Derek Parfit’s Reasons and Persons: An Introduction and Critical Inquiry 2019045380, 2019045381, 9781138595118, 9781138595132, 9780429488450

Citation preview

DEREK PARFIT’S REASONS AND PERSONS

Derek Parfit (1942–​2017) is widely considered to be one of the most important moral philosophers of the twentieth century. Reasons and Persons is arguably the most influential of the two books published in his lifetime and hailed as a classic work of ethics and personal identity. Derek Parfit’s Reasons and Persons:  An Introduction and Critical Inquiry is an outstanding introduction to and assessment of Parfit’s book, with chapters by leading scholars of ethics, metaphysics and of Parfit’s work. Part I provides a much-​needed introduction to key topics and themes in Reasons and Persons that will be useful for those new to Parfit’s complex work.These include Parfit’s idea of self-​defeating theories, rationality and time, personal identity, future generations and well-​being. Part II explores various debates generated by Reasons and Persons, including its connections with Buddhism, metaethics, rationality, transformative choices and further developments in personal identity and metaphysics such as conativism. Combining clear exposition of the major topics and arguments in Reasons and Persons with scholarly perspectives on more advanced themes, this book is ideal for students of ethics, metaethics, metaphysics and anyone interested in Derek Parfit’s philosophy. Andrea Sauchelli is Associate Professor of Philosophy at Lingnan University, Hong Kong. His book Personal Identity and Applied Ethics:  A Historical and Philosophical Introduction is also available from Routledge.

DEREK PARFIT’S REASONS AND PERSONS An Introduction and Critical Inquiry Edited by Andrea Sauchelli

First published 2020 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 selection and editorial matter, Andrea Sauchelli; individual chapters, the contributors. The right of Andrea Sauchelli to be identified as the author of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-​in-​Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-​in-​Publication Data Names: Sauchelli, Andrea, editor. Title: Derek Parfit’s Reasons and persons : an introduction and critical inquiry /edited by Andrea Sauchelli. Description: Abingdon, Oxon ; New York, NY : Routledge, 2020. | Includes bibliographical references and index. | Identifiers: LCCN 2019045380 (print) | LCCN 2019045381 (ebook) | ISBN 9781138595118 (hardback) | ISBN 9781138595132 (paperback) | ISBN 9780429488450 (ebook) Subjects: LCSH: Parfit, Derek. Reasons and persons. | Ethics. | Rationalism. | Self. Classification: LCC BJ1012.P393 D47 2020 (print) | LCC BJ1012.P393 (ebook) | DDC 170–dc23 LC record available at https://lccn.loc.gov/2019045380 LC ebook record available at https://lccn.loc.gov/2019045381 ISBN: 978-​1-​138-​59511-​8  (hbk) ISBN: 978-​1-​138-​59513-​2  (pbk) ISBN: 978-​0-​429-​48845-​0  (ebk) Typeset in Bembo by Newgen Publishing UK

CONTENTS

Notes on contributors  Acknowledgements  Introduction to the collection  Andrea Sauchelli

vii ix 1

PART I 

11

1 Toward a unified theory of morality: An introduction to Part One of Reasons and Persons  Ben Eggleston

13

2 Introduction to Part Two: Rationality and time  Brian Hedden

30

3 Introduction to Part Three: Personal identity  Andrea Sauchelli

48

4 Parfit, population ethics and Pareto plus: Reasons and Persons Part Four  M. A. Roberts

68

5 An opinionated guide to “What Makes Someone’s Life Go Best”  Chris Heathwood

94

vi Contents

PART II 

6 In a different league: Intransitivity, betterness, and league-​based satisficing  Chrisoula Andreou

115 117

7 Conativism about personal identity  David Braddon-​Mitchell and Kristie Miller

129

8 Reasons and conscious persons  Christian Coseru

160

9 Transformative choice and the non-​identity problem  Nilanjan Das and L. A. Paul

187

10 Prudence and self-​concern  Dale Dorsey

209

11 Reductionism, self-​constitution, and the moral significance of personal identity  Carol Rovane

227

12 Non-​identical and impersonal  J. David Velleman

249

Index 

261

NOTES ON CONTRIBUTORS

Chrisoula Andreou is a professor in the Philosophy Department at the University

of Utah, USA. David Braddon-​Mitchell is a professor in the Department of Philosophy, University

of Sydney, Australia. Christian Coseru is a professor in the Department of Philosophy, College of

Charleston, USA. Nilanjan Das is a lecturer in the Department of Philosophy, University College

London, UK. Dale Dorsey is professor of philosophy at the University of Kansas, USA. Ben Eggleston is a professor of philosophy at the University of Kansas, USA. Chris Heathwood is an associate professor at the Department of Philosophy,

University of Colorado Boulder, USA. Brian Hedden is a senior lecturer in philosophy at the University of Sydney,

Australia. Kristie Miller is an associate professor in the Department of Philosophy, University

of Sydney, Australia. L. A. Paul is a professor of philosophy and cognitive science at Yale University, USA.

viii  Notes on contributors

M. A. Roberts is a professor of philosophy at The College of New Jersey, Ewing,

New Jersey, USA. Carol Rovane is a professor of philosophy at Columbia University, USA. Andrea Sauchelli is an associate professor at the Department of Philosophy, Lingnan

University, Hong Kong. J. David Velleman is a professor of philosophy at New York University, USA.

newgenprepdf

ACKNOWLEDGEMENTS

I would like to thank all those people who made this project possible: First, the contributors to this volume (in particular, many thanks to Ben Eggleston and Brian Hedden, who have been especially generous with their comments and suggestions) and, second, the reviewers of the chapters in Part II (Chrisoula Andreou, Derek Baker, Rafael De Clercq, Shyam Nair, Dan Marshall, Dan Waxman). Many thanks also to Tony Bruce for suggesting the format of this collection and to Adam Johnson, Sarah Green and her team for their help during the publication process. On a personal note, I thank my wife Hyelin and my family for their support. I regret that I have never met Derek Parfit, but I like to think that he would have appreciated this further recognition of the fecundity of his work had he been given the opportunity to see it.

INTRODUCTION TO THE COLLECTION Andrea Sauchelli

This volume is divided into two parts: the first introduces Derek Parfit’s Reasons and Persons (henceforth ‘R&P’), whereas the second includes chapters that critically discuss recurring ideas in R&P. The chapters in this collection were written by different authors, and their styles and approaches slightly differ from each other. As the editor of this volume, I decided against imposing any strict requirements on its contributors, with the exception of reminding the contributors to the first part that their chapters are supposed to help the readers better understand the content of Parfit’s book. Some of these writers adopted a more critical style, whereas others chose a more illustrative and exegetical approach. I think that they have all achieved the aim of introducing Parfit’s book clearly, albeit in different ways. The chapters in the second part were commissioned with the intent of collecting works in various fields of philosophy that further elaborate on some of R&P’s principal themes and ideas. As will emerge from this brief introduction, the variety of the areas of research discussed in R&P is remarkable. Parfit’s book has become a contemporary classic, widely read both by philosophers and scholars in other fields (e.g. psychology and even economics). Parfit made several changes to the first edition of R&P published in 1984—​the introduction to the 1987 edition contains a brief summary of these alterations.1 In its 1987 version, R&P comprises four parts and ten appendices. Regarding its content, R&P elaborates on several works that Parfit published from the early 1970s to the beginning of the 1980s. In fact, entire chapters are based on earlier material, albeit modified in light of the criticisms and suggestions Parfit received from an astonishing number of other influential philosophers (the long list includes the likes of Amartya Sen, Shelley Kagan, Larry Temkin, Bernard Williams and John Broome). Among the authors whose published works have more conspicuously influenced R&P, whether directly or indirectly, we may list: Henry Sidgwick, Thomas Nagel, David Wiggins and Bernard Williams.The success and enduring popularity of R&P

2  Andrea Sauchelli

may be partially explained by its impressively high standard of argumentative rigour, the inventiveness and perspicacity of the case studies discussed, and the interesting and controversial theses defended throughout the book. Furthermore, R&P touches on a variety of different topics in apparently disparate fields of philosophy—​among others, rational choice theory, normative ethics, personal identity and population ethics—​and brings together various problems and issues that others (with rare exceptions) have discussed only separately. Although R&P ‘contains multitudes’, there are recurring themes and unifying threads that run through it; for example, Parfit’s attempt to show that one popular version of what he calls the Self-​interest theory (S) is false.2 To a first approximation, this theory about individual rationality tells us that each person has a supreme rational ultimate aim, namely, that her life go for her as well as possible. Because there are different conceptions of how a life can go well, and Parfit aims to provide arguments sufficiently general to apply to several versions of S, he painstakingly explores the applicability of his arguments to the various ways in which S can be further understood.3 In turn, the recurring criticism of S is developed ‘from different fronts’. More specifically, in the first part of R&P, Self-​Defeating Theories, Parfit suggests that S, along with consequentialist theories of morality (C), may be indirectly self-​defeating and possibly self-​effacing. A theory T is directly individually self-​defeating when there are cases in which it is certain that, if someone successfully follows T, she will thereby cause her own T-​g iven aims (the aims given to her by the theory itself) to be more poorly achieved than they would have been if she had not successfully followed T.  Parfit argues that S is not directly but indirectly self-​defeating because there are people for whom it would be worse if they were disposed never to do what they believe would be worse for them. This point does not show that S fails on its own terms because, as a theory of individual rationality, S does not claim that each individual should never act irrationally. In fact, Parfit suggests that in certain cases it may be rational to act irrationally (there can be cases of rational irrationality)—​perhaps just for a short time—​and this is compatible with S. Although not directly self-​defeating, S and C are self-​effacing because they may both imply that we should try to believe in some other theory. For instance, C implies that we should believe the theory such that, if believed, the outcome would be best—​and, crucially, this point is compatible with not believing C itself (a similar reasoning applies to S). In addition, and perhaps even more importantly, S can be collectively self-​defeating. In particular, consider situations involving more than one individual in which: (i) the achievement of each person’s T-​g iven aims partly depends on what others do, (ii) what each person does will not determine what the others do, and (iii) T is agent-​relative; that is, it gives to different agents different aims. Parfit claims that there are many cases in which if each person rather than no-​one does what will be better for herself, or her family, or those she loves, there will be a worse outcome for everyone. Collectively, we would be better off if not everybody acted in a self-​interested way. However, because S is a theory of individual rationality, we may argue that such cases do not prove that it is decisively refuted. The above criticism (being collectively self-​defeating) also applies to what

Introduction to the collection  3

is termed common-​sense morality—​the reason being that this form of morality is generally regarded as including the idea that we have special moral obligations towards members of our family, and that following such obligations may bring about situations that are collectively worse than those in which these obligations are not followed. As highlighted by Ben Eggleston in his introductory chapter, the discussion in Part One is functional to the outlining of the general traits of a new moral theory that does not suffer from the above problems and that unifies consequentialism and certain aspects of common-​sense morality (see Eggleston’s contribution for more details). After having discussed in Part One some arguments that do not seem to directly refute S, in Part Two (Rationality and Time), Parfit proposes other arguments that are supposed to be sufficient to reject it. This section of R&P does not question our non-​reductionist intuitions about our nature and continuity over time (more on non-​reductionism later) and attempts to prove that S should be rejected for reasons that are compatible with different theories of personal identity. As outlined by Brian Hedden in his contribution, in this part of R&P Parfit offers three main arguments against S. In particular, Parfit suggests that it may be rational not to care most about one’s own well-​being and to care at least as much for other things, the pursuit of which we may believe is not conducive to the best possible outcome for ourselves. Examples of desires for these things include the desire to sacrifice oneself (or, at least, not to maximise our well-​being) for moral reasons or desires for achievements (or, better, for some achievements in certain circumstances). The latter are specified in a vaguely Nietzschean fashion because Parfit includes among them the desire to produce a great work of art despite regarding the fulfilment of such a desire as not leading to what is best for oneself (within reasonable limits). Parfit’s point is that these desires may be no less rational than the desire for what the relevant agent deems best for herself.The second line of reasoning against S is focused on one of its alleged faulty structural features, namely, the fact that such a theory is agent-​relative (in specifying the aim that is rational for an agent to pursue, the theory makes essential reference to the agent herself) but time-​neutral (in considering what is best for an agent, said agent should count the well-​being of each temporal part of her life equally). Against this general structure, Parfit suggests that there are reasons to prefer a theory that is either fully neutral or fully relative. The third argument against S is based on the idea that it may not be irrational to be time-​biased—​for instance, to care more about some future parts of our lives rather than those parts in the past. As for the rest of R&P, the subtle thought experiments and ingenious reasoning used to argue for these points have been highly influential and have helped to shape the contemporary debate—​for more details, see Hedden’s chapter. The third part of R&P (Personal Identity) contains another important and recurring theme of the book: the idea that changing our beliefs about our nature and persistence may have important consequences for various issues in moral theory and applied ethics. More specifically, Parfit’s achievements in this part are at least twofold:  first, he clearly delineates and forcefully defends a version of what he will later call Constitutive Reductionism (a family of theories of personal identity)

4  Andrea Sauchelli

and, second, he investigates in some detail the practical and moral consequences of adopting such a view.4 In this respect, Part Three decisively contributes to the debate on whether the holding of the relation of personal identity is a necessary component of the what-​matters relation (R). In this context, R can be understood as the relation that determines the extension of our rational self-​concern: when P at t1 is R-​connected to Q at t2, P’s well-​being is part of Q’s. One of Parfit’s most debated theses is that, contrary to the opinion of many other philosophers, personal identity is not what matters. In this paragraph, I will briefly summarise only part of his lengthy reasoning for this conclusion. Parfit thinks that relation R is exhaustively composed of two more fundamental relations that only partly compose personal identity, namely psychological continuity and psychological connectedness when they hold in the right way.5 The amount and relevance of direct psychological connections between two persons at different times, P and Q, determine the degree of psychological connectedness between P and Q. In several versions of the psychological view, when a strong degree of psychological connectedness is established, and chains of such connections hold between P and Q, we can say that P and Q are psychologically continuous. Examples of direct psychological connections between P at t1 and Q at t2 include P’s experiencing of an event at t1 and Q’s recollection of it at t2, Q’s acting at t2 out of P’s intention at t1, and so on. Crucially, given the nature of the relevant grounding relations, personal identity may be a matter of degree (this view contrasts with the theory that personal identity depends on a non-​physical and non-​psychological entity that is always determinate [e.g. a Cartesian Ego]). Due to the fact that personal identity includes a non-​branching condition—​roughly speaking, the relevant psychological relations should hold between at most two persons each at different times—​and only the proper holding of the relevant psychological relations matter, personal identity is not a necessary condition for what matters. In short, personal identity includes psychological connections, continuity, the non-​branching condition and, on some versions of this criterion, a condition regarding how these relations are supposed to hold (e.g. R should hold in virtue of the continuity of parts of the relevant person’s brain). However, R does not necessarily include the non-​branching condition. Therefore, there are cases in which personal identity and R do not coextensively hold. Parfit argues for this conclusion by elaborating on a thought experiment previously discussed by Sydney Shoemaker and David Wiggins. The upshot is that there are cases of symmetric fissions—​cases in which an individual’s relevant psychological connections existing at t1 are equally distributed between two different persons each existing at a later time t2—​the outcomes of which it may be irrational to regard as being as bad as death. In this part of R&P, Parfit also elaborates on the consequences of adopting a reductionist view (the psychological account of personal identity delineated above is one form of reductionism) for other issues in moral theory. For example, he explores the idea that we may assign a different weight and scope to certain principles of distribution (e.g. equality) proportionally to the degree of psychological connectedness holding intra-​or inter-​personally. See my introductory chapter for more details.

Introduction to the collection  5

Future Generations, the fourth part of R&P, begins with the claim that it is of utmost importance that a moral theory should address how we ought to behave towards future generations. In particular, an acceptable unified moral theory (perhaps of the kind outlined in Part One) should solve a series of puzzles and problems addressing, among other things, harm and beneficence towards future people. Some of these problems partly derive from the fact that our present choices affect not only the number and quality of life of future people but also their identity. For instance, the famous non-​identity problem stems from an attempt to reconcile apparently plausible principles, some of which involve the existence of future people. In particular, some philosophers claim that an act can be wrong only if it makes things worse for some existing or future people (bad must be bad for someone), and that an act is not bad for someone if the act brings about the existence of such a person, provided that the life of this person is at least worth living (or, at least, existence-​conferring acts, acts unavoidable for the existence of an individual, do not make the existence they bring about worse). Now, Parfit puts forward some cases involving actions that we would intuitively judge to be wrong but that are simultaneously unavoidable for generating lives that are at least worth living. For example, take the case of a 14-​year-​old girl who decides to have a child and whose socioeconomic situation clearly suggests that she is unable to provide her child with a good start in life. Had she waited for several more years, she would have been able to give a better start in life to the other child she would have had. The life of the child she gives birth to is worth living but significantly worse than the life she could have given to the other child she would have had if she had waited several more years. Many people agree that the girl should have waited but can we say that, by not waiting, she has thereby harmed her actual child? How can we explain our initial intuition that the girl should have waited? According to Parfit, a satisfying moral theory should solve this problem and meet other requirements. These requirements include: (1) Avoiding the Repugnant conclusion—​roughly speaking, the conclusion that it is better to have a large population of people whose lives are barely worth living than a population of significantly fewer people but with a much higher quality of life; (2) Avoiding the Absurd conclusion—​consider two scenarios: in the first, there is a huge population at t1 with a quality of life higher than our planet now in which one person in 10 billion has a life of uncompensated suffering, whereas in the second scenario, there is a collection of populations of 10 billion each (as before, one person in 10 billion has a miserable life) that do not interact with each other (e.g. each group of 10 billion of these people lives at times after t1). If we impose a local limit on the value of positive quantity but not on negative quantity (for example, if we believe that there is a limit to the positive value that an increase in quantity can have at a specific time but also think that the disvalue of an increase in uncompensated suffering has no upper limit), then the first scenario is bad (because the quantity of suffering is not outweighed by the increase of quantity of positive value), whereas the second scenario is good (because the increase of quality outweighs the quantity of uncompensated suffering). However, this asymmetrical evaluation is absurd. After a painstaking discussion of these problems and possible solutions, Parfit claims that,

6  Andrea Sauchelli

in R&P, he has found no theory that satisfies all these requirements. See Roberts’s chapter on Part Four for a critical discussion of these and related issues. In one of the final sections of R&P, Parfit selects this general point as a common theme or lesson to be learnt from his book: ‘our reasons for acting should be more impersonal’. As it has partially emerged in the brief introduction above, this idea has taken different forms—​for example, the application of his reductionism in personal identity to morality, and his rejection of person-​affecting principles to solve the non-​identity problem. In a way, the rest of Parfit’s career can be seen as an increasingly enriched attempt (mostly at a rather theoretical level) to further refine and improve on the conclusions reached in R&P. The second part of this collection comprises new original papers on some of the ideas in R&P.6 In particular, Chrisoula Andreou’s chapter discusses some theoretical consequences of Parfit’s quandaries (and later elaborations by other philosophers) on puzzles and problems in value theory. In particular, Andreou considers the transitivity of the “better than” relation, using Parfit’s work on the Repugnant Conclusion as her starting point. Andreou considers the possibility of betterness cycles and the implications of accepting the intransitivity of “better than.” She argues that if betterness cycles are indeed possible, then a distinctive form of satisficing that involves reasoning in terms of leagues, plays a crucial role in proper reasoning about what to do. David Braddon-​Mitchell and Kristie Miller’s contribution outlines the conceptual terrain of what they call conative accounts of personal identity. These views have in common the idea that personal identity over time depends on conative phenomena such as desires, behaviours and conventions. In particular, the authors distinguish these conative views along three dimensions, namely, on the basis of (1)  what role the conations play, (2)  what kinds of conations play that role and (3) whether the conations that play that role are public or private. Braddon-​Mitchell and Miller also evaluate such theories by adopting two key desiderata: accommodating faultless disagreement and accommodating our practical concerns. Christian Coseru addresses the following questions: What justifies holding the person that we are today morally responsible for something we did a year ago? Further, why are we justified in showing prudential concern for the future welfare of the person we will be a year from now? Coseru suggests that these questions cannot be systematically pursued without addressing the problem of personal identity. His chapter considers whether Buddhist Reductionism, a philosophical project grounded in the idea that persons can be reduced to a set of bodily, sensory, perceptual, dispositional and conscious elements, provides support for Parfit’s psychological criterion for personal identity. Coseru examines the role that self-​consciousness plays in mediating both self-​concern and concern for others, offering an argument for how reductionism about substantive or enduring selves may be reconciled with the seemingly irreducible character of self-​consciousness. Nilanjan Das and L.A. Paul investigate some philosophical aspects of a subclass of acts, namely, those acts that change who we are (personally transformative acts). A personally transformative act is one that brings into existence a future self that is

Introduction to the collection  7

radically different from who the agent previously was. In some of these cases, the agent may be antecedently certain that the existence of this future self, although worth having, will be unavoidably flawed, even if the future self values its existence. However, if the agent does not perform the transformative act, she will not change so radically, so her unchanged future self may indeed be better off than her transformed future self. In their chapter, Das and Paul argue that situations of this kind raise a problem that is structurally similar to the non-​identity problem. In his contribution to this collection, Dale Dorsey unravels some important theoretical issues related to the Self-​interest theory (or prudence). In particular, he discusses a problem associated with the idea that, although the Self-​interest theory is not the whole story about practical rationality, many philosophers find it entirely plausible to hold that prudence is the best theory of rationality when it comes to normative self-​concern, the idea being that, when our decision concerns only us, we have the strongest reason to promote our welfare to the greatest extent. However, prudence can seem alienating, especially in cases in which we are called upon to abandon deeply valued projects for the sake of projects we may have already taken on (or have yet to take on)—​and yet, prudence seems precisely correct in cases of other, less significant welfare goods. Dorsey argues that this puzzle can be solved by holding that self-​concern is not prudential. In particular, he claims that self-​concern is not (or need not be) welfarist in nature. Carol Rovane focuses her attention on Christine Korsgaard’s early critical response to Parfit’s Reasons and Persons, in which Korsgaard pointed out that Parfit’s reductionist account of personal identity did not take due account of the fact that persons are agents. In her contribution, Rovane offers a reductionist account of personal agency that takes this into account. In particular, Rovane’s reductionism holds that the existence of a person consists in nothing but a certain sort of intentional activity that stands in the right sorts of relations. The account also claims that persons are self-​constituting in much the way that Korsgaard suggests. Rovane’s form of reductionism, however, does not support Korsgaard’s Kantian ambition to derive and ground an unconditional imperative of morality. Nor does it support the Kantian conception of the person of an end in itself, for it entails that persons, qua agents, exist for the sake of the ends that their existence makes it possible to pursue—​the ends for the sake of which they constitute themselves. Rovane’s account agrees with Parfit’s claim that we must revise our common-​sense notions about the moral significance of the individual person.Yet it does not invite the consequentialist orientation that Parfit thought his own reduction invited. The last chapter of the collection, David Velleman’s ‘Non-​identical and impersonal’, discusses several topics through the lens of a broadly Kantian approach to ethics. In particular, Velleman offers a solution to the non-​identity problem that resorts to the Categorical Imperative, thus rejecting some of the utilitarian assumptions that have characterised the debate so far. Velleman claims that rather than focusing on the notions of harm and benefit towards particular people, we should consider the idea that personhood itself can be disrespected. Velleman’s chapter also contains a criticism of Parfit’s theses on what matters—​a criticism that

8  Andrea Sauchelli

Velleman advances from the perspective of his imaginability-​based account of what matters and personal identity.

Notes 1 In particular, Parfit (1984/​87: x). 2 Even a lengthier summary of the book would be inadequate for capturing the richness of R&P. The brief introduction in the main text will sidestep many important issues and be imprecise in certain important aspects. 3 Parfit discusses in Appendix I, thoroughly analysed in Chris Heathwood’s contribution, various different theories of well-​being or welfare. 4 See Parfit (1999). 5 A more precise formulation of relation R is given in a later essay, that is, Parfit (2007). 6 With the exception of the short introduction to Velleman’s chapter, the descriptions of the chapters in Part II in the main text are abridged versions of the abstracts sent by the authors.

References Parfit, D. 1984/​87. Reasons and Persons. Oxford: Clarendon Press. Parfit, D. 1999. Experiences, Subjects, and Conceptual Schemes. Philosophical Topics 26, 1–​2: 217–​270. Parfit, D. 2007. Is Personal Identity What Matters? Marc Sanders Foundation, 31 December. Retrieved at:  www.marcsandersfoundation.org/​wp-​content/​uploads/​paper-​Derek-​Parfit. pdf

Additional resources Dancy. J. ed. 1997. Reading Parfit. Oxford: Wiley. A volume of important critical essays specifically focused on R&P. Parfit’s replies are not collected in the same volume, and some have only been published online (e.g., Parfit, 2007). MacFarquhar, L. 2011. How To Be Good. New Yorker, 29 August. Accessed 26 June 2019. www.newyorker.com/​magazine/​2011/​09/​05/​how-​to-​be-​good An interesting profile that includes details on Parfit’s life. Rabinowicz, W. 2016. Derek Parfit’s Contributions to Philosophy. Theoria 82: 104–​109. A concise and helpful summary of R&P and of some of Parfit’s other achievements. Symposium on Derek Parfit’s Reasons and Persons, Ethics 96, 4. Parfit modified the 1987 edition of R&P also in light of some of the comments and criticisms contained in this special edition of Ethics. Williams, B. 1984. Personal Identity. London Review of Book 6, 10: 14–​15. Bernard Williams’ review of R&P.

Parfit’s publications prior and relevant to Reasons and Persons Parfit, D. 1971. Personal Identity. Philosophical Review 80, 1: 3–​27. Parfit, D. 1972. On ‘The Importance of Self-​Identity’. Journal of Philosophy 68, 20: 683–​690.

Introduction to the collection  9

Parfit, D. 1973. Later Selves and Moral Principles. In A. Montefiore, ed., Philosophy and Personal Relations, 137–​169. Routledge. Parfit, D. 1976a. Lewis, Perry, and What Matters. In A. Rorty, ed., The Identities of Persons, 91–​107. University of California Press. Parfit, D. 1976b. Rights, Interests, and Possible People. In S. Gorovitz et al., eds., Moral Problems in Medicine, 369–​375. Prentice-​Hall. Parfit, D. 1976c. On Doing the Best for Our Children. In M. D. Bayles, ed., Ethics and Population, 100–​115. Schenkman Pub. Co. Parfit, D. 1978. Innumerate Ethics. Philosophy and Public Affairs 7, 4: 285–​301. Parfit, D. 1979a. Is Common-​ Sense Morality Self-​ Defeating? Journal of Philosophy 76, 10: 533–​545. Parfit, D. 1979b. Prudence, Morality, and the Prisoner’s Dilemma. Proceedings of the British Academy 65: 539–​564. Parfit, D. 1980. An Attack on the Social Discount Rate. Philosophy & Public Policy Quarterly 1, 1: 8–​11. Parfit, D. 1982a. Personal Identity and Rationality. Synthese 53: 227–​241. Parfit, D. 1982b. Future Generations: Further Problems. Philosophy and Public Affairs 11, 2. Parfit, D. 1983a. Energy Policy and the Further Future:  The Social Discount Rate. In D. MacLean & P. G. Brown, eds., Energy and the Future, 31–​37. Rowman and Littlefield. Parfit, D. 1983b. Energy Policy and the Further Future: The Identity Problem. In D. MacLean & P. G. Brown, eds., Energy and the Future, 166–​179. Rowman and Littlefield. Parfit, D. 1984. Rationality and Time. Proceedings of the Aristotelian Society 84, 1: 47–​82.

PART I

1 TOWARD A UNIFIED THEORY OF MORALITY An introduction to Part One of Reasons and Persons Ben Eggleston1

Introduction Part One of Reasons and Persons discusses a wide variety of topics, and a first-​ time reader could be pardoned for not seeing the topics as unified in any clear way. This part of the book can seem, instead, like a tour de force of disconnected insights:  brilliant, but haphazard. The closing pages of Part One reveal, however, that nearly all of it can be seen as articulating and arguing for a particular theory of morality. In retrospect, this gives Part One a coherent agenda that, for most readers, is not initially evident. The moral theory advocated blends consequentialism and common-​sense morality, and Parfit’s support for it has three main components: rebutting a particular set of objections to a standard form of consequentialism, arguing that this standard form of consequentialism should be modified in a particular way, and claiming that common-​sense morality should also be modified in a way that makes it similar to the modified form of consequentialism. It would have been fitting for Parfit to call Part One something like “Toward a Unified Theory of Morality” (which is why I have chosen that phrase for the title of this chapter). Why, then, does he call it “Self-​Defeating Theories”? That title is apt for two interconnected reasons. First, both the standard form of consequentialism that Parfit discusses and common-​sense morality are self-​defeating in ways that Parfit describes at length. Indeed, practically all of Part One is concerned with self-​ defeating theories. The second reason is that many of the arguments of Part One focus on the self-​defeating character of the theories being discussed. In the case of the standard form of consequentialism that Parfit discusses, Parfit’s main purpose is to argue that it is not self-​defeating in a way that warrants its rejection. In the case of common-​sense morality, his purpose is the opposite.

14  Ben Eggleston

In fact, according to Parfit it is not just those two theories that are self-​defeating in certain ways: he writes that “all of the best-​known theories are in certain ways self-​defeating” (p.  3).2 He goes on to say, however, that some theories are self-​ defeating in harmless ways while the self-​defeating character of others necessitates their revision or rejection. To appreciate these subtleties in Parfit’s views, some terminological points should be kept in mind. When Parfit first mentions the idea of a theory being self-​defeating, he implies that such a theory thereby “fails in its own terms, and thus condemns itself ” (p. 3). In later passages, however, Parfit diagnoses some theories as self-​defeating, but also holds that they do not fail in their own terms, and do not condemn themselves. In other words, contrary to initial appearances, self-​defeat does not imply self-​condemnation –​the latter verdict depends on what kind of self-​defeat the theory suffers from. I mentioned above that Parfit’s argument for a unified theory builds on discussions of both a standard form of consequentialism and common-​sense morality –​emphasizing, in each case, concerns about self-​defeat. He begins, however, by exploring several aspects (especially the self-​defeat) of yet another theory: the Self-​interest Theory.

The Self-​interest Theory What the Self-​interest Theory is Parfit characterizes the Self-​interest Theory in terms of several claims, the “central” one being that “For each person, there is one supremely rational aim: that his life go, for him, as well as possible” (p. 4). This theory is not a moral theory, and Parfit does not treat it as providing source material for his unified theory. It might seem odd, then, that Parfit would devote a substantial fraction of Part One to this theory. It turns out, however, that the various aspects of self-​defeat that Parfit is concerned to examine in connection with consequentialism and common-​sense morality also arise in connection with the Self-​interest Theory. Thus, Parfit’s discussion of this theory serves mainly  –​at least in Part One  –​to acquaint the reader with self-​ defeat-​related concepts and lines of reasoning whose real importance materializes in connection with those moral theories. (Note, however, that the Self-​interest Theory serves a further purpose outside of Part One, playing a more central role in Part Two.)

How the Self-​interest Theory is indirectly self-​defeating Parfit observes that an obvious way in which the Self-​interest Theory is self-​ defeating is that when people attempt to make their lives go as well as possible, they often fail, due to false beliefs –​for example, a thief ’s overconfidence about being able to steal something without getting caught. Parfit quickly judges that “These cases are not worth discussing” and that this type of self-​defeat is “no objection” to

Toward a unified theory of morality  15

the theory because the problem is caused by the agent’s incompetent implementation of the theory rather than anything internal to the theory itself (p. 5). The scenarios that Parfit takes more seriously involve not just isolated acts in which a person attempts to make her life go as well as possible, but an underlying disposition toward performing such acts. One might expect Parfit to focus on the disposition to always try to make one’s life go as well as possible, and he does mention this disposition (pp. 5–​6). But he claims that we should also include, as pertinent to assessing the Self-​interest Theory, many acts that people perform while “acting on a more particular desire” than the desire to make one’s life go as well as possible. For example, Nina might help Oren make a cake because she loves him, not because she has any opinion about how that activity will affect her life. In such a case, as long as Nina does not regard her act as making her life go worse, we should (according to Parfit) count her decision-​making as aligning with the Self-​interest Theory. So, the disposition Parfit elects to focus on is that of being “never self-​ denying” (p. 6). A person with this disposition might sometimes be inattentive to the goal of making her life go well and might sometimes lack beliefs about her acts’ promotion of that goal, but she will (and this is the defining trait of the disposition) never do what she does believe will be worse for her (p. 6). Parfit presents several cases showing that if a person is never self-​denying, then her life can go worse than if she had a non-​self-​interested disposition. Here is one such case: Suppose that I am driving at midnight through some desert. My car breaks down.You are a stranger, and the only other driver near. I manage to stop you, and I offer you a great reward if you rescue me. I cannot reward you now, but I promise to do so when we reach my home. Suppose next that I am transparent, unable to deceive others. I cannot lie convincingly. Either a blush, or my tone of voice, always gives me away. Suppose, finally, that I know myself to be never self-​denying. If you drive me to my home, it would be worse for me if I gave you the promised reward. Since I know that I never do what will be worse for me, I know that I shall break my promise. Given my inability to lie convincingly, you know this too. You do not believe my promise, and therefore leave me stranded in the desert. This happens to me because I am never self-​denying. p. 7 Parfit’s example might seem rather peculiar and far-​fetched, but it is an instance of a large class of cases that exhibit a well-​established and widely applicable concept known as the paradox of happiness (though Parfit does not use this term). This is the initially surprising –​but generally acknowledged –​fact that “regarding happiness as the sole ultimately valuable end or objective, and acting accordingly, often results in less happiness than results from regarding other goods as ultimately valuable (and acting accordingly).”3 Parfit’s example illustrates the paradox perfectly: if he (upon having been driven home) were able to regard keeping his promise as more

16  Ben Eggleston

important than advancing his happiness, then he (earlier, in the desert) could sincerely make that promise and he would thereby advance his happiness more than he actually can, given that he is never self-​denying. Hence, being never self-​denying is worse for him than having some other disposition (i.e., one that allowed occasional self-​denying acts, such as keeping certain promises) would be. According to Parfit, in cases such as these the Self-​interest Theory is “indirectly individually self-​defeating” (p. 7). As Parfit defines this term, it applies to a theory T (here the letter “T” is a variable) when the following is true: “if someone tries to achieve his T-​given aims [such as making one’s life go as well as possible], these aims will be, on the whole, worse achieved” (p. 5). This is Parfit’s first major conclusion concerning the self-​defeating character of any of the theories he discusses.

The failure of several objections to the Self-​interest Theory Having shown that the Self-​interest Theory is indirectly individually self-​defeating, Parfit then asks whether this means the theory “fail[s]‌in its own terms,” or whether it “condemn[s] itself ” (p. 7). To answer these questions, Parfit notes that the theory implies many things about what people have reason to do, and one of these is the following: what each person has most reason to do is to cause himself to have, or to allow himself to keep, any of the best possible sets of motives [which, for Parfit, include dispositions], in self-​interested terms. These are the sets of motives of which the following is true.There is no other possible set of motives of which it is true that, if this person had these motives, this would be better for him. p. 8 Because of the paradox of happiness, being never self-​denying is “for many and perhaps most people” (p. 17) definitely not among the best possible sets of motives. Consequently, Parfit explains, the Self-​interest Theory does not tell these people to have that disposition. Thus, when people suffer the ill effects of having that disposition, they are violating, not complying with, the Self-​interest Theory. So, the theory is not failing in its own terms (p. 11).4 Parfit goes on to consider other questions about the Self-​interest Theory, such as whether there might be other grounds for saying that it fails in its own terms (p. 11), or whether it is objectionable because it directs people to be disposed to sometimes act irrationally (p. 12), or whether it is objectionable because it implies that we cannot entirely avoid acting irrationally (p. 16), or whether it is objectionable because it might sometimes tell agents to believe a revised version of itself (p. 19), or whether it would be objectionable if it were self-​effacing, in the sense of telling “everyone to cause himself to believe some other theory” (p. 24), or whether it is objectionable because the outcome would be bad for a group of people if they believed it and were unable to change their beliefs or dispositions (p. 43), or whether it is objectionable because it treats acting rationally as a mere means (p. 45).

Toward a unified theory of morality  17

Parfit rejects all of these. Note, however, that for our purposes, as we keep in mind the trajectory of Part One as a whole, Parfit’s defence of the Self-​interest Theory against these objections has less to do with his support for the Self-​interest Theory per se than with his desire to discredit the logic of these objections. For he will want to entertain and dismiss them yet again in his discussion of consequentialism, in order to establish the latter theory as a worthy contributor to his unified theory.

Practical dilemmas The Self-​interest Theory figures prominently in one other major topic of Part One: situations that Parfit calls “practical dilemmas.” These include not only traditional, two-​person prisoner’s dilemmas, but many-​person variants as well. The latter, like the former, are characterized by the fact that “it is certain that, if each rather than none of us does what will be better for himself, this will be worse for everyone” (p. 59). Parfit holds that many-​person cases occur more frequently than, and matter more than, two-​person cases: “Though we can seldom know that we face a Two-​Person Prisoner’s Dilemma, we can very often know that we face Many-​Person Versions. And these have great practical importance. The rare Two-​ Person Case is important only as a model for the Many-​Person Versions” (p. 59). Parfit gives about a dozen examples of such many-​person dilemmas (pp. 61–​62). One example whose resonance has perilously increased in the decades since the publication of Reasons and Persons pertains to climate change: It is better for each person if he or she uses as much energy as is optimal for his or her lifestyle, but if everyone makes this same decision, the result is worse for each person than if everyone exercises some restraint. Many-​ person dilemmas are critical to Parfit’s discussion of common-​ sense morality, as we will see below. But they also aid Parfit in exploring one further objection to the Self-​interest Theory that (like the objections mentioned above) Parfit dismisses as unsuccessful. This objection is based on another kind of self-​ defeat: direct collective self-​defeat. In defining this notion Parfit gives several criteria of increasing precision, but the essential idea is that compliance with the theory by each of a group of people ultimately frustrates their achievement of their aims (p. 55). For example, the energy-​consumption example just mentioned shows that the Self-​interest Theory is directly collectively self-​defeating because compliance with the theory by each person results in energy consumption of such a magnitude that the result for each person is worse than if they had each exercised some restraint, in defiance of the Self-​interest Theory. Parfit grants that the Self-​interest Theory is directly collectively self-​defeating (p. 88). But, as with the indirect individual self-​defeat discussed above, he holds that this does not underwrite a successful objection to the theory. This is because the theory is not concerned with the collective level –​it is only concerned with the individual level. Consequently, its collective self-​defeat does not entail that it either fails in its own terms or condemns itself (p. 92).

18  Ben Eggleston

Consequentialism What consequentialism is In contemporary moral philosophy, consequentialism is generally taken to refer to the principle that acting rightly is a matter of choosing the act that leads to the best possible outcome.5 In Reasons and Persons, Parfit discusses a theory that he calls “Consequentialism,” whose central claim is that “There is one ultimate moral aim:  that outcomes be as good as possible” (p.  24). Another tenet he attributes to this theory echoes the general consequentialist principle mentioned above:  “What each of us ought to do is whatever would make the outcome best” (p. 24). This theory forms half of the basis for the unified theory that he proposes at the end of Part One. Accordingly, as indicated above, he both defends this theory against several objections and argues that it should be modified in a particular way. Before turning to those matters, it is worth pausing briefly to observe that the way Parfit states the central claims of the Self-​interest Theory and Consequentialism imply that they operate in different domains, rather than being rivals of each other. The central claim of the Self-​interest Theory refers to the supremely rational ultimate aim, while the central claim of Consequentialism refers to the ultimate moral aim. Although Parfit writes that moral theories and theories about rationality both give answers to the question “What do we have most reason to do?” (p. 3) and touches on the opposition of morality and self-​interest (p. 88), he does not position the Self-​interest Theory and Consequentialism against each other. In fact, the most salient connection between them, in Parfit’s discussion, is that they are both targets of largely the same set of unsuccessful objections.

How consequentialism is indirectly self-​defeating Above I claimed that the main purpose of Parfit’s discussion of the Self-​interest Theory in Part One is to acquaint the reader with self-​defeat-​related concepts and lines of reasoning whose real importance materializes in connection with the moral theories he discusses. One basis for this claim is that Parfit’s discussion of the Self-​interest Theory is closely echoed by his discussion of Consequentialism, as he acknowledges at the beginning of the latter discussion (p. 24). Let us review those concepts and lines of reasoning as they arise in relation to Consequentialism. Paralleling his consideration of the dispositions that we should associate with the Self-​ interest Theory, Parfit identifies a disposition to associate with Consequentialism: that of being a “pure do-​gooder” –​someone who is disposed to “always try to do whatever would make the outcome as good as possible” (p. 27). Parfit then argues that if everyone were a pure do-​gooder, then the outcome would probably be worse than if some people had some other dispositions. The essence of his argument runs as follows:

Toward a unified theory of morality  19

Most of our happiness comes from having, and acting upon, certain strong desires. These include the desires that are involved in loving certain other people, the desire to work well, and many of the strong desires on which we act when we are not working.To become pure do-​gooders, we would have to act against or even to suppress most of these desires. It is likely that this would enormously reduce the sum of happiness. p. 27 According to Parfit, in cases such as these Consequentialism is “indirectly collectively self-​defeating” (p. 28). This term applies to a theory T (again, using that letter as a variable) “when it is true that, if several people try to achieve their T-​g iven aims, these aims will be worse achieved” (p. 27).

The failure of several objections to Consequentialism As with the Self-​ interest Theory, Parfit argues that Consequentialism is not discredited by the scenarios that show that it is indirectly self-​defeating. Just as the Self-​interest Theory requires us to have the motives that will make our lives go best (not necessarily the disposition of being never self-​denying), Consequentialism requires us to have the motives that will have the best consequences generally (p. 26). Assuming these do not include being a pure do-​gooder, Consequentialism “tells us that it would be wrong to cause ourselves to be, or to remain, pure do-​ gooders.” Thus, Consequentialism does not fail in its own terms or condemn itself (p. 28).6 Parfit goes on to consider a series of questions about Consequentialism that closely follow the further questions about the Self-​interest Theory that he considered. These include whether Consequentialism is objectionable because it directs people to be disposed to sometimes act wrongly (p.  32), or whether it is objectionable because it implies that we cannot entirely avoid acting wrongly (p. 36), or whether it is objectionable because it tells us to cause ourselves to do what it claims is wrong (p. 37), or whether it would be objectionable if it were self-​effacing (in the sense specified in the discussion of the Self-​interest Theory) or esoteric, in the sense of “telling those who believe it not to enlighten the ignorant majority” who do not believe it (p. 41), or whether it is objectionable because the outcome would be bad for a group of people if they all believed it and were unable to change their beliefs or dispositions (p. 43), or whether it is objectionable because it treats acting morally as a mere means (p. 45). Although these issues have spawned further discussion,7 the key point to appreciate for our purposes is that Parfit is maintaining the viability of Consequentialism because it forms half of the basis for the unified theory that he proposes at the end of Part One. Almost everything about Part One that I  have described so far is from the first of its five chapters, which occupies nearly half of it. (The only exception is the material pertaining to practical dilemmas, which is from Chapter  2.) In the preface to Reasons and Persons, Parfit writes that the first chapter is the only one in

20  Ben Eggleston

which he does not “try to challenge what we assume.” Instead, in that chapter, “I cannot avoid repeating what has been shown to be true.” Consequently, he says, it is “dreary” (p. x).

Practical dilemmas In my overview of Parfit’s discussion of the Self-​interest Theory, I mentioned that practical dilemmas show that that theory is directly collectively self-​defeating, but that Parfit holds this to be acceptable for that theory since it is concerned with the individual level, not the collective level. No such reply would be available to excuse any direct collective self-​defeat that Consequentialism could be shown to exhibit, since the collective level is precisely where Consequentialism is meant to operate (since it is a moral theory). But Parfit nips this objection in the bud: he claims that Consequentialism cannot be directly collectively self-​defeating because compliance with the theory by each person in a group of people will definitely result in the best possible outcome –​that follows directly from what the theory requires (p. 54). Having stated this simple and apparently decisive point, however, Parfit goes on to investigate a question that poses a different threat to Consequentialism: In practical dilemmas that involve such large numbers of people that “any single altruistic choice would make no difference,” can we “explain why we should contribute by appealing to the consequences of our acts”  –​i.e., without appealing to any nonconsequentialist considerations (p.  67)? If we cannot, then Consequentialism might not imply that we should contribute in such cases. Depending on how obvious we take it to be that we should contribute in such cases, this fact about Consequentialism (if it turns out to be a fact) might be taken to indicate that Consequentialism is seriously flawed as a moral theory.

Mistakes in moral mathematics According to Parfit, we can indeed appeal to the consequences of our acts in order to explain why we should contribute, if we avoid what he calls “mistakes in moral mathematics” (p. 67). He discusses a total of five such alleged mistakes.The first two concern how to credit or blame people for the combined effects of their individual acts (pp. 67–​73), and Parfit’s claims here have generated some discussion.8 The third mistake is to ignore possible events that are highly unlikely, such as the possibility that one’s vote will determine the outcome of a presidential election or the possibility that one of many nuclear-​reactor components will fail and cause a catastrophe. Parfit writes that such possibilities must not be ignored in decision-​making, though of course it is appropriate to discount them according to how improbable they are (pp. 73–​75). However, for the trajectory of Part One as a whole, the most important mistakes are the fourth and fifth ones. The fourth mistake is to regard very small effects as morally insignificant (p.  75). For example, in the energy-​consumption example mentioned above, each consumer might think that because the effects of his or

Toward a unified theory of morality  21

her consumption decisions on other people are very small, those effects are morally insignificant, and thus provide no reason for him or her to exercise restraint. The fifth mistake is similar to the fourth, but refers to imperceptible effects instead of very small effects (p. 75). Parfit declines to address the fourth mistake separately and, instead, argues against the fifth mistake at length. This suffices to address the fourth mistake if one thinks –​as seems plausible –​that if imperceptible effects matter in the moral assessment of an act, then surely any perceptible (even if very small) effects also matter in such an assessment. Parfit’s argument for the moral significance of imperceptible effects includes several artfully constructed hypothetical examples. Probably the most important of these is “The Harmless Torturers” (p. 80). In this case, there are a thousand torturers and a thousand victims. Each victim is connected to a machine that causes pain, where the intensity of the pain is determined by the position of a dial on the machine. The dial has a minimal setting of 0, which corresponds to mild pain, and can be advanced by a thousand increments. Every one-​increment turn of the dial is imperceptible to the victim, but the maximal setting of 1,000 is extremely painful. There is a button that simultaneously advances each of the thousand dials by one increment. One day, all of the dials are initially set at 0. During the day, each of the thousand torturers presses the button once. By the end of the day, each victim is in severe pain. (This example closely resembles an example created by Jonathan Glover and M. J. Scott-​Taggart,9 which Parfit cites. He writes that his discussion of the five mistakes, and “especially” his Harmless Torturers example,“derives entirely from the stimulus of this brilliant example” (p. 511, n. 44).) Parfit judges that “the torturers are clearly acting wrongly” (p.  80), and he supports this conclusion with a disjunctive argument hinging on whether it is, or is not, possible for a person’s pain to worsen imperceptibly (which Parfit acknowledges is debatable). If this is possible, then the argument for the wrongness of the actions of the torturers is straightforward: each torturer causes a lot of pain, even though no individual victim perceives the worsening of pain inflicted by any individual torturer (p. 80). On the other hand, if it is not possible for a person’s pain to worsen imperceptibly, then although no torturer harms anyone, we can conclude that each acts wrongly because “they together impose great suffering” (p. 80). Either way, by the original construction of the example, the effects of each torturer’s act are imperceptible, so Parfit concludes from this example that imperceptible effects can be morally significant. As mentioned above, this suffices to show that small effects are morally significant, too. I mentioned at the beginning of this section that Parfit’s discussion of these “mistakes in moral mathematics” provides the key to explaining why we should contribute in many-​person practical dilemmas.We can now see how it does this, by considering once again the energy-​consumption example. If Parfit is correct to hold that small and even imperceptible effects of acts can be morally significant, then the greenhouse-​gas effects of even a single act of energy consumption cannot justifiably be ignored in an accounting of the consequences of that act. Such an accounting would include those effects and would supply the elements of a consequentialist

22  Ben Eggleston

argument in favour of exercising some restraint. Thus, Consequentialism is saved from implying the unpalatable verdict that it is morally permissible for a person to consume energy unrestrainedly.

Common-​sense morality The third and final theory that plays a major role in Part One is common-​sense morality. Parfit argues that this theory is self-​defeating in a way that should persuade even its proponents to embrace a small but meaningful modification of it –​leading toward the unified theory that he proposes at the end of Part One.

What common-​sense morality is In ordinary parlance (at least among moral philosophers), common-​sense morality is understood as differing from consequentialism in virtue of designating some particular kinds of conduct as being especially bad –​if not prohibited absolutely, then at least having a strong presumption of wrongness and therefore being justifiable only in exceptional circumstances. Such kinds of conduct might include lying, stealing, breaking promises, racial discrimination, religious persecution, torture and murder. According to common-​sense morality, the badness of these kinds of conduct is not just a matter of their tending to have bad consequences, though that of course does matter greatly. In this way, common-​sense morality contrasts with consequentialism since the latter assesses conduct strictly through the lens of its consequences, and therefore does not designate some kinds of conduct as being especially bad in and of themselves. Although Parfit acknowledges this element of what is generally called common-​sense morality (p.  112), he focuses on a different, but perhaps no less integral, component of that moral outlook in order to characterize the theory he calls Common-​Sense Morality. He writes that “Common-​Sense Morality largely consists in” the following moral belief: Most of us believe that there are certain people to whom we have special obligations. These are the people to whom we stand in certain relations—​ such as our children, parents, friends, benefactors, pupils, patients, clients, colleagues, members of our own trade union, those whom we represent, or our fellow-​citizens. We believe that we ought to save these people from certain kinds of harm, and ought to try to give them certain kinds of benefit. p. 95 Part of the content that Parfit sees these special obligations as having is that they can require a person to save a child (or a parent, friend, etc.) from some harm even at the cost of saving a stranger from a larger harm. For example, Common-​ Sense Morality might well require that a parent save her child from a broken arm even if she knows that she is thereby forgoing the opportunity to save a stranger

Toward a unified theory of morality  23

from a different injury that she knows will be more serious. Parfit adds, however, that even according to Common-​Sense Morality, “This priority is not absolute. I ought not to save my child from a cut or bruise rather than saving a stranger’s life.” Nevertheless, there is some priority: according to Common-​Sense Morality, “I ought to save my child from some harm rather than saving a stranger from a somewhat greater harm” (p. 95). It is useful to situate Parfit’s characterization of Common-​Sense Morality in the context of his distinction, which has proven influential, between agent-​neutral theories and agent-​relative theories (p. 27). Consequentialism is an agent-​neutral theory because it assigns the same moral aims to all agents: that outcomes be as good as possible. In contrast, Common-​Sense Morality is an agent-​relative theory because the aims it assigns to an agent (and their relative weights) are relative to that agent’s values, relationships, or other attributes. For example, in some situations, Common-​Sense Morality might assign Parfit the (primary) aim of saving his child from a particular harm even at the cost of some other person incurring a somewhat greater harm, while it might assign the other person’s parent the (primary) aim of averting that harm even at the cost of Parfit’s child incurring a harm. Parfit’s characterization of Common-​Sense Morality is clearly rather idiosyncratic and schematic, focusing on the structural feature of agent-​relative obligations (especially ones that diverge from producing the best possible outcomes) rather than the kind of substantive rules and values that we should want children to be taught by their parents and teachers. The reason for this eccentric focus is that this structural feature is the essential ingredient in Parfit’s argument that Common-​ Sense Morality is self-​defeating, which is a key claim in his argument for his unified theory.

How Common-​Sense Morality is directly self-​defeating To set the stage for Parfit’s comments on the self-​defeat of Common-​Sense Morality, let us recall his self-​defeat-​related conclusions about the previous two theories. Both the Self-​interest Theory and Consequentialism are indirectly self-​defeating at the levels with which they are concerned –​the individual level and the collective level, respectively. But indirect self-​defeat is not a serious flaw. As for direct self-​defeat (which can be a serious flaw), the Self-​interest Theory is directly self-​defeating only at the collective level, but that is a non-​issue since the Self-​interest Theory is not concerned with the collective level. And by the logic of Consequentialism, it cannot be directly self-​defeating at the collective level. So, each theory avoids direct self-​defeat at the level where that attribute would be a serious flaw. Matters are different with Common-​Sense Morality: Parfit argues that at the level with which it is concerned –​the collective level –​it is directly self-​defeating. He shows this by presenting several hypothetical examples that are ingenious variations on the traditional case of the prisoner’s dilemma. Several of these examples are what he calls “Parent’s Dilemmas” (p. 96), and the simplest one has the following form (though I have rewritten it to make some of its key features explicit).

