in making sense of moral obligations that grant to the agent wide discretion as to how they are to be satisfied, we will thereby have rendered intelligible one feature that is often attributed to imperfect moral duties. But this limited result would not, in itself, amount to a vindication of the doctrine of imperfect moral duties, nor do I aspire in the remarks that follow to offer such a vindication. Agential discretion, in the sense relevant to my discussion, needs to be understood in qualitative and not merely quantitative terms. It is a matter of having available a range of alternatives for choice that differ from one another in evaluatively significant ways.³ So construed, agential discretion is present to some degree with a great many of the duties that morality intuitively assigns to us as individuals. Our negative duty not to harm others as we make our way through the world can be satisfied in any number of ways—e.g., by engaging in activities that risk harm conscientiously and in compliance with all pertinent standards of due care, or alternatively by avoiding such activities in the first place—and it is a matter of indifference to morality which of these paths we decide to follow. Similar remarks apply to many positive obligations. If I have promised to meet you at the café for lunch at 1 p.m., it is generally up to me to figure out how I am to live up to the affirmative commitment I have undertaken; I might set out for the café on foot, or ride my bike, or combine our lunch with a visit to a nearby museum in the morning.⁴ Agential discretion, like other features traditionally taken to distinguish imperfect moral obligations, is a matter of degree. What is theoretically puzzling, I take it, is not the idea that morality grants us some discretion as to how to live up to the requirements it lays down on us, but that there are cases in which we seem to have especially wide discretion in this matter. I want to look at two specific classes of moral obligation that intuitively exemplify the kind of wide agential discretion that is my quarry, namely duties of gratitude and of mutual aid. The aim will not be to offer a comprehensive account of these interesting and complex phenomena, but to sketch a general framework for thinking about them, one that helps us to ³ The kind of discretion at issue here involves the ability to choose amongst options that are adequate, in Joseph Raz’s sense, for an autonomous life; see Raz, The Morality of Freedom (Oxford: Clarendon Press, 1986), ch. 14. I return to the theme of autonomy below. ⁴ Compare Michael Stocker, “Acts, Perfect Duties, and Imperfect Duties,” Review of Metaphysics 20 (1967), pp. 507–17; also Daniel Statman, “Who Needs Imperfect Duties?,” American Philosophical Quarterly 33 (1996), pp. 211–14.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

52

.  

make sense of their salient features. The framework I favor incorporates an essentially relational conception of moral obligation, which I lay out in the first section of the paper; the second and third sections then apply that conception to the special cases of gratitude and mutual aid. It will emerge that wide agential discretion makes sense against the background of an understanding of morality as a set of directed obligations that we owe to each other, and that make it possible for individuals to relate to each other on a basis of autonomy and equality.

1. Moral Obligation: A Relational Approach Reflection about the idea of discretionary moral obligations needs to be situated within the context of a broader understanding of the nature and sources of moral obligation in general. But what are the desiderata that we should expect a notion of moral obligation to satisfy?⁵ I believe that two considerations are especially significant. First, from the perspective of deliberation, obligations function in a way that sets them apart from other kinds of normative consideration. They are not merely reasons that “count in favor” of options that are open to the agent, of the kind that might be weighed in the balance in deciding what to do. Rather, they define constraints on action that are presumptively decisive in their force. This shows itself in the way obligations interact with other reasons, functioning as what Joseph Raz has called exclusionary reasons; normative considerations of this kind do not outweigh reasons for action on the other side, but exclude them from deliberative consideration from the start.⁶ In these and other ways, obligations enter the deliberative field of agents in a highly distinctive guise, representing defeasible constraints on action, rather than considerations that feed into a further process of deliberative calculation. The second important dimension of obligations is their role as a basis for interpersonal accountability. Moral obligations are suited to be internalized by us in our relations with other agents, in ways that connect to our practices of moral responsibility. When people act with disregard for their moral obligations, their attitudes have special significance for other parties,

⁵ I draw, here and in what follows, on material from my book The Moral Nexus (Princeton: Princeton University Press, 2019), chs 2, 3, and 6. ⁶ See Joseph Raz, Practical Reason and Norms, second edition (Princeton: Princeton University Press, 1990).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

53

giving them reasons to respond with blame and the behaviors that give expression to it. We accordingly hold each other to account for actions that flout moral obligations, in a way that contrasts with other normative standards that individuals might or might not fall short of. There is a broadly relational approach to morality that can help us to make sense of these twin dimensions of moral obligation.⁷ This approach construes moral obligations as a special class of relational duties, where duties of this kind exhibit the following features. First, they are directed obligations, which are owed to another party in particular. To be subject to an obligation of this kind is to stand in a normative relation to another person, whereby compliance with the obligation is owed to that person. Second, the individual to whom a relational obligation is directed may be understood to have a claim to the agent’s compliance with it, one that is held against the agent by the person to whom the obligation is owed; indeed, this claim could be understood as the other side of the normative relation between agent and the person to whom the agent’s obligation is specifically owed. Third, there is the idea that agents who disregard a relational obligation have not merely done something that is wrong or normatively defective in individual terms, but changed their relationship to the individuals who hold a claim against them to performance. They will have wronged those individuals, flouting or disregarding their claims, and thereby giving the aggrieved parties a privileged basis for objecting to what they have done. These elements of relational obligation are familiar parts of the broader normative domain. They are present, for instance, in the structure of Hohfeld-style claim rights, such as those implicit in the obligations brought into existence through promises and contracts. In providing another party with promissory assurance, I will ordinarily be understood to have done something that creates a new normative relation between me and the other party. I will have undertaken a new obligation, but one that is owed to the promisee in particular, rather than to the community of moral agents at large, and that corresponds to a claim held against me by said promisee. Furthermore, if I flout the promissory commitment, I will not merely have acted in a way that is wrong or objectionable, but will have wronged the recipient of my promise, in ways I will not have wronged others, such as third parties who might stand to benefit incidentally from my promissory fidelity on this occasion.

⁷ The approach is developed and defended in my The Moral Nexus.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

54

.  

The relational model that I find attractive interprets the larger domain of interpersonal morality as a set of relational obligations of this kind. According to this way of understanding morality, it is a matter of what we owe to each other, not insofar as we stand in special relationships to each other, but just insofar as we are each persons whose interests matter equally. To understand morality in these terms is to extend the relational paradigm much more widely than even the proponents of Hohfeld-style claim rights are typically willing to do. It is, for instance, to understand there to be directed moral obligations in cases in which we would not ordinarily speak of correlative moral rights at all, including cases of discretionary moral obligation that are my ultimate quarry in this paper. Before turning to the implications of the relational account for these cases, however, I want to highlight its theoretical advantages as a general paradigm for understanding the requirements of interpersonal morality. The basic idea here is that the relational approach is uniquely wellpositioned to illuminate the two distinctive normative roles that I have attributed to interpersonal moral duties. In the perspective of individual deliberation, directed obligations are well-understood to be considerations that intelligibly function in the manner of presumptive constraints on what we do. If I have entered into a promise or contract with you, then I owe it to you to comply with its terms; the directed obligation that links us to each other is our common possession, so to speak, and it isn’t open to me to discount it unilaterally in my reflection about what I am to do. Debts that are owed to another party are naturally understood to resolve one of the questions we face about what we are to do, representing what is often referred to as the original notion of an obligation. For these and other reasons, directed duties are rightly registered within deliberation as presumptive constraints on our agency, rather than as considerations that are to be set in a balance of pro and contra reasons. The relational model also offers an extremely promising account of the distinctive interpersonal role of moral obligations. As I noted above, these have normative significance for other parties, of a kind that sets them apart from other kinds of normative consideration. Moral obligations are thus suited to structure relations of interpersonal accountability, and their disregard by the agent provides other parties with warrant for such familiar blaming reactions as resentment. But directed obligations are considerations that are tailor made for this interpersonal role. Insofar as they are owed to another party, disregard of them by the agent amounts to disregard of someone’s claims, which is something that is of obvious significance for

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

55

the claimholder. It thus makes sense that those who hold claims against others would assert them interpersonally, adopting the expectation that other agents would acknowledge and honor their claims, and reacting with a characteristic form of grievance when this expectation is flouted. Indeed, the paradigmatic form of warrant for the reactive attitude of resentment is widely understood to be the fact that one has been wronged by another person, something that presupposes the directed model of obligation that is central to the relational interpretation of it. For these reasons, the relational model strikes me as the interpretation of morality that makes best sense of it, as a domain of interpersonally significant obligations. In the space that remains to me, I wish to consider the idea of agential discretion through the lens of this interpretation of the moral domain.

2. Obligations of Gratitude Suppose that another person does you a kindness, at significant personal cost; perhaps they volunteer to read your new book manuscript conscientiously, offering detailed and extremely helpful comments that seriously set them back with their own summer research project. The bestowal of benefits in a case of this kind is ordinarily understood to change your moral relations to the person who has provided them, generating a duty of gratitude. Furthermore, as I noted earlier, duties of gratitude are among the moral requirements that exhibit wide agential discretion. There isn’t any particular thing that you have to do to discharge the debt of gratitude that you owe in cases of this kind; rather, you have wide latitude to decide for yourself how you will repay the kindness that was done to you. My question is how these twin elements might be combined: the new moral obligation that is brought into existence through the bestowal of benefits, and the wide discretion that the agent has as to how the obligation is satisfied. I have suggested that we think of moral obligations in relational terms. This fits easily with some aspects of our intuitive understanding of the case of gratitude. It is perfectly natural, for instance, to say about cases of this kind that the beneficiary of a significant kindness owes a debt of gratitude to their benefactor, and to speak in this way is to conceptualize the obligation in question as directed to the benefactor in particular, in the way that is characteristic of relational obligations. On the other hand, the gratitude case also lacks at least one feature that is present in some of the paradigmatic

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

56

.  

examples of relational obligation, in that does not correlate with a right on the part of the person to whom the duty is owed. Nobody has a moral right to the gratitude of those who make significant sacrifices to assist them in situations of need or dependency. But it might be wondered whether the relational model clearly applies in cases in which allegedly directed obligations do not correspond to rights to performance on the part of those to whom they are owed. I will eventually want to argue that directed moral obligations correspond not to moral rights, narrowly construed, but to moral claims, and that the relevant notion of a claim is one that has application in the case of duties of gratitude. Before we get there, however, it will help to reflect briefly on the question as to why it seems out of place to speak of rights in cases of this general kind. There are two considerations that may be significant in this connection. The first has to do with alienability. Many moral rights can be waived by the person to whom they are assigned, through the right-holder’s consent to the actions that would otherwise count as infringements. I can rescind my right against you that you not engage in acts of trespass by inviting you to collect pine cones in my wooded back yard, or by consenting to the surgical procedure whereby you would repair my injured foot. But benefactors cannot in the same way waive the obligations that beneficiaries are under to reciprocate with gratitude for the kindness that was done to them.⁸ Of course, some moral rights are popularly understood to be inalienable, such that they cannot be rescinded through voluntary acts on the part of the individual who is their bearer. But inalienable rights, if indeed there are such, are plausibly grounded in personal interests on the part of the right-holder that are especially significant, perhaps on account of their connection to the right-holder’s moral standing. But the interest that benefactors have in displays of gratitude on the part of those whom they have helped does not appear to be a weighty moral interest of this kind. The second feature of standard moral rights that appears to be missing in the cases of gratitude is just the element of wide agential discretion. Rights are ordinarily understood to correlate with fairly determinate duties on the ⁸ There are certainly contexts in which it makes sense for benefactors to say something like “Think nothing of it!” to the persons they have helped, and to discourage them from doing anything to reciprocate. Protestations of this kind are most at home in situations in which the benefactor’s act of assistance did not really involve a significant sacrifice, so that it is questionable that a genuine duty of gratitude even obtains. Outside of such contexts, however, interventions of this kind can easily seem patronizing, insofar as they denigrate the agency of the beneficiary. This underscores the point that there is a genuine duty in place here, and that it is not the place of the benefactor to determine whether it obtains or how it is to be satisfied.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

57

part of the agent against whom they are held, where there is a specific category and occasion of performance that the agent must engage in to count as living up to the obligation that they are under. Thus rights to property or bodily integrity go together with duties not to engage in unauthorized trespass, and promissory rights involve claims against promisors that they will keep their word. But as I have noted, there is nothing so specific in play in the gratitude case; it isn’t as if the benefactor can insist, say, that the person benefited should arrange for them to attend an NBA playoff game, even if a gesture of that kind would count as satisfying the beneficiary’s duty of gratitude.⁹ Of course, the discretion that seems to be present in these cases, though wide, is not unlimited. Though it isn’t for the benefactor to say exactly how and when the debt of gratitude should be repaid, it would be a fault on the part of the beneficiary to disregard completely the benefactor’s wishes regarding the forms of assistance they might receive. We are all familiar with cases in which displays of gratitude become burdensome or intrusive for the individual at whom they are targeted, and these failings can often be chalked up to insufficient attentiveness on the part of the agent to the genuine interests of the person to whom gratitude is owed. In a different kind of case, the nature of the kindness that was originally bestowed by the benefactor, or the relationship between the benefactor and the beneficiary, might set limits on the kinds of performances through which gratitude is properly repaid. Consider the earlier example in which a person sacrifices valuable research time to provide detailed feedback to a friend or colleague on their manuscript. In the context of this kind of exchange, there is some presumption that gratitude should be shown through acts that reciprocate the kind of benefit that was originally bestowed. The beneficiary should, if possible, offer a comparable level of assistance to the benefactor with their own scholarly projects, rather than insisting on helping the benefactor in some other way (e.g., by building a new shed for their garden). This is true, it seems to me, even if the beneficiary is much more skilled at carpentry than at

⁹ As I observed in Section 1, even in cases of rights to positive performance, there is some discretion that agents have to determine specifically how they are going to fulfill their directed duties. Having promised to meet you at the airport, I have leeway to decide for myself which bridge to take to get there, how long in advance I will arrive, etc. The general idea is that moral claim rights will seem firmly in place in proportion to the degree to which the directed duties connected with them are more determinate. The duties at issue in cases of gratitude, it seems, are at the less determinate end of the spectrum.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

58

.  

commentary on scholarly manuscripts, so that they assistance they might provide the benefactor with the garden project would ultimately be more valuable to the benefactor than the feedback they are able to offer on the benefactor’s academic writing. These limits to agential discretion, it seems to me, illustrate the importance of equilibrium between the parties in cases in which debts of gratitude are owed, a theme to which I shall return momentarily. But how can there be a directed duty of gratitude if there isn’t a strict moral right to which it corresponds? I believe that directed duties correspond to normative entitlements of some kind; but as I mentioned earlier these can be understood as moral claims rather than rights, and it seems to me that the general notion of a claim has application even in the case of gratitude. The moral claim that is place in situations of this kind is not a claim to some specific action or range of actions on the part of the beneficiary of the kind that would alone count as discharging the debt. It is, rather, a claim that the beneficiary should exercise their discretion—within the limits that obtain in any given case—to repay the debt that was incurred when the kindness was originally bestowed. This is a claim that is naturally understood to be held against the beneficiary, and it represents the normative residue of the directionality that we make explicit when we say that the debt of gratitude is owed by the beneficiary to the benefactor. Furthermore, this structure of directed duties and claims helps us to understand why the moral considerations at work in these discretionary cases represent obligations. A duty that is owed to the benefactor is something that the beneficiary will rightly treat as a presumptive constraint on their deliberations, rather than as the kind of reason that is to be weighed in the balance within moral reflection. Acknowledging that the duty obtains, the beneficiary will reasonably form the intention to repay the debt that they owe as they move forward with their life. True, the element of agential discretion entails that this intention will perhaps be underspecified at the point in time when it is originally formed. But in this, it is no different from many other kinds of future-directed intentions that people reasonably arrive it—such as the intention to go to Sicily this summer, before one has figured out how one will get there and where one is going to stay once one arrives. To have such an intention is to acknowledge a constraint that one takes to be defeasibly fixed as one proceeds to fill in one’s plans; whatever else one does, one will have to work out how one is going to travel to Sicily in the summer, just as one will have to figure out a way of repaying the debt of gratitude one has incurred.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

59

Discretionary considerations of this kind also exhibit the interpersonal significance that is characteristic of moral obligations. The benefactor, in holding a claim against the beneficiary that gratitude be shown, is in a position to be wronged by the latter if they turn out to be ungrateful. The benefactor cannot reasonably expect that gratitude be shown in some particular manner, given the agential discretion that we have seen to be characteristic of this case; but they can certainly expect that the beneficiary will acknowledge the kindness that was shown them, and do something to repay the debt that was thereby incurred. A reflection of this is the fact that it would seem perfectly fitting or warranted for a benefactor to resent a beneficiary who is ungrateful about a favor that was done for them at significant personal cost to the benefactor. The colloquial language of a debt of gratitude suggests that there is an imbalance of some kind between the parties that is restored to equilibrium through the action that repays the debt. This is certainly apt, to some extent, in the case of gratitude, though the model of figurative debt can in other respects mislead. To start with the ways in which the model is helpful, obligations of gratitude are generally created by beneficial actions that go above and beyond what people are anyway expected to do for each other.¹⁰ We don’t owe a debt of gratitude to the friend who takes us to the airport as promised—though we might be grateful that they made the promise in the first place—but we would ordinarily feel very grateful if they dropped everything to get us to the airport when we called them in a panic after our own car refused to start. The original imbalance in our relations is created by the benefactor’s taking on significant personal burdens to provide us with valuable assistance when they are not otherwise required so to act, and if this element is lacking, then the idea that there is a debt of gratitude that has been created would often lack application. Thus, it would often appear insulting or sarcastic to display gratitude for beneficial actions of another that merely meet reasonable interpersonal expectations we hold each other to (like ostentatiously gesturing to thank someone for yielding to us when we anyway have the right of way). An exception to this generalization is perhaps presented by cases in which individuals benefit us through actions that, though costly to them, nevertheless discharge a moral duty that they are under. In the kind of direct

¹⁰ On this element in a Kantian conception of the grounds of gratitude, see Houston Smit and Mark Timmons, “The Moral Significance of Gratitude in Kant’s Ethics,” Southern Journal of Philosophy 49 (2011), pp. 295–320.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

60

.  

rescue situation to which I shall return below, we might well feel very grateful to the person who rescues us from drowning, even if that person was a very good swimmer who was morally required to do what they could to help us in the situation we found ourselves in. Here, though we could reasonably expect the swimmer to assist us, we can also acknowledge that it was their bad luck to be the person on the scene who was called on to do the job. In these and other ways, duties of gratitude presuppose a background set of expectations and moral entitlements, relative to which a debt is understood to be incurred in the first place. Similarly, the actions that discharge the debt of gratitude are standardly ones that serve to benefit the benefactor in some significant way, compensating them, so to speak, for the extraordinary costs that the benefactor undertook. This is the superficial way in which equilibrium is restored in the relationship that has become unbalanced through the original exertions on the part of the benefactor. But it is at this point that the model of a figurative debt can also start to become misleading. If the problem to be resolved through gratitude is merely to compensate the benefactor for extraordinary costs undertaken to assist the beneficiary, then it should be open to the benefactor to simply name a benefit that would reimburse the original costs; it should also be open to the beneficiary to repay the costs in a grudging or reluctant or perfunctory way. But neither of these possibilities really applies to duties of gratitude. The former is at odds with the element of wide agential discretion for which these obligations are exemplary. And the latter neglects the expressive dimension that is equally important to gratitude. The actions that discharge duties of gratitude must be ones that display gratitude, showing that the beneficiary acknowledges the significant burdens that the benefactor originally undertook in helping them in the first place, and that they appreciate that sacrifices were made. A sincere expression of thanks is almost always the initial thing that we owe to those who have gone out of their way to help us with our projects, and the expression of gratitude remains an essential element of any further actions through which our debts in this area are eventually repaid. These aspects of the case call attention to the importance of agency in the relationship between the original benefactor and beneficiary. These individuals are not merely vectors of causal force that operate to disturb the prevailing distribution of benefits and burdens, in ways that occasionally call for correction and calibration. They are persons who are conducting their lives within the framework of the interpersonal obligations that they owe to each other. When benefactors take on a significant burden to assist

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

61

beneficiaries, they are exercising their capacities for agency to advance the beneficiaries’ activities in ways they could not do on their own. This is something that calls not only for compensation, but also, and more importantly, for acknowledgment. A relation between moral equals should be one in which the parties are cognizant of each other as independent agents with lives of their own to lead. Part of the significance of interpersonal morality derives from the fact that honoring its requirements is a way of recognizing the standing of other agents, as individuals with their own projects and relationships whose interests are no less important than ours. But this same recognitional imperative should lead us to acknowledge and to honor the extraordinary burdens that other agents sometimes accept in their own lives in order to assist us with the conduct of ours. Recognition and acknowledgment of such burdens is important to the maintenance of relations between moral equals, and it helps to make sense of the essentially expressive dimension of gratitude. The expressive aspect of gratitude also helps, to some degree, in understanding the agential discretion that is characteristic of it. Where actions have an essentially expressive meaning, it is important that they result from the exercise of some discretion on the part of the agent who performs them. Thus, a present is typically more effectively expressive of attitudes of care and concern and love if it is not prescribed in advance exactly what one is to give, but if there are many options, and if the giver selects the gift with thoughtful attention to the needs and interests of the givee. Its expressive significance is enhanced through the agential exertions that went into its selection and presentation. But there is a distinct element to the agential relations between benefactor and beneficiary that Barbara Herman has called attention to, which sheds some additional light on the discretion displayed by duties of gratitude.¹¹ The original act of generosity, while it confers a benefit of some kind on the beneficiary, also takes something away from the beneficiary, namely their active role in shaping the relationship with the benefactor on common terms. They become dependent on the benefactor, rather than individuals who are able to lead their lives actively on their own terms. What is needed, given this situation, is not merely compensation for the burden that the

¹¹ I am indebted here to the illuminating discussion in Barbara Herman, “Being Helped and Being Grateful: Imperfect Duties, the Ethics of Possession, and the Unity of Morality,” Journal of Philosophy 109 (2012), pp. 391–411.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

62

.  

benefactor originally took on, but a mechanism of repayment that at the same time corrects the disequilibrium of dependency and agency that has emerged between the parties. By reciprocating for the favor that was bestowed, beneficiaries assert agential control as partners to the relationships in which they stand to their benefactors, something that is not really possible so long as gratitude is not displayed or performed. But if gratitude is to function as a mechanism for effecting the transition from dependency to agency, it will be more effective to the extent that beneficiaries have wide latitude to determine for themselves the specific form that their gratitude will assume. It also makes sense that benefactors should lack discretion to waive the duty of gratitude that is created through their acts of kindness, since their doing so would only serve to exacerbate the dependency that the duty of gratitude is meant to help overcome. I noted earlier that there are cases in which agential discretion is significantly limited within contexts that call for displays of gratitude. When you make a personal sacrifice to provide valuable feedback on my manuscript, I might owe it to you not merely to assist you in some way or other with one of your undertakings, but to help out with one of your scholarly projects in particular. This makes sense if the role of gratitude is in part to rectify a disequilibrium of agency and dependency in the relationship to the parties. The rectificatory work to be done depends in part on the nature of the relationship between the parties, and in the case at issue it is a relationship between colleagues or fellow scholars that is at stake. It would do nothing to restore equilibrium within this relationship for the beneficiary to show gratitude by building the benefactor a new garden shed, even if the finished shed would, given the beneficiary’s carpentry skills, represent a greater net benefit to the benefactor. The fact that gratitude is in this way about the beneficiary as well as the benefactor also sheds light on some of its characteristic hazards. There are ways of reasserting agency within a relationship that turn the spotlight onto the person who is acting. As I noted earlier, one can choose to repay a kindness through actions that are ostentatious or obtrusive or otherwise annoying for the intended recipient. One is taking advantage of opportunities available to one in this kind of case, but doing so in a way that opens one to further moral complaints (having to do with the fact that one was thoughtless, or more concerned about one’s own appearance than about the interests of the person one is trying to assist). Displaying gratitude, like giving a gift, is challenging, and it is made so, at least in part, by the very feature of agential discretion that gives it much of its value.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

63

3. Obligations of Mutual Aid It is clear that we have important moral obligations to help those who are in serious distress, even when this is at least moderately costly for us. But there are vast numbers of people who need assistance, and nothing we might do as individuals would alleviate the suffering of more than a small number of them. Under these conditions, we naturally think that agents have significant discretion to decide for themselves how they will meet the requirements of mutual aid. It goes together with this element of wide discretion that we do not think that any of the individuals who are in dire need have a moral right to be helped by us in particular, despite the fact that we have a weighty obligation to provide assistance to people in their position. So once again, we have obligations that are not correlative with any specifically assignable moral rights. But the agential discretion that leads us to deny the assignment of rights in this case is more extensive than in the case of gratitude. With gratitude, there is at least a discernable individual to whom the obligation is owed; this provides an opening for supposing that there is a claim, if not a determinate right, that corresponds to the directed obligation. But with mutual aid we have discretion as to whom to benefit through the actions that comply with our moral obligation, which seems to frustrate the application to it of the relational model. To whom might the obligation be understood to be owed in a case with this structure? Unless this question can be answered persuasively, we will not be able to apply the relational model to the case of duties of mutual aid. But this in turn, I contend, will make it difficult to understand them as genuine moral obligations in the first place. A promising framework for understanding the discretionary duties at issue in these cases is to understand them in relation to structures of collective action. We have institutions that are organized to contribute assistance of various kinds to the many people who suffer extreme deprivation, and our moral responsibilities as individuals in relation to these institutions is, in the first instance, to contribute our fair share to their joint ability to meet their objectives. According to this familiar approach,¹² the first question we need to think about in determining what is required of

¹² The resulting account would have affinities with the approach sketched by Liam B. Murphy in Moral Demands in Nonideal Theory (Oxford: Oxford University Press, 2000). For critical discussion of this approach, see Garrett Cullity, The Moral Demands of Affluence (Oxford: Oxford University Press, 2006), ch. 5.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

64

.  

us is what level of contribution from people in our position would be required to address the basic human needs of the worst off, if everyone were doing their part. Our duty as individuals would then be to make contributions at this level, exercising discretion as to which of the many eligible aid organizations are to benefit from our efforts. As noted earlier, we might direct our fair share contributions to the Against Malaria Foundation, or we could use them instead to support organizations that assist the homeless in our own community or the global refugee population. This way of conceptualizing duties of mutual aid, however, delivers a strategy for identifying the parties to whom the duties are owed. Such claims might be assigned to all of those whose acute needs could potentially be alleviated by the beneficent efforts of the agent.¹³ Such potential beneficiaries have strong personal interests that—to put things in contractualist terms¹⁴— ground powerful objections to principles that allow the affluent to contribute nothing to collective efforts to alleviate the plight of people such as themselves. These objections easily defeat the much milder objections the affluent might have to principles that require them to do their part in these efforts. The result of this way of thinking about things would be that we owe it to all of the persons in the class of potential beneficiaries to contribute our fair share to supporting organizations that effectively work to address their welfare deficits. Each of those individuals, in turn, could be said to have a claim against us that we should make contributions at the relevant level to such collective efforts. The resulting moral claims, it has to be said, have some surprising features. For one thing, they are not claims that can be waived or alienated voluntarily by the individuals to whom they are assigned. This is connected to the fact that each individual claimholder belongs to a class of potential beneficiaries of aid that includes others who are equally in a position of acute need; it therefore makes sense that no one of them should have authority to waive entitlements that in any given case are liable to benefit other members in the class. But there is a further feature of the claims to mutual aid that is even more peculiar. Not only are they not claims to any specific kind of ¹³ For any given agent, we might ask two questions: (a) What is the range of eligible actions open to the agent to provide assistance to others in acute need (i.e., actions that are over some reasonable threshold of expected effectiveness)?; (b) Which specific individuals would benefit from the assistance that is made possible by each of those available actions? It is a familiar point that the resulting class of potential beneficiaries is extremely extensive, since there are so many easy actions open to us that would make available benefits to people who are remote from us. ¹⁴ T. M. Scanlon, What We Owe to Each Other (Cambridge, MA: Harvard University Press, 1998).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

65

performance on the part of the agent against whom they are held, given the wide discretion agents have to determine for themselves how they are to discharge their responsibility to contribute their fair share. They are not even claims that agents should act in ways that benefit the claimholder in particular, since the correlative duties could be discharged fully through discretionary efforts that end up assisting completely different members of the class of potential beneficiaries. These are among the considerations that make us reluctant to assign moral rights to individuals to our assistance in cases of mutual aid. But there is nevertheless a basis for applying the relational model of moral obligation to these cases, and it seems to me illuminating to do so. Those of us who are comparatively privileged arguably owe it to each of the individuals in the class of potential beneficiaries to do our fair share to support collective efforts to address their dire human needs. And each member of that class, in turn, has a claim against us that we should so contribute. We typically conceptualize moral claims as demands, compliance with which would benefit the claimholder in particular, but this feature does not seem to be built into the very meaning of a claim. The relational structure implicit in these cases goes together with the extremely natural idea that affluent agents who fail to contribute their fair share to projects of mutual aid will be unable to justify their conduct to the individual members of the class of potential beneficiaries. Those individuals may accordingly be thought to have been wronged when privileged individuals fail to contribute their fair share to collective efforts to assist people in their condition; this in turn gives them warrant for accountability ascribing reactions such as resentment. A common objection to this general approach to mutual aid is that it fails to track our strong intuitions about face-to-face encounters between agents and others in a position to be assisted by them. Thus, suppose that you are a highly skilled swimmer who regularly goes for walks along a remote beach; and suppose that, through freakish circumstance, you have already braved the surf to rescue five drowning swimmers this week while on your walks, ruining several different sets of clothing. You might, at this point, be thought to have contributed your fair share toward rescuing those who are in distress on this particular section of the beach. But this consideration would not relieve you from your responsibility to rescue a sixth person who is drowning if you happen upon them while out for a walk the next day. It might in some sense be unfair for you if cosmic circumstances in this way distribute disproportionally to you opportunities to rescue swimmers who are at risk of

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

66

.  

drowning, when there many are other excellent swimmers around who would, if present on the scene, be equally capable of helping. But that doesn’t get you off the hook when you are the person who is in a position to save drowning swimmer number six.¹⁵ I agree with this conclusion about the beach walker; but I do not think it casts doubt on the fair share approach to individual duties of mutual aid. What it calls for, instead, is a distinction between those duties and the part of morality that involves direct emergency assistance. Even in the most wellorganized social order, where adequate public provision is made for meeting the needs of all, there will be local situations in which people require the immediate assistance of other individuals. A swimmer gets pulled out by a rip current while bathing on a beach that is not served by life guards; or a pedestrian trips on the sidewalk and is now bleeding and concussed, requiring immediate attention; or someone’s car breaks down on a remote stretch of country road that is out of range of cellular phone service. In situations such as these, we cannot count on institutions to address the urgent needs of the individuals in distress, but are reliant on the kindness of strangers who happen to be on the scene. If those strangers are able to provide the necessary assistance at only modest cost to themselves, then they plausibly owe it to people in need of urgent help to provide it. This is not a duty merely to do one’s fair share, but simply a duty to rescue, and it is therefore in force even if one has the bad fortune to find oneself on the scene of such emergencies much more frequently than most people do. In “Famine, Affluence, and Morality,” Peter Singer famously argued that the differences between situations of direct emergency rescue and of famine relief are without moral significance.¹⁶ His contention is that if we must provide assistance in cases of the first kind, then we have a similar moral obligation to make similar sacrifices to assist people dealing with emergencies in countries remote from our own. I am maintaining, against Singer, that it makes a moral difference that our efforts in the famine case are ¹⁵ Compare Cullity, The Moral Demands of Affluence, pp. 75–7. Though he admits that a fair share view applies more naturally to collective projects than to direct emergencies, he ultimately rejects it even there. But Cullity draws the distinction somewhat differently than I have done. My idea is that we need different principles to explain our duties in regard to ongoing collective efforts to address structural problems of acute human need, and in direct emergency situations in which only a few individuals are even in a position to respond. Cullity, by contrast, distinguishes collective endeavors from individual direct action, and notes that some situations that require a collective response have the character of direct emergencies. On the approach I have sketched, the latter are not situations that call for a fair share solution. ¹⁶ Peter Singer, “Famine, Affluence, and Morality,” Philosophy & Public Affairs 1 (1972), pp. 229–43.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

67

mediated by charitable organizations, which are standing institutions that coordinate our individual efforts, providing ongoing assistance to those who face structural and economic challenges and intermittent crises of various kinds. We have a serious obligation to support these collective, institutional activities, but it is rightly understood to be a duty to do our fair share, not to provide assistance on each occasion when we are a position to do so (until we reach the point of diminishing marginal utility). The relational interpretation of mutual aid I have been developing takes as its benchmark the situation of full compliance. We are to consider what level of contribution by people in our position would be sufficient to address the significant needs of the most vulnerable, if everyone similarly positioned were to contribute at the same level. This defines our fair share; the duty of mutual aid that we owe to potential beneficiaries, then, is to make contributions to collective aid efforts that represent our fair share, exercising discretion as to which of the many such efforts we wish to support. Our moral obligation is to comply with this standard, even if, in the actual world in which we live, most of those in our position are doing nothing. A noteworthy aspect of these situations of partial compliance is that further contributions by us, beyond the level that represents our fair share, would make a real difference to the life prospects of actual individuals who are subject to acute need. The question, however, is this: Do prospective beneficiaries have moral claims against us to such additional contributions, given that so many others are doing nothing (or next to nothing)? There are reasons to doubt this.¹⁷ If there are moral claims in play here, as I am assuming there must be, they are primarily claims against those individuals who are not already doing their fair share in the collective efforts to provide assistance. Those individuals need to step up and begin doing their part. This way of thinking of things is reflected, further, in the accountability reactions that are warranted on the part of potential beneficiaries under circumstances of partial compliance. Members of this group certainly have a grievance in this situation, something that would provide grounds for resentment and other forms of blame. But it would be odd for them to direct these reactions at individuals who are already doing their fair share when so many in a similar position are doing nothing, and when contributions by those

¹⁷ See, again, Murphy, Moral Demands in Nonideal Theory, for a broadly similar view. Murphy emphasizes the fact that our duties of mutual aid give us a common, agent-neutral aim; my discussion differs in stressing the relational character of those duties and the nature and extent of the claims with which they correspond.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

68

.  

non-compliant agents would suffice to address the basic needs of the claimholders.¹⁸ Two further observations about partial compliance are called for. First, though the failure of others to do their fair share does not affect the level of contribution reasonably expected of us, it might have some effect on the forms of assistance that we should offer. There are surely some collective schemes where a failure of sufficiently many to contribute anything would make contributions by the rest of us pointless. Under these conditions, our fair share contributions should be directed to other organizations that will be able to make use of them to advance concrete charitable ends. Second, obligations of direct emergency rescue seem to me to function differently from obligations of structural mutual aid under conditions of partial compliance, which reinforces my contention that these domains make different moral demands on us. I have been arguing that the fact of partial compliance does not affect the level of contribution that we owe to those in a position to be helped by our assistance. But moral failures on the part of others can affect the level of sacrifice imposed on us by our duties of direct rescue. If a military confrontation leads to a mass refugee crises, then those in countries on the front lines of the crisis might find that they are called on to provide at least temporary emergency assistance to the civilians fleeing the conflict. This duty remains in place even if the crisis to which it is a response would not have arisen in the first place if others were living up to their moral responsibilities.¹⁹ The idea that we have wide agential discretion as to whom to assist with our fair share contributions goes together with the idea that we are not under a moral duty to maximize the welfare benefits that accrue from those contributions. If this were the nature of our fair share obligations, then it would generally be the case that there is a single charitable institution or organization that we are required to contribute to (or perhaps a small set of such organizations, if they are comparably effective in translating

¹⁸ Compare Jonathan Cohen, “Who is Starving Whom?,” Theoria 47 (1981), pp. 65–81. ¹⁹ The original rationale I sketched for the individual duty of direct rescue was that there would be situations that require individual response even in a world of full compliance with moral requirements, insofar as collective and institutional efforts cannot anticipate and respond to all emergencies that might arise. Under partial compliance, we have duties to contribute our fair share to collective efforts to alleviate the needs of the worst off, and duties of direct rescue to assist the persons who are not in a position to be helped by those institutions that currently exist (e.g., the refugees who have washed ashore on our island). My point is that our obligations of direct rescue might be much more demanding under conditions of partial compliance; but the basic distinction between the two still obtains.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

69

contributions into utility for those who are globally worse off ). It would follow, further, that the beneficiaries of those maximally effective contributions would be the members of the broader class of potential beneficiaries whom we are actually required to assist. It has lately become fashionable to argue that our obligations of mutual aid have this kind of maximizing character. Proponents of the so-called effective altruism movement think that we ought to discharge our moral obligations to charity by channeling them to the organizations that will do the most good with them in impartial terms.²⁰ (Indeed, some go still further, contending that we often have a conditional obligation to optimize our charitable contributions when we make them, even when it would be permissible for us not to contribute anything.²¹) For obvious reasons, this amounts to the denial of the common-sense idea that we have wide agential discretion in this area, including discretion to direct our charitable efforts in ways that will be sub-optimal relative to the general goal of addressing the acute human needs of those who are worst off. In support of this way of thinking about obligations of beneficence, it is customary to frame the issues in terms of the “burdens” or “costs” on the agent of diminished discretion, as compared to the benefits that might accrue to the individuals who stand to profit from their charitable efforts.²² Taking the occasion of charitable giving on its own, there are burdens on the agent that need to be taken into account if they are forced to perform the action that is maximally effective at addressing basic human needs. (It would perhaps be inconvenient or disadvantageous for them, e.g., not to have the freedom to support the homeless in their community rather than giving to the Against Malaria Foundation.) But these must be set over against the lives that could potentially be saved by directing one’s charitable contributions toward the organizations that are best at converting them into concrete welfare gains. Comparing these benefits and burdens one-by-one, it is very tempting to conclude that there will always be strong moral pressure to make the contribution that is most effective at advancing generally beneficent ends.²³ ²⁰ See, for example, William MacAskill, Doing Good Better (New York: Avery, 2015), and Peter Singer, The Most Good You Can Do: How Effective Altruism is Changing Ideas about Living Ethically (New Haven, CT: Yale University Press, 2015). ²¹ See Theron Pummer, “Whether and Where to Give,” Philosophy & Public Affairs 44 (2016), pp. 77–95. ²² See Pummer, “Whether and Where to Give,” pp. 80–1. ²³ Or if the personal burden on the agent of so acting is weighty enough to defeat this presumption on some occasions, that would merely seem to show that the optimific charitable

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

70

.  

But I think we should resist this conclusion, for two closely related reasons. First, to think about mutual aid in this way is to think of it in essentially individualistic terms, which ignores the collective dimension of the problem that is central to the fair share approach. Those in the class of potential beneficiaries have claims against us that we should do our part to support institutions or organizations that target, on an ongoing basis, the basic needs of the individuals who are worst off amongst us. If all were contributing their fair share, the result would be that these important moral claims to assistance would be met, for that is the standard by reference to which we determine what is a fair share contribution in the first place. But meeting this goal is a collective responsibility, something that is lost sight of when we are expected, as a matter of individual moral obligation, to fine tune our contributions so as to take up some of the slack created when others who are similarly positioned fail to do anything. This is, in effect, the upshot of the effective altruist approach, and it neglects the character of our duties of mutual aid as collective responsibilities. Second, the considerations that support the idea that there is wide discretion in relation to these duties are not in any case appropriately conceptualized merely in terms of trade-offs between costs for the agent and incremental gains for others that would accrue if those costs were borne. The rationale for wide discretion in this case appeals, in ways I have not yet emphasized, to the considerations about relationships of agency that we saw to play a role in the example of gratitude. The interests of ours that are especially significant when thinking about our obligations of mutual aid include interests in having a life of our own that reflects our projects and attachments and sense of what is engaging and important. By leaving scope for individuals to choose between a range of opportunities to contribute to collective efforts of mutual aid, morality enables them to integrate their efforts, to some extent, with the individual projects that help to define them as distinctive centers of agency. We understand ourselves to be members of local communities, for instance, and this can make it important to us to do something to help the homeless and destitute who immediately live among us, even if they are better off in absolute terms than many who inhabit less affluent parts of the world. Similarly, we might take a special interest in

action is supererogatory. The result would be that our determinate obligation is to perform the suboptimal action that is less burdensome for us, rather than that we are under a less determinate duty of mutual aid that leaves us with wide agential discretion as to how to meet it.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

71

efforts that are being undertaken to discover treatments for a disease that a person close to us has suffered from.²⁴ Proponents of effective altruism acknowledge that effects of these kinds are among the costs to agents that need to be taken into account in determining what they are obligated to do for others. But their significance is not merely as burdens on the agent, but as considerations that affect our relations with those who stand to benefit from our charitable efforts. In discussing gratitude above, I referred to the idea that morality should enable relations between moral equals, in which each party is aware of themselves as interacting with others who also have lives of their own to lead. Agential discretion allows each of us to identify fair share contributions that make sense in relation to the rest of our lives, and that do not merely result from the rote application of impersonal standards to our situation. We can even make it one of our projects to encourage people to engage in more effective forms of altruistic activity, if we so choose, writing books about such things and establishing communities of shared interest in pursuing this goal.²⁵ Adopting this as a personal ambition is something that could well help to give shape and meaning to person’s life, as one of their ground projects (in Williams’s familiar sense²⁶). But relations between benefactor and beneficiary align better with our moral ideals to the extent the contributions are understood to be expressions of the benefactor’s personality and attachments. Those relations become less a matter of noblesse oblige, and more actions that have a place within the benefactor’s own sense of what their life is about.

²⁴ The background assumption here is that, even under full compliance, there would be cooperative schemes that effectively address basic human needs, but that also give individuals discretion as to the nature of their fair share contributions. Such schemes might be marginally less efficient than alternatives under which the nature of individual contributions is precisely prescribed, resulting in a modestly higher level of required contributions from each. But individuals have compelling reasons, on their own behalf, to reject the idea that fair share contributions be understood in relation to a collective scheme that is maximally efficient. It is more important to be able to give shape to one’s life than to minimize the level of contributions one is required by morality to make. ²⁵ It is striking that the adherents of effective altruism construe it as a “movement,” and emphasize the ways in which it can lend meaning to your life to devote yourself to optimizing the good that comes of it. The attractions of this way of acquiring meaning, especially for the millennial generation, may lie in part in the crushing failure of political and other institutions to deal with the acute social and environmental challenges of the present; the lure of a certain style of technocratic iconoclasm; and the hostility of contemporary corporate culture to other forms of meaning in the workplace (such as loyalty, community, or commitment to long-term goals). ²⁶ See Bernard Williams, “Persons, Character, and Morality,” as reprinted in his Moral Luck (Cambridge: Cambridge University Press, 1981), pp. 1–19.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

72

.  

The relational alternative for which I have been pleading in this chapter enables us to make sense of mutual aid as one of several requirements that we owe it to other individuals to comply with, one whose distinctive features derive from the collective character of the institutions we have an individual obligation to support. Understood in this way, mutual aid is not a matter of maximizing the beneficial effects of our individual efforts, but of standing in relations of recognition with those who have specific claims against us. The wide discretion inherent in these duties of mutual aid has its real significance in this context. By allowing those who are better off to participate in their own way in the collective task of enabling the agency of the least advantaged, it promotes relations between benefactors and beneficiaries on a basis of agential autonomy and equality. We all have reason to care about interacting with each other on these terms, and this helps us to understand the significance of the agential discretion that inheres in our duties of mutual aid.²⁷

²⁷ For extremely helpful feedback on this (still very programmatic) paper, I am indebted to audiences at the Arizona Workshop on Normative Ethics in 2018; at the Bern–Zürich Workshop in Moral Theory in 2018; and at the University of Essen. I also received very helpful suggestions from an anonymous reviewer for Oxford University Press.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

4 Taking it Personally Third-Party Forgiveness, Close Relationships, and the Standing to Forgive Rosalind Chaplin

In ordinary language, it is not unusual to talk about forgiving wrongdoers for harms done to our loved ones. I might say to a friend, “I am not sure I will be able to forgive that woman for what she did to my brother.” Or a friend might say to me, “If I could speak to my father, I would want to tell him I’ve forgiven him for what he did to my mother.” In philosophical discussions of forgiveness, however, statements like these are controversial. For as many philosophers see it, victims are the only ones with the standing to forgive, and cases that appear to involve third-party forgiveness are in fact cases in which either (a) what looks like a third party is really a victim or (b) what looks like forgiveness is really not forgiveness.¹ This chapter challenges this common piece of philosophical dogma. Although there is something right about the intuition that not just anyone has the standing to forgive, the standing to forgive nonetheless is not limited to victims; some third parties can appropriately forgive (or refuse to forgive) wrongdoers for wrongs done to victims. In particular, I argue that third parties have the standing to forgive only when it is appropriate for them to take ‘personally’ wrongs done to victims.² This, in turn, is (typically) appropriate only when third parties are in close personal relationships with victims. I thus advance a ‘victim-relative’ account of third-party forgiveness: ¹ Scholars that explicitly reject third-party forgiveness (3PF) include Govier and Verwoerd (2002), Haber (1993), Murphy and Hampton (1988), Owens (2012), Walker (2013), Warmke (2016a), and Zaragoza (2012), among others. Many others focus exclusively on victim forgiveness, even if they do not explicitly reject 3PF. Exceptions to this are Bennett (2018), Griswold (2007), MacLachlan (2008 and 2017), Milam (2017), Norlock (2008 and 2009), Pettigrove (2009), and Radzik (2010). ² I borrow this term from MacLachlan (2008). Although I part ways with MacLachlan on what ‘taking it personally’ requires, my view is greatly indebted to her work. Rosalind Chaplin, Taking it Personally: Third-Party Forgiveness, Close Relationships, and the Standing to Forgive In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0004

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

74

 

although one can have the standing to forgive without having been wronged, the standing to forgive does not extend to all bystanders; in most circumstances, the standing to forgive is limited to victims and their loved ones. I begin in Section 1 by presenting a series of cases that provide intuitive support for a victim-relative view. Given cases of the sort I describe, there is prima facie support for the conclusion that third parties have the standing to forgive when they are in personal relationships with victims. Section 2 considers how opponents of third-party forgiveness would respond to these cases. Typically, opponents of third-party forgiveness argue that apparent cases of third-party forgiveness can always be adequately analyzed as covert cases of victim forgiveness. I argue that this strategy fails on two fronts. First, it fails to capture the nature of our concern for our loved ones when we forgive as third parties. Second, it fails to account for the special norms governing third-party forgiveness. Section 3 turns to the question why some but not all third parties have the standing to forgive. I argue that third-party standing to forgive is grounded in the standing to take wrongs done to victims ‘personally,’ which is (typically) limited to those in close personal relationships with victims. Section 4 considers the objection that the standing to forgive in fact extends to all third parties. As I argue, like victim-exclusive accounts, this view fails to account for the special norms governing third-party forgiveness. Section 5 highlights a further benefit of focusing on third-party forgiveness: it can help us adjudicate between competing accounts of the nature of forgiveness. Although I do not attempt to give conclusive arguments against any particular view, I suggest that attitude-based accounts are best able to accommodate our practices of third-party forgiveness.

1. Three Cases: Intuitive Support for a Victim-Relative Account To begin, consider the following three cases: Secret: You are attending a block party with your partner, and several of his family members are in attendance. As you stand around the grill with a cluster of neighbors, your partner’s sister intentionally reveals an embarrassing secret of his. You know your partner has asked her to keep the information private, and you’re angry with her for being so inconsiderate. You decide not to forgive your partner’s sister until she sincerely apologizes to your partner.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

75

Bad Ex: Anna and Mayra are close friends, and Mayra has recently gone through an ugly breakup. Mayra decides that she needs to forgive her ex so that she can move on with her life. Anna supports this decision but feels strongly that she herself should not forgive Mayra’s ex. It may be best for Mayra to forgive him, Anna decides, but as a staunch friend she will continue to hold her ex’s bad behavior against him. Woods: Phoebe and Frank spend weekend afternoons watching SportsCenter on ESPN. It’s 2009, and much of the news concerns Tiger Woods’s infidelity. After it’s announced that Woods has finally admitted to an affair, Phoebe turns to Frank and says, “You know, I really resent Tiger Woods for what he did to his wife. I don’t think I’m going to be able to forgive him!” Intuitively, there is an important difference between Woods, on the one hand, and Secret and Bad Ex, on the other. In particular, Phoebe is behaving inappropriately in claiming that she will not be able to forgive Woods, whereas the third parties in Secret and Bad Ex are behaving in morally appropriate ways. When you consider whether or not to forgive your partner’s sister, you are acting out of care and respect for your partner. Likewise, when Anna decides not to forgive Mayra’s ex, she is acting out of a desire to protect and defend her friend. In contrast, Phoebe seems to misunderstand that she is merely a bystander to Woods’s situation. It is not legitimate for her to think of herself as having a personal stake in Woods’s behavior, and when she claims that she cannot forgive Woods, she fails to appreciate this.³ If this assessment of the cases is correct, then we can make the following preliminary claims about third-party forgiveness: 1. Third-party forgiveness is possible, and in some cases it is morally appropriate for third parties to forgive or refuse to forgive wrongdoers for wrongs done to victims. 2. Some but not all parties have the standing to forgive. Normally, this standing is limited to those in close personal relationships with victims.

³ However, this is not to say that Phoebe cannot have any blaming attitudes toward Woods. E.g., it may be appropriate for her to feel impersonal moral indignation. The claim here is just that the subset of blaming attitudes involved in forgiveness are not appropriate for Phoebe to have.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

76

 

Claim 1 is supported by cases like Secret and Bad Ex, where third parties appear to be acting within the bounds of moral propriety when they consider whether or not to forgive a wrongdoer. Claim 2 is supported by the above discussion of the difference between Woods, on the one hand, and Secret and Bad Ex, on the other; not every third party can legitimately consider forgiving wrongdoers for wrongs done to victims, since not every third party is appropriately related to the victim.⁴ However, we can also draw a further conclusion from the above cases. Consider the following counterfactual possibilities. Suppose that in Secret you decided to forgive your partner’s sister before she apologized to your partner. Or suppose you decided to forgive before your partner was ready to forgive. Intuitively, these decisions would not be morally appropriate. You should show deference to your partner as the victim of the wrong, and forgiving before your partner is ready to forgive undermines this.⁵ Similarly, ceteris paribus, an apology to your partner should be a precondition of your forgiveness, for insofar as your concern and respect for your partner made forgiveness a question for you in the first place, your concern and respect for your partner should continue to determine your response to his wrongdoer. Similarly, in Bad Ex, it would be inappropriate for Anna to forgive Mayra’s ex before Mayra had forgiven him, and (if she forgives at all) she should do so only after he makes appropriate amends with Mayra. It may even be fitting for Anna to be less forgiving of Mayra’s ex than is Mayra. For as staunch friends, Anna and Mayra have especially strong duties to support and defend one another; plausibly, this means Anna should continue to hold the ex’s bad behavior against him even after Mayra has decided to move on.⁶ With these considerations in mind, we can draw a third conclusion about third-party forgiveness: 3. Our practices of third-party forgiveness are norm-governed. For example, we think third parties should not forgive until victims have forgiven (ceteris paribus), and in some relationships third parties should be less forgiving than victims, even after victims have forgiven. ⁴ I use the term ‘standing to forgive’ in this paper to indicate the status a person has when it is appropriate for her to consider whether or not she will forgive a wrongdoer. A person that lacks the ‘standing to forgive’ cannot appropriately take up the question “Will I forgive?” in the first place. ⁵ But one could imagine cases where the presumption against forgiving before the victim is overridden (e.g., when a victim is being wholly unreasonable in refusing to forgive). ⁶ Although she does not explicitly endorse 3PF, Priest (2016) offers a competing view according to which victims’ forgiveness renders inappropriate close associates’ continued blame.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

77

Note that claim 3 also provides additional support for claim 1. If third-party forgiveness were either impossible or always inappropriate, there wouldn’t be special norms distinguishing appropriate from inappropriate exercises of the third-party prerogative to forgive. This is not to deny that many of the norms governing third-party forgiveness concern the ways in which third parties should be responsive to victims’ decisions and circumstances, but this does not undermine the claim that some third parties have the standing to forgive. It simply shows that forgiving third parties are constrained in unique ways by considerations pertaining to victims.⁷ Note also that the claim that third parties sometimes have the standing to forgive should not be mistaken for the claim that third parties can forgive on behalf of victims (i.e., that they can co-opt the victim’s prerogative to forgive). As one would expect, third parties do not have the authority to forgive for victims or to put victims and wrongdoers back on good terms; whether or not victims forgive is up to victims.⁸ This may place additional constraints on our general account of the nature of forgiveness (see Section 5), but at present I remain neutral on the nature of forgiveness except to stress that third-party forgiveness does not undermine the victim’s prerogative to decide for herself whether she will forgive her wrongdoer.⁹

2. Arguments for Victim-Exclusive Views The discussion above has aimed to provide intuitive support for third-party forgiveness. But how would proponents of victim-exclusive views deal with cases of the sort just described? That is, how would they attempt to establish that forgiveness is the exclusive prerogative of victims? Broadly speaking, victim-exclusive accounts are motivated by two main kinds of argument. One kind of argument appeals to putative facts about the logic of forgiveness to establish that third-party forgiveness is conceptually incoherent (and so impossible). Another strategy appeals to claims about victims to establish that third-party forgiveness is always morally improper. ⁷ Note that if I’m correct that this standing is (normally) grounded in close relationships between third parties and victims, then it makes good sense to say that considerations pertaining to victims play important roles in the norms of 3PF. ⁸ An exception to this may be a case where a victim explicitly grants a third party the power to forgive as a proxy. ⁹ However, I do make some uncontroversial assumptions such as that forgiveness presupposes culpable wrongdoing such that one does not forgive as the result of discovering that a putative wrongdoer was not responsible after all.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

78

 

First consider the claim that third-party forgiveness is conceptually incoherent. A common argument for this conclusion runs roughly as follows: (P1) X can forgive Y for a wrong done by Y only if X has the standing to resent Y for that wrong. (P2) X has the standing to resent Y for a wrong done by Y only if X is the victim of Y’s wrongdoing. (P3) Therefore, X can forgive Y for a wrong done by Y only if X is the victim of Y’s wrongdoing. Premise 1 reflects an assumption about the nature of forgiveness: whether or not forgiveness always requires forswearing resentment, if a person has the standing to forgive, then she at least has the standing to resent. This claim is compatible with the thought that forgiveness sometimes involves forswearing attitudes other than resentment (e.g., hurt feelings),¹⁰ and it is compatible with the idea that forswearing negative attitudes may be neither necessary nor sufficient for forgiveness.¹¹ Premise 1 simply states that a person with the standing to forgive cannot be morally criticized for resenting if she resents (ceteris paribus). Premise 2 makes a claim about the nature of resentment: resentment is a response belonging only to victims. P. F. Strawson (1962) provides one way of motivating this claim when he says that resentment is “essentially [a] reaction[] to the quality of others’ wills towards us.”¹² If this is correct, then it simply makes no sense to say that I feel resentment when a wrongdoer shows ill will towards someone else. Jeffrie Murphy (Murphy and Hampton 1988) also endorses a claim like this when he says that resentment is essentially a response to wrongs against the self. As Murphy argues, “resentment (in its range from righteous anger to righteous hatred) functions primarily in defense, not of all moral values and norms, but rather of certain values of the self.”¹³ Thus, the argument goes, if I claim to resent a wrongdoer who has not victimized me, I must be confused about what ¹⁰ See Allais (2008), Bell (2008), Murphy (1998), Murphy and Hampton (1988), and Richards (1988). ¹¹ E.g., even if forgiveness is best understood in terms of, say, releasing wrongdoers from debts, one could still hold that a person has the standing to forgive only if she has the standing to resent. ¹² Strawson (1962, 11). ¹³ Murphy and Hampton (1988, 16). In 1998, Murphy allows for self-forgiveness and admits that attitudes other than resentment can play a role in forgiveness. However, Murphy does not modify his account of resentment itself or suggest that the other attitudes that can figure in forgiveness make room for 3PF.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

79

resentment is. Perhaps I am confusing Strawson’s “vicarious” attitudes such as moral disapprobation and indignation for resentment.¹⁴ Or perhaps I am using the claim that I resent the wrongdoer as a stand-in for the claim that the wrongdoer has done something worthy of resentment by the appropriate parties. Whatever the case, according to premise 2, only victims can actually resent, and if both premise 1 and premise 2 are correct, it follows uncontroversially that only victims can forgive. How does a conceptual argument like this fare against the cases discussed in Section 1? The central problem with the conceptual argument is that it rests on stipulative considerations about resentment, and considerations of this sort should not count as evidence against third-party forgiveness. Consider again Anna’s response to Mayra’s ex. According to the conceptual argument, Anna cannot forgive because she cannot feel resentment, and she cannot feel resentment because she is not a victim. But a proponent of thirdparty forgiveness can simply reject this assumption about resentment (i.e., she can reject premise 2). Anna can feel resentment, she can argue, for Bad Ex shows that resentment is not essentially a response to wrongs against the self. If one were to insist that it is simply a matter of definition that resentment belongs only to victims, then a proponent of third-party forgiveness would have a new reason to reject premise 1. For premise 1 is appealing only if we have not defined resentment so as to prejudge the question of third-party forgiveness. If it is not a matter of definition that only victims can resent, then Bad Ex (and Secret) should count as evidence that premise 2 is false.¹⁵ If the conceptual argument fails, does the moral argument do any better? Traditionally, proponents of the moral argument appeal to cases like Woods to show that third-party forgiveness is always morally untoward. As the argument goes, all third parties that take themselves to be in a position to forgive are like Phoebe; they fail to appreciate the difference between their ¹⁴ Strawson (1962, 11). ¹⁵ Here we might also consider a particular view about blaming attitudes, e.g., that they are forms of protest. The conceptual argument says that resentment by definition protests wrongs against the self; moral indignation protests wrongs done to others. The proponent of 3PF wants to show that cases like Bad Ex undermine this view of resentment and indignation; indignation may protest general wrongdoing, but resentment protests wrongs done not only to ourselves but also to our loved ones. For as the case shows, the protest “How dare you do that sort of thing to my friend!” has more in common with “How dare you do that sort of thing to me!” than with “How dare you do that sort of thing!” (i.e., it has more in common with attitudes protesting wrongs to the self than with attitudes protesting general wrongdoing) (see Griswold 2007, 45 and Priest 2016, 627 for related discussion). The conceptual argument appears unresponsive to these considerations.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

80

 

own situations and the situations of victims, and because of this, they overstep the bounds of moral propriety.¹⁶ As we have seen, however, cases like Secret and Bad Ex undercut a quick generalization from Woods-like cases to the conclusion that all third parties lack the standing to forgive; in Secret and Bad Ex, third parties seem to be acting in morally appropriate ways when they consider whether or not to forgive their loved ones’ wrongdoers. Does the opponent of third-party forgiveness have a response to cases like Secret and Bad Ex? One natural move is to argue that these cases merely seem like legitimate cases of third-party forgiveness and that the forgiving agents are in fact secondary victims. That is, one might attempt to deflate the significance of Secret and Bad Ex by arguing that they are disguised cases of victim forgiveness. As the argument goes, you can forgive (or refuse to forgive) your partner’s sister only because you are harmed when your partner is harmed. Likewise, Anna is in a position to forgive (or withhold forgiveness from) Mayra’s ex only because the injury to Mayra is also an injury to Anna. This response by the opponent of third-party forgiveness brings our attention to an important point: wrongs against our loved ones normally do harm us, for part of what it is to be in a close relationship with someone is to have one’s own well-being depend on that person’s. Nonetheless, to suggest that apparent cases of third-party forgiveness are disguised cases of victim forgiveness misrepresents the nature of our concern for those we love when we forgive as third parties.¹⁷ To see why this is so, consider again Secret. According to the proposal at hand, when you consider forgiving your partner’s sister, you are in fact acting in your capacity as a secondary victim; your partner’s sister has victimized not only your partner but also you, and this is what grounds your standing to forgive. Notice, however, that if this is correct, then when you consider whether or not to forgive your partner’s sister, you should in fact be considering whether or not to forgive her for what she did to you. For if you can forgive only in your capacity as victim, then presumably you can forgive only for wrongs done to you. But this misconstrues what you care about when you consider whether or not to forgive your partner’s sister. Even if you acknowledge that you were wronged by your partner’s sister, your primary concern is the wrong done to your partner, not the wrong done to yourself. It would be perfectly ¹⁶ Zaragoza (2012) offers an argument like this. ¹⁷ Pettigrove (2009, 588–90) also makes this point, as does MacLachlan (2017, 150).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

81

coherent (and in fact fitting) for you to say that although you were harmed by the wrong to your partner, the question whether you will forgive the sister for what she did to your partner is independent of the wrong done to you. Indeed, this helps to make sense of why if your partner’s sister apologizes to you for what she did, then unless she also apologizes to your partner, you will be unlikely to forgive her. Insofar as your primary concern is the wrong done to your partner, you may even consider the sister’s apology to you irrelevant.¹⁸ In short, accounts that make forgiveness the exclusive prerogative of victims fail to capture the important ways in which our concern for our loved ones can make it a question for us whether we will forgive a wrongdoer. And so while it is undoubtedly true that offenses often have multiple secondary victims, and these secondary victims are often those in close relationships with primary victims, it is not merely insofar as we are victims that we have the standing to forgive. One feature of being in close relationships with others is that our well-being comes to depend on the well-being of others, but it is also a feature of being in close relationships with others that we care about (and react emotionally to) wrongs done to others notwithstanding the impact those wrongs have on our own welfare. Hence, we need to acknowledge third-party forgiveness as a genuine form of forgiveness to capture the nature of our concern for those we love.¹⁹

3. What Grounds the Standing to Forgive? If the arguments above are correct, then some third parties have the standing to forgive. But what explains why not all third parties have that standing? In this section, I argue that third-party standing to forgive is grounded in the standing to take wrongs done to victims ‘personally’, which is normally limited to victims and their loved ones. I borrow the notion of ‘taking it personally’ from Alice MacLachlan’s “Forgiveness and Moral Solidarity” ¹⁸ A proponent of victim-exclusive views might point out here that it would make sense for you to consider an apology to the primary victim a precondition of your own forgiveness if the victim’s injury were partly constitutive of the injury done to you. Thus, your own injury might justify your resentment, but your focus might still be the harm done to the victim. This view is coherent, but it is a significant cost in that it forces us to divorce the justification of our resentment (in this case, the fact that we ourselves were injured) from the target of our resentment (the harm done to our loved ones). ¹⁹ Similar considerations explain why we cannot construe apparent third-party forgivers as harmed members of the moral community forgiving in their capacities as tertiary victims.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

82

 

(2008). According to MacLachlan, forgiveness always involves a claim to take a wrong personally, and third parties can legitimately make this claim when they meet two conditions: they must (a) be in caring relationships with victims and (b) empathize with victims’ experiences.²⁰ I agree with MacLachlan that the standing to forgive presupposes the standing to take an offense personally, but I propose a different account of the ground of that standing. Rather than requiring empathy, I argue that third-party standing to forgive is (normally) grounded in close relationships alone. Thus, victims can take wrongs personally because they were wronged; those with close relationships to victims can take wrongs done to victims personally because of their close relationships. So while some proponents of third-party forgiveness argue that we have drastically underestimated how far the standing to forgive extends, I argue that the standing to forgive extends only slightly farther than traditional accounts of forgiveness take it to; third-party forgiveness is victim-relative, even if it is not victim-exclusive. First consider MacLachlan’s basic insight about the connection between forgiveness and taking wrongs personally: whenever we consider forgiving, we make an implicit claim to take the wrongdoing ‘personally.’ This view is appealing in at least two respects. First, it allows us to map out the standing to forgive such that (i) victims always have the standing to forgive (barring defeating conditions),²¹ (ii) perfect strangers to victims rarely have the standing to forgive,²² and (iii) close family and friends of victims normally have the standing to forgive (again, barring defeating conditions). This coheres with our considered judgments about when it is morally appropriate for agents to consider forgiving wrongdoers. Second, insofar as we understand taking a wrong ‘personally’ in terms of reacting with personal as opposed to impersonal attitudes, MacLachlan’s suggestion is consistent with the idea that forgiveness presupposes the standing to resent (and not merely the standing to blame with impersonal attitudes such as indignation). When I am in the position legitimately to consider whether to forgive as a third party, I can react emotionally in ways that would not be appropriate if I had no relationship to the victim. This is explained by the fact that close relationships legitimize attitudes that express a special concern for the ²⁰ See MacLachlan (2008, 8–9). See also Priest (2016) for a discussion of “relationshipcentered blame” (or “Associate Blame”), which I take to be closely related to ‘taking it personally.’ ²¹ E.g., hypocrisy might undermine the standing to forgive if it undermines the standing to blame. ²² An exception to this is discussed in Section 4.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

83

well-being of the victim; resentment is one of these attitudes.²³ Thus, although we should resist the idea that forgiveness is appropriate only when we ourselves are harmed, the idea that forgiveness paradigmatically takes place in personal as opposed to impersonal situations is compelling. What about the idea that an ability to empathize with the victim is required for ‘taking it personally’?²⁴ In my view, this demand is too restrictive. For consider a case where a family member is seriously wronged: suppose you discover that your brother or son was abused by a teacher at school. The idea that your standing to forgive (or refuse to forgive) his wrongdoer depends on your ability to empathize seems inappropriate here; even if you are unable to empathize with the experience of your family member, you can legitimately consider whether to forgive his wrongdoer.²⁵ Or alternatively, suppose a friend is wronged but is either unaware of or unbothered by the offense. In a case like this, your friend lacks the kind of subjective experience that would be an appropriate object of empathy, but it seems you are still in a position to ask yourself whether you will forgive the wrongdoer. Thus, good empathizers and poor empathizers alike have the standing to forgive wrongs done to their loved ones, and we can forgive (or refuse to forgive) as third parties even when empathy cannot get a hold. In contrast, the idea that caring relationships ground third-party standing to forgive is compelling. Consider again Secret and Bad Ex. In both these cases, the relationships between third parties and victims make it appropriate for third parties to take a special interest in the victim’s well-being and to treat wrongdoers in ways that mere bystanders cannot appropriately treat them. In Secret, your relationship with your partner legitimizes your resenting his sister. Plausibly, this is because your relationship involves a commitment to standing up for your partner; insofar as resentment can protest his injury and socially sanction his sister, refusing to forswear resentment helps you satisfy your obligations to him. Similarly, in Bad Ex, Anna’s friendship with Mayra makes appropriate her decision to continue holding the ex’s ²³ For recall that refusals of forgiveness can help us stand up for and defend our loved ones. So insofar as I have special obligations to stand up for and defend my loved ones, refusing to forgive their wrongdoers can help me meet these obligations. See Radzik (2011, 593) for the related idea that refusals of forgiveness often function as forms of social sanction). ²⁴ Recall Maclachlan’s claim that “a third party prerogative [to forgive] is doubly grounded: first, in a prior relationship of identification with or care for the victim or wrongdoer, and second, in a careful and attentive empathetic engagement with the victim’s experience of having been wronged” (2008, 9). ²⁵ A similar criticism applies to MacLachlan’s more recent (2017) suggestion that forgiving third-parties must engage in “imaginative sympathy”; if I stand in the right relation to the victim, it doesn’t matter whether I’m able to imagine what her experiences are like.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

84

 

behavior against him; for if friends have special obligations to protect and defend one another, and if refusing forgiveness can affirm a victim’s value and defend her worth, then Anna’s refusing to forgive Mayra’s ex is part and parcel of her being a good friend to Mayra.²⁶ The picture I am recommending, then, is as follows. Our close relationships give rise to special obligations and entitlements, and these obligations and entitlements explain why third-party standing to forgive arises just where it does.²⁷ For instance, our relationships with our family members typically require us to take a special interest in their well-being; this requirement entitles us to the class of ‘personal’ attitudes that reflect this interest (a class that includes attitudes like resentment). Moreover, insofar as attitudes like resentment are especially powerful forms of social sanction and especially effective affirmations of value, they often prove indispensable to us in our ongoing efforts to fulfill the obligations of our close relationships. Without the ability to withhold forgiveness from our loved ones’ wrongdoers, we would be deprived of an important means of standing up for those we care about. Undoubtedly, there are forms of relationships that present borderline cases in which it is difficult for us to determine whether thirdparty forgiveness is appropriate, but this is to be expected insofar as it is sometimes unclear exactly which obligations and entitlements our relationships give rise to. To the extent that it is unclear whether a particular relationship gives rise to the obligations and entitlements that make ‘taking it personally’ appropriate, it will also be unclear whether third-party standing to forgive arises in that relationship.²⁸ Note also that on my view taking it personally is not equivalent to identifying with victims such that one sees oneself as a victim. Indeed, this would threaten the claim that forgiving third parties can forgive not only in ²⁶ Note: on my view, an account of 3PF is better off if it is permissive with respect to the function of attitudes like resentment; they can express protest, serve as social sanctions, stand in defense of a person’s value, and so on. Attending to the variety of roles played by these attitudes can help us see why they serve such an important function in our ongoing efforts to fulfill the obligations of our close relationships. ²⁷ I hope to remain relatively neutral on exactly which special obligations and entitlements particular relationships give rise to, as this no doubt varies greatly from case to case. It is enough for my purposes that relationships (such as friendship, romantic partnership, siblinghood, and so on) give rise to special obligations and entitlements and that these obligations and entitlements make ‘taking it personally’ appropriate. ²⁸ For example, can a parent forgive or refuse to forgive an estranged child’s wrongdoer? The central point here is just that when we are unsure whether a third party has the standing to consider forgiveness, our uncertainty should be explicable in terms of uncertainty with respect to the nature of the relationship between the victim and the third party and the obligations and entitlements to which that relationship gives rise.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

85

their capacities as victims but also in their capacities as third parties. Instead, as suggested above, I understand taking a wrong personally in terms of responding with attitudes that reflect a special, personal interest in the wellbeing of the wronged. When I take wrongs done to loved ones personally, my attitudes make an implicit claim that my loved ones’ well-being and hence their victimization is my business. My partner’s victimization is my business because I have made a special commitment to supporting him. My friends’ injuries are my business because of my special obligations to standing up for them. This is not to say that there are no limitations on the ways in which the lives of our relationship partners are our business, but, as a general matter, wrongs done to our loved ones are our concern.²⁹ And because they are our concern, we can legitimately consider forgiving our loved ones’ wrongdoers. Finally, note that we could also describe this distinction between personal and impersonal reactions to wrongdoing in Strawsonian terms. Just as Strawson argued that there is a difference between our demand that others express goodwill toward ourselves and our demand that they express goodwill toward all people, so too there is a difference between our demand that others show goodwill toward our loved ones and our demand that they show goodwill toward all people. Attitudes that pertain to our own treatment and the treatment of our loved ones are ‘personal,’ and we explain their appropriateness by appealing to special features of our own circumstances. When we ourselves are victims, our status as victim is the feature that explains the appropriateness of our response; when our loved ones are wronged, our special relationship with the victim plays this role.