24  Ben Eggleston

Quinn has a son, and so does Ramona. (But there are no relationships of special obligations between the two families.) Each child is vulnerable to two harms, a small one and a large one. Quinn can either save her son from the small harm or save Ramona’s son from the large harm. Similarly, Ramona can either save her son from the small harm or save Quinn’s son from the large harm. The harms are close enough in size that, according to Common-​ Sense Morality, each parent’s obligation to save her own son from the harm threatening him that she can avert (the small harm) outweighs her obligation to save the other boy from the harm threatening him that she can avert (the large harm). Parfit also presents Parent’s Dilemmas involving the conferral of benefits rather than the blocking of harms (pp. 96–​97), but in all cases, the upshot is the same: Common-​ Sense Morality requires the parents to prioritize their own children rather than avert the most harm or do the most good. The result is that every child suffers a larger harm than was necessary in the circumstances, or receives a smaller benefit than was possible in the circumstances. To see how this example shows that Common-​Sense Morality is directly collectively self-​defeating, recall the definition of that notion from the discussion of the Self-​interest Theory and practical dilemmas: the essential idea is that compliance with the theory by each of a group of people ultimately frustrates their achievement of their aims (p.  55). In the example of Quinn and Ramona, Common-​Sense Morality holds that each of them has a special obligation to shield her own son from harm (as well as an ordinary obligation to shield other people from harm).We may consider the safety of their children their aim, for the purposes of applying the definition of self-​defeat just given. And when Quinn and Ramona comply with Common-​Sense Morality, this aim is achieved worse than was necessary in the circumstances. Here, then, is a case in which Common-​Sense Morality frustrates the achievement of the aim it tells people to pursue. Thus, it “is here directly collectively self-​defeating” (p. 99). Parfit concedes that two-​person parent’s dilemmas do not arise often. But he claims that situations often arise in which there are not just two parents, but many parents, with each parent facing a choice like the ones in the parent’s dilemmas, such that if each parent gives priority to his or her child, then all of the children end up faring worse than they needed to.This is, for example, the problem of public goods (p. 98). Parfit adds that “Similar remarks apply to all similar obligations—​such as those to pupils, patients, clients, or constituents. With all such obligations, there are countless many-​person versions like my three Parent’s Dilemmas” (p. 98; see also p. 102). One feature of these cases that justifies the centrality that Parfit claims for them is that they arise even when a person is willing to justify his or her behaviour in moral terms –​not merely in self-​interested terms. Consider, for example, a person who is planning to buy a huge, gas-​guzzling SUV, imposing harms on others that exceed the benefits he will secure for himself (relative to, say, a more fuel-​efficient

Toward a unified theory of morality  25

car). Although we cannot accuse the person of acting irrationally in terms of his own self-​interest, we might think that if only he could be persuaded that he must also account for his behaviour in moral terms, then he would acknowledge that he is behaving wrongly. However, the same person might adapt some remarks that Parfit discusses (p. 100; see also p. 62) and say, “Oh, if it were just me, I would be glad to buy a smaller car. But I  also have to think about my children, and their safety and comfort. And my obligations to them are stronger than my obligations to people in general. Thus, all things considered, the moral reasons ultimately point in the direction of the bigger car.” Thus, the shift from a self-​interested point of view to a moral point of view is not enough to get people to make decisions that, collectively, result in the maximal achievement of their moral aims. Some moral points of view, such as that of Common-​Sense Morality, exhibit the same kind of collective self-​defeat that we saw in the case of the Self-​interest Theory. But whereas the Self-​ interest Theory can shrug off collective self-​defeat as being a non-​issue because it is only concerned with the individual level, collective self-​defeat is a serious flaw for Common-​Sense Morality or any other moral theory.10

Revising Common-​Sense Morality The collective self-​ defeat of Common-​ Sense Morality leads Parfit to propose revising that theory in certain ways. He proposes three distinct revisions, but they all basically boil down to the idea that people should deviate from the prescriptions of Common-​Sense Morality when those prescriptions frustrate the achievement of the aims of the theory. For example, in situations of the kind we have been considering, people should do what will cause those aims to be better achieved, not worse achieved (pp. 100–​103).To understand the precise character of this idea, recall that the aims in question are the aims that Common-​Sense Morality prescribes for people. Thus, the idea is not that people should feel free to ignore the demands of Common-​Sense Morality (or morality in general) in order to more fully achieve their self-​interest. Rather, the idea is that the aims of Common-​Sense Morality are worth promoting, and that people should promote them more thoughtfully than the dictates of Common-​Sense Morality prescribe. For example, a person should be willing to contribute to a public good that will benefit his or her child, even if it would be better for that child for the parent to decline to contribute and devote the same resources to that child individually. After specifying this revised version of Common-​Sense Morality, Parfit gives several arguments for the claim that proponents of Common-​ Sense Morality should find these revisions appealing. One argument begins by emphasizing the inherent senselessness of accepting that one’s moral theory is directly collectively self-​defeating. After all, the direct self-​defeat of Common-​Sense Morality means that the self-​defeat stems not from some agents’ failing to act in accordance with the theory, but from all agents’ successfully acting in accordance with the theory (p. 103). And on most views about the nature of morality, a moral theory –​unlike a theory of individual rationality such as the Self-​interest Theory  –​purports to

26  Ben Eggleston

offer a code for collectively achieving our moral aims (p. 106; see also p. 113). Thus, the direct collective self-​defeat of a particular theory is an absurdity that any proponent of that theory should want to remedy. Since the revisions Parfit proposes remedy Common-​Sense Morality’s self-​defeat while leaving the rest of its content intact, proponents of Common-​Sense Morality should regard his revisions as costless improvements (p. 106). A second argument proceeds independently of the concept of self-​defeat. Parfit invites us to imagine that in one of his parent’s dilemmas, two additional conditions hold: (1) the parents can communicate and (2) each parent can best advance the interests of his or her child by making a conditional promise to the other parents that if they refrain from giving priority to their own children, then he or she will refrain from doing that, too. (For example, Quinn and Ramona could each promise to save the other’s son from the larger harm, rather than save her own son from the smaller harm.) In situations in which these two conditions hold, a parent contemplating making the promise would not be contemplating deviating from Common-​ Sense Morality. Instead, making the promise would be required by Common-​Sense Morality: by hypothesis, that is each parent’s best option for advancing the interests of his or her child. And once the promise has been made, even Common-​Sense Morality would prescribe that it be kept, simply as an instance of its moral regard for promise-​keeping generally. In a sense, then, Common-​Sense Morality can be seen as endorsing exactly the kinds of revisions to it that Parfit proposes (p. 107). Parfit acknowledges that communication among the parties is not always possible. But he claims that people can fulfill the aims of Common-​Sense Morality more fully when they can communicate than when they cannot. This provides a reason for any supporter of the aims of Common-​Sense Morality to take the communication scenario as a reference point when thinking about how it would be desirable for people in such situations to act, even if communication is not possible. Thus, the impossibility of communication does not undermine the basic thrust of the argument just given: any supporter of the aims of Common-​Sense Morality would wish for people to act as if they had made, and were intent on complying with, the promises described above. So, we see again “a sense in which [Common-​ Sense Morality] itself tells us to accept this revised version of itself ” (p. 108).11

The unified theory of morality As I mentioned above, Part One culminates in Parfit’s proposing a new theory of morality. He prepares the reader for his statement of this theory by commenting on how his discussions of Consequentialism and Common-​Sense Morality reduce the “distance” or “disagreement” between those theories (p. 111). First, recall his discussion of Consequentialism  –​specifically, his discussion of its indirect collective self-​defeat. One lesson of that discussion is that instead of being pure do-​ gooders, people should have strong desires to benefit their families and friends, and to do their work well, which would result in their having strong desires to benefit their pupils, patients, clients and other people that they have special professional

Toward a unified theory of morality  27

relationships with. Additionally, people should have strong aversions to committing certain kinds of acts, such as murder and deception. All of these desires will impel people to be disposed to act as Common-​Sense Morality requires (p. 112). Second, recall Parfit’s discussion of Common-​Sense Morality. As we saw just above, Parfit argues that any adherent of Common-​Sense Morality should embrace the revisions he proposes. He writes that the revised theory “is Consequentialist, giving to all of us common moral aims” (p. 111) –​here Parfit may be claiming that the revised theory is agent-​neutral. He then says that moving from Common-​Sense Morality to the revised theory “reduces the disagreement between Common-​Sense Morality and Consequentialism” (p.  111, emphasis added).12 With Consequentialism and Common-​Sense Morality brought closer together in this way, Parfit conjectures that “We might be able to develop a theory that includes and combines revised versions of both” of them. “Call this the Unified Theory” (p.  112). Parfit actually says very little about the content of this theory, writing that developing it “would take at least a book” (p.  113). Instead, he emphasizes a major obstacle that the project would face: the differences in the moral judgements that follow from Consequentialism and Common-​ Sense Morality. Although Consequentialism can be brought closer to Common-​Sense Morality by way of dispositions to act in ways that Common-​Sense Morality prescribes, the two theories still have different contents. Thus, there will be cases in which complying with one theory means violating the other, and vice versa. In articulating the Unified Theory, “our greatest task would be to reconcile these conflicting beliefs” (p.  113). (Strangely, Parfit says nothing here about the extent to which the unification project could be furthered by the revisions to Common-​Sense Morality that he recommended. Recall that he argued that they eradicate much if not all of the content of Common-​Sense Morality that cannot be reconciled with Consequentialism.) Let me conclude this overview of Part One by describing a conversation that took place on 20 February 1865. On that date, John Stuart Mill spent time with John Russell, whose son Bertrand would be born seven years later and receive the honour of having Mill as his godfather. The conversation turned to the topic of moral progress, and Russell’s diary reports the following of Mill: It did one great good to hear him & raised one into a hopeful state of mind. He said the wish & intention to do good was good in itself—​and he said the great thing was to consider one’s opponents as one’s allies; as people climbing the hill on the other side.13 Parfit quotes Mill’s “climbing the hill on the other side” phrase in the penultimate paragraph of Part One, and characterizes his Unified Theory in that spirit (p. 114). Consequentialism and Common-​Sense Morality, rather than being seen as rivals, can be seen as the outlooks of people who are on different sides of a mountain, but who share the goal of reaching the summit and seeing the whole truth of morality.

28  Ben Eggleston

This metaphor remained a touchstone for Parfit. More than a decade later, he wrote a manuscript called Climbing the Mountain,14 which, although never published, was circulated widely among moral philosophers. In 2011, much of this manuscript appeared as –​or was superseded by –​Parfit’s two-​volume treatise On What Matters. There Parfit writes that although Kantians, Contractualists and Consequentialists are often seen as disagreeing deeply with one another, that perception is a mistake: “These people are climbing the same mountain on different sides.”15 More important than the metaphor, one of the main claims of On What Matters is the convergence of the three views just mentioned.Thus, the quest for a unified theory that we find in Part One of Reasons and Persons is significant not only because of the impact of that book, but because it is the first major articulation of what would prove to be one of the major themes of Parfit’s cumulative body of work.

Notes 1 I would like to thank Chris Heathwood, Brian Hedden and Andrea Sauchelli for helpful comments on an earlier version of this chapter. 2 Parenthetical in-​text page references are to Parfit (1984/​87). 3 Eggleston (2013a: 3794). 4 For criticisms of this defence of the Self-​interest Theory, see Dancy (1997: 4–​11) and Adams (1997: 255–​256). 5 Shafer-​Landau (2018: 122). 6 For a criticism of this defence of Consequentialism and Parfit’s reply, see Gruzalski (1986a:  771–​777) and Parfit (1986:  865, section 2). For further criticism, see Dancy (1997: 11–​16). 7 See, e.g., de Lazari-​Radek and Singer (2010) and Eggleston (2013b). 8 See, e.g., Jackson (1997), Eggleston (2000), Eggleston (2003) and Petersson (2004). 9 Glover and Scott-​Taggart (1975: 174–​175). 10 For criticism of this reasoning and Parfit’s reply, see Kuflik (1986) and Parfit (1986: 849–​ 854 and 865–​867). For further criticism, see Gruzalski (1986b:  150–​151) and Adams (1997: 256–​261). 11 For criticism of the inferences Parfit draws from the collective self-​defeat of Common-​ Sense Morality, see Mendola (1986). 12 For criticism of Parfit’s claim that his proposed revisions bring Common-​Sense Morality closer to Consequentialism, see Gruzalski (1986b: 145–​150). 13 Russell and Russell (1937: 373). 14 MacFarquhar (2011: 50). 15 Parfit (2011: vol. I, p. 419; see also vol. II, p. 259).

Bibliography Adams, R. M. 1997. “Should Ethics Be More Impersonal?” in J. Dancy (ed.), Reading Parfit (Oxford: Blackwell Publishers), pp. 251–​289. Dancy, J. 1997. “Parfit and Indirectly Self-​defeating Theories” in J. Dancy (ed.), Reading Parfit (Oxford: Blackwell Publishers), pp. 1–​23. de Lazari-​Radek, K., and P. Singer. 2010. “Secrecy in Consequentialism:  A Defence of Esoteric Morality.” Ratio vol. 23, no. 1 (March), pp. 34–​58.

Toward a unified theory of morality  29

Eggleston, B. 2000. “Should Consequentialists Make Parfit’s Second Mistake? A Refutation of Jackson.” Australasian Journal of Philosophy vol. 78, no. 1 (March), pp. 1–​15. Eggleston, B. 2003. “Does Participation Matter? An Inconsistency in Parfit’s Moral Mathematics.” Utilitas vol. 15, no. 1 (March), pp. 92–​105. Eggleston, B. 2013a. “Paradox of Happiness” in The International Encyclopedia of Ethics, edited by H. LaFollette (Blackwell Publishing Ltd., 2013), pp. 3794–​3799. DOI:  10.1002/​ 9781444367072.wbiee202. Eggleston, B. 2013b. “Rejecting the Publicity Condition:  The Inevitability of Esoteric Morality.” The Philosophical Quarterly vol. 63, no. 250 (January), pp. 29–​57. Glover, J., and M. J. Scott-​Taggart. 1975. “It Makes No Difference Whether or Not I Do It.” Proceedings of the Aristotelian Society, Supplementary Volumes, vol. 49, pp. 171–​209. Gruzalski, B. 1986a.“Parfit’s Impact on Utilitarianism.” Ethics vol. 96, no. 4 (July), pp. 760–​783. Gruzalski, B. 1986b. “Parfit’s Unified Theory of Morality.” Philosophical Studies vol. 50, no. 1 (July), pp. 143–​152. Jackson, F. 1997. “Which Effects?” in J. Dancy (ed.), Reading Parfit (Oxford:  Blackwell Publishers), pp. 42–​53. Kuflik, A. 1986. “A Defense of Common-​Sense Morality.” Ethics vol. 96, no. 4 (July), pp. 784–​803. MacFarquhar, L. 2011. “How to Be Good.” The New Yorker, 5 September, p. 42–​53. Mendola, J. 1986. “Parfit on Directly Collectively Self-​ Defeating Moral Theories.” Philosophical Studies vol. 50, no. 1 (July), pp. 153–​166. Parfit, D. 1984/​87. Reasons and Persons, “Reprinted with further corrections 1987” edition (Oxford: Clarendon Press). Parfit, D. 1986. “Comments.” Ethics vol. 96, no. 4 (July), pp. 832–​872. Parfit, D. 2011. On What Matters (Oxford: Clarendon Press). Petersson, B. 2004. “The Second Mistake in Moral Mathematics is not about the Worth of Mere Participation.” Utilitas vol. 16, no. 3 (November), pp. 288–​315. Russell, B., and P. Russell. 1937. The Amberley Papers:  The Letters and Diaries of Bertrand Russell’s Parents. New York: W. W. Norton & Company, Inc. Shafer-​ Landau, R. 2018. The Fundamentals of Ethics, 4th edition. Oxford:  Oxford University Press.

2 INTRODUCTION TO PART TWO Rationality and time Brian Hedden

2.1  Introduction Parfit describes his task in Part Two of Reasons and Persons as a sustained attack on S, the Self-​interest Theory. S makes its first appearance on the first page of the book, where Parfit says that ‘S gives to each person this aim: the outcomes that would be best for himself, and that would make his life go, for him, as well as possible’ (3). Restricting ourselves to evaluation of acts, S says that a person ought to perform act φ only if there is no alternative act ψ such that his life would go better, for him, if he were to ψ than if he were to φ.1 It bears emphasizing that S does not say that people ought to be selfish or egotistical, at least as we ordinarily understand these characteristics. One’s life might go best by having deep friendships, loving one’s children, and giving to charity. This can be true, albeit for different reasons, on any of the three kinds of theory of well-​ being that Parfit considers, namely hedonistic theories, desire-​fulfilment theories, and objective list theories (Parfit discusses these theories in depth in Appendix I). Parfit advances three main objections against S. First, it can be rational to care most about things other than one’s overall lifetime well-​being, such as doing one’s duty or achieving some goal, and so, contra S, it can be rational to perform some act that will not maximize one’s overall lifetime well-​being. Second, S is an objectionably hybrid view in that it is agent-​relative but time-​neutral. It is agent-​relative in that it gives to each agent a distinct aim, namely that his life goes as well as possible. But it is time-​neutral in that it prohibits privileging any temporal part of his life over any other. Insofar as well-​being can accrue to temporal parts of a life, the well-​being of each of those temporal parts is to count equally. In his ‘Appeal to Full Relativity,’ Parfit argues that, unlike S, a theory should be either fully neutral or fully relative. Third, time-​neutrality may not be rationally required. It may be rational to care more about some parts of one’s life than about others.

Rationality and time  31

I will examine each of these arguments in what follows. But first, we must look at the theory that constitutes the main foil for S, namely P, the Present Aim Theory.

2.2  The Critical Present Aim Theory and intrinsically irrational desires The Present Aim Theory, or P, ‘tells each to do what will best achieve his present aims’ (92). (Parfit also distinguishes various versions of P at the outset of Part Two; we will soon consider the version he favours.) P differs from S by being both agent-​relative and time-​relative, whereas S is agent-​relative but time-​neutral. This difference plays a key role in Parfit’s arguments against S, especially his Appeal to Full Relativity. But there is another important difference between P and S. S is about maximizing well-​being, whereas P is about achieving aims. Thus P and S are not a ‘minimal pair,’ differing only with respect to time-​neutrality. A theory that differed from S only by being time-​relative where S is time-​neutral would say that an agent A at time t ought to perform the act that will maximize the well-​being of A-​at-​t. This view may be implausible for two reasons: first, perhaps time-​slices cannot themselves have levels of well-​being (this is an issue we will revisit in section 2.4), and second, perhaps how A acts at t cannot affect the well-​being of A-​at-​t, but only that of A-​at-​t+E and successive time-​slices (this is an issue that will come up in section 2.5 in a slightly different form). A theory that differed from P only by being time-​neutral where P is time-​ relative would say that agent A at time t ought to perform the act that will best achieve all of the aims that A has throughout his life, not just those that A has at time t (Kagan 1986).This theory runs into trouble in cases where what A does at t affects what aims A will later have. Suppose that if I now enrol in business school, I will acquire the aim of making lots of money, while if I now enrol in a philosophy PhD programme, I will acquire the aim of writing a philosophy book. Setting aside all other aims I might have, our time-​neutral analogue of P says that what I ought to do depends on what I will in fact do. If I will in fact enrol in business school, then I will in fact later have the aim of making lots of money, and so I ought to enrol in business school, since this will best achieve the aims I have over the course of my life. Similarly, if I will in fact enrol in a philosophy PhD programme, then I will in fact later have the aim of writing a philosophy, an aim which I will now best help achieve by enrolling in a philosophy PhD programme. So, our time-​neutral analogue of P says that I ought to enrol in the philosophy PhD programme. Each of my possible decisions is thus self-​reinforcing in such a case. Worse, suppose that if I enrol in business school, I will acquire the aim of writing a philosophy book, while if I enrol in a philosophy PhD programme, I will acquire the aim of getting rich. Then, our time-​neutral analogue of P says that whichever of these actions I will in fact perform, I ought to perform the other. Each of my possible decisions is self-​frustrating in this case. Thus, our time-​neutral analogue of P has the odd implication that what one ought to do at time t can depend on what one will in

32  Brian Hedden

fact do at time t. This implication is certainly odd. It is a further question whether it constitutes an objection to the theory; see Hare and Hedden (2016) for discussion. I will follow Parfit in continuing to focus on S and P, but it is important to keep in mind that they are not a minimal pair, but rather differ along two dimensions, namely time-​ neutrality vs. time-​ relativity and well-​ being maximization vs. aim-​achievement. Parfit’s preferred version of P is the Critical Present Aim Theory, or CP. Parfit argues that some desires are intrinsically irrational, while others are (intrinsically) rationally required. Intrinsically irrational desires do not provide reasons for action. And while he is not explicit on this point, he presumably thinks that if one is rationally required to have some desire, then one has a reason to try to make true the content of that desire regardless of whether or not one in fact has that desire. Then, CP says that agent A at time t ought to perform the act that will best satisfy the set of desires whose members include the rationally permissible desires that A has at t as well as any rationally required desires that A lacks at t. In claiming that some desires are intrinsically irrational, while others are rationally required, Parfit is opposing Hume, who is commonly interpreted as holding that desires cannot be irrational unless they are based on a false (or better: irrational) belief. One’s ultimate ends cannot be criticized as irrational. What about Parfit’s argument for his anti-​Humean position? The first thing to note is that he does not actually argue for the second part of his claim, that some desires are (intrinsically) rationally required. He only argues for the claim that some desires are intrinsically irrational. Still, if one is convinced that some desires are intrinsically irrational, this should at least lessen one’s resistance to the thought that some desires are rationally required. Parfit’s argument that some desires are intrinsically irrational is based on examples, the most famous of which is the man with Future Tuesday Indifference: A certain hedonist cares greatly about the quality of his future experiences. With one exception, he cares equally about all the parts of his future. The exception is that he has Future-​ Tuesday-​ Indifference. Throughout every Tuesday he cares in the normal way about what is happening to him. But he never cares about possible pains or pleasures on a future Tuesday. Thus he would choose a painful operation on the following Tuesday rather than a much less painful operation on the following Wednesday. This choice would not be the result of any false beliefs. This man knows that the operation will be much more painful if it is on Tuesday. Nor does he have false beliefs about personal identity. He agrees that it will be just as much him who will be suffering on Tuesday. Nor does he have false beliefs about time. He knows that Tuesday is merely part of a conventional calendar, with an arbitrary name taken from a false religion. Nor has he any other beliefs that might help to justify his indifference to pain on future Tuesdays. This indifference is a bare

Rationality and time  33

fact.When he is planning his future, it is simply true that he always prefers the prospect of great suffering on a Tuesday to the mildest pain on any other day. 123–​124 Considered as a mere intuition pump, Future Tuesday Indifference is compelling but may not add much to the debate. After all, Hume himself gave an example of an intuitively irrational preference—​that of the man who preferred the destruction of the world to the scratching of his finger—​but insisted that the preference was not in fact irrational. And Rawls (1971) gives the example of someone whose main goal in life is to count the blades of grass in his yard. But Parfit’s key point is that Future Tuesday Indifference is arbitrary. It involves drawing a sharp line between cases that are similar in all respects worth caring about. The same is true of his other examples, Bias Towards the Next Year and Within-​A-​Mile Altruism. In arguing that such arbitrariness is irrational, Parfit thus provides more of an argument than we get from just considering intuition pumps like those of Hume and Rawls.2

2.3  The ‘best objection’ to S In criticizing S, Parfit leads with what he calls his ‘best objection’ to the theory. He argues that according to S, each person should be ‘governed by the desire that his life goes, for him, as well as possible,’ regardless of the costs to others (131). He refers to this desire as the bias in one’s own favour. And he argues that the bias in one’s own favour is not supremely rational; it can be rational to have other desires that are as strong as, or stronger than, the bias in one’s own favour. He considers two sorts of desires that might lead one to act in a way that violates S and that, he claims, are no less rational than the bias in one’s own favour. First are moral desires. One might desire to sacrifice one’s life in order to save several other people, even though this self-​sacrifice would make one’s own life go worse overall.3 Second are desires for achievement. One might desire to write a great novel, even knowing that one will thereby make one’s life go worse, given the stress, uncertainty, and self-​doubts that one will suffer while writing. Parfit claims, plausibly, that these desires are no less rational than the bias in one’s own favour, and that it would not be irrational for one to act on these desires, even in cases where doing so would make one’s own life go worse overall. Kagan (1986) objects to Parfit’s argument. On his way of reading Parfit, Parfit is objecting to S on the grounds that it ‘elevates one particular pattern [of concern]—​ the bias in one’s own favor—​and gives it a unique theoretical status,’ and S is therefore ‘intolerant in its attitude toward the rationality of different patterns of concern’ (750). But, he argues, P also elevates a particular pattern of concern—​we might call it the bias in favour of one’s present desires—​and gives it a unique theoretical status. The situation, as Kagan sees it, is this: Both S and P (and indeed any theory of what one ought to do) elevate some particular pattern of concern (a ‘metadesire,’

34  Brian Hedden

as Kagan terms it) to unique theoretical status. For S, this is the bias in one’s own favour. For P, this is the bias in favour of one’s present desires. But neither needs to deem other particular desires, such as moral desires, or desires for achievement, to be irrational.Thus, for Kagan, S and P are on a par, and Parfit’s ‘best objection’ gives no grounds for preferring P to S. As I see it, though, Kagan misinterprets Parfit’s objection. Parfit is not objecting to the mere fact that S elevates some pattern of concern or other to unique theoretical status. He is objecting to the fact that it elevates this particular pattern of concern—​ the bias in one’s own favour—​to this unique theoretical status. He thinks that it can be rational to act on certain desires, like moral desires and desires for achievement, even when doing so makes one’s life go worse overall. If this is so, then S is false.4 For S and P to be on a par in this respect, we would need to make an analogous case against P. We would need a case where performing some action is rational even though this action will not best satisfy one’s present desires, and even though one’s present desires are all rationally permissible and include all rationally required desires (since we are considering the Critical version of P).5 It is not clear what such a case would look like. And absent such a case, Parfit can claim that S and P are not on a par.

2.4  The appeal to full relativity We noted earlier that S combines agent-​relativity with time-​neutrality. Parfit objects to this structure in his Appeal to Full Relativity.6 Theories that combine agent-​ neutrality with time-​neutrality, or agent-​relativity with time-​relativity, are pure. They are either fully neutral, or fully relative. Examples of fully neutral theories include Consequentialism (C)  and the agent-​neutral modification of Common Sense Morality (N) that Parfit considers toward the end of Part One. The Present Aim Theory is a fully relative theory. By contrast, S is a hybrid theory, being neither fully neutral nor fully relative. It is incompletely relative. Parfit suggests that by virtue of this hybrid, incompletely relative structure, ‘S can be charged with a kind of inconsistency’ (140). But it is not clear why exactly incomplete relativity constitutes a kind of inconsistency. As Kagan (1986) notes, the most Parfit seems to do to argue that theories should be either fully neutral or fully relative is to note a formal analogy between personhood and time. The words ‘I’ and ‘now’ are both indexicals. The way in which the word ‘I’ picks out a person is similar to the way in which the word ‘now’ picks out a time. An utterance of ‘I’ refers to the speaker of the utterance, and the word ‘now’ refers to the time of the utterance. This is just a linguistic point, however. Parfit adds that ‘When each of us is deciding what to do, he is asking, “What am I to do now?” ’ (140), and concludes that a theory of rationality should treat ‘I’ and ‘now’ in the same way. This formal analogy is not entirely convincing. But let me first point out that Parfit has overlooked a way in which even C and N are not fully neutral. Just as there is a formal analogy between ‘I’ and ‘now,’ so there is a formal analogy between these two and ‘actual.’ ‘Actual’ is also an indexical, utterances of which refer to the possible world of the utterance. And when each of us is deciding what to do, he is

Rationality and time  35

asking not just ‘What am I to do?’, and not just ‘What am I to do now?’, but ‘What am I actually to do now?’ So one could argue that even C and N are incompletely relative. A fully neutral theory must be not only agent-​neutral and time-​neutral, but also world-​neutral. Such a theory would be implausible in the extreme. A genuinely fully neutral version of Utilitarianism, for instance, would say that we ought to act so as to maximize the sum-​total of happiness across all people, times, and possible worlds. But this sum-​total of happiness is fixed regardless of what we do (if we increase happiness in our world, we thereby decrease it in another, simply by making the former rather than the latter actual), and so it would say that all acts are on a par. Of course, if genuine full neutrality is absurd, this might just provide support specifically for P (and other fully relative theories). But it might also make us doubt Parfit’s contention that incomplete relativity is objectionable. Let us set aside modality and world-​neutrality. Why should the mere existence of a formal analogy between ‘I’ and ‘now’ mean that they should be treated alike by a theory of rationality? Perhaps simplicity and elegance are theoretical virtues, in normative theorizing as in scientific theorizing. And pure theories are simpler and more elegant than hybrid ones.This yields a pro tanto reason for preferring pure theories over hybrid ones. Thus, other things being equal, we should opt for a pure theory over a hybrid one.This, I think, is the most plausible interpretation of Parfit’s Appeal to Full Relativity.7 However, in Part Three Parfit provides further support for favouring pure theories over hybrid ones by arguing for reductionism about personal identity over time, according to which facts about personal identity over time are neither metaphysically deep nor normatively significant. But it also makes the Appeal weak, in that an opponent could respond that other things are not equal. Perhaps personhood and time are different in sufficiently important respects as to outweigh the pro tanto reason for preferring pure theories over hybrid ones. Brink (2011) responds along these lines, invoking the notion of compensation to break the analogy between ‘I’ and ‘now.’ If one person is harmed for the sake of providing a greater benefit to another person, the first person is not thereby compensated for the harm. This is the upshot of the so-​called separateness of persons emphasized by Rawls (1971) and Nozick (1974). Nor is there any larger entity, like the mereological sum of the two persons, that is both benefactor and beneficiary, and thus automatically compensated for the harm. As Nozick (1974, 32–​3) writes, ‘there is no social entity with a good that undergoes some sacrifice for its own good.’ By contrast, when one person is harmed at an earlier time for the sake of providing a greater benefit to the same person at a later time, the person is thereby automatically compensated for the earlier harm.8 According to Brink, this fact—​ that compensation is automatic when benefactor and beneficiary are the same person, but not when they are different persons—​justifies the hybrid structure of S. Brink addresses an important objection to his suggestion. Might we go further than appealing to the separateness of persons and insist on the ‘separateness of different periods within a person’s life’ (Brink 2011, 364)? We could then claim

36  Brian Hedden

that ‘me-​now is [no] more compensated for its sacrifices on behalf of me-​later than I am compensated by my sacrifices for you’ (ibid.). Compensation would then be no more automatic in cases of intrapersonal tradeoffs than in interpersonal ones. Brink’s own response is unsatisfactory in my view. Here is what he says: [T]‌his challenge to temporal neutrality requires thinking that we can and should adopt a sub-​personal perspective when reckoning compensation. But there are problems with this idea. Once we go sub-​personal and appeal to full relativity, there seems no reason to stop until we reach the sub-​personal limit—​a momentary time slice of the person. But notions of compensation have no application to momentary time slices, which do not persist long enough to act or receive the benefits of earlier actions. Moreover, many of the goods in life, especially the pursuit and achievement of worthwhile projects, seem to be realized only by temporally extended beings. 364 I agree with Brink that if we go sub-​personal in thinking about compensation, it would be arbitrary to stop short of the limiting case—​momentary time-​slices. But he is mistaken in thinking notions of compensation cannot be applied to such time-​ slices. First, he says that these momentary time-​slices do not persist long enough to act. But while time-​slices are not sufficiently long-​lived to perform physical actions, it is less clear that they cannot perform mental actions like making decisions and forming intentions. More to the point, it seems that what is crucial is whether a given entity can receive compensation, and it is unclear why this should require the ability to act. Second, he claims that many goods can be ‘realized only by temporally extended beings.’ But there is an ambiguity here.We must distinguish between which entities suffice to cause some good to exist and to which entities that good (understood as a unit of well-​being) accrues. There are many goods that cannot be realized by a single person, such as friendship. But this does not prevent the corresponding good—​the well-​being associated with engaging in friendship—​from accruing to each individual friend. Similarly, it may be that there are goods that require for their creation the existence and cooperation of many successive time-​slices. But that does not prevent the corresponding well-​being from accruing to each of those time-​ slices. And it is worth noting that none of the major theories of well-​being seem to entail that well-​being cannot accrue to time-​slices. Time-​slices can presumably have both phenomenal states and desires (meaning that they can have levels of well-​ being according to hedonistic and desire-​fulfilment theories), and they can have at least some of the items mentioned by objective list theorists, such as knowledge and health. So I  think that Brink is incorrect in thinking that notions of compensation make no sense when applied to momentary time-​slices. But a better response to the objection is available. The compensation theorist should insist that what is at issue, when one entity is harmed so as to give a greater benefit to another, is whether

Rationality and time  37

they comprise some larger entity that is both benefactor and beneficiary, and so is automatically compensated for the harm. Recall that, in arguing that it is impermissible to harm one person for the sake of benefiting another, Nozick emphasizes that ‘there is no social entity with a good that undergoes some sacrifice for its own good’ (see also Gauthier 1962, 126). The implication is that if there were such a social entity with the two persons as parts, it would be permissible to harm the one person to benefit the other, for the social entity would be both benefactor and beneficiary; this is so even though the one person is not compensated by the benefit gained by the other. If that is correct, then it is irrelevant, for Brink’s purposes, that me-​now is not compensated when it is harmed so as to provide a greater benefit for me-​later. What matters is rather that me-​now and me-​later are parts of a larger entity—​me—​ which is compensated for the early harm by the later benefit. (Of course, we might well deny that the existence of a larger entity, which is compensated when one of its parts is harmed for the sake of benefiting another, makes this harm permissible.This would make Nozick’s rejection of a social entity a non sequitur and would resuscitate the ‘separateness of time-​slices’ objection to Brink’s compensation argument.) Where does that leave us? Suppose we reject this separateness of time-​slices objection but take seriously the separateness of persons and the importance of compensation. In my view, this may still not suffice to justify S. First, there may be agent-​neutral theories that prohibit harming one person just for the sake of providing a greater benefit to another (or, more generally, for the sake of maximizing the good). A consequentialism of rights (of which Parfit’s N, an agent-​neutral modification of Common Sense Morality, may be an example) might have this implication. It would allow one agent to be harmed, in violation of his rights, only for the sake of reducing the overall number (and magnitude) of rights violations that occur. Second, the appeal to compensation may not justify time-​neutrality as a rational requirement. Lack of compensation may make it impermissible to impose a harm on someone for the sake of benefiting another (without the former’s consent), but it is far from clear that the presence of compensation can make it rationally required for one to undergo a harm for the sake of a greater benefit. A person might grant that it is him who will be compensated if he undergoes some harm now for some greater benefit later, but insist that it would nonetheless not be irrational for him to decline to do so. He might say that while the fact that he will be compensated for the harm makes it permissible to undergo the harm, it does not make it rationally required. And time-​relative theories can easily grant that doing so would be permissible. So while Brink has identified a potentially important disanalogy between ‘I’ and ‘now,’ facts about when compensation is automatic and when it isn’t do not clearly justify either agent-​relativity or time-​neutrality.

2.5  Time-​bias Amid the myriad fascinating observations about our attitudes to time, Chapter 8 includes perhaps the most powerful of Parfit’s arguments against S. S gains some of its plausibility from its condemnation of what Parfit calls the bias toward the near,

38  Brian Hedden

whereby we care more about our near futures than our far futures and are willing to trade a larger benefit in the farther future for a smaller one in the nearer future, and to trade a smaller harm in the near future for a larger one later on. This form of time-​bias is apt to strike us as irrational, and it has been widely condemned by philosophers. If we accept that the bias toward the near is irrational, this lends some support to S’s claim that time-​neutrality is a requirement of rationality. But Parfit notes that there is another form of time-​bias as well, namely bias toward the future. We prefer that our pleasures be in the future and our pains in the past, even if this means a worse overall balance of pleasure and pain throughout our lives. This time-​bias is illustrated in the case My Past or Future Operations: I am in some hospital, to have some kind of surgery. Since this is completely safe, and always successful, I have no fears about the effects. The surgery may be brief, or it may instead take a long time. Because I have to co-​operate with the surgeon, I cannot have anaesthetics. I have had this surgery once before, and I can remember how painful it is. Under a new policy, because the operation is so painful, patients are now afterwards made to forget it. Some drug removes their memories of the last few hours. I have just woken up. I cannot remember going to sleep. I ask my nurse if it has been decided when my operation is to be, and how long it must take. She says that she knows the facts about both me and another patient, but that she cannot remember which facts apply to whom. She can tell me only that the following is true. I may be the patient who had his operation yesterday. In that case, my operation was the longest ever performed, lasting ten hours. I may instead be the patient who is to have a short operation later today. It is either true that I did suffer for ten hours, or true that I shall suffer for one hour. I ask the nurse to find out which is true. While she is away, it is clear to me which I prefer to be true. If I learn that the first is true, I shall be greatly relieved. 165–​166 Parfit thinks not only that most (all?) of us have the bias toward the future, but that most of us think that it is rational. Indeed, many of us probably think that it is not just rationally permissible, but rationally required. (Parfit later argues that it would be better for us if we lacked the bias toward the future, but he notes, correctly, that an attitude can be rational even if it is bad for us.) Clearly, if the bias toward the future is rational, this is a problem for S.  But Parfit’s discussion of exactly whether it is a problem is confusing. Here is how Parfit seems to view the matter (see p. 153, 157, and 163): The rationality of bias toward the future does not entail the falsity of S. Rather, it undercuts one motivation for S. The S-​theorist must claim that we ought to give equal weight to the well-​being of all our present and future time-​slices. But he cannot support this claim by appeal to time-​neutrality. He cannot claim that a mere difference in timing has no rational significance. After all, time-​neutrality entails the irrationality of bias toward the

Rationality and time  39

future. And so the S-​theorist must find some other motivation for the claim that it is irrational not to give equal weight to the well-​being of all our present and future time-​slices. In a nutshell, Parfit sees S as committed to the irrationality of the bias toward the near, but not to the irrationality of the bias toward the future. But it is prima facie difficult to motivate the claim that the bias toward the near, but not the bias toward the future, is irrational (we will shortly consider a few attempts to do so). In my view, Parfit is understating his case. If the bias toward the future is rational, then this simply entails that S is false. Recall that ‘S gives to each person this aim: the outcomes that would be best for himself, and that would make his life go, for him, as well as possible’ (3). S does not say that each person should aim for the outcomes that would be best for his present and future selves, and that would make his life go, for him, as well as possible in the future. It says that each person should aim for the outcomes that would be best for himself simpliciter, and that would make his life simpliciter go, for him, as well as possible. There is nothing in the statement of S that allows it to acknowledge the rational permissibility of bias toward the future. Thus, Parfit should instead claim that the rationality of bias toward the future decisively refutes S, but that nonetheless one previously sympathetic to S could switch to defend a future-​directed modification of S, S*. But S* will be difficult to defend, since the S*-​theorist cannot condemn the bias toward the near by invoking full-​blown time-​neutrality, which would likewise implausibly condemn the bias toward the future. What might the S*-​theorist say in response? Is it possible to narrowly target the bias toward the near without overshooting and condemning the bias toward the future? Not obviously. Consider some ways to defend the bias toward the future. First consider control. The past is outside our control (we will shortly question this claim), while the future is at least somewhat within our control. But this is a bad justification for bias toward the future. It’s not clear why we should prefer that our pains be in the past and hence outside our control, and moreover, as Parfit notes (p. 168), we are not unconcerned about future pains that are outside our control in the way that we are unconcerned about past pains. Worse, even if considerations of control could justify the bias toward the future, they might also justify the bias toward the near, since we often have more control over the near future than the far future. Second, consider epistemology. We have different, and usually better, epistemic access to facts about the past than to facts about the future. But again, it is not clear why this would justify the bias toward the future. Why should we prefer that our pains be such that we are in a better position to know about them, and our pleasures to be mired in uncertainty (Hare 2013)? Moreover, this consideration (if it were a good one) would implausibly justify the mirror image of the bias toward the near. Since we typically have better epistemic access to facts about the near future than about the far future, we would be led to prefer that our pains be in the near future and our pleasures in the far future, rather than vice versa.

40  Brian Hedden

Third, consider the metaphysics of time. Parfit suggests that belief in the genuine passage of time, or ‘objective becoming,’ might help justify the bias toward the future. Without going too far into the details, he seems to be referring to the A-​ theory, on which properties like being past, being present, and being future are not reducible to relations like being earlier than, being simultaneous with, and being later than. On one version of the A-​theory, now is like a ‘moving spotlight’ that illuminates different times with a special property—​now-​ness—​as time passes. But as Parfit notes (p. 180), it is not clear what the argument would be from the truth of the A-​theory to the rationality of bias toward the future. Without such an argument, it is impossible to say whether the A-​theory, if it supports the bias toward the future, would also support the bias toward the near.9 And it is worth noting that popular ways of fleshing out the A-​theory do not seem to support bias toward the future. Presentism, the view that only present things exist, would seem to support an absolute bias toward the present if it supports any form of time-​bias at all. The growing block view, on which past and present things exist, with the block universe growing and adding more and more time-​slices as time passes, would seem to support a bias toward the past if it supports any form of time-​bias at all. As Hare (2013) notes, the only sort of A-​theory that seems to have any chance of supporting bias toward the future is the shrinking block view, on which present and future things exist, so that the block universe sheds time-​slices and thereby shrinks as time passes. But the shrinking block view is believed by almost no one. Parfit concedes that the S*-​theorist could insist that ‘in appealing to time’s passage, we do not need arguments’ and that it could be a ‘fundamental truth that, since time passes, past suffering simply cannot matter—​cannot be an object of rational concern’ (180–​1). This is of course true. One can always insist that some facts are brute. But such an appeal to brute facts would constitute a somewhat unattractive feature of S*. Perhaps it was a mistake to abandon S in favour of S*. Instead, the S-​theorist should stick to his guns and say that the bias toward the future, like the bias toward the near, is irrational. He could soften the blow by saying that the bias toward the future is harmless, since it can’t be acted upon. And that helps explain why we mistakenly think that it is rationally permissible. (It would hardly explain why we think it is rationally required, however.) Alternatively, but equivalently for present purposes, he could restrict S to a claim about rational action rather than about rational aims and preferences; then, since the bias toward the future can’t be acted upon, the S-​ theorist need neither condemn nor approve of it. But it is false that the bias toward the future can’t be acted upon. Parfit observes that on a desire-​fulfilment theory of well-​being, we can affect the well-​being of our past selves (and of past people generally). If, in the past, I had a desire to go skydiving at some point in my life, I can fulfil this desire, and boost the well-​being of my past self, by going skydiving. And he argues, persuasively, that most of us are biased toward the future with respect to desire-​fulfilment. We think that we may have no reason to try to fulfil such mere past desires.

Rationality and time  41

Parfit (p. 163) suggests that the S-​theorist could maintain that the bias toward the future cannot guide action by rejecting desire-​fulfilment in favour of a hedonistic or objective list theory of well-​being. But even here matters are complicated. Some of the items typically mentioned by objective list theorists are such that we can affect whether they were had by past selves. For starters, desire-​fulfilment itself often appears on such lists. Consider also accomplishment. If my earlier self laboured in obscurity to try to write a great novel, I can make it the case that my earlier self ’s labours were in service of a genuine accomplishment, namely by finishing the novel and getting it published. Next consider knowledge. Suppose my past self had a justified belief in P. I may be able to make it the case that this past self had knowledge, and not mere justified false belief, by making P true. Only if the S-​theorist strikes items like desire-​fulfilment, accomplishment, and knowledge from his objective list can he maintain, on an objective list view, that the bias toward the future cannot guide action. It may seem obvious that on a hedonistic view, the bias toward the future is practically inert, since we cannot causally affect the happiness of past selves.This is true at least on phenomenal conceptions of happiness. But if we consider decision-​making under uncertainty, the bias toward the future may nonetheless be action-​guiding. Dougherty (2011) argues that if we are both biased toward the future and risk averse, then this can affect how we act. For instance, consider a modification of Parfit’s case from above: A coin was flipped to determine which of two surgery regimes you will undergo. If it landed heads, you will undergo the Early Course—​4 hours of painful surgery on Tuesday and 1 hour of painful surgery on Thursday. If it landed tails, you will undergo the Late Course—​no surgery on Tuesday and 3 hours of painful surgery on Thursday. It is now Wednesday (and you know this), but you have been given selective amnesia whereby you don’t remember whether you had surgery on Tuesday.You are offered a gamble: If you accept the gamble, then if you are in the Early Course, your Thursday surgery will be extended by 30 minutes, while if you are in the Late Course, your Thursday surgery will be shortened by 30 minutes. If you are risk averse, you will prefer options that better your worst-​case scenario while worsening your best-​case scenario, at least if doing so makes no difference to your expected utility. Now, if you are risk averse in this way and also future-​biased, then you will accept the gamble, while if you are risk averse but time-​neutral, you will decline it. This is because what you deem your worst-​and best-​case scenarios differs depending on whether or not you are future-​biased. If you are future-​ biased, your worst-​case scenario is being in the Late Course, since that involves the most future pain. And accepting the gamble betters that worst-​case scenario while worsening your best-​case scenario, without affecting your expected total or future pain at all. So, being risk averse as well, you will accept the gamble. But if you are time-​neutral, then your worst-​case scenario is being in the Early Course, since that

42  Brian Hedden

involves more total pain (albeit less future pain). And accepting the gamble will worsen that worst-​case scenario while bettering your best-​case scenario, without affecting your expected total or future pain at all. So, being risk averse, you will decline the gamble. Thus, we have a case where, assuming you are risk averse, how it is rational for you to act depends on whether or not you are future-​biased. Thus, it seems future-​ bias can be action-​guiding even on a hedonistic view of well-​being. Greene and Sullivan (2015) object, arguing that we aren’t risk averse with respect to pleasures and pains in the standard sense of having diminishing marginal utility for pleasures and pains. It is not the case that each additional unit of pleasure or pain matters less to us than the one that came before, as is the case with dollars, say. Note, however, that if we adopt Buchak’s (2013) Risk-​Weighted Expected Utility Theory, then Dougherty’s argument goes through. Similarly, if we accept evidential decision theory, then the bias toward the future can be action-​guiding even given hedonism and even without any risk aversion. Evidential decision theory says that you should evaluate actions by looking at what evidence the performance of those actions would constitute for how good or bad things will be with respect to whatever you think matters (represented by a utility function). Formally, it says that you should perform the action with highest evidential expected utility, defined thus: EEU (A) = ∑i P (Oi | A) × U (Oi) where A is an action, the Oi are all the possible outcomes, P (Oi | A) is your credence that Oi is true, given that you perform A, and U (Oi) is your utility for Oi. Now suppose that there are two buttons in front of you.You’re told that people who had lots of pain in their childhood (since forgotten) tend to push the left button, while people who had lots of pleasure in their childhood tend to push the right button. Pushing the right button thus gives one strong evidence that one had a pleasure-​filled childhood, while pushing the left button gives one strong evidence that one had a pain-​filled childhood. (Neither button is such that pushing it gives one evidence about how things will be in the future.) Therefore, whether evidential decision theory tells you to push the left button or the right one depends on whether or not you are biased toward the future. If you are, then it says that you can push either button. But if you are not biased toward the future, then it says to push the right one, since pushing the right button gives you strong evidence that things were good for you in the past (which, being non-​future-​biased, you care about) and gives no evidence either way about how things will be for you in the future. Thus, given evidential decision theory and hedonism, whether or not you are biased toward the future can make a difference to how you ought to act. Thus, in order for the S-​theorist to defend the claim that the bias toward the future cannot be acted upon, he must endorse both (i) hedonism or an objective list theory that doesn’t include things like desire-​fulfilment, accomplishment, or

Rationality and time  43

knowledge, and (ii) non-​risk-​weighted causal decision theory. This is not necessarily a bad consequence; the latter is perhaps the orthodox decision theory, and the former are not implausible theories of well-​being. Nevertheless, it is important to be clear on what the S-​theorist must commit to in order to claim that the bias toward the future cannot be acted upon. We have been considering the option of the S-​theorist maintaining, against our intuitions, that the bias toward the future is irrational, but softening the blow by claiming that it is practically inert and hence harmless. But we have seen that it is difficult to defend the claim that bias toward the future is in fact practically inert. This defeats the S-​theorist’s attempt to soften the blow that comes with claiming that bias toward the future is irrational. But actually, the fact that the bias toward the future can be acted upon may actually help the S-​theorist’s case.This is because a policy of acting on one’s bias toward the future can be predictably disadvantageous. In the article mentioned above, Dougherty (2011) shows that if one is biased toward the future and also risk averse, then one will sometimes perform a sequence of actions that is predictably disadvantageous in the following sense: it is foreseeable that at the time each member of that sequence is available, one will prefer to perform it rather than not, but at all times, one will prefer performing no member of that sequence over performing all of them.This is the same sort of diachronic inconsistency that is involved in violations of other purported requirements of rationality, such as transitivity of preferences and Bayesian Conditionalization. The details of the case are somewhat intricate, so I relegate an overview to a footnote.10 For other arguments that bias toward the future is irrational because of the actions it recommends, see Greene and Sullivan (2015) and Dougherty (2015). Let us step back. Common sense suggests that bias toward the near is irrational while bias toward the future is rational. This constitutes a prima facie problem for the S-​theorist, who must condemn both. It seems to me that the best approach for the S-​theorist is to argue that bias toward the near and the bias toward the future are in fact both irrational, and moreover irrational for the same reason, namely that they sometimes recommend courses of action that predictably make one’s life as a whole go worse. This is a powerful argument. But I  do not think it is fully convincing. First, attitudes can be predictably disadvantageous despite being rational, or predictably advantageous despite being irrational; exaggerated self-​confidence may be greatly beneficial, but no less irrational for it. Indeed, Parfit himself argues that we would be better off if we were not biased toward the future, since it would yield more equanimity in the face of death (see Section 74). But he stops short of condemning the bias toward the future as thereby irrational. Second, in arguing that bias toward the future (and bias toward the near) are irrational because they can predictably make one’s life as a whole go worse, the S-​ theorist is begging the question against his opponent. An opponent who does not think that one must be concerned principally for how one’s life goes on the whole,

44  Brian Hedden

and who cares more about some parts of his life than others, will not be convinced. And a defender of full neutrality will agree with the S-​theorist’s condemnation of bias toward the future on the grounds that it can predictably make one’s life go worse overall. But he will go further and condemn the bias in one’s own favour on analogous grounds. For example, if two people are both biased in their own favour, then they will sometimes be led to perform a combination of actions that is predictably bad for both, in the sense that each would be worse off if each performs his part of that combination of actions than if neither does. The Prisoner’s Dilemma is one such example. See Parfit (1984, 187–​191) and Hedden (2015, Ch. 7) for further discussion.

2.6  Conclusion We have looked at Parfit’s main arguments in Part Two against the Self-​interest Theory, S. In my view, his arguments are, on the whole, successful, despite various troubles. But Part Two contains a multitude of observations about our attitudes to time that are fascinating and important independently of the debate over S, for instance about how our attitudes about the importance of past desires depend on whether they were based on deeply held values or mere tastes, about how our other-​directed time-​biases depend on our proximity to, and ability to communicate with, our loved ones. And his discussion of the relationship between rationality and time continues in Part Three, where he discusses the implications that the metaphysics of personal identity over time has for the rational permissibility of time-​bias.

Notes 1 Parfit uses masculine pronouns in stating such principles. Rather than employ generic or feminine pronouns, I will follow Parfit’s practice for the sake of avoiding confusion. 2 See Street (2009) for critical discussion of all these ‘ideally coherent eccentrics.’ See also Broome (1991) for another prominent argument againt Humeanism about preference. Broome argues that some substantive constraints on preferences are required if the formal constraints of transitivity and the like are to have any bite. For criticism of Broome’s argument, see Dreier (1996). 3 This self-​sacrifice might make one’s life go better in some respects. Perhaps if one did not sacrifice oneself, one would suffer unpleasant feelings of guilt or shame. Moreover, if one desires to sacrifice oneself, then satisfying this desire itself constitutes a boost in one’s well-​being. And some objective list theories say that acting morally itself makes one’s life go better. Thus, we must be careful in spelling out the case to ensure that the painfulness of one’s death, and the shortening of one’s life, sufficiently outweigh these benefits of self-​sacrifice. 4 As Parfit (1986, 845) says in his ‘Comments,’ P’s metadesire or ‘master function’ is not restrictive, whereas S’s metadesire or master function is restrictive. This is because P tells one to do whatever will best achieve one’s present aims, whatever those aims may be. By contrast, S tells one to do whatever will best achieve the aim of maximizing one’s overall

Rationality and time  45

lifetime well-​being, even when that conflicts with one’s other aims. Parfit then claims that in such conflict cases, S is committed to regarding those other aims as in some sense irrational or inferior to aim of maximizing lifetime well-​being, for otherwise it should be rational to act on those aims rather than the aim of maximizing lifetime well-​being. 5 Compare Parfit (1986, 845–​846). Note also that it might be possible to devise a case against IP, the instrumental version of P, which says that one ought to act so as to best satisfy one’s present desires, whatever those desires happen to be.We might think it would be rational for the man with Future Tuesday Indifference to schedule a less painful operation for next Wednesday rather than a more painful one for next Tuesday, even though doing so would not best satisfy his bizarre present desires. 6 The Appeal to Full Relativity appears briefly in Sections 34 and 35 of Part One. It has antecedents in Sidgwick’s The Methods of Ethics (1907) and also in Nagel’s The Possibility of Altruism (1970, 60–​71). 7 Toward the end of Part Two, Parfit goes beyond the formal analogy between ‘I’ and ‘now’ and notes that ‘with respect to some requirements of rationality, the relation between a person now and himself at other times is relevantly similar to the relation between different people’ (191). For instance, while consistency of beliefs at a time is a requirement of rationality, it is no more irrational for one person to believe P and another to believe P than it is for a person to believe P at one time but believe P at another time. Similarly, it is a requirement of rationality that one’s preferences be transitive at each time. It is thus irrational for a person at a time to prefer A over B, B over C, and C over A. But just as there need be no irrationality in one person preferring A over B and B over C while another prefers C over A, so there need be no irrationality in a person at one time preferring A over B and B over C while at another time preferring C over A. Does this observation bolster the Appeal to Full Relativity? I am doubtful. Parfit is correct that with respect to some requirements of rationality, the relation between a person now and himself at other times is relevantly similar to the relation between different people. But the Appeal claims that this is so with respect to all requirements of rationality. Thus far, then, we are left with just a pro tanto preference for pure theories over hybrid ones. In my view, the strongest arguments for favouring pure theories over hybrid ones come in Part Three, where Parfit argues for reductionism about personal identity over time. See also Hedden (2015). 8 This is not to say that such compensation always justifies the harm. It may be wrong to impose the earlier harm if the person does not consent to it, even if the later and greater benefit constitutes compensation. 9 For starters, if being past and being future are special properties not reducible to relations of earlier than and later than, it would seem that being in the near future and being in the far future are likewise special properties not reducible to such relations. 10 Consider a slight modification of the surgery case above. The setup is the same: A fair coin is tossed to determine whether you will undergo the Early Course (four hours of pain on Tuesday and one hour on Thursday) or the Late Course (none on Tuesday and three hours on Thursday). And you know throughout that on Wednesday you will be given amnesia which will make you forget whether or not you had surgery on Tuesday. In this new version of the case, you know that on Monday you will be offered Help Early and Help Late, respectively. If you take Help Early, then if you’re in the Early Course it will reduce your Thursday surgery by 29 min, while if you’re in the Late Course it will increase your Thursday surgery by 31 min. If you take Help Late, then if you’re in the Early Course it will increase your Thursday surgery by 30 min, while if you’re in the Late Course it will decrease your Thursday surgery by 30 min. If you take both pills,

46  Brian Hedden

the result is simply that you will suffer one more minute of pain on Thursday than if you had refused both pills. Hence it seems irrational to take both. But Dougherty argues that if you are biased toward the future and risk averse, then on Monday you’ll prefer to take Help Early and on Wednesday you’ll prefer to take Help Late, regardless of what you do on the other day. Why? Taking Help Early reduces the difference between the highest and lowest amounts of possible future pain while increasing your expected future pain by only one minute. (This preference for reducing the difference between the worst-​case and best-​case scenarios at a small cost to expected value is characteristic of risk aversion.) And taking Help Late reduces the difference between the highest and lowest amounts of possible future (i.e. future relative to Wednesday) pain without any change in expected future pain. Thus, Dougherty argues that the combination of bias toward the future and risk aversion is irrational since it yields the threat of diachronic inconsistency—​on Monday you prefer to take Help Early, on Wednesday you prefer to take Help Late, but on both days you prefer rejecting both pills over taking both of them. He regards risk aversion as clearly rational and therefore lays the blame at the feet of bias toward the future. As noted, Greene and Sullivan (2015) reject Dougherty’s argument on the grounds that we are not risk averse with respect to pleasures and pains in the sense of having diminishing marginal utility for pleasures and pains. It is true that bias toward the future can still yield diachronic inconsistency if one is risk averse in the sense of following Buchak’s Risk-​Weighted Expected Utility Theory. But Risk-​Weighted Expected Utility Theory yields diachronic inconsistency on its own (see Briggs 2015 and Buchak 2015 for discussion). So this diachronic inconsistency may not constitute a good argument against bias toward the future in particular.

References Briggs, R.A. 2015. ‘The Costs of Abandoning the Sure-​Thing Principle.’ Canadian Journal of Philosophy 45 (5):827–​840. Brink, D. 2011. ‘Prospects for Temporal Neutrality.’ In Callender, C. (ed). Oxford Handbook of the Philosophy of Time. New York: Oxford University Press, pp. 353–​381. Broome, J. 1991. Weighing Goods: Equality, Uncertainty, and Time. Wiley-​Blackwell. Buchak, L. 2013. Risk and Rationality. New York: Oxford University Press. Buchak, L. 2015. ‘Revisiting Risk and Rationality:  A Reply to Pettigrew and Briggs.’ Canadian Journal of Philosophy 45 (5):841–​862. Dougherty, T. 2011. ‘On Whether to Prefer Pain to Pass.’ Ethics 121 (3):521–​537. Dougherty, T. 2015. ‘Future-​Bias and Practical Reason.’ Philosophers’ Imprint 15 (30):1–​16. Dreier, J. 1996. ‘Rational Preference: Decision Theory as a Theory of Practical Rationality.’ Theory and Decision 40 (3):249–​276. Gauthier, D. 1962. Practical Reasoning. Oxford: Clarendon Press. Greene, P. and Sullivan, M. 2015. ‘Against Time-​Bias.’ Ethics 125 (4):947–​970. Hare, C. 2013. ‘Time:  The Emotional Asymmetry.’ In Bardon, A. and Dyke, H. (eds) A Companion to the Philosophy of Time. Wiley-​Blackwell. pp. 507–​520. Hare, C. and Hedden, B. 2016. ‘Self-​Reinforcing and Self-​Frustrating Decisions.’ Noûs 50 (3):604–​628. Hedden, B. 2015. Reasons without Persons:  Rationality, Identity, and Time. Oxford:  Oxford University Press. Kagan, S. 1986. ‘The Present-​Aim Theory of Rationality.’ Ethics 96 (4):746–​759. Nagel, T. 1970. The Possibility of Altruism. Princeton: Princeton University Press.

Rationality and time  47

Nozick, R. 1974. Anarchy, State, and Utopia. New York: Basic Books. Parfit, D. 1986. ‘Comments.’ Ethics 96 (4):832–​872. Rawls, J. 1971. A Theory of Justice. Cambridge, MA: Belknap Press. Sidgwick, H. 1907. The Methods of Ethics, 7th Edition. London: Macmillan. Street, S. 2009. ‘In Defense of Future Tuesday Indifference: Ideally Coherent Eccentrics and the Contingency of What Matters’ Philosophical Issues 19 (1):273–​298.

3 INTRODUCTION TO PART THREE Personal identity Andrea Sauchelli

3.1  Criteria of personal identity One of Parfit’s motivations to discuss our beliefs about what we are is to further criticise the Self-​interest theory (S), a theory of rationality holding that our rational aim is to bring about the outcomes that would be best for ourselves, and that would make our lives go, individually, as well as possible (R&P: 3).1 In particular, Parfit suggests that since our identity over time is not always determinate and may hold as a matter of degree, an unqualified version of S is not plausible.2 To motivate part of this argument, he explores several theories of personal identity, for example the Cartesian Ego View, the Physical View and the Psychological View. Parfit groups theories of personal identity into two families, that is, reductionist and non-​reductionist views. The family of views favoured by Parfit (i.e., reductionism, a family of theories according to which the holding of the relation of personal identity consists in physical and/​or psychological facts) can support a criticism of S on the basis that the special concern we almost instinctually have for, say, ourselves at an old age is at least not as important or deep as we think it is. Parfit’s idea is that it is not irrational to regard our prudential concern for some of our ‘future selves’ as proportional to the degree of psychological connectedness we may have with them.3 More on the notion of ‘psychological connectedness’ later. Becoming a reductionist should persuade us to revise not only our views on rationality but also on morality. For example, Parfit suggests that reductionism supports ethical theories that include prescriptions to care impartially and impersonally for the welfare of others (although it is compatible with other views). Parfit’s strategy to argue for these conclusions, especially those related to his views on personal identity, relies on a series of thought experiments.4 For instance, consider:

Personal identity  49

The Teletransporter Case. Suppose that you want to travel to Mars. Luckily, traveling from Earth to Mars takes only approximately 20 minutes now:  a scanning device on Earth can record the exact molecular states of your body and brain and send this information to a receiver on Mars. Once the information is recorded and transmitted, your body and brain on Earth are destroyed. The device then recreates on Mars a perfectly functional replica of your body and brain.The new body and brain on Mars will remember the last thing that you did on Earth before the information was recorded (presumably, entering into the device) and will have your character traits, personality, interests, and so on. These facts may imply that your life, and not someone else’s, will continue. Now, the device works exactly as expected: on Mars, you seem to continue the conversation you were having with your partner on Earth before the both of you were teletransported. The Malfunctioning Teletransporter Case. There is a new device that creates people on Mars without destroying the ‘originals’ on Earth. You try such a device, but it malfunctions: the transmission of your blueprint and creation of a new individual on Mars based on this information are successful but the scanning process has caused some damage to the body and brain on Earth. In a few days, the body and brain on Earth will cease functioning—​an outcome you may think appropriate to describe as causing your death. A few points about these cases. In both scenarios, the teletransportations are presented so as to involve causal processes. In other words, there is a causal connection between the body and brain on Earth and those on Mars—​although such a connection does not imply that the body and brain on Earth are identical with the body and brain on Mars. Also, these scenarios presuppose that mental states and subsequent occurrences of mental states, for example memories, depend on at least physical states.5 How should we understand and evaluate the previous scenarios? First of all, the kind of identity at issue is the numerical identity of persons through time, not their qualitative identity.6 More specifically, Parfit is discussing which metaphysical criteria of personal identity correctly describe that part of reality involving the identity of numerically the same person through time, not the identity conditions of person-​types. In this context, a criterion of personal identity tells us what personal identity consists in, that is, what is at issue here is a metaphysical criterion, not a criterion about how we know that personal identity holds (which would be an epistemological criterion). Parfit claims that many would argue that the best description of the outcome of the second scenario is that you die. Why is that so? One intuitive answer is that you on Earth and the individual on Mars are not physically continuous and so you and the replica are different persons (i.e., you and the replica may be qualitatively identical without thereby being numerically the same). Although there is a causal process—​we may even add that this causal process is usually reliable—​the two individuals do not seem to have the same body or even a relevant part of it in common. This point may suggest that the identity of a person through time depends on one

50  Andrea Sauchelli

form of physical continuity (e.g., brain continuity), or, at least, that physical continuity is a necessary condition of personal identity. According to what Parfit refers to as the best version of the Physical View (or Criterion), personal identity within t1 and t2 holds if and only if enough of the same functioning brain continues to exist within these times in a non-​branching form (e.g., if the same brain continues to exist, is not split in half and both halves are not transplanted into different bodies within the period under consideration).Why is this version better than other forms of the Physical View? There is a sense of ‘body’ that does not imply that a body must be composed of exactly the same bodily parts to persist; for instance, if we regard it as an organism. If you are a yakuza, and have to make amends and cut one of your fingers, this does not mean that you or your body will cease to exist. Organisms gain and lose parts.That is one of the ways in which they survive. On this view, after having cut your finger, you have made amends (hopefully) and you still have or are the same organism (albeit slightly reduced). How many parts of your organism can you cut away and still be the same? A plausible reply is that perhaps the only thing you cannot cut away (or replace) without thereby ceasing to exist is your brain or, at least, those parts of your brain responsible for the functioning of some of your psychological activities or the functioning of your organism.7 The parts that cannot be cut away without you thereby ceasing to exist—​your brain—​would be those the continuity of which, under the right circumstances, grounds personal identity through time, an idea captured by the best version of the Physical View. A different type of theory is based on the idea that personal identity is essentially dependent on psychological features (the Psychological View). Parfit elaborates on John Locke’s and Sydney Shoemaker’s versions of this approach.8 In particular, Parfit improves Locke’s account in at least two ways: first, he distinguishes between psychological connections and psychological continuity, and, second, he more explicitly claims that various psychological states and connections in addition to memories are relevant to our identity over time (a point already made by Shoemaker). For example, one relevant connection can be that between an intention (‘I want to read Frank Herbert’s Dune’) and the subsequent acting on this intention (my reading the novel). Other relevant psychological connections can be those between our character traits and their expressions in our behaviour. Parfit defines psychological connectedness as ‘the holding of particular direct psychological connections’ and psychological continuity as ‘the holding of overlapping chains of strong connectedness’ (R&P: 206). Psychological connectedness can hold to different degrees, depending on the number of direct connections obtaining. For example, one direct psychological connection holds between you-​ yesterday and you-​today if you-​today remembers the experience of you-​yesterday having dinner. For you-​today to be strongly connected to you-​yesterday, you-​today should have at least half the number of direct connections that hold, over every day, in the lives of actual people—​Parfit claims that this specific quantity of direct connections is simply a stipulative qualification of ‘strong connectedness’. The relation of strong connectedness is not a transitive relation: from the fact that A at t1 is strongly connected to B at t2, and that B is strongly connected to C at t3, it does not

Personal identity  51

follow that A at t1 is strongly connected to C at t3. For example, considering only memories for simplicity, you-​today may remember much of what you-​three-​days-​ ago did, and you-​three-​days-​ago may remember much of what you-​one-​week-​ago did, but it is possible that you-​today does not remember much of what you-​one-​ week-​ago did.9 Parfit defines the Psychological Criterion as follows: P at t1 is one and the same person as Q at t2 iff P at t1 is psychologically continuous with Q at t2, in case such a continuity has the right cause and has not taken a branching form—​a relation branches if for each relatum at a time, there is not at most one other relatum at any other time when the relation holds. There are various ways of specifying what a right cause is: on the Narrow version of this criterion, the right cause is the normal cause (i.e., continuity of the relevant parts of the body and brain responsible for the relevant psychological connections); on the Wide version, the cause only has to be reliable (e.g., a reliable teletransportation device); on the Widest version, the right cause can be any cause (e.g., an unreliable teletransportation device). On the Narrow Psychological Criterion, it is possible that you may forget everything you did, say, in 2008 and still be the same person in 2018, provided that the changes in your character or memory are brought about by their normal causes—​ and that there are intermediate steps that form strong chains of psychological connections between you-​in-​2008 and you-​in-​2018. Note that this view differs from the brain-​version of the physical criterion in that the latter claims that the continuity of enough of your brain is necessary and sufficient for your persistence as the same person over time. According to the Narrow Psychological Criterion, continuity of enough of a person’s brain may be at most a necessary condition for this person’s continuity as the same person. In a note added in 1985, Parfit claims that he has withdrawn his previous (1984) support for the Wide Psychological Criterion and, in the 1987 edition, he claims that we do not have to decide (or even that we should not try to decide) which version of the psychological criterion to adopt (R&P: x; 208–​209). However, according to Parfit, there is a family of theories that we should accept: reductionism. Both physical and psychological theories as discussed above are forms of reductionism partly because they claim that the identity over time of persons consists in only the continuity of relevant bodily parts or psychological states (properly connected). Let us elaborate on the distinction between reductionism and non-​reductionism. In R&P, Parfit seems to draw the distinction in metaphysical terms, more precisely, in terms of the difference in the alleged metaphysical dependence or constitution of the metaphysical fact(s) of identity. More specifically, Parfit maintains that, for a reductionist, what constitutes personal identity over time is only facts regarding bodies, brains and (eventually) psychological states properly connected. Parfit says that a reductionist may also make a conceptual claim, namely, that facts of personal identity can be described in an impersonal way. If it is true that these facts can be described impersonally, then the related analysis of personal identity may not involve vicious circularity. In contrast, non-​reductionists generally claim that our identity is not analysable and that it depends on a further particular and indivisible fact or entity. For example, according to the Featureless Cartesian View, such a

52  Andrea Sauchelli

separately existing entity is a purely mental or spiritual substance the only essence of which is being conscious. This entity is said to exist ‘separately’ because its existence is supposed not to depend on or consist in the existence of physical and psychological facts. Some non-​reductionists even argue that we are purely mental and indivisible subjects of experience and that we may know this solely by introspection. Parfit denies that he is introspectively or directly aware that he is an entity of that kind—​and even if he were, it is not the case that he would thereby also know that such a purely mental entity is a persisting subject of experience. Even though Parfit rejects non-​reductionism, he does not think that the concept of a Cartesian Ego is unintelligible—​after all, it might have been true that such entities existed. His claim is rather that we do not have sufficient evidence for believing in the existence of such purely mental and indivisible entities.10 What is Parfit’s preferred theory of personal identity in R&P? Parfit is a reductionist and, although he claims that we do not need to (or even should not) choose amongst reductionist theories, he consistently argued throughout his career that, if at all, personal identity only consists in physical and psychological facts or events—​ where the latter do not involve purely mental or spiritual substances. In addition, Parfit’s version of reductionism also holds these two claims: (1) a person’s existence consists solely in the existence of a body and brain, and the occurrence of several mental events, and (2) a person is an entity that is distinct from a body and brain, and the occurrence of several mental events. On this view, persons exist, but their existence consists in the existence of physical and psychological facts. So, although a person can be said to exist—​we can even claim that a person has thoughts and desires, a body, a brain, etc.—​a person is not an independently existing entity (i.e., an entity the existence of which consists in facts other than those which are physical and psychological).11 As a nation can be said (a) to exist, (b) to be an entity that is distinct from its citizens and its territory, and (c) to be an entity the existence of which simply consists in the existence of its citizens living together in a certain way and a territory, people can be similarly said to exist in the way previously described. Given the kind of facts and entities our identity and persistence consist in, there can be cases where it is indeterminate whether a person is the same over time—​Parfit also argues that ‘only if we are separately existing entities, can it be true that our identity must be determinate’ (R&P: 216).

3.2  An objection to the psychological criterion Some non-​reductionists have argued that memory connections must presuppose personal identity to count as proper memories. Today, John remembers having eaten an ice cream yesterday. Remembering such an experience—​and not simply seeming to remember eating an ice cream—​presupposes that John himself had the experience, or so the non-​reductionist argues. The point can be generalised to all instances of experiential or episodic memory.12 So, a certain kind of psychological connection (namely, experiential memory) that is essential to our persistence already presupposes personal identity and thus cannot be used in a non-​circular

Personal identity  53

analysis of personal identity.13 Parfit’s defence of the Psychological View relies on (Shoemaker 1970).14 In particular, against one understanding of the charge of circularity, Parfit claims that we can define psychological concepts (e.g., quasi-​memory) in a way that does not presuppose the notion of personal identity. Such concepts can be incorporated into the Psychological View without thereby making it circular. More specifically, a quasi-​memory of a past event involves (1) a person’s seeming to remember such an event, (2) someone’s having experienced such an event and (3) the person’s current relevant mental state being causally dependent, in the right way, on the experience in question. The causal dependency that Parfit has in mind here is roughly in terms of memory traces, which are taken to involve (or be) modifications in a certain number of brain cells. Parfit also seems to believe that a memory trace, intended as a configuration of brain cells, can be replicated in different brains and produce identical or similar experiences—​this presupposition is not essential to his other arguments. On this understanding of quasi-​memory, it is possible that someone can quasi-​remember ‘from the inside’ of an experience that was had by someone else. For example, suppose that Tonio has a conversation with Hans in Lubeck. Suppose that such an experience is recorded in Tonio’s nervous system—​it forms a memory trace. Years later, Lisaveta has the memory trace corresponding to Tonio’s experience copied into her brain. Now Lisaveta can have a vivid apparent memory from the inside of what Tonio said to Hans—​Lisaveta is quasi-​remembering conversing with Hans. This does not imply that the conversation has for Lisaveta the same meaning or relevance as it had for Tonio. Rather, the point is simply that, granted that one is aware of having undergone such an implant, it does not seem to be contradictory to claim that it is possible to have non-​delusory memories of experiences from a distinctive first-​person perspective that have not been experienced by the person who is currently remembering them. What ensures that such memories are not delusions is that they represent real experiences and they are transmitted through an appropriate causal chain. This idea implies that it is not necessary that a memory is veridical only if such a memory presupposes personal identity over time.15 The next step is the incorporation of quasi-​memory and other mental states (e.g., quasi-​intending) into the Psychological View:  Instead of saying that psychological connectedness involves memories, we should rather claim that, for example, psychological connectedness is determined also by overlapping chains of quasi-​ memories, quasi-​intentions and so on. This statement completes what seems to be Parfit’s favourite version of reductionism.

3.3  Is personal identity always determinate? It is not necessary to believe in Cartesian Egos to claim that personal identity is always determinate. In particular, also some supporters of the physical criterion have suggested that although personal identity does not depend on a further fact, nevertheless personal identity is determinate. For example, Bernard Williams holds  that  we cannot even imagine or conceive of situations in which

54  Andrea Sauchelli

it is indeterminate whether I personally survive—​that is, situations in which it is indeterminate whether personal identity holds.16 This claim about what is imaginable or conceivable, in turn, would make it difficult to believe that it is possible that personal identity be indeterminate. However, Parfit argues that even the bodily criterion cannot plausibly provide the required sharp cut-​off point that would make personal identity always determinate. In addition, Parfit claims that only non-​ reductionism may imply such determinacy; however, since non-​reductionism is false and that we should be reductionists, we should admit that there may be cases in which it is indeterminate whether personal identity holds. Now, Williams’s main purpose in (1970) was not that of showing that personal identity must always be determinate, rather, he wanted to argue that the physical criterion is better than the psychological one. In a nutshell, Williams holds that, in considering cases in which torture is presented to us as an inevitable future event, we would find little solace in being told that, say, prior to the ordeal, the torturer will also tamper with our brain and inscribe in it memory traces from the life of Napoleon. In general, our emotions and beliefs about our own future do not seem to align with beliefs regarding our psychological continuity. To the degree that our prudential concern for our future pain reveals the extension of personal identity, it seems that psychological continuity is not what determines our personal identity (or our attitudes towards our future). Parfit replies that Williams’ reasoning seems to presuppose that questions regarding our persistence through time must always have a determinate answer (i.e., that in principle you could always affirm or deny that you exist at a specific time (as the same person)). This idea is questioned by appealing to various ‘spectrum arguments’. First, Parfit re-​describes part of Williams’ argument in terms of what he calls the psychological spectrum. Call the sum of your psychological features relevant to personal identity your ‘psychological profile’. Suppose that a surgeon can activate a series of switches working as follows. The flipping of switch s1 causes a minimal change in your beliefs, memories and character traits. The change will involve only the insertion into your psychological profile of a few memories and one minor character trait that Napoleon had. The flipping of s2 would cause another insertion so minimal that, in itself we would not regard it as sufficient to determine the end of a person.When all switches have been activated, you would eventually be psychologically similar to Napoleon—​you will have his psychological profile. For Williams’ argument to be a successful attempt to show that psychological continuity is not necessary for personal identity, we should rule out other evaluations of the case at issue. In particular, Parfit says that there are at least three ways of evaluating this case: (1) the two individuals at the end of the spectrum are numerically the same person, (2) there is always a sharp borderline between the two ends of the spectrum with respect to which person exists, and (3) the reductionist solution. If we want to deny 1., Parfit and Williams seem to assume that we would need to claim that there is a sharp borderline between the two ends even in all of the central cases of the spectrum (a similar assumption is made in the evaluation of the other cases below). Now, option 2 is excluded as highly implausible—​there is no

Personal identity  55

plausible candidate to play the role of this (non-​conventionally established) cut-​off point. The reductionist’s solution (3)  would be to say that there are cases in the middle of the spectrum where there is no answer regarding whether the resulting person would be you or not. However, according to Williams, we cannot make sense of this solution when it comes to, for example, the anticipation of a future agony; in fact, he believes that, on reflection, we have a strong intuition that it will be either me or not-​me who will suffer. Besides, special concern for our future does not seem to become inappropriate simply because of small changes (say, one or two non-​important memories). So, we may believe that 1. is the correct reply: at the end of the spectrum, the resulting person is you albeit with a different psychological profile—​after all, you still fear the prospect of a future torture even after being told about the change of psychological profiles (R&P: 230). If the problem of the supporter of the Psychological View with the previous reasoning is that she may have to admit the existence of sharp borderlines and that this is implausible, then, Parfit argues, this is hardly a problem only of the Psychological View. In fact, we can construct an analogous Physical Spectrum against the Physical View. In this scenario, switching s1 will cause the replacement of 1 per cent of the cells in your brain and body with duplicates out of new matter, s2 of 5 per cent and so on until a 100 per cent substitution. At the end of the spectrum, we would have a result similar to that for the case of teletransportation.17 Is there in this scenario a precise threshold the trespassing of which determines the end of a person and the beginning of another? First, we may argue that it is plausible to say that the two persons at the end of the spectrum are the same—​after all, they display the same behaviour, have the same interests, and so on. If plausible, this conclusion would show that the physical criterion is false (this is the counterpart of option 1 in the previous scenario). Alternatively (option 2), we may claim that there is a sharp borderline, but again this does not seem very plausible—​how can a difference of just a few cells (or even an atom) determine the end of a body and the beginning of another? The last option is to argue that there are borderline cases in this spectrum—​the reductionist’s favourite solution. Still, we may be tempted to choose 1. over 3., since we may believe that the alleged psychological continuity between the two individuals can sustain personal identity. The combined spectrum scenario includes a new twist aimed at showing that the reductionist’s favourite solution is after all the best option. In this scenario, at the near end of the spectrum there are changes compatible with personal identity through time—​the resulting person would be you in the same way in which, in your actual life, it is you who will wake up tomorrow with some minor difference in your psychological profile and your body. At the far end of the spectrum, there is someone physically and psychologically entirely different from you, say, Charlize Theron (I assume). Suppose that, by flipping a series of switches (sn), a scientist can gradually substitute parts of your brain and body with corresponding parts of new organic matter modelled after Charlize Theron—​for example, s1 can replace a few bodily and brain cells, s2 a few more and so on. In considering this case, Parfit claims that the counterpart of the previous option 1 is not plausible in this scenario: we

56  Andrea Sauchelli

cannot sensibly claim that the resulting person (a replica of Charlize Theron) is you, as this person is physically and psychologically completely different from you. Since it is implausible to claim that there is a sharp borderline—​after all, the differences generated by the activation of a switch are sufficiently small that it is hard to believe that the persistence of an adult organism or psychological profile depends on, for example, only one cell or memory—​Parfit claims that our best option is the reductionist’s. On this view, in at least some of the central cases of the spectrum, it is indeterminate whether the existing entity is you or Charlize Theron’s replica. Parfit also claims that, according to the reductionist, asking whether the resulting person is you or not in the central cases may be an empty question in the sense that, once you know which physical and psychological facts hold, you know all there is to know. Whether the resulting person would be you in the borderline cases may have an answer, but such an answer may just be a stipulation about how to use or extend our language, in particular that part of our language involving personal identity. This extension, in turn, does not seem to be based on anything metaphysically deep.18 Besides, Parfit says, such a stipulation cannot determine anything intrinsically or directly morally or rationally important (it may be important how we use our language in certain legal cases, but the importance of such stipulations do not derive from the facts personal identity consists in) or, at least, it may not have the importance we originally ascribed to the difference between identity and non-​identity.

3.4  Divided minds Parfit claims that ‘recent’ findings (recent in the 1970s and 1980s) further support the reductionist view. In particular, he discusses several cases that seem to put into question traditional views of the connections among minds, consciousness and persons. Parfit’s discussion is indebted to Nagel (1971) because he deploys some of its points to argue in favour of reductionism as well as for other more radical theses.19 First of all, it is (technically) possible to separate the two hemispheres of our brain by cutting a bundle of fibres that connect them—​an operation (commissurotomy) that surgeons have performed to treat severe cases of epilepsy. Since each hemisphere controls one hand, one field of vision and several other specific abilities (e.g., linguistic abilities), it has been possible to study the effects of this ‘brain splitting’ operation. In particular, it seems that, in certain cases, we have evidence to conclude that the right hemisphere, which controls the left field of vision and the left hand, may perceive and express awareness of things taking place on the left side of the body in a way that can be insulated from what can be separately perceived by the other hemisphere. In addition, the expressions of awareness of one hemisphere may differ from what the other hemisphere might express; for example, one hemisphere may report, through the use of the related hand, that one object is in its relevant field of vision while the other hemisphere might report only the existence of a different object on the other side of the field of vision. This situation can be described as involving a person with two streams of consciousness, each unaware of the other.

Personal identity  57

Parfit suggests that, although our brains tend to specialise during development—​ that is, certain areas of the brain become specialised in performing certain tasks (e.g., the elaboration of language)—​it is possible that, in some individuals, their mental abilities can be equally distributed across the two hemispheres. Now, suppose that this last possibility is true of you and imagine that you are taking a difficult physics exam in an isolated room. After having read the exam, you think that there are two different ways of solving the problem in it, but you do not know which one is correct or better. Fortunately, you have a device that allows you to momentarily split the two hemispheres of your brain, a device you can activate almost at will.You split your brain and the result is two different streams of consciousness, each separately aware of working on one way of solving the problem. We might even suppose that, in addition to the splitting device, you also have a memory trace-​reduplicator—​ perhaps installed to amplify the vividness of your memory or as a back-​up if half of your brain is compromised—​that produces memory traces of your short-​and long-​ term memories in both hemispheres (or, at least, of those memories that will make a contribution to the decision-​making systems of your brain). After some time, the designated hemisphere activates the device and reunites your brain. Suddenly, you seem to remember having worked on two different calculations. While your brain (and mind) was divided, each single stream of consciousness had the feeling of being the same person before the split, apparently including the memory of deciding to split your brain to work on two separate ways of solving the problem and how to proceed in this regard. Parfit suggests that, since the brain-​division is brief, reversible, completely under your control or at least under the control of one designated stream, and that each separate stream’s experiences are accessible to the main stream after unification, it is plausible to claim that, in the context of this scenario, there is only one person in the room. Some have argued that each stream of consciousness is unified by a subject of experience and that each of us is a subject of experience. However, this idea is not appealing to those who claim that, in the physics-​exam scenario, you had two different streams of consciousness: how can you be the unifying subject of both if at any time when your mind was divided each stream was not conscious of at least one of the experiences of the other stream? Perhaps we should say that when your mind was divided, there were two different subjects of experiences.We might even have to claim that the case in question involved at least three subjects of experience: the owner of the pre-​division stream of consciousness (a person), and the two subsequent owners of the two different streams causally connected to the first one. A fourth individual can be added: the subject after the unification of the streams. Parfit claims that, on this interpretation, we may have to believe that the life of a person might involve two non-​person subjects of experience. However, when you divide, each stream feels continuous with the allegedly different previous subject of experience—​according to our scenario, each stream of consciousness need not be interrupted or experience disunity with the antecedent experiences. Explaining this unity of consciousness by saying that these experiences are experienced by the same subject that is not you seems implausible—​after all, each post-​division stream ‘feels’ that it was part of the pre-​division stream. How can

58  Andrea Sauchelli

this new subject not be you if the unity of each stream of consciousness is explained by the existence of one and only one corresponding subject of experience? At this point, the reductionist may claim that there is a simpler and better explanation. In particular, the reductionist can argue that the unity of consciousness in the case of division is explained by the fact that there are two single states of awareness that ‘connect’ certain thoughts, perceptions and feelings. When the mind is divided, there is no single state of awareness that spans over both sets of experiences at the same time. In addition, the reductionist can say that, in the case at issue, there are two streams of consciousness each co-​conscious of appropriately connected mental states and that this is all there is to say with regard to their unity. The unity of consciousness need not be explained in terms of a unifying subject, but can simply be explained in terms of the relations that connect the different mental states in each stream. In Parfit’s words, ‘a particular mental event occurs within some life in virtue of its relations to the many other mental and physical events which, by being interrelated, constitute this life’ (R&P: 252).20

3.5  Personal identity does not matter An extension of the previous thought experiment is the following. Suppose that you have an identical twin and that your body and your twin’s brain have been fatally damaged. Surgeons connect your brain to the nerves of your twin’s body successfully. Who is the person that wakes up after the operation? Both supporters of the Psychological View and of the brain version of the Physical Criterion may agree that the resulting person is you. Notice that you could have survived had only half of your brain survived the operation: after all, we believe that people who had a stroke that damaged an entire half of their brain can survive. David Wiggins gave this thought experiment a further twist.21 Suppose that you have another sibling with a body similar to yours, that your brain has been divided into two halves, and that each half is separately transplanted into one of your siblings’ bodies. Both resulting people will be psychologically continuous with you and believe to be you.22 What happens to you after this person-​splitting (or fission) case? (1) You do not survive, (2) You survive as one of the two people, (3) You survive as the other, or (4) You survive as both. Against 1., Parfit claims that since you would survive had only half of your brain been successfully transplanted, it is not plausible to hold, at the same time, that you do not survive if the other half were also successfully transplanted. In Parfit’s words, ‘[h]‌ow could a double success be a failure?’ After all, the relation between you and each of the resulting people is the same. Options 2. and 3. are implausible: given the relevant facts, there is nothing that would seem to make you one of the resulting individuals rather than another—​and thus regarding 2. (or 3.) more plausible than 3. (or 2.). On behalf of 4., we may say that describing the two resulting individuals as different persons is misleading:  since Parfit admitted that a person can have a divided mind, perhaps the two resulting individuals are parts of one person with a doubly-​embodied divided mind, that is, you may claim that you survive as both. However, Parfit claims, this supposition would involve too

Personal identity  59

great a distortion of our common concept of a person—​particularly when we also assume that the two individuals may decide to each go their separate ways and lead different lives, perhaps in two different parts of the world. Among the problems of this possibility for our concept of a person, claims of moral responsibility and other attitudes we think as connected to the concept of a person may, over time, be applicable only to one of the resulting individuals. Parfit suggests that the reductionist has a better solution. In particular, the reductionist would say that, in a sense, the different options above do not really describe different possibilities. In knowing that each individual will have half of your brain and will be psychologically continuous with you, you know everything there is to know with regard to the identity of the individuals in question—​and adding that, for example, it is you who survives would not add anything about reality. There is a better description of the outcome of the thought experiment, which Parfit claims to be that neither of the resulting people will be you, but this is not a description that adds anything to reality, at best it indicates what we regard as the most reasonable way of extending our concept of personal identity. In a sense, which option we choose (that is, which way we decide to describe the outcome) may simply be a linguistic or conceptual stipulation. Parfit maintains that a more important question—​more important than determining how to apply the concept of personal identity (and related concepts or terms, such as ‘you’) to this case—​concerns how you should regard, from a moral and rational point of view, your division. For example: Should you be prudentially concerned with the fate of these individuals?23

3.6  What matters when you divide? Parfit maintains that, when evaluating the case of your division, it would be irrational to regard the outcome as bad as death, even assuming that we describe the outcome as one in which you are not personally identical to any of the resulting individuals. The main reason is that the relations holding between you and each of the two different resulting people do not fail to individually contain anything important that is not included in your ordinary survival. If it is rational to have prudential concern for yourself, it may also be rational to be prudentially concerned for the two products of your fission. After all, each relation to the corresponding resulting person contains what we would independently regard as prudentially mattering. Certainly, a case of double survival is not similar to ordinary survival, but it does not mean that it does not contain what matters in normal survival. Rather, Parfit claims, what is important is relation R, where ‘R is psychological connectedness and/​or continuity, with the right kind of cause’ (R&P: 262).24 Parfit also suggests that, in this context, the ‘right kind of cause’ is any cause. In our world, relation R normally has a non-​branching form, that is, R connects no more than one person existing with another at a particular time, whereas, in the your-​division case, relation R has a branching form—​it holds between you and two different individuals,

60  Andrea Sauchelli

both existing at the same time—​hence, there is not a single future individual with whom you are identical. Some (e.g., Susan Wolf) objected that, far from being appealing, branching-​R scenarios would have horrifying consequences.25 In particular, according to Wolf, branching would disrupt some of the things we value most, such as some forms of exclusive personal relationships. In R&P, Parfit seems to address this issue by saying that the uniqueness of the holding of R may have some extra value, but it can make at most some difference—​perhaps a difference due to the psychological profile of the dividing person. However, Parfit claims that, on reflection, even such an individual should not rationally regard the prospect of division as bad as death. A different reply, for example the one discussed in (Parfit 2007), draws our attention to the distinction between questions of prudential concerns from questions regarding what makes survival desirable. The point of the case of your division is to discuss whether it would be irrational to extend to the R-​branching individuals the kind of prudential concern that a self-​interest theory would prescribe you to have for yourself. Extending such a concern does not imply that such an extension would be always desirable—​for example, you may be prudentially concerned for yourself even though you think that your future will not be desirable. Questions of the desirability of a future may not coincide with questions of prudential concern for a future. So, in evaluating your division with regard to the rationality of prudential concern for R-​branching individuals, we would better suppose that their future lives are as desirable as the one the pre-​branching individual is having.Wolf ’s worries may thus not be all relevant to the case at issue. What are some of the consequences of accepting the previous arguments about the non-​importance of personal identity? One is this: instead of saying ‘I shall be dead’, you may rather say that there will not be any future experiences related in certain ways to the experiences you are now having. This idea may bring you some solace at the perspective of incumbent death: death does not involve the irremediable destruction of something metaphysically deep. Rather, it should be seen as the interruption of a causally interrelated flow of experiences and bodily functions. If we hold that what matters is both R and physical continuity, our position is equivalent to saying that what matters is R with its normal cause since physical continuity (at least in its brain-​continuity version) is part of R’s normal cause. However, Parfit claims that the continuity of your brain should not matter if it is not the carrier of relation R. Brain-​continuity can be part of what matters, but its importance is only derivative. For, Parfit claims, our desire (if at all) to have a body or brain similar to those we have is not the desire to have a particular body. In fact, a functionally equivalent body—​functionally equivalent in its capacity to sustain strong psychological connectedness—​should, on reflection, suffice. Our ‘old brain and body’ may have sentimental value to us, so it may not be entirely irrational to care about their persistence, but such a care should not obfuscate what really matters or why this brain and body started to matter at all: the value of what they sustain. In considering whether both psychological connectedness and continuity matter, Parfit seems to believe that caring about only psychological continuity may involve

Personal identity  61

deep regret (R&P: 301). It is conceivable that, given sufficiently long periods of time, we may have psychological continuity between A and B with very few psychological connections between them; however, Parfit argues, most of us would regret the loss of some of our long-​past memories. A similar claim can be made about the persistence of other psychological features such as intentions and desires. Although you may regard psychological fluctuations and changes within the life of a person as compatible with psychological continuity, a significant number of them would reduce our psychological connectedness over time. Parfit’s conclusion is that psychological continuity certainly matters, but it is not true that it is the only relation that matters: psychological connectedness matters as well.

3.7  Personal identity and rationality How is the previous discussion on personal identity connected to Parfit’s criticism of S? According to S, for each person, there is one supremely rational ultimate aim, namely, that things go as well as possible for her. Some (e.g., Joseph Butler) have argued that if reductionism is true, we have no reason to be concerned about our own future (the Extreme Claim). If this claim and reductionism were both true, we would have a direct refutation of S. However, Parfit maintains, a reductionist can also plausibly argue that relation R may give us some reasons for special concern (the Moderate Claim). Perhaps a better argument against S can be devised. This new argument is based on the idea that what matters are psychological connectedness and continuity. Another crucial premise is a central point of S, the Requirement of Equal Concern: a rational person should be equally concerned about all parts of their future.26 Now, Parfit claims that there may be other reasons to support the idea that it may not be irrational to care less about one’s future other than simply because it is in one’s own future. In particular, Parfit holds that one of these reasons is that our concern for our future, may correspond to the degree of connectedness between me now and myself in the future. Connectedness is one of the two relations that give me reasons to be specially concerned about my own future. It can be rational to care less, when one of the grounds for caring will hold to a lesser degree. Since connectedness is nearly always weaker over longer periods, I can rationally care less about my further future. (R&P: 313) Although in general it is not true that to a decrease of the degree to which a relation holds we can always rationally believe that such a relation has less importance, it may not be irrational to care less about our own future because there is a reduced degree of psychological connectedness.27 Hence, Parfit argues, it may not be irrational for you to care less when there will be much less connectedness, given that such a connectedness is one of the reasons for having prudential concern. For example, suppose that you know that you will have a painful day tomorrow and

62  Andrea Sauchelli

one in 30 years.You-​now is likely to be more strongly psychologically connected to you-​tomorrow than to you-​in-​30 years. Since connectedness is one of the relations that justifies prudential concern, you cannot be irrational to comparatively care less when there is less connectedness. One consequence of—​some may say ‘an objection to’—​the previous reasoning is that, if Parfit’s view is correct, it might not be irrational to be greatly imprudent, for example, it may not be irrational to be an imprudent youth. However, we should criticise great imprudence. Parfit’s reply to this objection is that we should include the issue of imprudence in the moral territory—​that is, condemn great imprudence from a moral point of view. For example, we might appeal to one form of consequentialism, in particular, to a version including an impartial or agent-​neutral principle of beneficence. On this version of consequentialism, if your action now will overburden you-​old, you are increasing the overall sum of suffering. This increase of suffering is morally wrong, independently of whether it is you or an individual about whom you may rationally prudentially care less than yourself who will suffer as a result of your action. So, great imprudence can still be criticised. In addition, we may also claim that you stand in a special relation of duty or obligation to your future self, akin to the relation you may have with your children, pupils etc.28 In turn, this obligation can provide further ground to criticise imprudence.

3.8  Personal identity and morality Reductionism gives us reasons to change some of our moral beliefs. Among the claims or consequences of reductionism that can give us such reasons, Parfit focuses particularly on: (1) personal identity over time may not be a deep truth or fact, and (2) becoming a person and personal identity over time are a matter of degree.29 In evaluating the morality of early cases of abortion (e.g., a few days after fertilisation), the reductionist may believe that at the beginning of a pregnancy there may be nothing seriously wrong with a voluntary abortion, as it would not amount to the destruction of an existing person, and that having an abortion may become, ceteris paribus, gradually wrong as time passes and the foetus gradually nears personhood. Some non-​reductionists have maintained that only if personal identity is a further fact or involves a deep metaphysical truth, can we justifiably maintain a series of important moral and legal practices (the moral counterpart of the Extreme Claim for rationality). For example, non-​reductionists argue that if reductionism is true, we would not be justified anymore in punishing people for their past crimes (at least in certain cases). The reason would be that only if non-​reductionism is true can we truly say that a person is identical through time, and thus that the same person who committed a crime can be subsequently punished for it (a necessary requirement for punishment). Parfit suggests that a different claim is equally defensible, namely, that also psychological continuity may carry a certain degree of moral responsibility; hence, reductionism is compatible with some of our current beliefs about the scope of, for example, punishment and responsibility. If we also consider psychological connectedness as having moral and practical relevance, we may

Personal identity  63

also think that punishment should be proportional to the degree of psychological connectedness between, for example, the person at the time of the crime and the person at the time of conviction (or even later in their life). This conclusion seems plausible. There are other consequences of reductionism for morality, for instance, with regard to compensation and distributive justice. To see the connection between reductionism and these issues, we need to introduce the notion of the separateness of persons—​roughly, the idea that each person has a separate, individual life to live and that this fact is normatively relevant.30 Versions of this notion seem to have influenced important figures in the history of ethics, including some utilitarians. For example, Henry Sidgwick claims that there are two rational aims, S and Benevolence. In particular, Benevolence says that it is rational to desire or act such that ‘things go, on the whole, as well as possible for everyone’ (R&P: 329). Sidgwick then specifies the requirement of benevolence in terms of his Hedonism; more specifically, he suggests that ‘our ultimate moral aim is the greatest net sum of happiness minus misery’ (R&P: 330). On his version of utilitarianism, what morally matters is the amount of happiness and suffering—​how this distribution is performed among people makes no moral difference insofar as the utility is maximised. Benevolence may not be compatible with S and Parfit suggests that Sidgwick accepted the possibility of an irresolvable conflict between these two principles also because he believed in the separateness of persons. Given the separateness of persons, it may be rational to hold either of the two principles. Some deontologists also appealed to the separateness of persons. For example, John Rawls, Thomas Nagel and others variously objected to utilitarianism that, in addition to its principles of impersonal distribution, we should also introduce other principles of distributive justice the moral importance of which is not reducible to their utility and that are based on the separateness of persons. Now, utilitarians need not disregard distributive principles; rather, they may regard these principles (e.g., of equal distribution) as mere means to maximise overall or average utility, not as aims in themselves. Still, critics of utilitarianism have complained that this way of reasoning ignores ‘the boundaries between lives’ and their differences.31 Ignoring such boundaries would be wrong because, among other things, the relationship between (i) one person’s selves and those of other people and (ii) the selves of a single person, is radically different and incommensurable; for example, only the holding of the latter relationship can automatically compensate an individual for what she did at an earlier time. As David Brink noted in a similar context, ‘diachronic, intrapersonal compensation is automatic; interpersonal compensation is not.’32 Parfit claims that there are different ways in which reductionism can be combined with our preferred principles of distribution; for instance, reductionism may justify changes to the scope or to the weight of these principles. With regard to the first option, a reductionist may change the scope of the principles at issue (their ‘units of moral concern’) from entire lives to selves or even to people’s states at particular times.33 In particular, Parfit suggests that reductionism supports regarding the subdivision within lives as, in certain ways, like the divisions between lives (since the

64  Andrea Sauchelli

fact of non-​identity is not a deep metaphysical truth). As a consequence, a reductionist may apply distributive principles even within lives. For example, in a situation in which a child (the present self of a person) will be burdened with some hardship for a benefit in his adulthood (a future self of the same person), the reductionist may claim that such a distribution would be as unfair, or at least similarly unfair, as the allocation of a burden on a person to benefit someone else: If all the future selves of a person are, in certain ways, like selves of other people, the reductionist may regard trade-​offs within lives as potentially unfair as trade-​offs between lives (R&P:  333–​334). However, the reductionist can also suggest that, although the distributive principles should have a wider scope, their weight should be correspondingly reduced (R&P: 334–​345). For example, if distributive principles are based on the separateness of persons (the alleged non-​identity between people), the reductionist may argue that since there are cases in which intrapersonal and interpersonal boundaries are not metaphysically deep, we should give less moral weight to principles such as that of compensation in these cases. Parfit’s main point is that if we accept reductionism and come to believe that persons are not separately existing entities, the fact of their non-​identity should seem less important and ‘it becomes more plausible to be more concerned about the quality of experiences, and less concerned about whose experiences they are’ (R&P:  346). On the reductionist view, the impersonality of utilitarianism towards, for example, suffering—​roughly, the idea that it is not morally important who is suffering but the negative quality of the experience—​is more plausible. In general, Parfit’s lesson here is that a change in our beliefs about what we think we are and how we persist through time has significant consequences for our beliefs about rationality and morality.

Notes 1 I use ‘R&P’ for references to Reasons and Persons (1984/​87). 2 I here assume that personal ontology and the metaphysics of personal identity coincide. More recently, philosophers have argued that an answer to the question ‘What are we?’ is not immediately answered by an account of personal identity. For instance, some have suggested that we are entities that are not always persons (e.g., we are human organisms). See Olson (2007) and Sauchelli (2018: 13–​14). Theories of personal identity are here understood as aiming at providing at least the synchronic and diachronic conditions of identity of persons. 3 Parfit employs the terminology of ‘selves’, ‘future selves’, etc., but also claims that it is simply a manner of speech. A self is merely a unit of particularly interconnected psychological connections, but is not an independently existing entity. 4 Some scholars have objected in various ways to the thought experiment methodology. I do not have the space to properly address their arguments but I will assume that thought experiments—​even those that investigate the application of certain familiar concepts to situations alien to the ‘normal application of such concepts’—​are still useful. (R&P: 200). See Wilkes (1988) for criticism. 5 Parfit does not specify what he regards to be the metaphysical nature of mental states or of their content. Some philosophers of mind may argue that some of the presuppositions of his thought experiments are questionable on the basis of unstated internalist presuppositions on the metaphysics of content. See Burge (2003).

Personal identity  65

6 Two things may be qualitatively identical without thereby being numerically identical—​ or so it is generally assumed. Besides, one thing can be numerically the same at a later time without thereby being qualitatively identical to itself in the past (e.g., a baby growing into an adult). 7 See Olson (2007). 8 Locke (1694) and Shoemaker (1970). See also Shoemaker (1997) for a later development of his view. 9 See section 2 for a qualification of the kind of psychological connections Parfit has in mind. 10 Parfit has discussed and partially modified the distinction between reductionism and non-​reductionism in several other publications after R&P. See Parfit (1999) and (2007). 11 This version of reductionism, a slightly different version of which Parfit would later call Constitutive Reductionism, resembles Shoemaker’s and Baker’s theories. See Parfit (1999), Shoemaker (1984), (1997) and Baker (2000). More recently (2012), Parfit proposes a different approach (although he says that he has not rejected his previous version of constitutive reductionism). 12 Psychologists and philosophers have distinguished a variety of kinds of memory (episodic, semantic, etc.). See Bernecker (2010: 11–​45) for an extensive review. 13 This objection has been understood in other different ways, at least starting from Joseph Butler’s and Thomas Reid’s criticisms of Locke (Sauchelli 2018: 110–​112). 14 In turn, Shoemaker’s account is indebted to Martin and Deutscher (1966). In a later work, Shoemaker claims that Parfit’s example in R&P of an episode of quasi-​memory is not convincing. See Shoemaker (2004: 581). 15 See Schechtman (1990) and McDowell (1997) for further criticisms. Parfit (1999) contains a reply to McDowell. 16 The actual claim made by Williams is subtler than this. In particular, his point is that it is of little help in settling our emotions to know that there are indeterminate cases in our future when we project or imagine how such a future would be like—​under the presupposition that we know that we will have to undergo such a future. See Williams (1970/​ 73: 58–​59). 17 Some may argue that it makes a difference whether the change is gradual and continuous (both in space and time) or whether the original brain and body stop functioning (even for a very short time). In fact, we can argue that if the change is gradual and does not involve any interruption of function, we may dispute Parfit’s claim that at the far end of the spectrum your ‘brain and body are completely destroyed’ in the manner in which the teletransporter does. See Unger (1990: 123–​125) and McMahan (2002: 70–​72). 18 It is not always clear what the concept of ‘metaphysically deep’ is. One way of understanding it is that a difference is not metaphysically deep with respect to K if it is a difference that we do not generally regard as significant to determine the belonging of an entity to a kind K (whether natural or not). 19 Parfit cites some early works by Roger Sperry. See Gazzaniga (2005) on the topic. 20 Parfit calls this view ‘the bundle theory’ of self in Parfit (1987). 21 Wiggins (1967: 53–​55). Parfit says in various places that reading Wiggins’s example and the Brown-​Brownson thought experiment in Shoemaker (1963) greatly influenced his choice of studying philosophy. 22 We may further add that the two ‘receiving bodies’ are equally similar with yours with respect to the relevant facts (e.g., if one of your character traits is inherently connected to certain features of your kind of body and a certain physical appearance, we can stipulate that such a connection is maintained after the operation).

66  Andrea Sauchelli

23 The thesis that personal identity does not prudentially matter has generated considerable debate (e.g., Lewis (1976/​83), Brueckner (1993), Parfit (1993) and Johnston (1997)). Recent instances are Johansson (2010) and Gustafsson (2018). 24 In a more recent paper, Parfit recognises that this formulation contains an inaccuracy. In particular, he should have said that what matters is non-​branching R, where R is psychological connectedness and continuity (Parfit 2007: note 30). 25 See Wolf (1986). 26 See also Sidgwick (1907/​81: 124, n. 1). 27 It is not entirely clear to me whether Parfit has successfully addressed the objection discussed at (R&P: 314). 28 See Whiting (1986) for relevant considerations. 29 We should distinguish two claims, namely, that (1) it can be indeterminate whether an entity at a time is a person, and (2) it can be indeterminate whether P at t is the same person as Q at t*. Parfit seems to accept both (R&P: 321–​323). 30 Norcross (2009) provides an interesting (non-​sympathetic) attempt to understand this notion. 31 ‘To sacrifice one individual life for another, or one individual’s happiness for another’s is very different from sacrificing one gratification for another within a single life’ (Nagel 1970: 138). 32 Brink (1997: 108). 33 In replying to an objection raised by Bart Schultz, Parfit seems to suggest that, given reductionism and the argument on compensation, the relevant units of moral concern—​ the moral units that should be regarded as the proper subjects of distribution, etc—​would better be selves (temporal parts or stages of people, see note 3). See Schultz (1986) and Parfit’s reply at (1986:  840). Parfit clarifies his claim in Appendix H (1986). See Shoemaker (1999) for discussion.

References Bernecker, S. 2010. Memory. Oxford: Oxford University Press. Brink, D. O. 1997. Rational egoism and the separateness of persons. In Reading Parfit, ed. J. Dancy, 96–​134. Oxford: Blackwell Publishers. Bruekner, A. 1993. Parfit on what matters in survival. Philosophical Studies 70, 1: 1–​22. Burge, T. 2003. Memory and persons. Philosophical Review 111, 3: 289–​337. Gazzaniga, M. 2005. Forty-​five years of split-​brain research and still going strong. Nature Reviews Neuroscience 6: 653–​659. Gustafsson, J. 2018. The unimportance of being any future person. Philosophical Studies 175: 745–​750. Johansson, J. 2010. Parfit on fission. Philosophical Studies 150, 1: 21–​35. Johnston, M. 1997. Human concerns without superlative selves. In Reading Parfit, ed. J. Dancy, 149–​179. Oxford: Blackwell. Lewis, D. 1976/​83. Survival and identity. Reprinted in his Philosophical Papers, Volume I, 55–​77. Oxford: Oxford University Press. Locke, J. 1694/​ 1971. An Essay Concerning Human Understanding. 2nd edition. Oxford: Clarendon Press. Martin, C. B. & M. Deutscher. 1966. Remembering. Philosophical Review 75, 2: 161–​196. McDowell, J. 1997. Reductionism and the first person. In Reading Parfit, ed. J. Dancy, 230–​250. Oxford: Blackwell. McMahan, J. 2002. The Ethics of Killing. Oxford: Oxford University Press.

Personal identity  67

Nagel, T. 1970. The Possibility of Altruism. Princeton: Princeton University Press. Nagel, T. 1971. Brain bisection and the unity of consciousness. Synthese 22: 396–​413. Olson, E. 2007. What Are We? Oxford: Oxford University Press. Parfit, D. 1984/​87. Reasons and Persons. Oxford: Clarendon Press. Parfit, D. 1986. Comments. Ethics, 96: 832–​872. Parfit, D. 1987. Divided minds and the nature of persons. In Mindwaves, eds. C. Blakemore & S. Greenfield, 19–​28. Oxford: Blackwell. Parfit, D. 1993. The indeterminacy of identity. Philosophical Studies 70, 1: 23–​33. Parfit, D. 1999. Experiences, subjects, and conceptual schemes. Philosophical Topics 26, 1–​2: 217–​270. Parfit, D. 2007. Is personal identity what matters? Marc Sanders Foundation, 31 December. Retrieved at:  www.marcsandersfoundation.org/​wp-​content/​uploads/​paper-​Derek-​ Parfit.pdf Parfit, D. 2012. We are not human beings. Philosophy 87, 1: 6–​28. Rawls, J. 1971/​ 99. A Theory of Justice. Revised Edition. Cambridge, MA:  Harvard University Press. Sauchelli, A. 2018. Personal Identity and Applied Ethics. London & New York: Routledge. Schechtman, M. 1990. Personhood and personal identity. Journal of Philosophy 87, 2: 71–​92. Schultz, B. 1986. Persons, selves, and utilitarianism. Ethics 96, 4: 721–​745. Shoemaker, D. 1999. Selves and moral units. Pacific Philosophical Quarterly 80: 391–​419. Shoemaker, S. 1963. Self-​Knowledge and Self-​Identity. Ithaca, NY: Cornell University Press. Shoemaker, S. 1970. Persons and their pasts. American Philosophical Quarterly 7, 4: 269–​285. Shoemaker, S. 1997. Self and substance. Philosophical Perspectives 11: 283–​304. Shoemaker, S. 2004. Brown-​Brownson revisited. The Monist 87, 4: 573–​593. Sidgwick, H. 1907/​ 1981. The Methods of Ethics, 7th edition. Indianapolis, IN:  Hackett Publishing. Unger, P. 1990. Identity, Consciousness, and Value. Oxford: Oxford University Press. Whiting, J. 1986. Friends and future selves. Philosophical Review 95, 4: 547–​580. Wiggins, D. 1967. Identity and Spatio-​Temporal Continuity. Oxford: Blackwell. Wilkes, K. 1988. Real People. Oxford: Clarendon Press. Williams, B. 1970. The self and the future. The Philosophical Review 79: 161–​180. Wolf, S. 1986. Self-​interest and interest in selves. Ethics 96, 4: 704–​720.

4 PARFIT, POPULATION ETHICS AND PARETO PLUS Reasons and Persons Part Four M. A. Roberts

Introduction The work that has been done in population ethics since, and largely thanks to, Derek Parfit’s completion of Reasons and Persons positions us to understand that some of its most critical arguments float just beneath the surface of that great work. In this chapter, I  focus on Part Four of Reasons and Persons and particularly on the chapters on the Mere Addition Paradox and the Nonidentity Problem. It’s indisputable that the various texts that make up those parts of Reasons and Persons have taken us on some fun-​filled and occasionally harrowing adventures.Yet at this juncture we have become comfortable with many of the conclusions those texts explicitly aim for. They have become familiar to us and may well strike us as relatively modest –​as interesting, to be sure, and helpful in understanding the structure of morality, but nothing that upsets the apple cart. What I want to suggest in this chapter is that any such complacency in respect of Part Four is premature. Enough time has passed that we may well now be comfortable with many of the conclusions spelled out in Parfit’s texts. But it’s also true that enough time has passed that we are now in a position to understand –​decades into population ethics –​that the subtexts of Part Four give rise to conclusions that are far more disturbing –​conclusions that do upset the apple cart, and conclusions that, if correct, leave the whole of population ethics hanging in the balance. Most prominent among those conclusions is a principle we can call Pareto plus. According to that principle, the mere addition of the worth-​having existence makes things better.1 That is, the addition of the worth-​having existence, other things equal, makes one alternative outcome, or possible world or future, better than another. If we think that one future’s being better than another is tied in some important way to our evaluation of what we ought to do, then that same

Population ethics and Pareto plus  69

addition  –​under certain assumptions  –​can also make an otherwise permissible choice wrong and an otherwise wrong choice permissible or even obligatory.2 Those results are highly disturbing. Intuitively, we think it often makes things worse, and is often wrong, for agents to bring an additional person into a miserable existence or an existence that is avoidably less than it might have been. But we don’t (other things equal) see the wrong in the choice never to bring that same person into existence to begin with. Beyond intuition, we also have practical reasons to find Pareto plus disturbing. It both –​as we shall see –​puts climate ethics at odds with itself and strips the notion of a constitutional right of early abortion and contraception –​the notion, that is, of a constitutional right of procreative privacy –​of its moral foundation. The highly contentious Pareto plus may seem a position that Parfit himself nicely managed to sidestep in Part Four of Reasons and Persons even as he successfully demonstrated results that, while interesting and useful, strike us at this juncture as relatively modest. But we now know enough to know that that palliative thought involves a lot of hair-​splitting.With the texts come the subtexts; with the arguments Parfit explicitly spells out come still other arguments Parfit himself said little or nothing about. The upshot is that we need to go back to the beginning. Given all that is at stake, we need to examine the texts and the subtexts more carefully than we ever have before.

Plan for this chapter I’ll turn first to Parfit’s chapter on the Mere Addition Paradox. In that chapter, the explicit claim –​the claim Parfit at least provisionally seems to accept; the claim that we can call the Mere Addition Principle (MAP) –​is that, other things equal, the mere addition of the worth-​having existence doesn’t make things worse. More recent work in population ethics, however, forces us to appreciate that MAP unfolds, inevitably if not immediately, into Pareto plus: that, if mere addition doesn’t make things worse, then it must make things better. I’ll then turn to Parfit’s chapter on the Nonidentity Problem. An explicit conclusion Parfit claims for that chapter is that the person affecting, or person based, intuition (PBI), is false –​that is, that a future can be worse, or a choice wrong, even though it makes things worse for no existing or future person at all. The considerable work on the Nonidentity Problem that has been done in recent years at least suggests that the most powerful versions of Parfit’s argument depend in part on the idea that the better chance of a worth-​having existence makes things, in some sense, better.3 But if the better chance of existence makes things morally better, it seems entirely reasonable to suppose that the actual fact of existence must surely make things better still. And that –​again –​is just Pareto plus. In connection, then, first with the Mere Addition Paradox and then with the Nonidentity Problem, this chapter has two goals. The first is to identify the arguments –​the textual in combination with the subtextual –​that aim to compel us to accept Pareto plus. The second is to argue that those arguments are problematic.

70  M. A. Roberts

The upshot? That we aren’t after all compelled to accept Pareto plus –​we can set it aside if that’s what we at the end of the day want to do.

Why worry about Pareto plus? Of course, some philosophers will reject out of hand any thought that Pareto plus is disturbing in any way. They may not fully recognize the intuitive and practical concerns that weigh against Pareto plus. I  should thus underline those concerns before proceeding. It seems clear to me that the theory that instructs that the diminution of wellbeing of any one existing or future person4 makes a given future better or an otherwise permissible choice wrong or an otherwise wrong choice permissible or obligatory provided just that it’s done with the effect of bringing still another person into existence whose wellbeing gain at least counterbalances the wellbeing loss of the one person can’t be correct. Now, the addition described here isn’t, of course, mere addition. Rather, it comes on the condition that the one person is made worse off –​sustains, that is, a wellbeing loss.That means that Pareto plus itself doesn’t apply. However, if our otherwise plausible moral theory is already committed to Pareto plus –​to the position that mere addition, other things equal, makes things better –​ then it’s hard to see how that same theory can consistently tell us that addition, if done at any cost at all, suddenly, magically, is completely neutralized in terms of its morally salutary effects –​that suddenly, magically, that addition has no value at all and no capacity to make things better. It’s that extension of Pareto plus that puts climate ethics at odds with itself.According to climate scientists, one of the most carbon intensive choices individual agents often can make is to produce an additional child.5 But the wellbeing costs imposed by that choice –​the often minute losses of wellbeing sustained by many existing and future people –​will often be, we can estimate, more than counterbalanced by the wellbeing gains that the additional child can be expected to accumulate over the course of that child’s lifetime.6 Taking steps to reduce population growth becomes, on an extended Pareto plus, a far less morally credible tool for addressing climate change than it otherwise would be.The call to arms to help save our poor planet by way of avoiding overpopulation is effectively muted –​if not silenced altogether. But, even in its original, unextended form, Pareto plus produces disturbing results. Where the choice of contraception or early abortion affects no one other than (arguably) the potential child, Pareto plus implies that it’s a choice that makes things worse –​and a choice that is, accordingly, wrong.7 The woman is obligated to produce the (perhaps not yet even conceived) child. Things become still more troubling under the extended principle. For it will rarely –​and perhaps never in the real world –​be the case that the production of the additional child comes at such a high wellbeing cost to others –​e.g. the woman –​that that cost will not be more than counterbalanced by the wellbeing gain that existence itself can be expected to shower upon the child. Either way, we lose any moral foundation for the constitutional right of procreative privacy.8

Population ethics and Pareto plus  71

Thus much is at stake. If we do decide, at the end of the day, to reject Pareto plus, Parfit’s discussion in Part Four –​as we shall see –​nicely reveals the price we must pay. Two points in particular are of interest –​in part because they come as something of a surprise and in part because they tell us something about the structure of morality. First, we will need to accept that mere addition doesn’t just not make a given future better than another; rather, it often makes the one future worse than the other. And, second, while the better chance of the worth-​having existence can, in one sense, make things better, the bare fact of that existence on its own has no such effect.

Does mere addition make things better? Does it make things worse? The Two Outcome Case. Consider the following case Before proceeding, I  should say a word about how Graph 4.1 and later graphs are to be read. In the Two Outcome Case, c1 and c2 are choices available to the agents at a given time, and f1 and f2 are the outcomes, or possible worlds or futures that those choices, respectively, may bring about. c1 and c2 are mutually exclusive and exhaust the agents’ available choices. f2 is accessible relative to f1 as of that moment just prior to choice, given the history that f1 and f2 share and in light of any constraints on how any given future might then unfold; other futures, though they may be possible, remain inaccessible. (Depending on the notion of accessibility we think should be at play in this context, such constraints may be imposed by limitations on the abilities and resources available to the agents, by the laws of physics that govern the universe or by still other principles or conditions.9) Typically, whether a given accessible future will unfold under a given choice is undecided, or at least unknown, as of the moment just prior to choice. That, in other words, a given choice will give rise to a future that displays one collection of morally relevant features rather than an alternate future that displays an alternate collection of morally relevant features is typically a matter of probability. In the Two Outcome Case, however, it’s stipulated that the probability that f1 will unfold under c1 is exactly 1 and ditto for c2 and f2. The notion of wellbeing is itself left undefined –​perhaps it’s a matter of happiness; perhaps capability to achieve functionings –​except to say that (i) the future that offers the person more wellbeing will be the future that is better for that person, (ii) it’s raw wellbeing that is at stake GRAPH 4.1 Two Outcome Case

Probability Wellbeing +5 +0

c1

c2

1 f1 p1 –​ pn Ben*

1 f2 p1 -​pn, Ben

72  M. A. Roberts

here, unadjusted for values beyond that which makes life go well for the one who lives10 and (iii) a person’s wellbeing level in a future where that person never exists is, by assumption, just zero.11 A  person’s name written in bold means that that person does or will exist at the indicated future, and italics along with an asterisk that that person never exists at all. Here, then, it’s Ben’s coming into existence, and only Ben’s coming into existence, that is at stake.The rest of the crowd –​p1 -​pn –​ will exist at their nice little +5 wellbeing level however the choice under scrutiny is made. f1 and f2 are, then, exactly alike in all their morally relevant respects except that Ben exists in f2 but not in f1. Thus it’s a very simple case.The very simple question, then, we want to ask is just this: does Ben’s existence in f2 make f2 better than f1?