4. Why Not Extend the Standing to Forgive to All Bystanders? At this point, one might object that it is not clear why we shouldn’t extend the standing to forgive to all bystanders.³⁰ Two strategies suggest themselves here. First, one might resist the claim that only those in close relationships with victims have the standing to take wrongs done to victims personally;

²⁹ In contrast, Phoebe cannot take what Woods did to his wife personally because she does not have any of the obligations or entitlements that would make wrongs done to his wife her business. ³⁰ See Radzik (2010).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

86

 

perhaps there are cases in which perfect strangers to victims have good reasons to take wrongs against victims personally. Second, one might resist the claim that forgiveness requires taking wrongs personally at all; perhaps one merely has to have the standing to blame (impersonally) to have the standing to forgive. I consider these objections in turn. What considerations could motivate the idea that perfect strangers to victims have the standing to take wrongs against victims personally? One could imagine several different lines of argument. First, consider MacLachlan’s suggestion that common experiences can build relations of moral solidarity that make third-party forgiveness appropriate.³¹ For instance, victims of hate crimes or sexual assault may have moral solidarity with other victims such that it is appropriate for them to consider forgiving or refusing to forgive perpetrators for crimes against other victims. If common experience makes all crimes of the relevant type personal for past victims, then any bystander with the relevant type of shared experience can in principle forgive. Or consider the idea that taking it personally is appropriate whenever one is in a position to be sensitive to issues such as when deference to the victim is needed, when the victim has truly forgiven, and so forth. Arguably, it is simply a practical matter of fact that those in close relationships with victims are best able to appreciate these facts about victims, and for this reason, any third party can in principle forgive. Although I do not think third-party standing to forgive frequently extends beyond those close to victims, I am willing to concede that shared experiences may enable third parties to legitimately take wrongs done to victims personally.³² So long as ‘taking it personally’ in these cases reflects a legitimate special interest in the well-being of the victim (and does not simply involve conceiving of oneself as a secondary or tertiary victim of the wrong), a third party with an experience similar to a victim’s may have the standing to forgive.³³ However, I would more strongly resist the second proposal. For although it is true that people close to victims are typically best able to show deference and sensitivity to victims and to understand their ³¹ MacLachlan (2008, 9). ³² For this reason, my view is best described as a ‘victim-relative’ view of 3PF. Close relationships are the paradigmatic setting in which third-party standing to forgive arises, but an appropriate relation to the victim may sometimes suffice. ³³ If a third party takes a victim’s injury personally because her shared experience makes her feel newly threatened or degraded, then we may not have a genuine case of 3PF. However, if her shared experience engenders a legitimate, special concern for the victim, and if this concern reflects a legitimate sense in which the victim’s injury is her business, then the case may be a genuine case of 3PF.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

87

situations, it is not merely the information that they have that gives them the standing to forgive. Even if Phoebe were given all the information about Woods and his wife that a close family member had, her relationship with Woods’s wife is not the right sort of relationship to ground third-party standing to forgive. So while information and understanding may be the result of close relationships, they alone cannot ground third-party standing to forgive. A second objection is that third-party standing to forgive does not presuppose the standing to take wrongs personally at all. Perhaps forgiveness requires nothing more than forswearing moral indignation. If this is correct, then third-party standing to forgive extends to anyone who can appropriately express this attitude. And so while the standing to forgive might be undermined by defeaters such as hypocrisy, it would not depend on a putative forgiver’s relationship to the victim. However, this proposal is unsatisfying in two respects. First, it forces us to revise too many of our considered judgments about when it is appropriate for us to contemplate forgiving as third parties (recall our assessment of the differences between Secret, Bad Ex, and Woods). Even more importantly, however, proponents of universal standing to forgive cannot tell a unified story about how the standing to forgive relates to the norms governing third-party forgiveness. I have argued that many of the norms governing third-party forgiveness have their bases in the obligations associated with particular kinds of close relationships. For example, close friends should be relatively unforgiving of one another’s wrongdoers because they have especially strong duties to protect and defend one another. I have also argued that the very same relationships that explain these norms also make ‘taking it personally’ appropriate. If the proponent of unrestricted third-party forgiveness is correct, however, then the standing to forgive has a generic basis, and an explanatory story of this sort is not possible. That is, there would be a generic ground for third-party standing to forgive, and yet the norms of third-party forgiveness would (at least sometimes) have their grounds in particular forms of close relationships. While this is not impossible, an account according to which norms and standing share a common ground is explanatorily more compelling.³⁴ ³⁴ Here is another objection a proponent of universal forgiveness might raise. Suppose all third parties have the standing to forgive, but practical constrains prevent them all from engaging in the practice. Perhaps most third parties lack the information that makes following the norms of 3PF feasible, or perhaps ‘taking it personally’ is too emotionally burdensome for us to be doing all the time. Then, our intuitions about standing might be faulty, merely tracking the

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

88

 

5. Theorizing with Third-Party Forgiveness: The Nature of Forgiveness In the remainder of this chapter, I consider the implications of third-party forgiveness for our overall account of the nature of forgiveness. Although I do not aim to provide a full defense of any particular account of forgiveness, I hope to bring out some of the challenges facing alternatives to attitude-based accounts. As I argue, these challenges provide a defeasible reason to favor attitude-based views. As acknowledged at the outset of this chapter, I have assumed that the standing to forgive presupposes the standing to resent. Although this claim about standing is strictly speaking neutral between various accounts of the nature of forgiveness, it naturally suggests an account according to which forgiveness consists in forswearing negative reactive attitudes. After all, it would be strange for the standing to forgive to presuppose the standing to resent if negative attitudes like resentment played no role in actual acts of forgiveness. I have also further reinforced these suggestions by placing ‘taking it personally’ at the center of my account of third-party forgiveness. If taking it personally is best understood in terms of responding with personal attitudes such as resentment, then it makes sense to say that paradigmatic instances of forgiveness involve forswearing these attitudes. Hence, my discussion above fits well with ‘attitude-based’ accounts: in paradigmatic cases, forgiveness centrally involves foreswearing negative attitudes such as resentment.³⁵ But there are other ways in which one might think about the nature of forgiveness. Consider the view that forgiveness consists in a commitment to seeing wrongdoers in a new way.³⁶ At least on the face of it, it looks like this account can accommodate third-party forgiveness, since victims and third parties alike can commit to seeing wrongdoers in a new way (e.g., as separate from their wrong actions or as better than their wrong actions indicate they

facts about how far our practices of 3PF happen to extend. I think the existence of norms governing 3PF helps to answer this objection too; it would be surprising if the norms of 3PF were grounded in particular forms of close relationships, and yet the standing to forgive had a fully generic basis. ³⁵ See Hieronymi (2001) and Murphy and Hampton (1988), among many others. ³⁶ See Ware (2014). Allais (2008) also emphasizes the importance of ‘seeing in a new way’ in forgiveness, but for Allais this is meant to help flesh out the intentional content of the attitudes involved in forgiveness rather than to serve as an alternative to attitude-based accounts.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

89

are). However, this view fails to explain why only some third parties have the standing to forgive, for it seems perfect strangers to victims can legitimately commit to seeing wrongdoers in a new way. One might argue that those close to victims have access to a unique set of reasons to commit to seeing (or refusing to see) wrongdoers in a new way, but one would still need to explain why having just these reasons makes forgiveness appropriate. This is not a decisive strike against the view, but it shows that its proponents need to say more about why the standing to forgive does not extend to all third parties. Other views face still greater challenges in accommodating third-party forgiveness. Consider the view that forgiveness involves restoring relationships with wrongdoers. It is clearly possible for third parties to restore relationships with wrongdoers, but consider the following dilemma. In cases of third-party forgiveness, either the relationship that is restored is the relationship between the victim and the wrongdoer, or it is the relationship between the third party and the wrongdoer. For the same reasons that third parties do not have the standing to forgive on behalf of victims, forgiving third parties do not have the authority to restore relationships between victims and wrongdoers. But if this is true, the restoring-relationships model must hold that third-party forgiveness involves restoring relationships between forgiving third parties and wrongdoers.³⁷ On the face of it, this may look plausible, but recall that when we forgive as third parties, what we care about is the effect that wrongdoers’ actions have on victims. Suppose you decide to forgive your partner’s sister for embarrassing your partner. Although your forgiveness may have as a consequence a restored relationship between you and your partner’s sister, that relationship is not your fundamental concern; what you care about is the injury to your partner. One could insist at this point that third-party forgiveness consists in restoring relationships between wrongdoers and third parties even though that relationship is not the primary concern of the forgiving third party, but a compelling account of forgiveness should tell a more satisfying story about the link between the nature of forgiveness and the considerations that most concern forgivers. Indeed, in the case of third-party forgiveness, the issues that concern potential forgivers seem to be compatible with there being no relationship at all between forgiver and wrongdoer. If forgiving third parties are not motivated by the status of their relationship with wrongdoers and ³⁷ Pettigrove (2009) takes this route in his defense of relationship-restoration as an important part of 3PF.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

90

 

could in principle forgive (or refuse to forgive) in the absence of having any relationship with wrongdoers, the view that third-party forgiveness consists in restoring relationships is poorly motivated. Finally, consider the view that forgiveness involves releasing wrongdoers from moral debts.³⁸ Call a debt-release model ‘strict’ if it says that debtrelease is necessary for forgiveness. On a ‘strict’ debt-release model, thirdparty forgiveness must consist either in third parties releasing wrongdoers from debts owed to victims or in third parties releasing wrongdoers from debts owed to third parties. The first interpretation is objectionable for familiar reasons; third parties cannot cancel debts on behalf of victims. The second interpretation of the debt-release model is not straightforwardly objectionable like the first, but it is not clear how wrongdoers could incur debts to third parties in the absence of having wronged them. The proponent of third-party forgiveness must not appeal to wrongs done to third parties to explain the debts owed to them because she is committed to saying that third parties forgive in their capacities as third parties and not only in their capacities as victims. But this leaves it unclear exactly what generates the required debts or what they would be; wrongdoers do not seem to owe third parties apologies, and they certainly do not owe third parties compensation (at least not insofar as third parties are not also victims). If a ‘strict’ debt-release model cannot accommodate third-party forgiveness, perhaps a related account of forgiveness can. Consider Nelkin’s (2013) suggestion that the insight at the core of the debt-release model is the idea that forgiveness involves ceasing to hold offenses against wrongdoers.³⁹ If we take the notion of ‘ceasing-to-hold-against’ as fundamental (rather than debt-release per se), then perhaps we can accommodate third-party forgiveness. For instance, consider the idea that ceasing-to-hold-against involves relinquishing certain rights. If ceasing-to-hold-against is equivalent to rights-relinquishment, then perhaps forgiving third parties forfeits rights to demand that wrongdoers apologize to victims, compensate victims, make amends with victims, and so forth. Since third parties and victims alike can make these demands (even if only victims can release wrongdoers from obligations to victims), a rights-relinquishment model appears promising. Nonetheless, a rights-relinquishment model faces familiar problems when it comes to explaining why only some third parties have the standing

³⁸ See Warmke (2016b).

³⁹ See Nelkin (2013, 179).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

91

to forgive. For it seems that all members of the moral community have at least some prerogative to demand that wrongdoers apologize to victims, compensate victims, and so forth. What the rights-relinquishment model needs is a distinction between these rights (which all members of the moral community enjoy) and a special set of rights enjoyed only by victims’ family members, close friends, etc. It is plausible that victims’ loved ones have special kinds of rights to demand that wrongdoers fulfill their obligations to victims, but one would still need to explain why these rights (as opposed to the rights that are enjoyed by all bystanders) are precisely those rights whose relinquishment constitutes forgiveness. Again, this is not to say that a rights-relinquishment model cannot be appropriately developed, but it is to say that the model is unsatisfying as it stands. In contrast, the appealing feature of an attitude-based account is that it can tell a clear story about how the nature of forgiveness relates to the standing to forgive and the norms of third-party forgiveness. Close personal relationships make it appropriate for us to respond personally to offenses against our loved ones, and the personal attitudes involved in these responses are the very attitudes we forswear in paradigmatic cases of forgiveness. Moreover, as argued above, the relationships that originally make having these attitudes appropriate also explain the norms governing third-party forgiveness. Our relationships with our family members and close friends require that we take a special interest in their well-being, and this means that we should be relatively unforgiving of our friends’ and families’ wrongdoers, that we should not forgive before our friends and family members have forgiven, and so on. Finally, we can also note that refusing to forswear reactive attitudes like resentment helps us satisfy the demands of care and partiality to which our close relationships give rise. For retaining attitudes like resentment helps us demand of wrongdoers that they treat our loved ones with respect and goodwill. In short, attitude-based accounts can not only accommodate third-party forgiveness but also paint a compelling picture of how third-party standing to forgive and the norms of third-party forgiveness share a common ground in special relationships. And by appealing to the idea that reactive attitudes play a role in making moral demands, protesting wrongdoing, and affirming value, they can explain why decisions to forswear (or, more commonly, refuse to forswear) negative attitudes can help us fulfill our obligations to our loved ones.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

92

 

6. Conclusion I have argued in this chapter that the standing to forgive is not limited to victims and that third parties in close relationships with victims can appropriately consider forgiving wrongdoers for their wrongs. This is explained by the fact that third parties in close relationships with victims can appropriately take wrongs done to victims personally, which is itself justified by the special obligations and entitlements to which our close relationships give rise. I have attempted to diffuse several common objections to third-party forgiveness by showing that we need to acknowledge third-party forgiveness to account for the ways in which we respond to offenses against those we care about. Finally, I have also aimed to show that the existence of legitimate third-party forgiveness provides at least a preliminary reason to endorse attitude-based accounts of the nature of forgiveness. For attitude-based accounts not only accommodate third-party forgiveness, but they also tell an explanatorily compelling story about the way in which the norms of third-party forgiveness, the grounds of third-party standing to forgive, and the nature of forgiveness fit together.⁴⁰

References Allais, Lucy (2008). “Wiping the Slate Clean: The Heart of Forgiveness,” Philosophy and Public Affairs 36 (1): 33–68. Bell, Macalester (2008). “Forgiving Someone for Who They Are (And Not Just What They’ve Done),” Philosophy and Phenomenological Research 77 (3): 625–58. Bennett, Christopher (2018). “The Alteration Thesis: Forgiveness as a Normative Power,” Philosophy and Public Affairs 46 (2): 207–33. Govier, Trudy and Wilhelm Verwoerd (2002). “Forgiveness: The Victim’s Prerogative,” South African Journal of Philosophy 21 (1): 97–111. Griswold, Charles (2007). Forgiveness: A Philosophical Exploration. Cambridge University Press. ⁴⁰ For invaluable feedback on earlier drafts of this paper and numerous helpful conversations on the topic, special thanks to Lucy Allais, Claudia Blöser, David Brink, Kathleen Connelly, Cory Davia, Jonathan Knutzen, Per Milam, Leo Moauro, Dana Nelkin, Evan Tiffany, Shawn Wang, and Monique Wonderly. Many thanks also to the participants at UCSD’s MaPPS and GPC, the participants at the 2018 Arizona Workshop in Normative Ethics, and two anonymous reviewers, all of whose feedback has been extremely helpful.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

93

Haber, Joram Graf (1993). Forgiveness. Rowman & Littlefield. Hieronymi, Pamela (2001). “Articulating an Uncompromising Forgiveness,” Philosophy and Phenomenological Research 62 (3): 529–555. MacLachlan, Alice (2008). “Forgiveness and Moral Solidarity,” in Stephen Bloch-Shulman and David White (eds), Forgiveness: Probing the Boundaries. Inter-Disciplinary Press. MacLachlan, Alice (2017). “In Defense of Third-Party Forgiveness,” in Kathryn J. Norlock (ed.), The Moral Psychology of Forgiveness. Rowman and Littlefield International: 135–59. Milam, Per Erik (2017). “How is Self-Forgiveness Possible?” Pacific Philosophical Quarterly 98 (1): 49–69. Murphy, Jeffrie G. (1998). “Jean Hampton on Immorality, Self-Hatred, and SelfForgiveness,” Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition 89 (2/3): 215–236. Murphy, Jeffrie G. and Hampton, Jean (1988). Forgiveness and Mercy. Cambridge University Press. Nelkin, Dana (2013). “Freedom and Forgiveness,” in Ishtiyaque Haji and Justin Caouette (eds), Free Will and Moral Responsibility. Cambridge Scholars Publishing: 165–88. Norlock, Kathryn J. (2008). “Why Self-Forgiveness Needs Third-Party Forgiveness,” in Stephen Bloch-Shulman & David White (eds.), Forgiveness: Probing the Boundaries. Inter-Disciplinary Press: 17–29. Norlock, Kathryn J. (2009). Forgiveness from a Feminist Perspective. Lexington Books. Owens, David (2012). Shaping the Normative Landscape. Oxford University Press. Pettigrove, Glen (2009). “The Standing to Forgive,” The Monist 92 (4): 583–603. Priest, Maura (2016). “Blame After Forgiveness,” Ethical Theory and Moral Practice 19 (3): 619–33. Radzik, Linda (2010). “Moral Bystanders and the Virtue of Forgiveness,” in Christopher R. Allers and Marieke Smit (eds), Forgiveness in Perspective. Rodopi: 66–9. Radzik, Linda (2011). “Minding Your Own Business: Differentiating Accountability Relations within the Moral Community,” Social Theory and Practice 37 (4): 574–98. Richards, Norvin (1988). “Forgiveness,” Ethics 99 (1): 77–97. Strawson, Peter F. (1962). “Freedom and Resentment,” Proceedings of the British Academy 48: 1–25.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

94

 

Walker, Margaret Urban (2013). “Third Parties and the Social Scaffolding of Forgiveness,” Journal of Religious Ethics 41 (3): 495–512. Ware, Owen (2014). “Forgiveness and Respect for Persons,” American Philosophical Quarterly 51 (3): 247–60. Warmke, Brandon (2016a). “The Normative Significance of Forgiveness,” Australasian Journal of Philosophy 94 (4): 1–17. Warmke, Brandon (2016b). “The Economic Model of Forgiveness,” Pacific Philosophical Quarterly 95 (4): 1–20. Zaragoza, Kevin (2012). “Forgiveness and Standing,” Philosophy and Phenomenological Research 84 (3): 604–21.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

5 Persons as Things Mark Schroeder

Persons are things. We are biological creatures, things of flesh and blood, whose behavior is governed by the same principles that govern the behavior of any other social mammals, plus or minus the complications that come from the recursive possibilities of access to natural language. That much is fact.¹ But to be treated as a thing amounts to a deep insult. To be treated as a thing is to be minimized, rather than engaged with, predicted and controlled rather than reasoned with, written off as the product of our environment rather than appreciated for our unique contributions. We are deeply familiar with all manner of concrete ways of being treated as things, ranging from the straightforward sexual objectification of the entertainment press commenting on the appearance of actresses and female musicians, to the casual diminishment of the mansplainer who reduces his audience to an opportunity for display, to the well-meaning friend whose reaction to your latest accomplishment is that “see? I told you that you do better work when you’ve had caffeine.” There may be no single form to being treated like a thing, but we know it when we see it, and unless we manage ourselves well, we resent it. We expect—no, we demand— better from those around us. We expect to be treated as persons. Philosophers, more than anyone else, have waxed eloquent in their attempts to articulate the principle that we deserve to be treated as persons, and not merely as things. In Kant’s hands it is the principle that we must recognize one another as self-legislating co-rulers of a kingdom of ends, noumena who transcend the phenomenal appearance of being subject to predictive laws and act according to our conception of laws. Perhaps philosophers have taken this thought to heart most forcefully because it articulates what we seek for our own work—that others engage with our arguments, rather than merely taking them as evidence for what sorts of ideas are in vogue at west coast departments, or as a way to trace out the

¹ Compare Korsgaard (2011, 91). Mark Schroeder, Persons as Things In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0005

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

96

 

various strands of influence of dissertation advisors on their students’ second book projects. One of the central philosophical themes that we trace to Socrates is the conviction that ideas can and should be engaged with for ideas’ sake and without regard for their progeny, and writing and publishing philosophy is a way of aspiring to this ideal—to be engaged with as a rational being, in the domain of reasons and justifications, rather than merely as a psychological being, subject to prediction or explanation for what we say. So philosophers may tend to light on this observation about what we value from each other in ordinary life, in part because philosophers themselves value this so much in their own work. I believe that there is something deeply right and central about the idea that we deserve to be treated as persons and not merely as things. Indeed, I think that there is much more that is right about it than my title would suggest. But the thought that motivates this chapter is that this idea is often taken too far. Part of what I want to suggest is that not only is treating someone as a person compatible with treating them as a thing, but in fact, in many cases treating someone as the thing that they are is actually required in order to successfully engage with them as the person who they are. Both in philosophy and in life, I will be arguing, the fact that persons in fact are things places an important constraint on what it takes in order to most successfully relate to them as persons, and the mode of relating that is suggested by the strongest flights of Kantian fancy actually constitutes a particular kind of vice—a moral vice in our relations to others in ordinary life, and a philosophical vice in our relations to others through their philosophical work. If this much is right, moreover, then that leaves us with a burning question, to which I turn in the remainder of the paper. It is: Which are which? That is, what makes for the difference between the kinds of ways of recognizing and taking account for how persons are in fact things take us outside of the realm of persons, and the kinds that allow us to more successfully relate to one another as persons? My answer will be necessarily incomplete, but will attempt to explain a striking asymmetry.

1. Diminishing Returns Wednesday afternoon my wife, Maria, comes home from work, smiles and gives me a kiss, and compliments me on how the jasmine vine I’ve been cultivating is coming along. Suppose that I responded by wondering whether

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

97

she has found a coin in a vending machine return that afternoon. I take it that that would be a signal of a deep flaw in our relationship. It diminishes Maria for me to view her favorable opinion of my jasmine vine, which, after all, is still coming along only slowly, as the product of her environment, rather than as a sincere judgment about horticulture to be engaged with on its own terms—a compliment to be accepted with gratitude or demurred from, disagreed with or enthusiastically concurred with. The facts that Maria is a human being, that human beings are, as a matter of documented social science, more likely to compliment one another when in better moods, and that the surprise of finding a coin in a vending machine is a common way in which moods can perk up, do nothing to change our sense not only that there is something wrong with me, if this is how I relate to Maria, but that there is something suspicious about my even keeping this hypothesis in reserve, simply as part of my interpretive repertoire. Even to entertain this hypothesis, I contend, is to diminish Maria in some important way. As in life, so also in philosophy. Victor publishes his first paper in Ethics, a detailed attempt to articulate the way in which affirmative action policies wrong all successful members of minority groups by casting doubt on the merits of their accomplishments. Spencer reads Victor’s abstract and infers that Victor, who he knows to be a successful black philosopher, must be insecure about his own accomplishments. Like my interpretation of Maria, Spencer’s interpretation of Victor diminishes him—it takes the contributions of his paper outside the realm of rational engagement with his arguments and into the realm of psychological analysis and evidence about personal experience. The fact that Victor is a human being, and human beings are, we know, more likely to write about topics in which they have a particular interest, and that experience shows us that those who have experienced life belonging to historically disadvantaged racial groups are often more likely to have interest in the experiences of historically disadvantaged racial groups, do nothing to change, again, our sense that there is something wrong about Spencer’s engaging with Victor’s contribution in this way. We can say more about both Victor and Maria, about what is missing if Spencer and I interpret them in these ways. Gratitude, Strawson told us, is one of the participant attitudes—an attitude that helps to constitute our relationships with one another.² Though one may be grateful for all kinds of ² Somewhat surprisingly from a contemporary perspective in which following Wallace (1994) much attention has been paid to the attitudes of guilt, resentment, and indignation,

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

98

 

things, it doesn’t make sense to be grateful to a rock or a machine. It only makes sense to be grateful to a person with genuine agency. Or perhaps less contentiously, it only makes sense to be grateful to someone within the participant stance—as part of a collection of attitudes that together constitute taking them to be a someone, and not just a something. If that is right, then it gives us the tools for an explanation of why attributing Maria’s compliment to her finding a coin in a vending machine tends to crowd out being grateful to her for the compliment. To treat her compliment as something that is subject to non-rational prediction and explanation by background facts such as having found a coin in a vending machine is to treat her as a thing, rather than as a person. It diminishes her, treating her as a something, rather than as a someone. But that takes her outside of the participant stance, which in turn makes her compliment ineligible for gratitude. Similarly, we might say, rational persuasion is, if not an attitude exactly, nevertheless a move within the participant stance. Though we may change our minds in response to all kinds of events, we can only be persuaded by someone within the participant stance—the collection of attitudes that together constitute taking them to be a someone, and not simply a something. The contrast between changing one’s mind in a way that is stimulated by the sounds coming out of someone else’s mouth and being rationally persuaded by them is important; indeed, it is precisely this contrast that diminishes us when our interlocutor changes his mind only once he puts our point in his own words (something I’m afraid I do quite often).³ If this is right, it gives us an explanation of why attributing Victor’s thesis to his experience as a black American in the academy tends to crowd out being rationally persuaded by his arguments—which it does (though it is consistent with taking his experiences as evidence). Spencer’s attitude toward Victor takes him out of the participant stance, treating him as an object of prediction rather than engagement, and this is what makes his arguments ineligible to be exercises of rational persuasion. I haven’t yet argued for any of the claims in the last two paragraphs. But they are a package of attractive claims that seem to hang together, in cases

gratitude is actually Strawson’s first and one of his leading examples of a participant attitude. See especially Bero (2017) for discussion. ³ Compare Richard Holton’s (1994) similar distinction between trusting and coming to believe, also a case in Strawsonian terms.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

99

like these. The idea that gratitude is a participant attitude helps to explain the intuitive sense that my response to Maria’s compliment diminishes her, in some important sense, and in particular, treats her in a way that is paradigmatically morally objectionable within the context of an intimate interpersonal relationship, but whose moral objectionability extends in some attenuated way even to the context of interactions between strangers. Only a dick, as my undergraduate students might say, thinks about their wife in that way, and the regard that we expect within such relationships is worth aspiring to even among strangers. And similar points go for Spencer’s treatment of Victor. So much, I believe, Kant gets right, when he advises us to treat one another as legislators-in-themselves. It is this very insight by which Kant himself fails so dramatically in his correspondence with Maria von Herbert, as documented and explored so persuasively by Rae Langton in her wonderfully rich essay, “Duty and Desolation” (1992). According to Langton, when I come to accept a causal explanation of my neighbor’s behavior, I stop thinking of him as an agent, whose reasons, mysterious as they might be, I can in principle come to understand. My neighbour becomes a problem to be managed, an obstacle to be avoided, not a person to be argued with. He becomes just one more of the hazards of Elwood, along with the threat of the flooding canal. I have switched from the participant standpoint to what Strawson calls the objective. This is the attitude we have to things, items in the natural order, whose behaviour is explicable under causal laws, and manipulated if you know enough about them. To adopt it is to see a person as, perhaps, ‘an object of social policy; as a subject for what, in a wide range of sense, might be called treatment; [someone] to be managed or handled or cured or trained.’ [ . . . ] Strawson says that although the two attitudes are deeply opposed to one another, they don’t exclude each other. We can step back, and observe people as we observe the planets. We can observe a friend’s rising anger as if it were the rising of the canal waters—something to be feared and avoided, not to be understood and respected. We can cast on objective eye on our students, our friends, our lovers, and no doubt we often do, when the interactive stance proves too exhausting. Kant would say that when we do this, we fail to treat people as human, as agents in the kingdom of ends, as ends in themselves.

And Langton goes on, drawing on both Strawson (1962) and Korsgaard (1992), to document in compelling detail the ways in which Kant himself

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

100

 

takes the objective stance toward his correspondent, Maria von Herbert, describing her to another as “an ecstatical little lady” suffering from a “curious mental derangement,” which opens him up to using her as a mere means, passing her intimate correspondence on to another stranger without her consent. It is important not to gloss over the fact that both Langton and Strawson allow for, though they don’t emphasize, a kind of compatibility between the objective and participant stances. For Langton, you can go back and forth between treating someone with the objective and participant stances— perhaps when “exhausted.” And for Strawson, though they are “deeply opposed,” they don’t “exclude one another.” Indeed, I think that for Strawson this point is quite important. It is possible that these remarks are consistent with the point for which I want to argue in this chapter, but if so, this is I think too little emphasized, and correspondingly too often overlooked. When Langton, Strawson, and others identify the objective stance with prediction, subsumption under causal laws, and management, and then go on to suggest that the objective stance is in some way in tension with the participant stance, they are certainly at least suggesting a diagnosis of what would make it so diminishing, were I to respond to Maria’s compliment by interpreting it in causal terms. On this account, interpreting another’s behavior under causal laws precludes engagement with them under the participant stance, at least in this instance. If and when you interpret someone’s behavior in non-rational causal terms, this explanation goes, you take the objective stance toward them, and if and when you take the objective stance toward someone, you thereby preclude the participant stance.⁴ Call this two-pronged diagnosis of these first two cases the Kantian Diagnosis. Importantly, I do not claim that anyone—even Langton, Korsgaard, Strawson, or Kant—quite accepts this diagnosis. But there is something in the neighborhood of the Kantian Diagnosis that is deep and true, I believe—something that accounts for why it is so perennially tempting. And I ultimately want to suggest something very different from the Kantian Diagnosis: that because persons are in

⁴ In what follows, I will frequently refer to ‘causal’ interpretation, omitting the ‘non-rational’ qualification. I do not mean to imply that rational explanation is not a special case of causal explanation, but merely to contrast paradigmatic rational interpretation with other forms of causal explanation. Similarly, though I will refer throughout to ‘causal’ interpretation, everything that I say could also apply equally well to merely statistical prediction or to essentializing explanations. It would be an easy exercise to offer statistical or essentializing variants on all of my key examples.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

101

fact a kind of thing, it should not be surprising if the participant stance—the perspective par excellence from which we engage with one another as persons—does not preclude prediction, interpretation under a causal lens, or even, perhaps, at least some kinds of management.

2. Harder Cases The next week Maria comes home and snaps at me that my Jasmine vines are coming along too slowly. This time it occurs to me that she might have had a bad afternoon at work. After all, I know that she sometimes does have bad afternoons at work—her job as a cancer surgeon can be stressful, and sometimes she has to tell women with small children that they have advanced stage breast cancer. And frustrating encounters with colleagues are not completely unheard of. It would be no wonder if from time to time she took these work frustrations out on me; on the contrary, we know from the study of human behavior more generally that it is harder to be cordial when in a bad mood, and that other events can influence one’s mood. Indeed, this isn’t just my observation—in the past she has sometimes even apologized to me after the fact for taking work frustrations out on me. Of course, if every time Maria complained to me about something, I were to immediately speculate about her mood, that would be a sign of something wrong with our relationship. Likewise, there may be something problematic if it is my first interpretive hypothesis. But at some point—for example, if the vines about which she is complaining are the same ones she complimented last week, or if she won’t let it go even after I agree that we should have let the bougainvillea grow back instead, or if she easily transfers her complaints to other topics rather than engaging with me about the growth of the jasmine—it is not, at the very least, a flaw in our relationship, if I resort to the causal mode of interpretation of her behavior. Once more, philosophy is a microcosm for life in general. Fred publishes a thirty-page paper with a twelve-page digression developing a complicated set of new terminology that he uses in order to state his paper’s main thesis more precisely. But when Allison reads the paper, she concludes that the whole point of this section is that in order not to be subject to straightforward counterexamples, Fred’s thesis needs to appeal to the analytic/ synthetic distinction, but as a moral philosopher trained at Harvard in the 1960s, Fred’s internal Quine makes him too uncomfortable using those words, so what he has really done is to try to reinvent the analytic/synthetic

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

102

 

distinction in twelve pages while denying that that is what he is doing. Allison concludes that we should ignore this part of Fred’s paper, and so when she later references the main idea in Fred’s paper, she describes it in terms of analyticity even though she knows that Fred himself would disavow that description. Obviously if Allison treats it as a general rule that moral philosophers, or philosophers who studied at Harvard in the 1960s, are not to be interpreted as having interesting things to say about the analytic/synthetic distinction, then something has gone wrong. But that is not what Allison does. She explores the possible interpretive hypotheses about what makes the subtopic of these twelve complicated pages, initially reserves judgment about whether the right explanation rationalizes or merely predicts, and on the balance of the evidence, comes to the conclusion that the right explanation for these pages is Fred’s internal Quine. But she does not diminish Fred, or relegate him to the realm of things. One can imagine that even Fred, once it is pointed out to him or perhaps with a few years’ distance, comes to accept Allison’s hypothesis about what those pages were doing, and likewise screens them out when conceiving of the contribution of his own paper. Now, if the Kantian Diagnosis’s diagnosis of our earlier cases is on the right track, then interpreting someone’s behavior according to causal laws is sufficient for relating to them, at least in this instance, through what Strawson calls the “objective stance,” and relating through the objective stance precludes, at least in this instance, relating to them by way of the participant stance. But in these new cases, Allison and I both do interpret Fred and Maria through a causal lens. I interpret Maria’s behavior, ultimately, as better explained by an underlying bad mood than as a conversational move that requires a response. And Allison interprets a major portion of Fred’s paper which he obviously thinks is important to hang-ups about Quine rolling over in his grave, rather than as motivated by particular philosophical concerns that need to be addressed within the confines of his paper. So if the Kantian Diagnosis is right, then my reaction to Maria’s complaint and Allison’s diagnosis of Fred should be diminishing in the same way as my hypothetical reaction to Maria’s compliment and Spencer’s diagnosis of Victor. But there is nothing wrong, I say, about at least some cases of interpreting someone’s behavior through a causal, rather than rational, lens—even when alternative, rationalizing, interpretations are possible. Not only are such interpretations not wrong, moreover, they do not require leaving the participant stance. These cases are counterexamples, I claim, to the Kantian

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

103

Diagnosis. When I resort to the interpretive hypothesis that Maria had a bad day at work, I do not relinquish my interactions with her, for the time being, to the realm of prediction and control, and I certainly do not do so when I merely hold this hypothesis in reserve, as part of my interpretive repertoire. On the contrary, interpreting her complaint in this way is fully compatible with feeling gratitude to her for coming home from work early, or anger at her for forgetting my birthday, or both at the same time. Similarly, when Allison characterizes Fred’s thesis in terms of the analytic/synthetic distinction in order to credit him with an important insight that she means to draw on in her own work, she does not remove him from the participant stance or debar herself, even for a moment, from treating him as a serious interlocutor in the space of ideas. On the contrary, setting aside the alternative terminology and complications in the twelve-page digression are what allow her to engage more fruitfully with Fred’s real main contribution in his paper. Moreover, although I have used an interpersonal relationships example to illustrate the simultaneity of the participant stance and causal interpretation, and a philosophical example to illustrate the interconnectedness between the causal interpretation and identification of the agent’s contribution, the reverse kinds of examples are also easy to come by. When a six-year-old misgenders their sibling’s pronouns, it may be a provocation, but when an octogenarian begins to systematically misgender following a stroke, we are right to screen that out into the causal background, in order to latch onto what they really meant to say. These aren’t two attitudes that we take sideby-side, but rather, setting aside the pronoun genders as caused by the stroke is what allows us to see through them. And I can simultaneously marvel at a philosopher’s ingenuity in resolving the semantic paradoxes within classical logic while also recognizing that that is a paradigmatically Oxford thing to want so strongly to do. In all of these cases, I believe, it is either a poor or an impossible description of what is going on, to describe it as alternating between the participant and objective stances. The relationship between causal interpretation and the participant stance is too intimate to separate them in this way. Rational engagement cannot preclude recognition of how someone is situated in the space of causes, because we all know, as part of the background, that our actions are physically constituted by movements of our bodies, which are very much subject to causal laws. Since it should be common ground that persons are a special case of things, it should be no surprise that engaging with someone as a person cannot preclude the recognition that they are also a thing.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

104

 

But then what are we to make of the way in which such causal interpretation does seem to preclude participant attitudes toward the very features that are so interpreted? That is, why should my interpreting Maria’s complaint as caused by her bad mood or her rough work day preclude my resenting it, if it doesn’t push me out of the participant stance toward her? The only answer to this that I see, is that even if the participant stance is itself all or nothing, the participant attitudes themselves don’t come only as a package. It must be possible to exclude some aspects of someone’s behavior from the domain toward which one is open to participant attitudes, without excluding others. In some sense this should not be surprising; if excusing someone is, following Strawson, excepting that behavior from the domain of participant attitudes, we can surely excuse someone for some behaviors while simultaneously not excusing them for others—even for others performed at the very same time. Indeed, the case of Maria’s complaint about my gardening may fit exactly this mold. But I think the correct point here is really much more general and not concerned merely with excuses at all. When Allison interprets Fred’s paper, she is not excusing him for including a twelve-page digression from the paper’s main topic at all—indeed, she may actively resent him for not being better at monitoring to see that this is the sort of thing best delegated to the philosophers of language. When she interprets the reason this section appears in the paper in causal terms, she doesn’t excuse it, but rather simply sets it aside as neither here nor there for purposes of coming to grips with what Fred is actually contributing in his paper. She relegates it to the causal background in the very same way that we adjust visual perception to different lighting conditions. It is there, but it is something to be seen through, rather than visually inspected, and once you have properly incorporated it into the background—once your eyes have adjusted to the new conditions—it is effectively invisible, and you see objects as the same colors as before you took off your sunglasses.

3. Going Deeper So far I’ve been trying to draw our attention to a way in which, despite its manifest attractions as an explanation of a wide variety of wrongs, the Kantian Diagnosis is overly strong. Causal interpretation, I have been arguing, is not incompatible or even in the slightest tension with the participant stance, though causal interpretation of particular behaviors is

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

105

in tension with holding participant attitudes toward those very behaviors. There is nothing wrong with deploying such causal interpretation even with our dearest loved ones and closest interlocutors, and even if we can also step aside from such relationships and take the completely objective stance on our loved ones, as Strawson and Langton claim, we certainly do not need to do so in order to appreciate the very specific ways in which some of their behaviors are very much part of the causal order. But the path that we have taken toward reaching these conclusions leads me to think that we have also discovered something much stronger: that deploying causal interpretation as an interpretive tool is actually one of the key elements of successful interpersonal engagement. In order to most successfully engage with another specifically as a person—or put differently, in order to fully occupy the participant stance toward them—you must have the tool of causal interpretation at your disposal. Return, again, to the case of Allison and Fred. It is possible, a priori, that a twelve-page digression in the middle of a paper by a moral philosopher about the characterological implications of exploitative marketing practices (as let us suppose Fred’s thesis concerns) is the best place to look for special insight into the nature of the analytic/synthetic distinction, but it is unlikely. Fred didn’t write his paper in order to try to make progress on or replace the analytic synthetic distinction; he wrote it to make a contribution to our understanding of exploitative marketing practices. So it is by seeing through the complications introduced by Fred’s complicated and idiosyncratic terminology that Allison is able to see through to the real contributions of his paper and properly engage with him as a colleague in the space of ideas. And it is only because she is willing to set some of his words aside as part of the causal background, rather than part of his contribution, that she is in a position to do so. Suppose, in contrast, that Allison had set herself the constraint that in order to engage with Fred’s paper, she must not set any part of his contribution aside as merely caused, rather than rationalized. In that case, she would be treating it as an interpretive constraint that Fred’s alternative terminology must be about something other than the analytic/synthetic distinction. And that, in turn, would prevent her from accepting the best rationalizing account of why Fred’s thesis is qualified using this terminology. It would force her, in other words, to misunderstand Fred’s reasons. It is an unpromising start, in trying to engage with someone in the space of reasons, rather than the space of causes, if you start by playing by rules that prevent you from even recognizing what the other person’s reasons even are.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

106

 

Indeed, over-rationalizing the contributions of others in this way is a common philosophical vice, and particularly common among graduate students and others just entering the field. More than one PhD student I have worked with has struggled over what Michael Smith meant, in The Moral Problem (1994), by the claim that you have a reason to do something just in case your fully rational self would advise you to do it. The most salient interpretive problem arises from the fact that you can have reasons both for and against the same action—indeed, there are almost always reasons on both sides of every possible choice. But it seems unlikely that your fully rational self will always offer you conflicting advice. I once caught a student who had become so paralyzed by the thought that Smith’s view couldn’t be subject to such a simple counterexample that he had convinced himself that he had no idea what was going on in the book at all, and had become intellectually paralyzed for over a week. The answer, I think, is that when he wrote that book, Smith just was not clearly thinking about the difference between ‘a reason’ and ‘most reason.’ But this isn’t a rationalizing explanation—it is just the causal background against which rationalizing explanations are possible. Cases like this one are, I think, the rule, rather than the exception. If you have done philosophy for more than a few years and cannot identify dozens of places in which your earlier thinking was shaped by who you were talking to, which referees happened to be assigned to your paper submissions, or simple failures of imagination, then you are self-deceived. Moreover, you are failing to take advantage of the full resources available to you in order to remain open to the possibility of changing your mind—it is far easier to change your mind if you can identify non-rational contributing factors to the process that led to your earlier conclusion. We should all be open to recognizing non-rational influences on our own past thinking. And if we can do it to ourselves, we can have no principled right to complain, just because someone else does it to us—though we may still disagree about whether this interpretive tool is being applied correctly! What goes for philosophy also goes for life. When Deeksha begins obsessing about her weight, if Brandon doesn’t have available to him the hypothesis that she is merely displacing her anxiety about not hearing back about the jobs that she has applied to, then despite superficially talking about the same thing, Deeksha’s weight, the two of them are going to fail to connect on their concerns of fundamental importance. If Kat assumes that Wilbur’s sudden complaint about her tone of voice is an attempt to

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

107

open a conversation about the aesthetics of vocal qualities rather than a manifestation of ‘hanger’ better addressed by stopping for a snack, then the two of them are likely to spiral into worse misunderstandings. And if Leah doesn’t understand that what upsets Ginger about their change in plans was the way it reminds her of a childhood event, then she won’t understand how to reassure her, either. In all of these cases, I claim, agents need to understand the ways in which each other’s behavior is shaped by the causal background, in order to be able to identify each other’s rational concerns and contributions. The space of reasons doesn’t cross-cut the space of causation; it is located within the space of causation, and so if you can’t properly navigate the space of causation, then you simply can’t find it.

4. Interpretation So far, I have been arguing that despite its initial appeal, the Kantian Diagnosis is too strong. Not only is interpretation in merely causal, nonrationalizing, terms consistent with the participant stance, it is sometimes actually required, and the inability to resort to merely causal interpretation constitutes a distinctive vice both within philosophy and within interpersonal relationships, much more generally. What I would love to do, at this point, would be to offer a general answer as to when causal interpretation enables the participant stance, and when it disables it and diminishes. But as this is a very general question and I have only been considering a very limited diet of cases, anything I attempt to say about this will of necessity be vastly undermotivated, carrying implications about issues to which I have not paid attention only because I have not paid attention to them, rather than because I have reasons for thinking that things go in that way. Nevertheless, I am going to try to set out a suggestion that I find particularly fruitful, in thinking about the contrast between the Maria compliment/ complaint cases, and between Spencer and Allison. It is a suggestion that fits particularly well with the extended analogy that I have been drawing throughout this chapter between philosophy and life. I am well aware that it does not obviously encompass every case of diminishing interpretation or behavior, and I do not know whether that is more likely to be because it tracks only one important form which diminishment can take, or because it gestures toward the truth but only very imperfectly, or it turns out not to

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

108

 

be on the right track after all. But I am inclined to think that it is on the right track toward helping us to understand at least some of these cases, and will endeavor to gesture toward why. When Allison interprets Fred, I suggest, she does well by him as a philosophical interlocutor because she follows a form, broadly speaking, of the principle of charity. She endeavors to make the most of his article, to see it as making the most interesting and significant contribution that it can, compatibly with the textual constraints. She looks, that is, for the best interpretation. But it is important to be careful about what I mean by this. If there are three dancers, of whom one dances best, one is best to talk to, and one is best to have in a fight, we could pick out any of the three with the phrase ‘best dancer,’ depending on our purposes. But only one of them is best with respect to the internal standards governing dancing.⁵ Similarly, we may have many purposes for being interested in interpretations, and when these purposes are active, we may truly pick out many different interpretations with the phrase ‘best interpretation.’ But only one of these will be best by the internal standard governing interpretation. This is the interpretation that we seek, when we are charitable, in the very strictest sense, and which Allison seeks for Fred. What I suggest, about cases like this one, is not merely that many interpretations are possible and so as long as we are being charitable we do not wrong the author of a text, but rather that despite a considerable domain for reasonable disagreement, there are genuine facts of the matter about the contribution made by each interesting piece of philosophical work, and that charity is the appropriate principle because charity tracks truth. And charity can track truth only if the real contribution made by a particular piece of argumentative writing simply is the best, most charitable interpretation of that piece of writing—the one that allows its contribution to be greatest, given the background constraints. Call this the interpretive account of philosophical contributions.⁶ The interpretive account of philosophical contributions goes substantially further than the claim that we should exercise charity. But in addition to ⁵ ‘Best,’ as Geach (1956) points out, is an attributive adjective. The remarks in this paragraph are intended to allow for some of the important observations in Szabo (2001) without giving up the spirit of what animates Geach. ⁶ Here it is worth comparing Dworkin’s (1986) treatment of the law, though I do not mean to be endorsing any of Dworkin’s specific views about good interpretation. But if the law is itself a kind of collective contribution of all of the legislators and judges who contribute to its content, it should not be surprising, on my view, if the law does turn out to be identical to the best interpretation of the behavior of the legal system.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

109

making sense of why charity is not epistemically problematic, it also makes sense of why interpreting an author’s main point through a purely causal lens feels diminishing. When you think to yourself, ‘of course Schroeder has started thinking about wooly topics—he’s going through a mid-life crisis,’ instead of hearing my arguments out, that diminishes my contribution in the very real sense that it makes my contribution out to be less than it really is. This leads me to conjecture that something similar goes for persons. When I interpret Maria’s complaint through a causal lens, as the product of a bad mood caused by a difficult afternoon at work, this enables a more charitable interpretation of Maria, on the whole—an interpretation that makes her contribution to the world greater, better, or more significant. As with all interpretation, interpretation of persons is constrained by a text, and in the case of persons, that text is the totality of behavior. But though all text goes into the interpretation, not all text plays the same role within a good interpretation. Behavior while sleepwalking is still behavior, and maybe even reveals something about an agent’s subconscious anxieties or obsessions, but it is not properly attributable to the person in the fullest sense. So all behavior goes in, and self-avowals must be treated with deference, but even self-avowals can go wrong. Since the person you are is not a character constituted by your narrative of who you say you are, the best interpretation of this person is not constrained by self-avowals in the way interpretations of the actions of a fictional character are constrained by every letter of the text. On the contrary, if interpreting you is a matter of finding the best interpretation of your contribution to the world, then whenever your self-avowals are best explained in causal terms, even those self-avowals are not properly yours, and so they do not place a hard constraint on which behavior belongs as part of the story about what contribution you are actually making. In the case of philosophical writing, I went further, and suggested that charitable interpretation is not just a game that we play which permits us to find things in a text that are not really there, but rather can be thought of as a way of discovering the truth about an author’s contribution to the space of ideas, if only we accept the interpretive account of philosophical contributions, according to which there is a fact of the matter about what contribution is made by an argumentative piece of writing, and that fact is constituted by the best interpretation of what contribution it makes. Similarly, we can go further in the case of interpersonal interpretation, if we accept the interpretive

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

110

 

account of persons—the conjecture that a person is constituted by the best interpretation of what contribution their behavior makes to the world. As before, the interpretive account of persons goes substantially further than the mere claim that we should be charitable. But it doesn’t just make sense of why there is no epistemic defect in interpreting another’s behavior charitably; it also makes sense of why failing to do so feels depersonalizing. If I respond to Maria’s compliment by wondering whether she found a coin in a vending machine, that brings her down. It treats her as a mere thing. If persons simply are, at least in part, the best interpretation of the behavior of their bodies, then it is no wonder that this is so. Interpreting her behavior in this way is precisely failing to relate to her as a person.

5. Circumstantial Evidence The interpretive account of persons sounds wild. I acknowledge that—it turns my head. Yet the interpretive account of philosophical contributions is not, I think, wild, and the analogies between interpersonal interpretation and textual interpretation of philosophical work are, as I have been endeavoring to illustrate in this paper, fairly robust. It is also supported, I think, by a range of circumstantial evidence. The interpretive account fits with the asymmetries that we have observed between cases in which non-rationalizing causal interpretation of behavior is appropriate and inappropriate. The evidential bar for interpreting a compliment through a merely causal lens is higher than the evidential bar for so interpreting a complaint, because the range of circumstances in which a compliment helps to constitute a person’s contribution to a relationship or shared project or to the world more generally are wider than those in which a complaint does so. This is not a bare positive/negative asymmetry, or charitable interpretation would always attempt to screen out complaints. Many complaints do contribute in a positive way to shared projects—they constitute initial moves toward assistance at overcoming bad behavior, or toward reconciliation after frustration has created distance within a relationship, or a deserved reprimand for inappropriate behavior by another. But when complaints do not play one of these or similar roles, they can mar an agent’s contribution to a relationship or to the world. That is why the range of circumstances in which a compliment helps to constitute an agent’s positive contribution are wider than those for a complaint, and correspondingly why it is more often easier

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

111

to justify interpretation of a complaint in causal terms without diminishing the subject of interpretation. The interpretive account also makes sense of why it is that inappropriate deployment of causal resources in explaining someone’s behavior feels diminishing—and in particular why it seems depersonalizing, and to leave something out about recognizing who you are. If I too readily think of Maria’s compliment as something likely caused by an upbeat mood because she found a coin in a vending machine, then I am probably, in that instance, not in the business of charitable interpretation of her contribution at all. And if I am not, then I am not treating her as a person, but as some other kind of thing. So it should be no wonder, then, that it feels depersonalizing for me to treat her in this way. Moreover, the interpretive account even makes sense of why sincere attempts to engage with someone can feel depersonalizing or insulting. On this account, a person is constituted by the best interpretation of their behavior. But the question of which interpretation is best is an evaluative one, and like many evaluative questions, subject to a range of reasonable disagreement. It is a consequence of this that not only can two interpreters disagree about the best interpretation of a third person’s behavior, but it is even possible to reasonably disagree with a person herself about the best interpretation of her own behavior. Many such disagreements, of course, will not be reasonable, because a good interpretation is constrained to be identifiable as the person’s own contribution, and her own interpretation is one of the salient pieces of evidence about which contribution is her own. But if Fred at some later time can reasonably disagree with Fred at the time of authoring his article about the best interpretation of what is going on in his twelve-page digression, then similarly it must be possible for Allison to reasonably disagree with Fred. And similar points go for Maria at the time at which she is taking out her frustrations on the rate of growth of my jasmine vine. It will be no wonder if, in the heat of being still upset from her afternoon at work, Maria felt insulted or diminished if I refuse to engage with her about my problematic gardening—even if she later comes to accept my interpretation of the course of events. Our evaluative perspectives can be distorted by our emotions in ways that lend greater perceived importance to the proximal objects of our passions. So given that, it should be no wonder if Maria may, in the heat of the moment, reasonably disagree with me about the best interpretation of her behavior. And if she does, then if I reveal my interpretation to her through my words or actions, it will make sense for her to

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

112

 

feel either that I have failed to appreciate or acknowledge what she is genuinely trying to say to me, or that I am not even trying. I take it to be a distinctive virtue of the interpretive account, that it can make this kind of sense of the ways in which we can hurt one another even when we are well-meaning. Because interpretation is value-laden, interpretations will diverge whenever values diverge, and hence it will often be a vexed issue whether and to what extent we are relating to one another successfully within the participant stance. And when values diverge deeply, as in the social polarization that we increasingly observe in twentiethcentury North America, for example, that will place special tensions on the tenability of even sincere participants to recognize one another as sincerely engaged in the participant perspective. When such conditions arise, the interpretive account should lead us to expect, each participant will see themselves as sincerely engaged but the other as somehow detached, not sincere, or having little to contribute that is rational and not merely caused. Indeed, this leads to a diagnosis of many forms of distinctive relationship pathologies. Marissa, for example, places particular value on generosity and mentorship; growing up as a teenager, her parents became distracted by a protracted divorce and did not provide her with the advice and guidance that she craved through difficult and formative life choices. But she has done well for herself and takes pride in being able to share the financial fruits of her labor to help her nieces and nephews pursue opportunities that she would not have had as a child. Many of her nieces and nephews are grateful for this attention and the opportunities it provides, but Ingrid resents it. Ingrid is a student of Locke and Mill, more than of Rousseau; she values her personal autonomy and freedom, and feels that Aunt Marissa is abusing her role as benefactor in order to try to control Ingrid’s choices. Ingrid and Marissa’s relationship, I suggest, is completely normal. All interpersonal relationships involve clashes of values of one form or another, sometimes large, and sometimes small. And when these clashes happen, they engender clashes in each participant’s sincere interpretations of one another’s behavior. What Marissa interprets as generosity in her own behavior and values as one of her greatest contributions to her relationship with Ingrid, is seen instead by Ingrid as part of the causal background—a kind of compulsion Marissa faces to satisfy her own desire to be helpful, rather than listening to what Ingrid actually needs. This clash is why instead of feeling both gratitude and frustration, Ingrid has a hard time feeling gratitude at all. Likewise, the behavior that Ingrid interprets as an exercise

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

113

of personal autonomy that helps to constitute her personal contribution to the world is interpreted instead by Marissa as willful disregard for the wisdom of her elders. Marissa has a hard time acknowledging these choices as truly Ingrid’s own because she values independent measures of success over autonomy, and so she sees them as products of youthful indiscretion or the influence of problematic friends, rather than choices of which Ingrid is proud and for which she seeks recognition from her beloved aunt. The reason that pathologies like Marissa and Ingrid’s persist and create conflict within their relationship goes much deeper than the fact that they each exhibit behaviors that bother each other. It is that each values and identifies with their own behavior while the other sees it as something that is mere behavior that should be overcome. When we do things that bother our friends and partners but we don’t identify with, these are easy to overcome— we can and do adapt our behavior in order to make our interpersonal relationships work. It is the behaviors that our loved ones interpret as less central to us but which are more central to our own interpretation of our own lives that are so hard to give up, and form the basis of the kinds of persistent pathologies, like the one between Marissa and Ingrid, which in greater or lesser forms crop up in ongoing and meaningful human relationships.

6. Taking Stock We began with an attractive, broadly Kantian, Diagnosis that was motivated in order to make sense of a way in which we systematically wrong one another within close interpersonal relationships, and by extension, in human interaction more broadly. This wrong is constituted not by outward behavior or treatment, but by a failure to recognize one another as persons, or at least as the distinctive persons that we are. To the extent that we fail to recognize one another as persons, or as the distinctive persons that we are, we place obstacles to what we can achieve together. As attractive as it is, the Kantian Diagnosis is not quite true. Indeed, not only are there exceptions, in which it is permissible and does not wrong someone in any way to interpret their behavior through a causal lens even within the bounds of the participant stance, but this is actively required in order to most successfully relate to one another as persons. And this means that full adherence to the Kantian advice to interpret one another always as rational self-movers actually constitutes an opposing vice in interpersonal

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

114

 

relationships. It may be the less significant of two opposing vices, or as Aristotle might say, one for which we have no name. But it is a vice nevertheless. Success in relating to one another as persons requires recognizing the fact that we are also things. I have endeavored to lay out a brief sketch of one proposal—the interpretive account—for what kind of thing persons are, that might make sense of these other observations, but the main points that I wish to make transcend the details of this proposal. Whatever kind of thing persons are, I have suggested, it should be no surprise that success in relating to one another as persons requires recognizing the fact that we are also things—of course persons are a kind of thing. Since persons are a kind of thing, failure to appreciate the kind of thing that we are will of course constitute a failure to engage with us as persons.⁷

References Bero, Stephen (2017). “Responsibility and the Emotional Structure of Relationships.” PhD Dissertation, University of Southern California. Dworkin, Ronald (1986). Law’s Empire. Cambridge, MA: Harvard University Press. Geach, Peter (1956). “Good and Evil.” Analysis 17(2): 33–42. Holton, Richard (1994). “Deciding to Trust, Coming to Believe.” Australasian Journal of Philosophy 72(1): 63–76. Korsgaard, Christine (1992). “Creating the Kingdom of Ends: Responsibility and Reciprocity in Personal Relations.” Philosophical Perspectives 6: Ethics: 305–32. Korsgaard, Christine (2011). “Interacting With Animals.” In Tom Beauchamp and R. G. Frey, eds, The Oxford Handbook of Animal Ethics. Oxford: Oxford University Press, 91–118. Langton, Rae (1992). “Duty and Desolation.” Philosophy 67(262): 481–505. Smith, Michael (1994). The Moral Problem. Oxford: Basil Blackwell.