Pareto plus Pareto plus immediately answers that question in the affirmative. Pareto plus.Where a future x contains exactly the same people as an accessible future y and each person in x has exactly the same wellbeing level as that (same) person has in y except that an additional person exists in y and has an existence worth having in y, then y is morally better than x, that is, x is morally worse than y. It’s fascinating to me that, at that juncture in Reasons and Persons where we might most clearly expect Parfit to take a stand on Pareto plus or apply that principle to get to a conclusion we know he is after –​that is, in the chapter on the Mere Addition Paradox –​he says almost nothing about it at all.12 I see that as a brilliant display of restraint on Parfit’s part. It’s an indication that he understands just how contentious Pareto plus really is and elects to stay away from it. For Pareto plus is directly at odds with the person-​based approach in ethics and, in particular, with the person-​based intuition, that is, PBI. And that’s so, no matter how carefully we formulate PBI. Thus it seems to me that the most plausible versions of PBI –​those, that is, we should have any interest in defending –​are very narrow in scope. Specifically, they provide a condition just on when things are worse, not on when they are better. And the condition they provide is just a necessary condition on things being worse, not a sufficient condition.13 Thus consider the Very Narrow Person Based Intuition (VNPBI): Very Narrow Person Based Intuition (VNPBI):    A future x is worse than an accessible future y, only if there is a person p such that        p does or will exist in x and        x is worse for p than …; and

Population ethics and Pareto plus  73

    A choice c made at x is wrong, only if there is a person p and an    alternate choice cʹ at an alternate accessible future y such that        p does or will exist in x and        x is worse for p than y is. Now, VNPBI itself is obviously incomplete. We will turn to the question of just how the ellipsis in VNPBI is to be filled in shortly. Even as it stands, however, VNPBI tells us a lot. It tells us that there’s no moral deficiency in a future that makes things worse for a person only by way of leaving that person out of existence altogether.14 Applied to the Two Outcome Case, it tells us that it’s not the case that f1, where Ben never exists, is worse than f2 –​that is, that it’s not the case that f2, where Ben exists, is better than f1. It tells us, in other words, that Pareto plus is false. Putting the matter the other way around, Pareto plus tells us that VNPBI is false. Of course, elsewhere in Reasons and Persons –​in the chapter on the Nonidentity Problem –​Parfit explicitly sets out to destroy PBI. But the restraint he displays in constructing the Mere Addition Paradox without appealing to Pareto plus suggests that he did not think that an argument that, at least sotto voce, instructs that PBI is false –​an implication that he would need to accept if he meant to appeal to Pareto plus –​could play a role in the construction of an interesting or important paradox. PBI just has too firm a hold, and Pareto plus is just too contentious. Insisting on Pareto plus at this point of the text would strip the paradox of its paradox, leaving us with an argument to inconsistency that we could far too easily avoid by simply remaining neutral on Pareto plus. Parfit’s text thus at that critical juncture in Reasons and Persons leaves PBI  –​ including VNPBI –​untouched, unmentioned, unchallenged. At that critical juncture, Parfit refrains from promoting Pareto plus. At that critical juncture, he provides us with no basis at all to think that the theoretical apparatus he wants to put to work in connection with his construction of the Mere Addition Paradox relies on Pareto plus or that it rules out VNPBI. We thus seem at that juncture of Reasons and Persons perfectly free –​if we’re so inclined –​to retain VNPBI; to reject Pareto plus; and still to walk with Parfit down the garden path to the Mere Addition Paradox and from there on to the Repugnant Conclusion.

Mere Addition Principle (MAP) Instead of relying on the contentious Pareto plus, Parfit constructs the Mere Addition Paradox using the seemingly more modest, and less contentious, mere addition principle (MAP). Mere addition principle (MAP). Where a future x contains exactly the same people as an accessible future y and each person in x has exactly the same wellbeing level as that person has in y except that an additional person exists

74  M. A. Roberts

in y and has an existence worth having in y, then it’s not the case that y is worse than x. So let’s focus on MAP. Let’s shelve, for the moment, the question whether mere addition (other things equal) makes things better and focus just on the question whether mere addition (other things equal) makes things worse. Here, Parfit is very definite. Having said little about whether mere addition makes things better, he’s very definite that it doesn’t make things worse. MAP is most famously put to work in Parfit’s construction of the Mere Addition Paradox. The Mere Addition Case (Graph 4.2) can be used to demonstrate the paradox. Here, MAP implies that it’s not the case that A+ is worse than A. And for purposes of constructing the Mere Addition Paradox, that’s close to all that Parfit needs –​that innocuous, inoffensive, unobtrusive, tiny little point; not that A+ is better than A but rather that A+ isn’t worse than A. For we then can plausibly note that A+ is worse than B. But if A+ isn’t worse than A, but A+ is worse than B, then B isn’t worse than A. Yet –​we can also plausibly note –​B is worse than A. Thus the paradox: both B isn’t worse than A and B is worse than A.15 Other principles seem quietly at play here. For example, to obtain the result that B isn’t worse than A, we seem to need to know, not just that A+ isn’t worse than A, but also (the arguably distinct claim) that A+ is at least as good as A.16 But the key ingredient here seems to be simply MAP. Whence Parfit’s confidence in MAP? His main strategy is to rule out the most plausible among the competing principles that might be thought to undermine MAP, principles that would tell us that A+ is worse than A. And he interlaces that line of defense with appeals to raw intuition. Thus the “average principle” instructs that A+ is worse than A. But Parfit rejects that principle, as should we.17 As Parfit notes, “research in Egyptology cannot be relevant to our decision whether to have children.”18 Considerations of equality might seem to provide another basis for the claim that A+ is worse than A. Such considerations can take different forms. The egalitarian

GRAPH 4.2  Mere Addition Case

probability wellbeing +10 +5 +1 +0

c1

c2

c3

1 A A people

1 A+ A people

1 B A people and A+ people

A+ people A+ people*

Population ethics and Pareto plus  75

might, for example, favor maximin, the idea that the better outcome is the one in which the least well off are better off. But, as Parfit notes, plausible versions of maximin “cannot support the claim that A+ is worse than A.”19 Here, it’s not the underlying principle, but rather the inference, that fails. It’s just not plausible to think that A, which leaves the least well off out of existence altogether, makes things better for those people. The egalitarian might, instead, propose that the brute inequality we see in A+ as compared against the perfect equality we see in A  is what makes A+ worse than A.  Here the inference succeeds and it’s the underlying principle that Parfit objects to: “When inequality is produced by Mere Addition” –​when, e.g., additional people (the A+ people) are added to A to produce A+20 –​“it does not make the outcome worse.”21 Though we might find the argument itself a little thin, we may well share Parfit’s intuition. Surely our allegiance to equality isn’t so strong that any inequality produced under the conditions of mere addition will necessarily render one outcome worse than another. Even many egalitarians consider equality of wellbeing just one factor, among perhaps many, in deciding overall goodness.22 And to get the paradox off the ground, all Parfit needs is one case, one set of specifications, where other factors are sufficient to leave us unwilling to say that the inequality on its own dictates that A+ worse than A. Having countered some of the significant resistance to MAP, Parfit is ready to defend that principle. In that context, his language is sometimes cautious. (Thus he writes: “Suppose we believe that A+ is not worse than A.”23) And sometimes it’s a raw appeal to intuition and not cautious at all. (Thus he asks: “Can we honestly claim to believe, of the extra group in A+, that it would have been better if they had never existed?”24) Either way, the Mere Addition Paradox is launched, without a word in favor of the idea that A+ is better than A.25 Relying, that is, mainly on MAP, and without a word of support for Pareto plus, Parfit constructs the paradox. In many cases, MAP seems perfectly plausible. Thus in the Two Outcome Case we are confident –​however appealing we find VNPBI –​that Ben’s existence in f2 doesn’t make f2 worse. (Can we “honestly claim” to believe that f2 is worse than f1? No.) Since, at that moment, we don’t feel compelled also to say that Ben’s existence makes things better, we are left with the impression that we are perfectly free to retain VNPBI. That is all very cozy. We think we are on stable ground, and we may think we have even found common ground, when we say “let’s stay neutral on Pareto plus –​ and thus refrain from rejecting VNPBI out of hand –​but welcome MAP with open arms and see where it takes us.” But of course one case –​the Two Outcome Case –​does not a principle make. The ground shifts beneath us when we begin to make inquiries in connection with still other cases. Consider, for example, the Three Outcome Case.

76  M. A. Roberts GRAPH 4.3 Three Outcome Case

Probability Wellbeing +10 +5 +0

c1

c2

c3

1 f1

1 f2

p1 … pn Ben*

p1 … pn, Ben

1 f3 Ben p1 … pn

The Three Outcome Case –​at least on its face –​is just like the Two Outcome Case except we’ve added a third alternate choice c3 and third accessible future f3, where f3 is still better for Ben than f2. John Broome’s discussion of his own version of the Three Outcome Case was not aimed at demonstrating the linkage between MAP and Pareto plus. That discussion was instead aimed to show that what he calls the neutrality intuition is false.26 But the line of reasoning he proposes in that discussion nonetheless can be reworked, as follows. We start by assuming MAP in order to demonstrate the conditional claim that MAP implies Pareto plus, and we then use reductio to show that the consequent of that conditional, that is, Pareto plus, is true. Thus: 1. MAP (assumption) 2. Pareto plus is false; the mere addition of the worth-​having existence doesn’t make things better (assumption) On the basis of those assumptions, and taking for granted that to say that it’s not the case that x is worse than y is to say that x is at least as good as y (an assumption I do not contest here27), we then say: 3 . f2 is at least as good as f1 (from (1)) 4. f3 is at least as good as f1 (from (1)) 5. f1 is at least as good as f2 (from (2)) 6. f1 is at least as good as f3 (from (2)) 7. f1 is exactly as good as f2 (from conceptual truth, (3), (5)) 8. f1 is exactly as good as f3 (from conceptual truth, (4), (6)) 9. f2 is exactly as good as f3 (from conceptual truth, (7), (8)) 10. f3 is better than f2 (from Same-​People Pareto, an uncontroversial Pareto principle28) 11. Inconsistency (conceptual truths, (9) and (10)) _​_​_​_​_​_​_​_​_​_​_​ 12. If MAP is true, then so is Pareto plus (by reductio and then conditional proof) A commitment to MAP thus implies a commitment to Pareto plus. And thus the close proximity of MAP and Pareto plus.29

Population ethics and Pareto plus  77

So, while at this critical point in Reasons and Persons  –​in the chapter on the Mere Addition Paradox –​Parfit’s text omits any discussion of the principle we are here calling Pareto plus, the subtext that we have just excavated, thanks to work that Broome has done since Reasons and Persons first appeared, shows that Parfit’s seemingly modest, explicit commitment to MAP in fact compels us to accept the highly contentious Pareto plus as well. As noted earlier, Pareto plus itself rules out PBI –​including, specifically,VNPBI. Now that we see the connection between MAP and Pareto plus, should we complain that Parfit’s acceptance of MAP in effect begs the question in favor of Pareto plus and against PBI and, more generally, against a person-​based approach in ethics? Whether something is question-​begging is a matter not of science but of art. Whether we think that a MAP-based account of the Three Outcome Case begs important questions will depend on whether we think we are compelled to accept MAP. And many philosophers do feel compelled to accept MAP. But it’s a little mysterious to me why that’s so. For we know where MAP is trying to take us –​first to the Mere Addition Paradox and then farther on down the garden path to the Repugnant Conclusion. In any case, one thing we can clearly conclude is this: MAP is, after all, every bit as contentious as Pareto plus. Now, in itself that point of course isn’t an argument against MAP. But it does clearly suggest a viable alternate way of avoiding the inconsistency of line (11): we can reject both Pareto plus and MAP. That, in turn, opens the door for a person-​based account of the case. The question then becomes which of the two accounts –​the person-​based account or the MAP-based account –​is the more plausible. Let’s, accordingly, compare those two accounts. As already noted, MAP implies that f2 is at least as good as f1. And we know, by Same-​People Pareto, that f3 is better than f2. We thus infer that f3 is better than f1. The idea that the evaluation of the agents’ choices is connected in some important way to the ranking of futures adds to this picture. On that view, c1 and c2 are both wrong while c3 is obligatory. In contrast, VNPBI tells us that f1 is at least as good as f3 and that f3 is at least as good as f1. Thus we infer that f1 is exactly as good as f3. In addition, VNPBI implies that f1 is at least as good as f2, there being no one in f1 that f1 makes things worse for. But as it stands –​ellipsis intact –​VNPBI tells us nothing about whether f2 is at least as good as f1.To complete the person-​based account of the case, we first decide how the ellipsis in VNPBI is to be completed. It might seem that –​and many formulations of PBI take for granted that –​we should say that x is worse than y only if x is worse for (the existing or future) p than y. But if we complete the ellipsis in that narrow way,VNPBI in effect disqualifies itself. For VNPBI then generates the result that f2 is at least as good as f1 –​and with a little further work that f2 is exactly as good as f3, a result that is clearly false. If we

78  M. A. Roberts

think that the key to comparing f2 against f1 under a person based approach is to compare how well off Ben is in f2 against how well off Ben is in f1, we can’t avoid that clearly false result. If, on the other hand, we think that the key –​the person-​based key –​to comparing f2 against f1 is to compare how well off Ben is in f2 against how well off Ben is in each alternate accessible future, then we easily avoid that result. The latter is to complete VNPBI’s ellipsis in a more expansive way. There’s good reason to think that that reading alone is consistent with the person-​based approach. Comparing how well off Ben is in f2 only against how well off Ben is in f1 completely blinds us to the one obvious person-​based moral deficiency in f2, a deficiency that has nothing to do with how Ben fares in f1 and everything to do with how Ben fares in f3. A well-​formulated PBI should have the resources to take that deficiency into account. Expansive Very Narrow Person Based Intuition (EVNPBI) does just that.    Expansive Very Narrow Person Based Intuition (EVNPBI):       x is worse than y, only if there is a person p and an alternate accessible future z such that        p does or will exist in x and x is worse for p than z (where z may, but need not, be identical to y); and      A choice c made at x is wrong, only if there is a person p and an alternate choice cʹ at an alternate accessible future y such that         p does or will exist in x and         x is worse for p than y. EVNPBI’s necessary condition on f2’s being worse than f1 is nicely satisfied in the Three Outcome Case. Thus EVNPBI is perfectly consistent with (though doesn’t itself imply) the result that f2 is worse than f1. Moreover, EVNPBI, like VNPBI, easily instructs that f1 is exactly as good as f3. We can then appeal to Same-​People Pareto to determine that f2 is worse than f3 –​and from all that that f2 is, after all, contrary to Pareto plus and to MAP, worse than f1. That account of the Three Outcome Case seems far more plausible than the MAP-based account. We are surely obligated to treat the people we do bring into existence, other things equal, well. But that doesn’t mean that we are obligated to bring those same people into existence to begin with. Despite the fact that PBI has little chance of success unless it’s constructed as EVNPBI, it seems unlikely that Parfit himself considered that construction a viable option. Elegant writer that he was, Parfit himself rarely completed the ellipsis on the printed page.30 He may well have taken it for granted that we would all understand the ellipsis should be completed narrowly. Indeed he may have taken it for granted that that’s how the completion of that ellipsis must go.

Population ethics and Pareto plus  79

But in fact Parfit only glancingly addresses the issue. Later in his chapter on mere addition, he briefly argues that how A+ compares against A can’t have anything to do with the availability of B. (So: the average principle, as we have already noted, doesn’t work for that purpose, neither do considerations of equality, and now neither does appeal to B.) Attempting to ground that point, Parfit suggests that the claim that how we rank A against A+ may vary depending on B’s availability confuses the evaluation of choices with the ranking of outcomes, that is, possible worlds, or futures. Thus “[w]‌hether I ought to act in one of two ways may depend on whether it would be possible for me to act in some third way.”31 He then underlines that his topic is not the evaluation of actions, or choices, but rather the evaluation of outcomes.32 And, as to the comparison of one outcome against another outcome, he writes, “relative goodness … cannot depend on whether a third outcome, that will never happen, might have happened.”33 Here it’s at least possible that Parfit meant to invoke a version of the principle of the Independence of Irrelevant Alternatives (Independence).34 The problem with that argument is that accepting EVNPBI doesn’t in fact put us at odds with Independence. Let’s back up. The account of the Three Outcome Case that appeals to a collection of principles that includes EVNPBI and other principles that are themselves consistent with a person-​based approach  –​the account of the case we sketched above –​implies that f2 is worse than f1. And for the Two Outcome Case that same collection of principles implies that f2 is exactly as good as f1. Regardless of what we think about Independence, it may seem that this inconsistency on its own is enough to show that we must reject EVNPBI. But that’s a mistake. Parfit wants to focus on outcomes, that is, possible worlds, or futures. So let’s do that.The problem for Parfit’s argument is that worlds come with all their features intact. They have their features necessarily, and any distinction in detail between worlds x and y means that x and y are numerically distinct. We then ask the question what makes f3 an available, or accessible, outcome in the Three Outcome Case but not in the Two Outcome Case. If agents are involved, it would seem to be that the agents in f1 in the one case but not the other had (at the relevant time, just prior to choice) the ability and the resources to make f3 happen in place of f1. If agents aren’t involved, it would seem to be that there’s some sort of necessity (physical; metaphysical) in f1 that rules f3 out in the one case but not the other. Either way, whether f3 exists as an alternative accessible outcome relative to f1 is a fact that is itself embedded in f1. The upshot is that the “f1” we talk about in the one case has features that the “f1” we talk about in the other case doesn’t share –​that is, that our commonly titled “f1’s” are in fact numerically distinct. Thus we avoid the inconsistency.35 If that’s correct, then that means that we also avoid any violation of Independence. Thus our evaluation of f1 against f2 doesn’t change as we move from one case to the other; it doesn’t depend on whether f3 exists as an accessible outcome. f1 remains exactly as good as f2 in every case, including the Two Outcome Case, since there is no case, and can be no case, where f3 exists as an outcome accessible or available to f1 or f2. Moreover –​and now, for clarity, we adopt distinct constants for discussion of the Three

80  M. A. Roberts

Outcome Case –​f2' remains worse than f1' in every case since there is no case, and can be no case, in which f3 doesn’t exist as an outcome accessible or available to f1' or f2'.36 Pareto minus As we have just seen, in accepting the seemingly innocuous MAP we must both (i) accept the far more contentious Pareto plus and (ii) reject EVNPBI. The better view, it seems to me, is to reject MAP and acknowledge that, depending on our facts, the mere addition of a worth-​having existence can, other things equal, make things worse. This is to accept a principle we can call Pareto minus. Pareto minus shouldn’t be confused with the highly implausible notion that mere addition always, or almost always, makes things worse. For example, Pareto minus doesn’t claim that in the Two Outcome Case the mere addition of Ben to f2 makes f2 worse than f1. The claim rather is that mere addition sometimes makes things worse. For example, in the Three Outcome Case less is more: f1 is exactly as good as f3, while f2 is worse than both. It’s important to note that, in the Three Outcome Case and other cases where we want to say mere addition makes things worse, the future that proves that point –​f3 in the Three Outcome Case –​is a future that is not just possible but also accessible in the relevant sense; it’s a future that is not ruled out either by limitations on what the agents themselves have the ability and the resources to achieve or by the physical necessities that themselves govern how the future might unfold. Without that restriction, we would seem compelled to say that f2 is worse than f1 in the Two Outcome Case and that f2 is worse than f1 –​or more precisely that f2' is worse than f1' –​in the Three Outcome Case. For it seems that there will always be an alternate possible future, a logically possible if inaccessible future, that will make people at least a little better off than they are in any given future. The restriction, then, is critical.

If a better chance of existence makes things better, does the actual fact of existence also make things better? The Better Chance Claim We’ve just identified –​and rejected –​one subtextual argument for Pareto plus. In this part, I identify a second subtextual argument to that same conclusion. Here, it’s Parfit’s chapter on the Nonidentity Problem that we focus on. The claim that launches the argument  –​the claim that Parfit never explicitly identifies in his chapter on nonidentity but that we can now understand, decades into population ethics, as critical to his construction of the problem itself37 –​is the Better Chance Claim. Better Chance Claim: A better chance of existence can make things in some sense better; at least, it can make an otherwise wrong choice permissible. This claim seems at least plausible –​indeed, as we shall see, it’s a claim that we seem compelled to accept.

Population ethics and Pareto plus  81

What is of interest about this claim is how immediately it grounds an argument for Pareto plus. For it seems only natural to suppose that, if the better chance of existence can make things better –​if, that is, the higher probability that a given person will exist can make things better –​then surely the actual fact of existence will make things better still. Of course, the text of Parfit’s chapter on the Nonidentity Problem is all about PBI –​and why we must reject that intuition. I want to argue, however, that the subtext of that chapter is –​at least in large part –​all about the Better Chance Claim.38 Indeed, it seems to me that what gives Parfit’s chapter on the Nonidentity Problem its special zing is that the Better Chance Claim –​in contrast to MAP and certainly in contrast to Pareto plus –​really does constitute stable, common ground. We seem compelled to say that the better chance of existence in some sense can make things better –​at least, that a better chance of existence can convert a choice that would otherwise be wrong into a choice that is permissible. The Better Chance Case makes that point. Here, c1 –​the choice to take a fertility pill –​creates a certain small but significant probability –​0.1 –​that Harry will be conceived and born and have a wellbeing level of +8 –​that is, that f1 will eventuate in place of f2, where poor Harry never exists at all. In contrast, c2 creates a far smaller probability –​the relatively insignificant 0.0001 –​that Harry will be conceived and born –​that f3 will eventuate in place of f4. Now, it’s true that c1 at f1 makes things worse for Harry than c2 at f3 does. But the probability that Harry will eventually exist is much higher under c1. One factor seems clearly to make up for the other. And we conclude that c1 is permissible. I take that to be something we’d all agree on. Even person-​based theorists who would insist that there is nothing at all wrong with c2 –​ that c2 is perfectly permissible, whether performed at f3 or f4 –​are not going to claim that c1 is wrong. Thus the better chance of existence that comes with c1 –​in some sense –​makes things better. At least: it’s converted an otherwise wrong choice into a permissible choice. After thinking about how the logic of the most powerful versions of the Nonidentity Problem actually work –​those based e.g. on Parfit’s own risky policy and depletion cases, Kavka’s pleasure pill and slave child contract cases, cases

GRAPH 4.4  Better Chance Case

c1: take fertility pill probability wellbeing +10 +8 +0

0.1 f1

c2: take aspirin 0.9 f2

0.0001 f3 Harry

0.9999 f4

Harry Harry*

Harry*

82  M. A. Roberts

involving climate change and historical injustice and many others –​it seems clear to me that the construction of the arguments based on those cases makes use, however quietly, of the Better Chance Claim.39 To see that point, consider, for example, Kavka’s Pleasure Pill Case. The argument that gives rise to the nonidentity problem may already be familiar but an outline of that argument is in order here. We are to suppose the parent’s choice to take the pleasure pill in fact results in the coming into existence of a child, Jaime, whose wellbeing level is lower than the wellbeing level of any of the distinct possible children (Katie 1, Katie 2 and so on) who might have existed in place of Jaime had the parent taken the aspirin instead of the pleasure pill and nonetheless conceived a child. The problem is that, had the parent done anything other than what he or she in fact did, surely Jaime would never have existed at all. Since Jaime’s life, by hypothesis, is unambiguously worth living, we seem to have a case in which a choice that is clearly wrong –​the choice to take the pleasure pill –​makes things worse for no existing or future person at all, including Jaime –​ and thus a case that challenges even the most carefully articulated version of PBI, that is, EVNPBI. In fact, however, this version of the nonidentity argument is itself problematic. It includes the claim that “surely Jaime would never have existed at all” had the parent done anything other than what he or she in fact did. But, as the graph itself suggests, that’s an overstatement. The case can’t plausibly include the claim that Jaime could not have existed had the parent taken the aspirin in place of the pleasure pill –​that it’s, e.g., inconsistent with the laws of physics that govern the universe for Jaime to exist and for the parent to have taken the aspirin rather than the pleasure pill. The case can’t, in other words, plausibly include the claim that the future in which the parent takes the aspirin instead of the pleasure pill yet Jaime still exists, is inaccessible GRAPH 4.5  Pleasure Pill Case: Usual Construction

probability wellbeing

c1: parent takes pleasure pill

c2: parent takes aspirin

0.1 f1

0.0001 f3

0.9999 f4

Jaime

Katie 1 or Katie 2 or …

Katie 1* or Katie 2* and …

Jaime*

0.9 f2-​1 or f2-​2 or …

+10

+8

Jaime

+0

Katie 1* and Katie 2* and …

Katie 1 or Katie 2 or … Jaime*

Population ethics and Pareto plus  83

relative to the future in which the parent takes the pleasure pill instead of the aspirin.40 The point, rather, can only be that it’s highly probable that Jaime would not have existed had the parent taken the aspirin in place of the pleasure pill. Now, that point does not on its own show how we can avoid the result that the choice of the pleasure pill doesn’t make Jaime worse off. Still, it should be clear that our initial description of the case is deficient. To be sure, it’s a deficiency that seems innocuous enough –​one that we may think can be repaired without reducing the overall force of the argument. We need just shift from the surely false claim that Jaime could not have existed under any alternate choice to the entirely plausible claim (and the claim that Graph 4.5 reflects) that Jaime very probably would not have existed under any alternate choice.41 And it may seem perfectly clear that that repair indeed does not reduce the overall force of the argument.That’s so because probabilities, we think, matter morally.Thus the choice to take the fertility pill in the Better Chance Case doesn’t seem to make things worse for Harry; if anything, taking both alternate probabilities and alternate wellbeing levels into account, it makes things better. That’s so, in virtue of the fact that Harry’s chances of ever existing at all would otherwise have been so remote. It may seem that we can say the same thing about the choice to take the pleasure pill in the Pleasure Pill Case. The lower wellbeing Jaime has under that choice is counterbalanced by what we may take to be –​what we are, that is, supposed to take to be –​the higher probability, under that choice, that Jaime will ever come into existence at all. Person-​based theorists  –​philosophers who think that worseness and wrongdoing must be tied to things being made worse for some existing or future person –​ have a very hard time with these sorts of cases since the choices under scrutiny, once we take the probabilities into account, seem worse for no existing or future person at all. The question of how the probabilities in fact play out in the Pleasure Pill Case is one we shall put aside for the moment and return to below. For the moment, our sole question is whether the Better Chance Claim itself compels us to accept Pareto plus. Does the fact that a better chance of existence can make things better itself mean that the actual fact of existence can make things better as well? We avoided Pareto plus above by denying MAP in favor of Pareto minus and EVNPBI. But a parallel strategy  –​which would be to deny the Better Chance Claim –​isn’t available to us. For the Better Chance Claim, as we’ve seen, is, unlike MAP, not a claim we can plausibly deny. Here, the appropriate strategy will be to take a closer look at both EVNPBI and the Better Chance Claim to see whether they can be fit together. We turn to that work now. Revising EVNPBI to reflect the Better Chance Claim As EVNPBI is written, it doesn’t generate any results at all regarding how f3 compares against f1 in the Better Chance Case. The fact that f3 exists as an accessible future in that case and that f1 is worse for Harry than f3 means that the necessary condition that EVNPBI sets forth is satisfied.

84  M. A. Roberts

At the same time, the Better Chance Case does make an important point: that EVNPBI fails to capture a critical aspect of PBI. The Better Chance Claim itself is compelling, and it’s reasonable to think that any adequate formulation of PBI must embrace that claim. More simply: an adequate formulation of PBI should be strong enough to imply that, in the Better Chance Case itself, c1 at f1 is –​given the probabilities and notwithstanding Jaime’s +8 wellbeing level at f1 –​permissible. So that’s one task: to make sure that we’ve captured the heart of the person-​ based intuition and not just the spleen. But the more central task we face at this point is something else entirely. It’s to determine whether the revision of EVNPBI that is able to make sense of the Better Chance Claim also compels us to accept Pareto plus. Must we, after all, accept Pareto plus? Or can we instead revise EVNPBI in a way that explains how both the better chance of existence, in some sense, makes things better and the fact of mere addition doesn’t make things better? I think we can. Let’s start by reformulating EVNPBI. First, a definition. Where a choice c made at a future x creates a probability n that p will have the wellbeing level (WB) that x in fact assigns to p, we can say that x’s probable value (PV) for p under c is n(WB).42 Then, the principle itself: Expansive Very Narrow Person Based Intuition + Probable Value (EVNPBI+PV):    x is worse than y, only if there is a person p and an alternate accessible future z such that        p does or will exist in x and       x is worse for p than z (where z may, but need not, be identical to y); and    c at x is wrong, only if there is a person p and an alternate choice cʹ at an alternate accessible future y such that        p does or will exist in x and        x is worse for p than y and        PV of c at x for p < PV of cʹ at y for p. This principle allows us to have our cake and eat it too. Let’s first note that the revised principle nicely supports the Better Chance Claim. According to EVNPBI+PV, c1 at f1 is wrong only if the probable value of c1 at f1 for Harry is less than the probable value of c2 at f3 for Harry. But it isn’t; the condition is therefore failed; and we get the result that c1 at f1 is permissible, which seems exactly right.

Population ethics and Pareto plus  85

And let’s note, second, that there’s nothing in the revised principle that compels us to adopt Pareto plus. Consistent with the revised principle, we can, in other words, insist what on reflection seems entirely plausible: the fact that c1 is permissible in the Better Chance Case doesn’t mean that the future in which Harry exists and is worse off –​that is, f1 –​is better than or even at least as good as the future in which Harry exists and is better off –​that is, f3. Specifically, the fact that Harry’s probable value is greater under c1 doesn’t imply, under EVNPBI+PV, that there’s anything wrong with c2. Rather, in EVNPBI+PV we see multiple routes to permissibility.The person we are concerned about –​here, Harry  –​might never exist at all; might exist at a maximized wellbeing level; or might exist at a maximized PV level.43 The choice in each such scenario is one that EVNPBI+PV deems permissible. The upshot, then, is just this. There’s some important sense in which the better chance of the worth-​having existence does make things better, even as there’s another important sense in which the worth-​having existence doesn’t make things better. We are thus free to say that c1 –​just like c2 –​is permissible but that the future in which Harry (against the odds) exists at +8 is still worse than the future in which Harry (against even greater odds) exists at +10 and that the futures in which Harry never exists at all are exactly as good as the future where he exists at +10. That minor disconnect between how we evaluate choices and how we compare futures in respect of their overall moral goodness, if anything, seems clarifying rather than problematic. We could take the position that the probabilities embedded in f1 (given c1) themselves make f1 better than f3 (given c2) is. But then we obfuscate what may well be a morally important fact about f1 and f3: that f1 is worse for Harry than f3 is. Pleasure Pill Case revisited Despite the careful revisions reflected in EVNPBI+PV, it’s still a principle we will be forced to reject if it generates the result that the parent’s choice to take the pleasure pill in the Pleasure Pill Case is permissible. The question is whether our reasoning about the Pleasure Pill Case should simply track the reasoning that we readily accepted in connection with the Better Chance Case. It seems widely accepted that it should. Harry’s better chance of existence under the choice to take the fertility pill meant that that choice was permissible. So Jaime’s better chance of existence under the choice to take the pleasure pill means that that choice is permissible as well. That’s, at least, how it’s widely accepted we are to reason in the Pleasure Pill Case. We are supposed to understand that the Pleasure Pill Case nicely counterexamples any carefully formulated, reasonably complete version of PBI, including EVNPBI+PV. We are supposed to agree that the probability that Jaime will exist under the choice to take the aspirin is so small that the probable value of the choice to take the pleasure pill –​the choice in fact made –​has to be greater than that (despite the fact that Jaime’s wellbeing level, if he exists at all, will be lower). The main purpose of the present chapter isn’t to dissect the Nonidentity Problem. But it will hardly do to let the idea that the Pleasure Pill Case stands as an

86  M. A. Roberts

obvious and widely accepted counterexample against any well-​formulated version of PBI, including EVNPBI+PV, go unchallenged. Thus, I  will just note here that the way we are supposed to reason about the Pleasure Pill Case badly misunderstands, or misrepresents, the probabilities at stake. It’s thus a fallacy to think that, just because Jaime in fact exists under the choice to take the pleasure pill, that Jaime’s chances of coming into existence under that choice were any greater than Jaime’s chances of coming into existence under the choice to take the aspirin instead. Whether the parent chooses the pleasure pill or the aspirin, Jaime’s chances of ever coming into existence at all remain ridiculously low. Perhaps this point can be most clearly seen when we contrast the Pleasure Pill Case against the Better Chance Case. In the latter case, Harry’s chance of existence really is increased under the parent’s choice to take the fertility pill rather than the aspirin. That’s the whole point of taking the fertility pill. If the pleasure pill were a fertility pill, we wouldn’t be so clear that the agent can’t permissibly choose to take the pleasure pill. But it’s not. It doesn’t increase Jaime’s chances of coming into existence, and it does guarantee that if Jaime exists, Jaime will be less well off than he (just as easily) might have been. We should, in other words, reject Graph 4.5 above and put in its place a picture that more faithfully represents the probabilities that are in play in the case. On the surface of things, the move from Graph 4.5 to Graph 4.6 is very slight; we’ve just revised a single value, the probability that Jaime will exist, given c1. But from a substantive point of view that revision is highly significant: it means that the probable value of the choice to take the pleasure pill at f1 is after all lower than the probable value of the choice to take the aspirin at f3. That, in turn, means that EVNPBI+PV’s necessary condition on wrongdoing is in fact nicely satisfied –​and

GRAPH 4.6  Pleasure Pill Case: Probability Sensitive Construction

probability wellbeing

c1: parent takes pleasure pill

c2: parent takes aspirin

0.0001 f1

0.0001 f3

0.9999 f4

Jaime

Katie 1 or Katie 2 or …

Katie 1* or Katie 2* and …

Jaime*

0.9999 f2-​1 or f2-​2 or …

+10

+8

Jaime

+0

Katie 1* and Katie 2* and …

Katie 1 or Katie 2 or … Jaime*

Population ethics and Pareto plus  87

the door left open for other person-​based principles to step in and declare the choice to take the pleasure pill wrong.

Conclusions Above, I suggested that Parfit’s work on mere addition quietly argues for a highly contentious conclusion. I argued that accepting the seemingly innocuous MAP –​ the idea that mere addition doesn’t make things worse –​compels us to accept Pareto plus –​the idea that mere addition in fact makes things better –​as well. And that’s a result we should want to avoid –​and one that we can avoid, I have argued, by understanding that the seemingly innocuous MAP is false and should be rejected in favor of EVNPBI and –​with that –​Pareto minus, that is, the idea that mere addition can (on occasion) make things worse. I pointed out that the Better Chance Claim has a critical role to play in constructing the most challenging versions of the Nonidentity Problem.The Better Chance Claim –​unlike MAP –​isn’t one I think we can plausibly deny. As, however, EVNPBI+PV illustrates, there is no reason we can’t accept that claim –​accept, that is, that the better chance of existence in some sense can make things better –​and yet reject Pareto plus –​that is, the further claim that the actual fact of existence, other things being equal, makes things better. Reasons and Persons Part Four has captured our attention from the beginning. Now that we are decades into the work of population ethics, we are in a position to appreciate that with the texts come the subtexts. And we can also appreciate more clearly than ever how important it is to subject all the arguments –​the textual and the modest; the subtextual and the contentious –​to ever-​closer critical scrutiny.44

Appendix A Consider the following case. Incorporating the concept of expected value into PBI  –​in the form of EVNPBI+EV –​seems not to work. Such a principle implies that c1 at f2 is perfectly permissible, expected value having been maximized in virtue of the small chance of a very high wellbeing level being produced by c1. In contrast, EVNPBI+PV GRAPH 4.7  Bad Lives Case (from Dean Spears) (Why we need PV; why EV fails)

c1: take longevity pill probability wellbeing +20,000 +10 +0 -​10

0.009 f1

0.99 f2

c2: take aspirin 0.001 f3 Mindy

0.1 f3

0.9

Mindy Mindy*

Mindy* Mindy

88  M. A. Roberts

performs nicely in this context, providing no basis at all for the result that c1 at f2 is permissible.

Notes 1 Dasgupta 1993, pp. 382–​383. Pareto plus is an implication of any simple form of total utilitarianism (totalism), according to which the value of a given outcome, or possible world or future, is the summation of the raw, unadjusted wellbeing levels, or utilities, of all the people who do or will exist in that future. It is, nonetheless, controversial. I take for granted that the claim that one future is better than another implies that the one future is morally better than another  –​that is, that the reason we are interested in the betterness relation ultimately is grounded in our interest in the nature of morality. 2 I take for granted that our comparison of futures by reference to their overall betterness and our evaluation of choice are strongly connected. When, even in the most straightforward cases, we claim to divorce the effort to say when the future one choice leads to is worse than the future an alternate choice leads to from the evaluation of the choices themselves, it becomes entirely unclear what relation we are talking about when we say that the one future is overall worse than another. We do well, in other words, to retain a strong connection –​though, as I argue below, not a perfect correlation –​between worseness and wrongness. 3 The versions of the Nonidentity Problem that seem to me to be most powerful are those that most clearly challenge PBI  –​the idea, that is, that what is “bad” must be “bad for” someone. Parfit 1987, p. 363. They include versions of the problem based on Parfit’s risky policy and depletion cases, Kavka’s slave child contract and pleasure pill cases, cases involving historical injustices and the case of climate change. See, respectively, RP 351–​379, Kavka (1981), Sher (2005) and Broome (1992). See generally Roberts (2007), (2009) and (2019). 4 I understand the term person to include many but perhaps not all human beings and many non-​human animals as well. Continued consciousness seems critical here for determining when an animal (human or not) counts as a person. 5 Kimberly Nicholas, “A Better Climate Life,” Climate Ethics and Future Generations Kick Off Conference (Stockholm Institute for Future Studies (IFFS) Sept. 2018). 6 The potential parents themselves stand to accrue more significant wellbeing gains and to sustain more significant wellbeing losses than will other existing or future people.The potential parents, after all, may badly want a child –​or, depending on the circumstances, it may be the case that having a child, or an additional child, may create a significant burden for them. Even if the latter holds, however, that burden often will be substantially outweighed by the wellbeing gain that will accrue to the child if that child is in fact produced. (We take for granted that if the child exists resources will be shared –​by the parents; by the community –​with the child, such that the wellbeing gains accrued as a result of the child’s existence are nicely high. We don’t produce the child and then abandon him or her in a ditch somewhere.) 7 See note 2 above (assumption of important connection between worseness and wrongdoing). 8 A referee for this chapter proposes that there isn’t “any real difficulty in defending both Pareto plus and a constitutional right of procreative privacy.” For we can make the argument that the parents, and not the government, are in the best position to weigh the burdens and benefits of producing an additional child in their own case –​and specifically

Population ethics and Pareto plus  89

to determine whether Pareto plus applies to produce the result that it makes things better to bring the additional child into existence.The law, then, can be structured to recognize that surely correct point –​in effect, then, giving the potential parents the legal right to make the choice. It’s perfectly consistent to accept Pareto plus and to insist on a legal right of privacy; moral law and positive law are two different things. But that position does raise difficulties. It means that the moral evaluation that the potential parents themselves are to perform privately (without state interference) must –​if it’s to be a correct evaluation –​ reflect Pareto plus as well as the extended Pareto plus described above (if the cost to others, including the potential parents, is small relative to the gain the additional child will accrue if that child exists, it makes things better to produce the additional child and would be wrong not to produce the additional child). A system that, by law, leaves the choice to the potential parents to be made in private but also instructs that morality requires that those same potential parents produce the additional child seems highly unstable. There are two cases. In one case, we have the potential parents taking morality seriously and then opting to produce the additional child despite their constitutional right not to do so. In the other case, we have the potential parents not taking morality seriously and taking advantage of their legal right to do what is deemed morally wrong. Both scenarios seem unfortunate to me –​especially if, at the end of the day, we come to understand that there is no sound foundation for thinking that our constitutional right of privacy and what we are morally permitted to do need to be split off from each other in this way –​if we come to understand, in other words, that we should reject Pareto plus and its extension. 9 That a given future is improbable under a given choice does not indicate, however, that it’s inaccessible. The future in which the correct combination to the safe is dialed, against all odds, is a perfectly accessible future. This point becomes relevant below (discussion of the Pleasure Pill Case). 10 As the term is used here, wellbeing thus is raw, unadjusted wellbeing, not wellbeing (or the personal good, to borrow Broome’s term) that has been adjusted to reflect, e.g., our values of fairness or equality or, more relevant for purposes here, our existential values. See Broome (2015). 11 The assumption that a person’s wellbeing in any future in which that person never exists is just zero seems plausible –​no burdens, no benefits, no happiness, no misery –​but is controversial. 12 RP 430 (“Some people believe…”). 13 The case of the child whose existence is less than worth having shows that the principle that requires, as a necessary condition on things being better, that a person does or will exist for whom things are better fails.The principle that deems a choice wrong by virtue of the fact that it makes things worse for some of the people who do or will exist under that choice without making things better for any of those people also fails. One choice can be permissible even though it makes worse for everyone who does or will exist under that one choice than an alternate choice does. That can happen, when that alternate choice brings still other people into existence who are worse off than they are under still another alternate choice. Addition Plus is a case that makes that point. See Roberts (2010), p. 65. 14 PBI thus grounds Narveson’s idea that, while morality pushes us to make people happy, it’s completely “neutral about making happy people.” Narveson (1976) p. 73. 15 RP 426. I  have just presented one version of the paradox here. For others, including versions that Parfit believes lead to the repugnant conclusion, see RP 381–​390. For further discussion, see especially Temkin (2012), pp. 426–​427.

90  M. A. Roberts

16 Parfit himself questions whether the fact that x isn’t worse than y implies that x is at least as good as y. He, that is, considers the possibility that x’s not being worse than y might be true in virtue of the fact that x is roughly comparable to (other philosophers might say on par with) y but at the same time not at least as good as y. RP 430–​432. On that view, our options for comparing x against y go beyond the following three: x is worse than y, x is better than y and x is exactly as good as y. While I do think there are cases in which claims of rough comparability seem attractive (see e.g. Parfit’s case of the “one Novelist and two Poets,” RP 431), the additional person cases I consider in this chapter do not seem to me to be of that kind. For the purpose of analyzing those cases, that is, it seems to me that we should accept the principle that, if x isn’t worse than y, then y is at least as good as x. 17 According to the average principle –​that is, averagism –​the value of a given world is the summation of the raw, unadjusted wellbeing levels, or utilities, of all the people who do or will exist in that world divided by that number of people. At least some philosophers have associated averagism “with Mill … who warned against changing the world ‘for the mere purpose of enabling it to support a larger, but not a better or a happier population.’ ” See e.g. Ord, unpublished manuscript (citing Mill, Principles of Political Economics, 1848)[Chap. IV, sec. 3]. It seems, however, that Mill would have had objections to averagism as well as totalism and that it’s at least as plausible to think that a person-​based approach  –​though obviously one that recognizes that it is better to have more well-​off people rather than fewer, i.e., one that counts –​might do a better job capturing his most well-​considered views. 18 RP 420. We have other reasons as well to reject the average principle. See, e.g., Parfit’s “Hell Three.” RP 422. 19 RP 423–​424. The version of maximin that declares A better in virtue of the fact that the extra people in A+ never exist in A seems to Parfit (and to me) implausible. 20 A mere addition is defined by Parfit to mean, among other things, that the additional existence is worth having and no one else is affected. Where someone else is affected –​e.g., made worse off by the addition of the new person –​the addition isn’t a mere addition. RP 420. 21 RP 425. 22 Most, and perhaps all contemporary, egalitarians are, in other words, pluralists. See, e.g., Temkin (2012). 23 RP 426 (emphasis added). 24 RP 427 (emphasis added). 25 The most he says is this: “[s]‌ome people believe that Mere Addition makes the outcome better.They claim that A+ is better than A.” RP 430.The context here, however, suggests, not that this is a principle he thinks we should, or must, accept, but rather that it provides simply an alternate route to the conclusion he wants: that “B is better than A.” 26 In Broome’s version of the case, whether f3 is an alternative accessible future relative to f1 and f2 for the agents –​a future that they have the ability and the resources to bring about –​is left open. He does, however, regard f3 as possible, and one that we may compare in respect of its overall betterness against f1 and f2. Broome (2004), pp. 143–​146. 27 See note 16 above (on rough comparability). 28 According to Same-​People Pareto, when futures f1 and f2 contain exactly the same existing and future people, and f2 is better for at least some of those people and worse for none of those people, f2 is better than f1. 29 As noted earlier, my interest in Broome’s line of reasoning here is not to show that the neutrality intuition is false but rather to show that, if MAP is true, then so is Pareto

Population ethics and Pareto plus  91

plus. To avoid the inconsistency in (11), Broome himself (I think correctly) rejects the neutrality intuition, Broome’s own preferred formulation of the person-​based intuition. Broome (2004), pp. 143–​146. Broome’s formulation implies not just that f2 isn’t better than f1 but also that it isn’t worse. In contrast, the version of PBI I want to defend, the very narrow version,VNPBI, provides no condition on betterness whatsoever.VNPBI instructs only that f2 isn’t better than f1, not that it isn’t worse. 30 See, e.g., Parfit (2017), pp. 118–​119. 31 RP 429. 32 RP 429. 33 RP 429. 34 It’s not certain Parfit means to invoke Independence here; he may just be waving away possibilities that never materialize as irrelevant for purposes of determining how any two outcomes are to be compared. Independence itself is unconcerned with whether a given outcome ever “happen[s]‌” or not. 35 Ben Eggleston notes that the accommodation of Independence that I propose may be ineffective in a case in which we imagine the relevant futures unfolding from a common prior state f0. On the analysis I propose, when agents are able to bring about any of f1, f2 or f3, f2 is worse than f1. But if f3 is “dropped from the option set,” on the analysis I propose f2 is at least as good as f1. What this interesting case shows is that, to make my proposal work, I will need to say that whatever it was in f0 that made agents able to bring about any of f1, f2 or f3 –​and then, when f3 is “dropped from the option set,” unable to bring about f3 – is part of the identity of f1, f2 and f3. And it seems then that we can make the requisite distinction, since surely the “f0” in the case where agents have the ability to bring f3 about is distinct from the “f0” in the case where the agents don’t have the ability to bring f3 about. 36 We can be sure that this way of talking about things will hold up across all our cases since worlds have their properties, in all their detail, necessarily, a fact that grounds what I have elsewhere called the Accessibility Axiom. (Roberts (unpublished mss. a and b). For any worlds x and y, if y is accessible relative to x, then necessarily y is accessible relative to x. 37 See e.g. Roberts (2007) and (2009) and Roberts and Wasserman (2016). For critical discussion, see Green (2016) and Smilansky (2016). For early references to the role of probability in constructing the Nonidentity Problem, see Kavka (1982), p. 100 n. 15. 38 Parfit doesn’t explicitly discuss, or endorse, that claim, which I will call the Better Chance Claim. He seems, unlike Kavka, reluctant to talk about probability, preferring to put his points in counterfactual terms. (E.g.,“[i]‌f we had chosen the other policy, these particular people would never have existed”; and “after one or two centuries, there would be no one living in our community who would have been born whichever policy we chose.”) I have no problem with the counterfactuals Parfit stipulates in the examples he constructs for purposes of challenging PBI. My problem is that (i) he doesn’t need them –​the probabilistic assessments work just as well as the stipulated counterfactuals are meant to work for purposes of the argument –​and (ii) they demonstrably have no bearing on whether a particular person is made worse off by a particular choice. See note 41 below (shoot q in shoulder or heart example). The upshot is that the most lethal among the various versions of the Nonidentity Problem will include the Better Chance Claim. 39 See note 3 above (on different versions of Nonidentity Problem). 40 Thus, the fact that a given future is improbable does not imply that it is inaccessible. See note 9 above. The case where, e.g., the parent is force-​fed the pleasure pill is, after all, a different case altogether; the premise of Kavka’s case is that the parent has the choice whether to take the pleasure pill or do something else entirely –​here, I’ve simply added

92  M. A. Roberts

that that something else is “take the aspirin”; we could just as well have added “take a sip of water” or “count slowly to five” –​instead. 41 Can we simplify –​and ultimately strengthen –​the argument by stipulating, as part of the case, the following counterfactual: had the parent not taken the pleasure pill, Jaime would never have been born? We can do that –​and to prove a principle false, it, of course, only takes one counterexample. But adding that stipulation to the case doesn’t, in fact, support the claim that the choice to take the pleasure pill doesn’t make Jaime worse off. Suppose that p shoots q in the shoulder and that p would have shot q in the heart had p not shot q in the shoulder (p was that irritated with q). p’s choice to shoot q in the shoulder still makes q worse off –​ p harms q in the ordinary, comparative sense of that term by shooting q in the shoulder. Specifically, p makes q worse off, not worse off than how q would have been had q been shot in the heart, but rather worse off than q would have been had q not been shot at all. 42 In formulating the person-​based approach, I have in earlier work sometimes appealed to the concept of expected value. While such a formulation would yield what we take to be appropriate results for the cases under consideration here, it now seems to me that it fails when applied to still other cases. A case suggested to me by Dean Spears seems to me convincing that the concept of expected value will not work for purposes of providing a credible account of the Nonidentity Problem. Spears (2019). See Appendix A.  The revised principle –​EVNPBI+PV –​thus uses the concept not of expected but rather of probable value. 43 Any plausible person-​based approach will need to recognize still other routes to permissibility. For example, tradeoff scenarios (include tradeoff-​to-​exist cases) will require us to recognize that a choice may be permissible even though neither wellbeing nor probable value has been maximized. 44 I am very grateful for the extremely valuable comments on an earlier draft of this chapter provided by the participants in the workshop on Parfit’s legacy held in conjunction with the annual meetings of the International Society of Utilitarian Studies, Karlsruhe, July 2018. I am also very grateful for the extremely valuable comments on a more recent draft of this chapter provided by Ben Eggleston and by two anonymous referees for this collection.

Bibliography Broome, J. (1992). Counting the Cost of Global Warming, Cambridge: The White Horse Press. Broome, J. (2004). Weighing Lives. Oxford University Press. Broome, J. (2015). “General and Personal Good:  Harsanyi’s Contribution to the Theory of Value,” in I. Hirose and J. Olson, The Oxford Handbook of Value Theory. Oxford University Press. Greene, M. (2016). “Roberts on Depletion:  How Much Better Can We Do for Future People.” Utilitas 23(1): 108–​118. Huemer, M. (2008). “In Defense of Repugnance.” Mind 117(468): 899–​933. Kavka, G. (1981). “The Paradox of Future Individuals.” Philosophy & Public Affairs 11: 93–​112. Narveson, J. (1976). “Moral Problems of Population.” In Ethics and Population, ed. M. D. Bayles. Schenkman. Pp. 59–​80. Ord, T. (unpublished manuscript). “On Ranking Alternatives with Different Populations.” Roberts, M. A. (2007). “The Nonidentity Fallacy: Harm, Probability and Another Look at Parfit’s Depletion Example.” Utilitas 19: 267–​311. Roberts, M. A. (2009). “The Nonidentity Problem and the Two Envelope Problem.” In Harming Future Persons, eds. M. A. Roberts and D. Wasserman. Springer. Pp. 201–​228.

Population ethics and Pareto plus  93

Roberts, M. A. (2010). Abortion and the Moral Significance of Merely Possible Persons: Finding Middle Ground in Hard Cases. Springer. Roberts, M. A. (2011a). “The Asymmetry: A Solution,” Theoria 77: 333–​367. Roberts, M. A. (2011b). “An Asymmetry in the Ethics of Procreation,” Philosophy Compass 6/​11: 765–​776. Roberts, M. A. and D. T. Wasserman (2017). “Dividing and Conquering the Nonidentity Problem.” In Current Controversies in Bioethics, eds. M. Liao and C. O’Neil. Routledge. Pp.  81–​98. Roberts, M. A. (2019). “The Nonidentity Problem.” Stanford Encyclopedia of Philosophy (available on-​line). Roberts, M. A. (unpublished ms. a). Modal Ethics (draft available on-​line). Roberts, M. A. (unpublished ms. b). The Existence Puzzles. Sher, G. (2005). “Transgenerational Compensation,” Philosophy & Public Affairs, 33: 181–​200. Singer, P. (2011). Practical Ethics (3rd ed). Cambridge University Press. Smilansky, S. (2017).“The Nonidentity Problem: United and Strong.” In Current Controversies in Bioethics, eds. M. Liao and C. O’Neil. Routledge. Pp. 99–​114. Spears, D. (2019). Correspondence with author. Temkin, L. (2012). Rethinking the Good:  Moral Ideals and the Nature of Practical Reasoning, Oxford: Oxford University Press.

5 AN OPINIONATED GUIDE TO “WHAT MAKES SOMEONE’S LIFE GO BEST” Chris Heathwood

Introduction In the opening pages of Reasons and Persons, Parfit lays out the basic normative and evaluative concepts that he will take for granted and in terms of which he will formulate many of his doctrines and arguments.These are the concepts of having a reason to do some act, of an act’s being what one ought to do, of an act’s being morally wrong, of an outcome’s being good or bad, and, finally, “of what is in someone’s self-​interest, or what would be best for this person” (ix-​x).1 Parfit calls special attention to this last concept, thinking it necessary to say more about it. He does so in an appendix, “What Makes Someone’s Life Go Best,” a ten-​page mini-​essay that has taken on a life of its own quite apart from the body to which it is appended (493–​502). In that appendix –​Appendix I –​Parfit, among other things, introduces a tripartite taxonomy of theories of well-​being that has since become the orthodox taxonomy in the field (493–​494); argues that hedonist theories should take a certain distinctive form (493–​494); identifies a problem for desire-​fulfillment theories that still has no received solution (494); gives arguments for the theses that desire-​ fulfillment theorists should accept the possibility of posthumous benefit and harm (495) and that they should count only our global desires as being relevant to how well our lives go (496–​499); and discusses, perhaps for the first time in the contemporary literature, the advantages of a so-​called hybrid theory of well-​being (501–​502). Appendix I appears to be the most influential and important of the appendices to Reasons and Persons (it also happens to be the longest).2 The present chapter serves as a critical guide to it. I will explain, elaborate, and evaluate most of its main theses and arguments. For those interested in studying an issue further, I will provide references to some relevant literature. I hope to convey to readers the interest, importance, and richness of “What Makes Someone’s Life Go Best.”

“What Makes Someone’s Life Go Best”  95

Appendix I also deserves the label ‘seminal’. When Reasons and Persons appeared, the topic of what makes someone’s life go best, which is now usually referred to in contemporary ethics as the topic of well-​being, was written on and studied by anglophone philosophers as a self-​standing area of inquiry far less than it is today, when there are entire journals devoted to the topic.3 Parfit deserves some of the credit for the present robustness of this subfield of moral philosophy. Theories of well-​being, or of self-​interest, answer the question of “What would be best for someone, or would be most in this person’s interests, or would make this person’s life go, for him, as well as possible” (493). This question, the philosophical question of well-​being, can be clarified in several ways. In asking it, we are asking about a distinctive kind of evaluation, different from moral assessments, as when we wonder what kind of life is the morally best kind of life for someone to lead; and different from, for example, assessments of how meaningful some person’s life is. We are instead asking about self-​interest, about benefit and harm, about personal welfare. It is a matter of continued controversy, however, just what we are asking when we ask the philosophical question of well-​being. Though the appearance of Appendix I is motivated by the wish to shed light on the topic of self-​interest, Parfit does not address this issue head-​on. Like most philosophers of well-​being, he relies on the particular ways he puts the issue in ordinary language. We can also use as a guide the claims about self-​interest that Parfit finds intuitive, which help to reveal just what concept he is using.4 When Parfit wonders what makes someone’s life go best, the making relation that he is talking about here is not a causal relation. The purely philosophical question of well-​being is not the partly empirical question of what causes people to be better or worse off. House fires generally cause people to be worse off; access to clean water generally causes people to be better off. They do this by causing other, distinct events that “make” –​in a more direct way –​people better or worse off.This more direct kind of making is what happens when the event in question is intrinsically good or bad for a person, or good or bad in itself. Access to clean water and house fires are of mere instrumental value and disvalue for people. So what things are intrinsically good or bad for people? That is precisely Parfit’s question. Appendix I is divided into five unlabeled sections, the divisions being indicated simply with skipped lines.The first section (493–​495) begins by introducing Parfit’s now orthodox taxonomy of theories of well-​being, on which there are three main kinds of theory: Hedonistic, Desire-​Fulfillment, and Objective List.5 The section goes on primarily (i)  to argue that hedonism should take a certain form, (ii) to explore the Desire-​Fulfillment Theory, (iii) to argue that a certain breed of it –​the Success Theory –​is superior to an unadorned version of it, and (iv) to compare the Success Theory to a certain distinctive form of Hedonism, Preference-​Hedonism. The second and third sections address questions that arise for Preference-​ Hedonism and the Success Theory –​and in fact for any kind of Desire-​Fulfillment Theory. Should the preferences or desires that a person actually has be used to determine how well things go for them in counterfactual scenarios? The brief second section (495–​496) argues ‘No’. The third section (496–​499) considers in some

96  Chris Heathwood

depth whether these theories should take a Summative or Global form, arguing in favor of the latter. Section four (499–​501) introduces the Objective List Theory and compares it with the Success Theory and, to a lesser extent, Preference-​Hedonism. Parfit is here essentially exploring the debate, familiar to philosophers of well-​being and among the most central debates in the field, over whether well-​being is objective or subjective. The fifth and final section (501–​502) introduces a new category of theory of well-​being, one that combines subjective and objective elements. This “composite” account is nowadays standardly referred to as the Hybrid Theory of well-​being. Although Parfit refrains from committing to any particular theory of well-​being, one suspects that he finds the Hybrid Theory most attractive. Let’s now examine some of these issues more deeply. Since there is not space to give a thorough treatment of all of them, I will focus on what I take to be the most interesting issues and the topics that are less well trodden in the well-​being literature.

Preference-​hedonism and the Theory of Pleasure and Pain The first argument Parfit makes in the Appendix is an argument against a theory of well-​being he calls Narrow Hedonism. Parfit objects not to this theory’s central evaluative claim –​that what is most in a person’s interest is for their balance of pleasure over pain to be maximized, a claim made by any hedonist6  –​but to its central metaphysical claim:  its account of the nature of pleasure and pain. As Parfit unhelpfully formulates it, this is the view “that pleasure and pain are two distinctive kinds of experience” (493). Parfit appears to be talking about a theory of the nature of pleasure and pain that is sometimes called the Distinctive Feeling Theory. It was assumed by G.E. Moore (1903: §12) and has been defended recently by Ben Bramble (2013). According to it, pleasure and pain are single, uniform feelings or sensations, in the same category as the taste of cilantro, the feeling of nausea, or the smell of lilac. Parfit’s brief argument against this view, due initially to Sidgwick (1907: 127) and usually referred to now as the heterogeneity problem, is an argument from introspection: attending to one’s own phenomenology reveals that there simply is no single, distinctive feeling common in cases of, for example, satisfying an intense thirst or lust, listening to music, solving an intellectual problem, reading a tragedy, and knowing that one’s child is happy. 4937 Parfit believes that a more plausible theory of pleasure and pain appeals not to sensory feelings but attitudes, in particular desire: On the use of ‘pain’ which has rational and moral significance, all pains are when experienced unwanted, and a pain is worse or greater the more it is unwanted. 494

“What Makes Someone’s Life Go Best”  97

Parfit is here advancing the view that what makes an experience a pain, or a painful experience, is nothing about its intrinsic nature but about the stance the subject takes towards the experience (and similarly for pleasure).This view avoids the implication that the “various experiences [listed above] contain any distinctive common quality” (494).8 But aren’t some experiences intrinsically painful? Consider what it is like to step barefoot on a tack. Doesn’t it seem to be part of the intrinsic nature of that experience that it hurts, or is painful? Parfit’s theory of the nature of pleasure and pain must deny this. But some remarks later in the appendix, made for a different purpose, address this concern. There Parfit notes that After taking certain kinds of drug, people claim that the quality of their sensations has not altered, but they no longer dislike these sensations. We would regard such drugs as effective analgesics. 501 “This,” he says, suggests that the badness of a pain consists in its being disliked, and that it is not disliked because it is bad. 501 If it suggests this, then, since a sensation’s being painful seems sufficient for its being bad, it also suggests that the painfulness of a sensation consists in its being disliked –​ or unwanted while it is happening.9 Thus, if it appears to subjects that the painfulness of the sensation of stepping barefoot on a tack is intrinsic to this sensation, then this appearance may be an illusion. Perhaps it results from a kind of mental projection of one’s intense dislike of the sensation onto a sensation that is, considered in itself, hedonically neutral. Parfit uses his desire-​based theory of pleasure and pain to introduce a version of hedonism that he sees as preferable to Narrow Hedonism; this is Preference-​ Hedonism. Preference-​Hedonists agree that pleasurable experiences, and these alone, are intrinsically good for us and painful experiences, and these alone, are intrinsically bad. But it replaces the Distinctive Feeling Theory of pleasure and pain with Parfit’s preferred desire-​based account. Narrow Hedonists presumably have to say that the sensations that are experienced under the influence of the drug Parfit describes above –​sensations that their subjects don’t at all mind or care to avoid –​ are nonetheless painful and bad, and make their lives worse. Preference-​Hedonism, by contrast, avoids this counterintuitive implication.

The Unrestricted Desire-​Fulfillment Theory and the problem of remote desires One of the most important arguments put forth in Appendix I takes only a few lines to present. It concerns the Desire-​Fulfillment Theory, according to which

98  Chris Heathwood

“what would be best for someone is what, throughout his life, would best fulfil his desires” (493). The most straightforward Desire-​Fulfillment Theory holds that all desires count –​that the satisfaction of any desire is good in itself for a person and makes their life go better. This Parfit calls the Unrestricted Desire-​Fulfillment Theory. Parfit finds this theory unacceptable on the basis of the case of “the stranger on the train.” Actually, no train is mentioned in the version of the case that appears in Appendix I. But Parfit is here reprising a case introduced earlier in Reasons and Persons, in Part Two. He used it first to give an example of a desire that is not conditional on its own persistence (151), and next to illustrate how changes in one’s concerns do not require changes in what one believes worthy of concern (157). The original case begins, “Suppose that I meet some stranger on a train …” (151).10 In the version in our appendix, the stranger has what is believed to be a fatal disease. My sympathy is aroused, and I strongly want this stranger to be cured. Later, unknown to me, this stranger is cured. On the Unrestricted Desire-​Fulfilment Theory, this event is good for me, and makes my life go better. This is not plausible. We should reject this theory. 494 This simple counterexample illustrates a feature of desire-​fulfillment theories not yet emphasized: that desire fulfillment requires no feelings of fulfillment. All that is required is that the object of the desire obtains. In a book whose manuscript Parfit had seen while writing Reasons and Persons,11 James Griffin put more abstractly what may be the same concern: The breadth of the [desire] account, which is its attraction, is also its great flaw … . It allows my utility to be determined … by things that do not affect my life in any way at all. The trouble is that one’s desires spread themselves so widely over the world that their objects extend far outside the bound of what, with any plausibility, one could take as touching one’s own well-​being. Griffin 1986: 16–​17 This, which I call “the problem of remote desires,” has been explored in some depth since Reasons and Persons,12 but has, as yet, no received solution. Parfit’s own solution restricts the theory to count only desires that are about one’s own life. Somewhat cryptically, Parfit calls the theory so-​restricted the Success Theory. Since the desire that the stranger be cured is not a desire about Parfit’s own life, the Success Theory delivers the desired result that the stranger’s being cured is no benefit to Parfit. Incidentally, Preference-​Hedonism delivers this result, too, since the stranger’s being cured has no effect on Parfit’s experiences.

“What Makes Someone’s Life Go Best”  99

Parfit admits that when the Success Theory “appeals only to desires that are about our own lives, it may be unclear what this excludes” (494). But some discussion, including several examples, help shed some light on how Parfit understands what it is for a desire to be about one’s own life (494–​495). Interestingly, Parfit maintains that desires whose fulfillment or frustration turns on what happens after one is dead can nonetheless count as desires about one’s own life. This contrasts Parfit’s Success Theory with a similar theory proposed in 1980 by Mark Carl Overvold. Like Parfit, Overvold offers a restricted version of the Desire-​Fulfillment Theory, though in response to a different but related problem: the problem of self-​ sacrifice.13 Overvold’s view counts only desires with this feature: they are for states of affairs that can obtain at some time only if the subject of the desire exists at that time (Overvold 1980:  10n). Though Overvold didn’t craft the theory with this in mind, it delivers the desired result about the stranger on the train. And it does so without countenancing posthumous benefit and harm  –​as Overvold desired (Overvold 1980: 108). Whether we understand the restriction in the Parfitian or the Overvoldian way, the resulting theory seems to suffer from convincing counterexamples. Consider, for instance, the desire that the team one roots for win. For many people, this desire is as strong as, and as important a part of their identity, as many desires that are about their own lives. When their team wins and this desire is fulfilled, this seems like a good thing in their life. But the Success Theory implies otherwise. Note that no Hedonistic Theory would exclude the pleasure taken in the victory of one’s team; nor, it seems, should the Desire-​Fulfillment Theory be restricted to exclude the corresponding desire fulfillments.14

Actual preferences and counterfactual well-​being In the brief second section of the appendix, Parfit raises an interesting and not-​ often-​discussed question:  “Should we appeal only to the desires and preferences that someone actually has?” (495). If we endorse some kind of desire-​based theory of well-​being, such as the Unrestricted Desire-​Fulfillment Theory, the Success Theory, or Preference-​Hedonism, and we are trying to decide how well off some actual person would be in an imagined counterfactual scenario, should we look to the desires the person actually has, or to the desires that they have in the counterfactual scenario? A common way of thinking seems committed to holding that only one’s actual preferences matter, but Parfit shows the mistake in this. Suppose you decide to stay home and read King Lear rather than go to a party; and suppose that, throughout your evening, you continue to be glad that, or to prefer that, you stayed home to read King Lear rather than go to the party (495–​496). It is tempting to infer from this that you made the right choice –​that staying home to read King Lear gave you the better evening. But this inference is fallacious, for it could still be that if you had gone to the party, you would have, throughout your evening, been glad that you went to the party rather than stayed home to read King Lear. The common way of

100  Chris Heathwood

thinking would then imply not only that staying home to read King Lear gave you the better evening but also the contradictory thought that going to the party would have given you a better evening. Parfit thus claims that we should “appeal not only to my actual preferences, in the alternative I choose, but also to the preferences that I would have had if I had chosen otherwise” (496). In doing so, he makes it sound as if his answer to the question that animates this section - the question of whether it is actual or counterfactual desires that matter - is “both.” But that appearance is misleading. For he is there saying that only in making a comparative judgment between the actual course of events and some counterfactual course of events should the theory take into account both actual and counterfactual preferences. That is consistent with (and indeed explained by) the idea that to know how absolutely good one of these scenarios would be for the person, we look only at the preferences the person has within that scenario. Though for a different purpose, Eden Lin (2019) helpfully distinguishes the two main options here for desire-​based and other subjectivist theories: Same World Subjectivism (the approach Parfit favors), on which scenarios are evaluated according to the desires (or other favoring attitudes) that one has in that scenario, and Actual World Subjectivism, which evaluates all counterfactual scenarios using one’s actual desires.15 The considerations Parfit discusses –​concerning the choice to stay home and read King Lear or go out to a party –​do indeed tell in favor of a Same-​World rather than an Actual-​World approach. But there are considerations that Parfit does not discuss that might attract one to an Actual-​World approach. Consider The Brand-​New Life. An eccentric billionaire with an interesting drug offers you a Brand-​New Life. You will be relocated to a new city –​one that does not now appeal to you at all.You will be immersed in a new circle of friends –​ people with whom you now have no wish to associate. You will be given a new career –​one you now have absolutely no interest in. You will never be allowed to return to your current home, to see your current friends or family again, or to pursue your current career and other projects. But you will be given a drug –​a complacency pill16 –​that will gradually cause you to want to be in your new city once you are there, to want to be associating with your new friends, and to want to be engaged in your new career. The drug will also cause your longings for your old life to diminish and eventually cease altogether. As it happens, the life you will lead if you decline the offer –​your old life –​has its ups and downs, and, although it is a fine life by any reasonable standard, has its share of unfulfilled desire. But owing to the effectiveness of the complacency pill, the Brand-​New Life on offer will fulfill far more of the desires you will come to have if you lead that life.17 Would it be in your best interest to accept the offer? Would you be foolish to decline this Brand-​New Life?

“What Makes Someone’s Life Go Best”  101

I suspect that very few of us who think our lives at least minimally decent and feel at all attached to our friends, family, projects, and careers would even contemplate such an offer –​even though we recognize that we would be turning down a life that would deliver far more of what we would want if we were to choose that life. Though I wouldn’t contemplate doing this either, I believe that the Same-​World approach, which Parfit endorses, has the correct implication here:  that I  would benefit more, or would get a life that is more in my interest to get, if I were to choose the Brand-​New Life.The Actual-​World approach can give us the alternative answer that you will be better off remaining in your current life.That is because the Brand-​New Life rates poorly when judged by the standards of your actual desires. But the Actual-​World approach is problematic, and not only for what it implies about the decision whether to stay home and read King Lear or go to the party. Suppose that I currently have no desire to try a certain sort of unfamiliar cuisine; I prefer to stick with my usual type of food, though it barely excites me. Suppose it’s also true that if I were to try the unfamiliar cuisine, I would love it, devour it with gusto, and be very glad that I decided to try it. In other words, I would enjoy more desire fulfillment –​because my newly acquired desires would be much more intense than my desires for my usual meal –​than I would get by eating my usual meal. The Actual-​World approach nevertheless implies, implausibly, that opting for my usual meal would be better for me. The Actual-​World approach also has the following bizarre –​perhaps even incoherent –​implication about the Brand-​New Life: that if you don’t in fact choose the Brand-​New Life, we can say, correctly, that it would have been a worse life for you, but that if you do choose it, then we must say that it is better than the life you would have gotten (your old life) had you not chosen it. That is because, on the Same-​World approach, if you do not choose the Brand-​New Life, we judge it by the standards of your actual desires, but if you do choose the Brand-​New Life, it becomes your actual life, and so we judge it by the standards of the desires that you have in the Brand-​New Life. This is similar to Parfit’s objection, concerning King Lear and the party, that “This theory thus implies that each alternative would have been better than the other” (496). I therefore agree with the Parfitian Same-​World approach that I would be better off in the Brand-​New Life, no matter what I end up choosing. I did say, however, that I wouldn’t in fact choose this better life. Is this irrational? Theories of self-​interest have, on their own, no direct implications about rationality or reasons for action. But a theory of self-​interest such as a Same-​World version of the Desire-​Fulfillment Theory, together with certain auxiliary principles of rationality, will imply that it would be irrational for me to decline the Brand-​ New Life. One such auxiliary principle is the central claim of what Parfit calls the Self-​Interest Theory of rationality, one of the centerpieces of Reasons and Persons: (S1) For each person, there is one supremely rational ultimate aim: that his life go, for him, as well as possible. 4

102  Chris Heathwood

Parfit, however, spends Part Two and some of Part Three of Reasons and Persons explaining why he thinks that the Self-​Interest Theory of rationality is false.18 This is welcome news to those of us who are prepared to say that the Brand-​New Life is a better life.That’s because, despite this, it does not seem that a refusal to choose the Brand-​New Life is irrational. Just what might make this refusal rational is a question that will have to be left for another time.

Summative vs. Global Desire-​Fulfillment Theories The third section of Appendix I (496–​499) concerns again an intramural dispute among desire-​based theories of self-​interest. Parfit argues that such theories should take a “Global” rather than a “Summative” form. Summative versions of the Success Theory count all of one’s desires about one’s own life; similarly, Summative versions of Preference-​Hedonism count all of one’s desires about one’s present experiences. Each theory assigns, in proportion to the desire’s strength, a positive value to the fulfillment of its favored kind of desire and a negative value to its frustration. Then, to determine how well a life goes overall, the theory simply sums the values of the fulfillments and frustrations that occur in the life (496). An Unrestricted Desire-​ Fulfillment Theory can also take this Summative form. On any Summative view, the intrinsic welfare value of a person’s life is derived from the values of all of the fulfillments and frustrations contained within it.19 A Global Desire-​Fulfillment Theory, by contrast, appeals only to global rather than local desires and preferences. A preference is global if it is about some part of one’s life considered as a whole, or is about one’s whole life. 497 Before we consider Parfit’s arguments in favor of Global theories, we should be sure that we understand those theories. Unfortunately, Parfit’s definition of ‘global desire’ is not very helpful.The first disjunct, on which a desire is global “if it is about some part of one’s life considered as a whole,” is especially obscure. Is Parfit saying that it is about some part of one’s life-​considered-​as-​a-​whole (whatever that might mean), or is he saying that the part needs to be considered as a whole? The latter may seem the more natural interpretation, but, as we’ll see below, it may conflict with some of the work to which Parfit wants to put the concept. Nor is the second disjunct, on which a desire is global “if it is about one’s whole life,” unproblematic. The first paragraph in the front matter to Reasons and Persons begins, “Sixteen years ago, I travelled to Madrid with Gareth Evans. I hoped to become a philosopher.” Is this hope a global desire? Few desires seem “bigger” than the desire to have a certain career, but not even this desire fits Parfit’s second disjunct. The desire to become a philosopher isn’t a desire about one’s whole life. But we can see by means of his examples the work Parfit wants the concept to do, and this may assist us in discerning its contours.

“What Makes Someone’s Life Go Best”  103

The drug addiction case One such example is the drug addiction case, put forth as a counterexample to Summative Theories. Knowing that you accept a Summative theory, I tell you that I am about to make your life go better. I shall inject you with an addictive drug. From now on, you will wake each morning with an extremely strong desire to have another injection of this drug. Having this desire will be in itself neither pleasant nor painful, but if the desire is not fulfilled within an hour it will then become very painful. This is no cause for concern, since I shall give you ample supplies of this drug. Every morning, you will be able at once to fulfil this desire. The injection, and its after-​effects, would also be neither pleasant nor painful.You will spend the rest of your days as you do now. 497 Even if we had no concerns about side-​effects or about the logistics of administering to the addiction, probably most of us would decline Parfit’s offer. And if he injected us anyway, perhaps most of us would wish we didn’t have this addiction, benign as it is. But advocates of Summative Desire-​Fulfillment Theories cannot take refuge in these facts about our desires, because, Parfit notes, the negative value of these desire frustrations would be swamped by the positive values of the repeated daily desire fulfillments (497). Summative desire theories thus appear to imply that it would be in one’s self-​interest to accept Parfit’s offer, and that, despite one’s wish not to be addicted, it would be in one’s self-​interest to remain an addict. But “Global Theories,” Parfit claims, “give us the right answer in the case where I  make you an addict,” and thus save the Desire-​Fulfillment approach. Global Theories appeal only to someone’s desires about some part of his life, considered as a whole, or about his whole life … .You would prefer not to become addicted, and you would later prefer to cease to be addicted. These are the only preferences to which the Global Theories appeal.They ignore your particular desires each morning for a fresh injection, since you have already considered these desires in forming your global preference. This application of the concept of global desire shows that Parfit means that the relevant part of one’s life is what should be considered as a whole. Parfit must here have in mind something like this. Your being addicted to this drug is a part of your life, and it is one that you can consider either piecemeal –​as you might if you were to think of yesterday’s desire for the drug, then today’s, then tomorrow’s, and so on –​or as a whole –​as you do when you consider the more general fact that you are addicted to this drug, and realize that you don’t want to be addicted to it. Fair enough, but each morning, when you are desiring a fresh injection, would you not

104  Chris Heathwood

be considering that injection as a whole? Normally, you would be; you wouldn’t be considering its elements piecemeal. Your injecting the drug on some morning would thus be a part of your life that you would be (and in any case certainly could be) considering as a whole. A  Global Theory would consequently include such desires, and Parfit’s solution would be undermined. In the last sentence of the passage quoted above, Parfit suggests what seems to be a new and different definition of ‘global desire’: something along the lines of a desire that is not the object of a conflicting higher order desire. Depending on how this definition is clarified, it may be subject to an objection that Parfit himself raises, an objection to the view that a desire can be ignored if it is a desire you prefer not to have (497–​498). Rather than delve further into just how to understand ‘global desire’, let’s allow Parfit to apply the concept as he wishes, in the ways that suit his needs. Even giving him that, his arguments in this section face interesting challenges.20 One problem is that the drug-​addiction thought experiment itself may not bear scrutiny. It is, if you think about it, hard to imagine the case as described, and the case may in fact be metaphysically impossible. In particular, it is hard to imagine that a person might (i) have a very strong desire for a certain thing, (ii) be aware that the thing is occurring once it starts occurring, (iii) continue to want it to be occurring as it is occurring, yet (iv) experience no pleasure when it is occurring. But Parfit’s case requires that this be possible. For each morning you (i) will have a very strong desire to be injected, (ii) will be aware that the injection is occurring once it is occurring, (iii) will continue to want it to be occurring as it is occurring, yet (iv) supposedly experience no pleasure during any of this. The idea that this is possible is in fact in tension with the desire-​based theory of pleasure that Parfit endorses throughout the Appendix. According to this theory, “whatever someone wants or does not want to experience  –​however bizarre we find his desires  –​should be counted as being for this person truly pleasant or painful” (501).21 If so, then the experience of injecting the drug cannot fail to be pleasant, contrary to what Parfit’s thought experiment stipulates. This matters because if taking the drug is pleasurable, and being “addicted” to it is as benign as Parfit stipulates, it is, upon reflection, not very intuitive to think that it is bad to be in this way addicted. In fact, describing it as a case of addiction is simply inaccurate (hence the scare quotes above). “The defining features of addiction are significant distress or harm,” according to a standard psychology textbook (Kalat 2016: 362). The same book also notes that a person qualifies as being addicted to something only if it “cause[s]‌serious trouble in [their] life” (Kalat 2016: 497). It is therefore simply false that the subject in Parfit’s example suffers from addiction. Moreover, it is likely that attaching the label of ‘addiction’ to the case distorts our judgment about it, by causing us to assume that there must be something bad going on. In fact, it is hard to see what bad is going on. If you would get pleasure from each injection, it seems impossible to distinguish what Parfit does to you in making you “addicted” to this drug than what a friend does to you in turning you on to, say, a

“What Makes Someone’s Life Go Best”  105

new, in-​no-​way-​unhealthy food that you love and come to crave each day –​and are able to get each day (e.g., morning coffee, an apple a day). Being “addicted” to such things is bad only if you run out of supplies, which is already ruled out in Parfit’s example (cf. Heathwood 2019: §3b). There is admittedly the following difference between Parfit’s case and the case of craving and getting coffee each morning or an apple each afternoon: in Parfit’s case, you want not to be addicted. Fair enough, but it is not clear why such a desire frustration (one of an intuitively irrational desire, given the above analysis of Parfit’s thought experiment) should trump the stronger daily desire fulfillments. It seems similar to a case in which a person, due perhaps to a severe religious upbringing, has a preference against certain innocent pleasures that they regularly receive. Of course the best option would be to rid oneself of the religiously induced aversion. But if that is not possible, the next best option is to enjoy the innocent pleasure and put up with the –​by hypothesis weaker –​global desire frustration. Upon analysis, then, Parfit’s drug-​addiction counterexample, rich and worthy of study as it is, does not seem to be successful.

The single-​life Repugnant Conclusion Parfit’s other main thought experiment in this subsection is also terrifically interesting.This is “the analogue, within one life, of the Repugnant Conclusion” (498). You could live one of two lives. In one –​call it ‘(a)’ –​you get “fifty years of life of an extremely high quality”; you “would be very happy, would achieve great things, do much good, and love and be loved by many people.” In the other –​call it (z)  –​you would receive “an indefinite number of years that are barely worth living” (498). In later work, Parfit refers to a similar pair of lives as the “Century of Ecstasy” and the “Drab Eternity” (Parfit 1986: 160). Assuming a Desire-​Fulfillment framework, this amounts to a choice between (a) a fifty-​year life containing very many fulfillments of strong and important desires and few desire frustrations, and (z) an indefinitely long life containing occasional, very mild desire fulfillments and few desire frustrations. As Parfit puts it, (z) “would each day contain a few small pleasures” (498). Parfit does “not believe that the second alternative would give [him] a better life” (498). But Summative Desire-​Fulfillment Theories may seem to imply otherwise. For however much benefit (a) contains, the amount contained in (z) can eventually surpass it, since each additional day adds value to it. Global Theories, by contrast, appear to get just the result Parfit wants, for Parfit has a global preference for (a) over (z). In fact, however, it is not clear that Summative Theories imply that life (z) is better. For it depends on what Parfit’s global desires are in (z). Summative Theories, recall, “appeal to all of someone’s desires” (496, emphasis mine).22 They thus count global as well as local desires. Almost as an afterthought to this subsection, Parfit, to bolster the judgment that (a) is a better life than (z), adds the detail that, “It is likely that, in both alternatives, I would globally prefer the first” (499).

106  Chris Heathwood

This detail may help elicit the intuition that (a) is better, but it undercuts the other main premise in Parfit’s argument: that Summative Theories imply that (z) is better. This is because, in addition to the fact that (z)  would each day contain a few minor desire fulfillments, it now has, with this detail added, a continuous stream of global desire frustration. In (z), Parfit is continually wishing that he were leading life (a). Parfit never specifies the relative strengths and durations of the daily fulfillments and frustrations in (z) so described, but, given the ever-​present global desire in (z) for life (a) and the modest size of (z)’s desire fulfillments, it seems likely that the disvalue of its frustrations would exceed the value of its fulfillments, thus making (z) negative in value overall. This would make it worse than (a) according to Summative Theories and would thus undermine Parfit’s argument against them. To give Parfit’s argument a better chance of working, we can suppose that, in (z), Parfit does not have a global desire to be living (a) instead. Let’s suppose that, in (z), he has no global desires at all; he lives in the moment, taking things one day at a time. This will deliver Parfit’s intended result that Summative Theories imply that (z) is better than (a). Is this a problem for Summative Theories? At first blush it may appear so, but upon scrutiny, arguably not. For there is a powerful and by now familiar sort of argument for the initially dubious conclusion that the drab, indefinitely long life (z) is better than the half-​century of ecstasy (a). The argument begins by asking us to compare (a) to a certain other life, (b). (b) is twice as long as (a) and only slightly less good at each moment; let’s say that it’s about 95 per cent as good at each moment. Or if that sounds too artificial, we can say instead that each day, or week, or year of (b) is about 95 per cent as good as each day, week, or year of (a). Putting aside for the moment issues of global desires, which life is, intuitively, the better life to get? Obviously, life (b). Each year of (b) is almost as good as each year of (a), but there are twice as many of them in (b). Next consider life (c), which stands to (b) as (b) stands to (a). (c) is a 200-​year life, twice as long as life (b), and each year of (c) is almost as good as each year of (b). Which seems better to you and which would you rather have: a 100-​year-​long life in which you “would be very happy, would achieve great things, do much good, and love and be loved by many people,” (498) –​all while wanting exactly these things –​or a very similar life of almost as high a quality each year but in which you get to live twice as long? Again, the latter life, life (c), is clearly preferable. Of course we can repeat this reasoning, and when we do, we will find that life (d) –​twice as long as (c) and only slightly lower in quality –​is better than life (c); we will find that life (e) is better than life (d); and so on, until we reach the claim that life (z) is better than its predecessor, life (y), a life half as long as (z), but only slightly higher in average annual quality. Finally, because the relation of being better than is a transitive relation, it follows that life (z), contrary to initial appearances, is better than life (a).23 The implication of Summative Theories about lives (a) and (z) can thus be shown via an independent argument to be the correct verdict after all. Can Parfit appeal to global desires to block this line of reasoning? It seems not. In addition to each life in the sequence (after life (a)) being intuitively better than its

“What Makes Someone’s Life Go Best”  107

predecessor, each is also (and surely not unrelatedly) a more appealing life to get. For these reasons, it would be very odd for someone to have a global preference for one of these lives over its successor in the series. So, I assume that Parfit would in fact globally prefer (b) to (a), (c) to (b), and so on, eventually globally preferring (z) to (y). Once Parfit learns this about himself, then, since he is presumably not prone to blatant irrationality, he will reverse his initial global preference, and come to prefer (z) to (a). In this case, even his preferred Global Theory will imply that (z) would be a better life for him to get than would (a). Suppose, however, that despite preferring (b) to (a), (c) to (b), (d) to (c), and so on, right up to (z) to (y), Parfit digs in his heels and continues to prefer (a) to (z). Well, either his Global Theory will allow irrational sets of preferences such as this one to determine one’s well-​being, or it will not. If it allows it, then, since better than is transitive (even if Parfit’s global preferences aren’t), his feared result that (z) is better than (a) will remain. And all this view will have done is added the additional, contradictory result that (a) is also better than (z).This hardly seems like an improvement over the Summative Theory. If Parfit’s Global Theory doesn’t allow intransitive preferences, and so requires preferences to be laundered, this will presumably wash away Parfit’s recalcitrant preference for (a) over (z), and we will be back where we were a paragraph ago. It seems hard to avoid the (to some) repugnant conclusion that life (z) is better than life (a), and in any case the appeal to global desires does not appear to help in avoiding it. If life (z) is, contrary perhaps to initial appearances, better than life (a), this raises the question of why things should have appeared to some as if life (a)  is better. Though Parfit doesn’t cite him, the single-​life analogue of the repugnant conclusion was discussed generations before Reasons and Persons by J.M.E. McTaggart. Astonishingly, McTaggart even writes that “this conclusion” –​one similar to the conclusion that (z) is better than (a) –​“would … be repugnant to certain moralists” (McTaggart 1927: 453). McTaggart, for his part, accepted the conclusion that Parfit finds repugnant, and offered explanations for why this conclusion might appear false. “It must be remembered that men’s choice in such cases is very much affected by their imagination,” he said, adding that it is not easy to properly imagine enormously long durations of time. McTaggart also cited a bias that Parfit discusses extensively in Part Two of Reasons and Persons  –​the bias towards the near  –​in explaining why people might prefer the shorter, worse life:  “we are generally affected more than is reasonable by the present or the near future in comparison with the far future” (McTaggart 1927: 453).24 It is curious that Parfit didn’t think to consider whether his preference for (a) over (z) might be the result of a bias that he discusses extensively elsewhere in the book. Parfit spends much more time on the Repugnant Conclusion proper than on its single-​life analogue. (The Repugnant Conclusion proper is the claim that “For any possible population of at least ten billion people, all with a very high quality of life, there must be some much larger imaginable population whose existence, if other things are equal, would be better, even though its members have lives that are barely worth living” (388).) Parfit would reject the sort of “continuum argument”

108  Chris Heathwood

presented above if applied to whole populations in an attempt to establish the Repugnant Conclusion proper. Interestingly, however, at least some of Parfit’s reasons for rejecting a continuum argument in that context don’t carry over to the present context. In the context of the Repugnant Conclusion proper, Parfit would reject (or at least regard as dialectically illegitimate) the initial step, the one that claims that a world with a “population of at least ten billion people, all with a very high quality of life” is not as good as a world with twice as many people, all with a quality of life almost as good as in the first world. Parfit would reject this step in order to accommodate the view that it doesn’t make things better when we make happy people (as we would if we move from the first to the second world above) but only when we make existing people happy.25 But these reasons don’t apply to the single-​life case. There is at least some plausibility to the thought that “of the two ways of increasing the sum of happiness –​making people happy, and making happy people –​only the first” (394) is an improvement. But there is no plausibility to the thought that, of the two ways of increasing the sum of happiness in a single life –​making some fixed number of years better, and adding on additional good years –​only the first is an improvement. Parfit’s main views about how one might block certain arguments for the Repugnant Conclusion proper thus don’t make problems for the above argument for the single-​life repugnant conclusion.

Objective vs. Subjective vs. Hybrid Theories After considering these various controversies within Desire-​Fulfillment Theory, Parfit turns to the more fundamental question of whether we should accept a subjective theory of self-​interest in the first place. The main alternative is to accept the Objective List Theory, which holds that getting a good life is less a matter of how we regard or feel about the things we get in life (that is what subjectivists think is all-important) and more a matter of the nature of those things themselves. The good things in life on an Objective List Theory “might include moral goodness, rational activity, the development of one’s abilities, having children and being a good parent, knowledge, and the awareness of true beauty” (499). About many philosophical topics, the main theories agree about most or all ordinary cases. This is true for our topic as well. Items that would appear on many Objective Lists tend to be just the sorts of things that people want in their lives and would enjoy getting. Thus, for most actual people, objective and subjective theories will agree on how well their lives are going (though of course they will give different explanations for why they are going as well or as badly as they are going). For this reason, to decide among theories, we often need to test them using cases that aren’t found in the actual world. Thus Parfit says, “In choosing between these theories, we must decide how much weight to give to imagined cases in which someone’s fully informed preferences would be bizarre” (499). Parfit seems generally inclined to rely on such cases in deciding among theories (500). And that’s probably a good thing: if we didn’t so rely, it’s not clear how else we would decide.

“What Makes Someone’s Life Go Best”  109

Classic objections to subjectivist theories are based on cases featuring certain sorts of intuitively defective desires. One such case involves an imagined life full of satisfied pointless desires, as illustrated by Rawls’s grass counter (499–​500; Rawls 1971:  432). Another such case involves an imagined life full of satisfied immoral desires (500). Many fair-​minded people –​perhaps including Parfit, though he never quite says so (500) –​find it hard to accept that such lives are best for the people with these desires. But, as a passage in the fifth and final section of the Appendix shows, there are arguments on the other side as well. Consider an Objective List Theory that “claims that what is good for someone is to have knowledge, to engage in rational activity, and to be aware of true beauty” (501). “Would these states of mind be good, if they brought no enjoyment, and if the person in these states of mind had not the slightest desire that they continue?” (501). Many fair-​minded people –​perhaps again including Parfit –​find this hard to accept too. It is hard to accept that a life that leaves one completely cold can nonetheless be of great benefit to one. This argument is essentially appealing to an internalist doctrine about self-​interest according to which, as Peter Railton put it in 1986, “what is intrinsically valuable for a person must have a connection with what he would find in some degree compelling or attractive, at least if he were rational and aware” (Railton 1986: 9). We thus seem to have arguments against both main approaches –​subjectivism and objectivism –​that fair-​minded people find compelling. This causes Parfit to wonder if both approaches might be wrong, because “each side … saw only half the truth” (502). Perhaps the best theory of what makes someone’s life go best will combine objectivist and subjectivist elements. Perhaps what is best for people is a composite. It is not just their being in the conscious states that they want to be in. Nor is it just their having knowledge, engaging in rational activity, being aware of true beauty, and the like … . What is of value, or is good for someone, is to have both; to be engaged in these activities, and to be strongly wanting to be so engaged. 502 This idea, which is now referred to as the Hybrid Theory of well-​being, has been developed in recent years in different ways by a number of different philosophers.26 Although it is not obvious that, as Parfit hopes, Hybrid Theories do sufficient justice to internalist intuitions about well-​being,27 it is a promising category of theory that deserves the increased attention that it has been getting. Parfit never returned to explore in as much depth the question of what makes someone’s life go best. The notion of well-​being does feature prominently in some of his later work, most notably On What Matters, where he discusses at length the roles that well-​being should play in morality and our reasons for action more generally. Although it is possible that which theory of well-​being one endorses will affect the normative reasons that one thinks one’s well-​being provides for oneself and others, the question of what role well-​being plays in these matters can be carried

110  Chris Heathwood

out to a large extent in abstraction from the question of which theory of well-​being is true. Parfit’s rich and fertile Appendix illustrates, however, that the question of what makes someone’s life go best is worth exploring very much for its own sake.28

Notes 1 Bare page references are to the “Reprinted with further corrections 1987” edition of Reasons and Persons. 2 It has been reprinted as a standalone essay (e.g., in Shafer-​Landau 2012) and has its own Google Scholar entry. 3 Such as the International Journal of Wellbeing. 4 For overviews of the controversies surrounding the issue of just what the philosophical question of well-​being is, see Campbell 2016 and Lin forthcoming b. 5 So long as each category is understood broadly, so that, for example, a Happiness Theory counts as Hedonistic, an Aim-​Achievement Theory counts as a version of the Desire-​ Fulfillment Theory, and a Perfectionist Theory counts as a version of the Objective List Theory, this taxonomy seems close to exhaustive. It may exclude Value-​Realization Theories, on which what would be best for someone is what would best realize their values; proponents of these views are often at pains to emphasize that one’s values are not merely one’s desires. For Happiness Theories, see Sumner 1996, ch. 6 (Sumner would not classify his own view as Hedonistic) and Feldman 2010, Pt. II (Feldman would classify his Happiness Theory as Hedonistic). For a theory that includes aim achievement, see Scanlon 1998. For a Perfectionist Theory, see Kraut 2009. For Value-​Realization Theories, see Raibley 2010 and Dorsey 2012. Finally, for recent defenses of a Hedonistic Theory, see Crisp 2006; a Desire-​Fulfillment Theory, see Heathwood 2005; and an Objective List Theory, see Rice 2013. 6 The “adjusted” hedonistic theories put forward in Feldman 2004 violate this claim, but it is controversial whether these theories are genuine forms of hedonism. 7 Many philosophers endorse the heterogeneity objection (for references, see Heathwood 2007:  26, note 8). For recent replies to the heterogeneity objection, see Bramble (2013:  209–​211) and Lin (forthcoming a). The Distinctive Feeling Theory should be distinguished from a similar view, the Hedonic Tone Theory (Broad 1930: 229–​231, Kagan 1992: 172), which Parfit would also reject. 8 The view also helps Parfit later on in the Appendix. He relies on it without comment in arguing against a certain proposal for saving Summative Desire-​Fulfillment Theories from his drug-​addiction objection (see pp. 497–​498). I explain Summative Theories and that objection in section 5 below. 9 Parfit does not here fuss over any possible differences between disliked sensations and unwanted-​when-​experienced sensations, but he does fuss over this in later work (Parfit 2011: ch. 2, §6). There he prefers a liking-​based theory over a desire-​based theory. 10 Parfit actually applies the case to issues of well-​being before we reach Appendix I, in Appendix C (468). 11 See footnote 25 on p.  532 in the first edition of Reasons and Persons. Griffin’s then-​ forthcoming book was at that point entitled Welfare rather than Well-​Being. 12 See, for example, Scanlon 1998 (113–​123), Lukas 2010, Fletcher 2016 (§§ 2.3, 2.6), and Heathwood 2016 (141–​142). 13 On the problem of self-​sacrifice, see Overvold 1980, Rosati 2009, and Heathwood 2011. 14 I discuss this sort of counterexample in Heathwood 2016: 141.

“What Makes Someone’s Life Go Best”  111

15 I defend a Same-​World Subjectivist theory in Heathwood 2005, though, to distinguish it from idealized desire theories, I describe it as an “actualist” theory. A somewhat different Same-​World approach –​Global rather than Summative –​is defended in Bykvist (2006). 16 As in Bricker (1980, pp. 398–​400). 17 The drug changes you gradually rather than abruptly to ensure that you will survive the changes, in case a psychological theory of personal identity is true (see Andrea Sauchelli’s chapter in this volume). 18 See the essays by Hedden and Sauchelli in this volume. 19 Summative Desire-​ Fulfillment Theories are defended in Heathwood 2005 and Heathwood 2006. 20 Yet another problem with, or at least another thing worth noting about the appeal to global desires is that it is not clear how a Preference Hedonist would make use of the notion. 21 My own view is that this connection between pleasure and desire is true on only one sense of ‘desire’ (Heathwood 2019). 22 That is, all of someone’s desires after any upstream filters –​such as restrictions to desires about one’s life on the Success Theory or restrictions to desires about one’s experiences on Preference Hedonism –​have been applied. 23 Although it seems undeniable that better than is transitive, this has been denied. E.g., by Temkin (1987) and Rachels (1998). 24 We also might cite the idea that “we have no reason to trust anyone’s intuitions about very large numbers” (Broome 2004: 57–​59; see also Huemer 2008: 908–​909) and the fact that we commonly make intuitive mistakes in the compounding of small quantities (Huemer 2008:  909–​910; cf. the “mistakes in moral mathematics” Parfit discusses in §§27–​28). 25 See the chapter in this volume by Melinda Roberts for more and related details. 26 E.g., Robert Adams (1999:  93–​101), Fred Feldman (2004:  119–​122), Shelly Kagan (2009), and William Lauinger (2012). For more on Hybrid Theories (and a concept of them more capacious than the category of theory Parfit is delimiting here), see Woodard 2015. 27 For an argument that they don’t, see Heathwood 2010 (652–​653). 28 Thanks to Ben Eggleston, Brian Hedden, Eden Lin, Susanne Mantel, Andrea Sauchelli, and audiences at Saarland University and the University of Colorado Boulder’s Center for Values and Social Policy.

Bibliography Adams, R.M. 1999. Finite and Infinite Goods:  A Framework for Ethics. New  York:  Oxford University Press. Bramble, B. 2013. “The Distinctive Feeling Theory of Pleasure.” Philosophical Studies 162 (2): 201–​217. Bricker, P. 1980. “Prudence.” Journal of Philosophy 77 (7): 381–​401. Broad, C.D. 1930. Five Types of Ethical Theory. London: Kegan Paul. Broome, J. 2004. Weighing Lives.Vol. 116. Oxford University Press. Bykvist, K. 2006. “Prudence for Changing Selves.” Utilitas 18 (3): 264–​283. Campbell, S.M. 2016. “The Concept of Well-​Being.” In The Routledge Handbook of Philosophy of Well-​Being, edited by G. Fletcher, 402–​414. Routledge. Crisp, R. 2006. “Hedonism Reconsidered.” Philosophy and Phenomenological Research 73 (3): 619–​645.

112  Chris Heathwood

Dorsey, D. 2012. “Subjectivism without Desire.” Philosophical Review 121 (3): 407–​442. Feldman, F. 2004. Pleasure and the Good Life. Oxford: Oxford University Press. Feldman, F. 2010. What Is This Thing Called Happiness? Oxford: Oxford University Press. Fletcher, G. 2016. The Philosophy of Well-​ Being:  An Introduction. 1 edition. London; New York: Routledge. Griffin, J. 1986. Well-​Being: Its Meaning, Measurement, and Moral Importance. Oxford: Oxford University Press. Heathwood, C. 2005. “The Problem of Defective Desires.” Australasian Journal of Philosophy 83 (4): 487–​504. Heathwood, C. 2006. “Desire Satisfactionism and Hedonism.” Philosophical Studies 128 (3): 539–​563. Heathwood, C. 2007. “The Reduction of Sensory Pleasure to Desire.” Philosophical Studies 133 (1): 23–​44. Heathwood, C. 2010. “Welfare.” In The Routledge Companion to Ethics, edited by J. Skorupski, 645–​655. Routledge. Heathwood, C. 2011. “Preferentism and Self-​ Sacrifice.” Pacific Philosophical Quarterly 92 (1): 18–​38. Heathwood, C. 2016.“Desire-​Fulfillment Theory.” In The Routledge Handbook of the Philosophy of Well-​Being, edited by G. Fletcher, 135–​147. Routledge. Heathwood, C. 2019. “Which Desires Are Relevant to Well-​Being?” Noûs 53 (3): 664–688. Huemer, M. 2008. “In Defence of Repugnance.” Mind 117 (468): 899–​933. Kagan, S. 1992. “The Limits of Well-​Being.” Social Philosophy and Policy 9 (02): 169–​189. Kalat, J. 2016. Introduction to Psychology. Eleventh Edition. Boston, MA: Cengage Learning. Kraut, R. 2007. What Is Good and Why: The Ethics of Well-​Being. Harvard University Press. Lauinger, W. 2012. Well-​Being and Theism: Linking Ethics to God. Continuum. Lin, E. 2019.“Why Subjectivists About Welfare Needn’t Idealize.” Pacific Philosophical Quarterly. Lin, E. Forthcoming a. “Attitudinal and Phenomenological Theories of Pleasure.” Philosophy and Phenomenological Research. Lin, E. Forthcoming b. “Well-Being I: The Concept of Well-​Being.” Philosophy Compass. Lukas, M. 2010. “Desire Satisfactionism and the Problem of Irrelevant Desires.” Journal of Ethics and Social Philosophy 4 (2): 1–​25. McTaggart, J.M.E. 1927. The Nature of Existence. Vol. II. Cambridge: Cambridge University Press. Moore, G.E. 1903. Principia Ethica. Cambridge: Cambridge University Press. Overvold, M.C. 1980. “Self-​Interest and the Concept of Self-​Sacrifice.” Canadian Journal of Philosophy 10 (1): 105–​118. Parfit, D. 1984. Reasons and Persons. Revised edition. Oxford: Oxford University Press. Parfit, D. 1986. “Overpopulation and the Quality of Life.” In Applied Ethics, edited by P. Singer, 145–​164. Oxford: Oxford University Press. Parfit, D.. 2011. On What Matters.Vol. I. Oxford: Oxford University Press. Rachels, S. 1998. “Counterexamples to the Transitivity of Better Than.” Australasian Journal of Philosophy 76 (1): 71–​83. Raibley, J.R. 2010. “Well-​Being and the Priority of Values.” Social Theory and Practice 36 (4): 593–​620. Railton, P. 1986. “Facts and Values.” Philosophical Topics 14: 5–​31. Rawls, J. 1971. A Theory of Justice. Cambridge, MA: Belknap Press. Rice, C.M. 2013.“Defending the Objective List Theory of Well-​Being.” Ratio 26 (2): 196–​211. Rosati, C.S. 2009. “Self-​Interest and Self-​Sacrifice.” Proceedings of the Aristotelian Society 109 (1pt3): 311–​325.

“What Makes Someone’s Life Go Best”  113

Scanlon, T.M. 1998. What We Owe to Each Other. Cambridge, MA: Belknap Press of Harvard University Press. Shafer-​Landau, R. 2012. Ethical Theory: An Anthology. 2nd Edition. Wiley-​Blackwell. Sidgwick, H. 1907. The Methods of Ethics. 7th ed. Hackett Publishing. Sumner, L.W. 1996. Welfare, Happiness, and Ethics. Oxford: Clarendon Press. Temkin, L.S. 1987. “Intransitivity and the Mere Addition Paradox.” Philosophy and Public Affairs 16 (2): 138–​187. Woodard, C. 2015. “Hybrid Theories.” In The Routledge Handbook of Philosophy of Well-​Being, edited by G. Fletcher, 161–​174. Routledge.

PART II

6 IN A DIFFERENT LEAGUE Intransitivity, betterness, and league-​based satisficing Chrisoula Andreou

Introduction Derek Parfit’s legacy includes a variety of now-​famous puzzles and paradoxes in value theory. In many cases, the puzzles and paradoxes depend crucially on the traditional assumption that the relation “better than” (or, more precisely, “all-​things-​ considered better than”) is transitive, where “better than” qualifies as transitive if, given any alternatives w, x, and y, if w is better than x and x is better than y, then w is better than y. Parfit recognized the key role of the traditional assumption; and, even though he allowed and, indeed, argued that, in light of the possibility of options being “imprecisely equally good,” the relation “not better than” is not transitive, he sought to resolve the quandaries he raised without simply giving up on the idea that the relation “better than” is transitive.1 Still, Parfit’s quandaries, and closely related quandaries developed by others, have led some philosophers to question the transitivity of betterness. For, the quandaries have been interpreted by some as supporting the possibility of betterness cycles, where a betterness cycle can be represented as in Figure 6.1; and if betterness cycles are possible, then “better than” is not transitive.2 In this chapter, I contribute to this debate, using Parfit’s enduring engagement with the “Repugnant Conclusion” as my starting point. Ultimately, I take the possibility of betterness cycles seriously and consider the implications of accepting the intransitivity of “better than.” I argue that if betterness cycles are indeed possible, then a distinctive form of satisficing that involves reasoning in terms of leagues (understood as categories characterized by different grades or levels of quality) plays a crucial role in proper reasoning about what to do. Significantly, and as I will explain, some kind of focus on leagues turns out to be appropriate even if “better than” is transitive and Parfit is on the right track when, as we’ll see, he interprets what might seem like a(n) (all-​things-​considered) betterness cycle as, instead, a case

118  Chrisoula Andreou

>

>

b

>

a

c

...

> m FIGURE 6.1   Read “w

is better than w.”

involving some imprecisely equally good options. Indeed, a focus on leagues turns out to be appropriate even if Parfit’s appeal to imprecisely equally good options is unnecessary because there is actually an optimal option (one that is as good or better than any alternative) in all hard cases involving apparent betterness cycles but it is just beyond us, in practice or in principle, to reliably determine which option is optimal. I proceed as follows:  In section 6.1, I  discuss the “Repugnant Conclusion” that Parfit introduced and grappled with in Reasons and Persons as well as a relatively recent argument he put forward reinforcing how challenging it is to evade the conclusion. Although Parfit’s reasoning allows that “not better than” is intransitive―and, relatedly, allows that options can fail to be precisely comparable and yet still be comparable as roughly equally good―it presupposes that “better than” is transitive. In section 6.2, I consider the now familiar―even if still highly controversial―suggestion that “better than” is intransitive. I focus in particular on Larry Temkin’s defense of the idea, which he’s developed, in part, in response to Parfit’s puzzles. In section 6.3, I delve into the implications of the suggestion that “better than” is intransitive―and, more specifically, that betterness cycles are possible―with respect to our understanding of proper reasoning about what to do. In section 6.4, I relate my reasoning to Parfit’s relatively recent proposed solution to the problem of the “Repugnant Conclusion,” and end with some remarks regarding a potential point of agreement that captures some crucial practical guidance.

6.1  Routes to the Repugnant Conclusion and Parfit’s proposed evasion As part of his discussion of population ethics in part 4 of Reasons and Persons, Parfit (1987) argues against the “Impersonal Total Principle,” according to which “if other things are equal, the best outcome is the one in which there would be the greatest quantity of whatever makes life worth living” (387). The problem Parfit raises for the principle is that it implies the “Repugnant Conclusion,” according to which “for any possible population of at least ten billion people, all with a very high quality of life, there must be some much larger imaginable population whose existence, if other things are equal, would be better, even though its members have lives that are barely worth living” (388). Given the assumption that, for any quantity x of whatever makes life worth living, it is in principle possible for any population, whatever

League-based satisficing  119

its size, to be composed of individuals who each have quantity x of whatever makes life worth living, the connection between the Impersonal Total Principle and the Repugnant Conclusion is easy to see. For suppose that δ is some small quantity of whatever makes life worth living, and that life with that quantity of whatever makes life worth living is barely worth living. Then, for any quantity y of whatever makes life worth living, there is some positive whole number n such that n times δ is greater than y. It follows that, no matter how much of whatever makes life worth living there is in a population of individuals all with a very high quality of life, there can be more of whatever makes life worth living in a population of individuals all with lives that are barely worth living, so long as the latter group is big enough. We thus arrive at the Repugnant Conclusion. But, as Parfit’s label suggests, the Repugnant Conclusion is “hard to accept,” and so figures as a problematic implication of the Impersonal Total Principle (Parfit 1987, 388). At least intuitively, it seems clear that, other things equal, a very large population of individuals all with a very high quality of life is better than a much larger population of individuals all with lives barely worth living, no matter how enormous the latter population. Unfortunately, as Parfit makes clear, avoiding the Repugnant Conclusion is not a simple matter. For, even if one gladly gives up the Impersonal Total Principle, granting that quantity is not the only thing that matters (since distribution, maximums, minimums, and averages can matter as well), plausible alternative principles seem to run into the same problem or else generate other unacceptable results. Indeed, as is made apparent in Parfit’s recent follow-​up work on the Repugnant Conclusion, there is a positive argument for the conclusion that takes as its starting point a premise that is much more modest than the Impersonal Total Principle, namely the premise that “compared with the existence of many people who would all have lives that were equally worth living, there are some much larger numbers of people whose existence would be better, though these people would all have lives that would be slightly less worth living” (2016, 116). As Parfit explains, this premise accords with the idea that “a slight loss of quality [in lives that are worth living] could be outweighed by a sufficient gain in quantity,” which seems plausible unless one gives quantity no weight at all (120). Following Parfit, I will, at least for the sake of argument, grant the initial plausibility of this idea. As will soon become apparent, the issues that are raised by the idea are also raised by intrapersonal cases in which the quantity and quality of some positive or negative experience are both clearly relevant; and it is the more general issues that are raised, rather than the specific premise Parfit focuses on, that I am ultimately interested in. For now, notice that, applying the premise repeatedly, we can imagine a series of worlds A, B, C, …, Z, where each world after the first one in the series is better than the preceding one in the series, and in which A is a world with a large number of people (say ten billion) all with an equally very high quality of life and Z is a world with a much larger number of people whose lives are barely worth living. Assuming that “better than” is transitive, it follows that Z is better than A. The preceding reasoning for the conclusion that Z is better than A captures the “Continuum Argument for the Repugnant Conclusion.”3

120  Chrisoula Andreou

Ultimately, Parfit resists the Repugnant Conclusion via an appeal to “imprecis[e]‌ comparab[ility]” (2016, 116). Putting aside complications and variations that need not detain us here, Parfit’s reasoning can be roughly captured as follows: For at least one adjacent pair of worlds in the series A, B, …, Z, it is not true that the larger world is better. That is not to say that the larger world is worse or that the two worlds are exactly equally good. Rather, the two worlds are “imprecisely equally good” (121); otherwise put, though not precisely rankable as one better than the other or as exactly equally good, the two worlds are comparable as roughly equally good. Since these adjacent worlds are such that the larger world is not better, the argument for the conclusion that Z is better than A is blocked. Notably, Parfit allows that intuition speaks against his claim that, for at least some adjacent worlds in the imagined series, the larger world is not better. His defense of the claim rests on the idea that the claim is less implausible than the Repugnant Conclusion. One might worry that, to the extent that Parfit’s introduction of imprecise comparability is compelling, it seems like it could apply to every adjacent pair of worlds in the imagined series, and this threatens to lead us to the unwelcome conclusion that A is not better than Z. For, if each adjacent pair of worlds is such that the two worlds are “imprecisely equally good,” then, for each adjacent pair of worlds, the smaller world is not better than the larger world. And, assuming that “not better than” is transitive, it follows that A  is not better than Z.  But, as Parfit explains, “imprecisely equally good” is clearly not transitive, and so, once one grants the possibility of imprecise comparability, one should also grant that “not better than” is not transitive. Consider, for example, that “even if [due to imprecise comparability] your being a writer would not be better than your being a doctor, which would not be better than your being a slightly less successful writer, your being a writer would be better than your being a slightly less successful writer” (Parfit 2016, 120). So Parfit’s proposal for resisting the Repugnant Conclusion doesn’t force him to the conclusion that A is not better than Z.

6.2  An alternative way of evading the Repugnant Conclusion: The Continuum Argument for Intransitivity I turn now to an altogether different way of evading the Repugnant Conclusion, one that resists the Continuum Argument for the Repugnant Conclusion with a different continuum argument, namely the Continuum Argument for Intransitivity. I  focus in particular on Larry Temkin’s Continuum Argument for Intransitivity, which employs a variation of a counterexample to the transitivity of betterness developed by Stuart Rachels.4 This alternative continuum argument suggests that the continuum argument Parfit raises is not really an argument for the Repugnant Conclusion but an argument against the transitivity of “better than.”5 Like Parfit’s Continuum Argument, Temkin’s Continuum Argument is based on a seemingly modest premise, namely the premise that “for any unpleasant or ‘negative’ experience, no matter what the intensity and duration of that experience, it would be better to have that experience than one that was only a little

League-based satisficing  121

less intense but twice as long” (Temkin 1996, 179). Applying this premise repeatedly, we can imagine a series of equally long lives A, B, C, …, Z, where each life before the last one in the series is better than the following one in the series, and in which A contains two years of excruciating torture and Z contains many many years of an annoying hangnail, but no torture. It follows that either “better than” is intransitive or A (which involves two years of excruciating torture) is better than Z (which involves many many years of an annoying hangnail, but no torture). For Temkin, the Continuum Argument is plausibly interpreted as a counterexample to the supposed transitivity of “better than.” This is because, though the example may be in some ways controversial, it clearly “exemplifies that together a sufficient number of differences in degree can sometimes amount to a difference in kind”; and “the relative significance of factors relevant for comparing alternatives merely differing in degree, may differ from the relative significance of those factors for comparing alternatives differing in kind” (194). So, for example, when comparing pains that differ merely in degree (e.g., a hangnail versus a slightly more annoying hangnail), duration is a very significant factor, but when comparing pains that differ in kind (e.g., a hangnail versus torture), duration may be much less significant (even if it is still given some weight). If the possibility of such variation in the relative significance of factors depending on the alternatives compared is acknowledged, then we have every reason to believe that “better than” is intransitive, apart from any particular proposed counterexample to the transitivity of betterness. Returning to the Continuum Argument that Parfit raises, Temkin’s reasoning suggests that―assuming that the quantity of lives that are worth living matters, even if other factors (e.g., distribution, maximums, minimums, and averages) matter too―we should expect that the relative significance of quantity will vary, with it being more important in the comparison of “adjacent” worlds, in which there is only a slight difference in quality of life, than in “distant” worlds, in which quality of life in the two worlds are definitely in completely different leagues.6 But, if this is right, then the Continuum Argument that Parfit raises is just another counterexample to the transitivity of “better than” in disguise; it does not support the Repugnant Conclusion, even if Parfit’s appeal to imprecise comparability is off track.