⁷ Special thanks to audiences at the Arizona Normative Ethics Workshop and the 4th Congreso de Posgrado del Instituto de Investigaciones Filosóficas at the Universidad Nacional Autónoma de México, and to Rima Basu, Steve Bero, Renee Bolinger, Alex Dietz, Carolina Flores, Micha Gläser, Fernando Rudy Hiller, Joe Horton, Nathan Howard, Aaron James, Robin Jeshion, Zoë Johnson King, Matt King, Berislav Marusic, Shyam Nair, Michael Nelson, Alexander Prescott-Couch, Grant Rozeboom, Sam Shpall, Elise Woodard, and many others.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

115

Strawson, Peter (1962). “Freedom and Resentment.” Proceedings of the British Academy 48: 1–25. Szabo, Zoltan (2001). “Adjectives in Context.” In I. Kenesei and R. M. Harnish, eds, Perspectives on Semantics, Pragmatics, and Discourse. Amsterdam: John Benjamins, 119–46. Wallace, Jay (1994). Responsibility and the Moral Sentiments. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

6 Fitting Love and Reasons for Loving Christopher Howard

1. Introduction Suppose Sam happens by a stray boot in the park and spontaneously comes to love it deeply and dearly. He wraps the boot in a blanket before placing it in the front seat of his car and, prior to driving off, makes sure it’s buckled in safely. Sam tends to his beloved boot for years thereafter, relishing his time with it, and dreading any time he must spend in its absence. We feel Sam’s attitude is perverse. We want to say that his love for the boot is unintelligible, irrational, or crazy—that it’s unfitting to feel the way that he does about it. We would feel the same way about someone in love with sour milk, Trump’s literary style, or the number 246. This suggests that love is the sort of attitude that can be fitting or unfitting to its object. And this, in turn, raises the question: What sorts of considerations make love fitting? This chapter defends the following, simple answer: What makes love fitting are the lovable qualities of the beloved. Just as a person’s admirable qualities make that person fitting to admire, and an outcome’s desirable qualities make that outcome fitting to desire, the lovable qualities of the beloved are what make the beloved fitting to love. This is the so-called “quality view.”¹ The quality view promises a highly natural account of what kinds of considerations can make love fitting. Moreover, it has a number of theoretical virtues. For example, it offers a simple and elegant explanation of our reaction to Sam’s bizarre love for his boot: Sam’s love is unfitting because it’s based either on lovable qualities his (unlovable) boot lacks, or on qualities of the boot that don’t make it lovable.

¹ Proponents of (some version of ) the quality view include Velleman (1999), Keller (2000), Abramson and Leite (2011), Jollimore (2011), Clausen (2016), and Protasi (2016). Christopher Howard, Fitting Love and Reasons for Loving In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0006

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

117

But despite its attractions, the quality view is often quickly dismissed in the recent literature on love. This is because the view seems to face a battery of embarrassing difficulties. If you have lovable qualities that make it fitting for me to love you, does that mean that it’s fitting for me to love anyone who has those qualities? (The problem of promiscuity.) That if someone is more lovable than you, it’s fitting for me to love her more? (The problem of trading up.) That if you lose your lovable qualities, it would be unfitting to love you? (The problem of inconstancy.) That it’s fitting for anyone who’s aware of your qualities to love you, too? (The problem of universality.)² And that’s just a sample. In addition to the “big four” listed above, the quality view seems to face several other difficulties. But in this chapter I’ll address each of the putative problems facing the quality view. The defense is intended to be comprehensive.

2. Preliminaries Love is complicated. I won’t try to give an exhaustive account of its nature. Occasionally I will make a claim about love’s essence, but when I do the claim will be uncontroversial. Is love an emotion (Brown 1987)? Does it constitutively involve a desire for union with the beloved (Nozick 1989)? Does it lie in the essence of love that if you love someone, then you take his interests as reasons to serve those interests (Frankfurt 2004)? I don’t know. I won’t take a stand. For now I assume only that love is properly understood as a distinctive kind of attitude that we might take toward persons or things. Most parties to the debate concerning what kinds of considerations can make love fitting focus only on personal love—i.e., love that takes a person as its object. These authors offer accounts of what makes it fitting to love romantic partners, family members, friends, colleagues, etc. But I won’t limit myself in this way here. Although most of my examples will concern personal love, the version of the quality view that I’ll defend in this chapter can easily be generalized to cover love for favored animals, football teams, films, and fiction. I take this to be a considerable advantage of the account. A few words about what the quality view is and is not. The quality view, as I’ll understand it, claims that the kinds of considerations that can make love fitting are specifically intrinsic qualities of the beloved, such as his kindness, ² I borrow the names for these problems from Jollimore (2011), and their manner of presentation from Setiya (2014).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

118

 

wit, empathy, and grace. Some authors explicitly or implicitly regard themselves (or are regarded by others) as quality theorists, but suggest that extrinsic qualities of the beloved can make love fitting (e.g., Keller 2000, Protasi 2016). But I find this suggestion unattractive. One reason is that quality theorists who hold that only intrinsic qualities of the beloved can make love fitting can easily accommodate the natural and plausible thought that what makes love fitting must be something about the beloved in her own right—qualities the beloved possesses in virtue of the way that she herself, and nothing else, is. But quality theorists who claim that extrinsic qualities can make love fitting can’t also claim this virtue. Still, this isn’t a hill worth dying on. None of the arguments I’ll advance in this paper contradict the claim that, in principle, certain extrinsic qualities could make love fitting.³ Now for what the quality view is not. The kind of quality view I’m concerned to defend does not imply that lovable qualities are the objects of love. When you tell your beloved that you love her on account of her remarkable poise, you aren’t thereby claiming that the object of your love is your beloved’s poise. Instead, what you’re claiming is that you love your beloved, and that her remarkable poise partly grounds (or makes pro tanto fitting) that love. The version of the quality view that I’ll defend in this chapter doesn’t contradict this plausible diagnosis. The kind of quality view that I’m interested in claims that the lovable qualities of your beloved are what make your love for her fitting. It doesn’t claim, implausibly, that these lovable qualities are the objects of your love. Next, some remarks about fittingness. The fittingness relation can be glossed as the relation in which a response stands to an object when the object merits—or is worthy of—that response. The quality view is thus the view that what makes someone merit love—what makes him worthy of it— are the lovable (intrinsic) qualities he possesses. To characterize fittingness a bit further, we can note some connections it bears to certain other normative properties. There’s wide consensus that something is valuable just in case it’s fitting to value, and that parallel equivalences hold between more specific value properties and the fittingness of certain, correspondingly specific ways

³ The “certain” qualifier is important: later, I’ll consider and reject “the relationship view” of what makes love fitting, according to which your beloved’s having the extrinsic property of being in a valuable relationship with you is what makes your love for him fitting (Hurka 1997, Kolodny 2003). But this view is also rejected (for different reasons) by even those “quality theorists” who allow that certain extrinsic qualities can make love fitting. As far as I can tell, the only disagreement between these latter theorists and relationship theorists concerns exactly which extrinsic qualities can make love fitting.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

119

of valuing. For example, it’s very plausible that someone is admirable just in case she’s fitting to admire, and that something is desirable just in case it’s fitting to desire. There’s also some consensus that fittingness is a permissive rather than a requiring notion. You’re not required to value everything that merits being valued (let alone to love everyone who’s worthy of love), though it is plausible that, at least other things equal, you’re permitted to do so.⁴ And lastly, it’s widely held that facts that make attitudes fitting provide (normative) reasons for those attitudes. For example, if the fact that you’re intelligent makes you fitting to admire, then, plausibly, that fact provides a reason to admire you. Similarly, if the fact that you’re empathetic makes you fitting to love, then that fact plausibly provides a reason to love you. So there seems to be necessary covariance, at least in one direction, between the fittingness relation and the reason relation. But beyond these reasonably clear connections, the question of how fittingness metaphysically relates to other normative properties lies fairly open. In a way, this will be our main topic of discussion.⁵ Finally, let me be clear that, aside from the motivation for the idea sketched at the outset of this chapter, I’ll provide no positive argument that love can be properly assessed as fitting or unfitting to its object. I’ll simply be assuming that this is so in what follows. This assumption is somewhat controversial, though increasingly less so. Most authors who argue that love can’t be fitting (or unfitting) appeal to either (or both) of two considerations: (1) that no existing account of what makes love fitting is satisfactory; and (2) that love is non-voluntary (or, as it’s sometimes more poetically put, “blind”), and so isn’t subject to the relevant kind of normative assessment.⁶ But responses to the latter consideration have been rehearsed (ad nauseum) in the literature on love,⁷ and the defense of the quality view I’ll provide in this chapter yields a counterexample to the former.

3. The Big Four I’ll start by addressing the “big four” objections to the quality view: the problems of promiscuity, trading up, inconstancy, and universality. I’ll ⁴ For discussion, see especially McHugh and Way (2016). ⁵ For historical discussions of the notion of fittingness, see, among others, Brentano (1889/ 2009) and Ewing (1948). For more recent discussions of the notion, see in particular D’Arms and Jacobson (2000a), McHugh and Way (2016), and Howard (2018, 2019). ⁶ See, for example, Frankfurt (1999). ⁷ One of the earliest responses is due to Kolodny (2003), the most recent to Naar (2017).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

120

 

suggest a unified solution to all four problems, as opposed to piecemeal solutions to each. Then, a bit later (in Section 5), I’ll draw on certain components of my proposed solution to the “big four,” in order to address the remaining difficulties (purportedly) faced by the quality view. Let’s begin by rehashing the “big four.” The quality view claims that what makes love fitting are the lovable qualities of the beloved. This seems to imply (1) that if your beloved has lovable qualities that make her fitting to love, then it’s fitting for you to love anyone who has those qualities (the problem of promiscuity); (2) that if someone is more lovable than your beloved, it’s fitting for you to love him more (the problem of trading up); (3) that if your beloved loses her lovable qualities, it would be unfitting for you to love her (the problem of inconstancy); and (4) that it’s fitting for anyone who’s aware of your beloved’s lovable qualities to love him, too (the problem of universality). My response is straightforward: I accept all of these implications. I accept, for example, that if newcomer Max really is more lovable than your beloved partner George, then it is fitting for you to love Max more than George; that if George loses all of his lovable qualities, it would be unfitting for you to love him; and that if George has lovable qualities that make him fitting to love, then it would be equally fitting for you to love absolutely anyone else who happens to have those very same qualities. But what I deny, crucially, is that from the fact that it would be fitting (unfitting) for you to love someone, it necessarily follows that you have sufficient reason (not) to do so. To claim otherwise would be to fall victim to what Justin D’Arms and Dan Jacobson (2000a) call the “conflation problem”: the problem of conflating the fact that it’s fitting to feel a certain way with the fact that there is sufficient (or decisive) reason for that feeling.⁸ I submit that this move mitigates the force of the “big four” considerably if not entirely. The quality theorist can happily accept, e.g., that if someone is more lovable than your beloved, then it would be fitting for you to love that person more, without being committed to the implausible claim that you thereby have sufficient all-thingsconsidered reason to do so.⁹ ⁸ D’Arms and Jacobson (2000a) introduce and discuss the conflation problem in a different context. They present the problem as a difficulty for certain forms of neo-sentimentalism. But as many have noted, the conflation problem isn’t a problem only for sentimentalists—it’s relevant to a variety of other debates, as well. For discussion, see Schroeder (2010). ⁹ I owe the spirit of this kind of response to the “big four” to the work of, and conversations with, Ginger Clausen. Clausen (2016) is the first to formulate explicitly the quality view as a view about fitting love (as opposed, for example, to a view about reasons for love), and her principal rationale for doing so is that this helps to guard against what D’Arms and Jacobson (2000b)

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

121

So quality theorists should deny that facts about which attitudes are fitting necessarily amount to facts about which attitudes there’s sufficient reason to have. Is this plausible? Yes. Although rejecting the relevant equivalence between fittingness and reasons isn’t uncontroversial, the position has many proponents (see, e.g., Rabinowicz and Rønnow-Rasmussen 2004, Danielsson and Olson 2007, Rosen 2015, Howard 2019). And the work the distinction does for quality theorists, in particular, suggests that they should follow suit. And notice that the move in question doesn’t commit the quality theorist to denying the very plausible claim that considerations that make love fitting provide genuine reasons for love. The quality theorist can maintain that the lovable qualities of people provide genuine “fit-related reasons” for loving them; she just has to hold that fit-related reasons for love aren’t the only reasons for love that there are. What other sorts of considerations might provide reasons for love? Notice that each of the purportedly implausible implications of the quality view we’ve considered so far seems implausible only if we assume the presence of a valuable, loving relationship. Take the problem of trading up. If you’re not in a valuable, loving relationship with George, then why shouldn’t you love Max more than George, if Max is more lovable? And this point generalizes to all the “big four” problems facing the quality view. Consider the problem of promiscuity. If Maxine has lovable qualities that make her fitting to love, and you’re not already in a valuable relationship with someone else, then why shouldn’t you love Maxine, or anyone else who shares her lovable features? Or take the problem of universality. If you’re not in a valuable relationship with Max, then why shouldn’t anyone who’s similarly unattached and aware of his lovable qualities love him, too? And consider finally the problem of inconstancy. If Maxine is truly unlovable, and you’re not already in a valuable, loving relationship with her, then, barring reasons of beneficence, why think you have any reason to love Maxine at all? What all this reveals, I think, is that valuable loving relationships have normative significance for love.¹⁰ In particular, since loving someone with

term the “moralistic fallacy” (which is, roughly, the “fallacy” committed by those guilty of the “conflation problem”). In the course of offering her defense of the quality view, however, Clausen does not, in my view, fully exploit all of the resources that this insightful specification affords. I highlight the central points of contrast between Clausen’s account and my own picture below (see note 11). Many thanks to Clausen for helpful conversations and correspondence. ¹⁰ That valuable relationships have normative significance for love shouldn’t be all that surprising. As I noted earlier, a number of writers have suggested as much (see note 3). But

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

122

 

whom you’re in a valuable relationship is necessary for the continued existence of that relationship, the presence of the relationship provides derivative “value-related” reasons for love. For example, in cases in which, intuitively, I shouldn’t trade up, what explains this is the presence of valuerelated reasons, which are (ultimately) provided by the value of my extant relationship. Similarly, in a situation in which my beloved loses her lovable qualities, but it seems that I shouldn’t cease loving her, this is due to my value-related reasons to continue loving her, which derive from our valuable relationship. And similar appeals to the presence of value-related reasons provided by valuable, loving relationships are, I think, sufficient to adequately explain the rest of the relevant data. The picture that emerges is this. The considerations that make love fitting are the lovable qualities of the beloved. These qualities provide genuine (fitrelated) reasons for love. But the fit-related reasons provided by a person’s lovable qualities aren’t necessarily the only reasons to love him. If you’re in a valuable, loving relationship with the relevant person, then, in addition to the fit-related reasons to love him, you have value-related reasons to love him, which derive from the value of your valuable, loving relationship. This is the picture of fitting love and reasons for loving I think the quality theorist should adopt.¹¹

the specific account of the normative significance of valuable relationships for love I’ll proceed to offer is, in fact, a new suggestion. ¹¹ Clausen (2016) imparts a picture that’s in some respects similar. In particular, she suggests that lovable qualities of the beloved are what make love fitting, and that the norms internal to valuable relationships provide reasons (not) to act in ways that (threaten) facilitate the existence of those relationships, and to care for the people with whom we have such relationships. So on Clausen’s picture, if someone is more lovable than your beloved, then it’s fitting for you to love that person more, but it doesn’t necessarily follow that you have decisive (or sufficient) reason to behave toward the relevant person in any particular way (p. 18), or to care about them in the way that you do your current beloved (p. 20). Perhaps the most important difference between the picture I’ve suggested and Clausen’s is that, on my view, instrumental value-related reasons that derive from valuable loving relationships favor continued love, whereas on Clausen’s picture, the reasons provided by the norms internal to relationships favor the various kinds of action and care that characterize one’s participation in the relevant type of relationship. I take this difference to be an advantage of my account. It wouldn’t be of much comfort to hear from your beloved that although she (fittingly) loves some newcomer more than she loves you, she’ll refrain from acting on (but not feeling) intense love for her new beloved. Also, Clausen’s favored picture doesn’t clearly afford the resources necessary to resolve certain of the remaining problems for the quality view. But in Section 5, I’ll show that, in contrast, the picture I’ve proposed does in fact provide quality theorists with the relevant resources.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

123

4. The Relationship View On the picture I’ve suggested, the normative significance of a valuable relationship for love consists in its being a source of value-related reasons for love. Valuable relationships are normatively significant for love, on my account, because they contribute to explanations of why your continued love for your beloved would be somehow valuable, or good. But Niko Kolodny (2003) offers a different account of the normative significance of valuable relationships for love. He claims that the fact that you have a valuable relationship with your beloved is what makes your love for her fitting. This is the “relationship view” of what makes love fitting.¹² It is the quality view’s main competitor. But in this section I’ll argue that the relationship view is false: the fact that your beloved has a valuable relationship with you does not (and indeed, cannot) make your love for her fitting. My argument against the relationship view is simple. Part of what it is to love someone is to value him non-instrumentally, for his own sake. If you don’t value someone non-instrumentally, you don’t love him at all. So whatever suffices to make love fitting must also make fitting noninstrumental valuation of the beloved. But a fact can make it fitting to value someone (or something) non-instrumentally only if that fact makes that person (or thing) non-instrumentally valuable. And while there are likely many facts about your beloved in virtue of which she’s valuable for her own sake, the fact that she has a valuable relationship with you is not plausibly among them. So the fact that your beloved has a valuable relationship with you doesn’t make it fitting to value her for her own sake. It follows that this fact can’t make your love for your beloved fitting.¹³

¹² As noted above (see note 3), Kolodny isn’t the only defender of the relationship view. I focus on Kolodny, and his version of the view, because he’s offered the most sustained defense of it. That said, the argument that I provide in this section generalizes to apply to any version of the relationship view. ¹³ Several authors have argued that a valuable loving relationship isn’t necessary for fitting love, focusing on cases of fitting love at first sight, and fitting unrequited love (Setiya 2014, Protasi 2016, Clausen 2016). The arguments offered by these authors seem to me to be successful, but the argument I’ve just offered above is in an important way stronger, since it purports to show that a valuable loving relationship isn’t sufficient for fitting love. This is a stronger conclusion, because the claim that a valuable relationship isn’t necessary for fitting love is compatible with the possibility that fitting love is “multiply realizable”—that a person’s (intrinsic) qualities or a valuable relationship with that person could make love for that person fitting. But the claim that a valuable loving relationship isn’t sufficient for fitting love entirely rules out the possibility that such a relationship could (ever) make love fitting.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

124

 

I take it to be obvious that loving someone essentially involves valuing him for his own sake and that it follows from this that whatever suffices to make love fitting must also make fitting non-instrumental valuation of the beloved. These claims seem uncontroversial. I also take it to be obvious that a given fact can make it fitting to value someone (or something) non-instrumentally only if that fact makes that person (or thing) non-instrumentally valuable. This much follows from the equivalence of being (non-instrumentally) valuable and being fitting to value (non-instrumentally). The above argument, then, turns entirely on the substantive, axiological premise that your beloved isn’t made non-instrumentally valuable by the fact that she has a valuable relationship with you. Two clarifying points regarding this claim are worth making. First, in advancing the claim in question, I’m not suggesting that your beloved lacks non-instrumental value, and thus that it’s unfitting to value her for her own sake. Rather, I’m claiming only that the particular fact that your beloved has a valuable relationship with you isn’t among the facts that make her non-instrumentally valuable, and so fitting to value as such. Second, in claiming that the relevant relational fact about your beloved isn’t one in virtue of which she’s of non-instrumental value, I’m not presupposing that it’s impossible for non-instrumental value to be grounded in a relational fact. That would be (close to) question-begging against the relationship view, given the plausibility of the essentialist claim that loving someone constitutively involves valuing him for his own sake. Moreover, while I’m not particularly persuaded by the proposal, a number of philosophers at least tentatively endorse the idea that relational facts do sometimes ground non-instrumental value (Kagan 1998, Rabinowicz and Rønnow-Rasmussen 2000). For example, Shelly Kagan suggests that the pen that Abraham Lincoln used to sign the Emancipation Proclamation might be valuable for its own sake in virtue of the (relational) fact that it was used by Lincoln to sign the Emancipation Proclamation. And Kolodny (2003) himself suggests that human remains have non-instrumental value in virtue of the relation they bear to the human whose remains they are. These claims may seem plausible. But my suggestion is that it’s not similarly plausible that your beloved has non-instrumental value in virtue of the particular relational fact that she has a valuable relationship with you. So I’m not necessarily opposed in principle to the idea that relational facts can ground non-instrumental value; instead, I’m (officially) opposed only to this very specific instance of the idea’s application. My claim that your beloved’s being in a valuable relationship with you isn’t a consideration in virtue of which he’s valuable for his own sake is,

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

125

I think, intuitively compelling. However, one might think that at least Kolodny’s version of the relationship view has the resources to resist it. To see why, we need to get a little more of Kolodny’s view on the table. So consider first that, according to Kolodny, loving relationships have noninstrumental value in addition to any purely instrumental value they might provide. Moreover, loving relationships are individuated, in part, by their participants: your loving relationship with your beloved wouldn’t be the particular relationship that it is if either you or your beloved didn’t figure in it. So on Kolodny’s picture, the fact that your beloved has a valuable loving relationship with you entails the fact that she’s a constitutive part of a non-instrumentally valuable whole, viz., your loving relationship. And perhaps one might think that this relational fact about your beloved does in fact make her non-instrumentally valuable, since one might suspect that, as a fully general matter, the fact that something is a constitutive part of a non-instrumentally valuable whole suffices to make that thing itself non-instrumentally valuable. But this suspicion would be mistaken. The fact that something is a constitutive part of a non-instrumentally valuable whole does not thereby make that thing non-instrumentally valuable. To borrow a nice example from Nicholas Stang (2012), consider a Beethoven piano sonata with a particularly dramatic rest. The rest is a constitutive part of the sonata, and the sonata (let’s assume) is a non-instrumentally valuable whole. But the rest itself has no non-instrumental value—it’s not fitting to value for its own sake. Rather the rest is fitting to value only in virtue of its constitutive contribution to a larger whole—the sonata—which is itself fitting to value for its own sake. And as a perfectly general matter: If something is fitting to value in virtue of its constitutive contribution to a (non-instrumentally) valuable whole, that thing isn’t fitting to value for its own sake in virtue of its contribution. So the fact that something is a constitutive part of a noninstrumentally valuable whole doesn’t suffice to make that thing, itself, non-instrumentally valuable. And so contrary to the suggestion we’re currently considering, the fact that your beloved is a constitutive part of your non-instrumentally valuable relationship is not plausibly a fact in virtue of which she has non-instrumental value. I can think of only one other way in which relationship theorists might try to defend the idea that your beloved’s being in a valuable relationship with you is among the considerations that make him non-instrumentally valuable. They might admit that the relevant relational fact about your beloved is not one in virtue of which he’s of agent-neutral non-instrumental value, but

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

126

 

maintain that it is one in virtue of which he’s of agent-relative noninstrumental value. In particular, relationship theorists might propose that the fact that your beloved has a valuable relationship with you makes him non-instrumentally valuable relative-to-you. They might then suggest that once the relevant relativization is made explicit, the axiological claim under consideration seems far more plausible. This suggestion is difficult (for me) to evaluate. That is because I have no grip on the notion of agent-relative value, and so no idea what it would mean to say that your beloved is “valuable for her own sake relative-to-you” in virtue of the fact that she has a valuable relationship with you. But I am not alone in, or plausibly at fault for, my befuddlement. For as many authors have observed, the notion of agent-relative value is of questionable intelligibility.¹⁴ Indeed, as Mark Schroeder (2007) has forcefully argued, it’s not at all clear what it means to say that something (or someone) is valuablerelative-to-you, if this is meant to be distinct from saying, for example, that that thing is valuable for you, or something you believe to be (agentneutrally) valuable.¹⁵ But with no independent grasp of the notion of agent-relative value, it would seem difficult if not impossible to evaluate the relative plausibility of the proposal that your beloved’s having a valuable relationship with you makes him valuable for his own sake relative-to-you. And at any rate, insofar as agent-relative value does look philosophically dubious, the relationship view’s commitment to its existence would make the view correspondingly less attractive. And notice that neither the account

¹⁴ See, inter alia, Dancy (2000), Schroeder (2007), and Zimmerman (2011). ¹⁵ Schroeder (2007) canvasses and rejects several suggestions for how to understand ‘agentrelative value.’ I think it’s fair to say that no proposal about how to understand the notion has gained wide acceptance. And note that for the relationship theorist’s purposes, not just any (reasonable) proposal will do. For in order to respond adequately to the current objection, the relationship theorist’s account of agent-relative value must entail a certain connection between agent-relative value and fittingness. In particular, it must predict that facts that make something non-instrumentally valuable relative-to-someone make it fitting for that person to value that thing non-instrumentally. Perhaps the most obvious way to ensure this result would be to claim that while agent-neutrally valuable things are things that are fitting for everyone to value, agentrelatively valuable things are things that are fitting for someone, but not everyone, to value. However, as Schroeder (2007) convincingly argues, this way of drawing the distinction can’t be right. For suppose some outcome O is overall agent-neutrally valuable, but that O is overall disvaluable “relative-to-you.” If agent-neutral and agent-relative value are distinct kinds of value, then this situation should be possible. But on the way of drawing the distinction we’re currently considering, it isn’t. For on the proposal under consideration, if an outcome is overall agent-neutrally valuable, then it’s fitting for everyone, including you, to value. So in the situation described, it’s fitting for you to value O. But by hypothesis, O is overall disvaluable relativeto-you, and so not an outcome that’s fitting for you to value. So this way of drawing the distinction can’t be right.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

127

of what makes love fitting that I’ve offered above, nor the supplementary account of the normative significance of valuable relationships with which I’ve proposed to pair it, comes with this commitment. Indeed, no component of the picture that I’ve suggested requires commitment to the existence of agent-relative value, or to any other, similarly suspect “philosopher’s trinket.” So even if relationship theorists opt for the response under consideration, they do so only at the cost of accruing significant theoretical disadvantage. And given the current dialectical situation, theoretical disadvantage isn’t something that relationship theorists can afford to accrue, since the principal motivation for their view over the quality view lies in its purported ability to avoid the (putative) difficulties faced by the latter; difficulties that, as I’m currently in the process of arguing, the quality view does not, in fact, face.¹⁶ That concludes my defense of the claim that your beloved’s being in a valuable relationship with you isn’t among the considerations that make her non-instrumentally valuable. I can think of no other way for relationship theorists to resist this claim beyond those discussed. Since whatever makes love fitting must also make fitting non-instrumental valuation of the beloved, and since a fact can’t make it fitting to value someone noninstrumentally unless that fact makes her non-instrumentally valuable, it follows from the claim in question that the fact that you have a valuable relationship with your beloved does not (and indeed, cannot) make your love for her fitting. The relationship view is false.

5. Modes of Love and the Amnesiac Biographer I’ve just argued that the relationship view offers an unsatisfactory account of the normative significance of valuable relationships for love, since it claims that the normative significance of a valuable relationship for love consists in its making love fitting, and valuable relationships do not (and cannot) make love fitting. And just prior to that, I argued that by accepting a certain picture of fitting love and reasons for loving, quality theorists can easily avoid the “big four” problems standardly raised against them. However, two (putative) problems for the quality view remain: the problems of “modes” and “amnesia” (Kolodny 2003). So in this section, I’ll draw on certain

¹⁶ That the main attraction of the relationship view is its (purported) ability to avoid the putative difficulties facing the quality view is evident from the discussion in Kolodny (2003).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

128

 

components of my solution to the “big four” in order to address these remaining difficulties. Consider first the problem of modes. You probably do—and more importantly for our purposes, ought to—love your romantic partner in a way that’s very different from the manner in which, e.g., his mother ought to love him. For the quality theorist, the problem of modes is the problem of explaining why this should be. If your partner’s qualities are what make it fitting for both you and his mother to love him, then why should the particular ways in which you and his mother love your partner differ so considerably? The picture of fitting love and reasons for loving I’ve proposed suggests a simple answer. What explains the fact that you ought to love your partner in a way that’s very different from the manner in which his mother ought to love him has nothing to do with the fit-related reasons provided by your partner’s lovable qualities, but rather everything to do with the value-related reasons provided by the value of your loving relationship. On the account that I’ve suggested, you have value-related reasons to love your partner because your continued love for him is necessary for the continued existence of your valuable, loving relationship. And the same goes for your partner’s mother. But crucially, the value-related reasons provided by your respective relationships don’t favor loving your partner in precisely the same ways. This is for the simple reason that the kind of love that would facilitate the continued existence of your relationship with your romantic partner is very different from the kind of love that would facilitate the continued existence of his relationship with his mother. And so the value-related reasons provided by your respective relationships favor loving your partner in rather different ways. And that is why, despite the fact that your partner’s lovable qualities are what make it fitting for both you and his mother to love him, the particular kinds of love that you and your partner’s mother ought to feel for your partner differ so considerably. So quality theorists who accept the picture of fitting love and reasons for loving that I’ve proposed have a simple and intuitive answer to the problem of modes.¹⁷ Consider next the problem of amnesia. The problem of amnesia, as Kolodny characterizes it, is the problem of explaining why the loss of certain

¹⁷ For a plausible alternative answer to this problem, which appeals to the idea that the norms internal to different types of relationship provide reasons to express one’s love (via action or care) in correspondingly different ways, and which doesn’t involve commitment to the existence of instrumental, value-related reasons for love, see Clausen (2016).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

129

memories extinguishes love. Assuming (as we have been) that love is an attitude we can hold for reasons, a natural explanation of this phenomenon is, as Kolodny suggests, that the amnesia victim loses cognitive access to his reasons for love. And if this is right, then, according to Kolodny, quality theorists would need to explain amnesia’s effect on love by appeal to its effect on the lover’s awareness of the beloved’s qualities. But in some cases, such an explanation seems unavailable. For example, suppose Tom is somehow highly knowledgeable about someone’s personal qualities despite the fact that he’s never met her—perhaps, to borrow Kolodny’s example, Tom is a biographer who’s produced a “strikingly intimate portrait” of his subject’s life and character via the detailed testimonies of her family and friends (2003, p. 141). And suppose also that, eventually, Tom meets the person whose qualities he knows so well, that the two of them fall in love, and marry. But years later, Tom suffers severe memory loss. He can recall everything that happened to him up until a year before his relationship with his wife began, but his memories of everything that happened after are lost—Tom’s forgotten his relationship with his wife. In this case, Kolodny claims that “we would not expect [Tom] to love [his wife], and indeed it is hard to see how he could” and that “[t]his is so even though [Tom’s] beliefs about [his wife’s] personal qualities, and his confidence in them, have changed only slightly, if at all” (2003, p. 141). So, Kolodny concludes, in this case, amnesia’s effect on love can’t be explained by appeal to its effect on the lover’s awareness of the beloved’s qualities. So for quality theorists, the problem of amnesia is the problem of explaining why, in the kind of case just described—viz., a case in which the lover forgets his relationship with the beloved, but not the beloved’s qualities—the lover’s love disappears. My response to the problem of amnesia proceeds in two steps. First, I think quality theorists should simply reject the demand to explain why forgetting a relationship with someone (whose qualities you continue to know) would extinguish your love for that person, since that seems like no fact at all. It is surely possible to love someone without believing you have (or had) a relationship with her. This happens all the time.¹⁸ So I think quality theorists should reject Kolodny’s claim that, despite remaining

¹⁸ As I noted earlier, it seems very plausible that a valuable loving relationship isn’t necessary for fitting love (see note 13). And if it isn’t, then, a fortiori, such a relationship isn’t necessary for love, either.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

130

 

fully aware of your beloved’s qualities, forgetting your relationship would extinguish your love.¹⁹ Still, something close to Kolodny’s claim does seem plausible. For while it’s not certain or even necessarily likely that your love for your beloved would disappear in the kind of case we’re considering, it does seem reasonable to expect that the character or intensity of the love that you feel for your beloved would change. Indeed, we can easily imagine Tom, our amnesiac biographer, now loving his wife in the way that he did prior to their relationship, or in its early stages, rather than in the way that he came to love her after years of marriage. But if this is the datum that quality theorists need to be able to explain, then those who accept the picture of fitting love and reasons for loving that I’ve suggested can easily deliver—and this is the second step of my response. For as I just argued, value-related reasons provided by valuable relationships favor loving people in particular ways, viz., in ways that facilitate the continued existence of the relevant relationship. So, prior to Tom’s memory loss, we can assume that he possessed value-related reasons provided by his valuable relationship to love his wife in a particular way, viz., in a way that would help to ensure that their relationship persists. But, as an effect of his memory loss, Tom lost those reasons, in the sense that the relevant reasons are no longer in his epistemic ken. And this, I think, easily suffices to explain why, as a result of his memory loss, the kind of love that Tom feels for his wife is different from the kind of love that he felt before: from Tom’s epistemic perspective, there simply are no considerations that favor loving the woman before his eyes as his wife. And so he doesn’t. So quality theorists who accept the picture of fitting love and reasons for loving that I’ve suggested have a natural and attractive answer to the problem of amnesia, too.

6. Still Weird and Trading Up Redux I’ve now given my answers to each of the problems the quality view is often claimed to face. I know of no other challenges to the view in the literature. But all of my answers to the problems putatively faced by the quality view turn on the specific picture of fitting love and reasons for loving I’ve ¹⁹ I’m not denying that it’s possible that your love for your beloved would disappear if you forgot your relationship with her. But this (mere) possibility isn’t a philosophically interesting explanandum.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

131

proposed for it. So I’ll conclude by considering some objections to that picture: what I call the “still weird” objection and the problem of trading up redux.

6.1 Still Weird In response to the “big four,” I argued that quality theorists should deny that facts about which attitudes are fitting necessarily amount to facts about which attitudes there is sufficient reason to have. Rejecting this equivalence allows quality theorists to deny that, necessarily, from the fact that it would be fitting for you to love someone, it follows that you have sufficient reason to love that person. And I claimed that this gambit mitigates the force of the “big four” considerably if not entirely. But isn’t it still weird to think that it could be fitting for you to love someone else’s partner (or child!) more than your own? And doesn’t it sound odd to suggest that it would be unfitting for you to love someone whom you’ve loved for a lifetime if she lost the qualities that make her lovable, perhaps through no fault of her own? Questions like these constitute the core of the “still weird” objection. They rhetorically suggest that denying an equivalence between fittingness and reasons does not, in fact, do much to mitigate the force of the “big four.” In response, I think that once we’re equipped with a proper understanding of the notion of fittingness, none of the purportedly peculiar implications of the kind of quality view I’ve been defending really do seem all that weird. Remember, the fittingness relation is the relation in which a response stands to an object when the object merits, or is worthy of, that response. So to say that it would be fitting for you to love someone else more than your beloved is to say only that that person merits or is worthy of more love than your beloved—i.e., that the relevant person is more lovable. And it would seem naïve and romantic (in a bad way) to deny the possibility that someone could be more lovable than your beloved—to believe that your beloved is the most lovable person there is. And the beauty, of course, is that he doesn’t have to be; your beloved has other things going for him that suffice to make it the case that you have sufficient reason to love him more than any (even more lovable) alternative(s)—viz., he figures essentially in your valuable, loving relationship. And for broadly similar reasons, I don’t find it problematic to hold that if your beloved loses his lovable qualities, it would be unfitting to love him. Indeed, one of the most wonderful things about the sort of love that we feel for those we love most is that it can (rightly) persist

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

132

 

even if its object isn’t worthy of it—i.e., even if there’s nothing about our beloveds in virtue of which they merit the intense love that we feel for them. And I think that similarly plausible things can be said to mitigate any residual, apparent weirdness that might seem to attach to certain implications of the kind of quality view that I’ve been defending in this chapter.

6.2 Trading Up Redux I’ve argued that quality theorists can answer the problem of trading up in the same way they can answer the rest of the “big four” problems standardly raised against them: by rejecting an equivalence between the fittingness relation and the reason relation, and by supplementing their account of what makes love fitting with the substantive normative claim that valuable, loving relationships provide (derivative) value-related reasons for love. In a situation in which, intuitively, I shouldn’t trade up, what explains this is the presence of value-related reasons, which are provided, ultimately, by the value of my current relationship. But suppose that someone is more lovable than my beloved and that my relationship with that person would be even better than my current one. Why shouldn’t I trade up then? This is the problem of trading up redux. I’ll offer two distinct and incompatible types of response to this problem. I think either type of response provides an adequate answer to the objection, but I want to remain neutral between them. I leave it to the reader to choose her own adventure. The first type of response to the problem of trading up redux is what I call the “diehard romantic” response. According to the diehard romantic, the problem of trading up redux falsely presupposes the possibility of legitimate trade-offs between loving relationships. This is because, according to the diehard romantic, loving relationships have a kind of value that precludes this possibility—in Kant’s terminology, they have a dignity, as opposed to a price. And as a consequence, the kinds of justified trade-offs the possibility of which the problem of trading up redux presupposes are in fact impossible: your loving relationship is irreplaceable in the sense that it could never justifiably be sacrificed in exchange for another relationship (or anything else for that matter) of equal or greater worth. A corollary is that the value-related reasons provided by a prospectively valuable relationship can’t possibly justify loving someone more than, or instead of, someone with whom you already have a valuable relationship.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

133

And so for the diehard romantic, the problem of trading up redux in fact presents no problem at all.²⁰ The diehard romantic response is not implausible. We certainly tend to treat and regard our most valued loving relationships as irreplaceable with potential alternatives. Indeed, we tend to value our loving relationships (and the loving relationships of others) in precisely the ways that Kant (and various Kantians) think that it’s fitting for us to value things with a dignity—we respect, esteem, and revere them. So perhaps the correct response to the problem of trading up redux is that of the die-hard romantic. Still, some will likely find the diehard romantic response objectionably saccharine. Is there really no possible situation in which it’s permissible to swap for someone who might be a “better match?” Or some might simply prefer a less axiologically extravagant proposal—one which doesn’t posit the special (spooky?) kind of value that Kant calls a “dignity.” So to those who might harbor such worries, let me recommend an alternative response to the problem of trading up redux: what I call the “austere realist” response. In contrast to the diehard romantic, the austere realist simply accepts the apparently problematic implication to which the problem of trading up redux points: if a new loving relationship with someone else would be discernably better than the loving relationship you currently have with your beloved, then, other things being equal, you have decisive allthings-considered reason to trade up. (Alternatively, if a new relationship would be equally valuable to your current one, then the austere realist accepts that, other things equal, you have sufficient all-things-considered reason to trade “sideways.”) Now, to the diehard romantic, the austere realist response will likely seem harsh. But there are things that can be said to mitigate its initial abrasiveness. First, other things are not always equal. In many cases, ending a valuable relationship with your beloved in favor of a new relationship with someone else will come with heavy costs. The pains of heartbreak and loss are real. And at least sometimes, the great disvalue associated with these pains (and other possible costs of ending a valuable loving relationship) will be enough to tip the scales in favor of “making it work” with your current beloved.

²⁰ Of course, in order to guarantee the result that you ought never to trade up, the diehard romantic would also need to commit to the claim that the fit-related reasons provided by a prospective partner’s lovable properties couldn’t possibly justify your sacrificing an extant loving relationship, either. This is no doubt a strong claim, but presumably not too strong for the tastes of a “diehard romantic.”

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

134

 

Second, the austere realist might point out that, in many cases, even if a new relationship would in fact be better than (or equally as good as) your current relationship, this fact will lie beyond your epistemic ken. This is for the simple reason that, no matter how lovable someone is, we can in general have little to no idea ex ante of how valuable a loving relationship with that person would be. Accordingly, in many cases in which a new relationship would in fact be better than your current one, it won’t be the case that you ought to trade up relative to your perspective, i.e., your epistemic position. Now, admittedly, the relevance of this last fact can vary depending on our view of the relationship between what you ought to do relative to your epistemic position and what you ought to do, full stop. If we hold, as is increasingly popular, that what you ought full stop to do just is what you ought to do relative to your epistemic position, then it follows from our general ignorance about the value of prospective relationships that, in many cases, even if a new relationship would in fact be better than your current one, it won’t be the case that you ought to trade up, full stop. This kind of ‘perspectivism’ about what we ought full stop to do thus pairs nicely with the austere realist response.²¹ For if perspectivism is true, then the austere realist response predicts that you ought to trade up only if, relative to your current epistemic position, a new relationship would be better than your current one, and the value of the new relationship would be at least sufficient to outweigh whatever disvalue might be associated with leaving your current relationship. This may be unromantic, but it doesn’t sound all that implausible. But suppose we reject perspectivism, and instead accept objectivism—the view that what you ought full stop to do is determined by all the facts, irrespective of your perspective.²² Still, our general ignorance about the value of prospective relationships remains relevant, albeit in a different way. For it’s widely held (particularly among objectivists) that the facts about what we ought (and may) do should be separated from the hypological facts, the facts about what we can be fittingly blamed or praised for doing. And very plausibly, the hypological facts (as they pertain to you) are determined by your perspective (perhaps along with other variables). So even if objectivism is true, and the facts of your situation make it the case that you ought to trade up, you might blamelessly fail to do so. Indeed, in some such cases you might be blameworthy for trading up, if doing so is forbidden, relative to your

²¹ Recent defenders of (some version of ) perspectivism include Broome (2013), Lord (2017), and Zimmerman (2014). ²² Recent defenders of objectivism include Graham (2010) and Thomson (2008).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

135

perspective. So although the exact relevance of our ignorance concerning the value of prospective relationships can differ depending on whether we’re objectivists or perspectivists, the fact of our ignorance in this regard seems to mitigate significantly the abrasiveness of the austere realist response, either way. So the austere realist response isn’t as harsh as it may initially seem. Still, it does predict that there are possible situations in which you have decisive (or sufficient) reason to trade up. But maybe this is exactly right. Perhaps, contrary to the diehard romantic, if it truly would be (objectively or perspectivally) better to be with someone other than your current beloved, then you really ought, all-things-considered, to be with him. Now, as I said, I want to remain officially neutral between the austere realist response and that of the diehard romantic. The right route isn’t clear to me. But as I also said, I think that either type of response is sufficient to answer the problem of trading up redux. So, I don’t think this problem speaks decisively against the proposal of this paper.

References Abramson, Kate and Adam Leite (2011). “Love as a Reactive Emotion.” Philosophical Quarterly 61: 673–99. Brentano, Franz (1889/2009). The Origin of Our Knowledge of Right and Wrong. Abingdon: Routledge. Broome, John (2013). Rationality Through Reasoning. Hoboken, NJ: WileyBlackwell. Brown, Robert (1987). Analyzing Love. Cambridge: Cambridge University Press. Clausen, Ginger (2016). “Love and Organic Unities.” Ph.D. Diss., University of Arizona. Dancy, Jonathan (2000). “Should We Pass the Buck?” Royal Institute of Philosophy Supplement 47: 159–73. Danielsson, Sven and Jonas Olson (2007). “Brentano and the Buck-Passers.” Mind 116: 511–22. D’Arms, Justin and Daniel Jacobson (2000a). “Sentiment and Value.” Ethics 110: 722–48. D’Arms, Justin and Daniel Jacobson (2000b). “The Moralistic Fallacy: On the ‘Appropriateness’ of Emotions.” Philosophy and Phenomenological Research 61: 65–90.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

136

 

Ewing, A. C. (1948). The Definition of Good. London: Routledge & Kegan Paul. Frankfurt, Harry G. (1999). Necessity, Volition, and Love. New York: Cambridge University Press. Frankfurt, Harry G. (2004). The Reasons of Love. Princeton, NJ: Princeton University Press. Graham, Peter (2010). “In Defense of Objectivism about Moral Obligation.” Ethics 121: 88–115. Howard, Christopher (2018). “Fittingness.” Philosophy Compass 13: e12542. Howard, Christopher (2019).“The Fundamentality of Fit.” In Oxford Studies in Metaethics, vol. 14, edited by Russ Shafer-Landau, ch. 10. Oxford: Oxford University Press. Hurka, Thomas (1997). “The Justification of National Partiality.” In The Morality of Nationalism, edited by Robert McKim and Jeff McMahan, 139–57. New York: Oxford University Press. Jollimore, Troy (2011). Love’s Vision. Princeton, NJ: Princeton University Press. Kagan, Shelly (1998). “Rethinking Intrinsic Value.” Journal of Ethics 2: 277–97. Kolodny, Niko (2003). “Love as Valuing a Relationship.” Philosophical Review 112: 135–89. Lord, Errol (2017). “What You’re Rationally Required to Do and What You Ought to Do (Are the Same Thing!).” Mind 126: 1109–54. McHugh, Connor and Jonathan Way (2016). “Fittingness First.” Ethics 126: 575–606. Naar, Hichem (2017). “Subject-Relative Reasons for Love.” Ratio 30: 197–214. Nozick, Robert (1989). “Love’s Bond.” In The Examined Life: Philosophical Meditations, 68–86. New York: Simon & Schuster. Protasi, Sara (2016). “Loving People for Who They Are (Even When They Don’t Love You Back).” Journal of Philosophy 24: 214–34. Rabinowicz, Wlodek and Toni Rønnow-Rasmussen (2000). “A Distinction in Value: Intrinsic and for its Own Sake.” Proceedings of the Aristotelian Society 100: 33–51. Rabinowicz, Wlodek and Toni Rønnow-Rasmussen (2004). “The Strike of the Demon: On Fitting Pro-Attitudes and Value.” Ethics 114: 391–423. Rosen, Gideon (2015). “The Alethic Conception of Moral Responsibility.” In The Nature of Moral Responsibility: New Essays, edited by Randolph Clarke, Michael McKenna, and Angela M. Smith, 65–88. Oxford: Oxford University Press. Schroeder, Mark (2007). “Teleology, Agent-Relative Value, and ‘Good’.” Ethics 117: 265–300.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

137

Schroeder, Mark (2010). “Value and the Right Kind of Reason.” In Oxford Studies in Metaethics, vol. 5, edited by Russ Shafer-Landau, 25–55. Oxford: Oxford University Press. Setiya, Kieran (2014). “Love and the Value of a Life.” Philosophical Review 123: 251–80. Stang, Nicholas (2012). “Artworks Are Not Valuable for Their Own Sake.” Journal of Aesthetics & Art Criticism 70: 271–80. Thomson, Judith Jarvis (2008). Normativity. Chicago, IL: Open Court Press. Velleman, J. David (1999). “Love as a Moral Emotion.” Ethics 109: 338–74. Zimmerman, Michael (2011). “Partiality and Intrinsic Value.” Mind 120: 447–83. Zimmerman, Michael (2014). Ignorance and Moral Obligation. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

7 “But I Voted for Him for Other Reasons!” Moral Permissibility and the Doctrine of Double Endorsement Alida Liberman

We are told that Confederate memorials should be decommissioned because some black people, in particular, equate them with racism, and racism makes them upset (understandably), and therefore the offending symbols have no place in our public spaces. . . . Tens of millions of patriotic Americans honor the South. The vast majority of Americans, moreover, equate Confederate memorials, and the Confederate flag, with Southern pride, not racism. Nicholas Waddy, “In Defense of Robert E. Lee.”¹ The sweeping argument against voting for [Roy] Moore (or Trump) rests on the mistaken view that in voting one is expressing one’s faith or moral convictions in their totality—identifying oneself with everything about a candidate. But a vote is not an expression of agreement with everything about a candidate or a candidate’s views. . . . Thus there’s no shame in voting for someone with whom you disagree, no matter how significant the disagreement, as long as you do so for the right reasons. Tully Borland, “Why Alabamians Should Vote for Roy Moore.”²

¹ See http://dailycaller.com/2017/08/16/in-defense-of-robert-e-lee/. ² See http://thefederalist.com/2017/11/30/alabamians-vote-roy-moore/. Alida Liberman, “But I Voted for Him for Other Reasons! ”: Moral Permissibility and the Doctrine of Double Endorsement In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0007

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

139

1. Introduction It is widely assumed that you can meaningfully separate the good and bad characteristics of a symbol, person, activity, or work of art, and can permissibly choose to support the good features while denouncing the bad features. Such appeals are often offered as a justification for (or an attempt to deflect blame away from) one’s support of something that is morally troubling. For example, as Nicholas Waddy notes in the epigraph above, someone who supports leaving Confederate monuments in place might distance themselves from the racism and dehumanizing violence of slavery while instead valorizing Southern pride. Voters frequently pick a candidate because they strongly support one of the candidate’s policies or positions even though they disagree with another of the candidate’s other positions. A fan of American football might enjoy the athleticism, momentum, and skill of the game while denouncing the traumatic brain injuries faced by many players. And many of us enjoy jamming out to a song in virtue of its great beat but despite its sexist lyrics, or watching Crimes and Misdemeanors for its bleak existentialist vision and cynical humor while condemning the alleged crimes and misdemeanors of the film’s director and star. My goal in this chapter is to refine and assess this kind of commonsense moral justification for supporting problematic people, projects, and political symbols. Although such justifications are frequently offered both in private and public conversations, there has been little explicit philosophical attention paid to this topic. What presumptions about moral responsibility for our endorsements underlie such appeals? Are these presumptions plausible, or generalizable? If we are to know how to properly assign moral responsibility to ourselves and others, we must ask whether and when we can meaningfully separate the good and bad features of these things, and nonculpably endorse the good aspects of them while distancing ourselves from the bad. To answer these questions, I propose an analogue of the Doctrine of Double Effect (or DDE) that applies not to our intentions leading to actions, but to our attitudes and endorsements. DDE states that “it is sometimes permissible to bring about as a foreseen but unintended side-effect of one’s action some harm it would have been impermissible to aim at as a means or as an end, all else being equal” (FitzPatrick 2012: 83; emphasis in the original). I argue that when certain conditions are met, it is morally permissible to directly endorse some object (such as a symbol, person, or artwork) in virtue of its morally positive (or aesthetically positive and morally neutral)

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

140

 

properties while standing against its morally negative properties—even though it would be morally impermissible to directly endorse those negative properties themselves.³ I call this the Doctrine of Double Endorsement, or DDN for short. I am not attempting to offer a positive argument for the claim that we can permissibly endorse something’s positive features while standing against its negative ones. Rather, I take this claim—which is widely appealed to by ordinary folk, yet which remains under-theorized—as a starting point, and assess whether it can be justified or if such appeals will always be misguided or hypocritical. My goal is to develop a plausible version of the claim, and I defend a list of necessary (although perhaps not sufficient) conditions that must be met for a permissible endorsement to occur. In Section 2, I offer an account of endorsement. In Section 3, I outline three constraints on when DDN applies: (1) separability: the good and bad features must not be inextricably linked; (2) proportionality: the positive value of the good features must be significantly greater than the negative value of the bad features; and (3) constrained choice: there must not be other things that the agent could endorse instead that share the same positive features but are not saddled with the negative ones. I apply these criteria to a variety of cases, and conclude with a summary of which cases are subject to DDN justifications.

2. An Account of Endorsement To endorse something is to support it or otherwise stand with it in a way that affirms that you value the endorsed thing. Because talk of endorsement in the philosophical literature is often connected to theories of autonomy (e.g., whether one must endorse one’s actions in order to act autonomously, and what such reflective endorsement consists in), I want to clarify that I am not addressing endorsement in this sense, and I will not be analyzing reflexive endorsements of our own actions. Rather, I am trying to capture the intuitive notion of supporting, affirming, or identifying with something outside of yourself.

³ To directly endorse X is for the primary object of your endorsement to be X. DDN claims that direct endorsement of some object X (say, a film) because of X’s good feature Y (say, its humor) can sometimes be permissible, even though X also contains bad feature Z (say, racist tropes) and direct endorsement of Z is impermissible.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

141

2.1 What Are Endorsements? What endorsing something involves will depend on the kind of thing it is and the ways in which we typically interact with that thing. Because we generally watch movies and TV to be entertained (or to appreciate art, or to expand our horizons, etc.) an endorsement of a movie or show will involve watching it for the sake of enjoyment (or art appreciation, or horizonexpansion, etc.). Because we vote for politicians to carry out the policies that we think are best, an endorsement of a politician will involve supporting their policies. At a minimum, endorsing requires only a pro-attitude, which can be conveyed in many ways. Re-tweets and Facebook “likes” are often paradigmatic examples of endorsement, since they are public proclamations that you stand with something. Another common form of endorsement is displaying symbols on your person or property (e.g., wearing a campaign button or putting a bumper sticker on your car). Endorsements are sometimes private—e.g., seeing news of protests and wishing for them to succeed—and sometimes publicly expressed or shared—e.g., driving past a protest and honking to signal your support of it. Endorsements vary in strength. Generally, more public and more permanent symbols signal stronger endorsement: someone who gets a “Feel the Bern” tattoo endorses Bernie Sanders for US president in a more robust way than does someone who pins a button to their jacket, while someone who flies a “Make America Great Again” flag in front of their home endorses Donald Trump’s candidacy in a stronger way than does someone who hangs a Trump poster on their bedroom wall. Fundamentally, endorsements depend not on what actions an agent performs, but on the agent’s mental state. This means that it will not always be clear to an outside observer whether an endorsement has actually occurred. Not every form of engagement with an object counts as a holistic endorsement of it. One might interact with an object ironically or jokingly without genuinely standing behind it; a group of friends who watch a cheesy old movie in order to riff on it à la Mystery Science Theatre 3000 do not thereby endorse the film they watch. Or one might interact with an object without freely choosing to do so, as does the student who watches Birth of a Nation because it is required for their film history class. Because endorsements signal our values, an interaction counts as an endorsement only if it is chosen. It is also possible to endorse something in a qualified way, qua a particular type of thing—for example, to view the Nazi film Triumph of the

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

142

 

Will and consider it an excellent example of political propaganda without supporting it as a work of art, or to endorse George H. W. Bush qua fashion icon for his colorful socks without supporting him as a politician or a person.⁴

2.2 What Kind of Endorsements are Subject to DDN? Endorsements can be complete or partial. Complete endorsement involves fully supporting every feature of the thing in question. DDN justifications apply only to partial endorsements, in which the agent stands with the endorsed thing in light of some of its features, but is opposed to it in light of others. The stance of the partially endorsing agent who is subject to a DDN justification must not be unitary: while they have decided all-thingsconsidered to watch a film, support a monument, or vote for a politician, they must not be fully wholehearted in their support. Fundamentally, DDN is used to assess cases of mere endorsement, in which taking or refraining from taking a pro-attitude towards something will not have any serious consequences. In many cases, endorsing X leads to good or bad outcomes that can be directly assessed in their own right. Buying The Cosby Show on DVD puts money in the pocket of a sexual abuser through residual payments; wearing a t-shirt with a Confederate flag on it in a predominately black neighborhood makes those who encounter you feel hurt, offended, or scared. In assessing whether it is ethical to purchase the DVD or wear the shirt, we should take considerations like these into account. If the consequences of an endorsement are obvious and are serious enough, they alone will determine whether it is permissible. However, there are other cases in which the consequences of an endorsement are extremely minor, or in which there are no effects at all. In these cases, we must appeal to DDN. For example, consider watching a film produced by Harvey Weinstein using a streaming service that pays distributors a flat licensing fee regardless of how many users stream the content, meaning that Weinstein receives no additional money. Or consider someone who joins an already very large protest or adds another signature to a

⁴ See https://www.npr.org/2018/12/01/672548335/president-george-h-w-bushs-choice-of-boldwhimsical-socks-made-him-a-style-icon.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

143

petition that already has over two million signatures.⁵ Even voting can be a form of mere endorsement, as someone who knows that their vote will not affect the election results or influence the elected official’s mandate to govern might vote simply to signal support of a candidate.⁶ (Those who are skeptical that most acts of voting are merely expressive can grant that at least some are: consider the California resident in line for the polls at 7:30 pm Pacific time who learns that a national election has already been called, but nevertheless casts a vote for their preferred candidate to signal their ongoing support.) Cases like these are mere endorsements, and DDN can help us figure out whether they are permissible. To clarify, I am not claiming that the things endorsed in DDN cases are not morally bad, or that they never have serious negative consequences. Rather, I am highlighting how DDN offers a tool for analyzing cases in which a particular act of endorsement does not contribute to these harmful effects.

2.3 Moral Responsibility for Endorsement When do X’s bad features outweigh or become so intertwined with X’s good features that it is no longer permissible to stand with X, all-thingsconsidered? To begin to answer this question, we need a general account of moral responsibility for endorsements. Endorsements reflect what we value: we stand with the things that we admire or care about. In many cases, these things are morally neutral, and a wide range of endorsements is permissible. I might endorse watching The Great British Bake-Off while you prefer to watch Chopped; on the assumption that both shows are morally permissible, neither of us is subject to moral praise or blame in virtue of our endorsement. In other cases, the mere act of endorsing can itself be good or bad, regardless of the particular thing endorsed. For example, rooting for a favored contestant to win a (morally neutral) reality competition show can be valuable because it gives the show more significance and interest for you, regardless of which particular (non-villainous) contestant you choose to ⁵ I’m presuming that these actions do not have any obvious effects on the outcomes: that the protest would be equally powerful without an additional protestor, and that the petition already has enough signatures that one more won’t matter. ⁶ See Brennan and Lomasky (1993) for an account of voting as a form of expressing oneself, and signaling or demonstrating commitment to certain political identities or causes.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

144

 

endorse. At other times, this is pernicious; ceteris paribus, rooting for any particular dog to win in a fight is blameworthy because dog-fighting is morally abhorrent. In the cases that I am most concerned with, the things we endorse have obvious good or bad features that make our endorsements of them a reflection of our moral priorities and characters. In general, we can be blamed for valuing or aligning ourselves with morally bad things (or bad features of things), and praised for valuing or aligning ourselves with morally good things (or good features of things). We can also be blamed for failing to properly condemn salient bad things; feeling neutral rather than outraged when you hear a report of child slave labor being used in a factory is morally troubling. We may also in some cases be blameworthy for failing to properly value salient good things; it is not appropriate to feel neutral toward (rather than admiring of ) the heroism of the person who rushes into a burning building to save a stranger’s baby. It follows that endorsement of a good thing (or non-endorsement of a salient bad thing) is praiseworthy, while endorsement of a bad thing (or non-endorsement of a salient good thing) is blameworthy. In simple cases, moral assessment of an endorsement bottoms out in moral assessment of the things we endorse: dog-fighting is morally wrong, and so it follows that endorsing a particular dog in the fight is wrong. RuPaul’s Drag Race is morally and aesthetically praiseworthy, and therefore so is endorsing the show and rooting for Asia O’Hara to win. However, DDN justifications are not needed in simple cases in which it is easy to make a holistic judgment about whether the endorsed thing in question is valuable. Rather, DDN is relevant for mixed cases, in which features A and B of X are valuable, while features C and D are disvaluable. The core philosophical question is about whether and when endorsing X on the basis of features A and B thereby makes you culpable for an endorsement of C and D as well.

3. Constraints on DDN My main task in the rest of this paper is to describe the three constraints that any endorsement must meet if DDN is to justify it: separability, proportionality, and constrained choice. First, though, a clarification about the agent’s attitudes is in order. The agent must be supporting the endorsed entity because of its positive features, and must be in some way opposed to or otherwise standing against its

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

145

obvious negative features. DDN will not morally justify the endorsements of agents who remain neutral about the negative features of the things they endorse; complacency about the negative features of something cannot be mitigated by endorsement of (what you take to be) its positive features. This is analogous to Michael Walzer’s revised version of DDE, which states that it is not enough to simply refrain from intending the bad effect, and requires that the agent intend both “that the ‘good’ be achieved” and “that the foreseeable evil be reduced as far as possible,” even at the cost of accepting minor risks of harm to oneself (1977: 155). Nor will DDN work as a justification for agents who are culpably ignorant of the negative features of the things they endorse. (Agents who are nonculpably ignorant of the negative features of the thing they endorse will not need to appeal to a DDN justification, because they will be straightforwardly exculpated from blame in virtue of their non-culpable ignorance.⁷) I cannot here offer a comprehensive account of when ignorance is culpable, but we have a rough sense of what culpable ignorance looks like—for example, ignorance about important and relevant topics that is motivated by wishful thinking, laziness, or apathy. To endorse something is to value it in a certain way, and it is morally careless to value something if you have not thoroughly enough investigated the extent to which it is in fact valuable. Ignorance is in some cases more than just careless: white supporters of Confederate monuments who remain unaware of their racist history (again, in spite of much popular discussion of this topic) are likely engaging in what Gail Pohlhaus, Jr. calls willful hermeneutical ignorance, in which “dominantly situated knowers refuse to acknowledge epistemic tools developed from the experienced world of those situated marginally” in a way that enables them to misinterpret or ignore important features of the world (2012: 715).⁸ They are blameworthy for remaining willfully ignorant of the monument’s racist symbolism in a self-motivated way that enables them to avoid grappling with their region’s difficult history and present struggle for racial justice. The person who publicly endorses a problematic thing must also do their best to ensure that the endorsement is not misconstrued by those they interact with as an endorsement of the negative features of the thing.

⁷ For arguments that non-culpable ignorance of your actions excuses you from moral responsibility, see Ginet (2000) and Mele (2011). ⁸ See also Collins (2000), Mills (2007), and Medina (2013).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

146

 

Now that we have an understanding of what sorts of attitudes agents must take toward the positive and negative features of the things they endorse, we can turn to the further constraints that permissible endorsements must meet.