6.3  The Intransitivity Thesis and league-​based satisficing The suggestion that “better than” is intransitive (henceforth the Intransitivity Thesis) remains highly controversial. Nonetheless, I think enough work has been done by Temkin and Rachels, among others, to at least support Temkin’s “modest” plea that the Intransitivity Thesis and, in particular, the possibility of betterness cycles be “taken seriously” (176). My aim is to consider the implications of this possibility for practical reasoning. As indicated above, I will argue that if betterness cycles are indeed possible, then a neglected form of satisficing that involves reasoning in terms of leagues plays a crucial role in proper reasoning about what to do. As will become apparent in the next section, some kind of focus on leagues turns out to be appropriate even if “better than” is transitive.

122  Chrisoula Andreou

The first thing to recognize is that, insofar as some sets of options form betterness cycles, guiding oneself by a pairwise comparison of the options one faces (always choosing the better over the worse whenever this description applies) cannot be hailed as a foolproof strategy for choosing well, even given the simplifying assumptions that (i) every pair of options is such that one option is better than the other or the options are exactly equally good, and (ii) the options are readily available, as is any relevant information about them. For, suppose that, given the options O0, O1,…, Om, for each n between 1 and m, On is better than On-​1, and yet O0 is better than Om. Then, if an agent uses pairwise comparison to choose between the options, which option she ends up with can be completely controlled by the order in which the pairs are considered. For example, for any s between 0 and m, the agent will opt for Os if she starts her pairwise comparisons at the next setting in the loop (which will be O0 if s=m and Os+1 otherwise) and works her way around the loop until she makes it all the way to Os, discarding worse options until Os is the only one left. The option selected via pairwise comparison is thus, in an important sense, completely arbitrary. This might not be too problematic if the options are all in the same league (as when the options are, for example, all great). But if, as Temkin suggests, “together a sufficient number of differences in degree can sometimes amount to a difference in kind,” we cannot assume that all the options are in the same league. If, for instance, O0 is Temkin’s enduring hangnail option Z and Om is Temkin’s excruciating torture option A, then O0 may be acceptable while Om may be unacceptable. Since using pairwise comparison can lead just as easily to Om as to O0, it is not a foolproof strategy for choosing well. The moral, in short, is that if “better than” is intransitive and betterness cycles are possible, practical reason calls for a more holistic strategy for choosing than pairwise comparison―one that involves grouping options into (fuzzily bounded) leagues (with some options falling into fuzzily bounded gray areas).7 Grouping options into leagues is related to having what I call “categorical [or category-​based] appraisal responses” to them, as contrasted with “relational appraisal responses” to them (Andreou 2016, 567): In appraising an alternative, I might respond categorically with something like ‘X is terrible’ or I might respond relationally with something like ‘X is worse than Y.’ The first sort of response is categorical in the sense that it indicates the appraisal category that I see X as falling in. The second sort of response provides no such category information. To appraise X as worse than Y leaves completely open the question of what category I place X in on the spectrum from, say, terrible to fantastic. It indicates only how I  appraise X and Y in relation to each other.8 One can think of grouping options into leagues along the lines of assigning them grades that are defined qualitatively (as is done with, for example, eggs, meat, and student papers), where the grading system used can be more or less refined.9 In a very basic grading system, the only two available grades might be “satisfactory”

League-based satisficing  123

and “unsatisfactory.” In a more refined system, “satisfactory” might be subdivided into “excellent,” “good,” “fair,” and “marginal.” When what is crucial is whether something falls in a certain category rather than out of it, we can ask whether it “makes the grade”; and the answer can have serious consequences (determining, for example, eligibility to graduate).10 While the number of categories used might be somewhat arbitrary and the boundaries between them vague, even the crudest of grading systems is more useful than pairwise comparison in cases involving betterness cycles. For, insofar as one puts aside the strategy of pairwise comparison and settles on an option (say Os) in the highest available grade range (knowing that a better option, namely Os+1, is available), one will at least avoid unacceptable options when acceptable options are available. Significantly, given the possibility of betterness cycles, options, by hypothesis, cannot invariably be assigned numerical grades that allow them to be neatly ordered from best to worst. Still, as suggested above, insofar as some options in a betterness cycle can be acceptable while others are unacceptable, options in a betterness cycle need not all be in the same league, and so (qualitative) grading (with fuzzily bounded gray areas) remains possible; moreover, identifying and settling on a top-​ grade option (where the top grade might just be “acceptable” or “satisfactory”) can be potentially vital in terms of avoiding a problematic pattern of choices. Note that the cumulative transition from, say, an “acceptable”-​grade option to an “unacceptable”-​ g rade option that a betterness cycle might prompt―if one is focused on invariably choosing the better option in a series of pairwise choices―need not involve any single choice in which a lower-​g rade option is favored and selected over a higher-​g rade option. Indeed, insofar as leagues can be separated by fuzzily bounded gray areas, one can get from an acceptable option to an unacceptable option without ever encountering a sharp switchover point from an option that is not determinately unacceptable to an adjacent option that is determinately unacceptable. Note also that a grading system can be, normally is, and, to be helpful in relation to betterness cycles, must be such that the grades can be neatly ordered from best to worst. Consider, for example, that even if a set of options, some of which are acceptable, some of which are unacceptable, and some of which fall in a fuzzily bounded gray area, cannot be ordered from best to worst, the grades “acceptable” and “unacceptable” can be neatly ordered, with (if these are the only two grades available) “acceptable” being the top grade and “unacceptable” being the bottom grade. And it is precisely because these grades can be neatly ordered that it is not okay to end up with an unacceptable option when acceptable options are available, even in the context of a betterness cycle. Importantly, the holistic strategy that can be used, in the context of a betterness cycle, to protect against this pitfall, namely the strategy of settling on a non-​ optimal but top-​grade option, qualifies as a sort of satisficing, but one that can be distinguished from two philosophically familiar forms of satisficing that can also involve, at least loosely speaking, knowingly passing up a better alternative.11 One

124  Chrisoula Andreou

of the philosophically familiar forms of satisficing in question involves maximizing relative to the costs associated with searching for and inquiring into alternative options. Suppose, for example, that one is on one’s way out the door to buy a $30 box of dumplings for a party tonight. One then recalls that a store other than the store one usually shops at, though one can’t quite remember which other store, has the dumplings on sale for $28.00. Considering the costs in time and energy that it will take to find out which store has the dumplings on sale, one satisfices and accepts the “worse deal,” recognizing that, once the costs in time and energy of pursuing the “better deal” are taken into account, maximizing speaks in favor of accepting the “worse deal.” This form of satisficing is not at issue in the cases of intransitivity we’ve been considering, since we have allowed the cases to be ones in which the sorts of cost involved in the preceding shopping case are not in play. The other philosophically familiar, though much more controversial, form of satisficing in question involves a willingness to pass up an optimal option that is readily available and select an option that is merely good enough.12 One might, for example, pass over the best dessert at the dessert table and opt for a dessert that is not as good, but is nonetheless perfectly fine. This form of satisficing, which is often dismissed as irrational (or else as really maximizing relative to some cost that is not explicit), is also not at issue in the cases of intransitivity we’ve been considering, since these cases are such that there is no optimal option, though some of the options are in different leagues; indeed, it is precisely because there is no optimal option that settling on a non-​optimal but top-​g rade option is necessary.13 Notably, given an adequately refined grading system, the options that fall squarely within a category might be plausibly described as roughly equally good, even if some are better than others. So, interestingly, the view that I  have been developing regarding the significance of leagues is tied, as is Parfit’s view, to the significance of rough equality. In the next section, I relate my reasoning to Parfit’s response to the threat of the Repugnant Conclusion, and follow up with some remarks regarding a potential point of agreement that captures some crucial practical guidance.

6.4  Parfit’s proposed evasion revisited Parfit attempts to evade the Repugnant Conclusion by maintaining that, for some adjacent worlds in his imagined series A, B, …, Z (where A is a world with a large number of people all with an equally very high quality of life and Z is a world with a much larger number of people whose lives are barely worth living), it is not true that the larger world is better. Rather, the two worlds are not precisely comparable; neither is better than the other, nor are they exactly equally good, though they may be “imprecisely equally good.” If, alternatively, one takes seriously Temkin’s suggestion that “better than” is not transitive, one can instead allow that the larger world is better, but that this sort of pairwise comparison is not what one should be focusing on. Instead, one should think of the options as falling into leagues (each containing options of the same grade) and settle on a top-​grade option, where all such options are, relative

League-based satisficing  125

to the level of refinement of the grading system used, roughly equally good, though some are better than others. Significantly, the italicized portion of the preceding conclusion is a position that Parfit will presumably have to embrace as well, since, in claiming that some, and perhaps all, adjacent options are not precisely comparable, he too must recognize the inadequacy of using pairwise comparison to select an option. For, suppose that all adjacent options are such that either the larger world is better or the two worlds are imprecisely equally good. Suppose further that one neglects to think of options A, B, …, Z as falling into leagues and assumes it is always permissible to choose (i) better options over worse options, and (ii) any of two “imprecisely equally good” options; then one may still end up going from A to Z via a series of purportedly permissible moves, which is problematic, since, while A seems like a good world, Z seems fair at best (and so is in a different, distinctly lower league). In the end, any answer to the theoretical question concerning whether “better than” is transitive may require one to bite some bullet (in the form of accepting some deeply counter-​intuitive claim) or to find some distinction between senses of “better than,” each of which captures some of the intuitions in play.14 Pursuing the theoretical question might be revealing in various ways, but it is good to know that, in any case, the potential point of agreement laid out toward the end of the preceding paragraph provides some practical guidance. It might be suggested that practical guidance was never needed, since both Parfit’s series of worlds and Temkin’s series of lives do not figure as realistic alternatives that we might at some point have to choose between. But, as the literature on intransitivity has made apparent, there are plenty of mundane and familiar cases that are structurally similar to the cases Parfit and Temkin consider. Consider, for example, the case of someone who enjoys eating desserts but also cares about his appearance.15 His case might be such that there is a continuum of options O0 to Om, where O0 involves eating 0 treats this year and Om involves eating m treats this year, and where, intuitively, it seems like, for each n between 1 and m, On is better than On-​1, and yet O0 is better than Om. In such a case, proceeding naively is quite dangerous and some practical guidance is indeed needed. As suggested above, thinking in terms of leagues can help; for, even the crudest grading system of “satisfactory” versus “unsatisfactory” can at least keep one from behaving in a way that is grossly self-​defeating (assuming one settles on an option that falls squarely in the former category). It is worth emphasizing that the importance of thinking in terms of leagues persists even if one is tempted to resist the possibility of betterness cycles and imprecise comparability altogether and maintain instead that (i) apparent betterness cycles are purely illusory and that (ii) although, in hard cases involving apparent betterness cycles, it is admittedly unrealistic, or perhaps even impossible to identify an optimal option (understood as one that is as good or better than any alternative), there is some such option. Though this response presupposes, rather than supporting, the acyclicity of betterness, practically speaking, even if it were compelling, the need for

126  Chrisoula Andreou

league-​based satisficing remains, since our ignorance would leave us with the same practical challenge as in a case involving a genuine betterness cycle.

Conclusion Taking Parfit’s enduring engagement with the “Repugnant Conclusion” as my starting point, I have explored the associated controversy concerning the traditional assumption that “better than” is transitive. Ultimately, I’ve argued that if betterness cycles are possible, then a distinctive form of satisficing that involves reasoning in terms of leagues plays a crucial role in proper reasoning about what to do. In the end, even if betterness is transitive, the strategy for responding to apparent betterness cycles (illusory or not) that I’ve described, which involves grouping options into leagues (each containing options of the same grade) and settling on a top-​grade option, can provide helpful practical guidance, giving us a rational way of proceeding without having to resolve the theoretical question of how best to understand “better  than.”

Acknowledgments My thanks to Elijah Millgram, Andrea Sauchelli, Mike White, participants at the Halbert Network Conference on Collective and Temporally Extended Rights and Wrongs, participants at the Tennessee Value and Agency Conference, participants at the UTK/​ CRE workshop on Challenging Choice Situations, and an anonymous reviewer for helpful comments on an earlier draft of this chapter. I am also grateful for supporting research funds from the University of Utah and from the Charles H.  Monson Esteemed Scholar Award.

Notes 1 Interestingly, he did, in his unfinished but posthumously edited and published paper “Future People, the Non-​Identity Problem, and Person-​Affecting Principles” (2017), grant that there are some non-​transitive senses of “better than.” Still, he remained reluctant to simply abandon the idea that “all-​things-​considered better than” is transitive; relatedly, he did not cast any non-​transitive sense of “better than” as the key to evading the (to-​be-​ discussed) “Repugnant Conclusion” that he repeatedly sought to evade. 2 See, for example, Rachels (1998) and Temkin (1996) and (2012). 3 Note that I have glossed over many complications that Parfit delves into because, for my purposes here, they are tangential. This continuum argument is closely related to some earlier continuum arguments put forward by Parfit (1984), including, for example, his significantly more complicated argument for the third version of the “Mere Addition Paradox” in part 4, section 149 of Reasons and Persons. 4 Rachels’s original example was unpublished when Temkin cited it in 1996; but see, for example, Rachels (1998) for a published version of the example. 5 See, relatedly, Rachels (2004). 6 I will say more about leagues shortly. As will become apparent, in describing two options as definitely in completely different leagues, I do mean to suggest that one of the options is superior to the other.

League-based satisficing  127

7 The holistic strategy that I will explicitly lay out in this section is related to the revisionary conceptions of instrumental rationality and betterness that I propose in Andreou (2016) and Andreou (2019); but, things are significantly complicated by the fact that I there leave room for the view that “better than” is transitive, in which case there may be no place for the sort of satisficing that the strategy that I will describe presently involves, though there may be room for a variation on the strategy. In any case, and as will soon become apparent, features of the strategy can be recognized as important even given disagreement regarding the theoretical question of how best to understand “better than.” 8 See Andreou (2015) for a detailed discussion of the distinction between categorical (or category-​based) appraisal responses and relational appraisal responses, and Andreou (2016; 2019) for discussion of the significance of category-​based appraisal responses in cases involving rationally cyclic preferences and/​or moral preferability cycles. According to the conception of instrumental rationality I develop in Andreou (2016), “rationally-​governed choice will not lead one to an alternative that is in a lower appraisal category than another available alternative (at least if there are no unanticipated developments and the set of appraisal categories is finite)” (571). Points closely related to some of the key points I make about category-​based appraisal responses and preference cycles in that discussion are integrated into the discussion regarding betterness cycles in the remainder of this section. 9 See, relatedly, Hsieh (2005). Note that, unlike Hsieh, I do not insist that how two options compare is a function of the (resolution of the) grading system in play, with all options receiving the same grade counting as equal. 10 More dramatically, whether a prosecutor’s case makes the grade with respect to establishing guilt “beyond a shadow of a doubt” can determine whether punishment is legally permissible. More controversially, whether an outcome would make the grade in terms of qualifying as a “catastrophic moral horror” might determine whether violating the constraint against killing an innocent person, in order to avoid the outcome, is morally permissible. (Alon Harel [personal correspondence] highlights both of these examples as relevant to my focus on the importance of leagues.The question of whether certain side constraints “may be violated in order to avoid catastrophic moral horror” is mentioned in (Nozick 1974, 30).) 11 Not all cases of satisficing involve knowingly passing up a better alternative (even loosely speaking). For example, in some cases of satisficing, one opts for the first satisfactory alternative, where this may or may not involve passing up a better alternative, since there may or may not be a better alternative available―indeed, satisficing may be appealing at least in part because finding out if a better alternative is available is too cognitively demanding. See, relatedly, Simon (1955) on satisficing and “psychological limits” (101). 12 See, relatedly, Slote (1989, ­chapter 1) on “satisficing moderation.” 13 The form of satisficing that is at issue in the cases of intransitivity we’ve been considering cannot be described, even for an agent whose preferences track betterness, as satisficing with respect to utility, since the relevant preferences would not conform to the axioms that must be satisfied for there to be a utility function. See Dreier (2004) for reasons to doubt that satisficing with respect to utility makes sense (at least in finite cases). Still, if “better than” is intransitive, rational satisficing, where satisficing involves knowingly passing up a better alternative, remains a possibility (even with respect to a finite set of options). 14 My distinction in Andreou (2019) between “is rationally preferred to” and “is better than” might be recast as a distinction between two alternative construals of “is better than,” one that leaves room for betterness cycles and another that does not. As I there explain, the construal that leaves no room for betterness cycles and allows “is better than” to be transitive has the key advantage of fitting with the idea that respecting betterness judgments is crucial to making good choices. Still, there remains, in my view,

128  Chrisoula Andreou

an important role for an intransitive relation that some equate with “is better than” and that fits with my construal of “is rationally preferred to.” 15 This example is adapted from Quinn (1993).

References Andreou, C. 2015. “Parity, Comparability, and Choice,” Journal of Philosophy, CXII: 5–​22. Andreou, C. 2016. “Andreou. The Real Puzzle of the Self-​Torturer:  Uncovering a New Dimension of Instrumental Rationality.” Canadian Journal of Philosophy, 45: 562–​575. Andreou, C. 2019. “Better Than,” Philosophical Studies, 176: 1621–​1638. 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. Hsieh, N. 2005. “Equality, Clumpiness and Incomparability,” Utilitas, xvii, 2: 180–​204. Nozick, R. 1974 [2013]. Anarchy, State, and Utopia. New York: Basic Books. Parfit, D. 1984. Reasons and Persons. Oxford: Clarendon Press. Parfit, D. 2016. “Can We Avoid the Repugnant Conclusion?” Theoria, 82: 110–​127. Parfit, D. 2017. “Future People, the Non-​Identity Problem, and Person-​Affecting Principles,” Philosophy & Public Affairs, 45: 118–​157. Quinn, W. 1993. “The Puzzle of the Self-​ Torturer,” in Morality and Action. Cambridge University Press. Rachels, S. 1998. “Counterexamples to the Transitivity of Better Than,” Australasian Journal of Philosophy, 76: 71–​83. Rachels, S. 2004. “Repugnance or Intransitivity,” in The Repugnant Conclusion:  Essays on Population Ethics, eds. J. Ryberg and T. Tännsjö. Dordrecht: Kluwer Academic Publishers. Simon, H. 1955. “A Behavioral Model of Rational Choice,” Quarterly Journal of Economics 69: 99–​118. Slote, M. 1989. Beyond Optimizing:  A Study of Rational Choice. Cambridge, MA:  Harvard University Press. Temkin, L. 1996. “A Continuum Argument for Intransitivity,” Philosophy & Public Affairs, 25: 175–​210. Temkin, L. 2012. Rethinking the Good. Oxford: Oxford University Press.

7 CONATIVISM ABOUT PERSONAL IDENTITY David Braddon-​Mitchell and Kristie Miller

7.1  Introduction Conativism is an umbrella term for a large range of views about the nature of persons. Broadly, it’s the view that in some sense or other, it’s up to us which things are the persons, either synchronically, or diachronically. Conativist views include views according to which the same-​person relation (henceforth SP-​relation) in some sense depends on desires, behaviours, choices and other conative states, as well as organised systems of such behaviours and choices in the form of conventions: so conventionalism about persons is a species of conativism according to our taxonomy. It’s the view that which things stand in what we will call the SP-​relation (same-​ person relation) to each other is settled, in part, by which conations obtain.1 So the contrast with conativism is non-​conativism. Non-​conativism is the view that which things stand in the SP-​relation to each other is settled entirely by matters of non-​ conative fact, and not in any part by the existence of conations. Conativism, then, is a thesis in first-​order metaphysics. As we will sometimes put our preferred version of the view, it’s the view that conations settle which relations in the world are or realise, the SP-​relations. The thought is that there are various relations that obtain between person-​phases; relations (such as similarity, or causal connectedness) that obtain independent of any conations. On our preferred version of conativism it is the obtaining of certain conations, which determines which of these relations, is, relative to a pair of person-​phases, the SP-​relation. That’s because the correct semantics for ‘SP-​relation’ is that the SP-​relation is the relation that obtains between two person-​phases when some relation (more on this shortly) holds between those person-​phases, and that relation’s holding is desired, or otherwise conatively relevant. Then different relations can realise the SP-​relation relative to some pair of person-​phases, on the assumption that different relations can be desired by (or otherwise be conatively relevant to), different person-​phases. For

130  David Braddon-Mitchell and Kristie Miller

what makes a relation the SP-​relation relative to some pair of person-​phases, is that the relation is conatively relevant to at least one of those phases. The sense, then, in which conations settle which person-​phases the SP-​relation obtains between, is not the sense in which conations settle which person-​phases various relations (like similarity or causal connectedness) obtain between:  those relations obtain independently of the conations. Rather, the conations settle whether any of those relations is the SP-​relation, if so which ones, and hence in that way settle between which person-​phases the SP-​relation obtains. Finally, it is a matter of conceptual analysis that a relation’s being conatively relevant to some person-​phase, settles that the relation is, relative to that person-​phase, the SP-​relation: it’s the fact that as a matter of conceptual analysis, the SP-​relation is the relation that obtains between two person-​phases when some relation holds between those person-​phases, and that relation’s holding is desired, or otherwise conatively relevant. For those hostile to talk of conceptual analyses, there are other ways to understand our view, and we turn to these in section 7.2.2 For now, however, one can think of us as making a claim about the right conceptual analysis of ‘SP-​relation’.3 This chapter focuses on diachronic versions of conativism:  that is, conativism about the persistence conditions of persons.4 It is most natural to think of this as conativism about personal identity. We, however, will talk about the same-​person relation in order to avoid prejudging the question of whether the relation that obtains between person-​phases (more on this shortly) of the same person is an identity relation or not. So we intend the SP-​relation to be a very metaphysically neutral notion. It’s what corresponds to same-​person judgements that people form. Consider this question: “That women you had coffee with this afternoon (t2). Is she the same person as Annie, whom I met last week (t1)?” If the answer is yes, then the SP-​relation holds between Annie at t1 and Annie at t2. You might think that it follows from this that the SP-​relation is a numerical identity relation. We make no assumption either that it is, or that it is not. It’s whatever kind of relation it is which (sometimes, when things go well) makes judgements of this kind true. And that, of course, depends on the right metaphysics of personal-​identity. Indeed, if you think that talk of the personal-​identity relation is metaphysically neutral in this way—​that is, if you think what we have just characterised is the relation of personal-​identity, then for you, talk of the SP-​relation just is talk of the personal-​identity relation.5 If the SP-​relation is a numerical identity relation, then the SP-​relation holds between a person at one time, and a person at another time, when the former is numerically identical with the latter. By contrast, if worm theoretic perdurantism is true, then the SP-​relation is the relation that holds between person-​stages that are parts of the same person. If the stage theory (i.e. exdurantism) is correct, then the SP-​relation holds when one person-​phase is a temporal person-​counterpart of  the other, and so on. Although ultimately we are drawn towards a view on which the SP-​relation is not an identity relation—​a view we will outline later in the chapter—​we think that conativism can be understood even within the framework of supposing the SP-​relation to be an identity relation. Hence, in what follows we will often talk of the SP-​relation obtaining between person-​phases p1 and p2. As

Conativism about personal identity  131

we use the term, ‘person-​phase’ is neutral between picking out a person at a time, and picking out a person-​stage—​where person stage is understood as a short-​lived entity that is qualitatively much like a person at a time (leaving open whether persons are composed of person-​stages or a person has such stages as temporal counterparts). Hence if we say that the SP-​relation obtains between person-​phases p1 and p2, this is metaphysically neutral between its being the case that p1 and p2 are numerically identical—​they are ‘phases’ of an enduring person—​or p1 and p2 being numerically distinct person-​stages that are either parts of the same perduring person (in the case of four-​dimensionalism) or bear same-​person temporal counterpart relations to one another (in the case of exdurantism).6 Conativists, then, agree that in some good sense desires, behaviours or conventions7—​ henceforth conations, determine between which phases the SP-​relation obtains. Views in this neighbourhood have been variously called desire-​first accounts, practice dependent accounts and conventionalist accounts. For instance, suppose Martha cares very much about her physical continuity. She has no self-​interested desires about a future in which she has a psychological continuer, but no physical continuer. Let us suppose she would die in a teletransporter in which her physical, but not psychological, continuity ends. The usual story about why Martha has the patterns of desires she does, is that there is a metaphysical fact that the SP-​relation depends on physical continuity, and Martha sees that this is so, and she dreads lack of continuity because it will result in death. Such a view is a version of non-​conativism: there is a metaphysical fact of the matter regarding which relation is the SP-​relation, a fact that obtains independently of any person-​phase’s conative states. Indeed, on such a view the appropriateness of person-​phases’ certain conative states depends on the presence of these metaphysical facts. Person-​phases ought to reason prudentially about the interests of just those person-​phases to which they are SP-​related. Non-​conative views disagree about which facts determine the SP-​relation (psychological, physical, organismic, soul, brute, etc.) but they agree that there are such facts independently of most conations. Conativists, by contrast, find it plausible that once we specify the various non-​ identity involving relations that obtain between person-​phases (similarity relations, causal connections, social relations, social conventions, and so on) and specify the various psychological properties (particularly the cognitive and conative states of person-​phases) then we have specified all that there is which can be relevant to settling which person-​phases bear the SP-​relation to which other phases. There are no other metaphysical facts that outstrip these properties and relations. So conativists hold that Martha would die in the transporter not because of a fact about the SP-​ relation that holds independent of her (and others’) conative attitudes, but because, for instance, her attitudes, or the attitudes of others, towards a future in which she has no physical continuer, are what make it the case that physical continuity is the SP-​relation for her, and thus that she would die in the teletransporter. Her desires do not depend on the facts about the SP-​relation, but rather, the facts about which relation is the SP-​relation depend (in part, or entirely) on her desires.

132  David Braddon-Mitchell and Kristie Miller

The most natural way to understand conativism, then, is as a kind of reductionism about the SP-​relation,8 with the added claim that (at least according to our version of the view) different relations (even within a world) can realise the SP-​relation. So while we talk about the SP-​relation, which relation that is, relative to any pair of person-​phases, can vary. Here, following Parfit (1984 p. 210) we take reductionism to be the view that the obtaining of the SP-​relation between person-​phases just consists in the holding of more particular facts, and that these facts can be described without presupposing facts about identity. Further, we assume that reductionism only really makes sense if one holds that the SP-​relation is not the relation of numerical identity, since the relation of numerical identity is surely irreducible (of course there are views which pitch themselves as reductionism about numerical identity, but a way to make those views coherent is as reductionism about whatever identity talk is about:  not, it turns out, identity). Assuming that is right, then reductionism can be paired with a view of the SP-​relation according to which it is a non-​identity relation that obtains between person-​phases, and where the reductive base for that relation is the combination of certain non-​identity involving relations that obtain between person-​ phases, and the conative states of those person-​phases. By contrast, we suppose that those who hold that the SP-​relation is a relation of numerical identity will typically be non-​reductionists. Non-​reductionists will not think that the obtaining of the SP-​relation reduces to the obtaining of certain relations between person-​phases, and the obtaining of certain conative states by those person-​phases, but they can still think that there is some connection—​perhaps a necessary bi-​conditional—​between the obtaining of the SP-​relation, with the obtaining of these other relations and conative states. On that view, the obtaining of the latter is effectively apodeictic evidence (if the biconditional is necessary) that the former obtains. In what follows we will talk of the presence of these conative states either settling which relation is the SP-​relation, or settling between which person-​phases the SP-​ relation obtains. Reductionists, such as we are, can straightforwardly read this as a claim about the reductive base of the SP-​relation. Non-​reductionists will need to read it, instead, as a claim about some connection, possibly necessary, between the SP-​relation and the conations (in which case the term ‘settling’ is perhaps inapt). Given this, we have a strong preference to construe conativism as a kind of reductionism, and this is the view we will have in mind henceforth. Nevertheless, we think that non-​reductionist versions of the view are available, which is why we continue to use the metaphysically neutral language of person-​phases. Conativism is often motivated by the thought that there is some sense in which what we believe, or what practices we engage in, or what conative states we have, play some important role in settling which things the kinds of things we care about—​persons—​are across time. Conativism is naturally paired with person essentialism—​the view that we are essentially persons. Technically, conativism only tells us which person-​phases the SP-​relation obtains between, leaving it open that you and I  are not essentially persons, and hence might survive some event that

Conativism about personal identity  133

nevertheless causes the person that we are, to cease to exist. But since conativists think that what we care about settles which things the SP-​relation obtains between, this would be the view that although the person each of us is, might cease to exist, we might continue to exist even though we do not care at all about our future continuer phases. We, however, are of the view that to talk about survival is to talk about something about which we care deeply. In fact, we think that whatever the SP-​relation consists in, it had better be what matters to us. It ought not be that when we spell out what it is for one person-​phase to bear the SP-​relation to some other person-​ phase, some question remains regarding whether the former ought care about the latter, or, indeed, some question remains about whether, if the latter exists, then the former survives. More generally, we, at least, are motivated by the thought that there ought to be a particularly close connection between our practical concerns and facts about the SP-​relation: it should not be that the latter come apart from the former.9 Let’s call this the practical concern desideratum.10 Indeed, as we will see, our analysis of the SP-​relation yields a version of conativism that meets the practical desideratum constraint:  I cannot fail to care about the SP-​relation, because the SP-​relation just is whatever (appropriate—​more on this later) relation it is that connects my current person-​phase with other person-​phases, such that I care about that relation (or care only about person-​phases connected to me by that relation) in a certain manner. We suppose, then, that the best version of conativism will meet the practical concern desideratum. Given this, in what follows, we will assume that you and I are essentially persons, and hence that you and I survive only so long as there is some future person-​phase that is SP-​related to our current person-​phase. We thus can be seen as disagreeing with Parfit’s (1984), view, according to which the things that we care about in survival can come apart from numerical identity. We think there are three ways things might be metaphysically organised, and exactly what the conativist should say about this will depend on how each of them goes. The view we find most attractive does not have trans-​temporal numerical identity relations. So the SP-​relation is not an identity relation, but rather, a relation that holds between the temporal stages of a self-​identical person, or, perhaps is a temporal counterpart relation. On this way things might turn out, the conativist disagrees with Parfit, insofar as Parfit seems to assume that there are relations of numerical identity that obtain across time. For in this version the conativist denies that there is any relation of numerical identity that obtains across time: objects, or at least, complex ones, do not persist by enduring. So there is no trans-​temporal identity relation that can come apart from what we care about, and hence cannot come apart from the SP-​relation. The second view a conativist might have is that there is numerical identity of persons across time, but that it can’t come apart from the SP-​relation.This would either require the SP-​relation to reduce to the identity relation, or the identity relation to reduce to the SP-​relation, or that the SP-​relation is somehow apodictic evidence

134  David Braddon-Mitchell and Kristie Miller

for the holding of fundamental identity facts. None of these options appeals to us; but the obvious problems, such as fission and so forth, designed to show that identity must come apart from what we care about, have been well enough dealt with in the literature for us to think this is a live option for the conativist who is a friend of numerical identity across time.11 On this way things might turn out, the conativist also disagrees with Parfit because although there are relations of trans-​temporal numerical identity, these cannot come apart from what we care about (or from the SP-​relation). Finally, of course, Parfit’s metaphysics might be right: there is numerical identity across time, and the facts about this do come apart from the conative facts (of which Parfit’s R relation is but one instance). We don’t think that Parfit’s positive arguments stand, and thus that there is any reason to suppose this is so. But there is room in logical space for the view there are such things as persons that endure across time, and that their persistence conditions come apart from anything that we care about. How one would make that discovery is a mystery to us, but suppose it to be made. Most endurantist conativists will resist, and say that conativism tells us what persons are and what their persistence conditions are, and thus that we should retreat to the second view. But perhaps a conativist might give in and buy this view, in which case they would indeed become close to Parfit. They would view Parfit’s R relation as the prototype of the SP-​relation, and say that whatever these dull metaphysical facts about the numerical identity of persons over time are, our actual practices around identity, survival and decision, should go with the SP-​relation, not the identity relation. Now let’s move on to a second, related, motivation for conativism. This is the intuition that there are at least some faultless disputes over survival, between parties. Consider a case in which Martha takes herself to be able to survive teletransportation, and Phyllida, a somatophile12 takes herself to be unable to survive teletransportation. We can imagine cases like this in which Martha and Phyllida agree about all the empirical facts: they agree about what teletransportation consists in, and they agree about the fundamental relations that would obtain between each of their person-​ phases, and the person-​phases that exist post-​teletransportation; they agree that there are no mysterious further facts (such as the existence of souls) that could make it the case that Martha survives and Phyllida does not; they agree about what ‘survival’ means in English, and so on.Yet their dispute remains persistent and intractable. Conativists are motivated by the thought that in such cases it’s hard to see what Martha and Phyllida could be disagreeing about. A  natural thought is that what differs between Martha and Phyllida are not the metaphysical facts, but the conative facts. It is the difference between their most basic desires, about their attitudes to various different kinds of future person-​phases, which explains why there is a persistent and intractable dispute. Moreover, it seems natural to think that as long as different conative states are permissible, we can, and ought, accommodate its being the case that both Martha and Phyllida are right about their own survival,

Conativism about personal identity  135

or lack thereof, after tele-​transportation.13 This is to accommodate there being (at least some) faultless disagreements about the SP-​relation. Since we think it plausible that there are such disagreements, on our view the best version of conativism will accommodate them. We call this the faultless disagreement desideratum. We will have more to say about how conativism in general, and our preferred version in particular, accommodates this desideratum in the remainder of the chapter. Versions of conativism differ in the kind of conations they suppose settle which things the SP-​relation obtains between, and the manner in which they do the settling. The range of conations is very broad, and can include attitudes about who one is14 and what one will survive; first-​person I thoughts; various other conative states such as anticipation; practices of prudentially caring about other person-​phases; social practices of imputing responsibility and ownership; and so on. As a consequence, different conativisms vary as to whether they are public or private: whether SP-​relations depend on the conations of individual persons, or person-​phases, or, as in conventionalism, emerge at the level of communities of person-​phases. In what follows we will first provide a general sense of what we mean when we say that conations settle which person-​phases the SP-​relation obtains between, (section 7.2), before we briefly outline a few different views about the nature of those conations (section 7.3). What is notable about these different conations is the extent to which they can be expected to come apart from one another. We then outline our preferred version of conativism—​direct private conativism (section 7.4)—​along the way explaining how the view does justice to various of our intuitions about personal identity: in particular, how it accommodates the two desiderata just outlined. We then consider a number of objections that are specific to this version of conativism (section 7.5) before turning to a more general objection to conativism in any form (section 7.6). According to this more general worry, conativism raises the spectre either of there being lots of different notions of ‘person’, each associated with some different conation or set of conations—​person pluralism—​or of having to tell some story about which set of conations is the right one to determine the SP-​relation. In section 7.6 we introduce two rather different kinds of practical concerns: self-​regarding and other-​regarding and argue that in some sense our view is a kind of pluralism about persons. There, we not only respond to this general worry for conativism, but argue that our version of conativism is better placed to respond to the worry. Finally, in section 7.7 we return to consider a number of further issues about personal-​identity discourse that are raised by conativism.

7.2  How do conations settle the SP-​relation? So far we have said that conativists hold that conations settle which person-​phases the SP-​relation obtains between and that, on our view, they do this by settling which relation, or relations, that obtain between person-​phases, realise the SP-​relation.

136  David Braddon-Mitchell and Kristie Miller

In particular, we think that something like the following is the right conceptual analysis of the SP-​relation. Conceptual Analysis: SP relation=df the relation that holds between two person-​phases when some CPC holds between those person-​phases, and that CPC’s holding is desired, or otherwise conatively relevant. Here, following Braddon-​Mitchell and Miller (forthcoming) we distinguish candidate properties of conation (CPCs) from the conative attitudes towards those properties. The candidate properties of conation will often be straightforward non-​conative naturalistic properties and relations like physical continuity, psychological continuity, similarity, and so on. Roughly speaking, one can think of CPCs as the various candidates that philosophers have introduced for being the SP-​relation.The conative attitudes (more on these shortly) then apply to these candidate properties. So on our view, the SP-​relations are those candidate properties that are cared about in the right way. We will have more to say about these candidate properties in section 7.5.1, where we will argue that not just any old relation can be a CPC. Instead, we will suggest that something can be a CPC only if its being desired (in the relevant ways) is compatible with (but not entailed by) ideal rationality and relevant knowledge. So, for instance, some sort of weak similarity relation won’t count as a CPC if, after reflection, and in light of relevant facts, a person-​phase could not have the right kind of conations towards that CPC. For now, though, it suffices to just think of the CPCs as the kinds of candidates that philosophers have proposed, for being the SP-​relation. Before we talk more about this analysis, there are a few things to say here. First, our contention is clearly at odds with those, like Sider (2001a) who think that meaning is determined by eligibility plus use, and that in the case of the SP-​relation, meaning and use do not determine which relation is the SP-​relation. In particular, Sider argues that there is no fact of the matter regarding whether the relation of physical continuity or psychological continuity is the SP-​relation, because they are both equally eligible candidates, and use does not determine which relation is the SP-​relation. So, on the one hand, we disagree with Sider that there is no conceptual analysis of the SP-​relation that settles whether two person-​phases are SP-​related to one another. On the other hand, we agree that eligibility and use do not fix that one particular relation—​psychological continuity, physical continuity, or something else—​is the unique SP-​relation. We think, that is, that on a case-​by-​case basis there is a fact of the matter as to which relation (which CPC) is the SP-​relation, we just think that this fact varies between person-​phases, and it varies because different person-​phases bear different conations towards different CPCs. Second, we cannot, in this chapter, defend a full account of what conceptual analysis consists in, let alone defend the value of such analyses. We can say that we have in mind something very much like what is often known as “The Canberra Plan” (see Braddon-​Mitchell and Nola 2008). Moreover, we cannot fully defend that the above is the right conceptual analysis of the SP-​relation.What we can do is

Conativism about personal identity  137

show that the view that we end up with is plausible; in particular, we can show that it meets the two desiderata with which we began this chapter, and so, in a roundabout way, we can defend its plausibility. To do that, we first introduce a couple of competitor conativist accounts:  accounts that differ in their view about what the conations are which do the settling. In section 7.6, when we consider some general objections to conativism, we will argue that our preferred version—​direct conativism—​is superior to those views. Along the way we will argue that direct conativism is an attractive view when measured against the desiderata that we set out earlier. Third, we noted at the beginning of this chapter that for those who are hostile to conceptual analysis, there is another way of thinking about our account of the SP-​relation. One need not think that the definition of the SP-​relation that we gave above is an analysis of a concept. Instead, one might hold that we are simply giving a sort of subject naturalistic (in the manner of Price 2008) account of the phenomenon in question: an account that explains why, in the domain of discourse and practices pertaining to personal identity, we talk the way we do and have the practices we have, and so on.Viewed in that light, there is no need to try to articulate some internal concept or representation of the same-​person relation, and then to map that thing onto something in the world. Instead, once we have told the full story of why we have the practices (linguistic and otherwise) we do, one has done all that there is to be done. Hence, thought of in these terms, our view is that in order to tell the full story about why we have the practices we do, one need only appeal to the CPCs that obtain between person-​phases, and the conative states of those person-​phases: nothing more need be said. Then the account we offer of the SP-​relation is perhaps best thought of not as a definition, but rather, as a short hand description of the explanation we provide. We prefer the former way of conceiving of our project in terms of an analysis of a concept, but we are happy for others to think of it in terms of this latter methodology.

7.3  Kinds of conations In what follows we distinguish direct from indirect conations. Direct conations are first-​person conative states such as anticipation and regret, attitudes of prudential care, and so on. These are to be contrasted with indirect conative states that are structures like conventions or practices built up out of direct conations. Conativists disagree both about whether the conations that settle the SP-​relation are direct or indirect and, further, about what sort of direct, or indirect, conations these are. Those we will call paracognitive theorists think that the conations are first-​person I-​thoughts and related states that appear to be cognitive, not conative, states:  for instance, beliefs about what one will and will not survive.15 Those we call direct conative theorists think that the conations are first-​person conative states such as anticipation and regret. The contrast is with indirect conative theorists, who rely on structures like conventions or practices built up out of conations. Among the indirect theorists

138  David Braddon-Mitchell and Kristie Miller

are various social theorists who think that the conations are social conventions of, for instance, imputing diachronic ownership and responsibility.16 Also among them are what we will call practice-​based theorists who think that the conations are some mixture of conative states, first-​person I-​thoughts and apparent beliefs, and social practices.17 Some of these views about the nature of the conations entail that conations are public, and some entail that they are private. We will say that conations are private just in case the manner in which they obtain constitutively depends only on the individual person-​phase. It is this sense in which private conations are ‘up to’ individual person-​phases. Of course, it does not follow from this that individual person-​phases can choose, at will, which conations they instantiate. Conation in general, like desire in particular, is not voluntaristic. This is important to bear in mind, particularly in a context in which views of this kind have been labelled (including by us) as conventionalist. Many, including, for instance, Schechtman (2014), go to some lengths to distance their views from conventionalism, on the grounds that which things are persisting person is not, on their view, just some matter of choice.18 As we will see shortly, however, we characterise Schechtman’s view as a version of practice-​based conativism. As we have been at pains to stress elsewhere19 while conativists think that in some sense which relation is the SP-​relation is ‘up to us’, since it is settled by our conations, or by conventions that depend on those conations, this is not to say that these conative states are voluntaristic, or easily changed, much less changed on a mere whim. Conventions, in our sense, include a whole range of social and inter-​ personal practices which, given our psychologies and evolutionary history, may be such that they could not easily have been different; or at least, could not easily have been significantly different. This is in contrast to more paradigmatic cases of conventions—​such as which side of the road to drive on—​which could easily have been decided differently. Notice that public things like conventions might play a critical role in determining private conations:  it might be, for example, that social attitudes or conventions causally determine what an individual person-​phase’s conative states are like.This is consistent with the conations constitutively depending only on how the person-​phase is. So on our view this would still be a private conativism, since the public properties do not directly determine SP-​relations, they merely cause the private conations that do. Continuing on, a conative theory is public just in case it invokes conations that do not constitutively depend on any individual person-​ phases’ conative states, though they may depend on the totality of conative states of such person-​phases.20 Conations in this sense are public insofar as the alteration of the mental-​states or behaviour of a single person-​phase makes no difference (or no detectable difference) to the conations in question. Public conations of this kind are often best thought of as public conventions, and henceforth, this is what we will have in mind when we speak of public conventions. As we just noted, these conventions can run from the more paradigmatically ‘conventional’ end of the spectrum, including legal conventions of holding persons responsible, and ownership of property and so on,

Conativism about personal identity  139

through to social conventions that seem much less paradigmatically ‘conventional’ such as naming individuals, caring for them in certain ways, treating them in certain ways, within certain structures, and so on. Finally one might have a mixed conative theory that involves both public conventions and private conations, and some trade-​ offs between these. Practice-​based conativism is plausibly a kind of mixed conativism. On that view, what makes it the case that some set of person-​phases is united by the SP-​relation is determined by which candidate property of conation is most eligible, and eligibility, in turn, is determined by best fit with person-​directed practices. These person-​directed practices include both public and private conations: social practices of tracking persons over time, holding person-​phases responsible for the actions of other person-​phases, compensating person-​phases for past events, having rules for the ownership of property over time and so forth. It also includes practices of self-​concern such as anticipating the future, having prudential concern for future person-​phases regretting past actions, and owning and imputing the actions of some past person-​phase as one’s own.21 In what follows we will talk of a person-​phase organising her (relevant) conative states around some CPC. Very roughly, a person-​phase organises its conations around a particular CPC just in case that person-​phase has the relevant conative states only towards person-​phases that are connected to it via that CPC. Since person-​directed practices can be organised around different candidate properties of conation in different societies, there can be faultless disagreements between person-​phases regarding what each can survive. Two person-​phases, p and p*, can disagree about what events they will survive, and each can be right, so long as each organises its person-​directed practices about a different CPC. If one organises its practices around, for instance, the psychological continuity relation, and the other around the physical continuity relation, then the former will survive events such as teletransportation, which the latter will not. Moreover, practice-​based conativism appears to be able to accommodate the practical concern desideratum. After all, the various person-​directed practices include the very sorts of things that person-​phases care about. If these determine which CPC is the SP-​relation, then for any person-​phase, p, that person-​phase will care about phases that are SP-​related to it. We will return to consider this issue more fully in section 7.6. Both paracognitive conativism and direct conativism are versions of private conativism. Despite having ‘cognitive’ in its name, we take the former to be a form of conativism, as we will now explain. According to this view it is something close to the content of I-​thoughts that determines, for each person-​phase, which stages are its continuers:  namely, those stages are continuers, which make most sense of the I-​thoughts of that stage.22 The idea is that there are various candidate properties of conation, and which is the most eligible is determined by best fit with the I-​thoughts and beliefs of the relevant person-​phases. We take this to be a version of conativism because ‘fit with I-​thought’ means something like: the thing that goes along with the person-​phase’s imaginings and conative dispositions. It cannot be that the content of the I-​thoughts literally determines the content of the

140  David Braddon-Mitchell and Kristie Miller

I-​thoughts; that would be circular. Rather it seems that the conative dispositions determine the content of the I-​thoughts, which, in turn settle which person-​phases bear the SP-​relation to one another. Understood in this way, however, this view looks a lot like what we call direct conativism: the view that direct conative states of the relevant person-​phases settles to which other person-​phases those stages bear the SP-​relation.23 On that view, it is in virtue of some person-​phase, p, organising its conative states around the particular candidate property of conation that it does, that when p uses ‘I’, that term refers only to person-​phases that are connected to p in respect of that candidate property of conation. Henceforth, then, we won’t distinguish this kind of paracognitive version of conativism from direct conativism.

7.4  Direct private conativism We endorse a version of direct conativism. Direct conativism is the view that it is direct, private, conations of some person-​phase, p, that settle which CPC is the SP-​ relation relative to p. There is room for disagreement, here, about which private conations matter. But the direct conative attitudes that we take to be relevant to determining the SP-​relation are of two kinds: desires for extension of oneself where these extensions bear certain sorts of similarities to the current person-​phase—​ these are forward looking conative attitudes—​and desires to have been extended from (i.e. to have as an origin) person-​phases that bear certain sorts of similarities to the current person-​phase—​these are backwards looking conative attitudes. It is these attitudes that, according to our preferred version of direct conativism, settle, relative to some person-​phase, which CPC is the SP-​relation. Further, we think there are constraints on which relations can count as a CPC. We will have more to say about this later (section 7.5.1) but for now, we’ll just say that something can be a CPC only if its being desired (in the relevant ways) is compatible with (but not entailed by) ideal rationality and relevant knowledge. So, for instance, some sort of weak similarity relation won’t count as a CPC if, after reflection, and in light of relevant facts, a person-​phase could not have the right kind of conations towards that CPC. Since a person-​phase might desire to be extended by the relation of psychological continuity, and desire to have come from person-​phases connected by psychological continuity, the relation of psychological continuity is a CPC (so long as its being thus desired is compatible with ideal rationality and relevant knowledge). If a person-​phase does desire to be extended by the relation of psychological continuity, and desires to have come from person-​phases connected by psychological continuity, we will say that that person-​phase desires psychological continuity, and mutatis mutandis for other CPCs. Why do person-​phases desire certain CPCs? Here is one part of the story we find plausible: there are primitive conative attitudes of anticipation and of origin: primitive attitudes of, on the one hand, anticipating the experiences of certain future person-​phases and, on the other, of taking some past person-​phases to be one’s origin. Here, again, there is something of an inversion of the order of explanation. Rather than its being the case that person-​phases ought

Conativism about personal identity  141

to have attitudes of anticipation and origin towards other person-​phases because those person-​phases bear some particular relation (such as psychological continuity) to the person-​phase in question, instead, it is the attitudes that are primitive. Some CPC, such as psychological continuity, is one that a person-​phase cares about because, for that phase, believing that psychological continuity holds between itself and, say, a future person-​phase, triggers the primitive conation we call, for example, ‘anticipation’. Anticipation in this example is primitive: it’s just a way of thinking, feeling and imagining. So it’s not analysable in any way, most especially not in terms of identity or the SP-​relation. Rather, as should now be clear, the SP-​relation is defined in terms of it, and other conations, and how they track CPCs. Further, it is our view—​but not one we argue for here24—​that the SP-​relation is best thought of as coming in continuous degrees. After all, if the SP-​relation is the relation that holds between two person-​phases when at least one CPC holds between those person-​phases, and its holding is desired (by at least one of those phases), then the SP-​relation can be a matter of degree because it can be a matter of degree whether some CPC is desired and by its being a matter of degree whether some CPC holds between said person-​phases. In either case we might expect the SP-​relation itself to come in degrees.25 To be clear, one could impose some kind of threshold in both cases. This is the approach of most reductionists about the SP-​relation. For instance, Parfit (1984 pp. 206–​207) holds that the SP-​relation only obtains where person-​phases are psychologically continuous with one another, where psychological continuity consists in there being overlapping chains of strong connectedness. In turn, strong connectedness obtains between two person-​phases just in case at least half the number of psychological connections obtains between those phases, which are present in the life of nearly all adult humans, nearly every day. Regardless of the explicit details of such a proposal, one could hold that only when a CPC holds above a certain degree, and only when it is desired above a certain degree, does the SP-​relation obtain. But we see no good reason to impose such a threshold. Indeed, if one thinks, as direct conativists do, that the CPC in question matters for the obtaining of the SP-​relation because of the conative relations person-​phases bear to it, then there could be such thresholds only if person-​phases have fine-​g rained desires about such thresholds. That is, there could be such thresholds only if, for instance, a person-​phase cares about some person-​phase with which it bears psychological connections, only if 50 per cent of the usual psychological connections obtain, and otherwise, does not care at all about that person-​phase. Such patterns of care are possible but not, we think, actual, or at least, not common. So, on the assumption that person-​phases don’t care about particular thresholds in a strong way, it follows that the direct conativist account entails that the SP-​ relation comes in degrees. Importantly, if the SP-​relation is that CPC which is desired, then it is not simply that the patterns of practical concern determine which CPC is the SP-​relation: it is also that the patterns of practical concern cannot come apart from the facts about the SP-​relation. So if actual patterns of practical concern come in degrees—​which clearly they do—​then so does the SP-​relation. So

142  David Braddon-Mitchell and Kristie Miller

meeting the practical concern desideratum entails embracing the claim that the SP-​ relation comes in degrees, at least given plausible views about person-​phase’s actual conative states. Indeed, we think it an attractive feature of the resultant view that it accommodates the SP-​relation obtaining to different degrees between person-​ phases. It seems to us that this accommodates the sort of phenomena that we see in the real world, with person-​phases treating other phases (either of themselves, or someone else) as being the ‘same person’, and yet, at the same time, not entirely the same person (think about person-​phases that are very distant in the future and very different from your current phase, with entirely different values; think of some future person-​phase suffering dementia, or the person-​phase of a family member who has undergo serious brain injury). We think that the fact that the SP-​relation comes in degrees allows us to make sense of a range of behaviours—​both moral and prudential—​regarding the treatment of future person-​phases to which one is SP-​related (to some degree or other) as well as the treatment of person-​phases of other persons. Hence we think that this aspect of our view is pleasing, and is some reason to think that the analysis of the SP-​relation we offer, is correct. Another reason we think that our analysis of the SP-​relation is plausible, is that the resulting version of our view (direct private conativism) is one that can nicely accommodate the faultless disagreement desideratum. So long as person-​phases can bear the relevant conative states towards different CPCs, it follows that said person-​ phases can faultlessly disagree about which kinds of continuers they have, or could have, and can disagree about to what degree those person-​phases are SP-​related. Let’s start with Martha and Phyllida. We can explain why Martha is right to say that she will survive teletransportation, and Phyllida is right to say that she, Phyllida, will not, by noting that for Martha (or the current person-​phase thereof) the psychological continuity relation is the SP-​ relation—​because that is the relation around which she organises her conations—​ while for Phyllida, the physical continuity relation is the SP-​relation—​because that is the relation around which she organises her conations. In fact, on our view this is not really faultless disagreement at all, since there isn’t really any disagreement between Martha and Phyllida. But suppose that not only does Martha think that she will survive teletransportation, but she thinks that Phyllida will too, while Phyllida thinks not only that she will not survive teletransportation, but that neither will Martha. Now we do seem to have a case of disagreement. When Martha evaluates what Phyllida says about her, Martha’s, survival—​i.e. that Martha will not survive teletransportation—​from Martha’s own context of assessment—​that is, given Martha’s own conations—​she will say that what Phyllida says is false. Mutatis mutandis for Phyllida’s evaluation of what Martha says about her, Phyllida’s, survival. Here, our view can capture a sense in which Martha might feel inclined to argue with Phyllida, and to strongly suggest to her that she get into the teletransporter machine. Equally though, if Martha has a firm grip on the concept of the SP-​ relation, then she will recognise that given Phyllida’s own conations, what Phyllida says is, by Phyllida’s own lights, true. Moreover, since there is nothing more to settle the debate between them, than facts about the relations between certain

Conativism about personal identity  143

person-​phases, and the conations of those person-​phases, Martha will see that there is no fact of the matter as to whose conations, hers or Phyllida’s, are right. So she will recognise that in some good sense their disagreement is faultless: because, working with the very same concept of the SP-​relation, Phyllida makes claims which, given her own conations, are true. Indeed, on our view there can be faultless disagreement between person-​phases that are parts of the same person. In the starkest case it can be that from the perspective of one person-​phase, P, some other person-​phase, P*, bears the SP-​relation to P to degree 1, while from the perspective of P*, P does not bear the SP-​relation to P* (i.e. bears it to degree 0).26 For suppose P* is psychologically but not physically continuous with P. Suppose, further, that P desires psychological continuity in virtue of having certain primitive attitudes of anticipation and origin towards stages connected to P via psychological continuity. Suppose, also, that P* desires physical continuity in virtue of having certain primitive attitudes of anticipation and origin towards stages connected to P* via physical continuity. Since P* is connected to P by psychological continuity, from the perspective of P the SP-​relation holds between the two stages. But since P* is not physically continuous with P, from P*’s perspective it’s not the case that the SP-​relation holds between the two stages. In virtue of P and P* having the relevant conative attitudes towards different CPCs, it turns out that from the perspective of P, P* is not SP-​related to P, and from the perspective of P*, P is not SP-​related to P*. In order to accommodate this being the case, the SP-​relation must be non-​symmetrical. A question then arises as to how to think of persons, and their connection to the SP-​relation. Before we turn to this issue, unfortunately, there are some complications. The first is whether it is solely the conations of an individual person-​phase that determines the SP-​relations between that phase and other phases, or whether the conations of other person-​phases play a role. Consider the following version of the view: the conations of some phase, P, determine what the nearest successor phase, P*, is. But that ends their role. It’s then the conative states of P* which determine her successor phase, P**, and so on. So whether P is SP-​related to some other distant phase, P’, is a matter of whether there is a chain of phases between P and P’ in which the conative states of each phase determine that the ‘next’ phase is SP-​ related.This, of course, allows P to be SP-​related to some P’ even when, by P’s lights, P’ is not her to any degree at all. Consider the case where the relevant conative states of the intervening phases (between P and P’) has changed enough so that, for example, both P*, the phase just before P’, and P’ accept that teletransportation is SP-​preserving, whereas P does not. So, P would take herself to have died before P’ appears, despite the existence of the chain of phases each of which takes themself to have survived to the next stage. The view we favour is that the conative states of each person-​phase fix what the person is by their lights (meaning of course that there could be different persons, depending on from which phase we are evaluating —​cf Braddon-​Mitchell and West 2001). So we simply look to P’s conative states, and we see that P organises her

144  David Braddon-Mitchell and Kristie Miller

direct conations around the psychological continuity relation, and that settles that P is SP-​related only to phases with which P is psychologically continuous.That is true even if some of those phases, in turn, organise their conations differently—​around, say the physical continuity relation. Relative to those phases, the SP-​relation will be the physical continuity relation. Hence there will be overlapping, but distinct, persons: the person united by psychological continuity, and the person united by physical continuity. Note that this does not prohibit the view that the SP-​relation is determined by a chain of phases each with different views about the local SP-​relation, and thus different views about which phase is the successor phase. But on the view we favour this would only be true where, from the perspective of a person-​phase, it followed from the conative states of that person-​phase.That would be true if, for instance, some-​ person phase defers to the conative states of a successor person-​phase, allowing that successor’s conative states to determine still later successors, and so on.27 Of course being committed to this way of thinking of things commits us to there being no fact of the matter as to what the person is, except indexed to a person-​ phase and their conations. The exact metaphysics of this might be spelled out in some of the different ways we will shortly consider (many overlapping worms, centred worms etc.) but the crucial thing is that there is a set of SP-​relations fixed by each person-​phase. This does not, of course, imply that there are always, or even often, a bewildering variety of persons made from the same (or largely overlapping) set of person-​phases. If the conative attitudes do not change hugely, then (in the case of the worm view) all the worms may completely overlap, or all the centred worms may differ only in their centre. Let’s consider how to think about persons given what we have said so far. Elsewhere (Braddon-​Mitchell and Miller ms) we have considered a number of proposals, including the fairly natural proposal that we think of persons as person-​ stages who bear non-​symmetric temporal counterpart relations (which is what the SP-​relations would be on this precisification) to one another, in virtue of which claims about what the person did, or will do, are made true. Rather than pursuing that view here, though, we will focus on what we call the centred worm view. According to the centred worm view, persons are mereological fusions of person-​ stages combined with a perspective: those person-​stages that bear SP-​relations to one another, seen from the perspective of one of them. Hence persons are centred four-​dimensional worms. We need the apparatus of a centred worm in order to make sense of the different perspectives of the various person-​stages that are parts of the worm. After all, these person-​stages can disagree about which relation is the SP-​relation, and hence can disagree about what would and would not survive and about which other person-​stages are its continuers. The analogy is with centred worlds. A  centred world28 is a tool used in semantics where we distinguish how things are objectively, from how things seem from perspectives within a world. So objectively speaking, there are only whole worlds. Such a perspective leaves out the indexical facts: what the time is, where one is, who one is, what is nearby and so on. These perfect naturalistic features are only visible from a location within a

Conativism about personal identity  145

world. At that location a certain time is the time, the person-​phase is whoever is there located, the future is the part of the world that is a certain temporal direction from that location, and so on. So ‘centred worlds’ are introduced to make sense of this: they are ordered pairs of worlds, and locations within the world. We construct centred worms by analogy. Worms are just the fusion of a set of person-​stages. But this uncentred perspective can’t reveal who is who, in the case where the stages disagree about who is who. Centred worms bring back this perspective, as there are as many centred worms as there are ordered pairs of the worm and a person-​stage that is part of it. So just as on the centred worlds approach, there are many centres within a single world, each reflecting a different perspective of someone at that location in the world, so too, on the centred worlds approach there is a single worm, and many centres, each corresponding to the perspective of the person-​stage at that centre. Then just as we can assess the truth of a proposition at different world centres, so too we can assess the truth of a proposition at different worm centres. Hence it can come out as true when assessed at one worm-​centre, P, that some person-​stage, P* is SP-​related to P, and yet also be true, when assessed at worm-​centre P*, that P is not SP-​related to P*. Of course, propositions about the conative attitudes of each person-​stage, and propositions about the nature and degree to which CPC relations hold between person-​stages, are simply true or false simpliciter. It is only propositions about the obtaining, or not, of the SP-​relation between certain stages that can ever be evaluated as true or false at a centred worm. What, though, are persons on this view? There are really two options. On one view, for each person-​stage, P, there is a worm that is the fusion of all those person-​ phases to which P is connected via the SP-​relation, as it were, from P’s perspective. Relative to each such person-​phase, then, there is some person composed of just those person-​phases to which said person-​phase is connected by the SP-​relation. For any such worm, then, there are multiple persons: each distinct person being a centred worm. So there are many distinct centred worms, although there is only one worm. Whether this is ontologically parsimonious depends on your take on the ontology of centred worms. If you think centred worms are ontologically ‘innocent’29—​nothing costly added to the worm and its centres—​then you might think so. Otherwise not. Regardless, on this view there are many persons cohabiting where there is a single worm: one person for each centre of that worm. In addition, there will sometimes be overlapping worms, where from the perspective of one centred worm P* is a part of the worm in question, and from the perspective of P* it is not part of the worm in question. In such cases we don’t just have one worm with multiple centres, we have multiple overlapping worms with different centres. The alternative is to identify the person with a single worm (if you think there is a way to form a complete, maximal worm, the parts of which comprise the set of all the SP-​relevant stages by the lights of every relevant stage), and use the centred worm story as a formalism to describe the SP-​relations from the perspective of each person-​phase. On this view persons have a less direct relationship to the SP-​ relation. The person is something like a maximally connected set of person-​phases

146  David Braddon-Mitchell and Kristie Miller

connected by the SP-​relation in either direction. Then there is a single worm which includes both stages P and P*, since from P’s perspective P* is a continuer. On this view it is the SP-​relations that obtain between person-​phases, rather than the existence of the person, which most directly explain patterns of concern. After all, on this view P* is part of the same person as P, despite the fact that from P*’s perspective P* is not a continuer of P. The benefit of this view is economy. There is only one person, and the centred worms play only a formal role. The cost is that the relation between persons, and the relations that explain patterns of concern displayed by a person-​phase, is indirect. There is an important complication for this second view, which makes it less attractive by our lights. As we just saw with P and P*, it’s possible that perspectives from within a ‘worm’ disagree as to which stages should be included in the worm in any way. Thus from those perspectives, the worm which is the person, might, for example, be rather smaller than the uncentred worm. So the uncentred worm has to be understood as a fusion of person-​stages considered relevant by any of the stages. The centring process will then make different parts of that large worm the ‘person’ by the lights of that centre. But that still requires specifying which stages should be used in this construction process. There is no parallel with centred worlds, except in the case that you think that the criteria for being a world is up for grabs: let’s suppose it’s optional whether spatio-​temporally isolated regions can be part of the same world. In that case, it might turn out that if agents at centres disagree faultlessly about this, the world in one sense would have different extents at different centres, while there is an overall world comprised of all those regions regarded as bearing the same-​world relation to one another, by any agent.The same problem would arise then: you would need to decide which stages were relevant to world construction before constructing the overall world. But it is more plausible in the case of worlds that what constitutes a possible world is an objective metaphysical fact which does not depend on agent’s views, rather than a perspective dependent matter. So far, we have wondered how we ought determine who I—​the person—​am (i.e. which person-​phases are person-​phases of me). Is who I am determined entirely by my (i.e. this very person-​phase’s) conations, or is it partly determined by the conations of the phases to which I (this person-​phase) am SP-​related, for instance, some chain in which my conations determine my continuer phase, but that phase’s conations determine which phase is the next continuer, and so on. Our view is that each person-​phase’s conations entirely determine to which other person-​phases that person-​phase is SP-​related: the conations of the person-​phases to which that person-​phase is SP-​related do not matter, except for determining the existence of different persons comprised from the same person-​phases. We can now ask a similar sort of question, but instead of wondering who I am, we can wonder who I  would have been. Consider actual person-​phase p, and the modal counterpart of person-phase p, person-​phase c. Suppose that c’s conations are quite different from p’s, so that the CPC around which c organises its conations is different from the CPC around which p organises its practices. For instance, let’s

Conativism about personal identity  147

suppose that p organises its conations around the psychological continuity relation, while c organises it around the physical continuity relation. Now suppose that actually, one of p’s continuers suffers a terrible accident, and thereafter there are no phases that are psychologically continuous with p. Suppose, too, that the same accident befalls one of c’s continuers; but c survives, since there are physical continuers. Ought p to think that, had things gone differently, he would have survived the terrible accident? That is, ought p to think that had things gone differently, the physical continuity relation would have been the SP-​relation, and hence there would have been some post-​accident physical continuer? Or ought p to think that had things gone differently, he would have had false views about what he would survive, but, in fact, he would not have survived the accident? The latter would be the right attitude if it is p’s actual conations that settle the modal truths; the former is the right attitude if it is p’s counterfactual conations which settle the modal truths. To think that it is p’s actual conations that settle the modal truths is to endorse a rigid version of conativism; to think that it is the counterfactual conations that settle the modal truths, is to endorse a non-​rigid version of conativism. While we think there is scope for dispute here, we are tempted by the non-​r igid version: we are tempted to say that had things gone differently, conatively speaking, p would have been able to survive different events than p in fact p is able to survive. That brings us to the end of our articulation of the view. Shortly (section 7.6), we will have more to say about why we prefer direct private conativism to other versions of conativism, and, in particular, we will have more to say about the connection between the private conations central to our account, and the public conations that figure in other accounts. In particular, we will consider a very general objection to conativism: that since the various different conations can pull in different directions, that is, can be organised around different CPCs, any conativist view either ends up being objectionably pluralist about persons, or, alternatively, must tell some story about which set of conations are the ones that matter, something which, in turn, looks difficult to achieve. Before we turn to these more general worries, however, first we want to consider a couple of specific objections to direct private conativism.

7.5  Objections to direct private conativism 7.5.1  The objection from implausible consequences We noted above that we take there to be constraints on which relations can be CPCs. In particular, we said that something is a CPC only if its being desired (in the relevant ways) is compatible with (but not entailed by) ideal rationality and relevant knowledge. But why think that? Well notice that it might be that the primitive attitudes of anticipation and origin can be held towards just about anything (stages of tables, lemons, cockroaches etc.). To consider some examples, suppose there is a relation, R, that holds between a particular person-​phase we

148  David Braddon-Mitchell and Kristie Miller

will call Freddie, and some future dustbin stage—​D—​and its holding is desired (or otherwise conatively relevant) to Freddie in part in virtue of the presence of primitive attitudes of anticipation. In virtue of that, Freddie is SP-​related to D. But that is absurd. Or consider Felicity, the current person-​phase of Freddie’s girlfriend. Freddie cares a lot about Felicity’s welfare. There is some relation, R, that obtains between Freddie and a future person-​stage, Fr*, that is bodily and psychologically continuous with Freddie. There is also some relation, R*, that obtains between Freddie and a future person-​phase that is bodily and psychologically continuous with Felicity. Call that future stage Fe*. Suppose that for any goods, Freddie is indifferent between Fr* and Fe* having those goods. So it may seem as though there is a relation, namely R*, that holds between Freddie and Fe*, whose holding is desired (or otherwise conatively relevant) by Freddie in part in virtue of the presence of primitive attitudes of anticipation. In virtue of that Freddie bears the SP-​relation to Fe*. But that is, perhaps, absurd. We think the direct conativist ought to say that what matters for the obtaining of the SP-​relation is the obtaining of those primitive attitudes that it is possible to hold after ideal rational reflection, in the face of relevant knowledge. In particular, we think that something counts as a CPC only if its being desired is not incompatible with ideal rationality and relevant knowledge. Before we show how this proposal helps resolve these problems, it’s worth noting a couple of important clarifications. Our claim about CPCs is not the claim that actual person-​phases are capable of ideal reflection. Something counts as a CPC just in case there were things like us to engage in ideal reflection, then, in the face of relevant knowledge, we would have certain conations towards it. That only requires that there is some world in which we engage in ideal reflection, not that that world is the actual world. To see how this helps, consider the dustbin case. Let’s suppose that R is the relation that holds between the Freddie-​stage and the dustbin-​stage. We can imagine that Freddie desires to be extended by R in part in virtue of the presence of primitive attitudes of anticipation toward D. For perhaps Freddie can primitively anticipate being a future dustbin. But Freddie cannot ideally anticipate being a dustbin, because one can only ideally anticipate being something if there is some way that it is to be that thing. Perhaps Freddie can anticipate being a dustbin in some sense or other. But knowing all the facts about dustbins, and knowing all the facts about anticipation and consciousness, she could not anticipate being a dustbin. So, were Freddie ideally rational, and were she to know all the relevant facts—​facts about dustbins and consciousness—​she could not anticipate being D. So the R relation is not a CPC: thus it is false that there is an SP-​relation obtaining between Freddie and the dustbin-​stage. For, recall, the SP-​relation is the relation that holds between two person-​phases when some CPC holds between those person-​phases, and that CPC’s holding is desired, or otherwise conatively relevant. That brings us to the case of Freddie and Felicity. Plausibly, the relation R* that holds between Freddie and Fe* is a CPC. But does Freddie desire that relation in part in virtue of the presence of primitive attitudes of anticipation? No. While Freddie is indifferent between goods accruing to Fe* and to Fr*, that is no reason

Conativism about personal identity  149

to suppose that she is indifferent because she anticipates the experiences of both Fr* and Fe*. Plausibly, the reason Freddie wishes certain goods to accrue to Fe* is on the basis of entirely other-​regarding attitudes. We will have more to say about these attitudes in section 7.4. And of course, if Freddie does primitively anticipate the experiences of Fe* things are more complicated. For now, we need only note that we have no reason to think that Freddie bears the SP-​relation to Fe* just in virtue of being indifferent between the well being of Fr* and Fe*. But why care about what one might ideally be able to (for example) anticipate, and thus limit the list of CPCs? Indeed, one might worry that doing so threatens to undermine the extent to which direct conativism can meet either the faultless disagreement or the practical concern desideratum. With regard to the former, one might worry that genuinely faultless disagreement is not possible between ideally rational and relevantly knowledgeable person-​phases. For, one might think, under those conditions all person-​phases will organise their conations in the same manner. But we see no reason to suppose that this is so. If one is a rationalist about desire—​if one thinks that some fundamental desires are rationally mandated and others rationally impermissible—​then one will suppose that is so. We, however, are Humeans about fundamental desires:  while it can be more, or less, rational to have certain instrumental desires, one’s fundamental desires are neither rational nor irrational:  they just are what they are. If different person-​phases can have different fundamental desires—​as surely they can—​then even under conditions of ideal reflection and relevant knowledge, those person-​phases may well organise their conations around quite different CPCs. Hence this characterisation of CPCs is consistent with there being disagreement (and hence faultless disagreement) between person-​phases regarding which CPC is the SP-​relation. The rationality requirement will only eliminate different conations where those differences are due to either cognitive error, or instrumental desires that are not properly in the service of fundamental desire. The second worry is that this will undermine direct conativism’s ability to meet the practical concern desideratum. For some actual person-​phase might not actually desire what they would desire in conditions of ideal reflection and relevant knowledge. That, after all, is precisely the position in which Freddie finds herself with regard to D. These are cases in which what some person-​phase in fact cares about—​it’s actual particular practical concerns—​can come apart from facts about survival. Freddie’s practical concerns include concerns about D, but Freddie is not, on our view, SP-​related to D. So we are committed to saying that it’s only one’s idealised practical concerns that cannot come apart from survival. For while our account constrains which relations can be CPCs, by appealing to rationality and knowledge, we could instead have said that it is a person-​phase’s ideal conations—​ their conations after ideal reflection, and in the face of relevant knowledge—​which settle which relation is the SP-​relation. Of course, which conations we would have, after reflection, is in part determined by the conations we in fact have. So our actual conations play a role in setting which CPC is the SP-​relation: it’s just that the way they settle this goes via some idealisations. But that, we think, is as it should be. At a

150  David Braddon-Mitchell and Kristie Miller

more abstract level, our practical concern is not for momentary whims, or for things that we would not care about if we were a little better informed. It’s only a weak requirement of practical rationality to say that our practical concerns are with what we would desire when not utterly misinformed or otherwise confused. That brings us to a second specific objection to direct conativism.

7.5.2  The objection from empty concepts Here is what we take to be another important objection. If some set of person-​ phases bear the SP-​relation to Freddie because of Freddie’s conative states, then what sense can we make of Freddie trying to deliberate over what desires she should have with respect to some set of person-​phases? A hard line conativist would deny the coherence of any such deliberation. Freddie’s desires are all there are to fix the matter, and if, for instance, she has no such desires, then there is simply no fact of the matter which set of person-​phases bear the SP-​relation to her. If she does have such desires, then the matter is fixed: we know which person-​phases bear the SP-​relation to her, and no question can arise as to whether she ought to care about those stages, rather than some other stages. But we think this is too quick. Surely Freddie could think in this way: I care about survival. I also care about certain future person-​phases in the way that I should care about them if they are me. But am I right to care in this way? Perhaps for many people, this thought would be epiphenomenal—​it would have no effect on the desires that they would have. But let’s suppose that Freddie is especially sensitive to the teachings of philosophers, and her desires move with whatever she is most persuasively taught in philosophy class is the best theory of personal-​identity.When the lecturer is especially eloquent in favour of physical continuity, she comes to care about future physical continuers. When the lecturer is especially enthusiastic about psychological continuity, she is prepared again to sign up for teletransportation. By itself the example is no objection to an account of the nature of the SP-​ relation, which says the relation depends on relevant desires. It shows just that what those desires are might depend on many things—​including metaphysical views (false or otherwise) about the nature of the SP-​relation. And in turn, the relevance of those metaphysical views is itself determined by conations: it’s a conative fact that someone cares about best metaphysics. But it does, perhaps, show that there is something wrong about direct conativism as a straightforward account of the concept of personal identity, at least as possessed by Freddie. Earlier in the chapter, we suggested that although there are other ways of framing the account, at least for expository purposes, our preferred way of framing it is as a conceptual analysis. But this example suggests another way of thinking about conativism, in which understanding the concept of the SP-​relation plays a role in determining that conativism is the right account, but the role is of there being a best deserver (which is not conativism) which gets revised to conativism in the light of best philosophy.

Conativism about personal identity  151

So perhaps Freddie’s concept is not a concept of the relation that holds between two person-​phases when some CPC holds between those person-​phases, and that CPC’s holding is desired, or otherwise conatively relevant. For we know that she is prepared to countenance the possibility that she does not bear the SP-​relation to the very person-​phases she does care about in the required ways. So what might her concept be? She might have a concept individuated at least in part by the appropriate response to tokening that concept: the concept of ‘same person’ is the concept of the relation that holds between person-​phases when those stages should be related by conative relations of a certain kind. When should Freddie have such conative relations? This is something about which the concept is too uninformative, and this, the thought goes, explains the lack of convergence in the literature. If there were deep metaphysical facts about personal identity, then presumably Freddie should have those conations when those facts obtain. It is to the extent that Freddie believes this that she is tempted to defer to best philosophy on the matter of what her desires should be. But as we noted at the very beginning of this chapter, we believe that there are no such facts—​just the desires with respect to other person-​phases whose origin, in best Humean style, is varied according to personal and cultural difference, and not subject, within the limits we discussed, to rational scrutiny. The concept Freddie has when she is in her introspective, theoretical mode (and, we suspect, many people have) is clear about how you should behave and desire if you grasp that the SP-​relation holds, but is silent on when it holds. It is in this sense an ‘empty’ concept: empty of decisive guidance about its extension, though not empty about what one should do when one judges that it obtains. If this is Freddie’s concept, then our proposal has to be understood as revisionary in a certain way. On discovering that there is nothing in the world, or in the concept, which normatively determines when the relation obtains, but recognising the importance of organising her life around such a concept (which makes elimination hard), Freddie revises30 to the view that the relation holds whenever the relevant desires do in fact obtain (with only the vestigial rational constraints of the kinds we have discussed in section 7.5.1). Hence Freddie revises to conativism.

7.6  Different conations and the threat of pluralism One very general objection to conativism proceeds as follows. Different conativists identify different conations as being important in settling which person-​phases the SP-​relation obtains between. Indeed, some conativist proposals identify a set of many kinds of conations, including both public and private, as jointly settling this matter. Yet it seems as though these various conations can ‘come apart’ in the following sense: different conative states can be organised around different CPCs.31 Different public conations—​ conventions—​ can be organised around different CPCs:  legal practices regarding who is responsible for what, about who gets to own what, and so on, might be organised around a different CPC than, say, certain

152  David Braddon-Mitchell and Kristie Miller

other social and inter-​personal conventions regarding how we track people over time, how we treat people over time, and so on. For instance, various other-​regarding practices—​the practices we enact with regard to person-​phases that we don’t take to be SP-​relation to ourselves (such as our family, children, friends, parents, and so on) taking care of other individuals, treating some person-​phase as though it is a continuer of some earlier phase, and so on, might come apart from other social and legal conventions. Legal conventions might mandate that certain person-​phases are not responsible for certain behaviours (children, those with limited cognitive capacities, those suffering mental illness or dementia etc.) of earlier person-​phases, and yet other social conventions and other-​regarding practices might involve treating those same person-​phases as though they are continuers of earlier phases. Relative to some sets of practices infants, and those with limited cognitive capacities, might be treated as phases of a persisting person, and relative to other practices, they might not, suggesting that different sets of practices (and hence conations) are organised around quite different relations. Indeed, other-​ regarding practices might come apart from self-​regarding practices—​private conations that involve first-​personal conations such as anticipation and regret (and so on) and distinctively prudential attitudes. It is easy to imagine that some person-​phase, p, organises its self-​regarding practices around the psychological continuity relation, and does not anticipate the experiences of any person-​phase with which it is not psychologically continuous. Hence p thinks that he has no continuer phases that are in a persistent vegetative state; instead, p supposes that if such an event comes to pass, then that event is the cessation of there being any continuers of p (p does not survive). Yet p’s family (and indeed other social institutions) might organise their other-​directed practices in such a way that they treat the vegetative phases as though they are continuers of p. Thus a whole range of other-​regarding practices with regard to p, and p’s continuers, might be organised around a relation other than psychological continuity. If those conations are the right ones, then they settle that p does have continuers that are in a vegetative state. Shoemaker (2007) takes away from this the idea that rather than attempting to locate some unique relation that is the personal identity relation, instead, the only reasonable project is to articulate the various different ways in which our practical concerns are organised around various relations in the world. For accepting, as we do, that there are no further metaphysical facts that outstrip the various conative facts and the facts about which (non-​SP-​involving) relations obtain between person-​phases, there is no further fact to which to appeal, to determine which of these relations really is the personal identity relation. A  slightly different way forward (but perhaps largely bookkeeping) would be to embrace person pluralism, a view countenanced (but rejected) by Schechtman (2014; ­chapter 1) according to which there are different kinds of persons—​a legal person, a moral person, a familial person, and so on—​each corresponding to a set of person-​phases united by some relation around which the legal, moral, familial, and so on, conations are organised.

Conativism about personal identity  153

Schechtman rejects pluralism on the grounds that there really is just one person: we don’t act as though there is a whole bunch of largely overlapping persons; we don’t talk of moral persons, and familial persons, and legal persons, and social persons. We act as though there is just one locus, the person. Schechtman’s response to the worry is to suppose that persons—​who for her, are things that engage in person lives—​are things that under normal circumstances develop psychological capacities and features and engage in certain kinds of social interactions—​ that is, they are things typically subject to the self-​regarding and other-​regarding conations already mentioned. But this allows that there can be cases in which, for instance, the psychological capacities are missing: so long as the life in question is a person-​life, we have a person, and we have a person-​life if there are sufficiently rich other-​regarding conations. So it is possible for a persisting person to have none of these characteristic psychological capacities, and that is so because that person exists within a set of social structures and conventions, and practices, all of which ultimately depend on conations, some of which any particular individual might lack. On this view, then, there is a unique relation that realises the SP-​relation, and it is determined by the totality of conations, including, quite crucially, social institutions and structures and other-​regarding practices.32 Our concern with this approach is that it results in a gap between certain kinds of practical concern, and survival. Of course, there is no gap between the totality of conations, and survival, since the former determines the latter. But let’s revisit person-​phase, p, who organises his self-​regarding practices around the psychological continuity relation. P does not care about any future phases that are in a vegetative state: as far as p is concerned, those are not his continuer phases. Moreover, p wishes that were a cataclysmic event ever to befall his body, that body would not be kept alive by machine; after all, as far as p is concerned, he failed to survive that event. In the event he does not want his body kept alive; he wants his organs donated so that some good can come of the tragedy. According to Schechtman (2014 pp. 147–​148), however, when the cataclysmic event happens, p does in fact survive, on the assumption that the relevant social and other-​reading practices are as we just described them; his body is kept alive; his family visits and reads to him; they treat the resulting phases as though they were continuer phases of p, and so on. P’s conations do not matter: they play effectively no role in determining under what circumstances p survives. Of course, one might be attracted to such a view: but we find it unpalatable to think that p’s practical concerns—​including p’s idealised practical concerns—​entirely come apart from the facts about p’s survival. Quite generally, conativist views face a dilemma: they can either concede that pluralism about persons is right, or tell some story about why these, rather than those, conations, settle the SP-​relation. But if conations are all there are—​if there are no further metaphysical facts that settle the SP-​relation—​then it’s hard to see what account one could give, of why one set of conations are the right ones, and others are not. Even if one could provide some reason to prefer one set of conations to another, the resulting view would run the risk of falling foul of the practical concern constraint. If different practical concerns track different relations, and if all of

154  David Braddon-Mitchell and Kristie Miller

these practical concerns matter to us, then insofar as just one of those relations is determined to be to the SP-​relation, that SP-​relation will come apart from at least some of those practical concerns. Conativisms that appeal to a range of different conations, which can come apart from one another, face a related problem. Mixed conative views, which appeal to both private and public conations, are a case in point.These views face the problem that the very things that they take to settle, of which CPC is the SP-​relation, might pull in different directions when it comes to doing the settling. It might, then, be indeterminate, relative to certain person-​phases, which relation is the SP-​relation. Alternatively, such views might take something closer to Schechtman’s route, and in some sense privilege certain of these conations. One might say that all those conative states are necessary preconditions for there being persons, but that only some of them settle of which CPC is the SP-​relation.33 As we already noted, however, that returns us to the issue that at least some practical concerns will come apart from the SP-​relation. Here is what we want to say. We think that something like pluralism is the right response to this dilemma. But how is that consistent with our earlier defence of direct conativism? Well, one can think of this as a defence of direct conativism in a context in which we conceive of the SP-​relation as a self-​regarding relation. That is to say, when we talk of the SP-​relation, we are talking of the relation that obtains between person-​phases, and which is the relation tracked by certain distinctively first-​personal and prudential concerns. The SP-​relation, conceived like this, is clearly an important relation. ‘Same person’ in this sense, is the thing I care about, in a certain sort of way, when I care about having continuers of a certain sort. What settles whether this relation obtains between person-​phase p, and other person-​phases, are p’s direct conative states. None of this, however, is to deny that there are important other-​regarding practices, and that there are other senses of ‘same person’ that map onto those practices. A  natural way to put this is that there are both self-​ regarding SP-​ relations—​SP-​relations that are settled by distinctively first-​personal, agentive and prudential conative states—​as well as other-​regarding SP-​relations—​SP-​relations that are settled by distinctively third-​personal, and social, practices. This allows that different relations might realise the self-​regrading SP-​relation, depending on which direct conative states some person-​phase has, just as it allows that different relations might realise the other-​regarding SP-​relation, depending on which other-​regarding and social practices some person-​phase, or set of phases, has. Moreover, conceiving of things in this manner opens up the prospects for considering the ways in which self-​regarding and other-​regarding conations do, and ought, interact. For instance, we previously described a case in which p’s self-​ regarding conations are organised around a very different relation than are p’s family’s other-​regarding conations towards p. Whether one thinks that, by and large, other-​regarding conations about p, and p’s own self-​regarding conations, ought coincide (at least when appropriately idealised) will depend on broader ethical considerations. Communitarians, for instance, might see no reason why other-​regarding

Conativism about personal identity  155

conations with respect to p, ought mirror p’s own self-​regarding conations. Others might disagree, supposing that this is to fail to respect certain aspects of p’s autonomy. Our aim is not to resolve that issue here, but merely to note its importance. Our core claim is just that there is an especially important sense of ‘same-​person’ which is the self-​regarding sense that we have articulated, and is the key one relevant to deliberation about one’s future, and it is this sense, we think, that is settled by direct conations of the relevant person-​phases.

7.7  Conclusion The scope of different forms of conativism is wide, and there is much more to be said about the connection between different conations: public and private, self and other-​regarding. We hope though, to have provided an overview of some of the options, and gone some way towards showing why one might find something in the vicinity attractive.

Notes 1 Put this way the view is a form of cognitivism; one could have a non-​cognitivist version according to which judgements about the SP-​relation express certain conative states, but we will not discuss that version here. Much, but not all, of what we go on to say would apply to such a version. 2 We say that it’s a matter of conceptual analysis, in part, just to have one version of the story playing centre stage.The first order features of our account would be no different in some important ways if it were a metaphysical discovery that SP-​relations are relations-​ when-​conatively relevant, for example. 3 This is not to suggest that all conativists suppose that conativism is true as a matter of conceptual analysis. It is not clear, for instance, that this is exactly Kovacs’ (2016) view. 4 For discussion of synchronic conventionalist views see Braddon-​Mitchell and Miller (2004) and Kovacs (2016). 5 Since the personal-​identity relation comes with various different kinds of metaphysical baggage, we’ve introduced the SP-​relation in order to target the relation that we have in mind. 6 Though we use the term ‘phase’ in this context, we don’t intend the sort of heavy-​duty notion that Johnston introduces by his talk of phase sorts, or crypto phase sorts, etc. 7 We take conventions to be a kind of conation in that conventions are an organised system of behaviours and choices that reflect underlying conations. 8 See Lewis (1976); Sider (2001) and Shoemaker (1984, 2003) for reductionist accounts of the SP-​relation in terms of similarity and causal connectedness. We take it that ‘further fact’ or ‘simple views’ (as described by Parfit (1984 p. 210) and defended by, inter alia, Swinburne 1984) according to which the presence of these features is at best evidence for the presence of the SP-​relation are much more likely (though perhaps not required) to suppose that the relevant relation is identity. 9 In part that is because we think that what makes something an account of personal-​ identity (rather than something else) is that what it tracks, plays a certain role in our prudential and moral theorising (i.e. meets the practical constraint). Dummett (1981 p. 358) makes a similar point.

156  David Braddon-Mitchell and Kristie Miller

10 See Eklund (2004) and Shoemaker (2007 and 2016) for a discussion of the ways in which accounts of person-​identity can ‘come apart’ from accounts of the ways in which we ought to structure our prudential or moral concerns. Schechtman (2014; c­ hapter 1) is one theorist who takes seriously the practical concern desideratum. 11 See Miller (2006) for a defence of this view that deals with the obvious issues. 12 See Johnston (1989). 13 For a defence of this claim see Sider (2001a), Braddon-​ Mitchell and West (2001), Braddon-​Mitchell and Miller (2004) and Miller (2009). 14 An attitude about ‘who one is’ probably is an expressive matter. But the point here is that in such a version of conativism there is a fact of the matter about the SP-​relation which might in part depend on what states are expressed, rather than claims about the SP-​ relation having themselves an expressivist analysis. For work on ‘who one is’ (though not necessarily in the context of a conativist account of personal identity), see Schechtman (1996; 2011); Frankfurt (1987; 1988); Ludwig (1997); Lindemann Nelson (2001; 2002) and Rudder-​Baker (2016). 15 We say ‘appear to be’ beliefs because they are states without content: nothing about them fixes their intention, and if beliefs are individuated intentionally they aren’t beliefs. They only gain content if this is filled in by conative attitudes. 16 See Braddon-​Mitchell and West (2001). 17 See Braddon-​Mitchell and West (2001), Miller (2009) and Kovacs (2016). 18 Schechtman notes that on her view persons are unified loci of interaction and that ‘It is not up to us what conditions must be met for there to be a unified locus of interaction, or for a single locus to continue’ (2015; p. 196.) But, of course, it is up to us, at least in the sense that it depends on us, what our conative states, and the conventions and practices that depend on that states, are, and, moreover, were those conative states different in various ways, then the facts about the SP-​relation would be different. As far as we are concerned, that is all that is required for a view to be conativist or, more specifically, conventionalist. 19 Braddon-​Mitchell and Miller (2004). 20 Indeed, one might think that this is all they depend on. 21 Braddon-​ Mitchell and West (2001) characterize person-​ directed practices in this manner. Their view is probably best interpreted as a version of practice-​based conventionalism. See also Johnston (1989); Unger (1990), Nozick (1981) and Perry (1972). We think that Schechtman (2014) also endorses a version of practice-​based conativism, though she would not self-​describe as a conventionalist. It may also be that would allow somewhat less faultless disagreement than the view allows, for it may be that she thinks that any recognisably human society will have relevantly similar practices of this kind (or at least, practices that don’t differ in such a way that would result in quite different relations being the SP-​relation). Schechtman allows, however, that once the public conventions are in place (and private conations are required for this) the presence of these conventions, alone, can ground there being a persisting person, even if the person-​phases in question do not have the capacity to have the sorts of first-​personal conative states that are typically thought to be definitive of persons. See for instance (Schechtman 2014 p. 106). 22 This view is explicitly endorsed by Kovacs (2016). 23 Examples of conative versions of conventionalism, (some of which are weak conventionalism, and some strong conventionalism) include Braddon-​Mitchell and West 2001; Braddon-​Mitchell and Miller (2004); Miller (2009) and Johnson (1989). Views of this kind have also been gestured towards in Nozick (1981 p. 69), Unger (1990 p. 66) and Whiting (1986). See also D. Shoemaker (2003) for a related view according to which

Conativism about personal identity  157

‘our identity as functioning, well-​developed agents is constituted by our nexus of cares.’ Eklund (2004) calls views of this kind self-​concern relativism. 24 See Braddon-​Mitchell and Miller (2001) for a defence of this view. 25 As Parfit (1984), notes. 26 Braddon-​Mitchell and West (2001) and Sider (forthcoming) also argue for views with this feature. In addition, Braddon-​Mitchell and Miller (2001) argue for a version of the desire-​first view according to which the SP-​relation comes in continuous degrees. On that version of the view, it can be that although both P and P* are, from each of their perspectives, SP-​related to one another, nevertheless from P’s perspective the SP-​relation holds to degree 0.5, say, and from P*’s perspective the SO-​relation holds to degree 0.7. We won’t get into the complexities of that view here. 27 Having made the choice we just did, we are absolved from making decisions about whether the direct SP-​relations hold between phases and their neighbours, or between phases and certain other phases, or phases and all other phases. For on our view, which of these is the case (from the perspective of a phase) is an empirical matter of what that phase’s conations are like. A defender of the view that there is some independent metaphysical fact, but who thinks that various conations are relevant to determining the SP-​ relations, has the task of saying which of these views is true as a matter of independent metaphysical fact. 28 As first introduced by Lewis (1979). 29 By analogy with Lewis’ view about mereological fusion. 30 Of course on some accounts of the nature of concepts, it is sometimes part of a concept how one should revise on making various sorts of discoveries about actuality (two-​ dimensional concepts are one such example). In which case we get to preserve the idea that our story is a conceptual analysis, which says certain things about the SP-​relation conditional on there being no metaphysical facts necessarily connected with various conations. 31 Shoemaker (2007) makes this point. 32 As Schechtman puts it “according to PLV (the person life view) persons are loci of interpersonal interaction whose integrity as unified wholes results from complex and dynamic interactions among biological, psychological and social processes” (2014 p. 184). 33 This is one way of reading Schechtman’s proposal. One could, of course, go this route but disagree with Schechtman about which conations are relevant to settling the SP-​relation.

References Braddon-​Mitchell, D. and Miller, K. (2004) ‘How to be a Conventional Person’, The Monist 87(4): 457–​474. Braddon Mitchell, D. and Miller, K. (unpublished manuscript). ‘Survival, to a degree’. Braddon-​Mitchell, D. and Nola, R. (2008). ‘Introducing the Canberra Plan’, Conceptual Analysis and Philosophical Naturalism, MIT Press. Braddon-​ Mitchell, D. and West, C. (2001). ‘Temporal Phase Pluralism’, Philosophy and Phenomenological Research 62(1): 59–​83. Chalmers, D. and Jackson, F. (2001). ‘Conceptual Analysis and Reductive Explanation’, Philosophical Review, 110(3): 315–​361. Chalmers, D. J. (2012). Constructing the World. Oxford: Oxford University Press. Dummett, M. (1981). Frege:  Philosophy of Language, 2nd ed’n. Cambridge, MA:  Harvard University Press. Eklund, M. (2004). ‘Personal Identity, Concerns, and Indeterminacy’, The Monist, 87(4): 489–​511.

158  David Braddon-Mitchell and Kristie Miller

Frankfurt, H. G. (1987). ‘Identification and Wholeheartedness’. In F. D. Schoeman (ed.), Responsiblity, Character, and the Emotions:  New Essays in Moral Psychology. Cambridge: Cambridge University Press. Frankfurt, H. G. (1988). The Importance of What We Care About: Philosophical Essays. Cambridge: Cambridge University Press. Johnston, M. (1989). ‘Relativism and the Self ’. In M. Krausz (ed.), Relativism: Interpretation and Confrontation. Notre Dame University Press. Kovacs, D. (2016). ‘Self-​Made people’, Mind, 125(500): 1071–​1099. Kovacs, D. (unpublished manuscript). ‘Diachronic Self Making’. https://​davidmarkkovacs. weebly.com/​uploads/​5/​5/​9/​5/​55953415/​dsv_​-​_​website.pdf Lewis, D. (1979). ‘Attitudes De Dicto and De Se’, The Philosophical Review, 88(4): 513–​543. Lewis, D. (1986). On the Plurality of Worlds. Oxford: Blackwell. Lindemann Nelson, H. (2001). Damaged Identity, Narrative Repair. Ithaca Cornell University Press. Lindemann Nelson, H. (2002). ‘What Child is This?’, Hastings Centre Report. Ludwig, A. M. (1997). How Do We Know Who We Are?: A Biography of the Self. Oxford: Oxford University Press. MacFarlane, J. (2009). Nonindexical contextualism. Synthese, 166(2): 231–​250. Miller, K. (2006). ‘Travelling in Time: How to Wholly Exist in Two Places at the Same Time’, Canadian Journal of Philosophy, 36(3): 309–​334. Miller, K. (2009) ‘Deciphering Personal-​identity Conventionalism’, Special Issue Rivista di Estetica “Convenzioni” edited by E. Casetta and A.Varzi 44(2): 59–​85. Miller, K. (2013). ‘Personal Identity Minus the Persons’, Philosophical Studies, 166(1): 91–​109. Nozick, R. (1981). Philosophical Explanations, Cambridge, MA: Harvard University Press. Olson, E. (1997). The Human Animal. Oxford: Oxford University Press. Parfit, D. (1984). Reasons and Persons. Oxford: Oxford University Press. Perry, J. (1972). ‘Can the Self Divide?’ Journal of Philosophy, 59: 463–​488. Price, H. (2008). ‘Two Notions of Naturalism’ http://​philsci-​archive.pitt.edu/​4430/​1/​ Tilburg1.pdf retrieved November 2017. Rudder-​Baker, L. (2016). ‘Making Sense of Ourselves: Self-​Narratives and Personal Identity’. Phenomenology and the Cognitive Sciences, 15(1): 7–​15. Schechtman, M. (2014). Staying Alive: Personal Identity Practical Concerns, and the Unity of a Life. Oxford: Oxford University Press. Schechtman, M. (2011). ‘The Narrative Self ’, in S. Gallagher (ed.), The Oxford Handbook of the Self. Oxford: Oxford University Press. Schechtman, M. (1996). The Constitution of Selves. Cornell University Press. Shoemaker, D. W. (2003). ‘Caring, Identification, and Agency’, Ethics, 114(1): 88–​118. Shoemaker, D. W. (2007). ‘Personal Identity and Practical Concerns’. Mind, 116(462): 317–​357. Shoemaker, D. W. (2016). ‘The Stony Metaphysical Heart of Animalism’. In S. Blatti and P. Snowdon (eds.), Animalism. Oxford: Oxford University Press, pp. 303–​328. Shoemaker, S. (1984). Personal Identity. Blackwell. Shoemaker, S. (2003). Identity, Cause, and Mind:  Philosophical Essays. Oxford:  Oxford University Press. Shoemaker, S. and.Swinburne, R. (1984). Personal Identity. Oxford: Blackwell. Sider, T. (1996). All the world’s a stage, Australasian Journal of Philosophy, 74: 433–​453. Sider, T. (2001a). ‘Criteria of Personal Identity and the Limits of Conceptual Analysis’. Philosophical Perspectives, 15(s15): 189–​209. Sider, T. (2001b). Four Dimensionalism. Oxford: Oxford University Press.

Conativism about personal identity  159

Sider,T. (2018). ‘Asymmetric personal identity’. Journal of the American Philosophical Association, 4(2), 127–​146. Unger, P. (1990). Identity, Consciousness and Value. Oxford: Oxford University Press. West, C. (1996). PhD Dissertation, ANU. White, S. (1989). ‘Metapsychological Relativism and the Self ’, Journal of Philosophy, 86: 298–​325. Whiting, J. (1986). ‘Friends and Future Selves’, Philosophical Review, 95: 547–​580.

8 REASONS AND CONSCIOUS PERSONS Christian Coseru

Introduction What justifies holding the person that we are today morally responsible for something we did a year ago? And why are we justified in showing prudential concern for the future welfare of the person we will be a year from now? Whatever our answer to these questions, it seems that we cannot systematically pursue them without in one way or another referring to persons and their identity over time. But while there is widespread agreement that considerations about personal identity must be front and center in any such inquiries, such agreement falls short when it comes to specifying the criteria for personal identity, that is, what this identity necessarily involves or consists in. Part of the difficulty is that an investigation into the nature of personal identity brings us to metaphysical questions about persons, their ontological status, identity conditions, and persistence over time. The challenge, then, is to pursue these additional questions without losing sight of the practical concerns that prompted them in the first place. Few contemporary philosophers have confronted this challenge with more analytic skill, depth, and ingenuity than Derek Parfit. In engaging with Parfit’s work on personal identity, primarily his Reason and Persons, my aim is to reassess his Reductionist View of personal identity in light of Buddhist Reductionism, a philosophical project grounded on the idea that persons reduce to a set of bodily, sensory, perceptual, dispositional, and conscious elements, which alone are real. Parfit is not only familiar with this Buddhist conception of personal identity, but thinks that the reductionist, no ownership position he defends, which takes persons both to exist and to reduce to their components, is true, and that, as he famously puts it, “Buddha would have agreed” (1984:  273). My goal here is threefold:  first, to review Parfit’s Reductionism position and evaluate its main arguments; second, to assess the extent to which Buddhist Reductionism supports Parfit’s psychological

Reasons and conscious persons  161

criterion for personal identity; and finally, to suggest some ways in which Buddhist conceptions of mind and consciousness can help to advance the contemporary debate on personal identity.

Reductionism and personal identity What is reductionism and how does it relate to the question of personal identity? The literal sense of the term, derived from the Latin ‘reducere’ (lit. ‘bring back’, ‘lead back’), captures rather well the philosophical notion of reduction: to reduce is ‘to lead back’ to something more fundamental than the thing in question. To say that the mental reduces to the physical, that motion reduces to kinetic energy, and that psychology reduces to biology, is to capture what it means for one thing (mind, motion, or psychology) to be brought back to the other (physical, kinetic energy, or biology). Reduction, then, stands for the view that, if entity x reduces to entity y, then y is prior to, more basic than, or constitutive of x.1 As a philosophical term of art, ‘reduction’ is central both to metaphysical questions of personal identity and to questions in philosophy of mind about consciousness, agency, and the mind-​body problem. One is a reductionist if one takes a particular theory or phenomenon to be conceivable in terms of, or reducible to, another theory or phenomenon.2 Parfit is one such reductionist, and Reasons and Persons contains his vastly influential theory of personal identity, which argues against the commonsensical, non-​ reductionist view of persons. According to the non-​reductionist view, persons are distinct and discrete entities that exist over and above their bodies and psychological states. Their identity, then, is an irreducible, brute fact of existence, and cannot be explained or described in more basic terms. Whatever persons are, an account of their identity would have to employ person-​level descriptive categories of embodied experience.3 One paradigmatic example for person in this non-​reductive sense is the Cartesian Ego. For Parfit the view that there are such entities as Cartesian Egos or souls is representative of a particular intuition about personal identity, according to which we assume that questions of the sort ‘Will I survive the death of my body?’ or ‘Will I  be the same person if I  were to be teleported elsewhere?’ must have definitive answers. Regardless of whether or not we have answers to these questions at present, given their implications for personal identity, answers must in principle be available. There must be a way to settle these questions one way or another, perhaps on the basis of our very conception of what personal identity entails. What drives this intuition, argues Parfit, is the assumption that our identity must in some sense be determinate (1984: 214). If we reject this intuition and allow for the possibility that Cartesian Egos do not exist, then, argues Parfit, we are in a sense compelled to accept the view of Reductionism. One of the advantages of Reductionism is that it offers new possibilities for reconceiving the problem of personal identity on both metaphysical and empirical grounds. Since the body is the seat of our physical, affective, and mental lives, we may conceive of persons as (i) bodies or as (ii) entities that have bodies, thoughts, and emotions. The first conception can also be understood as an

162  Christian Coseru

endorsement of one version of the identity view (persons just are bodies), while the second makes the case for the ownership view (persons are the sort of entities that have bodies, thoughts, and other kinds of experiences). As Parfit makes clear, whereas the ownership or constitutive view of personal identity can be easily entertained, and may even fit classical conceptions of persons as property-​possessors, the identity view of reductionism opens the door for something more radical: Eliminative Reductionism, the best example of which is the mind/​brain identity theory. For the eliminative reductionist, persons, much like nations, are ultimately reducible to their constitutive parts. Just as the concept of a nation cannot be the concept of an entity that is distinct from its people and its territory, so also the concept of a person cannot be that of an entity that exists over and above its body, perceptions, thoughts, and feelings. We may thus conclude, argues Parfit, “that, in that case, there are really no such things as nations.There are only groups of people, living together in certain ways” (2002: 656). What does it mean to extend this analogy and to claim that there are really no such things as persons? On Parfit’s general characterization of reductionism, the existence of persons, then, “just involves the existence of a brain and body, the doing of certain deeds, the thinking of certain thoughts, the occurrence of certain experiences” (1984: 2011). The force of this reductionist move is captured by the ‘just consists in’ locution:  all of the facts that pertain to a person’s existence–​–​ their particular type of body, their psychological profile, the kinds of relations that obtain between their mental and brain states, all of these admit of an impersonal description that neither requires nor presupposes this particular person exist. As a façon de parler, however, we do say that, because we can ascribe thoughts to them, thinkers exist. But just because we can infer from the content of their experience that there are thinkers, it does not follow that their existence is somehow separate from their thoughts. Parfit endorses the impersonal description thesis (1984: 225) even as he acknowledges, with Peter Strawson and Bernard Williams, that any talk of experiences, and of the various relations that obtain between them, is impossible without reference to the persons whose experiences they are. Although he would subsequently disavow it (Parfit 1999), on the grounds that a thoroughly impersonal conceptual schema would be ineffective for the kind of creature we are, the impersonal description thesis plays an important explanatory function for the view that personal identity over time just consists in physical and/​or psychological continuity. To motivate both reductionism and the impersonal description thesis, Parfit provides the well-​known example of the simple teletransportation (STT) machine. Suppose that in a future century when science is far advanced I am able to travel to a distant place, say Mars, simply by having a machine scan every cell of my body and broadcast the information in real time to another machine on Mars. There is one catch, however: my current body will be destroyed in the process of being scanned. On the up side, a qualitatively identical but numerically different body will be created on Mars. The initiation of the whole process is marked by a loss of consciousness, which is immediately gained once the complete body has been recreated on Mars. In principle, argues Parfit, we should be able to provide a

Reasons and conscious persons  163

description of the person’s states at t1, before entering the Earth STT, at t2, while being teletransported, and at t3, after exiting the Mars STT, without referring to the persons whose states they are. But while imaginary scenarios such as this do appear to support the impersonal description thesis, they raise a different set of questions, specifically about whether personal identity could be determinate. Upon waking up on Mars, am I the same or a different person? Assuming that the two persons are similar enough psychologically, and that the latter’s personality, consciousness, and memories are caused by the former’s, we would have to conclude that they really are the same person, but at different stages. In this case, causal dependence would indeed serve as a necessary and sufficient criterion for personal identity. But as Parfit concedes, such scenarios are not entirely unproblematic for the psychological theory of personal identity. Consider the second scenario, where an advanced teletransportation (ATT) machine has the same scanning capabilities as the STT, plus the added feature of leaving the scanned body intact. In this case, having undergone teletransportation, and on the assumption that the person I am on Earth is psychologically very similar to the one I am now on Mars, one would find oneself in two places at once. One advantage of this second scenario is that it provides a compelling reason for adopting some version of the ‘body’ theory of personal identity. But neither scenario addresses the many objections to the impersonal description thesis such as, for instance, that reductionism cannot explain in a non-​question-​begging way how exactly we get persons from impersonal psychophysical elements. The teleportation examples aim to drive home the point that personal identity is indeterminate, and that reductionists must contend with a situation in which questions about the mode of existence of persons may yield no definitive answers.4 Notwithstanding these problems, Parfit thinks such examples are compelling enough to warrant that we take reductionism seriously, and presents us with two distinct conceptions of reductionism about persons: (1) a person just is a particular brain and body and a series of interrelated physical and mental events; and (2) a person is an entity that is distinct from a brain and body and such a series of events (1984: 211). Although he thinks that (1), which articulates the eliminativist version of reductionism, is justified in some instances (it is right to claim that “there were really no witches, only persecuted women” (2002:  657)), his conception of persons as rational and moral agents clearly compels him to accept (2). Furthermore, Parfit insists on the importance of thinking about persons not in terms of their mental states, which would entail that they must be the states of some entity, but in terms of the occurrence of certain experiences, deeds, and thoughts. Thus understood, all reductionists would accept the view that: (3) a person’s existence just consists in the existence of a brain and body, and the occurrence of a series of interrelated physical and mental events (1984: 211).

164  Christian Coseru

Now contrast this view with that of the non-​reductionist: (4) a person is a separate existing entity, distinct from their brain and body, and their experiences (1984: 210). But Parfit clearly distinguishes between two non-​reductionist views. The common one, already noted, is that of a person as a purely mental entity, a Cartesian Pure Ego, soul, or spiritual substance. On this view, “personal identity over time does not just consist in physical and/​or psychological continuity” (1984: 210). Rather, the Cartesian Ego is an altogether different metaphysical entity. The second non-​ reductionist view denies that persons are separately existing entities, distinct from their brains, bodies, and experiences: (5) though we are not separately existing entities, personal identity is a further fact, which does not just consist in physical and/​or psychological continuity (1984: 210). Parfit calls this the Further Fact View. The typical non-​reductionist view articulates a familiar conception of personal identity, whose basic premise is that human beings (and perhaps other forms of life) must be conceived in terms of something essential, without which we could not ascribe to them the properties of persons. Unlike the typical non-​reductionist view, the further fact view poses a conundrum: what we have here is both the notion that persons are not metaphysically distinct entities and the notion that their existence cannot be accounted for in purely impersonal terms. On the further fact view, then, the concept of person is simple and unanalyzable. Parfit’s account of personal identity, which had initially drawn on the reductionist no-​self view of Buddhist Abhidharma,5 may be more profitably associated with one Buddhist school of thought, Personalism (Pudgalavāda), that did indeed defend a conception of person as irreducible (more about the arguments adduced in its support in the next section). Before we turn to the Buddhist discussions of personal identity, a brief overview of Parfit’s psychological criterion of personal identity is in order.6 Parfit takes the ‘criterion’ for identity to stand for “what this identity necessarily involves, or consists in” (1984: 202). In the case of physical objects such as the pyramids or the Moon that simply refers to the concrete spatio-​temporal continuity of the object in question. So far as our physical body is concerned, the criterion of personal identity would simply consist in the persistence of this body and brain over time. Parfit offers what he assumes is a better version of the ‘physical criterion’: what is necessary for personal identity is the persistence of enough of the body and the brain over time, such that even if we were to lose some of the body and brain (through amputation or hemispherectomy) the same person would continue to exist. Nonetheless, on the physical criterion of personal identity, the case of STT would not count as persistence over time and neither would that of rebirth, which presuppose that something of oneself continues beyond the destruction of the body. This is largely

Reasons and conscious persons  165

the reason why Parfit thinks the physical criterion does not suffice, and why instead he puts forward a psychological criterion of personal identity, which he understands as involving two relations: Psychological connectedness is the holding of particular direct psychological connections. Psychological continuity is the holding of overlapping chains of strong connectedness. 1984: 206 Although Parfit takes connectedness, which is a matter of degree, to be a more important relation, he does not think it serves as an adequate criterion for personal identity. Strong connectedness is a transitive relation, since the person I am today is strongly connected to the one I was yesterday, and to the one I was three days ago. But it cannot be true that the person I am today is strongly connected to the one I was twenty years ago, and so, despite its importance, the strong relation of psychological connectedness is too problematic for personal identity: “Because identity is a transitive relation, the criterion of identity must also be a transitive relation. Since strong connectedness is not transitive, it cannot be the criterion of identity” (1984: 206). What about psychological continuity? Parfit uses the continuity relation to introduce his ‘psychological criterion’ of personal identity: There is psychological continuity if and only if there are overlapping chains of strong connectedness. X today is one and the same person as Y at some past time if and only if (2) X is psychologically continuous with Y, (3) this continuity has the right kind of cause, and (4) it has not taken a ‘branching’ form. (5) Personal identity over time just consists in the holding of facts like (2) to (4). 1984: 206 The key elements here are (2) psychological continuity and (3) right causal determination, and Parfit uses right causal determination to further distinguish between narrow, wide, and widest versions of the psychological criterion depending on whether the cause is identifiably the correct one, any reliable cause, or any cause whatsoever. These additional criteria provide a basis for introducing the memory element, since being able to remember past experiences as one’s own is essential to the psychological criterion of personal identity. Memory thus stands as an example of strong connectedness between, for instance, an experience once had and later remembered or an intention once formed and later acted upon. These sorts of direct causal connections that memory, intention, and action provide are typically of a short duration. I seldom act on intentions formed twenty years ago, and distant childhood memories have already been embedded in a complex self-​narrative that had undergone revision and embellishment over time (even though Parfit takes it to be a logical truth that we can only remember our own experiences).7 So, while

166  Christian Coseru

there may be few direct connections between, say, the 50-​year-​old adult and the 10-​year-​old child, they are still indirectly connected through multiple overlapping chains of direct connectedness. The overlapping chains model thus serves to illustrate Parfit’s understanding of the criterion for psychological continuity, whose functionality is such that we can talk of the child I was once and the adult I am today in one breath. While the narrow account of causal determination serves as a plausible way to understand persistence over time, it cannot accommodate the simple teletransportation case. For this reason, Parfit thinks we need the wide and even widest views of causality. The wide view allows any reliable causal chain to serve as a criterion for personal identity.The widest view is even more permissible, allowing for any causal chain to play this role (a requirement for the persistence of identity in the ATT case). Since Parfit claims that there is no good reason to prefer the narrow view over the others, we are confronted with a problematic notion of persistence: just what it means to say the person I am now on Mars is psychological continuous with the one I was on Earth an hour ago is not exactly clear. And it is even more puzzling to entertain what it would be like to talk of personal identity between the two versions of myself in the ATT case. What is clear is Parfit’s motivation for articulating these two views, both of which allow for a conception of persistence over time that reduces the person to certain chains of causal connectedness between mental and physical states. The issue is not whether the same individual persists over time, including in the ATT case. Rather, the problem is how best to guarantee there is enough resemblance between the different person stages. Even so, in articulating his psychological criterion, Parfit does not aim to secure a strong foundation for personal identity; rather, he wants to demonstrate that the persistence of identity is not an important factor when it comes to matters of moral and practical concern. Now that we have identified the two primary elements of Parfit’s theory of personal identity, let us consider to what extent Buddhist Reductionism supports Parfit’s psychological criterion for personal identity.