3.1 Separability A common objection to the Doctrine of Double Effect is the “closeness” worry, or the concern that the intention/foresight distinction is unclear or can be arbitrarily drawn, leading DDE to morally license practically any behavior. As Dana Nelkin and Samuel Rickless put it in an article arguing that the closeness problem is intractable, “the problem is that an agent’s intention can always be identified in such a fine-grained way as to eliminate an intention to harm from almost any situation, including those that have been taken to be paradigmatic instances in which DDE applies to intended harm” (2015: 377).⁹ Since DDE is not the focus of this paper, I do not have to adjudicate whether the closeness problem is an insurmountable objection to it. But is there an analogue of the closeness worry for DDN? Not in all cases, for the positive and negative features of a thing (call them “P” and “N” for short) are often easily distinguishable, and we can draw a sharp line between them. However, there are cases in which this is not true, and in which P and N are inextricably intertwined or inseparably bound up with each other. In these instances, P and N are indeed too close: an endorsement of P cannot meaningfully be distinguished from an endorsement of N, and DDN will not apply. Call this the separability constraint. Because they exist in the same object, there is a sense in which the positive and negative features of a thing are always inseparable. I have in mind something more substantive than this: P and N are separable if and only if the shared social meaning of P does not include N. This goes beyond P and N contingently co-existing in the same person or object. Rather, X is part of the shared social meaning of Y if most folks in a society routinely and strongly associate X with Y, such that the social significance of one cannot be fully understood without the other. This could be for a variety of reasons. Perhaps X and Y are strongly culturally connected to each other; think of how pink or blue clothing on infants is generally taken to signal that the baby is a girl or

⁹ See also Foot (1978), Anscombe (1981), and Bennett (1995).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

147

boy. Or maybe it is widely known that X currently is or historically was used to support or enable Y. For example, consider how the swastika—originally a widespread and morally innocuous symbol of good fortune stemming from Buddhist, Hindu, and Jain religious iconography—was used to stand for Nazism and is currently deployed by anti-Semites and neo-Nazis.¹⁰ The shared social meaning of a word, symbol, or concept can vary across cultures (or even sub-cultures), although it is never a matter of idiosyncratic individual preference. Adorning a building with a swastika in East Asia has a very different social meaning than does adorning a building with a swastika in Europe or North America. This is so even if the person using the symbol intends it to be used in the religious sense; a Buddhist temple in contemporary Berlin or New York City that decorates their building with swastikas conveys the Nazi meaning rather than (or perhaps in addition to) the good fortune meaning. We do not always get to decide for ourselves what our symbols mean if our society assigns a contrary meaning to them. Confederate monuments offer a good illustration of inseparability.¹¹ Assume for the sake of argument that these monuments have positive features that are indeed worthy of endorsement, including the symbolic representation of the American South’s valiant attempt to preserve state sovereignty against outside coercive forces. Historically, though, antebellum Southern states struggled to maintain sovereignty so that they could uphold slavery. And many Americans (especially black Americans) see this as part of the social meaning of Confederate monuments. Moreover, most of the monuments were built during the Jim Crow era or during anti-civil rights backlash, and were explicitly aimed at upholding white supremacy.¹² States’ rights are valuable in general, but a state’s right to legalize the ownership and violent oppression of black people (or to push back against their ongoing

¹⁰ Social meanings do not just depend on historical connections. They also depend on the extent to which people are presently aware of this history. For example, the auto-maker Volkswagen was originally founded by Hitler. But because this is not very widely known, and because present-day Volkswagen has no affiliation with Nazism, the social meaning of the company does not include anything about Hitler. ¹¹ I focus on an individual’s support of a Confederate monument as a mere endorsement, assuming this support does not determine whether the monument remains standing. As Timmerman (2019) argues, Confederate monuments lead to serious harm to those who are aware of their racist origins and/or are reminded of a terrible time in US history by their presence, and should be removed for that reason. My concern is with moral assessment of individuals who endorse such monuments when their endorsements have no impact on whether the monument remains standing and harms others. ¹² See https://www.vox.com/identities/2017/8/16/16151252/confederate-statues-whitesupremacists.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

148

 

struggle for liberation) is not. The social meaning of the P feature of the monument (state sovereignty) is inextricably historically and culturally connected to the N feature (state-sponsored racism and oppression). An endorsement of a Confederate monument qua symbol of Southern state sovereignty cannot be conceptually distanced from an endorsement of the monument qua symbol of the South’s employing that sovereignty to uphold slavery and white supremacy.¹³ While there may be people who sincerely intend the Confederate flag to stand merely for states’ rights, they cannot unilaterally create a new shared social meaning for the flag. This does not mean that shared social meanings are immutable. For example, we might imagine a future in which the Confederate flag is reclaimed as a symbol by African Americans, perhaps in a way analogous to how the word ‘queer’ has been reclaimed by some in the LGBT+ community. If this occurred, the flag would no longer have an inseparable connection to white supremacy or racism, and endorsement of it would be permissible.¹⁴ Separability gives us a principled way to determine which monuments are irredeemably racist, and which have some racist connotations but are subject to DDN justifications. Consider the following tweet from Donald Trump, which attempts to argue that removal of Confederate monuments is a slippery slope leading to the removal of statues of any historical figure who was a slave owner: Sad to see the history and culture of our great country being ripped apart with the removal of our beautiful statues and monuments. You . . . can’t change history, but you can learn from it. Robert E Lee, Stonewall Jackson—who’s next, Washington, Jefferson? So foolish! Also . . . the beauty that is being taken out of our cities, towns and parks will be greatly missed and never able to be comparably replaced! (Donald J. Trump (@realDonaldTrump) August 17, 2017)¹⁵

¹³ Because they are government sponsored or sanctioned, Confederate monuments (as well as flags on government buildings) have a distinctive social meaning that is not shared by private display of similar flags or memorials: they signal a state’s lack of equal respect for all citizens. ¹⁴ Percival Everett’s short story “The Appropriation of Cultures” envisages this: the black protagonist decides to reclaim the Confederate flag as a symbol, and the flag becomes widely adopted as a symbol of black pride. As a result, it is stripped of its racist power, and is removed from the South Carolina statehouse (see Everett 2004). Thanks to Pierre Le Morvan for this reference. ¹⁵ Ellipses in the text represent where one tweet ends and the next one begins. See https:// twitter.com/realDonaldTrump/status/898169407213645824.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

149

There are positive reasons to endorse a statue of George Washington—he was a skillful general who led the American revolutionaries to victory, and he set a valuable democratic precedent of peacefully resigning from his position and refusing to remain in power. But Washington also owned many slaves. As president, he signed the first fugitive slave act, and he (unsuccessfully) pursued an escaped slave of his own for years.¹⁶ However, Washington’s good deeds can presumably be separated from his bad ones: for all we know, there is nothing about his generalship or his stepping down from office that entails, implies acceptance of, or is in the direct service of slavery. The social meaning of the two is—at least currently—not conceptually intertwined. The same is not true for Confederate monuments, which explains why one can permissibly endorse a statue of Washington but cannot permissibly endorse a Confederate monument in an ordinary setting (rather than, say, in a museum).¹⁷ Cases of inseparability also arise when assessing wonderful art made by people who do terrible things.¹⁸ Can the artist be meaningfully separated from the artwork, such that we can permissibly endorse the artwork without thereby supporting or standing behind the misdeeds of the artist? One way to answer this question is to assess whether the positive features of the artwork—say, the compelling storyline, or the humor, or the beautiful imagery—contain some implicit or explicit affirmation of the artist’s morally bad behavior. If so, the N features of the artwork are part of the shared social meaning of the P features of the artwork, and the artwork violates the separability constraint. For example, consider the films of Woody Allen, a man who has been accused of the sexual abuse of his young adopted daughter, and who at age 57 began a sexual relationship with the 20-year-old adopted daughter of his current romantic partner. Most of Allen’s comedies do not condone such behavior. As Kathleen Stock puts it in a blog post, “arguably (with the possible exception of Manhattan, which I avoid discussing for the sake of

¹⁶ See https://www.nytimes.com/2015/02/16/opinion/george-washington-slave-catcher.html. Washington ultimately freed his slaves, but only after his death. ¹⁷ It would be permissible to endorse Confederate monuments as historical artifacts were they deliberately moved to a suitable setting, in the way that Soviet statues were moved to an outdoor museum (Szoborpark) on the outskirts of Budapest after the fall of Communism in Hungary; see http://www.mementopark.hu/pages/conception/. ¹⁸ For further discussion of this topic, see the responses from philosophers (originally compiled by Daily Nous and updated in Aesthetics for Birds) about whether we can separate the art from the artist: https://aestheticsforbirds.com/2018/12/06/can-we-separate-the-artfrom-the-artist/.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

150

 

not ruining my argument), there is no serious implication in any of [Allen’s] films, intended to be believed by the viewer, that paedophilia is acceptable or in any way permissible. There is nothing about paedophilia whatsoever, in fact. Hence there is no genuine possibility of interpreting those films as inviting, via imagining, some endorsement of a counterfactual about it.”¹⁹ If this is correct, then DDN will permit us to endorse, say, Annie Hall on the basis of its positive aesthetic features while disavowing the negative moral behavior of the film’s star and creator. But this may not be not true of all of Allen’s work.²⁰ The reason Manhattan threatens to undermine Stock’s argument is that it stars Allen playing a 42-year-old character (Isaac) who is dating a 17-year-old high schooler (Tracy). While it is possible for art to depict morally bad behavior without thereby condoning it, this does not seem to be what is happening in Manhattan: ultimately, the overall narrative of the film seems to stand behind this behavior.²¹ Because the central storyline mirrors Allen’s history of engaging in inappropriate sexual relationships with much younger women,²² the social meaning of the film is too tightly bound up with Allen’s behavior to be separated from it. The film appears to condone the sort of objectionable behavior that its auteur is condemnable for, and this condoning storyline is at the film’s center. This ¹⁹ Stock (2017a). Stock is addressing what she calls the “puzzle of imaginative resistance,” or how readers/viewers resist engaging in some (but not all) forms of counterfactual imagining in fiction. Stock is investigating the extent to which Allen’s films seem to invite a counterfactual endorsement of pedophilia/ephebophilia. See also chapter 4 of Stock (2017b). ²⁰ My argument that separability is a constraint on DDN in no way depends on this particular analysis of Manhattan. I may be mistaken about the film’s aesthetic or moral virtues; perhaps the film’s core narrative is more morally nuanced than I make it out to be, and the good and bad features of Manhattan are in fact separable. If so, then endorsing Manhattan is permissible after all. This would show us not that the separability constraint is flawed, but that my understanding of Manhattan as a piece of art is flawed. ²¹ In the final scene, Isaac returns to Tracy after he has left her to (ultimately unsuccessfully) pursue a relationship with the intellectual Mary, realizing that he is most “relaxed” around Tracy and has the “nicest times” with her. Tracy is about to leave to study in London for six months, and Isaac asks her not to, stating “I just don’t want that thing about you that I like to change.” To me, this reads as an endorsement of their relationship, and of Tracy’s youthful innocence as the primary valuable thing about her. However, there may be more sympathetic interpretations of this scene; for example, see https://arcdigital.media/manhattan-revisitedfb2644239fd5. ²² Mariel Hemingway (Allen’s co-star in Manhattan, who was 16 when the film was made) has stated that Allen invited her to go to Paris with him when she was 18 in an attempt to seduce her; when she asked for separate bedrooms for the trip, he revoked the invitation; see https:// www.vanityfair.com/hollywood/2015/03/woody-allen-mariel-hemingway-manhattan. Actress Christina Engelhardt has also recently gone public about her sexual relationship with Allen, which started in the 1970s when she was 16 (and under the legal age of consent); see https:// www.hollywoodreporter.com/features/woody-allens-secret-teen-lover-manhattan-muse-speaks1169782.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

151

makes the shared social meaning of the storyline and the author’s bad actions inseparable: it becomes nearly impossible to view the film without a vivid awareness of that behavior.²³ This is not to say that the film has no redeeming aesthetic features; the Gershwin soundtrack is lovely, and the banter about art is witty. And Allen’s morally troubling behavior doesn’t impact how beautiful and fitting the music is, or how funny the jokes about art interpretation are. So it would be permissible to endorse the soundtrack, or the scene in which Isaac and Mary (Isaac’s best friend’s mistress, with whom Isaac falls in love) are introduced to each other at an art exhibit. But full-stop endorsement of the film as a whole—with the condoning storyline as a central part of it—violates the separability constraint on DDN and is impermissible.

3.2 Proportionality The traditional Doctrine of Double Effect is subject to a proportionality constraint; according to the New Catholic Encyclopedia, in DDE cases “the good effect must be sufficiently desirable to compensate for the allowing of the bad effect” (Connell 1967: 1021). A similar constraint applies to DDN: the positive features (P) must be good enough that they sufficiently compensate for the badness of the negative features (N). This has both a threshold component and a comparative component. First, N must rise above a minimum decency threshold, which means that N cannot be overwhelmingly bad. A novel or film that contains an explicit, vile, and non-ironically racist or sexist message is not worthy of endorsement, no matter how beautiful the prose is or how good the acting is. Second, we must make a comparative judgment between N and P: the good features must be significantly weightier than the bad features. This comparative judgment will not usually involve a straightforward weighing using a single metric: it is not clear how we could weigh, say, the democratic value of George Washington’s decision to peacefully step down from office against the moral horror of his owning slaves. Rather, the proportionality constraint on DDN is analogous to proportionality as ²³ Would DDN permit endorsement of Manhattan were it created by someone who had not engaged in Allen’s alleged or admitted sexual misdeeds? The narrative of the film seems to condone Isaac’s behavior independently of Allen’s off-screen actions. We would have to assess whether this condoning narrative is itself separable from the P features of the film, as well as whether it passes the proportionality and constrained choice criteria discussed below.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

152

 

understood by theories of criminal punishment. The harshness of a criminal sentence is not measurable against the badness of the crime on a single scale; the badness of an act of theft is not directly equivalent to the badness of six months in prison. Rather, proportionality demands that the levels of badness roughly match: minor crimes should receive minor punishments, while more serious crimes receive increasingly harsher punishments. Similarly (although conversely), the proportionality constraint on DDN demands that the worse N is, the better P must be to outweigh it. Such assessments can be made without weighing N and P on a single scale. How exactly we engage in these assessments is a difficult question that I cannot fully address here. But I can offer a speculative suggestion. If P and N are incommensurable, we should independently assess P and N according to their own scales. Roughly how bad is N, assessed as the type of thing N is (e.g., an act of sexual harassment, or a reminder of racial oppression)? And how good is P, assessed as the type of thing P is (e.g, a comedic film, or a reminder of historical struggles for states’ rights)? If P is very good—say, the funniest show you’ve ever seen—and N is not terribly bad—say, perpetuating a slightly offensive stereotype—then the object will pass the proportionality condition. If N is really awful, or if P isn’t that great, then the object will not pass the proportionality constraint. For example, it would not be proportionate to endorse a candidate who pledges to attain the minor good of preserving local historic landmarks in spite of the candidate’s terribly bad commitment to gutting the social safety net. But it might be proportionate to endorse a candidate who wants to slash social welfare programs if they are likely to accomplish the massive good of passing comprehensive gun control laws. To illustrate, consider a voter in the 2016 US presidential election who condemns the fact that Donald Trump has bragged about committing sexual assault, and is likely to enact policies that are harmful to women, racial minorities, and refugees. However, she reluctantly votes for Trump anyway because she prioritizes ending political corruption over every other cause, and believes that Trump will “drain the swamp” as he has promised to. We can grant that swamp-draining is a morally valuable goal. But it is not so valuable that it outweighs the terrible characteristics of Trump’s candidacy, including the racist, sexist, and xenophobic attitudes that are likely to be incorporated into harmful or discriminatory executive orders and policies; the hostility towards the media that seriously risks leading to a less free press; the abandoning of the normal standards of civic discourse that decreases opportunities for bipartisan progress, etc. The reluctant

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

153

single-issue voter cannot appeal to DDN to justify voting for Trump because the negative features that she disavows are so very bad that they swamp the positive feature that she stands behind. This is why the claim in the epigraph quotation about Roy Moore—that “there’s no shame in voting for someone with whom you disagree, no matter how significant the disagreement, as long as you do so for the right reasons”—is not true. If the disagreement is significant enough, even the “right reasons” cannot outweigh it. And what Roy Moore stands for is, presumably, so dreadful that it is not compensated for by any positive features of his candidacy.²⁴ Contrast this case with another voter, who reluctantly votes for Hillary Clinton in the same election. This voter endorses Clinton because of her support for universal health care, her plan to raise the minimum wage, and her political experience. But he stands against Clinton’s previous support for “three strikes” laws, her ties to corporate interests, and her lack of discretion in using a private email server. A DDN justification can apply in this case, because the positive features of Clinton’s campaign (arguably) outweigh the negatives: Clinton is far from a perfect candidate, but her good features outweigh the bad to a large enough extent that an endorsement of her is permissible, all-things-considered. The proportionality constraint will not always be easy to apply in practice. For example, consider someone who endorses NFL football. Are the positive features of the game—the skill and athleticism, and the camaraderie that comes from being part of a dedicated fan base—good enough to compensate for the game’s negative features (brain injuries, exploitation of college athletes, silencing of players who engage in political protests during games, etc.)? Answering this requires addressing substantive questions about how valuable the game’s positive features are and how disvaluable the game’s negative features are, and we can expect there to be reasonable disagreement about this. But this is a feature of the proportionality constraint rather than a bug, as it explains why some debates about the permissibility of particular

²⁴ Among other things, Moore (1) was removed from his office as Chief Justice in 2003 for failing to follow a court order to remove a Ten Commandments statue from the Alabama Supreme Court (http://www.cnn.com/2003/LAW/11/13/moore.tencommandments/); (2) stated in 2005 that he thinks homosexual conduct should be illegal (https://www.c-span.org/video/? c4682986/roy-moore-2005-bill-press); (3) made a public speech claiming that America was last “great at the time when families were united—even though we had slavery—they cared for one another” (http://www.latimes.com/politics/la-na-pol-alabama-senate-runoff-20170921-story. html); and (4) has been credibly accused of seeking out sexual relationships with teenagers half his age and in some cases sexually harassing or assaulting them (https://www.washingtonpost.com/ news/politics/wp/2017/11/16/timeline-the-accusations-against-roy-moore/).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

154

 

endorsements are so entrenched. Whether an endorsement satisfies the proportionality constraint will depend on the truth about the respective value and disvalue of the positive and negative features of the endorsed thing. In some cases, the answer will be obvious; in others, it will not be.

3.3 Constrained Choice The standard formulation of the Doctrine of Double Effect permits acting only under conditions of necessity, and demands that if one “could attain the good effect without the bad effect he should do so” (Connell 1967: 1021). A similar condition holds for DDN, which permits endorsement of X only when the agent is in a situation of constrained choice, in which there is no good option Y or Z that the agent could endorse instead that shares X’s P features but that doesn’t involve X’s N features. That is, endorsement of X is permissible only when there is no non-overly burdensome way to attain the benefits that P provides without endorsing something that includes N. The constraint is deliberately flexible about the degree of burdensomeness involved. If the negative features of something are extremely bad, the threshold for burdensomeness will be higher than in cases in which N is less bad. Constrained choices arise most obviously in situations in which it is impossible to attain some essential good without serious drawbacks. In such cases, you are justified in endorsing the option that is less bad, even if it would be impermissible to endorse this option if you had better choices available to you.²⁵ For example, suppose you are choosing which packaged lunch to eat at a conference. Your lunch break is too short to order delivery or leave the venue. The lunches have already been purchased by the conference organizers, and because the event is not recurring, your choice will have no impact on what food is ordered in the future. Assume that all of the choices available to you have serious negative features: the meat options involve harm to animals, and the vegetables are picked by migrant laborers who are paid unfairly low wages. Because you need to eat lunch, and your ²⁵ Remember that the threshold proportionality constraint still holds: if your vote will have zero effect on the outcome, you shouldn’t endorse Very Evil Person A over Even More Evil Person B but should instead abstain from voting. If the endorsement is not “mere”—if you expect that your vote really might affect which of two evil people is elected, then it is morally permissible (and perhaps required) to vote for the less evil person for consequentialist reasons of harm minimization.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

155

choices are constrained, you are justified in endorsing the less bad option— say, choosing unethically harvested salad instead of harmful and unsustainable beef, even if it would not be permissible to choose the salad were a harm-free option available to you. It will not always be easy to determine whether the endorsing agent faces a constrained choice. Consider fans of American football. Are there activities they could endorse instead that do not involve brain injuries to players but that do involve athleticism and the camaraderie of a fan community? Prima facie, it might seem that there are: one could follow baseball instead, or find a supportive community among Star Trek fans. But the goods of various sports are not interchangeable; the skills and athleticism of football are different than those of baseball. And many football fans find special value in carrying on family traditions of supporting particular teams, which alternate fan communities will not help them to maintain. Whether a football fan is in a situation of constrained choice will depend on substantive questions about what the goods of football are and whether they can be achieved through other means. We can ask similar questions about the endorsement of art that involves problematic content or is produced by problematic artists. Are there unique or distinctively valuable aesthetic goods in, say, the stand-up of Louis C.K. (who has admitted to non-consensually masturbating in front of women)²⁶ that cannot be attained through the stand-up routines of people who are not sexual harassers? I presume that the goods of individual works of art will not always be fungible: different comedians might offer distinctive aesthetic goods. But this will depend on what kind of good you’re seeking. If you simply want to laugh, there are plenty of other funny stand-up specials that are not created by sexual offenders. Contrast this with another use of Louis C.K.’s stand-up. Before C.K.’s sexual misconduct was public, I showed a clip from one of his routines in my ethics class to illustrate Peter Singer’s arguments about famine relief. (C.K. says his life is “really evil” because he drives an Infiniti when he could trade in his car for a Honda and use the leftover money to feed starving children instead, concluding “every day, I make them die with my car.”)²⁷ So far as I know, no other comedians have incisive and illuminating bits about how affluent people are morally responsible for failing to alleviate global poverty.

²⁶ See https://www.nytimes.com/2017/11/09/arts/television/louis-ck-sexual-misconduct.html. ²⁷ See Jason Brennan’s take on the comparison here: https://fee.org/articles/luxury-andlouis-cks-really-evil-life.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

156

 

If I wish to illustrate Singer’s arguments through a comedy clip, there is no other way to do so. For the time being, I’ve decided to stop using the clip. But it seems that its use in an ethics course is subject to a DDN justification due to constrained choice (at least, so long as the clip is carefully framed so as to avoid giving the appearance of endorsing C.K.’s bad behavior). Harder to assess is whether the proportionality constraint is met in cases of constrained choice, for we must also consider the moral value of the constrained choice itself: not every end is valuable enough to merit being attained only via endorsement of something with serious negative features. The athleticism of football is valuable, but is probably not worth the brain injuries players receive. This parallels how we must assess the moral permissibility of the end aimed at in a DDE case. Consider the tactical bomber who intends to destroy an enemy munitions factory while merely foreseeing that this will result in a number of civilian casualties. We must ask whether a tactical bomber should be bombing the munitions factory in the first place, in addition to assessing the permissibility of the act given his attitude toward the civilian casualties that will result. Constrained choices might also intersect with separability. Using comedy to illustrate Singer is memorable, and it helps students let their guard down about a challenging topic. But at the height of the #MeToo movement— which criticizes the way in which men in positions of power engage in sexual harassment and assault with impunity—endorsing the work of a harasser has a distinctive social meaning that it might not have in another time or place. The sexual misdeeds of famous actors and comedians are extremely salient in this cultural moment. In this context, an endorsement of Louis C.K.’s clip risks violating the separability constraint: maybe the social meaning of engaging with C.K.’s work in any way after #MeToo involves a problematic ignoring of bad behavior by powerful men. Perhaps in a future in which harassers are routinely held accountable, this social meaning will shift, and a successful DDN justification will apply to use of the clip. Limited entertainment options in a patriarchal society rife with sexist art might also lead to constrained choices for consumers, who may be unable to find an abundance of great movies, books, and songs that are free from problematic features. As Julie Bindel (2015) argues: If feminists—especially those of us who prioritise the campaign to end male violence against women—restricted themselves to entertainment that was perfectly non-sexist, perfectly pure, we would be pretty miserable, and have very little to watch or listen to. . . . As a feminist, under the system of

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

157

patriarchy, to live a life without contradiction means I would have to wall myself off from the wider world of music, film and literature, something I’m not prepared to do.

Someone who enjoys the drama and vicarious thrills of crime procedural shows may not be able to find a series that does not occasionally depict violence against women in a sexualized or glorified manner. If so, a DDN justification can apply to watching these shows in virtue of their drama and despite their sexism—though we must also consider proportionality, and ask whether the end of enjoying crime procedural dramas is valuable enough to be worth pursuing despite this negative feature. A final example of constrained choice involves the lack of representation of women of color in film. The “Bechdel test” (inspired by a cartoon by artist Alison Bechdel) proposes assessing gender representation in a film by ascertaining whether it involves two named women characters who talk to each other at some point during the film about something other than a man.²⁸ Some people criticized the 2013 film Pacific Rim because it failed the Bechdel test.²⁹ The movie—which is about fighting interdimensional monsters using co-piloted robots—features only one significant female character: Mako Mori, a skilled pilot who teams up with her adopted father to avenge her murdered family. Mako is a fully developed character with an interesting narrative that is central to a plot that is not primarily about romance. Tumblr user spider-xan responded to criticism of Pacific Rim as follows: It’s really easy to throw away a film because of [the Bechdel] test . . . if you’re a white woman and can easily find other films with white women who look like you and represent you . . . But as an East Asian woman, someone like Mako—a well-written Japanese woman who is informed by her culture without being solely defined by it, without being a racial stereotype, and gets to carry the film and have character development— almost NEVER comes along in mainstream Western media. And honestly—someone like her will probably not appear again for a very long time. So you’ll understand why I can’t throw her and the entire film ²⁸ The test is not meant to assess whether a movie counts as “feminist,” or to determine whether it is permissible to endorse a film all-things-considered (although it is sometimes used in this way). Rather, it is a marker of unequal gender representation in film, which can be used to assess one negative feature a film might possess (lack of broad representation of women). See https://en.wikipedia.org/wiki/Bechdel_test. ²⁹ For example, see https://www.vulture.com/2013/07/does-pacific-rim-have-a-woman-prob lem.html.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

158

 

away as meaning nothing in terms of representation—because she’s all I really have right now.³⁰

The post inspired the “Mako Mori test,” which proposes assessing intersectional gender representation in a film by assessing whether there is: (1) at least one female character of color who (2) has her own narrative arc that (3) is not about supporting a man’s story.³¹ Under-representation of women of color in film leads to a situation of constrained choice. Pacific Rim has a negative feature: it fails the Bechdel test. But it also has an important positive feature: it passes the Mako Mori test. Because it is exceptionally difficult to find blockbuster Western films that pass both tests, DDN will justify an endorsement of Pacific Rim. Contrast this with the “sexy lamp test,” coined by comics writer Kelly Sue DeConnick, who says “if you can remove a female character from your plot and replace her with a sexy lamp and your story still works, you’re a hack.”³² The test criticizes stories that treat female characters in such objectifying and underdeveloped ways that they could literally be replaced by inanimate objects. A film that fails the sexy lamp test may have other virtues: it may be entertaining, or feature beautiful cinematography, etc. Luckily, we are not in a situation of constrained choice with regard to films that fail the sexy lamp test. It is not overly burdensome to find films that are entertaining with great cinematography and that portray women in a way that gives them at least minimal agency, which means that such films are not subject to a DDN justification.

4. Conclusion I have argued that it can be morally permissible to endorse something in virtue of its positive features while disavowing its negative ones, so long as the good and bad features are meaningfully separable, the bad does not disproportionately outweigh the good, and the agent is in a situation of constrained choice. The table summarizes these results as they apply to the examples I have discussed.

³⁰ See http://spider-xan.tumblr.com/post/58305944138/also-i-was-thinking-more-about-whywhite-women. ³¹ See https://www.dailydot.com/parsec/fandom/mako-mori-test-bechdel-pacific-rim/. ³² DeConnick mentions the test in print here: http://www.ign.com/articles/2013/06/20/kellysue-deconnick-talks-captain-marvel-pretty-deadly-and-the-sexy-lamp-test.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

159

Type of case

Successful DDN justification of:

Unsuccessful DDN justification of:

The unsuccessful case involves failure of:

Voting

Voting for Hillary Clinton in 2016 Admiring statues of George Washington

Voting for Donald Trump in 2016 Admiring Confederate monuments/flags Watching Manhattan Watching Louis C.K.’s standup special on Netflix Watching movies that fail the “sexy lamp test”

Proportionality

Monuments

Art

Art

Watching Annie Hall Using Louis C.K. to illustrate Singer (in some contexts) Watching movies that fail the “Bechdel test” but pass the “Mako Mori” test

Separability

Separability Constrained choice Constrained choice

The Doctrine of Double Endorsement refines the intuitive justifications of partial endorsements that are often made by pundits, politicians, and everyday folk, and gives us a principled way to distinguish when those endorsements are morally permissible. Ultimately, voting for a problematic candidate for “other reasons” can be morally permissible—but it is not in many of the cases in which people attempt to offer such justifications.³³

Works Cited Anscombe, G. E. M. 1981. Ethics, Religion, and Politics. University of Minnesota Press. Bennett, Jonathan. 1995. The Act Itself. Oxford University Press. Bindel, Julie. 2015. “I am a Snoop Dogg Fan. That Doesn’t Make Me Less of a Feminist.” May 28, The Guardian. https://www.theguardian.com/com mentisfree/2015/may/28/snoop-dogg-feminist-sexist-lyrics. Brennan, Geoffrey and Loren Lomasky. 1993. Democracy and Decision: The Pure Theory of Electoral Preference. Cambridge University Press. ³³ Thanks to audiences at the 2018 Arizona Workshop in Normative Ethics and at the College of New Jersey in 2018, as well as to Josh Crabill, David Shoemaker, Julie Tannenbaum, and two anonymous referees for helpful discussion of this paper.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

160

 

Collins, Patricia Hill. 2000. Black Feminist Thought: Knowledge, Consciousness, and the Politics of Empowerment. Routledge. Connell, F. J. 1967. “Double Effect, Principle of,” New Catholic Encyclopedia (Volume 4). McGraw-Hill: 1020–2. Everett, Percival. 2004. Damned If I Do. Graywolf Press. FitzPatrick, William. 2012. “The Doctrine of Double Effect: Intention and Permissibility.” Philosophy Compass 7.3: 183–96. Foot, Philippa. 1978. “The Problem of Abortion and the Doctrine of Double Effect.” In Virtues and Vices. University of California Press: ch. 2. Ginet, Carl. 2000. “The Epistemic Requirements for Moral Responsibility.” Noûs 34.14: 267–77. Medina, José. 2013. The Epistemology of Ignorance: Gender and Racial Oppression, Epistemic Injustice, and Resistant Imaginations. Oxford University Press. Mele, Alfred. 2011. “Moral Responsibility for Actions: Epistemic and Freedom Conditions.” Philosophical Explorations 13.2: 101–11. Mills, Charles. 2007. “White Ignorance.” In Race and Epistemologies of Ignorance, ed. Shannon Sullivan and Nancy Tuana. SUNY Press: 11–38. Nelkin, Dana Kay and Samuel C. Rickless. 2015. “So Close, Yet So Far: Why Solutions to the Closeness Problem for the Doctrine of Double Effect Fall Short.” Noûs 49.2: 376–409. Pohlhaus, Jr., Gaile. 2012. “Relational Knowing and Epistemic Injustice: Toward a Theory of Willful Hermeneutical Ignorance.” Hypatia 27.4: 715–35. Stock, Kathleen. 2017a. “Imaginative Resistance and the Woody Allen Problem.” 13 November. Thinking About Fiction. https://www.thinkingaboutfiction.me/ blog/2017/11/12/imaginative-resistance-and-the-woody-allen-problem. Stock, Kathleen. 2017b. Only Imagine: Fiction, Interpretation, and Imagination. Oxford University Press. Timmerman, Travis. 2019. “A Case for Removing Confederate Monuments.” In Ethics, Left and Right: The Moral Issues that Divide Us, ed. Bob Fischer. Oxford University Press. Walzer, Michael. 1977. Just and Unjust Wars: A Moral Argument with Historical Illustrations. Basic Books.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

8 On the Virtue of Taking Oneself Lightly Macalester Bell

In everyday life people often stress the importance of not taking oneself too seriously. Given the high esteem this trait is thought to merit, it seems plausible that many consider it an especially important virtue of character, although, admittedly, it is not usually extolled using the language of virtue. While not taking oneself too seriously, or to use a less familiar locution, the virtue of taking oneself lightly, is widely praised in everyday life, philosophers have devoted very little attention to it, and in this chapter I offer a general account of what I see as a neglected moral virtue. A proper understanding of the virtue raises some deep questions about the nature of respect, alienation, and the importance of affective resistance under oppression. For Kantians, the virtue of taking oneself lightly looks to be in tension with the respect that we are said to owe all persons. Persons have dignity and merit respect. It is impossible to respect persons too much, and because of this, some Kantians will insist, it is impossible to take oneself too seriously. Ultimately, I aim to show that the standard Kantian position here is mistaken; the virtue of taking oneself lightly is not in deep tension with the kind of self-respect we ought to cultivate. There is a long tradition in philosophy according to which unity of the self is an important regulative ideal. On this picture, the virtuous person will be psychologically coherent and unified. But the virtue of taking oneself lightly seems to be in tension with this ideal. For the person who manifests this virtue seems to take up a meta-attitude which undermines her psychological coherence and unity. To take oneself lightly one must be distanced from one’s self to some degree, and this distance may be thought to open the door to a morally troubling form of conflict or alienation. In the end, I argue that the ideal of unity and psychological coherence that would render the virtue of taking oneself lightly objectionable ought to be rejected.

Macalester Bell, On the Virtue of Taking Oneself Lightly In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0008

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

162

 

In the last section of the chapter, I consider whether taking oneself lightly should be considered a virtue worth striving for under conditions of serious injustice and oppression. Under these circumstances, it might be thought that taking oneself lightly has no place in the good life. For oppression ought to be resisted, and it is sometimes suggested that a virtuous person will respond with righteous anger to the injustices which merit anger, and the virtue of appropriate anger looks to be incompatible with taking oneself lightly. In response, I argue that despite its seeming triviality and apparent incompatibility with apt anger, taking oneself lightly is an especially important virtue to cultivate in oppressive contexts.

I. On the Vice of Taking Oneself Too Seriously and the Virtue of Taking Oneself Lightly A. Taking Oneself Too Seriously and Taking Oneself Lightly To begin thinking about the virtue of taking oneself lightly, let’s begin with an example of a person who evinces the opposing vice: Garrett the Environmentalist: Garrett is a committed environmentalist. He spends most of his weekends hiking or picking up trash on a nearby beach. He attends a Quaker meeting every Sunday and is dedicated to non-violent forms of communication. He believes all persons have inherent worth and equal dignity, and he is utterly respectful of everyone he meets. He often greets people with a serene smile, but he does not enjoy most comedy since he abhors any sort of mean-spiritedness. When he discovers that he had been inadvertently throwing away some types of plastic that his city added to the recycling program, he berates himself and commits to doing better at keeping up with the news. More generally, when he makes a mistake or discovers an inconsistency in his values, he doggedly resolves never to make the mistake again and attempts to render his values consistent going forward. He is not disposed to make self-deprecating comments, and when a friend lightly teases him about how such a committed environmentalist could have spent the last three months throwing away the city’s newly recyclable yogurt containers, he does not laugh; instead, he blushes and quickly finds an excuse to leave. While Garrett is a moral exemplar in many respects, there is something about him that many find objectionable. Garrett seems to take himself way

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

163

too seriously. While he recognizes his moral imperfections, he is not at all inclined to laugh at them, and he does not enjoy the good-natured ribbing of his friends, even when he recognizes that they are not motivated by malice or ill will. What do people mean when they criticize someone for taking themselves too seriously? Before attempting to answer this question directly, let’s pause here to briefly distinguish what seems to be objectionable about Garrett from other, more familiar, faults or vices. First, taking oneself too seriously should be distinguished from arrogance. While people who take themselves too seriously may well be arrogant, we can’t reduce the vice to arrogance. I won’t attempt to provide a full account of arrogance here, but it seems uncontroversial to suppose that an arrogant person is one who holds a comparatively high opinion of herself vis-à-vis others and is disposed to attempt to make her supposed superiority felt by demanding that others appreciate her high status. But we can’t capture what seems to be objectionable about Garrett by appealing to the vice of arrogance. He is a committed egalitarian and doesn’t take himself to be superior to anyone; he is thoroughly humble, and he never attempts to make others feel inferior. Second, taking oneself too seriously should be distinguished from moral sainthood.¹ While Garrett might be interpreted as being excessively committed to morality, and this fanatical commitment might be thought to be what is objectionable about his character, I think our concerns about his character would remain even if we added details to the case that made it clear that he is no moral saint. Let’s say that in addition to his commitment to environmentalism and non-violent communication, Garrett is also a trained sommelier and accomplished oboist. This more well-rounded version of Garrett still seems to take himself too seriously. Alternatively, we could imagine another example which does not involve an overarching moral commitment at all. Consider Jason: Jason the Philosopher: Jason is committed to philosophical excellence. At the urging of a colleague, he joins an interdisciplinary working group with faculty members from a variety of fields. One afternoon, in discussing a work-in-progress, he refers to a novel as “a romance” only to have a member of the literature department gently correct him; the term “romance” has a distinct, technical meaning in the study of literature, he is told. Later, ¹ See Susan Wolf, “Moral Saints,” Journal of Philosophy, vol. 79, no. 8, 1982, pp. 419–39.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

164

 

another philosophy colleague gently teases him about his misstep, but Jason is stony-faced: “It isn’t funny,” he mutters, “I hate making that sort of basic mistake.” Like Garrett, Jason also seems to take himself way too seriously, but Jason is committed to the non-moral end of academic excellence rather than some moral ideal. What is involved in taking oneself too seriously? First, and most obviously, people who take themselves too seriously evince a somber attitude toward themselves. They do not see themselves as appropriate targets of amusement and are not themselves amused by their follies and foibles. Second, they are not disposed to make self-deprecating comments or jokes at their own expense. Third, people who are thought to take themselves too seriously have projects that they pursue and commitments that are important to them. A person disinclined to laugh at himself or make selfdeprecating comments but who spends his days idly surfing the Web or watching reality TV might be criticized for leading a superficial life, but such a person wouldn’t be described as taking himself too seriously. To be criticized as taking oneself too seriously presupposes, at minimum, that one takes oneself at least somewhat seriously, and a person without any projects or commitments would fail to satisfy this condition altogether. To understand why these traits, taken together, are constitutive of a vice, I’d like to pivot and focus my attention on what I see as this vice’s corresponding virtue, which I will call taking oneself lightly. Someone who takes herself lightly is someone who has a range of projects (or, perhaps, a single, important, project), at least some of which are serious commitments, but such a person treats herself and her commitments in a distinctive manner; she sees herself as an apt target of amusement, and she is disposed to laugh at herself, especially in response to some of her mistakes and failings, and she isn’t overly somber. She will, on occasion, make selfdeprecating comments and jokes. In short, the distinguishing trait of someone who takes herself lightly is her disposition to take an attitude of amusement toward herself and her projects. Philosophers have defended a wide variety of general theories of amusement, and I will give a brief overview of some of these theories below. Some have argued that amusement should be understood in terms of an attitude of superiority to the object of amusement; to be amused is to see the object as inferior to you along some axis of comparison. In the Leviathan Thomas Hobbes defines laughter as follows: “Sudden glory, is the passion

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

165

which makes those grimaces called LAUGHTER, and is caused either by some sudden act of their own that pleaseth them, or by the apprehension of some deformed thing in another, by comparison whereof they suddenly applaud themselves.”² So understood, laughter, and the attitude of amusement it expresses, is closely connected to contempt. To be amused is to see the target of your amusement as contemptible. Like contempt, amusement is reflexive; to be amused is to make a comparison between yourself and the target of your amusement and to see yourself high and the object of amusement as low. On such an account, laughing at oneself might be thought to be a conceptual impossibility; amusement is reflexive, and it seems impossible to look down upon oneself in this manner. But Hobbes suggests that his superiority account of laughter can make sense of self-amusement. While we often don’t find jokes at our expense amusing, we can under some circumstances: “. . . the passion of laughter is nothing else but sudden glory arising from some sudden conception of some eminency in ourselves, by comparison of some infirmity of others, or with our own formerly: for men laugh at the follies of themselves past, when they come suddenly to remembrance, except they bring with them any present dishonor.”³ For Hobbes, selfamusement is coherent insofar as we are amused at our past foolishness and thereby simultaneously pleased at our improved, present, state of being. According to some commentators, Hobbes’ superiority theory has difficulty making sense of amusement at one’s own present state. As Noël Carroll writes: . . . [W]e may laugh when we are amiably teased but this is hard to explain in terms of feelings of superiority we supposedly have, since it is not some former self whose shortcomings are being tweaked but ourselves in the present. Likewise, we often laugh at ourselves precisely at the moment when we find that we are in the process of doing something foolish, such as putting sugar instead of parmesan cheese on our spaghetti when we are tired. In that case, we are not laughing at some former self but at the guy with the spaghetti in his mouth.⁴

While Carroll is right to insist that we can be amused at who we are in the present, I don’t think this poses an insuperable objection to the superiority ² Hobbes, Leviathan, Cambridge: Hackett Publishing Company, 1994, Part I, ch. 6. ³ Hobbes, Human Nature, Oxford: Oxford University Press, 2008, ch. 8, §13. ⁴ Noël Carroll, Humour: A Very Short Introduction, Oxford: Oxford University Press, 2014, pp. 11–12.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

166

 

theory. Amusement at oneself in the present moment can be rendered intelligible on the superiority theory. We can take up different perspectives on ourselves, and one way of understanding self-amusement is in terms of a shift in perspective: when we are amused at ourselves we are considering ourselves from the perspective of who we want or aspire to be.⁵ From that perspective, we can coherently look down upon who we’ve, in fact, revealed ourselves to be, in the present, through our actions or attitudes. Others have argued that amusement isn’t a feeling of superiority but a recognition of some sort of incongruity. Arthur Schopenhauer writes: “The cause of laughter in every case is simply the sudden perception of the incongruity between a concept and the real objects which have been thought through it in some relation, and laughter itself is just the expression of this incongruity.”⁶ According to Schopenhauer and others who defend incongruity accounts of humor, to be amused is to perceive an incongruity between the object of amusement and one’s expectations or presumptions about how the world is or how the world ought to be. On this view, amusement arises from a background of expectations or norms.⁷ These norms may be norms of logic or prudence or language or profession, and, more directly relevant for my purposes in this chapter, they can also be moral norms. Amusement tracks incongruity between these norms and behavior or attitudes that run counter to them. Of course, not all norm violations will be seen as amusing. Context matters a great deal as to whether some incongruity will merit amusement or be found amusing. Some people find “dead baby” jokes amusing, and incongruity theories of humor can explain this, but babies actually killed in horrific ways obviously don’t merit amusement. The formal features of the joke are what make these jokes amusing, to the extent that they are. The joke

⁵ As an anonymous referee pointed out, in some cases, we can be amused by our own desires; we can coherently laugh at ourselves qua person who wants something we see as ridiculous. In these types of cases, we presumably do not endorse our first-order desire, and when we are amused at our desires, it is from this perspective. ⁶ Arthur Schopenhauer, The World as Will and Idea, translated by R. B. Haldane and John Kemp, 6th edition, London: Routledge and Kegan Paul, 1907–9. ⁷ Noel Carroll, Humour: A Very Short Introduction, p. 18. As Carroll notes, the concept of incongruity is rather “baggy,” and the types of incongruities constitutive of the amusing is wideranging: “prototypical incongruities, then, include deviations, disturbances, or problematizations of our concepts, rules, laws of logic and reasoning, stereotypes, norms of morality, of prudence, and of etiquette, contradictory points of view presented in tandem, and in general, subversions of our commonplace expectations, including our expectations concerning standard emotional scenarios and schemas, our norms of grace, taste, and even the very forms of comedy itself ” (p. 27).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

167

context undermines the potential moral threat posed by the incongruity of babies murdered, and they encourage us to take up an attitude of comic distance which arrests the sympathetic responses we would normally have toward those who are victims of violence.⁸ Part of what makes jokes amusing is the spirit of play which they exemplify. This brings us to a third account of humor. According to those who defend a play theory of humor, we should understand humor as part of the activity of play. Max Eastman, in his book, Enjoyment of Laughter asserted that the “first law of humor is that things can be funny only when we are in fun.”⁹ When we are “in fun” and we encounter “disagreeable things,” the affective valence of our experience of them shifts, and things that would normally pain us give us pleasure. This shift in valence is brought about by taking a disinterested attitude toward the object of humor.¹⁰ Being in fun is a kind of playfulness, and humor is one dimension of play. While many philosophers who write on humor are interested in defending one single, unified account of amusement, which they hope can explain all tokens of amusement, I think each of these theories highlight important dimensions of the amusement characteristic of taking oneself lightly. The person who takes herself lightly notices, and takes some pleasure in, incongruities between who she aspires to be and who she, in fact, is, as revealed through her mistakes, foibles, and inconsistencies. This kind of amusement involves a person “looking down” on herself; one assesses oneself, and finds oneself wanting, from the perspective of who one aspires to be. This kind of amusement is best characterized as a form of play in which the person is “in fun” with herself and takes up a disinterested attitude toward herself. Someone who takes herself lightly is disposed to feel this attitude of amusement toward herself. Specifically, such a person will be disposed to be amused at her mistakes and failings, both moral and non-moral, as well as her idiosyncrasies and unusual habits. She will be disposed to be amused when her friends tease her good-naturedly and will play along with the ribbing. Such a person will also express this attitude, on occasion, through her own self-deprecating comments and jokes. As is the case with the other virtues, the virtue of taking oneself lightly lies in a mean between excess and deficiency, and someone who evinces the ⁸ Carroll, Humour, p. 31. ⁹ Max Eastman, Enjoyment of Laughter, New Brunswick: Transaction Publishers, 2009, p. 3. Originally published New York: Simon and Schuster, 1936. ¹⁰ See Aaron Smuts, “Humor,” Internet Encyclopedia of Philosophy, http://www.iep.utm.edu/ humor/.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

168

 

virtue will be amused at herself in the right way, at the right time, to the right degree, and so on. While we may fall into vice by taking ourselves too seriously and refusing ever to be amused at our follies, we may also fall into vice by being in a constant, indiscriminate, state of self-amused buffoonery. Someone who takes herself lightly is someone who is appropriately amused at herself.¹¹ Moreover, someone who takes herself lightly is not a person who simply luxuriates in a state of self-amusement. If you evince the virtue of taking yourself lightly, you strive to be a morally good person, so you don’t just respond with amusement at discovery of your faults; instead, you work to improve yourself. But you do this work with a playful, rather than solemn, attitude. Why think that taking oneself lightly is properly characterized as a virtue? The folk, when extolling the merits of not taking oneself too seriously, don’t refer to the trait as a virtue, and it isn’t commonly discussed by contemporary virtue theorists. There are, of course, different accounts of what makes some trait a virtue. For eudemonistic theorists, a virtue is a quality of character which is good for its possessor; whereas for other virtue theorists a virtue is a quality of character which is admirable or leads to good consequences. I don’t wish to take a stand on which virtue theory we should accept; the case could be made that taking oneself lightly is a virtue no matter what specific account of virtue one ultimately endorses. It makes sense to characterize taking oneself lightly as a virtue for a few reasons. First, it is related to, although distinct from, other widely recognized virtues, such as humility, modesty, graciousness, wittiness, and friendliness. Second, it is a mean between a number of seeming vices (i.e., self-righteousness and taking oneself too seriously on the one hand, and a kind of servile buffoonery or lack of self-seriousness on the other). Third, taking oneself lightly involves being disposed to feel amusement at oneself and make selfdeprecating jokes at the right time, in the right place, in the right way, and so on. It isn’t an attitude that is more admirable the more it is maximized.

B. The Moral Value of Taking Oneself Lightly Having offered a brief sketch of the virtue of taking oneself lightly, I’d now like to turn to a discussion of its value. ¹¹ See Aristotle, Nicomachean Ethics, trans. Terence Irwin, Indianapolis: Hackett Publishing Company, 1999, Book II, Chapter VII.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

169

At first glance, taking oneself lightly may appear to be more of a vice than a virtue; it looks like evincing it would lead us away from being excellent people, and some might conclude that it isn’t an admirable trait worthy of cultivation. Garrett, our imagined righteous environmentalist, would likely be aghast at the suggestion that it would be good for him to laugh and poke fun at himself for failing to recycle the newly recyclable yogurt containers. If you are a committed environmentalist, and you discover you have been failing to recycle, how does being amused at yourself help? Surely, Garrett might insist, solemnly correcting oneself and vowing to do better is how a virtuous person would go on in the face of moral failings. Laughing and poking fun at oneself might seem like a morally perverse reaction that actually encourages vice.¹² Despite the force of these worries, I think a case can be made for the moral admirability of taking oneself lightly. First, part of the value of taking oneself lightly is that it is a trait that is pleasing to others and demonstrates a willingness to be agreeable. Someone who can laugh at themselves is able to avoid a kind of priggish selfrighteousness that can be off-putting and potentially threatening. Why is it that many people enjoy being around those who take themselves lightly? This is obviously a question for psychologists, and the extant literature on self-deprecating humor suggests that some self-deprecators are evaluated positively because they are seen as especially confident and worthy of esteem. According to one study, high status people are seen as particularly attractive when they utilize self-deprecating humor.¹³ In a social environment where people constantly compete for status and esteem, those who are well positioned in the hierarchy can signal their comparatively high status by playfully putting themselves down through the use of self-deprecating humor. More generally, I suspect that people find those who take themselves too seriously unpleasant to be around because such people give rise to the worry that they are as stinting in their goodwill toward others as they appear to be toward themselves. In addition, even if such persons are utterly lacking in arrogance, they might be interpreted as lording their righteousness over

¹² Philosophers have long been suspicious of the potential negative effects of amusement. In the Republic, for example, Plato claimed that laughter threatened persons’ rational self-control and denied comedies any role to play in the guardians’ education. ¹³ Gil Greengross and Geoffrey F. Miller, “Dissing Oneself versus Dissing Rivals: Effects of Status, Personality, and Sex on the Short-Term and Long-Term Attractiveness of SelfDeprecating and Other-Deprecating Humor,” Evolutionary Psychology, July 1, 2008, https:// doi.org/10.1177/147470490800600303.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

170

 

others in an attempt to elevate themselves in comparison.¹⁴ The playful self-deprecating humor of a person who takes herself lightly blocks these interpretations. Second, taking oneself lightly is epistemically advantageous. A person who takes herself lightly is someone who is aware of her lack of integrity, and, more to the point, the pleasure that she gets from viewing her failures and foibles in a disinterested manner leads her to be more open to discovering ways in which she is failing to live up to her own standards. This is a consequence of two features of amusement: its tendency to interrupt routine and its pleasure. First, many of our inconsistencies and moral failings can pass by undetected in a blur of routine and habit. Amusement characteristically wrests us out of autopilot mode and can make us aware of inconsistencies we had taken for granted. Second, the pleasure habitually felt by someone who takes herself lightly will put her in a better position for selfknowledge. Consider Garrett. Given the pain that accompanies any realization that he has failed to live up to his ideals or meet his basic standards, it seems unlikely that he will be on the lookout for inconsistencies and failures going forward. After all, these moral failings pain him greatly, and human beings don’t generally seek out this sort of pain. A person who takes herself lightly isn’t immune from feeling guilt and shame, but she can escape these feelings, temporarily, by taking a more playful perspective on herself and her failings. Given this, she isn’t as vulnerable to devastation in the face of evidence of her own shortcomings and is, therefore, more open to discovering them. This protects her from fanaticism and rigidity. In short, a person who takes herself lightly is in a better epistemic position to know her faults and failings; without this kind of knowledge it is impossible to strive for integrity. We need to appreciate the incongruities between who we are and who we strive to be so that we know ourselves. Amusement signals that we understand the fact that there is never perfect congruity and is a way of appreciating our fallible nature. This sort of appreciation is not the same as reveling in or even endorsing the incongruities, nor is it the same as condoning them. Taking oneself lightly is compatible with holding oneself responsible for one’s faults and foibles. But no matter how we think an imperfectly virtuous person will go on in the face of these incongruities, clearly they must first be acknowledged. And the most generous and human ¹⁴ In Hard Feelings: The Moral Psychology of Contempt, New York: Oxford University Press, 2013, I argue that “superbia” is an attitude of taking oneself to be of comparatively higher status than others and a desire to make this superior status felt by others.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

171

way of acknowledging these incongruities between who we are and who we aspire to be is by taking oneself lightly. Third, a person who takes herself lightly sees herself as a social creature and morality as a social enterprise.¹⁵ The practice of morality, no matter what its metaethical foundation, is a social practice, and taking oneself lightly is a way of acknowledging this. As we have seen, taking oneself lightly involves occasionally making self-deprecating jokes and comments. By participating in these activities, we are inviting others to share in our perspective. That is, in seeing ourselves as an object of amusement, we regard ourselves as part of a community. When we laugh at ourselves and join in good-natured teasing at our own expense we demonstrate that we recognize that the moral community is a social community.¹⁶ Fourth, a person who takes herself lightly is a better recipient of moral criticism. Taking oneself too seriously makes it more challenging to give fitting and apt blame uptake.¹⁷ One is less likely to be open to and receptive to criticism if one shuts down and berates oneself. Some might worry that the account I’ve offered of the virtue of taking oneself lightly fails to take into account background social circumstances. Specifically, it could be argued that taking oneself lightly is virtuous when one enjoys a relatively high social standing but it is not especially virtuous when one endures a relatively low social position. For example, those who are successful on the philosophy job market might be thought to evince a kind of virtue in taking themselves lightly for this attitude might be a way of marking that they recognize their success was partly a matter of good fortune. But there is no corresponding reason for those who haven’t had success on the job market to take up this attitude of playful amused detachment in response to their predicament, and those who fail to secure jobs cannot be said to evince a fault in treating their predicament in a very serious, and somber, manner. More generally, we might think this example

¹⁵ This is a theme which animates much of Cheshire Calhoun’s work. See her recent essay collection, Moral Aims: Essays on the Importance of Getting it Right and Practicing Morality with Others, New York: Oxford University Press, 2015. ¹⁶ There are, admittedly, people who put themselves down with the express aim that others sing their praises in response. This sort of false modesty should be distinguished from the kind of self-deprecation characteristic of taking oneself lightly. People who engage in false modesty aren’t taking up a playful attitude toward themselves, and they don’t really see themselves as an object of amusement. Instead, they strategically use the language of self-deprecation in order to manipulate others into offering praise. ¹⁷ I am grateful to Robert Wallace for suggesting this point.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

172

 

shows that taking oneself lightly is only something like a virtue for those who enjoy a high social standing.¹⁸ In response, I think this characterization of what is good about taking oneself lightly mischaracterizes the virtue. It is not best understood as analogous to some sort of noblesse oblige of the privileged. In fact, on that characterization, the attitude at the heart of the virtue strikes me as more vicious than virtuous; there is something condescending about not taking oneself too seriously in order to make others feel better about their unfortunate lot in life. Moreover, even those who enjoy comparatively low social standing can take themselves too seriously. The philosophy graduate student who ends up without employment and blames the women and minorities for snatching up all the jobs can be both socially disadvantaged and evince the trait of taking himself too seriously. It is certainly true that background social conditions contour which traits are considered virtues and affect how we should understand and characterize the virtues. As I will go on to argue in Section III, I think taking oneself lightly should be considered a virtue even under oppressive social conditions.

II. Two Worries about Taking Oneself Lightly While taking oneself lightly is an important moral virtue, it seems to be a virtue that is difficult for many moral theorists to accommodate. In what follows I will consider two broad objections that may be raised against taking oneself lightly: a Kantian, respect-based, objection and an objection to the picture of the self which is presupposed by the virtue.

A. Kantian Respect-Based Objections A person who takes herself lightly is disposed to occasional self-deprecation, when the situation calls for it, and she can laugh at herself and treats herself as an object of amusement. This tendency toward self-deprecation and selfmockery will presumably worry some Kantians.

¹⁸ I am grateful to R. Jay Wallace for pressing this objection.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

173

I will bring this worry to the fore by considering some of Kant’s remarks about defamation and ridicule. While Kant is focused on how we can wrong others by making them a target of our derisive laughter, his claims also constitute a potential objection to my account of the virtue of taking oneself lightly. In the Metaphysics of Morals Kant distinguishes three vices that violate our duty to respect persons: arrogance, defamation, and ridicule. Arrogance, according to Kant, involves taking oneself to be superior to others and demanding that others come to share this assessment of their own comparative inferiority.¹⁹ Since every person is owed a basic form of respect, the arrogant person’s demand is contrary to our duty. Defamation is defined as “the immediate inclination, with no particular aim in view, to bring into the open something prejudicial to respect for others. This is contrary to the respect owed to humanity as such; for every scandal given weakens that respect, on which the impulse to the morally good rests, and as far as possible makes people skeptical about it.”²⁰ Rather than taking on the task of exposing the moral faults of others, which Kant worries will lead to misanthropy and general cynicism, he urges that we throw a “veil of benevolence” over the moral faults of others.²¹ Kant’s discussion of ridicule is the most interesting for our purposes. Kant distinguishes the vice of ridicule from innocuous banter: Wanton faultfinding and mockery, the propensity to expose others to laughter to make their faults the immediate object of one’s amusement is a kind of malice. It is altogether different from banter, from the familiarity among friends in which one makes fun of their peculiarities that only seem to be faults but are really marks of their pluck in sometimes departing from the rule of fashion (for this is not derision). But holding up to ridicule a person’s real faults, or supposed faults as if they were real, in order to deprive him of the respect he deserves, and the propensity to do this, a mania for caustic mockery (spiritus causticus) has something of fiendish joy in it; and this makes it an even more serious violation of one’s duty of respect for other human beings. (6:467)

A couple of things are worth highlighting about Kant’s remarks. First, as the distinction between banter and ridicule makes clear, there is, for Kant,

¹⁹ Immanuel Kant, The Metaphysics of Morals, edited by Laura Denis, translated by Mary Gregor, Cambridge: Cambridge University Press, 2017, 6:465. ²⁰ Kant, Metaphysics of Morals, 6:466. ²¹ Kant, Metaphysics of Morals, 6:466.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

174

 

nothing morally objectionable about lighthearted teasing when there is background good will and when the fault that is mocked is a trivial peculiarity rather than a genuine moral fault. For Kant, mocking real or supposed moral faults is a failure of respect because in doing so, we are comparing ourselves to the target of our ridicule and holding the other in contempt. This contempt is, Kant claims, incompatible with the respect we owe all persons. The kind of teasing characteristic of banter is not focused, Kant seems to imply, on real, central features of the target, and therefore it doesn’t count as holding the person in contempt. Second, while the arrogant person demands that others acknowledge his supposed comparative superiority, the defamer and the ridiculer attempt to knock others down a peg or two by exposing their moral faults, and, in the case of ridicule, taking pleasure in the attempt to lower the moral status of another. This pleasure in attempting to deprive someone of the very respect that they are due as a person reveals the ridiculer’s utterly false set of values. We’ve seen why Kant is troubled by ridicule of others, but do his remarks have any implications for our assessment of taking ourselves lightly? A person who takes herself lightly is disposed to laugh at her faults, but it is worth stressing that taking oneself lightly does not involve a disposition to be amused at every moral failing. Feeling wry amusement at one’s selfish betrayal of a friend, immediately after the betrayal, does not show that one takes oneself lightly in the sense that has been discussed here. In this case, finding humor in one’s part in such a betrayal simply makes the person more worthy of criticism. Nevertheless, taking oneself lightly involves finding humor in more than simply one’s fortitude at bucking the trends and fashions of the day, so it cannot be reduced to a disposition to engage in harmless banter, in Kant’s sense. A person who takes herself lightly will, on occasion, be disposed to engage in mocking self-deprecation about her own moral failures and failings. Which moral faults and failings will be the target of amused detachment? I think it is impossible to provide a general answer to this question. It might be thought that the person who takes herself lightly will only be amused at morally insignificant faults. Perhaps this is why many find Garrett’s response odd: his failure to properly recycle the yogurt containers isn’t really a failure at all due to its insignificance, and his unwillingness to laugh at himself seems perverse. But while I concede that Garrett’s “transgression” isn’t especially significant, I don’t think a person who takes herself lightly will be disposed to be amused only in response to her insignificant failures and failings. A person who evinces the virtue cares about her

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

175

transgressions, and she will take responsibility for her actions and attempt to right wrongs done to others by apologizing and expressing her contrition. These reparative activities call for a certain solemnity in attitude so as not to distract from the message being expressed. Because of this, playful selfdeprecation in the immediate aftermath of a serious betrayal of a friend would be inappropriate because the attitude expressed would fail to harmonize with the contrition which is partially constitutive of remorse. But if the fault or failing only affects the subject, or if it is years in the past, or if apologies have already been made and accepted, the person who takes herself lightly may respond with amusement even to significant moral failures. Kant’s discussion of ridicule is focused on one person mocking another, but given that Kant insists that we have duties to respect ourselves as well as others, if we accept Kant’s account of the moral badness of ridicule, we should draw a similar conclusion about the badness of mocking oneself. Taking pleasure in poking fun at one’s own moral faults is, according to this line of thought, objectionable for the same reasons. Perhaps it would be judged to be even more objectionable since a person poking fun at himself is often successful in eliciting the laughter of others as well; it is easier, and more fun, to join in the activity of ridiculing a person who is ridiculing himself. Since being disposed to mock oneself is, I’ve claimed, part of the taking oneself lightly, and since Kant argues that this kind of mockery is vicious, it looks like we have Kantian reasons to conclude that taking oneself lightly is, in fact, more vicious than virtuous. For a person who takes herself lightly is failing to respect herself and perhaps others in the moral community. I think this worry is misplaced. Despite what Kant writes about the dangers of ridicule, I think we have little reason to accept Kant’s assessment, even if we grant many of his starting points. As we have seen, Kant’s argument against ridicule depends on the assumptions that all persons are owed a basic form of respect, that ridicule involves the expression of contempt, and that contempt is incompatible with the respect we owe all persons. I will grant the first two assumptions and challenge the third. The kind of respect that we owe all persons is compatible with harboring contempt, and what Kant writes about demonstrating respect and the social virtues suggests that, despite what Kant claims, there isn’t such a clear tension between respecting a person, qua person, and feeling and expressing contempt for the person.²² ²² For a fuller defense of some of these claims, see my Hard Feelings: The Moral Psychology of Contempt, ch. 4.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

176

 

Kant explicitly claims that harboring contempt for a person is incompatible with respecting them. He writes: “To be contemptuous of others (contemnere), that is, to deny them the respect owed to human beings in general, is in every case contrary to duty; for they are human beings.”²³ But as contemporary philosophers have noted, there are at least two forms of respect, and Kant repeatedly runs these two forms of respect together in much of his discussion.²⁴ Stephen Darwall has pointed out that we can distinguish between what he calls recognition and appraisal respect. The former, consists in “a disposition to weigh appropriately in one’s deliberations some feature of the thing in question and to act accordingly.”²⁵ This type of respect can take any number of things as its object, including persons. But this form of respect should be distinguished from another attitude that is also referred to as “respect.” Appraisal respect, when it takes persons as its object, involves a positive evaluation of the person (qua person or qua occupier of some role) and comes in degrees. Harboring contempt for someone is incompatible with having this kind of esteem for them. If all persons are owed respect, it is recognition respect that they are owed, not appraisal respect. But having recognition respect for someone is not incompatible with harboring or expressing contempt for them. Insofar as Kant suggests otherwise, he is failing to properly distinguish these two forms of respect. We can grant that all persons qua persons deserve respect, but deny that this respect is incompatible with harboring contempt for them. Having said this, it is undeniable that Kant is deeply troubled by expressions of contempt. For example, he writes: “At times one cannot, it is true, help inwardly looking down on some in comparison with others (despicatui habere); but the outward manifestation of this is, nevertheless, an offense.”²⁶ When giving examples of social practices that express contempt, Kant tends to focus on grizzly forms of punishment, such as drawing and quartering and cutting off the offender’s nose.²⁷ These types of punishment “make a spectator blush with shame at belonging to the species that can be treated that way.”²⁸ While these punishments can be understood as expressive of

²³ Kant, Metaphysics of Morals, 6:463. ²⁴ Michelle Mason first makes this point in “Contempt as a Moral Attitude,” Ethics, vol. 113, no. 2, 2003, pp. 234–72. See also my Hard Feelings: The Moral Psychology of Contempt. ²⁵ Darwall, “Two Kinds of Respect,” Ethics, vol. 88, no. 1, 1977, pp. 36–49, p. 38. ²⁶ Kant, Metaphysics of Morals, 6:462. ²⁷ Kant, Metaphysics of Morals, 6:463. This passage is also discussed in Mason, “Contempt as a Moral Attitude,” n. 8. ²⁸ Kant, Metaphysics of Morals, 6:463.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

177

utter contempt for persons where this is to be understood as a complete disregard for the dignity of persons, it is difficult to see what this attitude has in common with the playful ribbing of a friend for some moral fault. We could go along with Kant’s condemnation of these forms of punishment yet still think contemptuous mocking or ridicule has a role to play in responding to at least some moral faults. While taking oneself lightly may be incompatible with what Kant says about ridicule and showing contempt, I have argued that the virtue is compatible with the kind of respect Kant thinks we owe all persons.

B. Taking Oneself Lightly and Alienation There is a long tradition according to which the unity of the self is a regulative ideal. There are various ways of understanding this unity, but, most generally, the ideal agent is thought to be psychologically coherent and therefore unified. One example of this focus on coherence and unity may be found in Harry Frankfurt’s work on the will.²⁹ I shall not aim to do justice to Frankfurt’s complex account here; instead, I will offer a very rough overview of a few aspects of his account in order to bring out the emphasis he places on psychological unity which might seem to be in tension with my account of taking oneself lightly. Frankfurt defends a hierarchical model of desire. On this model, we have our ordinary first-order desires to take some action (say, to recycle all one’s recyclables), but persons can also form second-order desires, i.e., we may form desires concerning our first-order desires (say, to desire to desire to recycle all one’s recyclables). When we have a second-order desire that some first-order desire be effective in action, we have a second-order volition. We can develop our second-order volitions through a process of identification. If I desire that my desire to recycle all my recyclables be effective in action, I have identified myself with this desire, and it thereby becomes my own desire in a way that other desires, that I don’t identify with, are not. Frankfurt allows that there may be conflict between a person’s secondorder desires. It also might be the case that a person is ambivalent about whether they truly identify with some desire. Both of these individuals fail to

²⁹ Harry G. Frankfurt, “Freedom of the Will and the Concept of a Person,” Journal of Philosophy, vol. 68, no.1, 1971, pp. 5–20.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

178

 

be wholehearted. They haven’t resolved a conflict in their psychology and fall short of an ideal of personhood. As a result, they are alienated in an important sense. The virtue of taking oneself lightly seems to be in tension with this ideal because a person who takes himself lightly will harbor a meta-attitude toward his self which looks as if it undermines his psychological coherence and unity. As we have seen, to be amused at oneself involves taking up a particular perspective: one considers one’s actions and attitudes from the perspective of the person one aspires to be. One comes to form a metaattitude of amusement regarding one’s first order attitudes and actions. But to do this, one must be distanced from one’s self to some degree, and this distance may be thought to open the door to a morally troubling form of conflict or alienation. For in taking oneself lightly, one is amused by a failure that, all things considered, one might think one shouldn’t be amused by. If you are, say, a committed environmentalist, you might think all your attitudes toward your environmentalism should cohere with one another. That’s what it means to be committed to your values. But amusement at your environmental lapses fails to cohere with your professed environmentalism. For to be amused at one’s failings is not to stand against them. As a result, you are in a state of deep and objectionable disharmony, and one might think that this type of disharmony is incompatible with virtue. In response, I’d like to push back on the idea that being amused at one’s failings is in conflict with being a person with serious commitments. One can be a committed environmentalist and disposed to feelings of amusement at one’s environmental lapses, if the amusement doesn’t undermine one’s commitment to the environment. A person who takes herself lightly won’t wallow in her self-amusement or act contrary to her commitments simply in order to have an occasion for amusement. What is distinctive about the person who takes herself lightly is her attitude toward her past failures, not her resolve not to do better in the future, and I see no reason to suppose that amusement in response to failure, by itself, undermines one’s dedication to one’s commitments. But even if this point about one’s resolve to do better in the future is granted, there might be an objectionable form of disharmony at the level of attitude in the person who takes herself lightly. Taking oneself lightly involves taking up, on occasion, a disinterested perspective on one’s self, and because of that, a person who evinces this virtue may be said to be alienated to some degree. However, I think this is a bullet that should be bitten. I see no reason why this sort of perspective taking, and any alienation

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

179

that comes with it, should be considered morally objectionable or somehow undermining of agency.³⁰

III. On Taking Oneself Lightly in Dark Times So far, I’ve offered an account of taking oneself lightly and shown how this under-discussed virtue raises questions about respect and how we should understand the self. But even some who accept my claims may have doubts about the moral importance of taking oneself lightly, especially under circumstances of oppression. It likely strikes many as, at best, a rather trivial trait, and unlike virtues such as courage, it might seem that taking oneself lightly has no place in a good life under circumstances of oppression. Some might go further and claim that under unjust social circumstances, there is nothing virtuous about taking oneself lightly. Members of historically oppressed groups have special reason to take themselves seriously, and the suggestion that it is virtuous to take up an attitude of detached amusement and engage in self-deprecation constitutes an utter failure to take the perspectives of the oppressed and their specific moral burdens into account. I will close by considering whether the virtue of taking oneself lightly is a virtue worth striving for under circumstances of serious injustice. One way to think about how circumstances of oppression shape how we ought to think about virtue is to consider the moral status of anger. In recent years, there has been a great deal written about the value of anger and other hard feelings under circumstances of oppression. Some have argued that anger is the primary way of protesting oppression. As a form of emotional insubordination, anger has been lauded as a way of protecting the selfrespect of the less powerful. Affective submission to oppression is thought to constitute a kind of acquiescence, whereas responding with anger signals to oneself and to others that one recognizes that one has been wronged and has sufficient self-respect to resist through one’s emotional responses. Others have argued that anger has direct epistemic value insofar as the experience of anger constitutes a distinct kind of knowledge of one’s ³⁰ As an anonymous referee helpfully pointed out, there is also a competing ideal which, due to space constraints, I’m unable to fully explore here according to which inconsistency and disharmony are ultimate ends and attempts to strive for harmony are seen as a kind of bad faith. See, for example, Kenneth Gergen, “The Healthy, Happy Human Being Wears Many Masks,” in The Fontana Postmodernism Reader, edited by Walter Truitt Anderson, London: Fontana, 1996.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

180

 

oppression. Anger is also thought to have indirect epistemic value insofar as its uptake conditions can help persons understand how others in society view their moral standing. Finally, anger is thought to be morally important insofar as it can motivate social change.³¹ Someone who manifested what we might call the virtue of appropriate anger will be keenly aware of when circumstances call for an angry response and when they do not. Aristotle describes what he calls the virtue of “good temper” as the mean with regard to anger.³² For Aristotle, the person who evinces this virtue will be neither too quick to anger, nor too slow to anger. Such a person will respond with great anger to great slights and moderate anger to lesser slights. Aristotle describes the person who does not respond with anger when the circumstances call for it as foolish and slavish because such a person is willing to let insults stand without resisting them, and the person who is too quick to anger is out of control and difficult to live with. The virtue of appropriate anger seems to be an especially important virtue to cultivate under circumstances of serious injustice since under these conditions it is imperative to challenge and resist the prevailing injustice. Taking oneself lightly might be thought to be incompatible with developing the virtue of appropriate anger. Anger, as many have pointed out, involves seeing oneself as wronged and justified in bringing that wrong to the attention of others through protest. The angry person is righteous in the sense that she takes herself and the perceived wrong done to her very seriously, and she sees herself as fully justified in judging that a wrong has been done. While someone with the virtue of appropriate anger will, obviously, respond with anger only when such anger is fitting, the sense of righteousness characteristic of anger may be thought to be incompatible with the virtue of taking oneself lightly. The playfulness and lack of solemnity of the person who takes herself lightly seems to be in tension with the righteousness of anger. If, under circumstances of serious injustice, we have reason to cultivate traits that will help us resist, it is unclear whether taking oneself lightly should be considered a virtue at all, let alone an important one. Despite the apparent cogency of the worry, I think taking oneself lightly is an important virtue to cultivate, even in dark times. First, as we have seen, ³¹ For a discussion of the moral value of the virtue of apt anger under circumstances of oppression and an elaboration of these points, see my “Virtue, Anger, and Oppression,” in Feminist Ethics and Social Philosophy: Theorizing the Non-Ideal; edited by Lisa Tessman, New York: Springer Science and Business Media, 2009, pp. 165–83. ³² Aristotle, Nicomachean Ethics, 1125b–1126a.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

      

181

part of what makes taking oneself lightly a virtue is that it involves an openness to other points of view and an ability to see and appreciate one’s own faults and inconsistencies. These attitudes are especially important to cultivate in circumstances of oppression since the focus on the wrongdoing of others, and the pleasures of resentment, can blind us to our own inconsistencies and failures. Second, many social problems are ones that require the cooperation of others to ameliorate, and taking oneself lightly is important since those who evince this virtue are pleasing to others and easier to engage and work with. To be clear, I’m not suggesting that the oppressed have special reason to make themselves pleasing to their oppressors, but positive social change necessitates cooperation among the oppressed and their allies.³³ Third, taking oneself lightly is not incompatible with harboring anger for another nor does it undermine anger’s aptness. I can take myself lightly and still judge that another has done me wrong and seek to protest the wrong done. But cultivating the virtue of taking oneself lightly helps to temper the tendency of anger to seem self-justifying. Insofar as I take myself lightly, I don’t simply presume that every token of anger I experience is fitting or morally appropriate. A person who takes herself lightly isn’t righteous, and this lack of righteousness does alter the experience of anger, but it doesn’t always extinguish it. Fourth, the disinterested attitude characteristic of the amusement of the person who takes herself lightly can help protect the subject’s psyche by offering a way to retreat from the pain that can accompany a relentless focus on the wrongdoing of others. P. F. Strawson famously remarked that we may occasionally and temporarily take up what he called the “objective attitude” when the “strains of involvement” become too much.³⁴ While taking oneself lightly is distinct from taking up the objective attitude, it too can offer a temporary retreat. When we take up a playful and disinterested attitude toward ourselves, as we are disposed to do when we evince the virtue, this can offer an escape from the strains of self-involvement which can sometimes plague those who seek to resist oppression. Finally, taking oneself lightly can protect one from certain forms of humiliation and can therefore be an especially valuable virtue to evince under circumstances of injustice where members of oppressed groups can

³³ I am grateful to an anonymous referee for suggesting that I clarify this point. ³⁴ P. F. Strawson, “Freedom and Resentment,” Proceedings of the British Academy, vol. 48, 1962, pp. 1–25.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

182

 

find themselves the targets of unfair and inapt mockery. Vaclav Havel famously quipped, “Anyone who takes himself too seriously always runs the risk of looking ridiculous; anyone who can consistently laugh at himself does not.”³⁵ In taking oneself lightly, one protects oneself from some of the potential harms of this type of derision. *

*

*

Despite the emphasis ordinary people place on it, taking oneself lightly has attracted little philosophical attention. I have tried to show that it is a virtue of great moral significance. While it might seem trivial, taking oneself lightly is an important virtue to cultivate, especially in dark times.³⁶

³⁵ Vaclav Havel and Karel Hvizdala, Disturbing the Peace: A Conversation with Karel Hvizala, New York: Vintage Books, 1990, p. 56. ³⁶ I am grateful to two anonymous referees, Mark Timmons, and my fellow participants at the 2018 Workshop in Normative Ethics and the 2018 Virtue and Moral Reasoning Under Oppressive Social Conditions workshop for their questions and comments on this essay.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

9 From Duty for the Right Reasons Teemu Toppinen

§1. Moral Worth Leading a massive horde of zombies away from a village, even at a significant risk to yourself, is, under many circumstances, the right thing to do. At least this is so if you are a superb sword fighter. But now consider two scenarios: Z 1. Michonne is a superb sword fighter. She observes a vast horde of zombies roaming toward the village. The measly wall surrounding the village won’t hold against the horde, and the villagers’ escape routes are completely blocked. Michonne believes that the only way to prevent the horde from stumbling into the village is to attract the zombies to another direction. This is going to be very dangerous, and is bound to involve fighting a considerable—but probably manageable—number of zombies. Michonne doesn’t really care about the villagers, but engaging with the zombies allows her to learn more of the Way of the Sword. And so she leads the zombies away from the village. Z 2. Michonne is a superb sword fighter. She observes a vast horde of zombies roaming toward the village. The measly wall surrounding the village won’t hold against the horde, and the villagers’ escape routes are completely blocked. Michonne believes that the only way to prevent the horde from stumbling into the village is to attract the zombies to another direction. This is going to be very dangerous, and is bound to involve fighting a considerable—but probably manageable—number of zombies. Michonne decides that given that the lives of the villagers are at stake, and that she’ll probably be able to handle what zombies she’ll encounter while distracting the horde, this is the right thing to do. And so she leads the zombies away from the village. In both scenarios, Michonne does the right thing. But in the first scenario her action is only accidentally right. Next time, the option that works Teemu Toppinen, From Duty for the Right Reasons In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0009

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

184

 

optimally with regard to learning of the Way of the Sword may turn out to be one that is not morally permissible. Michonne’s action reflects better on her in the second scenario. In this scenario, she does the right thing from what would seem like a good set of motivations. She wishes to save the villagers; she wants to do the right thing. When Michonne’s action is motivated in this way, it is no accident that she does the right thing. An action has moral worth, let us say—or is morally creditworthy—when it’s a morally right action (or close enough) and when it is motivated in such a way that its being right is not accidental. When an action is, in this way, non-accidentally right (or close enough), the agent is morally praiseworthy for doing the right thing. The action speaks well of the agent, in a certain kind of way. Competing accounts have been offered of how, exactly, we should cash out the idea of non-accidentally right, morally worthy actions. According to one kind of view—let’s call this the Right Reasons View—an agent performs an action with moral worth, or is praiseworthy for doing the right thing, roughly to the extent that she does the right thing for reasons that make it right to act in this way. According to another kind of view, actions with moral worth spring from the ‘motive of duty,’ or are based on the agent’s justified belief or knowledge that she ought to act in the relevant way. These are both attractive ideas. But given certain plausible assumptions, these ideas are in tension with each other. Thus there is some tendency, on part of those who are inspired by the first idea, to downplay the second, and vice versa. I will suggest that we can perhaps do justice to both ideas, and that metaethics is relevant here.¹ The extent to which the two attractive ideas about moral worth—roughly: that morally worthy actions are performed for the reasons that make them right, and that morally worthy actions spring from good moral thought—are in tension with each other depends on how we are to understand the nature of moral thought. If that’s right, this is of some relevance to normative ethics. I will first briefly present the Right Reasons View, and outline some challenges for it (§2), thus motivating views that give emphasis to the idea that morally worthy action springs from a motive of duty (§3). I then explore the difficulties that arise when we try to do justice to the idea of acting from a motive of duty in the context of the Right Reasons View (§4), and when we

¹ For a recent reconciliatory, pluralist, take on moral worth, see also Isserow ms, which came to my attention too late in order to be discussed here.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

185

try to capture the idea of acting for the right sorts of reasons in the context of views that emphasize the centrality of moral thought to acting with moral worth (§5). Such difficulties can be avoided to a significant extent, I argue, if we drop certain metaethical assumptions—more specifically, if we drop the assumption of cognitivism, or the idea that moral beliefs are to be understood as purporting to represent moral reality (§6). This isn’t as bad as it may sound; I close with some comforting remarks (§7).

§2. Right Reasons An attractive view of moral worth has been put forward by Nomy Arpaly and Julia Markovits, among others. In her book, Unprincipled Virtue, Arpaly (2003, p. 84) writes: For an agent to be morally praiseworthy for doing the right thing is for her to have done the right thing [ . . . ] for the reasons for which the action is right (the right reasons clause); and an agent is more praiseworthy, other things being equal, the deeper the moral concern that has led to her action (the concern clause). Moral concern is to be understood as concern for what is in fact morally relevant and not as concern for what the agent takes to be morally relevant.

Arpaly has revised the view in some respects. A more recent characterization is as follows (Arpaly and Schroeder 2014, p. 170): [A] person is praiseworthy for a right action A to the extent that A manifests an intrinsic desire (or desires) for the complete or partial right or good (correctly conceptualized) [ . . . ].

Acting on an intrinsic desire for the right or good (correctly conceptualized) is the same as acting for the right reasons. So, let’s suppose that Michonne’s action of drawing the zombies’ attention to herself is morally right because it helps to save the villagers’ lives. On this view, Michonne is praiseworthy for acting in the relevant way to the extent that her action manifests an intrinsic desire for saving the villagers’ lives. Markovits (2010, p. 205) advances a similar view: [ . . . ] my action is morally worthy if and only if my motivating reasons for acting coincide with the reasons morally justifying the action—that is, if and only if I perform the action I morally ought to perform, for the (normative) reasons why it morally ought to be performed. My motivating

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

186

 

reasons for performing some action in this case will not be the duty-based reason “that the moral law requires it” but the reasons for which the moral law requires it.

Arpaly (and Schroeder) and Markovits accept, then, both of the following:² R R N. It is necessary, in order for an action to have moral worth, that this action is performed for reasons that make it right to act in the relevant way. R R S. It is sufficient, in order for an action to have moral worth, that this action is performed for reasons that make it right to act in the relevant way. These theses seem pretty plausible to me. The first one, in particular—or something in its neighborhood—seems potentially worth holding on to. But both of the theses are controversial, and they both have been challenged, by Paulina Sliwa, for instance. She challenges the second thesis—the sufficiency claim—by arguing that an action’s being motivated by a desire for the rightmaking features leaves the connection to rightness too precarious. She offers the following case (Sliwa 2015, p. 398): E. Jean’s friend missed her bus to work and frets over being late to an important meeting; coming late would be a great embarrassment to her. Wanting to spare her friend a great embarrassment, Jean gives her a ride. Let’s assume that giving her friend the ride is the right thing to do in these circumstances and that the fact that it spares her friend a major embarrassment makes it right.

Sliwa continues: Given Jean’s motivation, it’s not a fluke that Jean spared her friend a major embarrassment. But it is a fluke that she did the right thing. This is because there are plenty of circumstances in which [ . . . ] Jean’s motivating de re desire would lead her to do the wrong thing. For example, Jean ought not to murder her friend’s ex boyfriend, even if doing so would eliminate a major source of embarrassment in her friend’s life.

A possible response would be that we need to look at Jean’s motivational state as a whole and thus take into account how she acts in nearby possible worlds. The thought would be that Jean’s motivational state must be such

² Way (2017) also proposes a congenial account of creditworthiness of actions.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

187

that her tending to do the right thing is modally robust. However, consider the following (see Sliwa 2015, p. 399; the point comes from Markovits 2010): D-. A dog-lover risks his life to save a drowning stranger but he could easily have been motivated differently. In particular, had his dog been present, he would have been unwilling to abandon the dog—or perhaps too distracted to notice the stranger.

In this case, the agent’s motivation to do the right thing is not modally robust. Yet we wouldn’t wish to deny that her action has moral worth. Or that’s, anyway, Sliwa’s take on the matter. She suggests that the moral worth of an action depends on its actual motivation rather than on the motivational state of the agent in nearby possible worlds. If this is right, it may seem that the Right Reasons View doesn’t capture the non-accidental rightness of morally worthy actions in quite the right way.³ Zoë Johnson King also takes issue with the idea that doing the right thing for the right reasons suffices in order for the action to have moral worth. She, too, thinks that the Right Reasons View leaves the connection between morally worthy action and rightness too accidental. According to Johnson King (2018), the defenders of the Right Reasons View must appeal to something along the following lines: N-. For someone to non-accidentally perform an act with property F it is sufficient that (a) she is motivated to perform it by the fact that it has another property G and (b) as a matter of metaphysical fact, the act’s having property G makes it the case that it has property F. This principle, Johnson King suggests, is vulnerable to counterexamples. The following is, I think, in the spirit of those supplied by Johnson King: S. Michonne must choose between two swords. One of them has a sharp edge, the other one doesn’t. Michonne intrinsically desires to have a sword with a sharp edge. Swords that have a sharp edge have, in virtue of this, the property of being well suited for slicing tomatoes cleanly.

³ Another possible response would be to suggest that in E, Jean is not plausibly acting for reasons that make it right for her to give her friend a ride. Perhaps giving the ride is the right thing to do because it maximizes happiness, or because it benefits her friend without significantly harming others. It is fairly plausible, in any case, that whatever the credible account of the right-making features of Jean’s action turns out to be, Jean’s being sensitive to these features won’t easily result in her killing her friend’s ex. But perhaps we could devise other, more plausible, cases along the lines suggested by Sliwa. This should be possible, assuming that what’s right-making in one context may easily fail to be so in another.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

188

 

Michonne (weirdly) isn’t aware of this, but as she is driven by a desire for a sharp sword, she chooses a sword that’s well suited for slicing tomatoes cleanly. Does Michonne accidentally choose a sword that slices tomatoes cleanly? Well, given her interest in sharp edges, she is a pretty reliable tracker of cleanly-slicingness in swords. Yet it could have been so that sharp edges did not slice tomatoes cleanly. It seems possible for tomatoes to have been such that you only got them sliced cleanly with a somewhat blunt edge. If things had been so, Michonne would not have been interested in cleanlytomato-slicing swords. There is, then, an element of accidentality in Michonne’s picking a cleanly-tomato-slicing sword. One might think that likewise, there is an element of accidentality in the way in which someone who acts on an intrinsic desire for the right-making features of actions does the right thing. At least this might seem to be so given that the rightmaking features of morally right actions do not necessitate the rightness of these actions. One could also question R R N. Sliwa (2015, p. 410) offers the following counter example: T. Anna’s older sister is struggling with alcohol addiction; she lost her job, blew through her savings and is several months behind on rent. She asks Anna to “loan” her some money. Anna is conflicted. On the one hand, she does not want her sister to end up homeless. But she wonders whether her sister needs to feel the full consequences of her addiction to finally seek treatment. Moreover, Anna’s financial circumstances are modest and she has her own family to look after. Anna is uncertain about what the right thing to do is. She turns to a friend whom she knows to be trustworthy and to have good judgment for advice. Her friend tells Anna that she shouldn’t give her sister money. Anna’s friend is right [ . . . ]. Anna trusts her friend and acts on her advice.

On Sliwa’s (2015, pp. 411–12) view, Anna is morally praiseworthy, but she cannot act for the reasons that make it right for her to act in the way that she acts. Perhaps what explains the rightness of her action is that it’s good for her sister in the long run. But (we may suppose) Anna doesn’t act for this reason. She acts for the reason that her friend’s advice is such-and-such. However, even granting that R R N thus fails, one could insist that Anna still doesn’t need to, and ideally shouldn’t, act from a de dicto concern to do the right thing. She could be acting on the basis of de re concern for the right-making features plus a disjunctive belief. Her reason

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

189

could be that refraining from giving her sister money is best for her sister, or best for everyone in the long run, or what follows from a categorical imperative, or . . . It’s not wholly implausible that this will have to be the case if we wish to see Anna as being praiseworthy for doing the right thing. Acting for this disjunctive reason would not seem to amount to acting for a right-making reason, but the truth of the relevant disjunctive consideration would at least be evidence that refraining from giving her sister money would be the right thing for Anna to do. We could then propose something along the following lines: R R/E N. It is necessary, in order for an action to have moral worth, that this action is performed for reasons that make it right to act in the relevant way or provide (non-trivial) evidence that this is the right way to act. In what follows, I will assume that R R N is a plausible thesis. But what I say could easily be revised in accordance with the idea that we should rather be operating with something like R R/E N. None of what I have said in this section is meant to suggest that the Right Reasons View is, in my view, doomed, or that I’m entirely persuaded by the arguments from Sliwa and Johnson King. But their arguments raise interesting concerns regarding the Right Reasons View and motivate the proposals of their own, to which I turn next.

§3. Motive of Duty Sliwa and Johnson King both draw, from their respective challenges to the Right Reasons View, the conclusion that morally worthy action springs from a motive of duty. Sliwa (2015, p. 394) proposes the following: K V. A morally right action has moral worth if and only if it is motivated by concern for doing what’s right (conative requirement) and by knowledge that it is the right thing to do (knowledge requirement).

Johnson King (2018), on her part, advances the following account: A V.⁴ An act has moral worth if and only if it is an instance of someone’s deliberately doing—of someone’s trying and succeeding in doing—the right thing.

⁴ This isn’t Johnson King’s name for the view.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

190

 

On both views, an action with moral worth must be based on a moral belief—on a belief that the action in question is the right or a morally permissible action. On neither view is it sufficient for an action to have moral worth that it is based on a suitable moral belief. But according to Sliwa, it suffices that the action is performed on the basis of knowing that it is the right thing to do. And according to Johnson King, it suffices that the action is based on the agent’s belief that it is the morally right action, and that the agent succeeds, when guided by this belief, in performing the morally right action. I noted above that I find the Right Reasons View attractive. I also find attractive the idea that morally worthy actions are performed on the basis of moral judgments, or at least I find very attractive the ideas that it is possible for morally worthy actions to be based on moral beliefs, and that an action’s being performed on the basis of a judgment that it is morally right need not be a strike against the agent, morally speaking. Indeed, it also seems that an action’s moral worth can sometimes be enhanced by the fact that it is based on a belief that performing this action was the right thing to do. Moral thought can and at least often does play a central role in acting with moral worth. But even if the critics of the Right Reasons View are right in emphasizing the role of moral thought in morally worthy action, we should perhaps question their sufficiency claims. That is, we should perhaps admit that acting for the right-making reasons is a necessary condition for acting with moral worth. In the next two sections, I explore the prospects for combining this idea with the idea that acting with moral worth requires guidance by moral thought, or is at least compatible with such guidance. I suggest that given certain plausible assumptions, it is not possible to smoothly combine these ideas.

§4. For the Right Reasons, from Duty? According to what might be called the Standard Account, when we act on the basis of a certain belief (or: when our action is caused in the right way by our having a certain belief), the content of this belief gives our motivating reason, or the reason for which we act. Consider the following: Z 3. Michonne believes that the only way to prevent a horde of zombies from stumbling into a village is to attract the zombies into

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

191

another direction. She leads the zombies away from the village because she wants to protect the interests of sentient beings and believes that doing so will best serve this aim. In this scenario, Michonne would seem to lure the zombies away from the village partly for the reason that this will best protect the interests of sentient beings. It is quite plausible, then, that things work the same way also when someone performs an action partly because she thinks that this action is morally right. Consider: Z 4. Michonne believes that the only way to prevent a horde of zombies from stumbling into a village is to attract the zombies into another direction. She lures the zombies away from the village because she wants to protect the interests of sentient beings and believes that doing so will best serve this aim, and also because she wants to do the right thing and believes that leading the zombies away is the right thing to do. In this scenario, Michonne would seem to lure the zombies away from the village for the reason that this protects the interests of sentient creatures as well as for the reason that this is the morally right thing to do. At least this seems to be what we should say assuming the truth of the very plausible Standard Account and a cognitivist view about the nature of moral belief. According to cognitivism, moral beliefs are like beliefs about the shape or color or function or price of some object. Their job is to represent things as being a certain way. On a cognitivist view, just as I may represent a PS4 console as angular, pretty slim, white, relatively pricey, and as having among its functions that of allowing people to entertain themselves by playing video games, I may represent an action as being morally right. My beliefs about colors and shapes have the telos of fitting the way the world is, and offer me a map with which to steer in pursuit of my goals. My beliefs about rightness and wrongness work the same way. Or that’s what the cognitivist view about the nature of moral belief says. The cognitivist may think, of course, that moral beliefs are special in some ways. But they share their representational function with beliefs about shapes, colors, prices, etc.⁵

⁵ Some views are sometimes called ‘cognitivist’ despite emphasizing that moral beliefs are not representational in this way. I’m thinking of Terence Horgan and Mark Timmons’s (2006) “cognitivist expressivism,” in particular. Their view is an expressivist one, in my book. T. M. Scanlon’s (2014) and Derek Parfit’s (2011, 2017) views may also fit the bill (see Dreier 2015). If so, they aren’t cognitivist, as I use the term.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

192

 

If this kind of cognitivist view is true, then we should say that in Z 4, Michonne leads the zombies away from the village for the reason that this is the right thing to do. Moreover, to the extent that we focus on Michonne’s performing this action under the guidance of her belief that it is the right thing to do, Michonne’s ultimate, non-derivative motivation is to do the right thing, and she acts ultimately for this reason. She also non-derivatively wants to protect the interests of sentient beings, and so, in acting on this desire, she also acts ultimately for the reason that leading the zombies away is the best way to protect the interests of sentient creatures. Or this is what we should say about the relevant scenario, given the Standard Account and cognitivism about the nature of moral belief. But now consider the Right Reasons View again. On this view, an action has moral worth to the extent that it manifests an intrinsic desire for the right-making features of the action. Let’s suppose that leading the zombies away from the village is made right by the fact that it is the best way to protect the interests of sentient beings. Michonne’s action, in Z 4, then manifests an intrinsic desire for the features of the action that make it right. But it also manifests a desire to do what is right, de dicto, and this property—the property of being right—is not one of the right-making properties of the relevant action. To the extent that Michonne’s action is being guided by her belief that it would be right to lead the zombies away, she is not acting for a right kind of, right-making, reason. Consequently, one might think that, given the Right Reasons View, being motivated by a belief that a certain course of action would be morally right is a problem, or a minus, when it comes to moral worth. Whereas in Z 3, Michonne is doing the right thing for the reasons that make it the right thing to do, in Z 4 Michonne’s action is driven also by other considerations. Wouldn’t it be better, then, for Michonne not to be guided by moral thoughts? Indeed, it is not clear that the fact that Michonne’s action would be morally right is any kind of reason to perform it (see, e.g., Dancy 2004, Stratton-Lake 2000). Perhaps this consideration—that the action is morally right—only serves to tell us that there are some other considerations that are the reasons for performing the action. If this is correct, then adding moral thought to the motivational mixture convicts the agent of acting for (motivating) reasons that are not (normative) reasons to act in the relevant way. Perhaps there is no deep problem here. It is far from obvious what we should say about whether the rightness of an action can be a reason to perform it (see, e.g., Johnson King 2019). And a defender of the Right Reasons View could, in any case, suggest that it is okay to be motivated by

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

193

one’s belief that acting in a certain way would be morally right, given that one’s action is motivationally overdetermined, and suitably manifests a concern for the right-making features of the action. However, this doesn’t seem entirely satisfying. Let us consider another variation on Michonne and the zombies: Z 5. Michonne observes a vast horde of zombies roaming toward the village. The measly wall surrounding the village won’t hold against the horde, and the villagers’ escape routes are completely blocked. Michonne believes that the only way to prevent the horde from stumbling into the village is to attract the zombies into another direction. This is going to be very dangerous, and is bound to involve fighting a considerable—but probably manageable—number of zombies. Bracketing moral considerations, Michonne doesn’t see much reason to save the villagers, to whom she bears no special relations. Nevertheless Michonne decides that given that the lives of the villagers are at stake, and that she’ll probably be able to handle what zombies she’ll encounter while distracting the horde, this is the right thing to do. And so she leads the zombies away from the village. Michonne’s action in this scenario seems completely morally praiseworthy.⁶ But her motivations are not overdetermined in that she would be willing to protect the villagers also independently of moral considerations. It takes her some moral thought to come up with sufficient motivation to risk her life by engaging with the zombies. It is only because she believes that leading the zombies away is the right thing to do that she is motivated to do so, to a sufficiently high degree. It seems, then, that being motivated by a judgment that it would be right to act in a certain way, or being moved by duty, is not just something that isn’t necessarily detrimental to the moral worth of an action. The moral worth of an action can be due to the fact that it springs from such motives. Moreover, even if Michonne’s desire to protect the villagers was sufficiently strong independently of her moral thought, her acting also on the basis of her judgment that leading the zombies away is the right thing to do would plausibly enhance the moral worth of her action. (I’ll spare you

⁶ Note that the description of the case doesn’t involve the claim that Michonne lacks nonderivative concern for the villagers’ well-being. The relevant concern of hers would just be a morally laden concern.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

194

 

the details of the relevant zombie case.) These phenomena seem to be difficult to account for, given the Standard Account plus cognitivism plus the Right Reasons View.

§5. From Duty, for the Right Reasons? We may also have a look at these issues from the perspective of the Knowledge View and the Accomplishment View—of views that emphasize dutiful motivation, as we might put it. These views also face problems given the Standard Account of motivating reasons and a cognitivist account of moral judgment. Or let’s rather put it this way: Even if morally worthy actions may be performed for the reason that they are right, and even if it is a necessary condition of an action’s having moral worth that it is dutifully motivated, one might still think that we should make room for the idea that acting for the right-making reasons is also essential to moral worth. And it’s not clear how we can make adequate sense of this given the combination of the Standard View and cognitivism. Let’s first consider the idea that it’s sufficient, in order for one’s action to max out in terms of moral worth, that the agent tries to do the right thing and succeeds in this. Or the idea that it’s sufficient, in order for one’s action to max out in terms of moral worth, that the agent does the right thing on the basis of her knowing that it’s the right thing to do. Assuming that we can have knowledge of what is right and wrong through testimony, Michonne should then be morally praiseworthy for doing the right thing in the following scenario: Z 6. Michonne observes a vast horde of zombies roaming toward the village. The measly wall surrounding the village won’t hold against the horde, and the villagers’ escape routes are completely blocked. Michonne’s Guru tells her that the right thing for her to do is to lead the zombies away from the village, but she doesn’t tell Michonne why this is right. Michonne intrinsically desires to do the right thing, and so she decides to lead the zombies away from the village. It seems problematic to suggest that Michonne’s action lacks nothing in terms of moral worth. This is a familiar idea from the literature on moral testimony. While many writers on the topic believe that it is possible to transmit moral knowledge through testimony, there is widespread (although

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

195

by no means universal) agreement that there is something problematic about deference in moral matters. It has been suggested that the problem has to do with the moral quality of the action, or with such action being faulty with regard to praiseworthiness or moral worth, and that the explanation for this is that acting on the basis of moral testimony precludes having understanding of why it is right to act in the relevant way (see Hills 2009, McGrath 2011). Even if this is not the problem with moral deference (see, e.g., Fletcher 2016), this is plausibly a problem. Of course there are dissenters. Perhaps there is nothing problematic about acting on the basis of moral testimony (see, e.g., Sliwa 2012, 2015). But the idea that action that is based on moral beliefs acquired through moral testimony therefore are lacking moral worth-wise is not implausible. One could propose that while acting from a motive of duty is a necessary condition for performing an action with moral worth, performing such action—or at least maxing out in terms of moral worth—also requires acting for reasons that make it right to act in the relevant way. However, simply being motivated to perform an action both by the fact that it is the right thing to do as well as by its having certain features such that make it the right thing to do doesn’t seem sufficient. Here’s Michonne again: Z 7. Michonne observes a vast horde of zombies roaming toward the village. The measly wall surrounding the village won’t hold against the horde, and the villagers’ escape routes are completely blocked. Michonne’s Guru tells her that the right thing for her to do is to lead the zombies away from the village, but she doesn’t tell Michonne why this is right. She omits the information that leading the zombies away is the right thing to do because it’s the best way to protect the interests of sentient beings. Michonne intrinsically desires to do the right thing, and so she decides to lead the zombies away from the village. She also intrinsically desires to protect the interests of sentient beings, and believes that leading the zombies away serves this goal, too. But she has no idea that leading the zombies away is the right thing to do because it is the best way to protect the interests of sentient beings. According to Johnson King (2018), when we act for the right-making reasons, our doing the right thing may nevertheless be accidental. So, perhaps in Z 3, Michonne is praiseworthy for protecting the interests of sentient beings, but she may nevertheless fail to be praiseworthy for doing the right thing. Likewise, one might suggest, when, in Z 7, Michonne

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

196

 

performs the right action for the reason that it is right, she is praiseworthy for doing the right thing. One might suggest that as she also performs this action for the reasons that make it right, she may, moreover, be praiseworthy for protecting the interests of the sentient beings. Yet she fails to be praiseworthy for doing the right thing for the right reasons. Or: she fails to be praiseworthy, in doing the right thing, for acting for the right reasons. And one might suggest that this amounts to a failure with regard to the moral worth of her action. We may also note that it’s not sufficient, in order to save the idea that an action has moral worth to the extent that it is performed both for the reason that it is right and for the right-making reasons, to add that the agent must believe that the right-making reasons make her action right. For consider one more variation on Michonne and the zombies. This is just like Z 7, with the following addition: Z 8. [Z 7 . . . ] However, as she starts heading off to the woods, making noise and drawing her sword from its scabbard, her Guru yells after her: “It’s the right thing to do because it best protects the interests of sentient beings!” As Michonne runs into the woods waving her sword at the zombies, her actions have not gained any extra moral worth thanks to her now having formed the true belief (we’re assuming) that leading the zombies away is the right thing to do because it best protects the interests of sentient beings. She is responsive to those considerations, alright. And she is guided by her moral judgment. But what is lacking, morally speaking, in Michonne’s action, is being guided by moral judgment in a way that would manifest responsiveness to the considerations that make it right to act in the relevant way. Again we see that combining the Standard Account with a cognitivist account of moral thought spells trouble for understanding moral worth. Of course some are happy with the Right Reasons View and see no need to accommodate any significant role for moral judgment when capturing the idea of moral worth. And some others might not be too troubled by the fact that this combination of views makes it difficult to see how one could act, in virtue of being guided by judgments about what’s right, for the reasons that make it right to act in the relevant ways. But these options seem costly. In the next section I suggest that things start to look quite different, and potentially more flattering to us qua moral agents, if we drop the assumption of cognitivism and try casting these issues in an expressivist light.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

197

§6. Metaethics Matters Suppose we accept the following: R . For S to believe that she ought to perform an action, φ, is (a) for S to have a (suitable) desire-like attitude toward φ-ing, where this attitude is based on S’s beliefs about the descriptive properties of φ-ing (and perhaps accompanied by certain other attitudes), or (b) for S to be in a ‘relational’ state realized by a (suitable) desire-like attitude toward actions with some descriptive property plus a belief that φ-ing would have that property (where this relational state perhaps needs to be accompanied by certain other attitudes). Let’s unpack that a little. First, this is a disjunctive view that covers both views that take having ought-beliefs to consist in having a certain kind of desire-like attitude (see, e.g., Blackburn 1998, Lenman 2003), and views that take it to consist in having one’s desires and descriptive beliefs related in a certain way (see, e.g., Schroeder 2013, Toppinen 2013, Ridge 2014). The former sort of view is sometimes called pure expressivism; the latter sort of view often goes by the name relational expressivism. The latter view perhaps is not very familiar. I should illustrate it a little. So, suppose Michonne and Rick both plan, if in Michonne’s shoes, to lead the zombies away from the village.⁷ But whereas Michonne plans to do this because she plans to protect the interests of sentient beings and believes that leading the zombies away from the village best protects the interests of sentient beings, Rick plans to do this because he plans to perform actions that follow from Categorical Imperative and believes that in Michonne’s circumstances leading the zombies away from the village follows from Categorical Imperative. While the contents of Michonne’s and Rick’s respective plans and descriptive beliefs differ, their plans and descriptive beliefs are similarly related to each other. They both are in the same relational state of mind in that they both plan to do actions with a certain descriptive property and believe that leading the zombies away from the village has that property. Their minds just realize this multiply realizable relational state in different ways. The pure and the relational view are quite similar. Just like the relational view, the pure view, too, involves the idea that in order for Michonne to

⁷ The talk of plans and shoes is, of course, inspired by Gibbard (2003).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

198

 

believe that she ought to perform some action, her desire-like attitudes and descriptive beliefs must be related in a certain kind of way. And neither of the views dictates what the contents of the relevantly related attitudes and beliefs are to be. Perhaps the pure view is a relational view of sorts; perhaps the relational view is not that impure. The disjunctive expressivist view, as stated above, does not say anything about specifically moral beliefs. Here’s a somewhat revised thesis: R  . For S to believe that φ-ing would be morally right in her circumstances is for S to have a (suitable) desire-like attitude toward φ-ing, where this attitude is based on S’s beliefs about suitable sorts of descriptive properties of φ-ing (and perhaps accompanied by certain other attitudes), or for S to be in a ‘relational’ state realized by a (suitable) desire-like attitude toward actions with some suitable sort of descriptive property plus a belief that φ-ing would have that property (where this relational state perhaps needs to be accompanied by certain other attitudes). If Michonne’s ultimate grounds for endorsing leading the zombies away have to do with protecting the interests of sentient beings, this already gives a moral flavor to her attitude. Also, we could demand that in order for Michonne’s attitude to count as a moral belief, this attitude would need to have certain further properties—such as being stable to some sufficient degree, or being accompanied by certain higher-order attitudes, or by attitudes demanding similar responses from others, etc. (cf. Blackburn 1998). The issues concerning how, exactly, normative judgments more generally, and moral judgments, in particular, should be understood within an expressivist framework are very much open to debate. It’s also not clear that any very good expressivist account is available. But I shall adopt the kind of view outlined above as my toy theory. I hope it’s useful enough in order to illustrate how casting the issues concerning moral worth in an expressivist light may turn out to have interesting results. Given an expressivist view of the relevant kind, Michonne’s belief that it would be morally right to lead the zombies away may consist, roughly, in her approving of leading the zombies away, where this approval is based on her belief that, among her options, this action would best protect the interests of sentient beings. Or it may consist in her approving of actions that best protect the interests of sentient beings, plus her belief that leading the zombies away has that property. By the Standard Account of motivating reasons, if Michonne now leads the zombies away, on the basis of her believing that this

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

199

would be the right thing to do, this is just a matter of her doing so for the reason that this would best protect the interests of sentient beings. Given that Michonne’s attitudes really count as having a belief that it would be right to lead the zombies away, the Standard Account suggests that Michonne performs this action also for the reason that it is the right thing to do. But on the expressivist view, Michonne’s acting on this reason amounts to nothing over and above her action being rationally explained by her having suitably related desire-like attitudes and descriptive beliefs (where these must perhaps be accompanied by certain other attitudes). In our example, Michonne’s action springs from her desire-like attitudes plus her belief that leading the zombies away best protects the interests of sentient beings. And so, Michonne’s performing the action of leading the zombies away for the reason that it’s the right thing to do amounts to nothing over and above Michonne’s being moved to act, through certain attitudes of a suitable sort, for the reason that leading the zombies away best protects the interests of sentient beings. By contrast with the cognitivist understanding of the case, Michonne’s being moved by a belief about the moral rightness of leading the zombies away doesn’t now add an independent motivating reason into the picture. Rather, Michonne’s action may be guided by her moral judgment, and be, in virtue of this very fact, performed for the reason that it optimally protects the interests of sentient beings. Michonne may act on the basis of her rightnessjudgment, and—in virtue of this very fact—act for a right-making reason.⁸ ⁸ How about cases where the agent is uncertain about what the right-making features are? We could utilize the idea that the agent may have a disjunctive belief about the potential rightmakers, plus the thought that it suffices for acting with moral worth that the agent is being sensitive to facts that give (non-trivial) evidence that she is acting in a morally right way (see §2, above). This isn’t wholly satisfying, though. We would also want to know how to understand (i) cases where the agent cannot articulate her grounds for thinking that it would be right to act in the relevant way, and (ii) cases where the agent is fundamentally morally uncertain, or uncertain about which features would be morally relevant, or potentially right-making. The former issue (i) may not be so difficult to deal with. Michael Ridge (2014, 2015), who defends a form of relational expressivism, proposes that the descriptive belief component in normative thought usually involves a demonstrative element and picks out the properties that the thinker is sensitive to without the identity of these properties being clear to the thinker. For analogy, consider hearing a piece of music and thinking “I just love music like that!” while being unable to articulate what it is about the music that is so awesome. Now, likewise, when Michonne thinks that distracting the zombies would be morally right, this might be a matter of her believing that distracting the zombies would be like that (mentally “pointing” toward the option of distracting zombies) or like those (mentally pointing toward certain actions that in her view are paradigms of right action)—where what it is to be like that or like those would be determined by Michonne’s relevant desire-like attitudes. What should we say, in this case, about the reasons for which Michonne acts? On the one hand, it is natural to say that her motivating reasons are rather inarticulate: she distracts the zombies because the action has that property—one that she

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

200

 

This allows us to see many of the challenges facing the different accounts of moral worth in a new light. In §5, I suggested that it is difficult, for someone attracted to the idea that acting with moral worth involves acting on moral thought (or from a motive of duty), to make entirely satisfying sense of the idea that when someone acts with moral worth, she is praiseworthy for doing the right thing for the right reasons. But as I have just explained, once we understand beliefs about what’s morally right in an expressivist vein, along the lines suggested above, capturing this idea becomes easy. In §4, I also suggested that if we are attracted to the Right Reasons View as well as to a cognitivist account of moral belief, we should worry about the idea that acting with moral worth requires acting on moral thoughts. On the cognitivist picture, when Michonne acts on her belief about moral rightness, this isn’t just a matter of her acting for potentially right-making reasons. This naturally raises the worry that acting on such belief is a minus, when it comes to moral worth—a manifestation of moral fetishism (cf. Smith 1994, pp. 71–6). Or at least this suggests that acting on such a belief cannot be essential to, or enhance, the moral worth of an action. Things look different, given an appropriate expressivist view. When Michonne acts on the basis of her belief that leading the zombies away from the village would be morally right, this may just amount to her acting for the reason that this is the best way to protect the interests of sentient beings. Michonne’s fundamental motivations can be entirely based on the right-making features of her action. Moreover, this can be so even if Michonne would not be (sufficiently) motivated to risk her life in protecting the villagers independently of moral considerations (see Z 5). We might now also be in a position to explain, in a way consistent with the idea that acting with moral worth involves acting for the right-making reasons, how it is that an action’s being motivated by considerations of

can only pick out using moral terminology. On the other hand, if what Michonne is being responsive to is the property of being best promotive of the interests of sentient beings, then it is also plausible to say that she distracts the zombies for the reason that it best promotes the interests of sentient beings. She’s just not aware of what it is that motivates her. Either way, Michonne may end up acting for the right-making reasons on the basis of her rightnessjudgment: we might think that her action is right because it is like that—or, if we are capable of a richer articulation: because it is best promotive of the interests of sentient beings (or whatever). We might then have differing views about the relevance of the level of articulation in the agent’s motivating reasons to questions of moral worth; I shall not attempt any discussion of this issue here. As for (ii), here I can only note, rather disappointingly, that questions regarding fundamental moral uncertainty seem to be very good questions for expressivists quite generally. It remains to be seen how these questions should be answered (for a recent take on this, see Ridge 2018), and what implications the answers will have for questions regarding moral worth.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

201

rightness can enhance its moral worth. Suppose that moral beliefs involve especially strong or stable attitudes, or attitudes that are coupled with resistance to change, or occupy an especially central place in our web of attitudes. Morally laden motivations will then be especially robust. If a right action’s being motivated in such a way that the agent could easily have ended up doing the wrong thing is a minus with regard to the action’s degree of moral worth, then an action’s being morally motivated would, on this picture, turn out to be a moral plus. (However, see Sliwa’s example of the dog-lover in §2, above.) Choice between cognitivism and expressivism may be relevant also for the prospects of the idea that acting for the right-making reasons is sufficient for acting with moral worth. For if we understand acting for a reason in a demanding enough way—as requiring sophisticated enough attitudes on part of the agent—it is not such a big stretch to claim that when someone (Huck Finn, say) acts for a certain reason (helps Jim for the reason that Jim is a human being just like himself), he at least implicitly believes, of the fact that is his motivating reason, that it is a normative reason for acting in the relevant way. This kind of idea would seem to be more plausible given an expressivist view of normative beliefs than it is on the assumption of cognitivism. If so, accepting that acting on the right-making reasons suffices for acting with moral worth might be somewhat easier for those who are inclined to think that one only acts with moral worth when one’s actions are guided by normative thinking.

§7. Conclusion My aim has been to suggest that our stance on the issue of cognitivism vs. expressivism in metaethics may have interesting consequences for our attempts to form an adequate understanding of what it is to act with moral worth, or to be praiseworthy for doing the right thing. In particular, I’ve suggested that some of the difficulties in forming such an understanding may evaporate given an expressivist account of moral thinking. Expressivism is, of course, super controversial. But it’s also— uncontroversially—a significant contender in contemporary metanormative theory. Importantly, accepting expressivism doesn’t require accepting that there are no moral truths, or that moral truth is subjective or relative, or that there’s no reason to act in accordance with the demands of morality. As is quite familiar these days, expressivists tend to be quasi-realists. They wish to

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

202

 

earn the right to speak as the realist does—to such extent that many now find it difficult to draw any distinction between expressivist and realist views. There is a distinction, though, in how these views explain the nature of normative thought (see above and, e.g., Dreier 2004). And this difference is significant in the context of theorizing about moral worth. If what I’ve said is right, expressivism gives us more leeway with regard to doing justice both to the idea that acting with moral worth is a matter of acting for reasons that make it right to act in the relevant way, and to the idea that acting with moral worth involves moral thought. Perhaps, if we keep the expressivist option on the table, this makes it easier to approach the insights produced by the work on moral worth in a reconciliatory spirit.⁹

References Arpaly, N., 2003. Unprincipled Virtue. New York: Oxford University Press. Arpaly, N. and T. Schroeder, 2014. In Praise of Desire. New York: Oxford University Press. Blackburn, S., 1998. Ruling Passions. Oxford: Oxford University Press. Dancy, J., 2004. Ethics without Principles. Oxford: Oxford University Press. Dreier, J., 2004. “Metaethics and the Problem of Creeping Minimalism,” Philosophical Perspectives 18: 23–44. Dreier, J., 2015. “Another World: The Metaethics and Metametaethics of Reasons Fundamentalism,” in R. N. Johnson and M. Smith (eds), Passions and Projections: Themes from the Philosophy of Simon Blackburn. Oxford: Oxford University Press, ch. 9. Fletcher, G., 2016. “Moral Testimony: Once More with Feeling,” in R. ShaferLandau (ed.), Oxford Studies in Metaethics, Volume 11. Oxford: Oxford University Press, ch. 3. Gibbard, A., 2003. Thinking How to Live. Cambridge, MA: Harvard University Press. Hills, A., 2009. “Moral Testimony and Moral Epistemology,” Ethics 120: 94–127. Horgan, T. and M. Timmons, 2006. “Cognitivist Expressivism,” in T. Horgan and M. Timmons (eds), Metaethics after Moore. Oxford: Oxford University Press, ch. 12. ⁹ Thanks to Ron Aboodi, Graham Bex-Priestley, Thomas Hurka, Zoë Johnson King, Jimmy Lenman, Mark Schroeder, Robert Stern, the rest of the audiences at WiNE 2018, University of Sheffield, and University of Turku, and the anonymous reviewers for OUP for helpful comments.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

     

203

Isserow, J., ms. “Pluralism about Moral Worth.” Johnson King, Z., 2018. “Accidentally Doing the Right Thing,” Philosophy and Phenomenological Research, https://doi.org/10.1111/phpr.12535. Johnson King, Z., 2019. “We Can Have Our Buck and Pass It, Too,” in R. ShaferLandau (ed.), Oxford Studies in Metaethics, Volume 14. Oxford: Oxford University Press, ch. 8. Lenman, J., 2003. “Non-Cognitivism and the Dimensions of Evaluative Judgment,” in J. Dreier and D. Estlund (eds): BEARS, http://www.brown.edu/ Departments/Philosophy/bears/homepage.html. McGrath, S., 2011. “Skepticism about Moral Expertise as a Puzzle for Moral Realism,” Journal of Philosophy 108: 111–37. Markovits, J., 2010. “Acting for the Right Reasons,” Philosophical Review 119: 201–42. Parfit, D., 2011. On What Matters, Volume 2. Oxford: Oxford University Press. Parfit, D., 2017. On What Matters, Volume 3. Oxford: Oxford University Press. Ridge, M., 2014. Impassioned Belief. Oxford: Oxford University Press. Ridge, M., 2015. “Summary,” Analysis 75: 433–42. Ridge, M., 2018. “Normative Certitude for Expressivists,” Synthese, https://link. springer.com/article/10.1007/s11229-018-1884-7. Scanlon, T. M., 2014. Being Realistic about Reasons. Oxford: Oxford University Press. Schroeder, M., 2013. “Tempered Expressivism,” in R. Shafer-Landau (ed.): Oxford Studies in Metaethics, Volume 8. Oxford: Oxford University Press, 283–314. Sliwa, P., 2012. “In Defense of Moral Testimony,” Philosophical Studies 158: 175–95. Sliwa, P., 2015. “Moral Worth and Moral Knowledge,” Philosophy and Phenomenological Research 93: 393–418. Smith, M., 1994. The Moral Problem. Oxford: Blackwell. Stratton-Lake, P., 2000. Kant, Duty and Moral Worth. London: Routledge. Toppinen, T., 2013. “Believing in Expressivism,” in R. Shafer-Landau (ed.): Oxford Studies in Metaethics, Volume 8. Oxford: Oxford University Press, 252–82. Way, J., 2017. “Creditworthiness and Matching Principles,” in M. Timmons (ed.): Oxford Studies in Normative Ethics, Volume 7. Oxford: Oxford University Press, 207–28.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

10 Deontological Decision Theory and the Grounds of Subjective Permissibility Seth Lazar

1. Introduction In virtue of what is an act subjectively permissible? What makes subjectively right acts subjectively right?¹ Obviously, one ground of subjective permissibility is the agent’s epistemic state, or the probabilities justified by her evidence, or something else that explains why this is subjective, rather than objective permissibility. Set that aside for now. Suppose that the basic formula for subjective permissibility is some function of the agent’s epistemic states, and some other input. What is that other input? In this chapter I consider two possibilities. Both assume that the other input is objective. Working out whether an action is subjectively permissible means, essentially, probabilistically discounting an objective value (or quasivalue) function. The first, which I call the “verdicts” approach, grounds that objective value function in verdicts about objective permissibility. In its simplest form, it considers the possible outcomes of your options, and assigns each a value based on whether you will have acted objectively permissibly, should that outcome be realized. I’ll discuss more complicated forms below. The second approach grounds the objective value function in your objective reasons for action. It ranks your acts according to the probability-weighted average of the objective reasons that would apply to them under the different possible states of the world. This presupposes that for any given act and any given state, we can assign some kind of weight (however imprecise) to the balance of objective reasons in favor of, or against, that act.

¹ I use “right” and “permissible” interchangeably. Seth Lazar, Deontological Decision Theory and the Grounds of Subjective Permissibility In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0010

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

205

I’m going to argue against the verdicts approach, and for the reasons approach. Matters are not cut and dried, however, and I will explain the initial appeal of the verdicts approach, and note that verdicts might ultimately have some role to play in our criterion of subjective permissibility, albeit alongside objective reasons.

2. Background Defining subjective permissibility is easier if we start with objective permissibility. By this I mean permissibility in light of all the facts. This does not mean, as has been claimed, that objective permissibility is equivalent to permissibility if the agent knew all the facts.² Sometimes the agent’s knowledge or ignorance can be relevant to objective permissibility, as when we think that one can act objectively wrongly in virtue of imposing a risk that does not result in (other) harm, for example by playing Russian roulette with a sleeping, unaware victim.³ By changing the agent’s epistemic states, we can change the salient facts. This is crucial. Subjective permissibility, then, is permissibility in light of the agent’s limited epistemic perspective. A full theory of subjective permissibility would say precisely how the agent’s perspective is limited. I won’t defend a position on that question here, but will advertise an allegiance: I favor evidence-relativity. I think that the morally most significant criterion of subjective permissibility deploys probabilities grounded in the evidence available to the agent if she did the research appropriate to the stakes, and met some minimal standards of reasoning.⁴ However, the rest of this chapter does not presuppose any particular sub-species of subjective permissibility. It applies to them all. My project is to develop a criterion of subjective permissibility for deontological ethics. I can’t hope to reach a universally satisfying definition of deontology. Instead, I will identify the kind of deontological ethics for which I want to develop a decision theory, while acknowledging that there are many other kinds. I like the slogan that the right is prior to the good—or, ² E.g. Derek Parfit, On What Matters (Oxford: Oxford University Press, 2011). ³ On this, see Seth Lazar, “Risky Killing and the Ethics of War,” Ethics 126/1 (2015), 91–117; Seth Lazar, “Risky Killing: How Risks Worsen Objective Rights-Violations,” Journal of Moral Philosophy 16/1 (2019), 1–26. ⁴ On “research appropriate to the stakes,” see Alexander A. Guerrero, “Don’t Know, Don’t Kill: Moral Ignorance, Culpability, and Caution,” Philosophical Studies 136/1 (2007), 59–97.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

206

 

more precisely, it is sometimes permissible, and sometimes required, not to perform the agent-neutrally best act. Standard consequentialists deny both of these claims.⁵ In more detail: an act is agent-neutrally best when it is best in virtue of properties that have the same force for everyone. My kind of deontologists think that often an act might be agent-neutrally best, but not best all-thingsconsidered, because of agent-relative reasons that outweigh the agentneutral reasons (agent-relative reasons either apply, or have particular force, to specific agents in virtue of facts about them that are not universally shared).⁶ For example, if an act involves using others for one’s own purposes, or intending harm, or breaching special obligations to those with whom we share valuable or promissory relationships, then it might be agentneutrally best, but not all-things-considered best—so we are required (and a fortiori permitted) not to do the agent-neutrally best act. At other times we are merely permitted not to do what is agent-neutrally best, most notably because it involves excessive personal cost, or because it is better only in virtue of benefits to ourselves.⁷ Of course, not all deontologists agree on all of these points, and not all consequentialists take the opposite view. Indeed, some claim the mantle of consequentialism while taking the apparently deontological stance on each of these claims.⁸ That’s okay. A rose by any other name, etc. Why, then, should this kind of deontologist want a criterion of subjective permissibility? All of our actual decisions are made under doubt, so obviously moral philosophers have to discuss risk and uncertainty. But why develop a criterion of subjective permissibility? One sometimes hears that moral philosophy should be action-guiding. But we need to be careful here. Moral philosophy can’t really be action-guiding in any literal sense.⁹ When ⁵ This way of understanding the divide is most clearly articulated by Samuel Scheffler, The Rejection of Consequentialism (Oxford: Clarendon Press, 1994). ⁶ This is probably more specific than I need to be; my version of decision theory will work for agent-neutral deontologists as well—they just have to come up with agent-neutral reasons that play in their theory the role that agent-relative reasons play in mine. See F. M. Kamm, Intricate Ethics: Rights, Responsibilities, and Permissible Harm (New York: Oxford University Press, 2007); Victor Tadros, The Ends of Harm: The Moral Foundations of Criminal Law (Oxford: Oxford University Press, 2011). ⁷ Shelly Kagan, The Limits of Morality (Oxford: Clarendon Press, 1989). ⁸ E.g. Douglas W. Portmore, Commonsense Consequentialism: Wherein Morality Meets Rationality (Oxford: Oxford University Press, 2011). ⁹ Fred Feldman, “Actual Utility, the Objection from Impracticality, and the Move to Expected Utility,” Philosophical Studies 129/1 (2006), 49–79; Holly M. Smith, “Subjective Rightness,” Social Philosophy and Policy 27/2 (2010), 64–110; Elinor Mason, “Objectivism and Prospectivism About Rightness,” Journal of Ethics and Social Philosophy 7/2 (2013), 1–21;

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

207

you have to act, typically you don’t have time to philosophize. When the stakes are high enough, doing so may even be inappropriate. We shouldn’t expect moral philosophy to provide us with a decision procedure that we can use in realistic circumstances.¹⁰ But a moral theory can guide action under uncertainty without offering a decision procedure, if it can evaluate our options under those conditions. Even if our computational limitations (among other things) prevent us applying our criterion of subjective permissibility, having one means that in cooler moments we can reflect on which was the right choice to make in the circumstances. This can help us calibrate the decision procedures that we do use, as well as to judge our actions ex post according to standards that we were able to meet at the time. Of course, we should care about ranking our options under other descriptions too. For some purposes, we want to know whether, in light of all the facts, we acted permissibly. Indeed, a moral theory is more powerful and more convincing the more adaptable it is to different ways of describing and constraining our options. Besides the objective and subjective rankings, we might also want a moral ordering in light of other feasibility constraints—or even in light of the agent’s unwillingness to do what they really ought to do. No doubt there are other morally relevant ways to describe our options in any given choice. I think that a moral theory is stronger the more comprehensive it is, in part because the more comprehensive your moral theory, the greater your prospects for coherentist justification. However, some deontologists might still resist repackaging their views in a criterion of subjective permissibility. They are generally happy to offer a piecemeal hodgepodge of principles, and enjoin us to balance them as best we can.¹¹ Many would scoff at the idea that we could succinctly, accurately, and informatively give necessary and sufficient conditions for an act’s being subjectively (or objectively) permissible. I think this is a mistake. First, because criteria of permissibility reveal important structural properties of a moral theory that we would miss if we focused only on the ad hoc principles. These structural properties can help Andrew Sepielli, “Subjective Normativity and Action Guidance,” in Mark Timmons (ed.), Oxford Studies in Normative Ethics, Vol. II (Oxford: Oxford University Press, 2012), 45–73. ¹⁰ Peter Railton, “Alienation, Consequentialism, and the Demands of Morality,” Philosophy & Public Affairs 13/2 (1984), 134–71. ¹¹ Canonically, of course: W. D. Ross, The Right and the Good (Oxford: Clarendon Press, 2002).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

208

 

us reach reflective equilibrium—if your ad hoc principles commit you to implausible structural properties, then that’s a problem. They can also help extend our theory to novel and tricky cases. Second, a criterion of subjective permissibility is a going concern only if we can systematically combine our epistemic states with the salient objective inputs. And it’s an open and extremely interesting question whether judgment under risk and uncertainty has that much structure. Decision theorists generally think that it does: objective considerations should be discounted in proportion to their probability of being actualized.¹² That is a compelling idea, and my working hypothesis is that it is correct—perhaps with some modification to allow for non-neutral attitudes to risk, and always remembering that facts about the agent’s beliefs when acting can be relevant to the objective assessment of her act. But if that working hypothesis is wrong, and we can’t develop an alternative systematic way to feed probabilities into a criterion of subjective permissibility, that too is a very useful lesson to learn. Last question: What do I mean by grounds? Nothing fancy. My question is: In virtue of what, alongside the agent’s epistemic states, is an act subjectively permissible? What are the inputs that, together with the agent’s epistemic states, determine subjectively permissible action?

3. Objective Verdicts The verdicts approach is driven by a simple idea: when we don’t know crucial facts pertaining to the objective permissibility of our actions, we should try to act objectively permissibly.¹³ Objective right and wrong are our guide, even when the path is imperfectly lit. This underlying idea has three main virtues: it can be developed into an initially appealing criterion of subjective permissibility, which delivers plausible verdicts on challenging test cases; it resonates with a more structural deontological approach to ethics; and it responds to something that we surely do care about. The early versions of the verdicts approach were not promising, however. A surprising number of philosophers believed that, when acting under risk

¹² For an overlooked general argument to this effect, see Graham Oddie and Peter Milne, “Act and Value: Expectation and the Representability of Moral Theories,” Theoria 57/1–2 (1991), 42–76. For some important objections, see Sergio Tenenbaum, “Action, Deontology, and Risk: Against the Multiplicative Model,” Ethics 127/3 (2017), 674–707. ¹³ Sepielli, “Subjective Normativity and Action Guidance.”

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

209

and uncertainty, we should do what is most likely to be objectively right.¹⁴ That view was famously debunked by Frank Jackson’s “Drug Example.”¹⁵ A doctor can choose from three treatments to remedy a patient’s uncomfortable but not life-threatening condition. Drug A will incompletely cure the patient, but has no adverse side effects. One of drug B or C—the doctor doesn’t know which—will completely cure the patient. The other will kill her. The doctor knows that prescribing drug A is objectively impermissible. In light of all the facts, only prescribing the drug that will completely cure the patient is permissible. But clearly it would be wrong for the doctor to prescribe B or C, given the significant risk of killing the patient. So we cannot confine our attention only to the probability that one’s action is objectively right. We have to take downside risks into account as well. In response to Jackson, some writers have amended the original verdictsfirst principle. They argue that all objectively permissible options are equally permissible, so should be ranked equally. But impermissible options are not all equally impermissible, and we should reflect this in our criterion of subjective permissibility. Peter A. Graham introduced—without endorsing—a proposal that Kristian Olsen subsequently took up, according to which we should “minimize expected objective wrongness” (MEOW).¹⁶ This assumes that objectively wrong acts can be wrong to different degrees. Some might baulk at this, arguing that there are no “degrees of wrongness,” although acts can be more and less seriously objectively wrong. Charitably, we can treat these locutions as interchangeable. MEOW handles Jackson’s case nicely. Prescribing drug A is objectively wrong, but not very seriously objectively wrong. If the doctor tries to act objectively permissibly, and gets it wrong, killing the patient, then she has acted very seriously objectively wrongly. It’s very likely, then, that prescribing A minimizes expected objective wrongness. Although MEOW has, at present, few advocates, a few other philosophers endorse a principle that is typically extensionally equivalent: maximize

¹⁴ For the honor roll, see Smith, “Subjective Rightness,” n. 19. ¹⁵ Frank Jackson, “Decision-Theoretic Consequentialism and the Nearest and Dearest Objection,” Ethics 101/3 (1991), 461–82. ¹⁶ Peter A. Graham, “In Defense of Objectivism About Moral Obligation,” Ethics 121/1 (2010), 88–115; Kristian Olsen, “Subjective Rightness and Minimizing Expected Objective Wrongness,” Pacific Philosophical Quarterly 99/3 (2018), 417–441. To be clear, only Olsen thinks of this as a criterion of subjective rightness. Graham is skeptical about the category of subjective rightness. He proposes this as one thing that a morally conscientious person rationally ought to take into account.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

210

 

expected objective deontic value (MEODV).¹⁷ MEODV is equivalent to MEOW as long as one assigns the same deontic value to all outcomes in which you have acted objectively permissibly, and different deontic values to those in which you have acted objectively wrongly, insofar as they involve different degrees of objective wrongdoing. Any deontologist who wants to adopt standard decision theory must make this move, to accommodate within a maximizing framework the fact that, if two options are objectively permissible, you may choose either one (if they didn’t have the same deontic value, then the one with less would be impermissible).¹⁸ Hereafter, then, I will treat these versions of MEODV as equivalent to MEOW. MEOW has more going for it than its ability to deal with thought experiments that confounded its predecessors. If we want a decision theory for deontological ethics, then we should prefer one that appropriately reflects the idea that the right is prior to the good. And MEOW seems to do that: it represents objective verdicts with a ranking—the ranking is “the good”; the objective verdicts are “the right.” The good is grounded in the right, so the right is prior to the good. This is an appealing result for deontologists. What’s more, the verdicts approach responds to something that we care about. Looking back on risky action, when we know the outcome, it makes sense to regret that, even though you did the best you could, it turns out in the end that you acted objectively wrongly. And if you narrowly escape acting objectively wrongly, then, as Graham says, it makes sense to think yourself lucky that you came so close to doing something seriously objectively wrong, but managed to avoid it.¹⁹ All-out subjectivists will deny that there is any meaningful category of objective permissibility at all. But that seems too uncompromising. Surely one important way to rank our acts is in light of all the facts? And shouldn’t this have some bearing on how we rank them when the facts are unknown? The verdicts approach is grounded in something that very plausibly matters.

¹⁷ Horacio Spector, “Decisional Nonconsequentialism and the Risk Sensitivity of Obligation,” Social Philosophy and Policy 32/2 (2016), 91–128; Michael J. Zimmerman, Living with Uncertainty: The Moral Significance of Ignorance (Cambridge: Cambridge University Press, 2008); Oddie and Milne, “Act and Value”; Mark Colyvan, Damian Cox, and Katie Steele, “Modelling the Moral Dimension of Decisions,” Noûs 44/3 (2010), 503–29. Note that Spector’s view has some complexities that I cannot do justice to here. The key point, though, is that he does not assign positive weight to permissions. ¹⁸ See Seth Lazar, “Deontological Decision Theory and Agent-Centered Options,” Ethics 127/3 (2017), 579–609. ¹⁹ Graham, “In Defense of Objectivism.”

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

211

4. Against the Verdicts Approach Much favors the verdicts approach; but much also tells against it. I have three kinds of objection. First, a series of cases in which MEOW and cognate principles deliver incorrect judgments. Second, a theoretical diagnosis of why this is happening. And third, a general reason for skepticism about the verdicts approach. My first three cases show that MEOW is intuitively overdemanding, requiring us to make sacrifices that are intuitively not subjectively required. Here’s the first: A’s life is in danger; she can save herself only by killing B. A has grounds to believe that B might be responsible for the threat that A now faces.²⁰ Let’s assume that letting oneself be killed, in cases like this, is always objectively permissible. That’s a reasonable assumption, provided one has no dependents or outstanding obligations to others. But killing B clearly risks objective wrongdoing, because there’s a chance that B is innocent. So, killing B involves some risk of objective wrongdoing, while letting herself be killed involves no such risk. So the former option minimizes expected objective wrongness, unless the probability that B is responsible for the threat reaches 1. This seems absurdly restrictive. The lethal use of self-preserving force against those who might be responsible for the threat one faces can be permissible, even if there’s some non-trivial chance that they are not responsible. Second case: your life is threatened by a culpable aggressor.²¹ Spread out before you are a loaded Rocket Propelled Grenade (RPG) launcher and two pistols. Only one of the pistols is loaded; each is equally likely to be the one. If you fire the RPG, you will save yourself, but kill the aggressor. If you fire the loaded pistol, you will save yourself, only wounding the aggressor. If you try to fire the unloaded pistol, you will be killed (you won’t have time to reach the other weapons). As before, we’ll assume that you have no outstanding obligations to others. Letting yourself be killed is sure to be objectively permissible. Firing the RPG is objectively impermissible—if you have the alternative of saving yourself by merely wounding the aggressor (with the loaded pistol), then killing him fails to satisfy the necessity requirement on defensive force, and is impermissible. Trying either pistol is sure to be objectively permissible—either it amounts to letting oneself be ²⁰ Lazar, “Agent-Centered Options.” ²¹ I introduced this case, to make a different point, in Seth Lazar, “In Dubious Battle: Uncertainty and the Ethics of Killing,” Philosophical Studies 175/4 (2018), 859–83.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

212

 

killed, or to averting the threat by inflicting the least harm effective. So, for MEOW, it’s subjectively permissible either to fire a pistol or to do nothing, but it is subjectively impermissible to fire the RPG. That again seems wrong. One isn’t required to bear a 0.5 risk of being killed to spare a culpable attacker the difference between being killed and being wounded. The third case moves from the ethics of self-defense to the relationship between the supererogatory and our duties of rescue. You can give up your life to save five, who will otherwise surely die. But there is some chance that you’re responsible for their predicament. If you’re responsible, then you are in fact objectively required to sacrifice your life, so it is objectively wrong to do nothing. If you’re not responsible, then sacrificing your life is supererogatory. But it would definitely be objectively permissible (again, we’ll assume that you have no outstanding obligations to others). So you know that sacrificing your life is objectively permissible, but doing nothing might be objectively wrong. As long as the probability that you’re responsible for the five’s fate is greater than zero, MEOW says that you are subjectively required to sacrifice your life. But this is again too demanding. I’m allowed to take some risk of acting objectively wrongly in order to preserve my own life. One might be tempted by these cases to amend the “M” in MEOW, and adopt satisficing instead. Perhaps it should be: “don’t take excessive risks of acting objectively wrongly.” Others have given ample reason to reject satisficing in general, and I’ve written elsewhere about the special problems raised for satisficing by moral decision-making under risk.²² So I won’t dwell on this response at length. Instead I’ll give some more cases, to show that the problems with MEOW can’t be solved by satisficing. They all put pressure on the idea of “excessive” in that adaptation of MEOW. In virtue of what is a given risk excessive? Surely, in part, because of the good that can be realized if one runs that risk. But MEOW cannot accommodate this upside possibility, because it assigns the same value to all outcomes in which one acts objectively permissibly. Imagine, then, that a building is alight, with six people trapped inside— five in one room, one in another. When you enter the building, you will have to choose to go either left or right. If you choose correctly, then you’ll save all six and escape the building with no further cost to yourself (suppose that the door to the corridor linking the two rooms will close irreversibly

²² See e.g. Ben Bradley, “Against Satisficing Consequentialism,” Utilitas 18/2 (2006), 97–108.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

213

immediately after you pass through it). If you choose incorrectly, then you’ll save only one—if you try to find some way to open the door, and save the other five as well, then you’ll all die. The problem: you don’t know whether to go left or right. You do know, however, that the costs to you of entering the building are enough that doing so is supererogatory. You’re not objectively required to bear those costs, even if you would certainly save all six lives. If you can save more rather than less people, at no additional personal cost, then you are objectively required to save more—even if you’re already engaged in supererogatory behavior.²³ It’s objectively wrong to save fewer lives when you could costlessly save more, even if it would be permissible to save none at all. If that’s right, then entering the building risks objective wrongdoing—you might take the wrong turn, and save only one person. By contrast, staying out involves no risk of objective wrongdoing. So you’re subjectively required to stay outside, saving nobody, because trying to save risks objective wrongdoing. This is extremely counterintuitive. That a potentially supererogatory act involves some risk of objective wrongdoing, however small, cannot make it subjectively impermissible. Here’s another similar case. You can donate some amount of money to a charity or keep it for yourself. Donating is merely objectively permissible, not required—it is supererogatory. But suppose, as is often the case, that there is a small chance that your donation might cause harm (for example, by enabling aid workers who sexually exploit members of the affected population). If that happens, then donating is objectively impermissible. Again, you can be sure of avoiding objective wrongdoing only by not donating at all. Donating risks objective wrongdoing; keeping your resources for yourself does not. So you’re subjectively required to do the latter. With the first set of cases, MEOW was too demanding: it required sacrifices that seem not to be subjectively required. In the second set of cases, MEOW is too restrictive: it prohibits making sacrifices that seem subjectively permissible. Does satisficing help here? Suppose that we can tolerate some risk of objective wrongdoing. At a first pass, it looks as though we could preserve some plausible responses in these cases—you’d be

²³ This is the least controversial thesis in this domain—on the broader literature, including some more controversial possibilities, see Theron Pummer, “Whether and Where to Give,” Philosophy & Public Affairs 44/1 (2016), 77–95; Kagan, Limits of Morality.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

214

 

permitted to engage in the supererogatory action, as long as it didn’t risk too much objective wrongdoing. However, if we adopt this approach, it makes no difference how much good you can expect to do if your act proves to be objectively permissible. And that can’t be right. Consider the burning building case. Suppose now that there are one hundred people in one room, and one in the other. The cost to you of entering the building is still great enough to make attempting the rescue supererogatory. But the additional ninety-five lives that you could save if things go well (relative to the first version of the case) should make a difference to whether you are permitted to enter the building. Other things equal, it should be easier to justify entering the building than when only six lives are at stake. But, on some plausible assumptions about our duties of rescue, MEOW would deliver exactly the opposite verdict. This is because the greater the number of people you could save at no additional cost to yourself, the greater your objective wrongdoing if you fail to save them. Failing to save ninety-five people when you could have done so without extra personal cost is objectively worse than failing to save five people in the same circumstances.²⁴ So according to MEOW, as we increase the number of people you could potentially save, it becomes harder, not easier, to justify entering the building. This is weird. The same problem arises in the charity case—on the satisficing approach, the amount of objective wrongdoing you are permitted to risk cannot vary depending on how much good your donation will do, if it is objectively permissible. One might try to salvage MEOW by arguing that our objective verdicts in these cases should be sensitive to our uncertainty when acting. Perhaps it is not objectively wrong to contribute to a charity that goes on to harm people, if the probability of its doing so is sufficiently low; perhaps it’s not objectively wrong to save fewer rather than more lives, when one lacks an alternative that has the same personal costs but better prospects of saving the greater number; perhaps we could run a similar line in the first three cases too.

²⁴ Here’s an interesting variation; I don’t know what MEOW should say about this kind of case. Suppose the number in the larger group is high enough that you’re actually objectively required to enter the building in order to save them. But you still don’t know whether you’ll save them by going left or right. I think MEOW should say that, in this case, you’re subjectively required to go into the building. But that has the weird result that as you increase the number of people in the larger group, the subjective requirement not to enter the building increases and increases in force, until suddenly the number is great enough that you’d be objectively required to save them if you could, and you’re subjectively required to enter the building.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

215

I doubt whether any adherent of MEOW would make this argument. The verdicts approach grounds judgments of subjective permissibility in verdicts of objective permissibility. If those verdicts of objective permissibility are themselves grounded in a balancing of probability-weighted objective reasons, then the verdicts approach is no longer distinctive. This would effectively build a distinct, objective-reasons-based criterion of subjective permissibility into the verdict of objective permissibility. If our account of objective permissibility is genuinely to be independent of our account of subjective permissibility, then it had better be possible for one to act subjectively permissibly, but objectively impermissibly, and vice versa. The proposed fix for MEOW steers perilously close to ruling that possibility out. Facts about one’s epistemic state when acting are sometimes relevant to how seriously objectively wrong an act is. But this proposal is more extreme, and defeats the whole object of grounding subjective permissibility in verdicts of objective permissibility. MEOW has implausible implications across a range of realistic cases. That is a problem. But what explains its failings? And is it a problem for the verdicts approach in general, rather than just with MEOW? I think there are two underlying problems for MEOW. First, sometimes we are quite sure that an act is objectively permissible given a particular set of facts, but equally confident that it is only permissible, not required. That an act ϕ is permitted does not merely entail that one is not required not to ϕ. It can, in some cases, mean explicitly that one is not required to ϕ. MEOW has no space for this distinct deontic status. It recognizes permissibility, impermissibility, and degrees of impermissibility. But it does not recognize mere permissibility adequately. Could we develop a better verdicts approach by building in this additional verdict, and including mere permissions, regular permissions, and impermissibility?²⁵ It’s possible, but there’s another problem around the corner. MEOW ignores the upside in some risky moral choices. In response to cases like Jackson’s, Olsen and others have seen the need to differentiate between different downside scenarios, recognizing degrees of wrongness. And this move helps in some other cases as well, since often the good news if things go right is just the counterpart of the bad news if things go wrong: we can ²⁵ This is a bit like Horacio Spector’s view. Note, though, that Spector adopts a “maximize expected deontic value” framework, and he still assigns mere permissions zero value within that framework. So while they are explicitly acknowledged, they don’t make any actual difference to his view’s verdicts on cases like those considered here. Spector’s view would deliver the same verdicts on these cases as MEOW.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

216

 

capture the moral importance of saving a life when you have a duty to do so either by focusing on the good results if you succeed or on the wrongfulness of failing to do so if you don’t, and get much the same results. But when the upside is part of a supererogatory action, then there is no downside risk to not performing that action—you know you won’t be doing something objectively wrong. And yet the amount of good your supererogatory action will lead to does make a difference to whether you may run the risk of objective wrongdoing. If you’ve got a chance of saving thousands of lives by doing something supererogatory, that should justify a greater risk of acting objectively wrongly than if you’re only going to save five lives.²⁶ The main failing of “do what is most likely to be objectively right” was its indifference to morally important information. It asks you to choose by thinking about only one thing: the probability that this action is objectively right. MEOW is better than that. It takes downside risk into account, factoring in both the probability that one’s action will be objectively wrong, and how seriously wrong it will be if so. But it still leaves out crucial information about whether that act should be required, rather than merely permissible, and about how well things will go if you act permissibly. And it does so intentionally and explicitly—Olsen and Graham both assert that there are no degrees of permissibility, even if there are degrees of impermissibility. And while their view has linguistic support—it’s odd to say that one act is more permissible than another—the cases in this section show that this is superficial. We do, and should, take into account the different properties of different permissible acts. We have to recognize that some morally permissible acts can be required, while others are merely permissible; and that some permissible acts are, simply put, better than others. We can’t do that if, as the verdicts approach says we must, we treat all objectively permissible acts the same. This leads me into my final objection to the objective verdicts approach. Think about the burning building case, and the difference between the variant in which you could save 101 people if you go into the building, compared with saving six people (and in both cases you might save only one). MEOW basically says that those additional ninety-five lives cannot add to the case for entering the building—they can only subtract from it, by increasing the degree of expected wrongness if you act the hero and fall short. But why,

²⁶ NB here Frances Kamm’s arguments that supererogatory actions can sometimes override duties. See F. M. Kamm, “Supererogation and Obligation,” Journal of Philosophy 82/3 (1985), 118–38. Thanks to Al Hájek for discussion on this point.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

217

ultimately, do we care about acting objectively permissibly? Simply because doing so is objectively permissible? Or because of the facts in virtue of which one’s act is objectively permissible? MEOW and the verdicts approach in general take an intermediate conclusion as an input into subjective permissibility. This is similar to a point made in the context of moral uncertainty; there is something oddly fetishistic about caring only about whether your action is in fact objectively permissible, and ignoring the underlying considerations in virtue of which it would be objectively permissible if it is.²⁷ If those considerations matter to objective permissibility, then shouldn’t they matter to subjective permissibility as well?

5. Objective Reasons This leads to my preferred approach. The grounds of subjective permissibility, alongside the relevant kinds of epistemic states, are objective reasons—that is, facts that would obtain if you ϕ and the world is in a particular state, which count for or against your action. These very facts determine whether an act is objectively permissible. But, rather than taking that intermediate judgment and using it to ground judgments of subjective permissibility, we should instead go directly to the source, and consider the objective reasons themselves. We then discount them for their probability of being actual—perhaps with some function to reflect non-neutral attitudes to risk—and that gives us the basis for our decision rule. The most obvious example of this kind of approach is a form of expected value utilitarianism. For such a theory, judgments of subjective permissibility are grounded in facts about how much utility would obtain in the different possible outcomes of one’s action, discounted for the likelihood of each outcome obtaining. But one doesn’t have to be a utilitarian to endorse this general approach. As long as we can rank acts in virtue of the reasons for and against them, given that a particular state of the world obtains, then this approach can get off the ground. So deontologists needn’t be put off. They might endorse, for example, the following criterion of subjective permissibility:

²⁷ Brian Weatherson, “Running Risks Morally,” Philosophical Studies 167/1 (2014), 141–163; Michael Smith, The Moral Problem (Oxford: Blackwell, 1994).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

218

 

COST: An act is permissible if and only if either: (a) there is no allthings-considered expectedly better act or (b) every all-things-considered expectedly better act either (i) involves unreasonable marginal expected costs to the agent or (ii) is better only in virtue of expected benefits to the agent.²⁸ In this case, one act is expectedly better than another in virtue of the probability-weighted balance of reasons in its favor (taking into account both agent-relative and agent-neutral reasons). But we must also consider the act’s expected costs to the agent—again, grounded in the objective facts that will obtain if you perform that act given the different possible states of the world. COST can handle all the cases that confounded MEOW. In the first selfdefense case, A is always subjectively permitted to let herself be killed, because even if the alternatives are expectedly better, they are so only in virtue of expected benefits to her. If the probability that B is responsible for the threat to her life is below some threshold p, then A is subjectively required to let herself be killed. If it is above p, but below a further threshold p’, then A is subjectively permitted to kill B, on the grounds that, while letting herself be killed is expectedly better, it is not better by enough for A to be required to bear the additional expected cost for the sake of realizing that expectedly better outcome (the marginal expected costs are unreasonable). And if the probability that B is responsible for the threat is above p’, then killing B is the expectedly best act, and for that reason subjectively permissible on those grounds. In the second case, one is permitted to fire the RPG rather than take a chance on one of the pistols because, in light of the attacker’s culpability, wounding her is not that much better than killing her. Her interests just don’t count for much, given her culpability. So you are not required to bear much expected cost for her sake. Perhaps shooting one of the pistols is the expectedly best alternative. Nonetheless, firing the RPG is subjectively permissible, because choosing a pistol involves unreasonable marginal expected costs—the moral benefit of not taking the culpable attacker’s life is not great enough for you to be required to bear a 0.5 risk of death for its sake. In the third case, perhaps you are objectively required to save the lives of the five, but it may also be merely objectively supererogatory. Whether you ²⁸ In unpublished work with Pete Graham, I have explored how to extend COST to accommodate lesser evil options—cases where one is merely permitted to impose costs on some to the benefit of others.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

219

are subjectively required to sacrifice yourself depends on whether the additional expected cost is reasonable for you to bear, in order to save the five. If the probability that you were responsible for their fate is above some threshold p, then you are subjectively required to save them. But if it is below that threshold, then it remains merely subjectively permissible. We can handle the burning building case in different ways. At a first pass, the worry about objective wrongdoing goes out of the window, and we simply consider your options given your current information. We know that staying outside is subjectively permissible, because the marginal expected moral benefits of going in are not great enough to make the marginal expected costs morally required. But staying outside is merely permissible, because entering the building is expectedly better. If we increase the number of people whose lives could be saved, then that increases the case in favor of entering the building, nudging it closer toward being morally required. Their lives are counted directly, not only the intermediate verdict. What about once you are inside the building, facing the decision to go left or right? On the assumption that neither alternative involves more personal risk, at that stage you subjectively ought to choose the alternative with the higher probability of saving the greater number. If the two directions are equally likely to lead to saving the greater number, then it is subjectively permissible to choose either. If you choose wrongly, then you may have acted objectively impermissibly—no doubt the circumstances mean that it is not the most serious objective wrongdoing you could commit. But you have acted subjectively permissibly throughout. The remaining risky supererogation case is also handily dealt with by COST. It’s subjectively permissible not to donate, because the alternatives aren’t expectedly good enough to make the marginal expected costs of donating morally required. And if the risk of the charity being involved in harming others is great enough, then it might be subjectively impermissible to proceed. But we need to weigh the risk of harm against the good that can be achieved if things go well. On the reasons approach, a charity that might use my donation to save ten lives warrants running a greater risk of causing (objectively wrongful) harm than one that would save only one life. Advocates of the verdicts approach have offered three main objections against this kind of view.²⁹ First, that there are no degrees of rightness. Second, that we really do want to talk about risking objective wrongdoing. ²⁹ See especially Olsen, “Minimizing Expected Objective Wrongness”; Graham, “In Defense of Objectivism.”

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

220

 

And third, that endorsing this kind of approach leads one inevitably away from deontology, toward something much more teleological. I’ll answer each objection in turn. The first objection can be quickly addressed. The worry about degrees of rightness derives, I think, from the linguistic noise mentioned above, and the assumption that if one act were “more right” than another, then it would be wrong to choose the less right act. But that’s a mistake: we must simply insist on a criterion of right action according to which sometimes one cannot be required to perform the morally better act. Of course, we need some explanation of that fact. I think the most plausible explanation arises when the better act involves unreasonable additional cost to the agent (or else benefits to the agent that she has a right to refuse). But there may be other possibilities as well. The key is to divorce the acknowledgment that one act is better than another from any presumptive deontic implications. The second worry can also be addressed, by being explicit about what I have not shown. First, I have not shown that we have no reason to care about objective permissibility in general. In fact, I think that our moral theory should be able to rank our options and say which are permissible in light of the morally relevant facts, as well as in light of the agent’s evidence when she acted. Even if objective permissibility is not a proper input into subjective permissibility, that does not justify ignoring objective permissibility in general. We can therefore accommodate the idea that, after acting in a way that came perilously close to being objectively wrong, we should feel regret at having risked objective wrongdoing. Second, I have not in fact argued that verdicts of subjective permissibility should never be grounded in verdicts of objective permissibility. I have only argued that we ought not to exclusively ground the former verdicts in the latter. There might yet be some room to ground subjective permissibility in verdicts of objective permissibility—for example, if we know for sure that some option is objectively permissible, then we might plausibly conclude that it must be subjectively permissible. Nothing in this chapter rules out that possibility. The final objection is no doubt challenging. But I think it can be answered. To do deontological decision theory, we don’t have to assign values to outcomes, and discount those values by the probability of the outcomes coming about. Instead, we need to consider the reasons for and against your different act-alternatives, given the different possible states that the world might be in. Of course, that means taking consequences into account, but it does not necessarily mean ranking states of affairs.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

  

221

This, roughly, is the strategy that Paul Hurley calls “deontologizing” moral theories.³⁰ Instead of representing them with value assignments to states of affairs, we consider them as sources of reasons for action. Both are equally compatible with orthodox decision theory, which only requires us to rank act–state pairs, but the second is much more amenable to deontologists who are skeptical about whether we can assign value to states of affairs. The question, then, is whether merely recognizing that we can rank our actions in terms of the strengths of the all-things-considered reasons for and against them involves committing to a teleological approach to ethics, which is fundamentally inconsistent with deontology. Is this the fundamental fault line in moral theory? I don’t think it is. We can accommodate practically any substantive moral disagreement within reasons-based frameworks like this. Saying that this terrain is uniquely consequentialist, and consigning deontologists to whatever the alternative to ranking acts is, amounts not just to moving the goalposts, but to relocating the entire pitch. The ability to rank our acts according to the all-things-considered reasons that favor them—remembering the diversity of dimensions of normative strength—is not uniquely consequentialist.

6. Conclusion Judgments of subjective permissibility are not grounded only in verdicts of objective permissibility. Criteria of subjective permissibility like Minimize Expected Objective Wrongness and Maximize Expected Objective Deontic Value are wrong. Acts are objectively permissible in virtue of objective facts—reasons that count for and against the action. They are subjectively permissible in virtue of the same facts or, at least, what would be the facts given the different possible states of the world when one acts, discounted for each state’s probability of being actual. Focusing on objective verdicts at the expense of objective reasons blinds us to morally crucial differences between supererogatory acts, and acts of mere self-sacrifice—the very properties in virtue of which those acts have their objective deontic statuses. There is, of course, much more to say. Deontological decision theory is in its infancy, and there are many corners yet to turn before we reach a viable ³⁰ Paul Hurley, “Consequentializing and Deontologizing: Clogging the Consequentialist Vacuum,” in Mark Timmons (ed.), Oxford Studies in Normative Ethics, Vol. III (Oxford: Oxford University Press, 2013), 123–53.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

222

 

theory. Strictly speaking, I have shown only that subjective permissibility is primarily grounded in objective reasons. There may be some further role for verdicts of objective permissibility, for example when one is certain that an option is objectively permissible. We also need some account of how to deal with intermediate objective verdicts—for example, those to do with proportionality in the ethics of harm. Should we care about the properties in virtue of which an act is objectively proportionate, or about probability of objective proportionality simpliciter?³¹ There are many such questions to address before we have a viable deontological decision theory. I have shown, though, that no sensible approach can afford to ignore our objective reasons, or to take them into account only insofar as they affect the objective permissibility or impermissibility of our available acts.³²

³¹ On proportionality, I owe this line of thought to discussions with Patrick Tomlin, who defends the view that we should use objective proportionality judgments for decision-making under risk in his “Subjective Proportionality,” Ethics, 129/2 (2019), 254–83. ³² I presented this paper at NUS, York, and at the Oxford Studies in Normative Ethics workshop. Many thanks to the audiences at those presentations for their searching questions. For discussion and comments, thanks in particular to Renee Bolinger, Pete Graham, Tom Hurka, Holly Smith, and Doug Portmore.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

11 Agent-Relative Prerogatives and Suboptimal Beneficence Ralf M. Bader

1. Suboptimal Beneficence Agent x has two options: A = do nothing B = save y’s left arm at some cost to himself Both A and B are permissible. Given that the cost − α that x has to incur to save y’s arm is sufficiently large relative to the benefit + β conferred on y, x is permitted but not required to benefit y. Helping y is good but not required and hence classifies as supererogatory. A = do nothing C = save both of y’s arms at the same cost to himself Although the benefit is greater than in the previous case, helping y is not required if the cost to x is sufficiently large. Both A and C are then permissible. A = do nothing B = save y’s left arm at some cost to himself C = save both of y’s arms at the same cost to himself When all three options are available, however, only A and C are permissible. B is impermissible.¹ As Parfit notes, saving only one arm when one can ¹ Taking the default to be A, we can represent the choice situation as follows (NB nothing hangs on construing A to be the default—this is simply done for illustrative purposes). A B C

x

y

0 −α −α

0 +β + ð2 × βÞ

Ralf M. Bader, Agent-Relative Prerogatives and Suboptimal Beneficence In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0011

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

224

 . 

save both at the same cost is “grossly perverse” (cf. Parfit 1982: 131; Kagan 1989: 15–17, 240–1). Suboptimal beneficence of this kind is impermissible. Consequently, neither supererogation nor agent-relative prerogatives can be understood in terms of threshold models.² Such models include options that are above the threshold, but are nevertheless suboptimal in this objectionable way and hence impermissible. In addition, the impermissibility of suboptimal beneficence is of significant practical relevance.³

2. Sequential Choice The intuitive verdict that only A and C are permissible when choosing from the triple {A, B, C} receives further support from sequential cases. An agent first decides whether to do nothing = A or to incur the cost = D. If the cost is incurred, he faces a subsequent choice between two (by then costless) beneficent actions: saving only y’s left arm ¼ B + or saving both arms ¼ C +.

A

0:0

–α

:0

D + B+

β 0:+ 0:+

(2 ×

β)

D + C+

The following three diachronic choice consistency conditions, namely (i) separability from choice history, (ii) dynamic consistency, and (iii) reduction of sequential choice,⁴ together with the substantive claim that (iv) in a choice between costless versions of B and C, namely B + and C +, where

² Nor in terms of moving bands (cf. Hurka and Shubert 2012). ³ Cf. Pummer (2016) who has argued that these cases support effective altruism. ⁴ Cf. Cubitt (1996) for formal statements of these conditions.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

225

doing nothing is not an option, only C + is permissible, imply that in the nonsequential case only A and C are permissible whereas B is impermissible.⁵ First, separability implies that the choice between B + and C + as part of the larger decision tree has to be the same as that between these options in an isolated tree. The agent has to choose at node n1 in the subtree in the same way as at node n0 in a corresponding isolated tree.⁶

0 0:+

B+

: +β

(2 ×

β)

C+

This choice is between two costless acts of beneficence. (Given a counterfactual baseline for evaluating benefits, only C + will classify as an act of beneficence.) There is no reason (and no excuse) for choosing the suboptimal one. B + is impermissible. Separability then implies that B + is likewise impermissible in the subtree at node n1 . First incurring the cost and then performing the suboptimal beneficent action is as problematic as choosing B + when deciding between two costless acts of beneficence. The fact that a cost was previously incurred that could have been avoided by doing nothing does not affect the permissibility of this choice.⁷ Second, dynamic consistency requires that choices at subsequent nodes are consistent with the strategies adopted at prior nodes. Plans adopted at later nodes have to be continuations of the original plan. Subsequent

⁵ The same reasoning applies when the agent first decides whether to incur the cost in order to save one arm and, if the cost is incurred, then faces a subsequent decision whether to save the second arm at no additional cost. ⁶ This argument only requires separability from choice history, not separability from chance history. The latter applies when deciding under uncertainty and is rather contentious, given its connection to the independence axiom of expected utility theory. Cf. Cubitt (1996) for a precise characterization of these separability principles. ⁷ Some complications arise. We will see in Section 7 that a commitment to agent-relative prerogatives is only compatible with dynamic consistency when using global choice principles. Global choice, however, is incompatible with separability, which seems to undermine the first step of the argument. Yet, we can specify a restricted separability principle that is compatible with global choice and that allows us to establish the impermissibility of suboptimal beneficence.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

226

 . 

evaluations have to agree with the initial evaluation. By requiring the choice at n1 to be a continuation of a permissible strategy at n0 , dynamic consistency establishes that the only strategy that goes via n1 that is permissible from the outset is the one that results in C +, given that one has to choose C + rather than B + at the subsequent choice node. Third, reduction of sequential choice requires the sequential case to be treated analogously to the nonsequential case. Sequential choice has to agree with nonsequential choice. The permissible strategies in the complex tree have to correspond to those that are permissible in the nonsequential tree that has the same normal form.

A

0:

0

–α : +β

B

–α (2

:+ × β) C Since {A, B, C} and fA, D + B +, D + C +g have the same payoffs, reduction implies that B is permissible iff D + B + is permissible. Since B + is impermissible (given separability), the strategy D + B + is impermissible at n0 (given dynamic consistency), which implies that B is impermissible in the nonsequential case (given reduction). Suboptimal beneficence is, accordingly, impermissible, even when doing nothing is permissible, both in the sequential and the nonsequential case. Those, such as McMahan 2017, who consider all three options to be permissible when choosing nonsequentially from {A, B, C}, yet consider the equivalent of B to be impermissible when choosing sequentially from fA, D + B +, D + C +g, have to reject dynamic consistency or reduction of sequential choice. It will then be permissible in certain cases to “pay” (either by incurring a greater cost, or by foregoing to confer a greater benefit) in order to achieve an outcome that is available but that otherwise could not permissibly be brought about. The agent can permissibly reach the outcome of B − but not the superior outcome of D + B +.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

227

A

0:

0

–(α + є) : +β

B–

–α

D + B+

:0

β

0:+ 0:+

(2 ×

β)

D + C+

Here one is allowed to bring about benefit β but only at a larger cost to the agent, namely − ðα + ϵÞ, but not at the smaller cost of only − α.

A

0:

0

–α : +(β – є)

B–

–α :0

D + B+

β

0:+ 0:+

(2 ×

β)

D + C+

Here one is allowed to incur cost − α in order to confer benefit + ðβ − ϵÞ but not allowed to incur this cost in order to confer a slightly greater benefit + β.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

228

 . 

3. Agent-Relative Prerogatives The prohibition on suboptimal beneficence follows from an agent-relative theory that understands permissible actions in terms of a dominance principle defined over both the agent-relative and the agent-neutral ordering. This theory incorporates agent-relative prerogatives, allowing agents to act on the basis of the agent-relative evaluation that is centered on the agent. The agent can permissibly privilege (i.e. be partial towards) himself as well as those who are close to the agent and is not required to be impartial and do what is agent-neutrally best. An agent-relative axiological theory consists of an agent-relative and an agent-neutral ordering. Alternatives can be evaluated in two different ways. The evaluation is either indexed on the perspective of the agent, in which case it is sensitive to identity facts and thereby allows for partiality. Or it abstracts from the identity of the agent, making it an impartial evaluation that is permutation-invariant. These two evaluations are connected. The agent-relative evaluation results when applying a modification function to the agent-neutral evaluation.
 Vrelative ðyÞ ¼ k × Vneutral ðyÞ, where k ¼ f dðx, yÞ The modifier k is a function f of distance d in moral space between the agent x on whom the evaluation is centered and the person y whose well-being is evaluated, where distance itself is a function of morally significant relationships between the persons. The strength of the multiplier is inversely related to distance. The greater the distance, the smaller the multiplier. The closer someone is to the agent, the more that person counts from the agent’s perspective. For a complete stranger we can set k = 1, i.e., f(d(x,stranger)) = 1, whereas max k ¼ f ðdðx, xÞÞ, i.e., no one is more closely related to the agent than the agent himself.⁸ We have two standpoints from which things can be evaluated, each giving rise to a complete ordering that weighs up all the costs and benefits: the agent-relative and the agent-neutral standpoint. Although the two evaluations are connected via the modification function k, the resulting orderings cannot be weighed up against each other (cf. Bader 2016: section 4.1 and

⁸ For an account of modification functions, cf. Bader (2016). For a detailed treatment of this theory of agent-relativity, cf. Bader (ms), “Personal, General, and Impersonal Good.”

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

229

Bader 2015).⁹ This is not merely an epistemic concern about our ability to precisely weigh up the different values. Nor is it an issue of incommensurability, given that the modification function ensures that the two evaluations are commensurable. Instead, the issue is that the values are internal to their respective standpoints and that there is no further standpoint that subsumes them both. There is no further perspective that allows us to combine these evaluations by determining their relative weights. All we have are these two standpoints, neither of which is privileged over the other.¹⁰ The top-elements of the two orderings are endorsed by their respective standpoints. Since neither standpoint is subordinated, in particular since neither standpoint is privileged over the other, there is nothing that can counteract their respective endorsements. The agent can hence permissibly perform those actions that are undominated on at least one ordering. He can either do what is agent-relatively best or what is agent-neutrally best. The following dominance principles defined over both orderings yield the set of permissible actions.¹¹    CðXÞ ¼fx ∈ X : :∃y:∃z ∈ Xðy >relative x ∧ z >neutral xÞ ∧ :∃w ∈ X½ðw ≥relative x ∧ w >neutral xÞ ∨ ðw >relative x ∧ w ≥neutral xÞg    CðXÞ ¼ fx ∈ X: :∃y:∃z ∈ Xðy >relative x ∧ z >neutral xÞg The strict dominance principle rules out those alternatives that are not topelements on either ordering. The weak principle also rules out alternatives ⁹ There would be no room for prerogatives if there were to be a correct way of weighing them up. Cf. Hurka and Shubert (2012: 3–5) who argue that permissions cannot be generated when operating with values or reasons and hence advocate positing underivative permissions. ¹⁰ Given that there is not one way to weigh them up, one might consider all possible ways of weighing them up and taking an action to be permissible if it is permissible according to some way of weighing them up. This amounts to taking the union of the optimizing choice sets of all convex combinations of the two evaluations, i.e., Vcombined ¼ λVrelative + ð1 − λÞVneutral . The difference between λ being in the closed or open unit interval corresponds to the difference between strict and weak dominance principles: strict if λ ∈ ½0, 1 and weak if λ ∈ ð0, 1Þ. (The two asymmetric principles correspond to λ ∈ ð0, 1 and λ ∈ ½0, 1Þ.) This way of weighing up the two evaluations, however, corresponds neither to the pure nor the mixed dominance principles. It is more permissive than the pure principle and less permissive than the mixed principle. Moreover, it does not satisfy dynamic consistency. ¹¹ These principles are somewhat similar to the dual-ranking theories developed by Sider (1993) and Portmore (2008), which are based on the self v. other distinction rather than the agent-relative v. agent-neutral distinction (thanks to Doug Portmore for pointing out these articles to me).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

230

 . 

that are tied for top-element on one ordering with something that dominates them on the other ordering. The weak principle treats the agent-relative and the agent-neutral orderings symmetrically. There are also two asymmetric principles that require only one of the disjuncts featuring in the weak dominance principle: either :∃w ∈ Xðw ≥relative x ∧ w >neutral xÞ or :∃w ∈ Xðw >relative x ∧ w ≥neutral xÞ. In order to argue in favor of the weak principle (over the strict principle as well as over the two asymmetric principles), two possibilities need to be ruled out. First, is one allowed to perform an action that is agent-relatively worse yet tied in terms of agent-neutral value? Here the question is whether one can set aside agent-relative reasons and take up an agent-neutral perspective, even when this does not make things impartially better. For instance, is it permitted to help a stranger rather than a friend when the benefit is the same, so that these options are equally good from the agent-neutral perspective, yet the latter is preferable from the agent-relative perspective? By citing the importance of friendship one can justify to the stranger that one is not helping him but instead one’s friend. Yet it would seem that one cannot justify to one’s friend why one is not helping him. Helping the stranger could be justified if doing so would bring about more agent-neutral good. However, when the alternative is not impartially better, then this justification is not available. Second, is one allowed to perform an action that is suboptimal agentneutrally but tied in terms of agent-relative value? Performing such an action looks like a gratuitous failure to promote the impartial good. The alternative effectively amounts to a case of “costless” beneficence, insofar as there is no net cost in terms of agent-relative good (though it may involve costs to the agent), and hence would seem to be required. Given that the two actions are equally good agent-relatively, one cannot cite any agent-relative considerations to justify not performing that action which is agent-neutrally optimal. These considerations suggest that we should adopt the weak rather than strict dominance principle. The arguments of this chapter, however, go through on both versions. In each case the dominance principle can account for the permissibility pattern with which we started. When choosing between A and B, both are permissible. The agent-relative ordering is: A > B. The agent-neutral ordering is: B > A. This ensures that neither option is dominated, so that both options end up in the choice set. The same reasoning explains why both A and C are permissible if one is to choose between them. When choosing between all three alternatives, however, the agent-relative ordering is: A > C > B, whereas the agent-neutral ordering is:

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

231

C > B > A. The dominance principles allow the agent to either pick option A (which is agent-relatively best) or option C (which is agent-neutrally best). Options, such as B, that are suboptimal on both orderings, by contrast, are impermissible. The agent-relative prerogative justifies x in choosing A rather than C, but does not justify choosing B rather than C, thereby ensuring that A and C end up in the choice set whereas B is excluded.

4. Mixed Dominance The reasoning that renders B impermissible when choosing from {A, B, C} also applies when the two different ways of helping do not involve the same cost, but when the agent needs to incur a slightly larger cost to confer the greater benefit. x y A B C−

0 −α − ðα + εÞ

0 +β + ð2 × βÞ

A and C − are the only permissible options as long as ϵ is sufficiently small. B does become permissible when the additional cost is sufficiently large. For instance, B is no longer impermissible if the additional cost is as large as the original cost, i.e. when C −− involves a cost of − ð2 × αÞ for x and a benefit of + ð2 × βÞ for y.

A B C

x

y

0 −α − ð2 + αÞ

0 +β + ð2 × βÞ

The pure dominance principles, however, exclude B from the choice set since B is suboptimal on both orderings. Yet, intuitively, B should be permissible when C −− is the alternative. This becomes clear when thinking of an analogous sequential choice situation, whereby x is faced on two separate occasions with the option of doing nothing or conferring benefit β at cost α. A B B  C −−

x

y

0, 0 0, − α − α, 0 − α, − α

0, 0 0, + β + β, 0 + β, + β

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

232

 . 

0:0

0:0 –α

: +β

–α :

+β

0:0 –α :

+β

≈A

≈B ≈B

≈ C––

Intuitively, both alternatives are permissible at each node. That x has (not) helped at the first choice node does not imply that he has to (not) help again at the second node. Although we will reject separability in Section 7, the reasons for doing so do not apply to this case, given that the initial choice makes no difference to the rest of the decision tree. The subtree that one faces at the second node is the same independently of whether one has helped at the initial node. The choice at n0 has no effect on what options are available at n1 : in each case one can either do nothing or confer benefit β at cost α. Moreover, if separability were to fail in this case, it would support rather than rule out mixed strategies, since it is understandable that an agent does not help again if he has already helped in the past. It is, accordingly, permissible for the agent to help only once. Since helping once is the analogue of choosing B from fA, B, C −− g, reduction of sequential choice implies that B is likewise permissible. A more nuanced approach is thus required. The pure dominance principles allow one to either adopt the relative or the neutral perspective. The agent can then perform actions that are permissible according to at least one point of view. We can weaken these principles and allow the agent to perform any action that is not dominated by some alternative on both orderings.¹²    CðXÞ ¼ fx∈X: :∃y∈X½ðy≥ relative x∧y > neutral xÞ∨ðy > relative x∧y ≥ neutral xÞg    CðXÞ ¼ fx ∈ X : :∃y ∈ Xðy >relative x ∧ y >neutral xÞg

¹² The principles are weaker in that they are more permissive. It is easier for things not to be dominated since the condition being specified by the dominance principle is stricter such that fewer things can satisfy this condition.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

233

If some alternative dominates an action from both the agent-relative and the agent-neutral perspective, then that action is impermissible. By contrast, if both orderings agree that an action is suboptimal but they disagree on the reasons for which it is suboptimal, then it is not impermissible. To be impermissible, the orderings not only have to agree that something is dominated, i.e., that it is not a top element, but also have to classify it as being suboptimal for the same reason, i.e., they have to agree on which alternative dominates it. Unless they agree, the reason for which the one perspective rules out a given action will not be admissible from the standpoint of the other, and vice versa. There will then be no agreed reason for ruling it out. To be responsive to the fact that an action is suboptimal is to not perform this action but instead one that dominates it. If there is no agreed reason, the agent cannot be responsive to both perspectives. There will be an action that dominates it on the agent-relative ordering and an action that dominates it on the agent-neutral ordering, but no action that dominates it on both orderings. The agent can then either be responsive to the agent-relative or the agent-neutral perspective, but not to both. Even though the other point of view agrees that the action is to be ruled out, the agent cannot be responsive to both of the reasons for which it is ruled out. Instead, the agent will end up siding with one perspective and will do something that the opposing point of view deems to be worse than the action that both points of view consider to be suboptimal. It is for this reason that one should operate with the mixed dominance principles and consider actions to be permissible, even if they are suboptimal on both orderings, as long as there is no agreed reason for considering them to be suboptimal. These mixed dominance principles ensure that, whilst B is impermissible when A and C − are the other alternatives, B is permissible in the case of C −− . As long as the additional cost is sufficiently small, C − is above B in both rankings. The cost has to be so small that the agent would have to incur − ϵ to confer benefit + β. Helping is not optional but required when the value of the additional benefit is greater than the additional cost, even from the agent-relative perspective, i.e., k × ϵ ≤ β. (We need a strict rather than weak inequality if we are using the weak rather than strict mixed dominance principle.) However, if the additional cost is sufficiently large, i.e., if k × ϵ > β, then it is optional and not required to incur this cost to confer benefit β.¹³ The mixed principle then renders B permissible. Whilst B is ranked below C − ¹³ The modification function, in this way, determines precisely how much greater to x the additional cost has to be in order for B to be permissible.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

234

 . 

on both orderings (relative: A > C − > B and neutral: C − > B > A), B is ranked above C −− on the agent-relative ordering (relative: A > B > C −− and neutral: C −− > B > A). There is hence no option that dominates B on both orderings. Whilst being dominated on both orderings, B is dominated by one option, namely C −− , on the agent-neutral ordering, yet by another option, namely A, on the agent-relative ordering. The mixed dominance principles which require the different orderings to agree allow us to construct a single combined evaluation, namely an intersection quasi-ordering. Since it is an incomplete ordering, we have to apply a maximizing rather than optimizing function to determine the choice set.¹⁴ The difference between the weak and the strict mixed dominance principles over the two orderings concerns the construction of this quasi-ordering. The difference is whether x ≫ y only requires x to be strictly better than y with respect to one of >relative and >neutral whilst being weakly better with respect to the other, or whether x needs to be strictly better on both orderings.   x ≫ y ¼df x ≥

relative y ∧ x ≥neutral

y ∧ :ðx ≡ yÞ

where x ≡ y ¼df x ¼relative y ∧ x ¼ neutral y   x ≫ y ¼df x >relative y ∧ x >neutral y The choice set then contains all alternatives that are not strictly dominated with respect to the intersection quasi-ordering. CðXÞ ¼ fx ∈ X: :∃y ∈ Xðy ≫ xÞg

5. Prerogatives and Beneficence This account explains not only why it is impermissible to help suboptimally, yet permissible to not help. It also explains why it is impermissible to help suboptimally, even though it is permissible to not do what is impartially best ¹⁴ An optimizing choice function, which selects options that are at least as good as all the alternatives, presupposes completeness. A maximizing choice function, which selects undominated options, is compatible with incompleteness (cf. Sen 1997).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

235

but instead help those who are close to the agent. A moderate agent-relative theory that uses a modifier k that is a continuous function of distance in moral space gives rise to an agent-relative prerogative that not only allows the agent to privilege his own interests but also the interests of those that are close to him.¹⁵ Accordingly, one can help a relative rather than a stranger, even if the latter would benefit more. Yet one is not permitted to engage in suboptimal beneficence either with respect to strangers or with respect to one’s relatives. x y r * A 0 0 +γ * B 0 0 + ð2 × γÞ C* 0 +β 0 D* 0 + ð2 × βÞ 0 Since the modifier varies with distance in moral space, greater weight is assigned to the well-being of those who are closer to the agent than those who are more distant. If
the relative r is sufficiently close to the agent x,  then the modifier k ¼ f dðx, rÞ can ensure that, even though ð2 × γÞ < β, it is nevertheless the case that k × γ > 2 × β (where y is a stranger so that the multiplier = 1). The agent is then permitted to take up either the agentneutral perspective and perform D* or the agent-relative perspective and perform B*. A* and C*, however, are impermissible. They are suboptimal on both orderings. Agent-relative as well as agent-neutral suboptimal beneficence are ruled out. We thus have a unified account that encompasses both partiality to oneself and to others.¹⁶ The prohibition of this kind of suboptimal beneficence, by contrast, cannot be explained in terms of conditional obligations. Whilst one might be able to explain the impermissibility of B when choosing from {A, B, C} by arguing that, even though one is not required to help, one is obligated to help in the optimal way on condition that one does help, this approach cannot account for the prohibition of suboptimal beneficence (i.e., the impermissibility of A* as well as C*) when choosing between helping relatives or helping strangers.

¹⁵ The modification function k does not only apply to the agent but to everyone. ¹⁶ Unlike accounts that are based on a self v. other asymmetry, e.g., Sider (1993) and Portmore (2008), and accounts that build in a separate principle of partiality, e.g., Hurka and Shubert (2012) and Horton (2017).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

236

 . 

The conditional obligations approach can be spelled out either: (1) in terms of obligations that are conditional on helping, or (2) in terms of obligations that are conditional on incurring a cost (the idea here is that one can justify doing A rather than C because of the cost involved in C, yet one cannot appeal to this cost to justify doing B rather than C, given that both options involve this cost). Both interpretations run into difficulties. Since all four options involve helping others, it is of no use to bring in obligations that are conditional on helping. Similarly, since the agent is not incurring any costs in this situation, bringing in obligations that are conditional on incurring costs is to no avail.¹⁷

6. Agent-Sacrificing Options The cost incurred by x in doing B or C has to be sufficiently large relative to the benefit conferred upon y for it to be permissible that x does nothing (= A).¹⁸ From the agent-relative perspective, the cost to x has to be at least as large as the benefit to y: Vrelative ðAÞ ≥ Vrelative ðCÞ, i.e., k × α ≥ k × 2 × β.¹⁹ Otherwise, A will not be a top-element in the agent-relative ordering. The cost is also not allowed to be too large (relative to the benefit), i.e. α ≤ β. The benefit would otherwise not be worth the cost from the agent-neutral perspective. Otherwise, A would be the (only) top-element in the agentneutral ordering and B would be impermissible when choosing between A and B. Accordingly, the following inequalities have to be satisfied: k×α≥2×β>β≥α (When using the weak dominance condition, A needs to be the only topelement in the agent-relative ordering if it is to be permissible. The agentrelative evaluation then needs to satisfy a slightly stronger condition: Vrelative ðAÞ > Vrelative ðCÞ and hence k × α > 2 × β. Likewise, A is not allowed to be a top-element in the agent-neutral ordering for B to be permissible when ¹⁷ In addition, conditional accounts lack the resources to address the dynamic inconsistencies identified in Section 7, since this requires a suitable set of orderings in terms of which strategies can be classified as being suboptimal by a global evaluation that takes counterfactual alternatives into consideration. ¹⁸ What classifies as sufficiently large depends on the size of the benefit. There is no absolute threshold but a relative threshold. If the benefit is very large, then an agent can be required to incur a large cost. The agent is required to incur the cost as long as β > k × α. ¹⁹ Given that y is a stranger, the agent-relative modifier for y is: k ¼ 1. We hence have the weak inequality: k × α ≥ 2 × β.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

237

choosing between A and B. The agent-neutral evaluation then needs to satisfy a slightly stronger condition: Vneutral ðBÞ > Vneutral ðAÞ and hence α < β.) Since α ≤ β, agent-relative prerogatives only allow for departures from what is impartially best by justifying agent-favouring options whereby the agent can be partial to himself and those who are close to him. They do not, however, justify agent-sacrificing options whereby the cost that is incurred by the agent is greater than the benefit that is conferred.²⁰ Such behavior is suboptimal from both perspectives. It can neither be condoned from the agent-neutral perspective nor from the agent-relative perspective. The prohibition on agent-sacrificing behavior can be explained by noting that the well-being of the agent matters morally²¹ and generates reasons in the same way as the well-being of anyone else.²² The benefits hence have to be sufficiently large in order to justify the costs that the agent has to incur. The costs to the agent speak against the action and there have to be sufficiently strong countervailing considerations that speak in favor of incurring these costs. Cases of agent-sacrificing whereby the benefit is smaller than the cost do not satisfy this condition. Engaging in agent-sacrificing is not merely imprudent but morally problematic. By bringing about less value, both agentrelatively and agent-neutrally, than if the action had not been performed, such an agent is not suitably responsive to moral reasons. Difficulties would arise in making sense of agent-relative prerogatives if one were to treat the agent’s well-being differently. If one identifies the agent-neutral point of view as the moral point of view, then it is difficult to see how it can be permissible to depart from this point of view.²³ Accordingly, we should not think of prerogatives in terms of non-moral reasons competing against moral reasons. Instead, we are dealing with two moral points of view from which things can be evaluated. Since neither

²⁰ Thanks to Tom Hurka for pressing me to say more about agent-sacrificing options. ²¹ Agent-relative good is not a form of prudential good but of moral good. The agent-relative v. agent-neutral distinction is not to be confused with the prudential v. moral contrast. What is good relative to x is also not to be conflated with what is good for x. This is particularly clear when concerned with benefitting people who are close to the agent. Benefitting them is not a matter of prudence (or some other form of non-moral reason) but a matter of partiality. ²² This account differs sharply from theories based on a self v. other asymmetry. Portmore, for instance, argues that “there is no requiring reason to promote one’s own wellbeing per se” (Portmore 2008: 416) but instead only a justifying reason. ²³ This problem is closely related to the “paradox” of supererogation. Cf. “It is hard to see how it could be permissible, from the moral point of view, to refrain from doing something that you have an undefeated reason (from that very point of view) to do. . . . Supererogation, according to this way of seeing things, turns out to be impossible” (Dreier 2004: 148).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

238

 . 

perspective is privileged, these two points of view must agree that an action is dominated by another alternative for that action to be impermissible. One can respect that one’s own happiness gives rise to moral reasons yet nevertheless incorporate agent-sacrificing options into the theory by including a patient-relative ordering. This ordering corresponds to an evaluation that is centered on the patient—one person’s agent-relative ordering is another person’s patient-relative ordering. One can then apply dominance principles over the triple.    CðXÞ ¼ fx ∈ X: :∃y ∈ X½ðy ≥ agent x ∧ y ≥neutral x ∧ y >patient xÞ ∨ ðy ≥ agent x ∧ y >neutral x ∧ y ≥patient xÞ ∨ ðy >agent x ∧ y ≥ neutral x ∧ y ≥patient xÞg    CðXÞ ¼ fx ∈ X: :∃y ∈ Xðy >agent x ∧ y >neutral x ∧ y >patient xÞg This makes it permissible not only to act on the basis of the agent-relative evaluation alongside the agent-neutral evaluation, but also to occupy the position of another. The agent is allowed to put himself into the shoes of someone else, namely the person that the agent is acting upon, and act on the basis of the evaluation centered on that person’s perspective. If that person could permissibly bring about a certain distribution of costs and benefits by maximizing that person’s agent-relative ordering, then it is likewise permissible for the agent to do so, i.e. to maximize the patientrelative ordering.²⁴ (Given a suitable set of agent-relative constraints, one can explain why it is permissible for the agent to sacrifice himself in order to benefit someone else, yet not permissible for the stranger to impose the same costs on the agent in order to incur the same benefit.) This allows for limited agent-sacrificing (contrary to the self-other asymmetry). Although the costs that the agent can permissibly incur can be greater than the benefits that are thereby conferred, the costs are nevertheless not allowed to be excessive. Instead of α ≤ β, the condition is now that k z, x × α ≤ k z, z × β.²⁵ The fact that the modifier k applies to α ensures that ²⁴ The dominance principles rule out suboptimal agent-sacrificing (both the gratuitous conferring of a smaller benefit and the gratuitous incurring of a greater cost). Such behavior is classified as suboptimal on the patient-relative ordering.   ²⁵ The first k corresponds to the  distance  between the patient and the agent: k ¼ f dðz, xÞ , whereas the second is maxk ¼ f dðz, zÞ . Since d does not satisfy the symmetry axiom, it is a pseudo-metric. Morally significant relationships need not be symmetric: x can be relatively close to z, without z being close to x.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

239

how much x can sacrifice in order to benefit z depends on how close z is to x. How the cost α is evaluated from the patient’s perspective depends on the distance between z and x, which in turn depends on the nature and strength of the morally significant relationships in which z stands to x. The closer z is to x, the more significant α is from z’s perspective. Relative to a given benefit β that accrues to z, the cost α that can be incurred by x will be smaller the closer z is to x. Those who are close to the agent cannot as easily condone self-sacrificial behavior since the agent’s well-being matters more from their perspective. In the same way that benefits to those who are close to the agent matter more from the agent-relative perspective, so costs to the agent matter more from the patient-relative perspective when the beneficiary is close to the agent.²⁶ More is at stake from z’s perspective so that the sacrifice is more easily classified not only as agent-relatively as well as agent-neutrally suboptimal but also as patient-relatively suboptimal. Difficulties arise if multiple people are affected by the agent’s actions. Once multiple patients are involved it becomes unclear whose perspective the agent can take up. Can the agent identify with particular individuals (or particular pluralities of individuals)? Or is the only patient-perspective that the agent can permissibly adopt that of all the patients taken collectively? If the former were the case, then the prohibition of suboptimal beneficence would be significantly weakened.

A B C

x

y

0 −α −α

0 +β + ð2 × βÞ

In the original case, B would still be impermissible. However, the situation would be different in a variant case in which the beneficiary of the suboptimal act of beneficence differs from the beneficiary of the optimal act of beneficence. A0 B0 C0

x

y

z

0 −α −α

0 +β 0

0 0 + ð2 × βÞ

²⁶ If z is required to incur a cost − α to confer a benefit + β on x, where α < β, then x cannot sacrifice − β in order to confer + α on z (assuming the distance between x and z is symmetrical).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

240

 . 

The agent’s ranking is: A0 > C0 > B0 and the neutral ranking is: C0 > B0 > A0 . Even though the ranking centered on the patients taken collectively is likewise: C0 > B0 > A0 , the relative ranking centered on the individual patient y is: B0 > C0 > A0 . When taking this final ranking into consideration, all three alternatives are permissible. Unlike B in the original case, B0 will then not be impermissible. This suggests that the relevant contrast is between the agent (the one doing the acting) and the patients taken collectively (all those who are acted upon). What the agent can do is to take up their collective perspective rather than identify with particular individuals. One then has to find a way of combining their perspectives and construct a plural perspective of the various patients taken collectively. We can do this by allowing the distance function to take plural arguments, letting the multiplier k be a function of the average distance. The distance of the zz’s to the agent x, i.e. d(zz,x), which determines the modifier k that applies when evaluating x’s well-being from the perspective of the zz’s will be the average distance: X dðzi , xÞ z ≺zz i

jzzj

7. Dynamic Inconsistency A and B are permissible when X ¼ fA, Bg, yet only A and C but not B are permissible when X0 ¼ fA, B, Cg. Expansion consistency condition , accordingly, fails. This condition requires that if two options are in a given choice set, then one of them is in the choice set of any superset of the original set of alternatives iff the other is also in the choice set. BETA: ½x, y ∈ CðXÞ ∧ X  X0  ! ½x ∈ CðX0 Þ $ y ∈ CðX0 Þ Whilst violating condition , contraction consistency condition  is satisfied.²⁷ ALPHA: ½x ∈ CðXÞ ∧ x ∈ X0  X ! x ∈ CðX0 Þ This violation of  is analogous to that which arises when dealing with incommensurable values, where two non-comparable options can be in a ²⁷ Expansion consistency conditions  and  are likewise satisfied.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

241

choice set yet only one of them fails to be in the choice set of a superset of the set of alternatives, since this superset involves a further alternative that dominates only this option but is non-comparable to the other. Although the agent-relative axiology does not involve incommensurability, since the agent-relative and agent-neutral evaluations are connected by a modification function and hence are comparable, we are nevertheless dealing with choice functions that are defined over two separate orderings. The fact that agent-relative prerogatives imply -violations is disconcerting since this gives rise to dynamic inconsistencies.²⁸ If an agent makes use of myopic choice principles (which are purely forward-looking) or sophisticated choice principles (that use backwards induction), then there are cases in which he can, by means of a sequence of permissible choices, bring about an outcome that is deemed to be impermissible from the outset. The prohibition on suboptimal beneficence thus appears to conflict with dynamic consistency. This already came up in the sequential choice situation in which an agent is faced on two occasions with the possibility of either doing nothing or conferring benefit β at cost α. Dynamic inconsistency arises when applying the pure (rather than mixed) dominance principle.²⁹ Only the pure strategies, namely (1) not helping in both cases or (2) helping in both cases, are permissible from the outset. However, when evaluated at the second choice node, both helping and not helping are permissible independently of how the agent has chosen at the first choice node (as long as one is using choice principles that satisfy separability). Only pure strategies are admissible at the outset, yet mixed strategies can permissibly result when applying the choice principles afresh at the second choice node. This dynamic inconsistency is

²⁸ -violations also have implications for conditional obligations. When X ¼ fA, B, Cg the choice set is CðXÞ ¼ fA, Cg. One is required to do either A or C. Given that one is not going to do C, it might seem that one is required to do A, i.e., that one is subject to a conditional obligation to do A on condition that one is not going to do C. This would be counterintuitive, cf. Horton (2017: 94). However, given that  is violated, this conditional obligation does not follow. The choice set of a restriction of the set of alternatives need not correspond to the (non-empty) restriction of the choice set. This is brought out clearly by an alternative formulation of  that highlights the corresponding contraction consistency condition: X  X0 ∧ CðXÞ \ CðX0 Þ 6¼ ∅ ! CðXÞ  CðX0 Þ. Instead of restricting the choice set, i.e., CðXj:CÞ ¼ CðXÞ\C ¼ fAg, we have to apply the choice function to the restricted set of alternatives, i.e., CðXj:CÞ ¼ CðX\CÞ ¼ fA, Bg. We then end up with the conditional obligation to do either A or B on condition that one is not going to do C. (If the evaluation fails to be context-free, then the fact that C is a possible option can affect the evaluation of B such that CðXj:CÞ need not be the same as CðX\CÞ.) ²⁹ Sider’s self-other utilitarianism (cf. Sider 1993), for instance, suffers from this kind of dynamic inconsistency.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

242

 . 

avoided by means of the mixed dominance principles, since they classify mixed strategies as being permissible at the outset. Some problems, however, remain. Dynamic inconsistency looms large in the following choice situation. Agent x has to choose between A, B, and a third option D = save y’s left arm at a cost slightly greater than α, but then face a subsequent choice between E = do nothing and F = save y’s right arm at a cost slightly less than α, i.e., D + E is the dynamic equivalent of B −, whereas D + F is the dynamic equivalent of C.³⁰

A B D0 + E0 D0 + F0

x

y

0 −α − α, 0 − α, − α

0 +β + ðβ − εÞ, 0 + ðβ − εÞ, + ðβ + εÞ

A

0:

0

–α : +β

–(

α

+

є)

:+

β –(α

B

0:0

D+E

– є)

D+F

: +β

³⁰ The same reasoning applies if one alters the benefit conferred on y, rather than the cost to the agent: A B D+E D+ F

x

y

0 −α − ðα + εÞ, 0 − ðα + εÞ, − ðα − εÞ

0 +β + β, 0 + β, + β

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

243

D + F is impartially best. D + E, by contrast, is suboptimal both agentrelatively and agent-neutrally. Since it is dominated on both orderings by B, it is impermissible according to the mixed dominance principles. Hence, one is only permitted to choose D rather B at n0 given that one is going to do F rather than E. Yet, the agent can permissibly perform D + E when choosing myopically, i.e., when applying the choice principle afresh after having chosen D and when faced with the choice between E and F. Once the agent has done D and is choosing between the continuations E and F, the additional cost that was involved in D is a sunk cost that does not affect this subsequent choice. The evaluation at the original node n0 does not cohere with that at node n1 . We are thus confronted with dynamic inconsistency. The strategy D + E is impermissible at the outset, yet can be implemented by means of a sequence of permissible choices. A case similar to that considered in Section 2 is even clearer. This time the agent does not first decide whether to incur the cost but instead decides whether to do C or do nothing. If the agent decides to do nothing at n0 , he then faces a subsequent choice at n1 of again doing nothing or choosing B.

0:0

0:0

–α :

–α

: +(

2×

+β

D+A

D+B

β) C

The strategy D+B is impermissible from the outset. Yet it can be implemented by means of a sequence of permissible choices. In the isolated tree both A and B are permissible. Separability then implies that B is a permissible continuation of D at node n1 , conflicting with the evaluation at n0 . The separability principle, accordingly, has to be restricted if one is to ensure dynamic consistency. It has to be weakened so that the choice between A and B in the isolated tree can conflict with the choice in the subtree. Fortunately, we can restrict separability such that it is compatible with the global choice principle developed in the next section, yet strong enough to underwrite the argument in Section 2 against suboptimal beneficence.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

244

 . 

Restricted separability applies if the maximal elements (i.e., those that are undominated with respect to the intersection quasi-ordering) are comparable in terms of the intersection quasi-ordering, i.e., the agent-relative and agent-neutral orderings agree on how they are to be ranked (which happens if they stand in ≡ to one another). If a subtree is such that the maximal elements at ni are comparable with each other, then the choice in the subtree has to agree with the choice at n0 in a corresponding isolated tree. The members of the choice set are comparable in that case and the maximizing and optimizing choice functions yield the same choice set. This does not presuppose that all alternatives are comparable (i.e., it is not required that 8x, y ∈ Xðx ≫ y ∨ x ≡ y ∨ x  yÞ).³¹ This is because it is possible for two non-comparable alternatives to be both dominated by the same option, such that any x, y in X are either comparable or both dominated by a further option that is comparable to each of them. The maximal elements can thus be comparable to all other elements, without the non-maximal elements having to be comparable with respect to each other.³²

8. Global Choice To ensure dynamic consistency one has to appeal to global choice principles that consider the entire decision tree and not only those nodes that are accessible from the current choice node.³³ Global choice principles are required because the choices are not independent. Unlike in the situation where the agent is faced on two occasions with the possibility of helping and where the choices are independent, in the case of {A, B, D + E, D + F} the choices are connected. The additional cost involved in choosing D rather than B is incurred precisely so that one can choose a certain option later on.

³¹ This only follows if comparability is an equivalence relation. ³² In fact, we can accept a slightly stronger condition: any maximal subset of comparable members of the choice set at node ni will be the choice set at node n0 of an isolated tree consisting of all those options that are in the set of alternatives X and that are comparable to the options in the subset of the choice set of X. ³³ This should be unsurprising given that incompleteness in general requires global choice. E.g. myopic choice principles allow one to zigzag through a sequence of non-comparable alternatives to end up with drastically suboptimal outcomes, even when dealing with a local form of incompleteness that is to be understood in terms of an intersection quasi-ordering generated from two dimensions of evaluation that can be integrated in different ways. Cf. Bader (ms), “Incompleteness and Dynamic Consistency.”

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

245

The rationale for choosing D is to be able to choose F later on. There is no point in choosing D unless one subsequently chooses F. Choosing E directly contravenes this rationale. By undermining the rationale for D, choosing E makes the prior choice irrational and pointless. Choosing E at n1 is impermissible because doing so makes having chosen D impermissible. D is only permissible as part of the D + F combination but impermissible as part of the D + E combination. D inherits its (im)permissibility from the (im)permissibility of the course of action of which it is a part. Accordingly, whether one chooses E or F determines whether or not D is impermissible. The principle that it is impermissible to do something that renders a past action impermissible then implies that, even though E and F are permissible when considered by themselves, only F is permissible qua continuation of D. Unlike in many other cases of dynamic inconsistency, sophisticated choice is not the solution. A sophisticated chooser cannot avoid inconsistencies arising from incompleteness by using backward induction in order to anticipate what he is likely to choose in future choice situations and then decide amongst available alternatives in light of the anticipated choice behavior. After all, both E and F are permissible when evaluated myopically. There is thus no reason to think that the agent will choose one rather than the other. Instead, the choice function has to take past choices into consideration. One has to choose in light of what choices one has already made and what options were previously available. Dynamic inconsistency arises when the evaluations at ni and nj conflict, i.e., when there is a difference between what was deemed to be permissible and what is now deemed to be permissible (where this is neither due to a change in information nor a change of objectives nor a difference in terms of means–ends reasoning, but entirely due to a change in perspective). This kind of conflict can be avoided by switching from local to global evaluations. As long as deliberation and evaluation are sensitive to the past, the evaluations at the different nodes will cohere. A change of temporal perspective has no effect on an all-encompassing evaluation that is not only futuredirected but also takes the past into consideration on an equal footing. The different evaluations will, accordingly, not conflict. The thought underlying global choice is that agents are diachronic beings. We should be concerned with the agent considered as a whole. We should evaluate choices not only locally, but also globally by asking how they fit into the agent’s life as a whole. By basing one’s choice on the most encompassing evaluation, one ensures that one is maximally responsive to reasons, not qua

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

246

 . 

synchronic time-slice, nor qua forward-looking agent, but qua unified diachronic agent. Privileging the diachronic agent over the present timeslice as well as over the future segment of the agent explains why the past can normatively constrain the present evaluation. Although an agent can only choose amongst alternatives that are currently available, he can evaluate these alternatives by comparing them to other alternatives that were previously available. Comparison, unlike choice, need not be restricted to available alternatives. It can also encompass counterfactual alternatives. Previously available options can cast a historical shadow. One has to expand the comparison class by considering counterfactual alternatives alongside available alternatives when deciding which of the latter to choose. An evaluation that considers options that the agent could have chosen is not restricted to local features of the choice situation but considers the choice in light of the agent’s life as a whole. This diachronic perspective allows alternatives that are undominated by the available alternatives to nevertheless be suboptimal due to being dominated by options that the agent could have chosen. Counterfactual options matter because what the agent does in the present can affect our evaluation of the agent’s past choices. We do not only evaluate isolated actions but also complex courses of actions. An action f can be permissible as part of one course of action, yet impermissible as part of another. How we are to evaluate x’s having f-ed can thus depend on what the agent does at later times. The agent’s subsequent actions determine whether f-ing is extended in such a way that a permissible course of action results. This kind of impermissibility only shows up from the diachronic perspective. Past choices matter from this perspective, rather than being sunk costs that are to be ignored. The past matters for diachronic evaluations. Although the past is fixed, i.e., the agent cannot change the fact that he has f-ed, the evaluation of f-ing is not fixed. For instance, the agent can affect its evaluative status by extending it in a way that renders having f-ed impermissible. When including counterfactual alternatives in the evaluation, an available alternative can be dominated by something on a different branch of the tree that is no longer available but that was accessible at a prior choice node. One considers not only available alternatives that one can choose, but also counterfactual alternatives that one could have chosen. In the case at hand, E is not dominated by any of the available alternatives (although F is agent-neutrally better, it is agent-relatively worse). However, when E is considered as a continuation of D, it is dominated by the counterfactual

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

247

alternative B. Although E is not dominated by an available alternative, it is dominated by an alternative that could have been performed had one chosen differently at a prior choice node. As a result, F is the only undominated available alternative from the perspective of global choice and hence is the option that is to be chosen. Dynamic inconsistency is thus avoided. The set Xi contains the alternatives that are accessible from ni . Relative to a subsequent node nj some of these alternatives will be merely counterfactual, namely those in Xi n Xj . They are not in Xj , but could have been reached had the agent acted differently at ni . Although these alternatives are no longer accessible and hence cannot be chosen, they can nevertheless be used for evaluating available alternatives. Being undominated in Xj , i.e., :∃yðy ∈ Xj ∧ y ≫ xÞ, is then a necessary but not sufficient condition for ending up in the global choice set. In addition, alternatives have to be undominated in a suitably expanded set Xi that also contains counterfactual alternatives. The choice set relative to a given counterfactual expansion is selected by means of a myopic choice function CM ðXj − n Þ. It contains those members of Xj that are undominated relative to the counterfactual expansion resulting from going back n steps in the decision tree. We select alternatives from Xj on the basis of being undominated by those in Xi where j − i ¼ n. CM ðXj − n Þ ¼ fx: x ∈ Xj ∧ :∃yðy ∈ Xi ∧ y ≫ xÞg where j − i ¼ n We go back in the decision tree and expand the set of options with which the available alternatives are compared. Available alternatives are compared to counterfactual alternatives that could have been reached from the choice node that precedes Xj by n steps. The further back we go, the more counterfactual alternatives are included. Since contraction consistency condition  is satisfied, it follows that what is dominated in one set will also be dominated in every superset. CM ðXj − n Þ  CM ðXj − m Þ where n > m The global choice function evaluates whether available alternatives are undominated in a suitably expanded set containing counterfactual alternatives (where this includes the degenerate expansion in which no alternatives are added). The global choice function CG , accordingly, does not select an option that is not likewise selected by the myopic choice function CM . The global choice function considers a superset of alternatives, yet can only select available but not counterfactual alternatives from this superset.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

248

 . 

Since the myopic choice function CM selects precisely from these available alternatives, the choice sets stand in a subset relation. CG ðXÞ  CM ðXÞ There cannot be a switch from an alternative being unchosen when choosing myopically to being chosen when applying the global choice function. Instead, it is only possible for something to be chosen by a myopic agent yet rejected by a global agent. Accordingly, if the myopic choice set is a singleton, then there is no need to consider any counterfactual alternatives. Any superset Xj − n which includes counterfactual alternatives and for which the choice function CM ðXj − n Þ yields a nonempty choice set will contain this unique alternative. CM ðXÞ ¼ fxg ! CG ðXÞ ¼ fxg If some alternative is undominated in X0 , which contains the alternatives that could have been reached from the original choice node, then the global choice set contains all options that are undominated in this maximal counterfactual expansion. CM ðX0 Þ 6¼ ∅ ! CG ðXÞ ¼ CM ðX0 Þ Problems arise when the agent has made a mistake and picked a dominated alternative at some point in the decision tree. Mistakes ensure that CM ðX0 Þ ¼ ∅. Every available alternative is then dominated by some counterfactual alternative. This can happen in two ways. Either an agent makes a myopic mistake by choosing an option that is dominated by available alternatives—one goes down a branch that is such that all the outcomes that can be reached from that branch are dominated by outcomes that can be reached by taking a different branch; or an agent makes a global mistake by choosing an option that, though not being dominated by available alternatives, is nevertheless dominated by counterfactual alternatives (whereby the agent did not make any myopic mistakes at the choice nodes involved in the counterfactual expansion). If the agent has made a mistake and there are a number of different undominated alternatives such that CM ðXj Þ is not a singleton, then we cannot go back to the original choice node. Instead, we can only go back to the choice node nj − m that has a non-empty choice set, i.e., CðXj − m Þ 6¼ ∅ and that is such that all preceding nodes, i.e., Xj − n where n > m, yield an empty choice set. In the case of a myopic mistake at node ni one can only

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-    

249

expand up to node ni + 1 . By contrast, in the case of a global mistake at node ni one can expand up to and including node ni but not beyond that. We then choose myopically from Xj − m .³⁴ CG ðXj Þ ¼ CM ðXj − m Þ such that CM ðXj − m Þ 6¼ ∅ ∧ CM ðXj − n Þ ¼ ∅ for all n > m The options that remain after the set of counterfactual alternatives has been maximally expanded, without rendering the choice set empty, are the actions that are to be chosen and that are maximally rational from a diachronic perspective.³⁵

References Bader, R. M. 2015. “Kantian Axiology and the Dualism of Practical Reason,” in The Oxford Handbook of Value Theory, ed. I. Hirose and J. Olson. Oxford University Press, 175–204. Bader, R. M. 2016. “Conditions, Modifiers, and Holism,” in Weighing Reasons, ed. E. Lord and B. Maguire. Oxford University Press, 27–55. Cubitt, R. 1996. “Rational Dynamic Choice and Expected Utility Theory.” Oxford Economics Papers 48(1): 1–19. Dreier, J. 2004. “Why Ethical Satisficing Makes Sense and Rational Satisficing Doesn’t,” in Satisficing and Maximizing—Moral Theorists on Practical Reason, ed. M. Byron. Cambridge University Press, 131–54. Horton, J. 2017. “The All or Nothing Problem.” Journal of Philosophy 114(2): 94–104. Hurka, T. and Shubert, E. 2012. “Permissions to Do Less than the Best: A Moving Band.” Oxford Studies in Normative Ethics 2: 1–27. Kagan, S. 1989. The Limits of Morality. Oxford University Press. McMahan, J. 2017. “Doing Good and Doing the Best,” in Philanthropy and Philosophy: Putting Theory Into Practice, ed. P. Woodruff. Oxford University Press, 78–102. Parfit, D. 1982. “Future Generations: Further Problems.” Philosophy and Public Affairs 11(2): 113–72. ³⁴ CM ðXj − n Þ where n > j is defined as ∅, since the relevant set of counterfactual alternatives is undefined. This ensures that one goes back to the original choice node, such that the choice set is CM ðX0 Þ, if the agent has not made any mistakes by the time he reaches node nj . ³⁵ Thanks to Roger Crisp and especially to Kacper Kowalczyk for helpful comments. Thanks also to participants of the Arizona Workshop in Normative Ethics and the Oxford Population Ethics seminar.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

250

 . 

Portmore, D. 2008. “Dual-Ranking Act-Consequentialism.” Philosophical Studies 138: 409–27. Pummer, T. 2016. “Whether and Where to Give.” Philosophy and Public Affairs 44(1): 77–95. Sen, A. 1997. “Maximization and the Act of Choice.” Econometrica 65(4): 745–79. Sider, T. 1993. “Asymmetry and Self-Sacrifice.” Philosophical Studies 70(2): 117–32.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

12 Well-Being, the Self, and Radical Change Jennifer Hawkins

It is a well-known fact that people change over time. Sometimes the change is gradual. Sometimes it is swift. And some changes are deeper than others, in the sense that they alter more significant features of the person. In this chapter I explore cases of “radical change.” I define radical change as change where either (1) several of a person’s core values change, or (2) some deeper feature of her psychology changes (which will typically also result in value changes). An example of a deeper psychological change might be a change in personality—e.g., a change in one of the big five personality traits discussed by psychologists. Or it might be some other even deeper feature of the individual that changes. As should be clear, radical change as I define it, is a degree concept. Within the category of radical change there are more or less radical changes. I take it that most really radical change is bad for the person who undergoes it. Consider Phineas Gage, the nineteenth-century railroad worker.¹ An explosion sent a tamping iron right through Gage’s skull, yet miraculously he survived. However, he was dramatically changed. His fundamental personality traits were altered forever by the alterations to his brain, and his life went poorly from then on as a result. But there are also clearly cases in which radical change can be prudentially good. So what interests me is the question: What explains why some radical changes are prudentially good and others prudentially bad?² It is important to emphasize that I am discussing ¹ The story of Phineas Gage is related in the first chapter of Antonio R. Damasio, Descartes’ Error: Emotion, Reason and the Human Brain (New York: HarperCollins, 1994). My comment simply relies on the reader’s probable familiarity with Damasio’s way of telling the story, which has been influential. However, I thank an anonymous reviewer for drawing my attention to the fact that the details of the extent to which Gage changed remain disputed. See, e.g., R. Griggs, “Coverage of the Phineas Gage Story in Introductory Psychology Textbooks: Was Gage No Longer Gage?” Teaching of Psychology, 42(3) (2015): 195–202. ² This is a metaphysical question, not to be confused with the epistemic question of how we can come to know which changes are good and which bad. Jennifer Hawkins, Well-Being, the Self, and Radical Change In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0012

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

252

 

prudential change, the kind of change that improves (or lowers) the welfare of the individual who changes. I set aside for now admittedly complicated questions about radical change and morality.³ One might suppose there is little point to my question because we so seldom have control over radical change. Certainly Gage could neither anticipate nor control what happened to him. However, some types of radical change do fall within our control, and in the future, with the development of new technologies, more forms of radical change may become possible. Even now, it is often true of us that we could change ourselves in particular ways if we chose to. But while many of us engage in smaller-scale projects of selfchange, few of us seek to radically change ourselves because it rarely occurs to us that a radical change would be best. However, that may simply be an unfortunate feature of our limited personal perspectives. My own thinking about such questions has been deeply influenced by a particular picture of prudential facts I find attractive. I shall call this view the “future-based reasons view” or FBR. This is a very general metaphysical view about what makes it the case that such and such would be a good choice (prudentially) for someone. It has two components. First, whether a choice is prudentially good for an individual depends on the good (or bad) things that various possible futures hold for that individual. If choosing x now will lead to the greatest net future welfare for me, then I have the strongest prudential reason now to choose x. Second, these normative facts about future welfare are not in any way dependent on the subject’s attitudes or desires either in the present or the past. What makes it the case that x is good for A in the future is not the fact that A now desires x, nor the fact that in the past A desired x for her future self. Nor is it the fact that a fully informed A would desire it for her lesser informed self. Instead, what determines facts about future good are either facts about A as she is in the future or facts about the world A will inhabit at that time, or most plausibly, the interaction between both. In other words, the full explanation of what makes something good for a person at a certain time depends on facts about the person and the world at that time.

³ It is, of course, possible that a person might change in ways that make her more or less moral. Moreover, a morally good change might not be prudentially good, and a prudentially good change might not be morally good. But the question of how to resolve conflict between prudential reasons and moral reasons is a huge topic far beyond the scope of this chapter. When I say a person has most prudential reason to do x, I leave it open whether she has most reason to do x all-things-considered.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

253

To be clear, what I am calling the FBR view is not a theory of welfare, but rather a framework upon which one might build a theory. Various contemporary approaches to welfare are compatible with it. Both hedonism and objective list theories are. But then so would be any subjective theory that focused on current appreciation of things (as opposed to the prospective pro-attitudes desire theories typically stress). But not all theories are compatible. As should be obvious, the view I am describing rules out desiresatisfaction theories of welfare including the informed desire versions.⁴ It also rules out the idea, accepted by many theorists, that part of the welfare value of a life is determined by the overall shape of that life.⁵ It is beyond the scope of this chapter to defend these exclusions at any length, but the framework I favor is compatible with a variety of otherwise diverse theories such that what I have to say should be of interest to many theorists of welfare. FBR has the virtue that it easily accommodates cases of good radical change. On this account, a radical change represents the best choice (and a subject has most reason to undergo it) if it leads to the possible future with the greatest net welfare value. But by itself, without qualification, FBR also appears to have the problem that it allows too many cases of radical change to count as good. Thus it seems that a defender of any version of FBR will need to seriously consider questions about radical change and what explains the difference between the good cases and the bad.

⁴ Many theorists have defended some version of desire-satisfactionism, but the most popular versions among philosophers have been informed desire theories. One of the most developed such views is that of Peter Railton. See Railton, “Facts and Values,” Philosophical Topics, 14 (1986): 5–31; and Railton, “Moral Realism,” The Philosophical Review, 95 (1986): 163–207. It is beyond the scope of this essay to explain my exclusion of desire theories, but I have explained my opposition to them elsewhere. See Hawkins, “Well-Being, Time and Dementia,” Ethics, 124 (2014): 507–42; and Hawkins, “Internalism and Prudential Value,” in Oxford Studies in Metaethics, Vol. 14, ed. Russ Shafer-Landau (Oxford: Oxford University Press, 2019), ch. 5. For good general overviews of the desire satisfaction approach, see L. W. Sumner, Welfare, Happiness and Ethics (Oxford: Oxford University Press, 1996); Ben Bradley, Well-Being (Malden, MA: Polity, 2015); Guy Fletcher, The Philosophy of Well-Being: An Introduction (New York: Routledge, 2016). ⁵ Certain theorists reject additive views (or as some say “intra-life aggregation”) because they take seriously the idea that the shape of a life matters prudentially and assume the two claims are incompatible. See, e.g., Michael Slote, “Goods and Lives,” in Goods and Virtues (Oxford: Oxford University Press, 1984), 9–37; David Velleman, “Well-Being and Time,” Pacific Philosophical Quarterly, 72 (1991): 48–77; Joshua Glasgow, “The Shape of a Life and the Value of Loss and Gain,” Philosophical Studies, 162 (2013): 665–82. Dale Dorsey, “The Significance of a Life’s Shape,” Ethics 125 (2015): 303–30, has shown that, depending on why you think shape matters, taking shape seriously may be compatible with aggregation. Still it is clear that my approach is incompatible with common ways of understanding the shape of a life thesis.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

254

 

My overall aims are really quite modest. My first is simply to articulate the issues vividly, and thereby enable philosophers to feel the force of a certain kind of problem that, as I see it, has all too often been overlooked. Second, I hope to convince philosophers that some of what seem like the more obvious ways to resolve the problem—attempts that invoke one or another notion of identity—fail. No doubt many, once convinced that the simple solutions are flawed, will think that the obvious next move is to adopt a more sophisticated and nuanced version of the same general strategy. Yet I hope to make it clear that any such move must try to answer a deeper, as yet unanswered question, about the relationship between the self and welfare.

§1. Good Radical Change I begin by presenting a case where it seems intuitively plausible that radical change would be prudentially good. Consider Sharon. Sharon is a creative, artistically talented painter. Unfortunately, however, she is also predisposed to unipolar depression. Beginning in young adulthood she experiences recurrent episodes of what psychiatrists call “major depression.” When she is depressed she can’t do her artwork, or much of anything else. But when she emerges from a depression she can return to her art. Importantly, the depressive episodes take a toll not only on her creative work, but also on her personal relationships. Significantly, Sharon has a certain image of herself to which, over time, she has become deeply attached. It is the description under which she now values herself. She values art, of course. But she also values being an artistic person, which she equates not simply with painting but with a certain lifestyle and with a certain degree of social nonconformity. In the past, when struggling with her depressions, she has often comforted herself with the thought that, at least in certain people, artistic creativity and depressive mental illness appear to be somehow linked.⁶ As a result she has brought herself to a point where she thinks of her depression as the price

⁶ It is well established that poets, fiction writers, visual artists, and musicians are much more likely than ordinary people to suffer from either manic-depressive illness (bi-polar disorder) or unipolar depression. For example, one study found rates of depression eight to ten times higher among artists and writers than in the rest of the population. See Kay Redfield Jamison, Touched with Fire: Manic-Depressive Illness and the Artistic Temperament (New York: Simon & Schuster, 1996). The exact nature of the association, however, is not understood and remains a topic of dispute and continuing inquiry.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

255

she pays for artistic creativity. Not only is she resigned to her illness, but she has even become somewhat proud of it because it seems (to her) a mark of other qualities she values. When Sharon contemplates the lives of various people she knows, she judges their lives to be (in contrast with hers) lacking in depth and meaning. Now let us suppose a medication exists that could really help Sharon by putting a halt to the extreme emotional cycles she currently experiences. However, Sharon is skeptical about trying it. As we have seen, she has already formed a self-narrative in which depression is partly explained as a special sign of giftedness. And because of this she is now reluctant to give it up. She is also afraid that the medication might change her in ways that she can’t anticipate, but which she is pretty sure she (as she is now) wouldn’t like. She is both worried that it will make her “shallow” as opposed to “deep” and more concretely, that it may interfere with her artistic development. She fears she won’t be creative in the same way or to the same extent if she takes medication. These are Sharon’s fears. Now let’s assume they are well-founded. Let’s assume that if Sharon doesn’t take the medication she will continue to suffer major depressions and this will continue to limit her creative work. It will continue to make her life difficult in a variety of ways, and it will rule out certain kinds of long-term meaningful relationships. If she does take the medication, however, it will have a variety of effects on her. On the one hand, it will lift the depressions completely. However, it will simultaneously alter her personality in subtle ways, and this, in turn, will lead to her abandoning her current work. Though she will always appreciate art, for a variety of reasons, the person she would become with medication would not pursue an artistic career. However, she would find other projects that would be just as fulfilling for the person she would then be as artistic pursuits currently are for the person she is now. Moreover, once she is no longer living the chaotic life of depressive ups and downs, she will be able to form and maintain loving relationships and accomplish more in her new line of work than she would accomplish in the old. I take no stand for the moment on whether the goods of her new life are intrinsic or instrumental. Hedonists are free to see them as instrumental, and to think the medicated life will have more overall pleasure or happiness. And objective list theorists are free to imagine that her life with medication will contain more objective goods. And so on for other types of theorists. I will stipulate only one thing, which is that Sharon will appreciate the life

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

256

 

she thereby gains if she takes the meds. She will not in retrospect view the change as a mistake, but instead as an excellent choice. The change is radical in my defined sense. It will alter many of her values, and will do so in part by altering certain aspects of her personality (this is why she will no longer pursue an artistic career). Even though she herself has no current desire to change, and would not now view the possible life with medication as in any sense good for her, it is plausible that it would be good for Sharon to undergo this change. I suspect most readers will agree with me. If that is so, then they will also agree that at least some cases of radial change are prudentially good and we should want a theory of welfare that can accommodate that.

§2. Good Radical Changes and Future-Based Reasons The future-based reasons view is able to handle certain cases of radical change quite well. But only certain cases. To understand why this is so, it is helpful to have a more developed understanding of the view. The FBR view is first and foremost a way of thinking about the relationship between good prudential choice and prudential value itself. In ordinary life when we try to make a self-interested decision (or try to see what would be the best decision for someone else), we take it that there are various things that could be done. Associated with the different things that could be done are different outcomes that might follow. And these outcomes could lead to others and so on and so forth. Indeed, summarizing this familiar way of thinking, we can say that associated with each practical choice a person makes there are various possible futures, possible ways that the life she has been living up until then might continue. The FBR assumes that these possible futures have a certain welfare value for the subject in virtue of the various good and bad things those futures contain for her. What that value is will depend, of course, on one’s preferred account of prudential value and how much value various things are thought to have. A hedonist, for example, will want to know of each possible future how much happiness it contains, on the one hand, and how much suffering, pain, or unhappiness it contains on the other. The overall net value (negative or positive) of that possible future will depend on whether there is more happiness than suffering or vice versa. Alternatively, an objective list theorist would want to know of each possible future what kind of objective goods (friendship, achievement, knowledge, etc.) that life contains as well as what

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

257

kinds of objective bads. No matter the theory of prudential value, each possible life continuation will have a positive or negative net value depending on whether the positives (as construed by that theory) outweigh the negatives (as construed by that theory), or vice versa. One implication of this approach is that possible futures can, at least in principle, be ranked from best to worst. The FBR thesis assumes, as do many philosophers, that practical reasons—in this case prudential practical reasons—stem from facts about prudential value. For any two possible futures, x and y of an agent A, if x has greater net welfare for A than y, then A has more reason to choose x. But A has most reason, i.e., the strongest prudential reason, to choose the best possible future or the highest ranked one, whichever one that is. Of course, this is a theory of what reasons we have, and not a theory of what reasons we can easily know about or perceive. It may be that we often simply don’t know about certain options, or don’t think of them, and so fail to identify the best ones. That is an epistemic problem, and a serious one. But then it seems right to me that it should turn out to be extremely difficult—perhaps even impossible—to know what the best path through life is. We do reasonably well when we are able to recognize the more obvious paths forward, assess their relative strengths and choose one of the better ones. Finally, it is because of these epistemic barriers that normative reasons and motivating reasons so frequently come apart. It is important to see that not all good choices lead to what we would ordinarily call good outcomes. One kind of case where this is so involves bad lives. Sadly, some lives are so filled with suffering or other negatives that the negatives outweigh any positives the life might contain. I shall refer to all such lives where net value is negative as bad simpliciter. But of course, even within the category of lives that are bad simpliciter, there can be lives that are better or worse than one another. Now suppose someone faces a grim choice between two lives each of which is bad simpliciter in virtue of containing great suffering. Still, one life has less suffering than the other. If these are the only options, then it still makes sense to choose the less bad life, even though the life chosen would not count as good as people ordinarily think of “good.” This is a case where the best choice does not lead to a good life. This idea that there is a point at which the negatives begin to outweigh the positives (however construed) suggests for many a way to draw a line between a life worth living (or a continuation that is worth living) and one that is not. And some go further and suggest that when the only options available are lives not worth living, then death is preferable. Of course death

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

258

 

is not a “possible future” but rather the absence of a future. Still, it is fairly easy to adjust the comparison of futures to make it a comparison of lives that are the same up to a certain point. If a life that ends at that point is better than any of the possible lives where it continues, then the FBR can allow that in such a case death would be better for the individual in question.⁷ However, even when death is the better option, it is not always a real, practical option, in which case it still makes sense, insofar as there are choices, to choose the future that is least bad. Let us now consider what the FBR says about radical change. As we saw at the outset, some radical changes are not chosen, but simply happen to us. But when choice is involved and we can either change radically or not, then the important question will be about the overall welfare value of various possible futures with the change and without it. In some cases FBR will not rank highly any possible futures involving radical change. But sometimes it will. It can certainly occur that the best possible future for an individual is one that includes a radical change. If that’s so, the individual in question will have most prudential reason to undergo the change. Sharon’s case is like this. The best possible futures for her are those where she takes the meds. She will be much happier in the future with meds, and even non-hedonists typically grant significant prudential value to happiness. She will also have more success in her endeavors, even though those endeavors will be different in her new life than in her old. And she will have loving, lasting relationships. These three features will be absent from the non-med life, which will also contain much suffering from depression. If these are the only options, then it is clear that in welfare terms Sharon has most reason to take the medications. Here FBR seems to give the right verdict. But now consider a very different type of case, where the FBR view also recommends change. Consider the case of Chloe who is offered the following opportunity, which is presented to her (by the eager scientist who developed it) as “the magnificent alteration.”⁸ The purpose of the procedure is to optimize future welfare ⁷ This is just the familiar thesis (known as the deprivation view) that death is a comparative harm and that the proper unit of comparison is whole lives. For a detailed introduction to such views and their problems, see Steven Luper “Death,” The Stanford Encyclopedia of Philosophy (Summer 2016 Edition), ed. Edward N. Zalta, https://plato.stanford.edu/archives/sum2016/ entries/death/. ⁸ This example was inspired by Jeff McMahan’s example “The Cure.” See Jeff McMahan, The Ethics of Killing (Oxford: Oxford University Press, 2000), 77. In McMahan’s case an individual must choose between certain death or continued life after going through an equally radical change. McMahan’s purposes are different from mine. He aims to illustrate the fact that many

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

259

prospects. However, the change required to do so is very radical. A person who undergoes this procedure has her mind wiped clean, thereby losing all her beliefs, memories, values, etc. When she awakes she will not remember her past life. However, the changes go beyond alterations of mental content. The procedure will also alter her natural psychological dispositions in ways that “optimize” her personality, giving her just the right degree of (for example) extroversion, or conscientiousness. This will be the precise degree of extroversion (or conscientiousness) that most often leads to happiness (or success as defined by other theories of prudential value). Other traits will be similarly fine-tuned. After the procedure is over, she will have no real reason to try to recreate her old life because, given the deeper psychological changes she has undergone, that life will no longer fit her. Chloe, understandably, is appalled by this description and wonders why this crazy man thinks she would want it. But his answer is that it creates for her a possible future with much greater overall well-being. As she is now, her life is going fairly well, but there are still some definite, long-standing problems. The person she is currently faces a lifetime welfare ceiling, one she could surpass if she changed. Currently, her values and her personality sometimes undermine her ability to achieve her goals. Studies of people who have previously undergone the magnificent alteration show that it works wonders. A fresh start with a different psychological profile will enable Chloe to have a life much higher in overall welfare value than any life open to her as she is. The example is deliberately extreme. Yet it is not crazy to suppose that for some people really radical change would enable them to live much better lives from the standpoint of welfare. In such a case, the best possible prudential future is the one that includes radical change and so FBR says that the individual has the strongest prudential reason to undergo the change.⁹ And yet, most people, including myself, have the intuition that it

people care less about a future self that is radically different from their current self. He goes on to use these facts about care to support his theory of time-relative interests. However, even if he is right about most people’s feelings, I doubt these feelings have the normative significance he assigns them. In an unpublished companion essay to the present one I argue against McMahan’s theory of time-relative interests. ⁹ McMahan says of his example, “The Cure,” that “most of us would at least be skeptical of the wisdom of taking the treatment and many would be deeply opposed to it” (McMahan, The Ethics of Killing, 77). However, FBR will clearly say that one ought to take the cure, because the only options are radical change or death, and it is clear that the life after radical change is well worth living, i.e., it has net positive welfare value for the subject.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

260

 

couldn’t be prudentially best to undergo such a change. Not only does Chloe not have most reason to choose it, she may have little or no reason to given that her alternative, while not perfect, is still a perfectly good life overall. How do we explain this in a way that preserves our original intuition about Sharon, namely, that sometimes radical change can be the best option, the one we have most reason to choose?

§3. Why Identity is Not the Solution: Part I. Numerical Identity Our question then is why the magnificent alteration is not the prudentially best choice for Chloe. What explains this fact? When presented with this kind of case, most philosophers reach for a certain kind of answer, an answer framed in terms of identity. The basic gist of the response is that this is not a good option for Chloe, because the individual post-change is no longer Chloe. Now it may be that the answer lies somewhere in this area. But there are at least two ways of understanding or interpreting this idea that strike me as quite unpromising. I want first to explain what these are and why they fail. This will set the stage for my explanation of why I remain skeptical of the entire strategy. There are at least two quite different things philosophers talk about using the language of identity. These are what I shall here call numerical identity and character identity.¹⁰ I examine numerical identity first. Questions about numerical identity are questions about the persistence conditions over time for a certain type of entity. It is true that if a change were so radical that it caused the individual who changes to go out of existence, we could not, in the cases that interest us, say that this was good for the individual. But is that what radical change does in a case like Chloe’s? Does it destroy one of us? That depends on what you think we fundamentally are.

¹⁰ Schechtman draws the same basic distinction in terms of two questions: the re-identification question and the characterization question. See Marya Schechtman, The Constitution of Selves (Ithaca, NY: Cornell University Press, 1996). DeGrazia draws the same distinction in terms of numerical identity and narrative identity, but since I want my argument to apply to a broader set of theories of self than just narrative views, I avoid his terminology and refer to “characterization identity.” See David DeGrazia, Human Identity and Bioethics (Cambridge: Cambridge University Press, 2005).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

261

A theory of numerical identity for x’s presupposes an understanding of what an x essentially is. Thus, closely allied with theories of numerical identity are theories of our essence, theories that answer the question of what we most fundamentally are. There are (at least) three popular answers to the question of our essence. To begin with, we are human beings, a type of animal. Animalist theories of identity over time explain our persistence conditions in terms of the persistence of a particular animal or organism.¹¹ We are also, however, sentient creatures in virtue of having a brain that persists through time and produces conscious experiences. Embodied mind accounts of identity explain our persistence conditions in terms of the continued existence of a brain capable of conscious experience.¹² And finally, we are persons in the philosophical sense of beings with certain complex psychological capacities such as higher-order desires, and awareness of ourselves as temporally extended beings.¹³ On these views our persistence through time consists in the persistence through time of certain psychological connections, often described as overlapping chains of psychological connections. For a long time the question about the persistence conditions of beings like you and me was referred to as “the problem of personal identity.” But this description is misleading. It runs together two distinct questions. One is the question of what it takes for the same person to survive a change. The other is the question of whether we are essentially persons, such that the destruction of the person I now am would be the death of me, or whether I could survive as a living non-person. For a person essentialist an answer to the first question is also an answer to the second. But for other types of theorists—for example, animalists and embodied mind theorists—the questions are importantly distinct. For these theorists it is quite possible to accept as a theory of what it takes for a person to continue one of the familiar psychological accounts of psychological continuity, even while rejecting the idea that you and I are essentially persons. For these theorists personhood is a phase of our life, but it does not necessarily set the boundaries of our existence.

¹¹ For examples of such a view, see, e.g., Eric T. Olson, The Human Animal: Personal Identity Without Psychology (Oxford: Oxford University Press, 1997) and Peter van Inwagen, Material Beings (Ithaca, NY: Cornell University Press, 1990). ¹² For several defenses of this kind of view, see McMahan, The Ethics of Killing; DeGrazia Human Identity and Bioethics. ¹³ There are many such views: e.g., Derek Parfit, “Personal Identity,” Philosophical Review 80 (1971): 3–27 and Reasons and Persons (Oxford: Oxford University Press, 1984); Sydney Shoemaker, “Self and Substance,” in Philosophical Perspectives, ed. J. Tomberlin (1997), 11: 283–319 and “Self, Body, and Coincidence” Proceedings of the Aristotelian Society (Supplementary Volume), 73 (1999): 287–306.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

262

 

Importantly, for a long time the two questions were routinely conflated, in part because so many theorists unreflectively assumed person essentialism. There are various diagnoses of what led to this. For example, Eric Olson contends that for a long time philosophers conflated issues of practical concern—ethical issues—with metaphysical ones. Psychological continuity theories may help us answer many of our important ethical concerns, but they are very poor accounts of what we are metaphysically. I myself am inclined to think I could survive as a non-person (for example, if I ever develop dementia and live to the end stages of the disease) or as a different person (for example, if I went through a procedure like the magnificent alteration). It also seems plausible to me that I started life as a non-person (a fetus). On the view I favor “person” is just something I am during a certain phase of my life, much as I am a parent now. Saying this need not imply that persons are unimportant or that it is not typically better to be a person than to be a non-person. However, the important point for the purposes of this chapter is that radical change of the sort that disrupts psychological continuity enough to potentially create a new person or a nonperson, will only disrupt numerical identity if it turns out that we are essentially persons. So we can only hope to use numerical identity to explain the problem with Chloe’s case if we are person essentialists. I have so far not explained why I think we should reject person essentialism, but I will now. As I see it, the decisive considerations have to do with the familiar notion of anticipation. Anticipating our own future experiences is very different from knowing that someone else will experience certain things in the future. And when we look to intuitions about which future experiences we can anticipate having, these suggest that one of us would survive even radical changes of the sort described in Chloe’s case. Bernard Williams offered the following famous example that highlights the issue. He asked his reader to imagine being captured and then told the following: “A series of things are going to be done to you, but the upshot is that the contents of your mind will be fully erased. Once the procedure is complete, the living individual who remains—an individual who will have the same brain as you but who will not think of himself as you nor have any memories of your life—will gradually be fed a series of false memories and other mental contents so that he comes to think of himself as a different person. After this is all complete, this individual will be tortured.”¹⁴ ¹⁴ Williams stops short of drawing the strong conclusion that I draw, namely the conclusion that numerical identity can survive the destruction of the person. His paper instead emphasizes

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

263

Williams asked whether or not it makes sense to be concerned about what will happen to this later individual. Obviously you will be upset by the prospect of procedures that will erase your mind. But by hypothesis, you can’t stop that. Should you fear the torture that the post-procedure individual will experience? Williams suggests (and I believe) that fear of the torture makes perfect sense. But this, in turn, suggests that numerical identity is something that can survive the destruction of all one’s psychological features. It can survive quite radical change of the sort that would alter or even destroy the original person. Admittedly, not all theorists will agree with me, as these issues remain deeply contested. Anyone committed to person essentialism will think that the destruction of the person is the death of me, and that there is no sense in which I can be said to experience that later pain. However, for those who accept my rejection of person essentialism, it should be clear that we can’t explain the badness of radical change in Chloe’s case by appeal to numerical identity. On the view I favor, Chloe is still around and has not changed her (numerical) identity even after such a radical alteration.

§4. Why Identity is Not the Solution: Part II. Character Identity There is another very common use of ‘identity’ according to which our identity is, roughly, our self. An account of our identity in this sense is an answer to the question “Who am I?,” a question Marya Schechtman labels “the characterization question.”¹⁵ There are actually two closely related concepts here. Sometimes philosophers are interested in what they call “the true self.” An account of the true self is an account of which character traits, values, beliefs, etc. are truly definitive of a particular person. Sometimes, however, philosophers are more interested in an individual’s

the ways in which our intuitions can apparently go quite different ways depending on how a case is described. However, many philosophers (including myself ) have subsequently been willing to use examples like this one, which comes from Williams, to draw stronger conclusions than Williams did in the essay where this first appeared. See Bernard Williams, “The Self and the Future,” The Philosophical Review 79(2) (1970): 161–80. ¹⁵ See Schectman, The Constitution of Selves and “Staying Alive: Personal Continuation and a Life Worth Living,” in Practical Identity and Narrative Agency, ed. Kim Atkins and Catriona Mackenzie (New York: Routledge, 2008), 31–55.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

264

 

own self-conception, the particular way that she answers for herself the characterization question. Either way, however, on most such views, character identity is primarily defined in psychological terms. Who I am depends on things like my personality, my values, my characteristic responses to things, and so on and so forth. Although interesting questions can arise about who someone really is if or when her self-conception and the facts about her psychology diverge, for the purposes of this project such issues are not really important. We are simply interested in how much change a self (or a self-conception) can undergo before we have a new self (or before a person’s self-conception is so different as to be a new self-conception). The notion of self may seem more relevant to explaining the badness of certain radical changes. Unfortunately, these views, though interesting and important in many respects, tend to be vague when it comes to the questions that interest me. How much change and what kind of change is compatible with being the same self or the same person in the characterization sense? What accounts for the difference between a case where a change occurs and what we have afterwards is the same person with some different qualities (call this “oldself-modified”) and cases where so much has changed that now we have a new person entirely (call this “new-self-formed”)? It is presumably a matter of degree, and different theorists will answer differently. Obviously, because the Chloe case is so extreme, almost any view will see it as a case of new-self-formed. But recall that we want an account that allows us to draw a plausible line between a case like Chloe’s and a case like Sharon’s. We want to continue to be able to say that radical change is the best option for Sharon and that she has most prudential reason to choose it. We don’t want to purchase the right answer for Chloe at the price of giving up our claim about Sharon. Thus everything turns here on whether theories of the self would see Sharon after she takes the meds as a new-self-formed or as old-self-modified. Consider then a cluster of popular theories of self united by the emphasis they place on a person’s core values. Call these value theories of self. For example, many theorists, inspired by Harry Frankfurt have come to understand the self in terms of commitments (and values) that are made (or adopted) by a higher-order self (or perhaps the rational self ).¹⁶ Frankfurt

¹⁶ Schechtman (The Constitution of Selves and “Staying Alive”), among others, cites Frankfurt as an example of someone who offers a theory that can be read as an answer to the characterization question. Another example she cites is Christine Korsgaard, The Sources of Normativity (Cambridge: Cambridge University Press, 1996) and Self Constitution: Action,

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

265

actually says more about personhood than selfhood.¹⁷ He gives us an account of what it takes for a person to remain a person (as opposed to a non-person). Roughly, for Frankfurt, personhood is tied to the capacity for developing stable higher-order attitudes. Lose that capacity and you will cease to be a person. However, many people interested in the self have also been inspired by Frankfurt’s ideas, and it is clear why. There are materials here that seem well suited to answering the characterization question. Who I am at a given time may also be determined by the attitudes of the higherorder self, or the rational self. What I come to value in this reflective way may serve as an account of who I am, most fundamentally. However, this still doesn’t help us to answer the question of when a person (without ceasing to be a person) becomes different enough to count as a new or distinct person. Presumably I could maintain throughout life the capacity for higher-order thought (and so always remain a person), even while changing my mind about some of the things that matter. I doubt any advocate of such a view would deny that. But what they don’t say is how much such change is compatible with remaining the same person in the characterization sense. Alas, there is no precise answer to be found. So it remains unclear whether this kind of approach would count Sharon’s changes as sufficiently radical to make her a new person. Still, somewhat worrisomely, it seems plausible that a philosopher working in this vein might say that post-medication she is a new-self. A few years out from her change (if she undergoes it) she will care very little about most of the things that seemed important to her before. People who knew her before will be struck by the depth of the change. Our goal, however, is to find a way to use the notion of identity to distinguish between Sharon and Chloe. So only if we are certain that such a view of self can accommodate the intuition that Sharon’s change is best for her should we appeal to it.

Identity and Integrity (Oxford: Oxford University Press, 2009), though Schechtman writing in “Staying Alive” cites the version that, at the time, was available on Korsgaard’s website. ¹⁷ See, e.g., Frankfurt, “Freedom of the Will and the Concept of a Person,” Journal of Philosophy 68(1) (1971): 5–20. For other Frankfurt essays key to the interpretation of theorists like Schechtman, see “The Importance of What We Care About,” Synthese: An International Journal for Epistemology, Methodology, and Philosophy of Science, 53(2) (1982): 257–72; “Identification and Wholeheartedness,” in Responsibility, Character and the Emotions: New Essays in Moral Psychology, ed. Ferdinand Schoeman (New York: Cambridge University Press, 1987), 27–45; and “Identification and Externality,” in The Identities of Persons, ed. A. Rorty (Berkeley: University of California Press, 1976), 239–52.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

266

 

Now consider briefly a very different approach to understanding the self and answering the characterization question. This is a type of view that has become quite popular in recent years: the narrative self view.¹⁸ Though there are many variants of this approach, the basic idea is easy enough to grasp: it is that a person’s self is constituted by a narrative. This narrative is the story she tells herself about herself and about what has happened to her and why. It is composed of much less than all the things that have happened in her life. Rather it incorporates those happenings that seem important to her, and that stand in special narrative relations to other events in her life. Nor are these events limited to just the ones that seem good to her. Rather, the narrative incorporates events both good and bad that need to be understood or that help to make other events intelligible. The themes and patterns that emerge help individuals create meaning out of what might otherwise seem like a random series of events. This selfconstituting narrative allows a person to make sense of where she has been and where she is going. As before, we want to know whether the narrative-self view can handle Sharon. It is pretty clear that it will deem Chloe to be a new-self after the magnificent alteration. After all, there could be no more thorough disruption to a narrative. Yet presumably the narrative self view, like value based theories of self, can accommodate a certain degree of change as long as there is a way to make sense of the change within a single, coherent story. The story must, however incorporate what has happened before the change and after it. This is not a huge constraint, since stories can develop in so many ways. For any particular beginning there are presumably many coherent continuations. And yet, even so, I don’t know whether such a theory could be trusted to give the right verdict about Sharon. Sharon has a self narrative that she has developed over time. Indeed, as explained earlier, part of her resistance to taking the medications is that she worries it will change her in ways that, from the perspective of her current narrative, are incoherent. Suppose she takes the meds. For her to reap the benefits of the change, sooner or later she will need to acknowledge that the changes she thought would be bad are not really so. But this will require abandoning key elements of the old story.

¹⁸ Again, there are many examples, but Schechtman, The Constitution of Selves, is a main defender. DeGrazia, Human Identity and Bioethics, also accepts a narrative view of self.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

267

For example, she may eventually have to admit that her earlier self was not nearly as “deep” as she took herself to be, nor were others so “shallow.” I would thus not be surprised if it were difficult or even impossible to find a single coherent narrative that could incorporate all of this in the right way. Again, until we are sure the narrative self view can give the right verdict about Sharon, we should not appeal to it to draw the line between good and bad radical change.

§5. Persons, the Self and Welfare I have described a problem that arises in one form or another for many theories of welfare, namely, the problem of radical change—the problem of capturing deeply held intuitions to the effect that certain kinds of extreme changes could either never be prudentially best or could only rarely be so. I have also tried to explain why what might seem like the most obvious response to this problem will not straightforwardly work. In this last section I present considerations that suggest that the whole general strategy of appealing to characterization identity is misguided. I then sketch an alternative approach. Ordinarily we tend to suppose that we have at least a reason to resist too much change to our selves, to who we are. Of course, this reason can be overridden if enough factors point a different way. But it would presumably take a lot to override it. Yet, what makes us think we have such a reason? There may be moral reasons in the vicinity. But even so, if that is all we can say, we have not done anything to block the conclusion about welfare. The intuition we started with, however, was precisely that radical change was bad for Chloe. What we would need instead is a prudential reason to remain as we are. But why think remaining the same has general prudential value? If we set aside temporarily the cases we have been discussing, we can see that in many real-life situations radical change is prudentially bad. Moreover, radical changes are risky because in real life it is so hard to know what the net effects of such a comprehensive change will actually be. These facts may explain our general presumption against radical change, but they don’t support a reason not to change. And they don’t help us with Chloe. There are other reasons why we are usually strongly resistant to change. On most views of the self a person’s values are a key component of who she is. But it is at least partly constitutive of genuinely valuing something that

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

268

 

one takes oneself to be correct in doing so.¹⁹ In other words, if I value a certain kind of project, I not only approve of it or see value in it, but I may feel that I would be mistaken not to approve of it and also mistaken not to have it in my life. While understandable, this seems to conflate the value something has in itself (in terms of which it may merit a positive evaluation) and its prudential value for me. Even if the thing has great value, it might not be a mistake to remove it from my life. Again, however, this explanation of our attitudes does nothing to justify them or show that we have reason to resist self-change. Perhaps one might argue that radical self-change just is, in itself, a brute kind of prudential bad, one that has significant negative weight and which gains more negative weight as change becomes greater. Of course, it would still be just one of many prudential goods and bads that would factor into an assessment of the net value of a possible future, but because it is so very bad, really radical change would rarely (if ever) turn out to be prudentially best. This yields the right answer at the price of great mystery. What could motivate such a move? If that general line of thought goes nowhere, then perhaps we should reconsider the resources available within the FBR. One important kind of consideration can be brought to bear to explain why many radical changes would not be prudentially best. Whenever a person undergoes a radical change there will be transition costs. The greater the change the greater the transition costs. For example, if I have changed dramatically, I may not enjoy or even have the aptitude for many of the things I once did. So I will need to find new projects to engage me. I also might need to find new friends or form new relationships. I might need to find a new career. Yet all of these things take time. While I am working to recreate my life there will be a period of time where I have lost the old goods and do not yet have the new ones. So there will be a period of very low, perhaps even negative, welfare value. The thing to notice is that any possible future that involves radical change and yet has net positive welfare value is one where the good that follows the transition must be so good it can compensate for the bad and still tip the balance to the positive. If the transition costs are big enough then the goods

¹⁹ For example, Agnieszka Jaworska writes, “one would always view the possibility of not valuing something one currently values as an impoverishment, loss, or mistake.” See Jaworska, “Respecting the Margins of Agency: Alzheimer’s Patients and the Capacity to Value,” Philosophy and Public Affairs 28(2) (1999): 105–38.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

-,  ,   

269

that follow must be tremendous. Finally, if a possible future is best (as opposed to just net positive), then it will have to be good enough to cancel the transition costs and still come out with more positive value than other relevant possibilities, and there will typically be a number of these all of which involve less change. This is a steep requirement. However, I don’t deny that some theories will meet it. Yet which theories meet it and how often they recommend radical change will depend on the details of the theory of prudential value. Consider briefly a simple form of Benthamite hedonism. Pleasure is the only within life intrinsic value, and it is construed as a simple feeling the value of which is exhausted by facts about intensity and duration. Transition costs are typically highest when we try to recreate post-change complex goods like a network of friends or an engaging career, things that typically take years to establish. But if things like friendships and careers have value only instrumentally insofar as they lead to pleasure, then it may well be that the best strategy after a radical change is to look for simpler, more easily attainable sources of pleasure. Assuming there are some, as seems plausible, then there will be a fair number of cases where radical change is part of the best possible future for a person. So she will have most reason to select it. But those possible futures will look quite different from futures in which we work to establish new networks of friends. In short, simple hedonism would presumably recommend radical change far more often than other theories because it has ways to minimize transition costs. But its ways of minimizing transition costs will not strike many as really prudentially good, thus revealing another implausible consequence of simple hedonism. Suppose, however, we assign prudential value directly to things like engagement in valued projects and the development and maintenance of good relationships. And suppose we allow that such things have much more value when (in the case of projects) they engage more of our faculties or when (in the case of relationships) our feelings are strongest and returned, and the relationship lasts a long time. On such a view the transition costs of radical change begin to look formidable. Even so, I imagine there will be cases where radical change turns out to be worth it. Someone who is very young has years ahead to refashion a life, including time to develop and nurture new relationships, and so on and so forth. If that life, once created, would be dramatically better, then it might prudentially speaking be worth it for this person to start over. Even for someone who is not quite so young but who has no very good relationships or no valuable, engaging projects, it may make sense to start over. Still, once

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

270

 

we adopt a more sophisticated theory of value, transition costs can go a long way towards explaining why radical change is rarely the best option. Such an approach can explain why Sharon’s change is good. She is young, her depression is undermining her relationships and she has the potential to do more and relate better if she changes. As for Chloe, how we view the case may depend on details that were not initially provided. It was simply stipulated when the case was presented that it was one where the prudential value of the possible future with radical change was clearly much higher than any others. But is that even possible? It is of course logically possible, but is it going to turn up a real possibility given the world as it is and the facts of human psychological being what they are? Perhaps it would if her life to date were miserable, and she had no other possibilities that would lead to engaging work and deep relationships. Then it might be plausible that the greatest net value lies on the other side of the magnificent alteration, even given the huge transition costs. But it now seems like a very rare kind of case indeed. Obviously I have not solved these problems entirely. It remains to be seen whether this approach can resolve all the problems. But it seems more promising to me than a focus on the importance of identity. For it seems to me that the importance of identity is not a brute prudential fact. Rather it matters, when it does, because of what else depends on it. Some of my prudential goods require that I be me. And since usually they are the best goods open to me, it remains important that I remain me.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

13 From Teleosemantics to Normative Ethics Jacob Ross

The morally wrong actions, it seems, are the actions that are worthy of moral disapproval. Hence, one way to approach normative ethics is to ask the following question: Toward what kinds of action is moral disapproval fitting or correct? In this chapter, I will argue that we can answer this question by examining the cognitive content of the attitude of moral disapproval. For if moral disapproval is an attitude with cognitive content, then this attitude will be fitting or correct just in case this cognitive content is satisfied. Hence, by determining the cognitive content of moral disapproval, we can determine what kind actions it is fitting to have this attitude toward. Of course, if all we can say about the cognitive content of moral disapproval is that this attitude represents its object as morally wrong, then we will have made no progress in understanding moral wrongness. What we need is a theory of cognitive content that allows us to identify the content of moral disapproval in naturalistic terms. Much of this chapter, therefore, will be concerned with sketching and defending such a theory. I will proceed in three stages. In the first part of the chapter, I will defend a general account of the cognitive content of attitudes. This will be a teleosemantic account, in the sense that it will explain cognitive content in teleological terms, and ultimately in terms of natural selection. The account I defend will imply that the cognitive content of an attitude depends on its functional role. Accordingly, in the second part of the chapter, I will defend an account of the functional role of moral disapproval. Then, in part three, I will combine the results of the first two parts in order to determine the cognitive content, and hence the correctness conditions, of moral disapproval. I will conclude with some remarks about the normative implications

Jacob Ross, From Teleosemantics to Normative Ethics In: Oxford Studies in Normative Ethics Volume 9. Edited by: Mark Timmons, Oxford University Press (2019), © the several contributors. DOI: 10.1093/oso/9780198846253.003.0013

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

272

 

of the resulting view. In particular, I argue that this view supports a kind of moral relativism.¹

1. Toward a Teleological Theory of Content 1.1 The General Strategy In what follows, I will assume that some attitudes have cognitive content, and that these attitudes are correct or fitting just in case this content is satisfied. Many authors have held that this is true, for example, of the attitude of fear. In fearing something, it seems, we represent this object as dangerous to us. Hence, our fear of X is fitting or correct just in case X is indeed dangerous to us. I will assume that something similar is true of moral disapproval. Thus, I will assume that if someone disapproves of promisebreaking, they represent promise-breaking as having some property, and their attitude is fitting or correct just in case promise-breaking has this property. Now suppose we have a naturalistic theory of content, according to which an attitude A has content X just in case A stands in some natural relation R to X. Since, presumably, attitudes stand in natural relations only to natural properties, natural states of affairs, etc., such a theory will ascribe contents in naturalistic terms. Hence a naturalistic theory of the contents of attitudes should enable us to specify, in naturalistic terms, the conditions under which it is fitting to disapprove of something. The kinds of naturalistic theories of content I will be focusing on are teleosemantic theories. The basic idea of teleosemantics is this. The notion of cognitive content is normative: mental state with cognitive content can either get things right or get things wrong. In this respect, this notion resembles teleological concepts that figure in biology, such as the concept ¹ This chapter explores the possibility of deriving normative ethical implications from a teleosemantic framework. For interesting discussions of the metaethical implications of a teleosemantic framework, see Harms (2000) and Sinclair (2012). One important difference between my approach and that of Harms is that I apply teleosemantics to a moral attitude (namely, moral disapproval) in order to determine its fittingness conditions, whereas Harms applies teleosemantics directly to moral utterances in order to determine their truth conditions. One reason for favoring my strategy over Harms’ is brought out in Joyce’s (2001) response to Harms, where Joyce points out that Harms’ argument requires the assumption that natural selection equipped us with a special “moral signaling system” that underlies our moral discourse. However, Joyce argues that it is more plausible that “what the pressures of natural selection favored (in humans) was a range of moral emotions” (2001, p. 726). My argument requires only the latter assumption, not the former, more contentious assumption.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

273

of a biological function. And we can give reductive, Darwinian accounts of the normative concepts that figure in biology. Hence, if the normativity of content could be understood in terms of biological normativity, then we may be able to give a reductive account of content. In order to motivate the teleological view I will ultimately defend, I will begin by considering what are currently the two main families of teleological theories of content. In thinking about these two families of views, we should keep in mind a point made by Ruth Millikan, who says that teleological theories of content are often best understood “as specific forms . . . of other kinds of theories” (2009, p. 394). For “the teleologist can take any naturalistic theory of representation at all and turn it into a teleological theory of content.” (2009, p. 396). I will use this idea to classify the teleological theories I will be examining. I will first consider teleosemantic theories that can be seen as versions of success semantics, and then turn to teleosemantic theories that can be seen as versions of indicator semantics.

1.2 Teleological Success Semantics According to success semantics, the content of a belief is that condition that must obtain in order for the kind of behavior motivated by this belief to be successful. Traditional versions of success semantics understand successful action in terms of desire-satisfaction (Whyte 1990, 1991). Hence, they are not fully reductive, since they understand the content of belief in terms of the content of desires. Teleological versions of success semantics, however, are more ambitious. On such views, cognitive content is ultimately explained in terms of the organism’s biological ends. Theories of this kind were developed independently by Ruth Millikan (1984, 1989, 2009) and by David Papineau (1984, 1993). On the simplest teleological version of success semantics, the cognitive content of a mental state is that condition that must obtain in order for the kinds of behavior motivated by this mental state to achieve the biological ends they serve. I will focus on this simple version, but the worries I raise about it are quite general and apply, in modified form, to other teleological versions of success semantics. A key feature of teleological success semantics, which differentiates it from the kind of view we will be considering in the next section, is that it implies that our cognitive representations can be generally mistaken. Millikan illustrates this point with the example of the hoverfly (1989,

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

274

 

p. 291). There is a mental state, F, belonging to male hoverflies with the following features: (a) F is triggered by the sight of objects moving across the hoverfly’s visual field in a certain characteristic manner. (b) F causes the male hoverfly to pursue the object in question and, if it turns out to be a female hoverfly, to initiate mating behavior. State F can be triggered by all sorts of things, including male hoverflies, insects of other species, or even distant birds, and F is triggered by female hoverflies only in a small minority of cases. (1990, p. 331). Nonetheless, Millikan maintains that F represents a passing female hoverfly (1989, p. 291; 1990, p. 336). For the purpose of the chasing behavior is to mate with female hoverflies—this is the biological end served by the behavior which explains why the behavior evolved. And the chasing behavior normally promotes this end only when the object in question is a female hoverfly. Hence, the view under consideration implies that F represents the presence of female hoverflies, since this is the condition that must obtain in order for the behaviors motivated by this state to promote their end. I will now argue that this theory of content to ascribe cognitive contents to mental states in such a way as to generally make sense of the behaviors they motivate. This can be seen by considering the deer. A deer can respond to potential predators either by the flight response or by the freeze response. The flight response is more effective at avoiding predators, but it is also more costly, since flight requires a significant expenditure of time and energy. Thus, if deer engaged in the flight response every time there was evidence that a threat might be present, they would be unable to carry out the other tasks necessary for survival. Consequently, deer evolved to engage in the flight response only when there is strong evidence of predators, and to engage in the freeze response when the evidence is weaker. Where M1 represents the mental states that cause the freeze response, what does M1 represent, according to teleological success semantics? It seems it must represent the presence of predators, or some stronger condition that entails the presence of predators. For it is only in the presence of predators that freezing benefits the deer—if no predators are present, the deer would be better off carrying on with its regular activities. However, if M1 represents the presence of predators, or some stronger condition that entails this, then the content of this state does not make sense of the behavior that this state motivates. For, if the deer really believed that a predator is

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

275

present, it wouldn’t make sense for the deer to freeze. After all, if a predator really were present, then it would be best for the deer to take the most effective means of avoiding the predator, namely, fleeing. The content that would rationalize the freezing response is not the presence of a predator, but rather something like a middling chance that a predator is present. For it is when it is likely, but not extremely likely, that a predator is present that freezing rather than fleeing makes sense. However, teleological success semantics cannot assign this content to the mental state. For whether a given behavior benefits the organism depends on what is actual, and not what is likely given the organisms’ evidence. No matter how likely it may appear that a predator is present, so long as no predator is actually present, freezing will not benefit the deer. I will mention one other problem facing teleological success semantics. This view can only ascribe cognitive content to mental states that motivate behaviors that promote the ends of the organism. But it seems there could be contentful mental states that motivate the organism to act irrationally or against its (reproductive) interests. This seems possible for the following reason. In order for a disposition to engage in a given type of behavior to be selected for, all that is required is that this disposition promote reproductive success. And a disposition to engage in a given type of behavior can promote reproductive success even if these behaviors themselves do not, and even if these behaviors have the opposite effect. As an extreme illustration, consider a disposition, belonging to a world leader, to respond to a nuclear strike with a counterstrike so cataclysmic that it destroys all life on earth. Obviously, engaging in this behavior would not promote one’s interest. But being known to be disposed to engage in this behavior may promote one’s interests, since it may serve as an effective deterrent. Thus, assuming that others have a reasonably reliable ability to discriminate whether one has this disposition (this ability to discriminate needn’t be perfect), it can be in one’s interest to have this disposition, even though the behavior to which it disposes one is clearly counterproductive. A number of theorists, following Robert Frank, have argued that emotions frequently work like this.² For many emotions, such as anger, depression, etc., often seem to motivate irrational, counterproductive behaviors. It has been suggested that these emotions are adaptive not because the behaviors they produce are adaptive, but rather because being known to have these

² See Frank (1988) and Greenspan (2000).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

276

 

dispositions transforms our strategic situation in advantageous ways. Later I will argue that something similar may be true of moral disapproval. But if this strategic model of these attitudes is correct, then we can’t determine their content using teleological success semantics. For this theory only applies to mental states that produce behaviors that promote the organism’s ends. Thus, teleological success has serious problems, and arguably does not apply the mental states we are interested in. Let us therefore turn to the other main kind of teleological theory of content, namely teleological indicator semantics.

1.3 Teleological Indicator Semantics In its simplest form, indicator semantics maintains that a mental state represents whatever condition it tracks or reliably indicates. The teleological version of indicator semantics can be seen as an attempt to address a major problem for this simple version, a problem concerning the possibility of systematic misrepresentation. Suppose frogs have a mental state, F, that represents the presence of flies and that causes the frog to snap out its tongue at whatever is in front of it. It seems that the frog could be disposed to mistakenly enter this state in the presence of BB pellets and hence to misrepresent BB pellets as flies. But in this case, the state will reliably indicate the presence of flies-or-BB-pellets. Hence, the simple version of indicator semantics will imply that it represents flies-or-BB-pellets. And so this theory will imply that, when the frog enters this state in response to a BB pellet, there is no misrepresentation after all.³ The teleological version of indicator semantics aims to solve this problem. On this view, first proposed by Fred Dretske, what a mental state represents is not what it in fact reliably indicates, but rather what it has the biological function of reliably indicating.⁴ More precisely, this view implies that mental state M represents X just in case it is the biological function of the mechanism that regulates M to do so in such a way that M reliably indicates X. Here the biological function of a mechanism is the effect of that mechanism that it was selected for producing. Hence, so long as the mechanism regulating M was selected for doing so in such a way that M reliably indicates flies ³ See Fodor (1990). ⁴ See Dretske (1986). This kind of view has been developed in Neander (1995, 2017).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

277

(and not flies-or-BB-pellets), teleological indicator semantics will imply that M represents flies, not flies-or-BB-pellets. But there is a problem. Consider the frog’s mental state F that causes them to snap their tongues out at passing objects. This mechanism causes state F to covary with several conditions, including: C1. The presence of objects snapping at which would promote the frog’s reproductive success. C2. The presence of frog food. C3. The presence of flies. C4. The presence of small dark objects flying through the air. C5. The presence of a certain pattern of retinal stimulation. Note that these effects of the F-regulating mechanism are arranged hierarchically: this mechanism causes the F to covary with the C4 by causing F to covary with C5. And this mechanism causes F to covary with the C3 by causing F to covary with C4, etc. Similarly, this mechanism causes F to covary with C5 in order to cause F to covary with C4, and it causes F to covary with C4 in order to cause F to covary with C3, etc. Since all these effects figure in the explanation of why the F-regulating mechanism contributes to the fly’s fitness, all of these effects are functions of this mechanism.⁵ Thus, the teleological version of indicator semantics faces a problem of underdetermination. Of all the conditions that it is the function of a given mental state to reliably indicate, which of these conditions does the mental state represent? It seems natural to think that, if we want a principled, nonarbitrary answer to this question, we should look either to the top or to the bottom of the hierarchy. One answer, sometimes called “low-church teleosemantics,” looks to the bottom of the hierarchy.⁶ This view identifies the content of a mental state with whatever condition it covaries with most immediately and reliably. On this view, F represents C5. For it is in virtue of causing F to covary with C5 that the F-regulating mechanism causes F to covary with all the other conditions. Thus, if we were to put the individual organism in an environment in which C5 came apart from the other conditions, then F would continue to covary with C5 but it would cease to covary with the other conditions (C4, C3, etc.). State F covaries with these other conditions only when and because these conditions covary with C5. ⁵ See Neander (1991). ⁶ The “low-church”/“high-church” terminology is introduced in Neander (1995).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

278

 

The opposite answer, sometimes called “high-church teleosemantics,” identifies the content of a mental state with whatever condition it is its ultimate goal to covary with. On this view, the content of F would be C1. For it is in order to cause F to covary with C1 that the F-regulating mechanism causes F to covary with all the other conditions. Thus, if C1 came apart from the other conditions, then it would only be useful for F to track C1, and so there would only be evolutionary pressure for it to track C1. There is evolutionary pressure for F to track the other conditions only when and because these conditions covary with C1. Unfortunately, identifying the content of mental states with conditions at either end of this hierarchy has unacceptable implications. If we adopt lowchurch teleosemantics, and identify the content of a representation with whatever condition it covaries with most immediately and reliably, then every mental state will represent proximal rather than distal stimuli. Further, we will destroy the distinction between a representation’s being true or correct and a representation’s being rational or justified by the evidence. Moreover, we will fail to solve the problem of misrepresentation that the teleological version of indicator semantics was introduced to solve. For we will no longer be able to say that the frog is misrepresenting when it enters state F in response to a passing BB pellet, so long as the passing BB produces the same pattern of retinal stimulation as a passing fly. If, in the other hand, we adopt high-church teleosemantics, and identify the content of a representation with whatever condition it has the ultimate goal of covarying with, then we will face another problem. For, just as lowchurch teleosemantics collapses the distinction between truth and justification, high-church teleosemantics collapses the distinction between truth and utility. The ultimate goal of the mechanism regulating a given mental state is to produce this mental state under conditions where it, or the responses it produces, will be useful or promote reproductive success. Hence, highchurch teleosemantics implies that every mental state represents this condition. But this is surely a mistake, since a representation can clearly be useful without being correct or correct without being useful. Of course, there may be some mental states that represent conditions at either of the two ends of this hierarchy, but surely not all mental states do so, and so we need a theory of content that allows us to avoid these two extremes in at least some cases. Moreover, we should seek a theory that allows us to do this in a principled, well-motivated manner. I will attempt to provide such a theory in the next section.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

279

1.4 Adaptation Semantics The basic idea of the teleological theory of content I will propose is this: the cognitive content of a representation is the condition to which the behavioral pattern produced by this representation is an adaptation. To clarify this idea, let’s begin by considering an example from Sober and Wilson (1998, p. 127): Imagine a dangerous aquatic habitat with a predator lurking around every corner and a safer habitat from which predators are excluded. A fish species that lives entirely in the dangerous habitat will evolve cautious behaviors to avoid being eaten. A second fish species that lives entirely in the safe habitat will evolve to behave more boldly . . . Now imagine a third fish species, whose environment includes both a safe and a dangerous habitat . . . If individuals can detect which habitat they occupy, they should evolve to behave like the cautious species in the risky habitat and like the bold species in the safe habitat.

Let’s focus on the third fish species. This fish detects which of the two kinds of habitat it occupies and adjusts its behavior accordingly. We may assume, therefore, that there are mental states that mediate this differential behavior. Let M1 be the mental state that results in cautious behavior, and let M2 be the mental state that results in bold behavior. Thus, the fish has the following two dispositions connecting its behavior with its mental states: D1. D2.

To behave cautiously when in state M1. To behave boldly when in state M2.

How could these two dispositions have evolved? Let’s focus on the first disposition, D1. Presumably, part of what explains D1 is that it’s adaptive to be disposed to behave cautiously in dangerous environments. This fact, together with the fact that M1 covaries with dangerous environments (so that the fish is only likely to be in M1 when in a dangerous environment), will explain why it’s adaptive for the fish to behave cautiously when in state M1. And this, in turn, can explain how D1 could have evolved. Similarly, we can explain how D2 could have evolved by appealing to the adaptiveness of the disposition to behave boldly in safe environments together with the covariation between D2 and safe environments. Because these dispositions can be explained in this manner, we can say that the kind of behavior motivated by M1 is an adaptation to dangerous

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

280

 

environments, and the kind of behavior motivated by M2 is an adaptation to safe environments. And, for this reason, it seems that M1 stands in for, and hence serves to represent, dangerous environments, whereas M2 stands in for, and hence serves to represent, safe environments. Let us generalize. For any mental state M belonging to an organism O, to determine the content of M we begin with a proposition describing the complete pattern of behavioral dispositions resulting from O’s being in state M. This proposition will have the form “being in state M disposes O to behave in manner B.” Call this proposition a behavioral pattern description. One way to explain the truth of this behavioral pattern description is to appeal to (i) the historical covariation between mental state M and some condition, C together with (ii) the adaptiveness of acting in manner B under condition C. Let us call an explanation of this form a mediated adaptation explanation (I call it a “mediated adaptation explanation” because the mental state mediates the adaptation of the organism’s behavior to the environmental condition). We can now state our proposed theory of content as follows: Adaptation Semantics: A simple mental state M has cognitive content C just in case C is the condition that figures in the best mediated adaptation explanation of M’s behavioral pattern description. (As currently stated, adaptation semantics applies only to simple mental states that represent states of affairs. It is to such mental states that adaptation semantics applies in the first instance. A full account, which I will not attempt to provide here, would need to apply to complex mental states with compositional structures.) I will now argue that adaptation semantics solves the problems we have seen for the other teleological theories of content we have considered. First, unlike teleological success semantics, adaptation semantics does not imply that the deer’s mental state that causes the freeze response represents the presence of predators. For adaptation semantics would imply this only if, contrary to fact, it were adaptive for deer to be disposed to freeze in the presence of predators. Instead, the view I have proposed implies, much more plausibly, that this mental state represents something like a middling chance that predators are present. For it is adaptive for deer to be disposed to freeze when there is middling chance that predators are present. And this fact, together with the fact that the deer’s mental state covaries with conditions in which there is a middling chance that predators are present, can explain the evolution of the freezing disposition.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

281

Second, unlike teleological success semantics, adaptation semantics is able to ascribe content to mental states that motivate actions that do not promote reproductive success. For adaptation semantics does not require that the actions motivated by a mental state be adaptive. All it requires is that the behavioral dispositions associated with this mental state be adaptive. Third, adaption semantics allows us to avoid the extremes of the highand low-church versions of teleological indicator semantics, and to do so on a principled way. To see why this is so, we should first note that, even when several environmental conditions covary, it is often the case that one of these conditions provides a better explanation for a given trait than the others. Consider, for example, the fact that polar bears have thick fur. Now consider the following three explanations of this fact: Explanation 1: 1a. Historically, polar bears have lived in geographical region R (consisting of northern Canada, Russia and Greenland). 1b. It’s adaptive to have thick fur in geographical region R. Explanation 2: 1a. Historically, polar bears have lived in cold climates. 1b. It’s adaptive to have thick fur in cold climates. Explanation 3: 3a. Historically, polar bears have lived in environments where having thick fur promotes reproductive success. 3b. It’s adaptive to have thick fur in environments where having thick fur promotes reproductive success. It seems clear that, among these three explanations of why polar bears have thick fur, Explanation 2 is the best. For Explanation 1 lacks counterfactual robustness, since it appeals to highly contingent facts that are false in lots of nearby possible worlds in which polar bears have thick fur. And Explanation 3 is unilluminating, since the explanans is too close to the explanandum. After all, if we are looking for an evolutionary explanation of why polar bears have thick fur, we already know that thick fur was adaptive in their ancestral environment; we want to know what it is about this environment that makes thick fur adaptive. By contrast, Explanation 2 is reasonably robust and highly illuminating, and so it is the best explanation. Thus, the polar bear’s thick fur is best explained as an adaptation to cold temperatures, not as an adaptation to region R or to conditions in which it’s useful to have thick fur.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

282

 

Similarly, we can give several mediated adaptation explanations of why mental state F disposes frogs to engage in snapping behavior. Consider, in particular, the following three explanations: Explanation 1: 1a. Historically, state F has covaried with the presence of retinal pattern R. 1b. It’s adaptive to be disposed to engage in snapping behavior in the presence of retinal pattern R. Explanation 2: 2a. Historically, state F has covaried with the presence of flies. 2b. It’s adaptive to be disposed to engage in snapping behavior in the presence of flies. Explanation 3: 3a. Historically, state F has covaried with conditions in which snapping promotes reproductive success. 3b. It’s adaptive to be disposed to snap under conditions in which snapping promotes reproductive success. Once again, the second explanation seems clearly to be the best. For, again, the first explanation lacks counterfactual robustness, as it appeals to facts that are false in plenty of nearby worlds in which state F causes snapping. And the third explanation is unilluminating, since the explanans is too close to the explanandum. And if Explanation 2 is indeed the best explanation, then adaptation semantics will imply that F represents the presence of flies. Thus, adaptation semantics appears to have some advantages over the other teleological theories of content we have considered. Now, according to adaptation semantics, in order to determine the content of a mental state, we must first identify the pattern of behavioral dispositions associated it. Hence, in order to apply this account to moral disapproval, our next task will be to determine the pattern of behavioral dispositions associated with moral disapproval.

2. The Behavioral Consequences of Moral Disapproval In what follows, I will begin in Section 2.1 by offering a descriptive characterization of the behavioral consequences of moral disapproval. Then, in

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

283

Section 2.2, I will argue that this behavioral profile makes sense from an evolutionary point of view.

2.1 Descriptive Analysis Suppose we morally disapprove of some type of action, say, killing, lying, or incest. How will this disapproval influence our behavior? For one thing, it’s likely to give us some motivation to avoid the kind of action in question. (I say some motivation since this motivation may be outweighed by contrary motivations.) Moreover, if we disapprove of some type of action, this will give us intrinsic rather than instrumental motivation to avoid it. That is, we don’t simply avoid such actions out of self-interest, e.g., so as not to suffer punishment or reputational damage. Rather, we have some motivation to avoid such actions even when we believe they would promote our self-interest. Morally disapproving of a given type of action also affects our behavior toward others. This is so in two ways. First, if we disapprove of f-ing, we tend to treat those who f less favorably. In particular, we are less disposed to befriend them, to cooperate with them, and to promote their interests. In short, we are disposed, to a greater or lesser degree, to shun them. And again, this motivation is intrinsic. When we disapprove of f-ing, we have some motivation to shun those who f even when doing so is against our interests. The second way in which disapproving of f-ing affects our behavior toward others is that it motivates us to treat less favorably, not only those who f, but also those who don’t have the right attitude toward f-ing. Thus, if we disapprove of torture or child molestation, and we discover that someone does not engage in these practices simply because he lacks the opportunity to do so, or simply because he wants to avoid punishment, we will be disposed to treat this individual less favorably. We will likewise be put off by someone who has intrinsic motivation of the wrong kind to avoid such actions. Thus, if we morally disapprove of torture, we will not look favorably on a fin de siècle aesthete who has no moral qualms about torture but simply avoids the practice because it doesn’t agree with their aesthetic sensibilities. In general, morally disapproving of f-ing give us some intrinsic motivation to shun, or to treat less favorably, those who fail to disapprove of f-ing. Lastly, moral disapproval motivates us to engage in certain expressive behavior. Some of these expressive behaviors are voluntary. Thus, if we disapprove of f-ing, we are likely to denounce f-ing as well as those who engage in it, and to apologize when we ourselves f. However, much of the

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

284

 

expressive behavior resulting from disapproval is not under our voluntary control. Thus, when we denounce f-ing or apologize for engaging in it, our disapproval of f-ing will manifest itself in our tone of voice, demeanor, and facial expressions, where the latter often involve facial musculature that cannot be activated at will. In general, disapproving of f-ing disposes us to display this very disapproval, often in ways that are not under our voluntary control, and hence in ways that cannot easily be mimicked by those who fail to disapprove of f-ing. These seem to be the main behavioral consequences of disapproving of a given type of action. In the next section I will argue that evolutionary considerations support this account.

2.2 Evolutionary Considerations If moral disapproval plays the role I have suggested, then this will enable us to explain something that would ordinarily be difficult to explain within an evolutionary framework. It seems that, when we morally disapprove of a given type of action (such as murder, rape, theft, etc.), we are often motivated to avoid this kind of action even when performing it would promote our reproductive success or the proliferation of our genes. Or, as we may put it, disapproving of a given type of action often results in our behaving in ways that are strongly altruistic. Hence, moral disapproval presents an evolutionary puzzle, since it seems to make us act against our reproductive self-interest. There are various ways in which one might try to explain how such moral motivation could be adaptive. One might, for example, appeal to reciprocity,⁷ punishment,⁸ reputation effects,⁹ or kin selection.¹⁰ Each of these approaches attempts to reconcile morality with evolution by arguing that, properly understood, moral behaviors aren’t contrary to self-interest after all, since these behaviors serve to promote one’s interests or perpetuate one’s genes in indirect ways. For, if it makes sense for me to behave morally so that others return the favor, or so that I avoid being punished, or so that my reputation is preserved and people treat me better in the long run, then ⁷ See Trivers (1971), Alexander (1987), and Nowak and Sigmund (1998). ⁸ See Boyd and Richardson (1992), Fehr and Gächter (2002), Sripada (2005), and Boyd et al. (2010). ⁹ See Milinski et al. (2002) and Panchanathan and Boyd (2003). ¹⁰ See Grafen (1984) and Alonso (1998).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

285

behaving morally is in my interest after all and is not strongly altruistic in the sense defined above. And, similarly, if it makes sense for me to behave morally in order to promote the reproductive success of my close relatives who share my genes, then again my acting morally promotes the proliferation of my genes and so it is not strongly altruistic in the sense defined. Is there any way to give an evolutionary explanation for strongly altruistic behavior, that is, behaving in accordance with moral norms at a net cost to the proliferation of one’s genes? It seems we can, by appealing to a point we made earlier, namely, that a disposition to engage in a given type of behavior can be beneficial, even when this behavior itself is harmful. In particular, the attitude of moral disapproval may be adaptive, even when the behaviors it motivates, such as refraining from cheating, are not. Not only is this a conceptual possibility, but it is precisely what we should expect if moral disapproval involves the pattern of behavioral dispositions outlined in the previous section. To see why this is so, let f be some type of behavior that is beneficial to the agent (e.g., cheating when no one is watching). In this case, refraining from f will be against one’s interest, and so the attitude of disapproving of f will motivate behaviors that are against one’s interests. Nonetheless, on the account of disapproval I have proposed, so long as others disapprove of f, it will often be in one’s interest to do the same. For, according to this account, disapproving of f motivates expressive behaviors that cannot easily be mimicked by those who do not disapprove of f. Hence, we should expect others to be able to discern, with a reasonable degree of reliability, whether one disapproves of f. Moreover, on the account of disapproval I have proposed, disapproving of f motivates one to shun others who are known not to disapprove of f. Consequently, on the view of moral disapproval I have proposed, if others disapprove of f, then if one fails to disapprove of f then there is a reasonable chance that others will (i) recognize this fact and accordingly (ii) shun one for failing to disapprove of f. Hence, disapproving of f may be beneficial despite the fact that refraining from f is harmful. Thus, the account of moral disapproval I have sketched, in addition to being independently plausible, also allows us to explain the existence of strongly altruistic behavior within an evolutionary framework.

3. The Cognitive Content of Moral Disapproval In what follows, I will apply the adaptation semantics sketched in Section 1.4 to the attitude of moral disapproval. Recall that, according to adaptation

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

286

 

semantics, in order for X to be the cognitive content of moral disapproval, X must play the right role in the best mediated adaptation explanation of the behavioral dispositions associated with moral disapproval. Consequently, the following must be true: (i) The behavioral dispositions associated with moral disapproval must be adaptive in relation to X. (ii) Moral disapproval must have historically covaried with X. My goal, in what follows, will be to identify such an X. I will begin, in Section 3.1, by characterizing a feature of actions such that it is adaptive to be disposed to behave disapprovingly toward actions with this feature. Then, in Section 3.2, I will argue that it is plausible that moral disapproval covaries with this very feature. Hence, I will argue that this feature is a plausible candidate for being the cognitive content of moral disapproval. I conclude, in Section 3.3, with a brief discussion of the normative picture that emerges from this account of the content of moral disapproval.

3.1 Toward What Kinds of Action is it Adaptive for Us to Behave Disapprovingly? Recall that, in Section 2, we identified four kinds of behavioral disposition that appear to be associated with moral disapproval: (i) the disposition to avoid the actions we disapprove of, (ii) the disposition to shun those who engage in such actions, (iii) the disposition to shun those who fail to disapprove of such actions, and (iv) the disposition to display this very disapproval in ways that are hard to mimic. Let us say that someone behaves disapprovingly toward a given type of action just in case they manifest these four dispositions toward it. Our present task is to identify a feature of actions in relation to which this pattern of dispositions is adaptive and that could plausibly figure in the deepest explanation of these very dispositions. I will argue that the feature in question is a complex one since, for any action type f, there are several factors that are typically relevant to whether it is adaptive to be disposed to behave disapprovingly toward f-ing. Hence, it will typically be adaptive, all things considered, to be disposed to behave disapprovingly toward f-ing just in case f-ing has the right combination of these factors. In order to identify these relevant factors, I will begin by considering the costs and benefits of behaving disapprovingly in relation

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

287

to a given type of action, and I will then consider the factors that contribute to these costs and benefits. For any action type f, the main evolutionary benefit of the disposition to behave disapprovingly toward f-ing is that this disposition will enable one to avoid being shunned by others who disapprove of f-ing. Let A(f) represent the magnitude of this benefit, which we may call the shunning-avoidance value. The main evolutionary cost of the disposition to behave disapprovingly toward f-ing is that this disposition is likely to prevent one from acting in the most instrumentally rational manner. In particular, it is likely to prevent one from f-ing even when f-ing would promote one’s interests, and it is also likely to motivate one to shun others who f, or who fail to disapprove of f-ing, even when such shunning is not in one’s interest. Let B(f) represent the magnitude of this cost, which we may call the behavioral cost. The third, and generally least significant, respect in which it can be beneficial or harmful to disapprove of f-ing is that doing so may influence others to disapprove of f-ing, and this in turn may influence their behaviors in ways that may be beneficial or harmful to oneself. Let P(f) represent the magnitude of this benefit or cost, which we may call the persuasion value (if it is a cost, the value of P(f) will be negative). Thus, overall, it seems the disposition to behave disapprovingly in relation to f will be adaptive just in case the following sum is positive: AðfÞ  BðfÞ þ CðfÞ We must now determine what factors are relevant to each of the three terms, for a given person S in relation to a given type of action f. Beginning with the first term, the shunning-avoidance value, the two following factors appear to be highly relevant: F1. The commonality of disapproval of f-ing among those in S’s society, or with whom S interacts. The more widely f-ing is disapproved of in S’s society, the more people there are who are likely to shun S if S fails to disapprove of f-ing, and so the more beneficial it will be for S to be disposed to behave disapprovingly in relation to f-ing. F2. The strength of the disapproval of f-ing among those who disapprove of it. The more strongly they disapprove of f-ing, and hence the more they are disposed to shun those who fail to disapprove of f-ing, the more beneficial it will be to be disposed to behave disapprovingly in relation to f-ing.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

288

 

Turning now to the second term, behavioral costs, the three most relevant factors would appear to be the following: F3. The beneficiality of the practice of f-ing. The more beneficial it is for S to be able to f when f-ing is in S’s interest, the costlier it will be for S to avoid f-ing, and so the costlier will be the disposition to behave disapprovingly in relation to f-ing. F4. The commonality of f-ing among those with whom S interacts. For the more people one encounters who f, the more frequently one is likely to end up engaging in costly shunning behavior if one is disposed to behave disapprovingly toward f-ing. F5. The commonality of approval of f-ing among those with whom S interacts. For, again, the more people one encounters who fail to disapprove of f-ing, the more frequently one is likely to end up engaging in costly shunning behavior if one is disposed to behave disapprovingly toward f-ing. Lastly, in relation to the third factor, persuasion value, the two most relevant factors appear to be the following: F6. The harmfulness (or beneficiality) to S of others’ f-ing. The more harmful it is to S when others f, the more S stands to benefit from others refraining from f-ing, and hence the more S stands to benefit from others disapproving of f-ing. Hence, insofar as S’s behaving disapprovingly in relation to f-ing may motivate others to disapprove of f-ing, the more harmful it is to S for others to f, the more beneficial it will be for S to be disposed to behave disapprovingly in relation to f-ing. Conversely, the more beneficial it is to S for others to f, the more harmful it will be for S to have this disposition. F7. The acceptability of the disapproval of f-ing from other’s perspective. If it makes no sense from other’s perspective to disapprove of f-ing (e.g., because refraining from f-ing would be extremely costly for them), then the mere fact that S disapproves of f-ing is unlikely to motivate others to disapprove of f-ing. Hence, the more sense it makes, among those in S’s society who do not yet disapprove of f-ing, to come to disapprove of f-ing, the more likely it is that S’s behaving disapprovingly toward f-ing will affect their attitudes, and so the greater will be the benefits or harms resulting from such influence.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

289

While I don’t claim that this list is exhaustive, it may serve as a first approximation of the set of factors relevant to the adaptiveness of behaving disapprovingly toward f-ing. Thus, the disposition to behave disapprovingly toward f-ing will generally be adaptive whenever f-ing has the right combination of these properties. Let X represent the complex property consisting in having the right combination of the features just outlined (a full account would need to specify the nature of this combination, but I won’t attempt to do so here). If X covaries with moral disapproval, then X can figure in the kind of explanation we are looking for, and so X will be a plausible candidate for the content of moral disapproval. And so our next task is to determine whether it is plausible that moral disapproval covaries with X.

3.2 The Covariation between this Property and Moral Disapproval Since the precise nature of X has not yet been specified, we are not yet in a position to show that moral disapproval covaries with X. We can, however, argue that it covaries, in the manner we would expect, with the various factors constitutive of X. I will support this claim with the following seven observations, one for each of the factors constitutive of X. F1. It is undeniable that we tend to disapprove of what others in our society disapprove of. A major predictor of whether someone disapproves of, e.g., cousin marriage is whether others in her society disapprove of this practice. F2. We are especially likely to disapprove of what others in our society strongly disapprove of. Thus, we are more likely to disapprove of cousin marriage if others in our society regard it as an abomination than if others regard it as a trivial offence. F3. If avoiding some kind of action would be extremely costly for us, then we are unlikely to disapprove of this kind of action. It has been observed that “there is a natural feeling of distrust and even disapproval of wealth, especially on the part of those who have never possessed it” (Higginson 1897). Once we come into wealth, avoiding wealth would be very costly for us, and so we are less likely to disapprove of it.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

290

 

F4. The more common it is for people in our society to engage in a kind of action, the less likely we are to disapprove of it. Thus, if everyone we know smokes cigarettes, but no one we know smokes hashish, then we are more likely to disapprove of the latter than of the former. F5. The more people we know who approve of some kind of action, the more likely we are to approve of it. This is simply the flip side of our first observation. F6. The more we stand to gain from others’ avoiding a given type of action, the more likely we are to disapprove of it, and the more we stand to lose from others’ avoiding a given type of action, the less likely we are to disapprove of it. Thus, music producers are more likely than others to disapprove of the practice of copying and sharing music, since music producers would benefit from the avoidance of this activity. And cattle farmers are less likely than others to disapprove of meat consumption, since they would be harmed by the avoidance of this activity.¹¹ F7. We are more likely to disapprove of f-ing if we can defend this disapproval in discussions with others in ways that make sense to others. Hence, we are more likely to disapprove of f-ing when this attitude would make sense from others’ points of view. While far from demonstrative, these observations lend some support to the claim that the attitude of moral disapproval covaries with a complex property X consisting in the right combination of the seven factors outlined in the previous section. Moreover, since the behavioral dispositions associated with moral disapproval appear to be adaptive in relation to this property, such a property can figure in a mediated adaptation explanation for these behavioral dispositions. Hence, such a property is a plausible candidate for the cognitive content of moral disapproval.

3.3 Normative Implications Suppose we’re on the right track, and moral disapproval represents the kind of complex property we’ve been describing. If this is right, what are the normative implications?

¹¹ These phenomena are discussed in DeScioli et al. (2014).

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

291

Perhaps the most obvious implication is that we must reject all nonrelativistic normative ethical theories. For, according to any such theory, actions are either right or wrong simpliciter. On such a view, if an action is wrong simpliciter, then it would be fitting for anyone to disapprove of it, whereas if an action is right simpliciter, then there is no one who could fittingly disapprove of it. But the account I have offered implies otherwise. On this account, whether it is fitting for S to disapprove of f-ing depends in part on features of S, such as how commonly f-ing is disapproved of in S’s society and how costly it would be for S to refrain from f-ing. Hence, on this view, whether it is fitting to disapprove of f-ing can vary from person to person. This view, however, avoids some of the problems facing standard forms of relativism. For many forms of relativism imply that a sufficient consensus is infallible, so that, necessarily, if everyone disapproves of f-ing then it is fitting to do so. Hence, many forms of relativism imply that we could conclusively determine that f-ing merits disapproval simply by conducting a poll. But the view under consideration avoids this implication. On this view, since the prevalence of disapproval of f-ing is but one of the many factors relevant to whether such disapproval is fitting, everyone could be wrong in disapproving of f-ing. Does this view also have relativistic implications about moral wrongness claims? Likely yes. For, given the close connection between the fittingness of disapproving of f-ing and the correctness of asserting that f-ing is wrong, it seems that if we adopt a relativistic account of the former we should also adopt a relativistic account of the latter. The details of such an account, however, would require further investigation. No doubt many readers will find the relativistic implications of the view I have proposed to be highly counterintuitive. However, as I will now argue, there are two reasons to welcome these relativistic implications. The first such reason is that, in virtue of its relativism, the view I have proposed enables us to avoid skeptical worries. If we held a non-relativistic view according to which actions are right or wrong simpliciter and so everyone should disapprove of the very same actions, then it seems that reflecting on how our moral beliefs are formed should make us doubt their accuracy. We know from observation and from empirical investigation that our attitudes of moral disapproval, and our corresponding moral beliefs, are responsive in large measure to the kinds of factors discussed in the previous section. We are much more likely to disapprove of actions that others in our society disapprove of; and, in forming our moral attitudes, we are strongly influenced by our interests and our position within society. On a non-relativist

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

292

 

view, it seems we should regard all of these as off-track influences that are not generally conducive to truth. Hence, it seems we should regard any attitudes formed in such a manner as unlikely to be correct. By contrast, on the view I have proposed, the factors that guide our moral attitudes are precisely the ones that typically make these attitudes adaptive, and so they are the very factors that make these attitudes fitting. A second reason for welcoming the relativistic implications of the view I have proposed is that it allows for a more unified account of our affective attitudes and their fittingness conditions. For, if we consider our various affective attitudes, we find that, in general, their fittingness conditions appear to relativistic. For example, whether it is fitting for S to fear X depends on whether X is dangerous to S. Whether it is fitting for S to hope for some occurrence, or for S to be pleased or displeased by some occurrence, depends in part on the bearing of this occurrence on S’s interests. Whether it is fitting for S1 to be grateful toward, or angry at, S2 depends in part on what S2 has done to S1. Whether it is fitting for S1 to envy or pity S2 depends in part on how well S2 fares in comparison to S1. And since admiring someone involves looking up to them whereas despising someone involves looking down on them, the fittingness of S1’s admiring or despising S2 likewise depends in part on how S2 compares to S1. The account of content, and hence of fittingness, that I have offered allows us to explain these fittingness conditions, and it explains why all of them are agent-relative. For, on the view I have proposed, the content of an attitude will be some condition in relation to which the pattern of behavioral dispositions associated with this attitude is adaptive. And, in general, whether it is adaptive for us to be disposed to behave in a given manner toward an object depends on how this object is related to us. Whether it’s adaptive, for example, to be disposed to behave fearfully in relation to X (e.g., by fleeing or avoiding X) depends on whether X is dangerous to S—and so on for the other kinds of behaviors typically motivated by affective attitudes. Hence, this account predicts that, in general, affective attitudes will have agent-relative fittingness conditions. On the very same grounds, we should expect moral disapproval to have agent-relative fittingness conditions. For whether it is adaptive for S to behave disapprovingly toward f-ing depends, in the kinds of ways I’ve described, on how f-ing is related to S. Thus, we should not be surprised to find that moral disapproval has relativistic fittingness conditions. To the contrary, we should be surprised if the orthodox, non-relativist view were true. For, in this case, moral disapproval would be an anomaly among affective attitudes.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

    

293

References Alexander, Richard D. (1987). The Biology of Moral Systems. Hawthorne, NY: Aldine de Gruyter. Alonso, W. J. (1998). “The Role of Kin Selection Theory on the Explanation of Biological Altruism: A Critical Review.” Journal of Comparative Biology 3(1): 1–14. Boyd, Rob, Herbert Gintis, and Samuel Bowles (2010). “Coordinated Punishment of Defectors Sustains Cooperation and Can Proliferate When Rare.” Science 328(5978): 617–20. Boyd, Robert and Peter J. Richardson (1992). “Punishment Allows the Evolution of Cooperation (or Anything Else) in Sizable Groups.” Ethology and Sociobiology 13(3): 171–95. DeScioli, Peter, et al. (2014). “Equity or Equality: Moral Judgments Follow the Money.” Proceedings of the Royal Society B 281(1797): 2014112. Dretske, Fred (1986). “Misrepresentation,” in Radu Bogdan (ed.), Belief: Form, Content and Function, New York: Oxford University Press, 17–36. Fehr, Ernst and Simon Gächter (2002). “Altruistic Punishment in Humans.” Nature 415(10): 137–40. Fodor, Jerry (1990). A Theory of Content and Other Essays, Cambridge, MA: MIT Press. Frank, Robert (1988). Passions within Reason: The Strategic Role of the Emotions, New York: Norton. Grafen, Alan (1984). “Natural Selection, Kin Selection and Group Selection,” in J. R. Krebs and N. B. Davies (eds), Behavioural Ecology: An Evolutionary Approach, Sunderland, MA: Sinauer Associates, 62–84. Greenspan, Patricia (2000). “Emotional Strategies and Rationality.” Ethics 110(3): 469–87. Harms, W. F. (2000). “Adaptation and Moral Realism.” Biology and Philosophy 15(5): 699–712. Higginson, Thomas (1897). Book and Heart: Essays on Literature and Life, New York: Harper & Brothers. Joyce, Richard (2001). “Moral Realism and Teleosemantics.” Biology and Philosophy 16(5): 723–31. Millikan, Ruth (1984). Language, Thought and Other Biological Categories, Cambridge, MA: MIT Press. Millikan, Ruth (1989). “Biosemantics.” Journal of Philosophy 86(6): 281–97.

OUP CORRECTED PROOF – FINAL, 16/9/2019, SPi

294

 

Millikan, Ruth (1990). “Truth, Rules, Hoverflies and the Kripke-Wittgenstein Paradox.” Philosophical Review 99(3): 232–53. Millikan, Ruth (2009). “Biosemantics,” in Brian P. McLaughlin (ed.), The Oxford Handbook of Philosophy of Mind, Oxford: Clarendon Press, 394–406. Milinski, M., D. Semman, and H. J. Krambeck (2002). “Reputation Helps Solves the ‘Tragedy of the Commons.’ ” Nature 415 (January): 424–6. Neander, Karen (1991). “Functions as Selected Effects.” Philosophy of Science 58(2): 168–84. Neander, Karen (1995). “Malfunctioning and Misrepresenting.” Philosophical Studies, 79(2): 109–41. Neander, Karen (2017). A Mark of the Mental: In Defense of Informational Teleosemantics, Cambridge, MA: MIT Press. Nowak, Martin A. and Karl Sigmund (1998). “Evolution of Indirect Reciprocity by Image Scoring.” Nature 393 (June): 573–7. Panchanathan, Karthik and Rob Boyd (2003). “A Tale of Two Defectors: The Importance of Standing for Evolution of Indirect Reciprocity.” Journal of Theoretical Biology 224(1): 115–26. Papineau, David (1984). “Representation and Explanation.” Philosophy of Science 51(4): 550–72. Papineau, David (1993). Philosophical Naturalism, Oxford: Blackwell. Sinclair, Neil (2012). “Metaethics, Teleosemantics and the Function of Moral Judgements.” Biology and Philosophy 27(5): 639–62. Sripada, Chandra S. (2005). “Punishment and the Strategic Structure of Moral Systems.” Biology and Philosophy 20(4): 767–89. Sober, Elliott and Wilson, David Sloan (1998). Unto Others: The Evolution and Psychology of Unselfish Behavior, Cambridge, MA: Harvard University Press. Trivers, Robert L. (1971). “The Evolution of Reciprocal Altruism.” Quarterly Review of Biology 46(1): 35–57. Whyte, J. T. (1990). “Success Semantics.” Analysis 50(3): 149–57. Whyte, J. T. (1991). “The Normal Rewards of Success.” Analysis 51(2): 65–73.

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

Index Abramson, Kate 116 n. action agent-sacrificing 237 and DDE 139 and discretion 50, 50 n., 52–53, 58–63, 69–71 and duty of mutual aid 64 n., 66 n. and endorsement 140–141, 151 and evaluation vs. guidance 207 and forgiveness 88–89 and Frankfurt’s hierarchical model of desire 177–178 and love 122 n., 128 n. and moral worth 183–186, 186 n., 187, 187 n., 188–199, 199 n., 200, 200 n., 201 and motives 10, 14, 16, 18–19, 20 n., 21, 25–28 and permissibility 208–210, 216, 216 n., 217, 220–221, 228–230, 232–233, 238–239, 245–246 and principle of utility 38 and responsibility 145 n. and waiver of rights 56 belief-desire model of 20 n. beneficent 224–225 moral disapproval of 271, 283–284, 286–291 adaptation semantics 279–282, 285 agency and attitudes of moral disapproval 285, 292 and DDE 146 and discretion 50, 50 n., 51–51, 55–56, 58–63, 68–69, 70 n., 71–72 and duties of mutual aid 70, 72 and endorsement 141–142, 144–146, 154–155, 158 and gender 157, 157 n., 158 and moral praiseworthiness 184–185, 187, 190, 192, 194, 196, 199 n., 200 n., 201

and motive 21–23 and participant stance 98, 99, 103, 107, 109–110 and taking oneself lightly 179 and testimony 47 n. and unity of the self 161, 177 agent-neutrality and aims 67 n. and love 125 n. and prerogatives 228–229, 229 n., 230–231, 233–237, 237 n., 238–239, 241, 243–244, 246 and reasons 218, 230 and subjective permissibility 206, 206 n. and values 125, 126 n. agent-relativity and fittingness conditions 292 and love 126, 126 n., 127 and prerogatives 224, 225 n., 228, 229 n., 230–231, 233–236, 236 n., 237–239, 241, 244 and reasons 206, 206 n., 218, 230 and values 126–127 akrasia 23 Alexander, Richard D. 284 n. alienation 161, 177–178 Allais, Lucy 78 n., 88 n. Alonso, W. 284 n. altruism 15, 69, 71, 71 n., 224 n. amusement 164–166, 166 n., 167–169, 169 n., 170–171, 171 n., 172–175, 178–179, 181 Anscombe, G.E.M. 30, 37, 41, 46, 146 n. apt/appropriate anger 78, 162, 179–180, 180 n., 181 Aristotle 114, 168 n., 180, 180 n. Arpaly, Nomy 185–186 Austin, J.L. 29, 37, 43 authority and forgiveness 77, 89 and promising 15

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

296



authority (cont.) service conception of 40 n. to waive entitlement 64 autonomy 52, 72, 112–113, 140 autonomous life 51 n. backwards induction 241, 245 bad faith 37, 38, 179 n. Bader, Ralf M. 228, 228 n., 229, 244 n. Bell, Macalester 78 n. beneficence 21, 79, 131 suboptimal 223–224, 226, 228, 235, 239, 241, 243 Bennett, Christopher 73 n. Bennett, Jonathan 146 n. Bero, Steve 98 n. Bindel, Julie 156–157 Blackburn, Simon 197–198 blame 33–35, 53, 67, 75 n., 76 n., 79 n., 82, 82 n., 86, 134, 139, 143–145, 171–172 Boyd, Robert 284 n. Bradley, Ben 212 n., 253 n. Brennan, Geoffrey 143 n. Brennan, Jason 155 n. Brentano, Franz 119 n. Broome, John 134 n. Brown, Robert 117 Calhoun, Cheshire 171 n. Carroll, Noël 165, 165 n., 166 n., 167 n. categorical imperative 189, 197 choice and autonomy 112–113 and discretionary moral duties 51 and request 45 constrained 140, 144, 151 n., 154–159 dynamic consistency 224–225, 225 n., 226, 229 n., 240, 240 n., 241, 241 n., 242–245, 247 global 225 n., 236 n., 243, 244, 244 n., 245, 247–249 myopic principles 241, 243, 244 n., 245, 247–249 optimizing 229 n., 230, 234–235, 239, 244 separability 224–225, 225 n., 226, 232, 241, 243–244 sequential vs. non-sequential 224–226, 231–232, 241

sets 240–241, 244, 250–251, 251 n., 254, 254 n., 257–259, 259 n. sophisticated principles 241, 245 Clausen, Ginger 116 n., 120 n., 121 n., 122 n., 128 n. cognitivism 185, 191, 191 n., 192, 194, 196, 199–201 Cohen, Jonathan 68 n. collective action 63 Collins, Patricia Hill 251 n. Colyvan, Mark 210 n. conflation problem 120, 120 n. Connell, F.J. 151, 154 consent 30, 56, 100, 150 n. consequentialism 50, 50 n., 154 n., 206, 221 contractualism 64 counterfactuals 76, 150, 150 n., 236 n., 246–249, 281–282 Cox, Damian 210 n. Cubitt, R. 224 n., 225 n. Cullity, Garrett 63 n., 65 n. Cupit, Geoffrey 32 n., 47 n. D’Arms, Justin 119 n., 120, 120 n. Damasio, Anthony R. 251 n. Dancy, Jonathan 126 n., 192 Danielsson, Sven 121 Darwall, Stephen 30 n., 33 n., 48 n., 176, 176 n. decision theory 205, 206 n., 210, 210 n., 220–222 decision tree 225, 232, 244, 247–248 DeGrazia, David 260 n., 261 n., 266 n. deliberation 16, 24 n., 27, 52, 54, 58, 154, 245 dependence thesis 40 n. deontic status 215 value 210, 221 deontology 205–206, 208, 210, 220–222 DeScioli, Peter 290 n. desire and amusement 166 n., 170 n. and autonomy 112 and forgiveness 75 and love 116–117, 119 and motive 18, 18 n., 19, 19 n., 20, 20 n., 23 and prudence 252–253 and radical change 252–253, 256 and right reasons 185–188, 192–195, 197–199, 199 n.

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

 and semantics 273 and welfare 272 as a system 22 hierarchical models of 177, 261 diachronicity 224, 245–246, 249 dignity 132–133, 161–162, 177 discounting 214, 218, 227, 230–231 discretion constraint 30, 44–45 dispositions 18 n., 19 n., 164, 174, 176, 259, 275–276, 279–282, 285–290 doctrine of double effect 139, 145, 146, 151, 154, 156 Dorsey, Dale 253 n. Dreier, Jamie 191 n., 202, 237 n. Dretske, Fred 276, 276 n. duty/duties (see also obligations) accessoric 13 and agent responsibility 26 and amusement 173, 175–176 and capacity 18 and guidance 15 and moral worth 184, 186, 189, 193, 195, 200 and motive 14, 17–18, 24–25 and supererogation 216 n. and system view of motives 19 co-instance of 12, 12 n., 13 conflict of/between 12 n., 13 content of 13, 15, 19 directed 52–57, 57 n., 58, 63 discretionary 50, 50 n., 51, 52, 54–57, 57 n., 58–60, 62–63, 65, 67–69, 70 n., 71, 71 n., 72 imperfect 10, 28, 50, 50 n., 51 negative 10, 14, 51 of due care or non-negligence 10–13, 13 n., 14–18, 18 n., 24–26 of gratitude 51–52, 55–56, 56 n., 57, 57 n., 58–59, 60–63, 70–71 of mutual aid 51–52, 63–67, 67 n., 68–70, 70 n., 72 perfect 14, 50 positive 10, 51 primary 13, 15–16, 18 n., 25 promissory 15 secondary 13, 13 n., 17 to have a feeling or motive 17 to rescue 70, 75–76, 78, 222, 224 value of 12, 15–17, 22, 25 Dworkin, Ronald 108 n.

297

Eastman, Max 167, 167 n. effective altruism 69, 70–71, 71 n., 224 n. empathy 82–83 endorsement (see also permissibility: and endorsement and reasons: for endorsement) all-things-considered 142–143, 153, 157 n. and constrained choice 140, 144, 151 n., 154–159 and ignorance 145, 145 n. and proportionality 140, 151, 151 n., 152–154, 154 n., 156–157, 159 and seprability 140, 144, 146–149, 150 n., 151, 151 n., 156, 158–159 doctrine of double endorsement 140, 140 n., 142–146, 148, 150, 150 n., 151, 151 n., 152–154, 156–159 Enoch, David 32 n. entitlement and close relationships (see also obligations: of close relationships), 84–85, 92 and debts of gratitude 56 n., 60 and directed duties 58 and duties of mutual aid 64 and requests (see also requests; obligations: and requests; reasons: and requests) 43 equality 50 n., 52, 54, 61, 71–72, 133, 148 n., 162, 209, 233, 236, 236 n. Ewing, A.C. 119 n. expressivism 191 n., 196–199, 199 n., 200, 200 n., 201–202 Fehr, Ernst 284 n. Feldman, Fred 206 n. FitzPatrick, William 139 Fletcher, Guy 195, 253 n. Fodor, Jerry 274 n. Foot, Philippa 38 n., 146 n. forgiveness (see also obligation: and thirdparty forgiveness) and personal wrongs 73–75, 83–86, 88, 88 n., 91 attitude-based accounts of 74, 88, 91–92 debt-release models of 90 ground of standing to 74, 77 n., 80–83, 83 n., 87, 87 n., 91–92 rights-relinquishment models of 90–91 third-party 13, 73, 73 n., 74–84, 86–92

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

298



forgiveness (cont.) victim-exclusive accounts of 74, 77, 81 n., 82 victim-relative accounts of 73–74, 82, 86 n. Frank, Robert 285, 285 n. Frankfurt, Harry G. 117, 119 n., 177, 177 n., 264, 264 n., 265, 265 n. Gächter, Simon 284 n. Geach, Peter 108 n. Gergen, Kenneth 179 n. Gibbard, Allan 197 n. Ginet, Carl 145 n. Glasgow, Joshua 253 n. good will and amusement 174 and requests 46, 47 n. Govier, Trudy 73 n. Grafen, Alan 284 n. Graham, Peter 136 n., 209, 209 n., 210, 210 n., 216, 218 n., 219 n. gratitude and agency 56 n., 60, 61–62, 97–99, 103, 112 duties of 51–52, 55–56, 56 n., 57, 57 n., 58–59, 60–63, 70–71 entitlement to and debts of 60 grounding of 59 n. promises and 59 Greengross, Gil 169 n. Greenspan, Patricia 275 n. Griggs, R. 251 n. Griswold, Charles 73 n., 79 n. Guerrero, Alexander 205 n. guilt 33, 37, 97 n., 170 Haber, Joram Graf 73 n. Hampton, Jean 73 n., 78, 78 n., 88 n. harmony of ends 42–46 Harms, W.F. 272 n. Havel, Vaclav 182, 182 n. Hawkins, Jennifer 253 n. hedonism 253, 255–256, 258–269 Herman, Barbara 61, 61 n. Hieronymi, Pamela 88 n. Higginson, Thomas 289 Hills, Alison 195 Hobbes, Thomas 164–165, 165 n. Hohfeld, W.F. 53–54 Holton, Richard 98 n.

Horgan, Terence 191 n. Horton, Joe 235 n., 241 n. Howard, Christopher 119 n., 121 Hume, David 18 n., 19 n. Hurka, Thomas 118 n., 224 n., 229 n., 235 n., 237 n. Hurley, Paul 221, 221 n. Hvizdala, Karel 182 n. identity 16, 228, 254, 260, 260 n., 261–262, 262 n., 263–265, 267 implicit bias 23 incommensurability of values 50 n., 229, 240–241 indignation 33–34, 75 n., 79, 79 n., 82, 87, 97 n. insults and agency 95, 111 and anger 180 and gratitude 59 interpersonal accountability 52, 54 interpretive account of persons 108–112, 114 irrationality and love 116 and negligence 12 n. of behavior 275 Isserow, Jessica 184 n. Jackson, Frank 219, 219 n., 225 n. Jacobson, Daniel 119 n., 120, 120 n. Jaworska, Agnieszka 268 n. Johnson King, Zoë 187, 189, 189 n., 190, 192, 195 Jollimore, Troy 116 n., 117 n. Joyce, Richard 272 n. Julius, A.J. 40 n., 47 n. justice and agency 17 and oppression 145, 162, 179–181 Kagan, Shelly 124, 206 n., 213 n., 224 Kamm, Frances 206 n., 217 n. Kant, Immanuel 13, 13 n., 18 n., 22 n., 42, 44–45, 59 n., 95–96, 99–100, 102, 104, 107, 113, 132–133, 161, 172–173, 173 n., 174–176, 176 n., 177 Kolodny, Niko 118 n., 119 n., 123, 123 n., 124–125, 127, 127 n., 128–130 Korsgaard, Christine 95 n., 99–110, 264 n. Kukla, Rebecca 32 n.

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

 Lance, Mark 32 n. Langton, Rae 99, 100, 105 Lazar, Seth 205 n., 210 n., 211 n. Leite, Adam 116 n. Lenman, Jimmy 197 Little, Margaret Olivia 32 n. Locke, John 112 Lomasky, Loren 143 n. Lord, Errol 134 n. love and belief 129 and fit 116–118, 118 n., 119, 119 n., 120, 120 n., 121–122, 122 n., 123, 123 n., 124–125, 126 n., 127–128, 129 n., 130–133, 133 n., 134 and forgiveness 80–81, 85, 91 and grounding 118, 124 and ignorance 134–135 and problems of inconstancy 117, 119–121 and problems of promiscuity 117, 119–121 and problems of trading up 117, 119–121, 130–134 and problems of universality 117, 119–121 and reasonable harmonies 42 and value-related reasons 122, 122 n., 123, 128, 128 n., 130, 132 quality view 116, 116 n., 117–120, 120 n., 121, 122 n., 123, 127 127 n., 130–132 relationship view 118 n., 123, 123 n., 124–127, 127 n. luck and duties to rescue 60 and negligence 11 and risk 210 Luper, Steven 258 n. MacAskill, William 69 n. MacLachlan, Alice 73 n., 80 n., 81–82, 82 n., 86, 86 n. Markovits, Julia 185–187 Mason, Elinor 206 n. Mason, Michelle 176 n. McGrath, Sarah 195 McHugh, Connor 119 n. McMahan, Jeff 236, 258 n., 259 n., 261 n. Medina, José 145 n. Mele, Alfred 145 n. mental states 141, 272–282 metaethics 171, 184–185, 197, 201, 272 n. Milam, Per Erik 73 n.

299

Milinski, M. 284 n. Mill, John Stuart 112 Miller, Geoffrey F. 169 n. Millikan, Ruth 273–274 Mills, Charles 145 n. Milne, Peter 208 n., 210 n. moral agency 17 moral concern 26, 185 moral deference 195 moral disapproval 271–272, 272 n., 276, 282–284, 286, 289–292 evolutionary explanations of 284–285, 287 fittingness of 271–272, 291–292 moral judgment 194, 196, 199 moral motivation 284 moral norms 166, 285 moral relativism 272, 291–292 moral responsibility 52, 139, 143, 145 n. moral standing 56, 180 moral worth 184, 184 n., 185–187, 189–190, 192–196, 198, 199 n., 200, 200 n., 201–202 and right reasons view 184–185, 187, 189–190, 192, 194, 196, 200 grounds for 198–199 moralistic fallacy 120 n. Moran, Richard 40 n., 47 n. motive/motivation and belief 17, 185, 188, 190–191, 191 n., 192–193, 195–199, 199 n., 200–201, 273 and duty of due care or nonnegligence 14–15, 17–18, 25–26 and impermissibility 9, 10, 27 and instinct 18, 18 n., 20 and moral agency 17–21, 24 and moral disapproval 283–285, 287–288, 292 and negative duties 10 and negligence 14 and positive duties 10 and positive law 9–10, 17 and psychology 14 n., 17 n., 20, 22 n. and value 18, 21, 24–25, 27–28 and wrongness 9–11, 14–28 as a system state 18, 18 n., 19–20, 20 n., 21–22, 22 n., 23–24, 24 n., 26–27 de dicto 188, 192 de re 186, 188 difference between being motivated and being moved 23–24

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

300



motive/motivation (cont.) mental states and 275, 279–281, 283–285, 287–289, 292 moral 18 n., 19, 21, 22, 22n., 24, 26, 189–190, 193–195, 194–197, 198–199, 199 n., 200, 200 n., 201 motivating reasons vs. normative reasons 257 of duty 184, 189, 195, 200 rational 21, 24 Murphy, Jeffrie 73 n., 78, 79 n., 88 n. Murphy, Liam 63 n., 67 n. Naar, Hichem 119 n. natural selection 271, 272 n. naturalism 20–21, 271–272 Neander, Karen 276 n., 277 n. needs and agency 112 and gratitude 61 and mutual aid 64–68, 68 n., 69–70, 71 n. and requests 43 and value sensitivity 24, 24 n. negligence (see also duty of due care or non-negligence) 11–12, 12 n., 15–17, 25 and motive 14 and strict liability 15 n. distinguished from accident 13 duty of non-negligence 10, 10 n. moral fault of 13 n. Nelkin, Dana Kay 90, 146 Norlock, Kathryn J. 73 n. normative ethics 184, 271, 272 n., 291 Nowak, Martin A. 284 n. Nozick, Robert 117 objective stance 99–100, 102–103, 105 obligations (see also duties) and beneficience 235–236, 241 n. and protected reasons 32 and reactive attitudes 33–35 and requests (see also requests; entitlement: 1and requests; reasons: and requests) 32, 32 n., 35–39, 41, 45, 47 n. and third-party forgiveness (see also forgiveness) 83, 83 n., 84, 84 n., 85, 87, 90–92 conditional 69, 235–236, 241 n. moral 50, 50 n., 51–52, 54–55, 63, 65–67, 70

of close relationships (see also entitlement: and close relationships) 83, 83 n., 84, 84 n., 85, 87, 92, 206 of emergency assistance 66, 66 n., 67–68, 68 n., 69–70 prima facie 44 pro tanto 44 relational view of 52–56, 63, 65, 67, 67 n., 72 when choosing between actions 206, 211–212 Oddie, Graham 218 n., 229 n. Olsen, Kristian 209, 209 n., 225, 226, 229 n. Olson, Eric T. 261 n., 262 Olson, Jonas 121 oppression 147–148, 152, 161–162, 172, 179–180, 180 n., 181 options and minimizing wrongness 211 and permissibility 220, 222, 225, 230–231, 234, 241–242, 244, 247–248 and radical change 260, 264, 270 as constrained choice 154–155 Owens, David 32 n., 73 n. pain and amusement 167, 170, 181 and love 133 and motive 21–22 and radical change 256, 263 Panchanathan, Karthik 284 n. Papineau, David 273 Parfit, Derek 27, 191, n., 205 n., 223–224, 263 n. participant stance 97, 97 n., 98–105, 107, 112–113 performatives 29–30 permissibility and beneficence 223–226, 228–229, 229 n., 230–233, 233 n., 234–237, 237 n., 238–243, 245–246 and DDE 139–140 and DDN 142–143, 146, 148–150, 150 n., 151, 153–159 and discretionary moral duties 69 and dominance principle 228–229, 229 n., 230–232, 232 n., 233–234, 236, 238, 238 n., 241–243 and effective altruism 69

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

 and endorsement 142–143, 146, 148, 149 n., 150, 150 n., 151, 153–154, 154 n., 155–156, 157 n., 158, 159 and ignorance 205 and interpreting behaviour 113 and love 133 and moral worth 184, 190 and motive 9–10, 10 n., 27, 184 and participant stance 113 and reasons for action 184, 190 criteria for subjective 209–214, 214 n., 215, 215 n., 216–219, 221 grounds of 204–205, 208, 210–211, 215, 217–218, 220–222 objective 204–205, 208–217, 220, 221–222 subjective 204–209, 212–213, 215, 217, 219–221 Pettigrove, Glen 73 n., 80 n., 89 n. Plato 169 n. pleasure and amusement 167, 170, 174–175, 181 and motive 12, 20, 22n., 25 and requests 47 n. hedonism 255, 269 Pohlhaus, Jr., Gail 145 Portmore, Douglas W. 206 n., 229 n., 235 n., 237 n. practical reason or deliberation 16 preferences 24, 147 Priest, Maura 76 n., 79 n., 82 n. principle of affective organization 22 principle of charity 108 principle of independence 40 n. principle of partiality 235 n. privacy 16, 25 projects amusement toward 164 and agential discretion 70–71 and self-change 252, 255, 268–269 promises and agential discretion 51, 63 and directed obligations 54 and gratitude 59 and Hohfeldian rights 53 and interpersonal normative performance 30 and moral disapproval 272 and motive 27 and negligence 11

301

disapproval of breaking 272 promissory duties/relationships 15–16, 53, 57, 206 Protasi, Sara 116 n., 118, 123 n. prudence/prudential goods 237 n., 251–252, 252 n., 253 n., 254, 256–260, 264, 267–270 and requests 35 future-based reasons view of 252–253, 256–259, 259 n., 268 Pummer, Theron 69 n., 213 n., 224 n. punishment 152, 176–177, 283–284 Quine, W.V.O. 101–102 Rabinowicz, Wlodek 121, 124 Radzik, Linda 73 n., 83 n., 85 n. Railton, Peter 207 n., 263 n. Rainbolt, George 50 n. Rawls, John 42 Raz, Joseph 31–32, 34, 34n., 40 n., 51 n., 52, 52 n. reactive attitudes 17–18, 33, 35, 55, 88, 91 reasons all-things-considered 34–37, 120, 133, 135, 221, 252 n. and context insensitivity 9 and directed duties 58–60 and dispositions 19 n. and endorsement 146, 149, 154, 159 and love 117, 119–120, 120 n., 121–122, 122 n., 123, 127, 128, 128 n., 129–133, 133 n. and moral praiseworthiness 184–185, 188–19, 193–196, 200–201 and participant stance 99, 105–107 and permissibility 204–206, 206 n., 215, 217–222 and prerogatives 229 n. and reactive attitudes 33 and requests (see also requests; obligations: and requests; entitlement: and requests) 30–32, 34–36, 36 n., 37–40, 40 n., 41–47, 47 n. and third-party forgiveness 86, 89 exclusionary 32, 52–53 for action 30–31, 34, 39, 40 n., 47, 52, 204, 221 for endorsement 153 Kantian 175

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

302



reasons (cont.) moral vs. non-moral 237 moral vs. prudential 252 n., 257 pro tanto 32–33, 35–36, 38–40, 44 protected 32 prudential 252, 252 n., 257–259, 264, 267 responsiveness to 18, 245 right 153, 169, 184–185, 187, 189–190, 192, 194, 196, 200 second-personal 30 n. reasonable person standard 17, 27 reasoning and agency 21 instrumental 15, 24 n. Redfield Jamison, Kay 254 n. reflective equilibrium 208 regulative ideal 42, 161, 177 requests (see also entitlement: and requests; obligations: and requests; reasons: and requests) and ignorance 41 compliance constraints on 35, 44–45 discretion constraints on 30, 44–45 good reason constraints on 37, 40, 46 mere addition account of 31–32, 32 n., 33–36, 38–39, 39 n., 40, 40 n., 41, 44–45, 47, 47 n. normative effect constraints on 30, 46 reasonable harmony account of 42–46 resentment 33–34, 36–39, 43, 54–55, 65, 78, 78 n., 79, 79 n., 81 n., 83–84, 84 n., 88, 91, 95, 97 n., 104, 112, 181 respect 16, 22, 22 n., 42, 75–76, 133, 148 n., 161–162, 172–177, 179 Richards, Norvin 78 n. Richardson, Peter J. 284 n. Rickless, Samuel 146 Ridge, Michael 196, 199 n., 200 n. ridicule 173–175, 177 right action 184–185, 189–190, 196, 199 n., 201, 220 rightness 186–188, 191–192, 200–201, 206, 209 n. degrees of 219–220 rights against trespass 56–57 alienability of 56, 64 and duties 15–16, 26 and duty of due care 14 and forgiveness 90–91

and harmony of ends 42 and motives 10, 27 and obligations 53–54, 57 n., 58–59, 63 duties of 44 grounding of 34 n. Hohfeldian 53–54 interest theory of 34 n. moral 54, 56, 63, 65 promissory 57 to bodily integrity 57 to privacy 16 to property 57 risk and DDE 145 and duty not to harm 51 and motive 27, 183, 187, 193, 200 and objective permissibility 205, 222 n. and radical change 267 and subjective permissibility 206, 208–220 responsibility for 15 n. Rønnow-Rasmussen, Toni 121, 124 Rosen, Gideon 121 Ross, W.D. 207 n. Rousseau, Jean-Jacques 112 Scanlon, T.M. 27 n., 40 n., 64 n., 191 n. Scheffler, Samuel 206 n. Schopenhauer, Arthur 166, 166 n. Schroeder, Mark 109, 120 n., 126, 126 n., 185–186, 207 semantics adaptation theories of 279–282, 285–286, 290 indicator theories 273, 276–278, 281 success theories 273–276, 280–281 Sen, Amartya 234 n. sentimentalism 120 n. Sepielli, Andrew 206 n., 207 n. Setiya, Kieran 117 n., 123 n. shame 38 n., 170, 176 Shiffrin, Shauna 10 n. Shoemaker, Sydney 261 n. Sider, Ted 229 n., 235 n., 241 n. Sigmund, Karl 274 n. Sinclair, Neil 272 n. Singer, Peter 66, 66 n., 69 n., 156–156, 159 slavery 139, 147–149 Sliwa, Paulina 186–187, 187 n., 188–190, 195, 201

OUP CORRECTED PROOF – FINAL, 4/9/2019, SPi

 Slote, Michael 253 n. Smit, Houston 59 n. Smith, Holly 206 n., 209 n., 221 n. Smith, Michael 106, 200, 217 n. Smuts, Aaron 167 n. Sober, Elliot 279 social meaning 146, 147, 147 n., 148, 148 n., 149–151, 156 social practices 171, 176 social sanction 83, 83 n., 84, 84 n. Socrates 96 Spector, Horacio 210 n., 225 n. Sripada, Chandra S. 284 n. Stang, Nicholas 125 Statman, Daniel 51 n. Steele, Katie 210 n. Stock, Kathleen 149, 150n. Stocker, Michael 51 n. Strawson, P.F. 17–18, 33, 78, 78 n., 79, 79 n., 85, 97, 98 n., 99–100, 102, 104, 105, 181, 181 n. Sumner, L.W. 253 n. sunk costs 243, 246 supererogation 70 n., 212–214, 216, 216 n., 218–219, 221, 223–224, 237 n. sympathy 21, 83 n., 167 Szabo, Zoltan 108 n. Tadros, Victor 206 n. teleology 191, 220–221, 271–282 Tenenbaum, Sergio 208 n. testimony 46–47, 47 n., 48 n., 129, 194–195 Thomson, Judith 134 n. Timmerman, Travis 147 n. Timmons, Mark 59 n., 191 n., 206 n., 221 n. Tomlin, Patrick 222 n. Toppinen, Teemu 197 transparency 12, 39–40, 40 n., 46, 47 n. trust 15, 27, 47, 47 n., 98 n., 188, 266, 289

303

utilitarianism 38, 217, 241 n. utility 38, 67, 69, 217, 225 n., 278 van Inwagen, Peter 261 n. Velleman, David J. 116 n., 253 n. Verwoerd, Wilhelm 73 n. vice 96, 106–107, 113–114, 162–164, 168–169, 173 virtue and virtuous person 161–162, 169–170 duties of 44–45 of taking oneself lightly 161–162, 164, 167–169, 171–175, 177–180, 180 n., 181–182 theory 168 Walker, Margaret Urban 73 n. Wallace, R. Jay 97 n. Walzer, Michael 145 Ware, Owen 88 n. Warmke, Brandon 73 n., 90 n. Way, Jonathan 119 n. Weatherson, Brian 217 n. welfare 22, 64, 68–69, 252–254, 256–259, 259 n., 267–268 desire satisfaction theories of 253, 253 n., 273 objective list theories of 253, 255–256 Whyte, J.T. 273 Williams, Bernard 71, 71 n., 262, 262 n., 263 Wilson, David Sloan 279 Wolf, Susan 163 n. wrongness and cognitivism 191 and motive 19–20, 20 n., 27, 38 and teleosemantics 291 and utility 48 degrees of 209–210, 215–216, 219–221, 225–226, 230 Zaragoza, Kevin 74 n., 80 n. Zimmerman, Michael 126 n., 134 n., 210 n.