Buddhist reductionism, personalism, and the no-​self view As is well known, in an effort to demonstrate that his views are not the product of a particular culture and epoch, but rather apply to all people at all time, Parfit turns to, inter alia, Buddhism, where he finds an early reductionist stance not unlike his own. As he notes: I claim that, when we ask what persons are, and how they continue to exist, the fundamental question is a choice between two views. On one view, we are separately existing entities, distinct from our brain and bodies and our experiences…The other view is the Reductionist View. And I claim that, of these, the second view is true… Buddha would have agreed. 1984: 273

Reasons and conscious persons  167

Whether or not Buddha would have agreed depends in large measure on whether Buddhist Reductionism articulates something close to a psychological criterion of personal identity. I will argue that it does, although in a way that emphasizes the structural dynamics of conscious experience over accounts of psychophysical causal relations (of the sort that Parfit’s reductionist view of persons endorses). One of the clearest articulations of Buddhist Reductionism about persons is found in The Questions of Milinda (Milindapañha), an eloquent para-​canonical work in the style of a Platonic dialogue detailing an exchange between Nāgasena (a Buddhist monk) and Milinda (better known as the Bactrian King Menander I) aimed at articulating a nominalist conception of personal identity. In a much-​quoted passage, Nāgasena explains that, although he responds to a certain appellation–​–​his name–​–​there is no corresponding referent: Sir, I  am known as ‘Nāgasena’; my fellows in the religious life address me as ‘Nāgasena’. Although my parents gave (me) the name ‘Nāgasena’…it is just an appellation, a form of speech, a description, a conventional usage; ‘Nāgasena’ is only a name, for no person is found here. Horner 1963: 34 Parfit is familiar with The Questions of Milinda, and appeals to this and to several other excerpts from Buddhist texts8 to make the case that “the Reductionist View is not merely part of one cultural tradition. It may be, as I have claimed, the true view about all people at all times” (1984: 273). That Parfit’s view and that of the Buddhists bear a strong resemblance is both obvious and well documented.9 What is less obvious is the extent to which Buddhist metaphysics supports the Reductionist View. To answer this question, I will turn to one of the most influential attempts to show how Buddhist resources could help Parfit address some of the difficulties of his reductionist account while working out the implications of a Parfitian-​style conceptual framework for Buddhist philosophy in general, and for Buddhist Reductionism in particular: Siderits’ Personal Identity and Buddhist Philosophy (1997/​2015).10 Following Parfit, Siderits too acknowledges that reductionism comes in a variety of forms, and that when situated in the Buddhist context one must not overlook the (historical) fact that not all Buddhists are reductionists. Nonetheless, within the context of early Buddhist scholasticism (viz., Abhidharma), all Buddhist philosophers who, in addition to being reductionists about composite entities such as chariots, forests, and armies, also take a reductionist stance with regard to persons, may be identified as Buddhist Reductionists. As Siderits explains, a distinctive feature of Buddhist Reductionism is the two truths framework, which takes sentences to be either conventionally or ultimately true depending on whether they lead to successful practice or to what is taken to be ultimately real (2015: 16). For instance, Buddhists take partite entities such as chariots to be conceptually constructed and thus not ultimately real. Thus chariots reduce to ‘axle’, ‘felly’, and ‘linchpin’, which in turn reduce to their elemental parts, and so on. On this account, only impartite unanalyzable entities are ultimately real. Siderits reworks the two truths framework

168  Christian Coseru

into an account of semantic insulation, which he uses to justify the Buddhist Reductionist view that partite entities cannot be ultimately real. As he himself admits, this view, which articulates the position of mereological nihilism, “will strike many as giving an implausibly austere ontology” (2015:  14). Siderits finds some of the arguments adduced in its support compelling, though, as I will argue, for reasons that are not entirely clear.What, then, motivates Buddhist Reductionists’ mereological nihilism? The answer lies in the widely-​shared but, according to the reductionist, obviously mistaken non-​reductionist stance that most people adopt: When the Buddhist Reductionist claims that strictly speaking chariots do not exist, that ‘chariot’ is a mere conventional designator for a collection of parts arranged chariot-​wise, this is meant to serve as an example of widely held belief that turns out not to be strictly speaking true, but also of a belief whose acceptance is perfectly understandable given the demands of everyday life. The point is not to get us to stop believing there are chariots. The point is to help us see how we could all be mistaken about ‘I’. For the mechanism that generates belief in a real ‘I’ is the same; hypostatization. What we have in both cases is a many masquerading as a one. Siderits 2015: 98 Does use of the first-​personal pronoun and handling of the middle size dry goods that populate our mundane existence have such dramatic effects on belief generating processes and embodied practices? And if they do, how exactly can the dualist semantics of the two-​truths framework undo our ingrained hypostatization proclivities? The mereological nihilist singles out the effects on belief that configuration processes–​–​by which various parts in the world come to be arranged chariot-​wise and person-​wise–​–​can have when the demands of everyday life constrain one’s ontological outlook. Speedy and effective long-​distance travel may demand that one avails oneself of various modes of transportation. Should we admit cars, trains, and airplanes in our final ontology? Sure, the car in the commercial is more than a wheeled and motorized mode of transportation; it is also a symbol of comfort, elegance, efficiency, and status. It has its own event horizon, and it promises to augment human experience in profound ways. Nonetheless, such demands should be recognized for what they are: alluring invitations to reify that which is ultimately only a collection of parts. The belief ’s acceptance may well reflect conventional linguistic practices that, once fixed, make possible assertions in which it is actually true that there are chariots and cars. The problem, as Siderits carefully explains, is that with such customary linguistic practices in place it becomes possible to assert both that there are such things as chariots and that there are certain parts that can be arranged chariot-​wise. Semantic insulation between the two types of discourse may indeed give rise to anomalies, especially if one were to assert that both types of entities are real. But Siderits, following Thomasson (2007), thinks such anomalies indicate that the conditions under which something comes to be called a ‘chariot’ are not

Reasons and conscious persons  169

always fixed by rules that apply to the discourse about parts, and so the existence of chariots, and of ordinary objects in general, can be vindicated. However, it is an open question whether the dualist semantics of the two truths framework brings Buddhist Reductionism in line with the particular type that Parfit favors. Siderits thinks that it does, and considers a version of the teleportation scenario that aligns the two accounts, but in a way that does not eliminate the problem of indetermination.The question whether parts arranged a certain way are identical with or distinct from the chariot does not admit of a determinate answer, and neither does the question whether the swapping of parts from one to another chariot-​wise arrangement results in an identical or different arrangement. Siderits’ proposed solution involves metalinguistic analysis: What we can do is ascent to a meta-​language and discuss those relations between our uses of chariot on the one hand and ‘axle’ and the like on the other that explain why, when the parts are assembled in a certain way, we say there is a chariot, and why there may be cases where we do not know whether to say there is a chariot, or to say that there is diachronic identity between some present chariot and an earlier chariot. We can talk about the whole, or we can talk about the parts, but we cannot talk about both in the same breath. 2015: 110 The metalinguistic analysis solution may seem compelling enough in the case of chariots and other inanimate objects, but it confronts a paradox:  since the parts taken together in some degree of combination do not add up to a chariot, there are no such things as chariots. Parfit confronts the paradox while answering an objection from Bernard Williams against the view that we can get personal identity from a psychological spectrum. The objection concerns the sorites paradox. Run in the reverse, the argument considers a scenario in which, by means of small enough surgical interventions, a surgeon could cause a person to cease to be psychologically continuous with previous stages of herself. If a surgeon could alter in small increments the neurochemistry of the brain in such a way as to profoundly transform the psychological features that make up who you are, does that mean you have ceased to exist? Parfit’s response considers the semantics of vague terms, and his solution to the sorites paradox comes in the form of an appeal to stipulation: we simply decide how many grains of sand to call a heap. But whereas stipulation may work for natural kinds such as forests it does not work for artifacts such as chariots. A chariot is not simply an arbitrary number of parts, but rather a definitive set of parts arranged such as to function as a ‘chariot’. These may vary depending on whether one considers extra features and embellishments, but one could agree on a minimum number of parts without which the concept chariot could not be meaningfully applied to parts arranged chariot-​wise. But even if we adopted such a provisory stance, does that mean the argument would work in the case of persons? If Parfit is right, the physical criterion is not

170  Christian Coseru

enough for personal identity, which is why we need to consider the full spectrum of psychological states. Since persons are defined primarily in terms of the subjective and phenomenal character of their conscious mental states, how does Buddhist Reductionism solve the personal identity problem? It would seem that any account of persons requires person-​level descriptive categories, a point that, as we noted above, Parfit himself concedes. Even Reductionists do not deny that people exist. And, on our concept of a person, people are not thoughts and acts. They are thinkers and agents. I am not a series of experiences, but the person who has these experiences. A Reductionist can admit that, in this sense, a person is what has experiences or the subject of experiences. This is true because of the way we talk. What a Reductionist denies is that the subject of experiences is a separately existing entity distinct from a brain and body, and a series of physical and mental events. 1984: 223 But in retaining a meaningful conception of ‘persons’ as thinkers and doers, Parfit’s view departs from mainstream Buddhist Reductionism only to join the ranks of the Buddhist Personalists. As expounders of the ‘personalist view’ (pudgalavāda), Buddhist Personalists do agree with the general tenet of Buddhist Reductionism that there are no such things as enduring substantive selves. They also retain the bundle view and take ‘persons’ to be conceptualized in dependence upon the psychophysical aggregates. Where they differ, however, is in admitting that persons, although neither identical nor different from the aggregates, are nonetheless real. It is unclear whether for the Buddhist Personalists persons enjoy the same ontological status as the aggregates, in part because they exhibit novel properties that neither of the aggregates do. What is clear, however, is that although the Personalists admit that the notion of ‘person’ is conceived in dependence upon the aggregates, they do not regard it as a conceptual fiction. Sustained efforts to reject Buddhist Personalism as an orthodox position form an integral part of Buddhist Reductionism. Historically, Buddhist Personalists were understood to have forgone commitment to the no-​self doctrine. However, it is now possible to accommodate their views on an emergentist or supervenience account of phenomenal properties.11 It is instructive that what makes Buddhist Personalism the target of criticism is a certain insistence on taking what we would now call the first-​person perspective seriously, particularly as it finds articulation in discussions of the relation between consciousness and content. The key issue under dispute is the concept of person itself: if it stands simply for the collection of aggregates, then the Buddhist Reductionist is right to insist on its fictional status. However, if the concept of person is meant to capture something determinate, such as the view that persons are self-​conscious subjects, then it is hard to see how the reductionist can make the case that the concept of ‘person’ lacks a proper referent. Consider, for

Reasons and conscious persons  171

instance, Vasubandhu’s refutation of personalism, and the argument that, since an individual can only possess ‘person-​properties’ on account of possessing some sort of essence (svabhāva), and since Buddhist metaphysics is anti-​essentialist, there is no ontological basis for admitting entities defined in terms of their possession of person-​properties.12 On the argument put forth here, person-​properties are those very things that allow us to use the first-​person pronoun meaningfully in sentences such as ‘I am walking to my office’ or ‘I am thinking about the conversation we had yesterday’. The motivation for resisting such language use, according to Vasubandhu, reflects its unintended consequences: conceiving of the aggregates together as forming a person is a slippery slope to thinking there are such things as selves. To be clear, Vasubandhu does not reject the view that we do appear to ourselves to be more than just a body, or a bundle of feelings and thoughts. Rather, he rejects the notion that feelings, thoughts, and the body are something that we possess. Such a view would entail an existence apart from the aggregates. But since we are nothing but thoughts feelings, perceptions, memories, and bodies of a particular shape, size, and gender, no such residual existence is anywhere to be found. In claiming that persons are conceived in dependence upon or in reliance upon the aggregates, though neither identical nor different from them, the Personalists invoke the conceivability principle. The dispute, then, hinges on whether or not the object of conception must be identical with its causal basis. If no such identity relation were to obtain, then conceiving would lack reference, and would not point to anything in the world. But even on a strongly non-​referential view, conceiving must be regarded as an effective epistemic practice that entails, inter alia, reference to nonconceptual entities (regardless of their ontological status). On a referential view of conception, thus, to conceive of persons is to conceive ourselves as persons. In conceiving of becoming a father I conceived of myself as a father. In conceiving of unicorns, I conceive of them as possible or impossible. On the personalist view, the aggregates may serve as a causal basis for our conception of person without being identical with it. Conception is indeed reliant on the presence of some manifest phenomena, but conceiving of ‘ourselves’ in dependence upon the aggregates does not make ‘us’ identical with them. We conceive of fire in relation to fuel and of friendship in relation to common interest, but fire is not identical with fuel, nor is friendship identical with common interest. The Buddhist Personalist, pace Vasubandhu, wants to make the case that the object of conception is not always identical with its causal basis, which is why we can both retain a meaningful conception of person and be reductionists about selves. It would seem that this particular version of Buddhist Reductionism, which retains the concept of person (though not that of self) would support Parfit’s conception of personal identity in the STT case, though not in the ATT case. Since the person is conceived in dependence upon the causal series, though not identical with it, I am sufficiently connected to the previous instance of myself prior to entering the Earth STT.The conceivability criterion of personalism makes the ATT case problematic, given that there is no way to differentiate between conceiving of

172  Christian Coseru

myself in dependence upon the causal series now present on Mars as opposed to those still on Earth. Might the person so conceived by Buddhist Personalists be a further fact, and could that be a mitigating factor against mereological nihilism? Siderits considers this objection by engaging some recent work in ontology, specifically Merricks’ (2001) defense of eliminativism. Merricks’ view is that strictly speaking there are no such things as statues, only “atoms arranged statuewise.” In denying the existence of macroscopic objects, Merricks is not simply arguing against conventional uses of language, of the sort that take statement such as “there exists a statue” to mean “there exist atoms arranged statuewise” (2001:  3). Nor is he denying the existence of statues because statues are nothing over and above atoms arranged statuewise, and thus not mereologically distinct from their parts. Rather, he puts forward a complex argument from causal overdetermination: admitting that there exist such things as statues causally overdetermines their effects (because whatever effect can be attributed to a statue is already an effect of its microphysical parts). Since such overdetermination is not observed, we must admit that there are no such things as statues.13 However, Merricks think the causal overdetermination argument does not apply in the case of human beings, because humans have causal powers that exceed those of their constitutive psychophysical parts. Most notably, humans instantiate consciousness, a property that does not supervene on the body’s microphysical properties. Merricks offers an interesting argument in support of this view operating on the premise that conscious and subjective mental properties bear the mark of being intrinsic. ‘Intrinsic properties’ are those properties “that an object can exemplify even if that object and its parts (if any) are the only objects that exist” (2001: 92). What is distinctive about humans is precisely that there is no metaphysical necessity that their intrinsic causal powers be implied by the existence of intrinsic properties of, and spatiotemporal and causal interaction among, their constituent elements. The argument goes as follow: suppose P, a conscious being, accidentally slices off her left index finger and thus shrinks. Suppose that at the very same moment of amputation, the atoms that compose her remain just as they were before amputation. The implication is that those atoms that are constitutive of P after amputation, are also constitutive of it immediately before amputation. Now, consider the pre-​ amputation finger-​complement that is not identical with P (on the grounds that P has a part, the finger-​complement, which the finger-​complement itself simply lacks). If that is the case, and if we assume a relational view of consciousness (as a composite of atoms grouped by spatiotemporal and causal interrelations), then we must concede that before amputation there are two consciousnesses present, one for P and the other for the finger-​compliment. But since we only ever have one such entity, P, the thesis that there are two consciousnesses must be false. Merricks thinks that to postulate the existence of a conscious pre-​amputation finger-​complement is to engage in an unacceptable multiplication of persons. If it is true that the schema of the overdetermination argument does not apply to the effects a human causes by virtue of being conscious, then humans cannot be

Reasons and conscious persons  173

treated as epiphenomenal entities. As composite entities capable of causing effects that our parts cannot redundantly cause by virtue of being conscious, we humans can resist the eliminative sweep of the reductionist. Now, Siderits is willing to grant that, were the argument from overdetermination to succeed, and the effects of conscious states are not systematically overdetermined, then we would have a forceful case for there being persons with causal effects that are irreducible to those of their constitutive psychophysical elements (2015: 112). This argument may not rehabilitate composite entities such as chariots, baseballs, or trees. But it would establish Parfit’s further fact version of Non-​Reductionism:  persons are not just suitably arranged psychophysical aggregates that exhibit novel causal powers. Rather, persons are what these specific aggregates are constituted as, which is why persons do not reduce to the aggregates in the same way that the chariot reduces to its component parts, and the forest to the collection of trees. The difference between Buddhist Reductionism and Buddhist Personalism should be clearer by now: whereas the reductivist claims that mental states strongly supervene on physical states and thus are reducible to the latter, the personalist takes them to possess novel causal powers that cannot be ontologically analyzed in terms of the properties of the physical states upon which they supervene. It is possible to interpret Buddhist Personalism as endorsing an emergentist view of persons and not simply a non-​reductive physicalist view, but that would be dependent on how we interpret the ‘in reliance upon’ clause. Siderits’ response to the arguments against causal overdetermination takes up the problem of compatibility between four different positions Buddhist Reductionism must consider when debating the ontological status of persons: (1) dualism, (2) reductive physicalism, (3) non-​reductive physicalism, and (4) emergentism.When considered in the context of debates about wholes and parts, these positions correspond respectively to (1) mereological realism, (2) reductive mereological nihilism, (3) non-​reductive mereological nihilism, and (4)  emergentism (2015:  114). Drawing on contemporary debates on these issues,14 Siderits admits that the non-​reductive physicalist position (viz., mental states supervene on physical states but are not reducible to them) ought to, at least in principle, be compatible with emergentism (viz., mental states do supervene on physical states, but since they exhibit novel causal powers, physicalism must be false). The question, however, is whether conscious mental states, and by extension the person whose states they are, are fundamental despite being dependent upon the intrinsic properties of, and the spatiotemporal and causal interaction among, their constituent elements. Siderits thinks attempts to answer this question that appeal to the notion of downward causation face an explanatory gap, and so efforts to close that gap inevitably “push the non-​ reductivist in the direction of reductionism and identity” (2015:  116). So, the position is best classified as a version of the “non-​reductive supervenience view” (2015: 117). But he does not agree that Buddhist Personalism offers a legitimate way to preserve the irreducibility of person-​properties necessary for, say, negotiating moral desert, and finds functionalist accounts of mental content compelling enough the resist the non-​reductionist move.

174  Christian Coseru

It is debatable whether the further-​fact view of Buddhist Personalism is as inherently unstable as Siderits claims. In the next section, I  will argue that it is not, and that considerations about self-​consciousness threaten certain aspects of the Reductionist View, and thus render the conception of person (as conceived in dependence upon the aggregates, though neither identical nor different from them) essential for psychological accounts of personal identity.

Reflexivity, agency, and the unity of conscious experience The much-​quoted passage from The Questions of Milinda (Milindapañha) discussed above–​–​in which Nāgasena argues against the ultimate reality of persons–​–​stands in sharp contrast to an equally important, but less celebrated, critical statement from his interlocutor, King Milinda, concerning the moral and pragmatic consequences of the no-​self doctrine: If, most revered Nāgasena, no person is apprehended, then who gives you the offerings that you receive as a Buddhist monk? Who enjoys those things? Who practices disciplined conduct? Who enters into contemplation? Who attains to the goal of nirvāṇa, which is the outcome of the path of cultivation? If, revered Nāgasena, one was to slay you, his would not be the crime or murder. Horner 1963: 100 As this passage makes clear, the problem of personal identity is not exclusively metaphysical, but engages central issues in ethics and moral psychology. Do the moral and mental forms of cultivation at the heart of Buddhist practice demand a robust notion of agency? Buddhism is indeed unique in articulating a theory of action that, it seems, dispenses altogether with the notion of agent causation. Even though practices of moral and mental cultivation form an integral part of the Buddhist path, there is no stable self or agent who bears the accumulated responsibility for initiating those pursuits, and seemingly no normative framework against which some dispositions, thoughts, and actions are deemed felicitous, and thus worthy of cultivation, while others are not so deemed. The agent-​neutral metaphysical picture of Buddhist scholasticism thus challenges traditional conceptions of moral agency. One way to address this challenge is to claim that the discourse of ‘persons’ and the discourse of ‘causes’ are compatible in so far as they belong in two distinct and incommensurable domains.15 Compatibilists who adopt this position typically cite evidence from social and cognitive psychology to show that any robust notion of agent causation must be incoherent.16 But, as I have argued elsewhere, such solutions compromise traditional conceptions of moral responsibility and render ethical conduct indistinguishable from merely pragmatic acts (Coseru 2016). Indeed, despite the dominance of an ultraminimalist account of agency, there are good and compelling reasons to give Buddhist Personalists credit for insisting

Reasons and conscious persons  175

that in so far as the aggregates operate in a person-​constituting way, persons are ineliminable from our discourse about agency and moral responsibility.17 Does Buddhist Personalism provide a closer analogue for Parfit’s theory of personal identity (and its psychological reductionism) than Buddhist Reductionism, with its stated mereological nihilist view that there are no such things as composite entities? I think that it does, for conceivability reasons: the psychological criterion is plausible only to the extent that I can conceive of myself in dependence upon a sufficiently similar series of aggregates. But the conceivability principle concerns the epistemological, rather than the ontological, dimension of personal identity, which brings up an altogether different set of considerations, specifically about the relations that obtain between self-​referential mental states (those that presuppose the notion of oneself as an agent) and self-​consciousness. The epistemological dimension is framed by a different set of questions that pertain not to what awareness supervenes on but to its structure and specific properties, specifically:  What in particular accounts for a mental state becoming an instance of self-​consciousness? Does self-​consciousness require that its self-​referential relation is present to itself as an object? If we can answer these questions, we can make progress in understanding the relation between self-​referential mental states and self-​consciousness. And if we can get clarity about the nature and character of self-​consciousness, we are in a better position to understand what it is that makes us persons.18 Recall the schematic analysis of the five aggregates that informs the Buddhist Reductionist account of personal identity. In this analysis, only the body is a physical aggregate stricto sensu. Feelings, perceptions, dispositions, and consciousness can acquire an objective aspect, but are not properly speaking empirically tractable phenomena. Nor are they things, that is, abstract entities with well-​defined properties and functional characteristics.What this reductive analysis of persons in terms of their constitutive features is meant to capture is not what persons are made of, but rather what human experience is constituted as: specifically, as a series of intentional and self-​ referential mental events. Consider the paradigmatic example of pain: as a sensation, pain is not reducible to the physical substrate, say a finger, in which it is instantiated (nor presumably to a mere physiological response). Rather, pain is constituted as a distinctly qualitative phenomenon whose intentional content cannot be dissociated from its subjective aspect.There is no such thing as generic or impersonal pain (understood strictly in terms of, say, the activation of Aδ-​and C-​fibers following an intense stimulation of nociceptors) apart from phenomenologically foregrounded sensations of some kind: of burning, stinging, or throbbing. Feelings may define the quality of the impressions that result from contact with an object, with the implication that they perhaps stand in a causal relation with these objects. In the schematic analysis of Buddhist scholasticism, they are categorized as mental states conditioned by habitual tendencies (vāsanā), which, in turn, they condition. Likewise, apperception (saṃjñā), the capacity to make intelligible or cause to be understood, although dependent on a multiplicity of psychological factors, captures the datum of experience only as unified into a single percept (since what Abhidharma psychology understands as a

176  Christian Coseru

simple apprehension and identification of a sensible trope actually involves a significant amount of unification of sensory input that goes on behind the scenes).Volitions too fit the same profile, with one important difference: rather than attending to the object at hand or providing a sort of transcendental unity of apperception, they bring forth future states of existence. As dispositions to act in certain ways, they cleave the mental domain into two classes of conditioned phenomena: those that are internal to consciousness, such as, for instance, obsessive dispositions like greed and delusion, and those that are dissociated from it, usually taken to refer to latent dispositions typically comprising various biological and physical traits.19 This dynamic model of personal identity is not incompatible with the notion that there are phenomenally primitive features of experience, features that any analysis of the structure of self-​consciousness must ultimately reveal. When the Buddhist Personalists insist that a mere functional account of the aggregates will not suffice to explain why an action counts as murder they draw attention to the specificity and individuality of a given bundle of aggregates, and hence, of its actions and consequences: “If the person were absolutely non-​existent, then there could not be killing nor (nor?) would the killer have killed anything. There would be nothing like theft and robbery … good and bad would yield neither freedom nor bondage; even bondage would have no one bound. There would be neither the doer nor the deed not any result thereof ” (Venkataramanan 1953:  177). Indeed, understanding why something is categorized as murder and not simply as the rearranging of aggregates presupposes a conception of intentional action that is unintelligible without reference to persons. If living beings were mere conventional designations, as Nāgasena contends, then there would be no non-​arbitrary way to assign guilt in the case of killing a sentient being, since killing a living ox would be no different from destroying a clay ox.20 Unlike clumps of clay, living beings are characterized primarily in terms of their capacity for responsive and intentional action: they can both do things and have things done to them in a way unavailable to insentient objects. Persons, unlike mountains or chariots, are not simply generic unities of aggregates, but self-​disclosing wholes. Persons are what they are by virtue of the fact that their aggregates are perceived to belong together.21 Aggregates grouped together in a person-​constituting way are specific to themselves in a way that mountains grouped together or chariot parts assembled together are not. What are the implications of this position for Buddhist Reductionism in particular, and for the Reductionist View of personal identity in general? There are various responses to this question. One may take the reflexive structure of self-​ consciousness to point to deeper aspects of consciousness as conscience,22 or delineate several ways in which pronominal and indexical uses of the ‘I’ can engender a false sense of self.23 One may also claim that the perceived unity of consciousness results from certain patterns of attention to, and grasping of, various features of experience.24 Whether one takes self-​consciousness to have a distinctive structure and clearly specifiable content, for the Buddhist Reductionist no minimal account of agency can escape the antirealist stance of the no-​self doctrine and its implications, namely that all modes of ‘I-​making’ (and ‘I-​sensing’) characteristic of

Reasons and conscious persons  177

self-​referential mental states must be conceptually constructed. This is precisely the reason why Siderits, for instance, thinks that we should take only causally efficacious particulars as providing metaphysical grounding (Siderits 2015: 115). Buddhist Reductionism does indeed pose a significant challenge for conceptions of phenomenal consciousness that take consciousness to consist in more than just a succession of associated ideas or a construct of post-​hoc rationalization. However, by treating its contents as transient episodes arising within a continuum of causally interconnected states, Buddhist Reductionism itself confronts an epistemic explanatory gap: if all there is to the experiential domain is a stream of momentary, object-​directed mental events, just what it means for an object to be known, and by whom? Can Buddhist Reductionism, with its account of primitive atom-​like ‘qualitons’ of experience, provide an adequate conceptual basis for mapping out the distinctly self-​intimating character of knowledge episodes? To answer this question let me briefly consider an epistemological attempt to negotiate the difference between egological and non-​egological accounts of self-​consciousness, drawing principally on the work of Dignāga. Building on the Abhidharma analytic project, Dignāga, not unlike Brentano, advances a dual-​aspect theory of mind, which takes the subjective or qualitative aspect (svābhāsa) and the object-​oriented or intentional aspect (viṣayābhasa) to be constitutive features of the structure of cognitive awareness.25 In large measure, this theory rests on three distinct claims: (1) that we are directly aware of each occurrent mental state; (2) that each occurrent mental state has a dual aspect: it has both subjective and objective content; and (3) that each occurrent mental state is also reflexively self-​conscious. The first claim goes against the view that occurrent mental states are ultimately impersonal or anonymous. To hold such a view would be akin to claiming that experience lacks a distinctly subjective character or possesses such character only by virtue of distortions embedded in the structure of ‘I-​making’ and ‘self-​grasping’ tendencies.26 The second claim identifies subjectivity and intentionality as distinct structural features rather than discrete elements of the mental domain. The subjective aspect is constitutive of an implicit openness to what is given, while the objective aspect captures what the mental state is about: an object or mental content of some kind. Finally, the third claim is intended to capture the mode of presentation of all conscious cognitive states. Incidentally, the first and third claims can also be read as making the case that effortful self-​knowledge–​–​of the sort gained through introspection, intersubjective reports, or reflective attitudes–​–​ depends on tacit or non-​propositional modes of acquaintance. This understanding of the structure of conscious experience is not unlike that put forward by those who argue that our subjectivity is immersive rather than egological. Consider, for instance, this statement from Zahavi, who takes Heidegger’s lead on this issue: “I am acquainted with myself when I am captured and captivated by the world. Self-​acquaintance is indeed only to be found in our immersion in the world, that is, self-​acquaintance is always the self-​acquaintance of a world-​ immersed self ” (Zahavi 2005:  82). On this view, self-​acquaintance is not something that occurs apart from our immersion in the world and its complex set of

178  Christian Coseru

intentional and intersubjective relations. Reflecting on the mode of presentation of our mental states, P. F. Strawson likewise observes, “our desires and preferences are not, in general, something we just note in ourselves as alien presences. To a large extent they are we” (Strawson 1992: 134). To claim, thus, that access to our mental lives is always mediated in some fashion or another, perhaps by participation in a shared domain of language and reflection–​–​is to ignore these essential features of self-​consciousness. Dignāga’s account of reflexive self-​consciousness (svasaṃvitti) does raise some legitimate concerns, for instance, about the absurd consequence of admitting that all cognitions, by virtue of being self-​presenting, are epistemically warranted.Yet the main issue, as I have argued at length elsewhere (Coseru 2012, Chapter 8), is specifying the criteria under which occurrent mental states become epistemically salient. A  reductive causal account that treats consciousness as emerging from elements that themselves lack phenomenal properties is necessarily incomplete if it does not explain how those elements acquire the phenomenal properties that they do when grouped together. Of course, not all groupings of microphysical elements need necessarily exhibit a reflexive structure.27 But those that do call for an explanation that considers the whole person and not just its constitutive elements. In putting forward an account of consciousness as self-​intimating, Dignāga is thus concerned to explain how a conscious mental episode, which is irreducible, can become the vehicle of self-​knowledge. In that sense, he shares a common ground with, inter alia, Aristotle, Descartes, Locke, Brentano, Husserl, Sartre, and many contemporary defenders of the view that the nature of consciousness is such that all conscious mental states are non-​representationally or pre-​reflectively aware of themselves.28 The picture they present corresponds to what some have termed non-​reductive “one-​level theories of consciousness”:  that is, theories which propose that consciousness is essentially a matter of having or being an awareness of a world that does not require a prior (representational) awareness of our own mental states (Thomasson 2008). Reflexivism or the thesis that self-​ consciousness consists in mental states being implicitly conscious of their occurrence thus serves as a grounding principle, enabling the intentional and subjective aspects of experience to emerge co-​constitutively in each instance of cognitive awareness. Can such reflexive self-​consciousness provide enough metaphysical grounding for thoughts of the sort ‘I am in pain’? Do such self-​ascriptions require independent criteria (or the normative, indexical, or ownership kind) for individuating streams of conscious episodes to manifest as occurring for me? Can conscious mental states occur without any sense of whose states they are? Can there be consciousness without self-​consciousness? If reflexive self-​consciousness has a distinctive character, a specific givenness or for-​me-​ness, then even misascriptions (as in the case of thought insertion or various pathologies of the self such as ego dissolution) would be unintelligible if thought were transparent with regard to its occurrent for-​me-​ness (Bermudez 1998; Zahavi & Kriegel 2015). Reflexivism, then, is simply a statement about the

Reasons and conscious persons  179

self-​intimating character of conscious mental states. It is the condition for the possibility of self-​knowledge, rather than the achievement of self-​knowledge. Because agency entails awareness of action and its consequences, persons (as self-​conscious agents) are ultimately real in a way that chariots and forests are not. Reflexivism allows for a minimal conception of subjectivity, specifically one that works against the idea that there could be experiences that are generic or impersonal until they are attended to in reflective or introspective thought. As such, it addresses not only the problem of self-​consciousness, but also the problem of subjectivity, of why it is that experiences present themselves as being not only about something, but for someone. It is only to the extent that experiences exhibit what has been called the ‘dative of manifestation’–​–​the fact that every experience is necessarily an experience for (or given to) someone (Prufer 1975)–​–​that we are in a position to understand what it is that makes us persons. We may conclude that what makes the further fact view of there being persons plausible is the structure of consciousness itself. This way of understanding persons is not unlike Locke’s view of persons as self-​conscious rational subjects. Neither a biological entity, nor an immaterial soul, a person is what it is by virtue of the fact that it, can consider it self as it self, the same thinking thing, in different times and places; which it does only by that consciousness, which is inseparable from thinking, and, as it seems to me essential to it: It being impossible for any one to perceive, without perceiving, that he does perceive. When we see, hear, smell, taste, feel, meditate, or will any thing, we know that we do so. … And as far as this consciousness can be extended backwards to any past Action or Thought, so far reaches the identity of that Person. Locke 1689 2.27.9, 335 The Buddhist Personalist may eschew the requirement that the self-​conscious subjectivity in play be manifest only in thought as rational deliberation and action, but agree that without self-​consciousness it is hard to understand how self-​concern, which is essential for moral cultivation, would be possible. For in looking to reward or punish persons for their deeds we do not look for a particular body, mental state or an immaterial substance of some sort, but rather “to connections between the consciousness of the one who did the deed and the one who is being held accountable” (Ainslie and Ware 2014: 248).29 Buddhist Personalists may also agree with Hume and the Parfitian version of the Reductionist View that our perceptions or sensible intuitions reveal no such thing as a fixed or stable self. Awareness of the inner contents of our mind lacks a necessary relation to a unified self. But while it may be true that we do not need to be self-​conscious in a higher order or metacognitive sense in order to perceive and think, the operations of perception and thought do presuppose that conscious experiences are unified in such a way that their contents can be synthesized. And it is the unity of conscious experience, achieved in the constitutive operations of the aggregates, that Buddhist Personalists point to when they claim that persons are ultimately real.

180  Christian Coseru

Conclusion Parfit’s motivation for rejecting the Non-​Reductionist View concerns the claim that there is more to being a person than what accounts of physical and psychological continuity can satisfactorily provide. The Non-​Reductionist makes the ultimately unsupported claim that “the unity of consciousness at any time is explained by the fact that several experiences are being had by a person” (1984: 275). Split brain cases and teleportation thought experiments, however, make it hard to explain unity by insisting on the importance of maintaining personal identity over time across different streams of consciousness. In conceiving of unity in terms of the ascription of different experiences to a particular subject, we are confronted with the unwarranted assumption that “there can be, in a person’s life, subjects of experiences that are not persons” (ibid.). Such cases are better explained, argues Parfit, by appealing to the Reductionist Psychology Criterion, which claims that “at any time, there is one state of awareness of the experiences in one stream of consciousness, and another state of awareness of the experiences in the other stream” (ibid.). In developing the criteria for psychological reductionism, Parfit’s theoretical perspective points in the direction of a formal but variant structure of awareness. This structure ought to, at least in principle, be able to accommodate continuity without the sameness or identity that ascribing such states to something like a Cartesian Ego would entail. It is unlikely that Buddha and most Buddhist Reductionists would have agreed with this part of his theory. Although it is in keeping with the principles of the Buddhist no-​self view, Parfit’s insistence on an external criterion for individuating or indexing awareness to a particular stream of consciousness may be at odds with what I  have argued are certain salient and ineliminable features of phenomenal consciousness, specifically its self-​reflexivity. If self-​reflexive conscious episodes lack an intentional structure, then they cannot provide the minimal sense of internal distance necessary for subjectivity. On the dual aspect model of consciousness sketched above (see also Coseru 2015, 66, 79), mental streams are differentiated by being covariant with intentional behavior, which presupposes that subjectivity and intentionality are structural features of consciousness, rather than constructs or relations among discrete mind moments (as presupposed by mereological nihilists). This chapter has considered whether Buddhist Reductionism provides support for Parfit’s psychological criterion for personal identity given considerations about the seemingly irreducible character of phenomenal consciousness. Taking Buddhist Personalism as a point of reference, I have argued that reductionism about selves can be reconciled with the seemingly irreducible character of self-​consciousness. Furthermore, Buddhist philosophers who follow in the footsteps of Dignāga and his successors provide a viable, if not altogether unproblematic, model for how this reconciliation might proceed. Whether or not this epistemological project is constrained by ontological assumptions about the constitutive elements of persons is open to debate.What is less controversial is that in rejecting a permanent locus for experience, the Buddha created an opportunity for the problem of personal identity

Reasons and conscious persons  181

to be explored not only on metaphysical but on empirical and phenomenological grounds as well. What sets apart non-reductive theories of persons as self-conscious subjects from their alternatives, then, is the notion that personal identity is not really a matter of the persistence or continuity of enough of the constitutive (physical and psychological) elements of persons, as demanded by the psychological criterion. Rather, on the model put forward here, it is a matter of understanding that the principle of individuation is neither external nor transcendent but internal to the structure of consciousness and its operations.

Notes 1 Within philosophy of mind, reductionism is concerned mainly with solving the mind-​ body problem, rather than, as was the case with Carnap (1934) and Neurath (1931), with the unity of scientific theories. Hence, the main motivation for reductionism with regard to questions of personal identity is ontological parsimony. 2 In the case of intertheoretic reduction, the question is whether the atomic and continuum scale models can be bridged.While some reductionists think that in principle it is possible to derive continuum scale entities from atomic scale details (Batterman 1995), others (in particular those who favor top-​down modeling) take the view that what we can hope for at most is identifying a set of intertheoretic relations (Berry 2001) rather than a hierarchy of theories operating at different scales. The upshot of this debate is that philosophical theories of reduction, which are typically modeled on scientific theories, must come to terms with the fact that reduction is problematic even in the case of the natural sciences. 3 One way to understand the difference between the non-​reductionist and the reductionist views of personal identity is along the simple/​complex divide:  the non-​reductionist favors the simple, soul or Cartesian Ego, view, whereas the reductionist prefers the complex view that entails relations among physical and psychological states. Holding a soul view, of course, does not necessarily amount to holding a brute fact view, although in the absence of non-​circular criteria for personal identity (of the sort required by the complex view) it is hard to tell them apart. What motivates recent defenders of the simple view (e.g., Baker 2013; Lowe 2009, 2013; Nida-​Rümelin 2013, Swinburne 2013) is not commitment to a Cartesian Ego, but rather the notion that a specific, perhaps non-​ conceptual and pre-​reflective, type of self-​awareness seems indispensable to framing any account of personal identity. I address this point below, in §4. 4 Parfit supports his claim regarding indeterminacy with the case of the Combined Spectrum, which, as the name suggests, involves all the possible variations of physical and psychological connectedness between such substantially altered person stages that the person I am now would in no way be connected with the one at the further end of the spectrum. 5 The first to draw attention to their similarity was Collins (1985), though in a subsequent article (Collins 1997) he softened his position, regarding the two views as similar rather than identical. Similar attempts to draw together the two views and to adopt a Parfitian conceptual schema in pursuing various projects in Buddhist philosophy are found in Bastow (1986), Stone (1988, 2005), Giles (1993), Duerlinger (1993), Basu (1997), Siderits (1997, 2003), Perrett (2002), Ganeri (2007), and Sauchelli (2016). 6 Parfit first articulated his views on personal identity as early as his 1971 paper “Personal Identity,” so that by the time Reasons and Persons was published, his views were widely known, and had already generated a great deal of discussion.

182  Christian Coseru

7 Parfit gets around the problematic issue of the relation between episodic and semantic memory by appealing to q-​memory (quasi memory–​–​a new category of memory first introduced by Shoemaker (1970)). According to Parfit, “I am q-​remembering an experience if (1) I have a belief about a past experience which seems in itself like a memory belief, (2) someone did have such an experience, and (3) my belief is dependent upon this experience in the same way in which a memory of an experience is dependent upon it” (1971: 15). 8 This text, along with Vasubandhu’s Treasury of Metaphysics (Abhidharmakośabhāṣya) and Buddhaghosa’s Path to Purity (Visuddhimagga), serve as his main source of Buddhist ideas about personal identity. Parfit includes several passages from these texts, excerpted from Stcherbatsky’s (1919) seminal work on the Buddhist no-​self doctrine. He also cites Collins’ (1982) landmark study of the early Buddhist doctrine of anatta. See also Collins (1997) for a response that considers the extent to which support for a reductionist account of personal identity on the Parfitian model can indeed be found in the Buddhist tradition. 9 For an overview of the impact that Parfit’s categorical framework and theoretical perspectives (on metaphysics and ethics) have had on Buddhist philosophy, see Hanner (2018). 10 All references are to the second edition, Siderits (2015). 11 Buddhist Personalists (Pudgalavāda) include the Vātsīputriyas and the Sāṃmitīyas, though their exact views on persons are hard to reconstruct since they are mostly preserved in unsympathetic treatments by their critics, mainly in the Points of Controversy (Kathāvatthu),Vasubandhu’s Treasury of Metaphysics (Abhidharmakośabhāṣya), Harivarman’s True Attainment Treatise (Satyasiddhiśāstra), Devasarman’s Body of Consciousness (Vijńānakāya), and Śāntarakṣita’s Compendium of Principles (Tattvasaṃgraha). The only source that presents their views unbiasedly is the Sāṃmitiya Nikāya Śāstra, now preserved only in Chinese (an English translation in Venkataramanan (1953)). The prevalent interpretation, gleaned from the Points of Controversy, is that Pudgalavādins take persons to be ultimately real. According to most of the other sources, Pudgalavādins think persons occupy an ontological position somewhere between conventional and ultimate reality. See Priestly (1999) for detailed studies of Buddhist Personalism, and also Châu (1999) for a general survey and reconstruction of their views. 12 The refutation, known as Treatise on the Negation of the Person, is appended as Chapter 9 of Vasubandhu’s Treasury of Metaphysics (Abhidharmakośabhāṣya). See Duerlinger (2003) for a detailed treatment of this section of Vasubandhu’s treatise. 13 For a response to the argument from overdetermination, see Sider (2003). 14 In particular, Crane (2010), Noordhof (2010), and Kistler (2010). 15 Proponents of this move include, inter alia, Siderits (1997, 2008), Flanagan (2002), and Meyers (2014). 16 See Caruso (2012), Smart (2006), and Wegner (2002) for various attempts to prove the illusory nature of experiences of mental causation. While not conclusive, Nahmias et al. (2004) review experimental data that seems to favor compatibilist over incompatibilist accounts of free will. 17 See Carpenter (2015) for a sympathetic treatment of Buddhist Personalism that proposes a new, constitutive or developmental model for understanding the Pudgalavāda, aimed at extending the categorical framework of Buddhist metaphysics. On this model, the aggregates (of perception, consciousness, etc.) relate to each other in a person-​constituting way, which she regards as “the most minimal explanation possible” that must be accepted if we are “to do justice to the phenomena without adopting a substantialist theory of self ” (2015: 27).

Reasons and conscious persons  183

18 A similar strategy is at work in Garrett (1998), though the focus there is on the connections between ‘I’-​judgments and self-​consciousness. Garrett finds Anscombe’s view that the first-​person singular pronoun is not a referring term unsupported, and builds a strong case for language use as an intentional activity: although not all language use exhibits intentional activity, self-​consciousness must be understood as consisting “in the presence of an intention to self-​refer” (Garrett 1998: 96). 19 See, e.g., Treasury of Metaphysics (Abhidharmakośabhāṣya) 2.23–​34 in de la Vallée Poussin (1980: 45–​62). 20 For Harivarnan’s discussion of some of the ethical implications of the no-​self view in Satyasiddhiśāstra, see Priestly (1999: 96). 21 Carpenter (2015:  23f) thinks the “belonging together” clause is not just a matter of spatiotemporal proximity, a feature shared by all aggregates that belong together, but something having to do with their functionality and development, even though in the end she admits that functionality alone is not sufficient to distinguish persons from chariots. 22 This is the strategy adopted by Harvey (1995). 23 Collins (1982) offers precisely such an analysis. 24 This position, first articulated by Albahari (2006), is developed at length by Ganeri (2017). 25 See, e.g., Compendium of Epistemology (Pramāṇasamuccaya) 1.10, in Hattori (1968: 29). See also Coseru (forthcoming) for a discussion of Dignāga’s theory relative to the problem of self-​knowledge. 26 This view is attributed to Asaṅga. As he observes in the Compendium of the Great Vehicle: How does one know that mind (manas) in the sense of ‘afflicted mind’ (kliṣṭa-​ manas) exists? Without it, there would be no uncompounded ignorance, that is, a basic ignorance not yet associated with all the faults (doṣa), but serving as their base (āśraya). Indeed, introspective awareness (mano-​vijñāna) must also have a simultaneous basis, as do the five types of empirical consciousness whose support are their material organs. Such a simultaneous support can only be found in the ‘afflicted mind.’ Mahāyāna-​samgraha 1.6–​7 in Lamotte 1938: 36 27 Micropsychism, roughly the view that human consciousness derives from a combination of micro-​conscious entities, might be taken as the exception here (see, e.g., Strawson 2006). But micropsychism faces the well-​known combination problem (Seager 1995):  how exactly do micro-​level entities with their own very basic forms of consciousness combine together to form human consciousness? In response to the combination problem, some defenders of panpsychism reject the atomistic metaphysics of micropsychism in favor of cosmopsychism, which holds that individual human consciousness is derived from, or an expression of, a single unified field of consciousness (Jaskolla & Buck 2012; Shani 2015, 2018; Nagasawa & Wager 2016; Goff 2017). Cosmopsychism faces the opposite problem, namely the decombination problem: how do we get from a unified field of consciousness to individual human and animal consciousness. 28 See Strawson (2015) for a defense of the self-​intimation thesis that draws together these major figures. 29 As Ainslie and Ware make quite clear, Locke not only defends the view, typically associated with Sartre, that consciousness is reflexive (that is, consciousness entails self-​ consciousness), but also (and more controversially) that consciousness is not “limited simply to our occurrent mental states” but rather extends its contents from one conscious moment to the next (2014: 249).

184  Christian Coseru

References Ainslie, D. C. and Ware, O. 2014. “Consciousness and personal identity”. In A. Garrett (ed.), The Routledge Companion to Eighteenth Century Philosophy. Routledge, 245–​264. Albahari, M. 2006. Analytical Buddhism: The Two-​Tiered Illusion of Self. Hampshire: Palgrave Macmillan. Baker, L. R. 2013. “Personal identity: a not-​so-​simple view”. In G. Gasser and M. Stefan, Personal Identity: Complex or Simple. Cambridge: Cambridge University Press, 179–​191. Bastow, D. 1986. “Self-​construction in Buddhism”. Ratio 28(2): 97–​113. Basu, A. 1997. “Reducing concern with self: Parfit and the ancient Buddhist schools”. In D. Allen (ed.), Culture and Self: Philosophical and Religious Perspectives, East and West, 97–​109. Boulder, CO: Westview Press. Batterman, R. W. 1995. “Theories between theories:  Asymptotic limiting intertheoretic relations”. Synthese 103: 171–​201. Bermudez, J. L. 1998. The Paradox of Self-​Consciousness. Cambridge, MA: MIT Press. Berry, M. W. 2001. “Chaos and the semiclassical limit of quantum mechanics (is the moon there when somebody looks?)”. In Quantum Mechanics:  Scientific Perspectives on Divine Action, R. J. Russell, P. Clayton, K. Wegter-​McNelly and J. Polkinghorne (eds.), Vatican Observatory CTNS Publications, 41–​54. Carnap, R. 1934. The Unity of Science. London: Kegan Paul, Trench, Trubner, and Co. Carpenter, A. 2015. “Persons keeping their karma together: the reasons for the Pudgalavāda in early Buddhism”. In K. Tanaka, Y. Deguchi, J. Garfield and G. Priest (eds.) The Moon Points Back. New York: Oxford University Press, 1–​44. Caruso, G. D. 2012. Free Will and Consciousness: A Determinist Account of the Illusion of Free Will. Lanham: Lexington Books. Cassam, Q. 1995. “Introspection and bodily self-​ascription”. In J. L. Bermudez, A. Marcel and N. Eilan (eds.) The Body and the Self. Cambridge, MA: MIT Press. Châu, T. T. 1999. The Literature of the Personalists of Early Buddhism. Translated by S. Boin-​ Webb. Delhi: Motilal Banarsidass. Collins, S. 1982. Selfless Persons:  Imagery and Thought in Theravāda Buddhism. Cambridge: Cambridge University Press. Collins, S. 1997. “A Buddhist debate about the self; and remarks on Buddhism in the work of Derek Parfit and Galen Strawson”. Journal of Indian Philosophy 25(5): 467–​493. Coseru, C. 2012. Perceiving Reality:  Consciousness, Intentionality, and Cognition in Buddhist Philosophy. New York: Oxford University Press. Coseru, C. 2015. “Perception, causally efficacious particulars, and the range of phenomenal consciousness”. Journal of Consciousness Studies 22(9–​10): 55–​82. Coseru, C. 2016. “Freedom from responsibility:  agent-​neutral consequentialism and the Bodhisattva ideal”. In R. Repetti Buddhist Perspectives on Free Will:  Agentless Agency? New York: Routledge, 92–​105. Coseru, C. (forthcoming). “Whose consciousness?  reflexivity and the problem of self-​ knowledge”. In M. Siderits, K. Ching and J. Spackman Buddhist Philosophy of Consciousness. Crane, T. 2010. “Cosmic hermeneutics vs emergence: the challenge of the explanatory gap”. In G. Macdonald and C. Macdonald (eds.), Emergence in Mind. Oxford: Oxford University Press,  22–​34. de la Vallee Poussin, L. 1980. L’Abhidharmakośa de Vasubandhu.Vol. 1. Bruxelles: Institut Belge des Hautes Études Chinoises. Duerlinger, J. 1993.“Reductionist and nonreductionist theories of persons in Indian Buddhist philosophy”. Journal of Indian Philosophy 21(1): 79–​101. Duerlinger, J. 2003. Indian Buddhist Theories of Persons: Vasuabndhu’s “Refutation of the Theory of a Self”. London: Routledge Curzon.

Reasons and conscious persons  185

Flanagan, O. 2002. The Problem of the Soul: Two Visions of Mind and How to Reconcile Them. New York: Basic Books. Ganeri, J. 2007. The Concealed Art of the Soul: Theories of Self and Practices of Truth in Indian Ethics and Epistemology. Oxford: Oxford University Press. Ganeri, J. 2017. Attention, Not Self. Oxford: Oxford University Press. Garrett, B. 1998. Personal Identity and Self-​Consciousness. London: Routledge. Giles, J. 1993. “The no-​self theory: Hume, Buddhism, and personal identity”. Philosophy East and West, 43(2): 175–​200. Goff, P. 2017. Consciousness and Fundamental Reality. Oxford: Oxford University Press. Hanner, O. 2018. “Buddhism as reductionism:  personal identity and ethics in Parfitian readings of Buddhist philosophy; from Steven Collins to the present”. Sophia 57: 211–​231. Harvey, P. 1995. The Selfless Mind: Personality, Consciousness and Nirvāṇa in Early Buddhism. Richmond: Curzon Press. Hattori, M. 1968. Dignāga, On Perception. Cambridge, MA: Harvard University Press. Horner, I. B. 1963. Milinda’s Questions. Vol. I. London: Luzac & Company. Jaskolla, L. J. and Buck, A. J. 2012. “Does panexperientialism solve the combination problem”. Journal of Consciousness Studies, 19(9–​10): 190–​199. Kirchin, S. (ed.) 2017. Reading Parfit. London: Routledge. Kistler, M. 2010.“Strong emergence and freedom–​–c​ omment on A. Stephan”. In G. Macdonald and C. Macdonald (eds.), Emergence in Mind. Oxford: Oxford University Press, 240–​251. Lamotte, E. 1938. La Somme du Grand Véhicule d’Asaṅga (Mahāyānasaṃgraha). Louvain-​la-​ Neuve: Université de Louvain. Locke, J. 1689/​1975. An Essay Concerning Human Understanding. P. Nidditch (ed.). Oxford: Oxford University Press. Lowe, E. J. 2013. “The probable simplicity of personal identity”. In G. Gasser and M. Stefan, Personal Identity: Complex or Simple. Cambridge: Cambridge University Press, 137–​155. Merricks, T. 2001. Objects and Persons. Oxford: Clarendon Press. Meyers, K. 2014. “Free persons, empty selves: freedom and agency in light of the two truths”. In M. Dasti and E. F. Bryant (eds.) Free Will, Agency, and Selfhood in Indian Philosophy. New York: Oxford University Press, 41–​67. Nagasawa,Y. and Wager, K. 2016. “Panpsychism and priority cosmopsychism”. In P. Brüntrup and R. Jaskolla (eds.), Panpsychism. New York: Oxford University Press, 113–​129. Nahmias, E., Morris, S., Nadelhoffer, T. and Turner, J. 2004. “The phenomenology of free will”. Journal of Consciousness Studies, 11(7–​8): 162–​179. Neurath, O. 1932/​1983.“Sociology in the framework of physicalism’. In Neurath, M. Philosophical Papers 1913–​1946, R. S. Cohen and M. Neurath (eds.), Dordrecht: Reidel, 58–​90. Nida-​Rümelin, M. 2013. “The non-​descriptive individual nature of conscious beings”. In G. Gasser and M. Stefan, Personal Identity:  Complex or Simple. Cambridge:  Cambridge University Press, 157–​176. Noordhof, P. 2010. “Emergent causation and property causation”. In G. Macdonald and C. Macdonald (eds.), Emergence in Mind. Oxford: Oxford University Press, 69–​98. Parfit, D. 1971. “Personal identity”. The Philosophical Review, 80(1): 3–​27. Parfit, D. 1984/​87. Reasons and Persons. Oxford: Clarendon Press. Parfit, D. 1999. “Experiences, subjects, and conceptual schemes”. Philosophical Topics 26(1–​2): 217–​270. Parfit, D. 2002. “Reductionism and personal identity”. In D. J. Chalmers (ed.), Philosophy of Mind: Contemporary Readings. Oxford: Oxford University Press, 655–​661. Perrett, R. 2002. “Personal identity, minimalism, and Madhyamaka”. Philosophy East and West 52(3): 373–​385. Priestley, L. 1999. Pudgalavāda Buddhism: The Reality of the Indeterminate Self. Toronto: Centre for South Asian Studies, University of Toronto.

186  Christian Coseru

Prufer, T. 1975. “An outline of some Husserlian distinctions and strategies, especially in the Crisis”. Phänomenologische Forschungen 1: 189–​204. Sauchelli, A. 2016. “Buddhist reductionism, fictionalism about the self, and Buddhist fictionalism”. Philosophy East and West 66(4): 1273–​1291. Shani, I. 2015. “Cosmopsychism:  a holistic approach to the metaphysics of experience”. Philosophical Papers 44(3): 389–​417. Shani, I. 2018. “Beyond combination: how cosmic consciousness grounds ordinary experience”. Journal of the American Philosophical Association, 16(41): 390–​410. Shoemaker, S. 1970. “Persons and their pasts”. American Philosophical Quarterly 7(4): 269–​285. Sider, T. 2003. “What’s so bad about overdetermination?” Philosophy and Phenomenological Research 67(3): 719–​726. Siderits, M. 1987. “Beyond compatibilism:  a Buddhist approach to freedom and determinism”. American Philosophical Quarterly 24(2): 149–​159. Siderits, M. 1997. “Buddhist reductionism”. Philosophy East and West 47(4): 455–​478. Siderits, M. 2008. “Paleo-​compatibilism and Buddhist reductionism”. Sophia, 47: 29–​42. Siderits, M. 2015. Personal Identity and Buddhist Philosophy:  Empty Persons (2nd ed.). Aldershot: Ashgate. Smart, J. J.  C. 2006. “Metaphysical illusions”. Australasian Journal of Philosophy 84(2): 167–​175. Stcherbatsky, T. 1919. “The soul theory of the Buddhists”. Bulletin de l’Academie des Sciences de Russie, 13 : 823–​854; 937–​958. Stone, J. 1988. “Parfit and the Buddha:  Why there are no people”. Philosophy and Phenomenological Research, 48: 519–​532. Stone, J. 2005. “Why there still are no people.” Philosophy and Phenomenological Research 70: 174–​192. Strawson, P. F. 1992. “Freedom and necessity”. In Analysis and Metaphysics. Oxford: Oxford University Press. Strawson, G. 2006. “Realistic monism:  why physicalism entails panpsychism”, Journal of Consciousness Studies 13(10 –​11): 3–​31. Strawson, G. 2015. “Self-​intimation”. Phenomenology and the Cognitive Sciences 14: 1–​31. Swinburne, R. 2013. “How to determine which is the true theory of personal identity”, in G. Gasser and M. Stefan, Personal Identity:  Complex or Simple. Cambridge:  Cambridge University Press, 105–​122. Thomasson, A. L. 2007. Ordinary Objects. Oxford: Oxford University Press. Thomasson, A. L. 2008. “Phenomenal consciousness and the phenomenal world”. The Monist 91(2): 191–​214. Venkataramanan, K. 1953. Sāṃmitīyanikāya Śāstra. Visva-​Bharati Annals, 5: 155–​243. Wegner, D. 2002. The Illusion of Conscious Will. Cambridge, MA: MIT Press. Wittgenstein, L. 1973. Philosophical Investigation. 3rd ed., (tr.) G. E.  M. Anscombe, Oxford: Blackwell. Zahavi, D. 2005. Subjectivity and Selfhood: Investigating the Fist-​Person Perspective. Cambridge, MA: MIT Press, 2005. Zahavi, D. and Kriegel, U. 2015. “For-​me-​ness:  What it is and what it is not”. In D. Dahlstrom, A. Elpidorou and W. Hopp (eds.), Philosophy of Mind and Phenomenology. London: Routledge,  36–​53.

9 TRANSFORMATIVE CHOICE AND THE NON-​IDENTITY PROBLEM Nilanjan Das and L. A. Paul

Some of our acts change who we are. Call them personally transformative acts. When an agent performs a personally transformative act, she brings into existence a future self that is radically different from who she previously was. In some of these cases, the agent may be antecedently certain that the existence of this future self, though worth having, will be unavoidably flawed, even if the future self values its existence. If the agent doesn’t perform the transformative act, she won’t change so radically, so her unchanged self may indeed be better off in the future than her transformed future self. In this chapter, we argue that situations of this kind raise a problem that is structurally similar to the non-​identity problem.1 Here is the plan for this chapter. We begin by explaining what we mean by personally transformative acts (section 9.1).We next argue that transformative acts involve the creation of selves that wouldn’t otherwise exist (section 9.2). Then, we show how two plausible principles about prudence raise a problem analogous to the non-​identity problem with respect to transformative acts (section 9.3). Finally, we consider some responses to this problem (section 9.4).2

9.1  Transformative acts In recent work on transformative experience, L. A. Paul (2014) distinguishes two kinds of transformative experience. Some experiences are epistemically transformative:  before undergoing those experiences, the agent doesn’t know what it would be like to undergo those experiences. Imagine a person who acquires a new sense-​modality, e.g., someone who has been blind since birth, but gains ordinary vision. Such a person could now encounter qualities, e.g., luminance and colour, which she didn’t see before,3 and didn’t know what it would be like to encounter those qualities before she underwent that experience.4

188  Nilanjan Das and L. A. Paul

Some experiences are also personally transformative: they change the core beliefs, desires, character traits, and other mental states that determine who the agent is. Examples include: “experiencing a horrific physical attack, gaining a new sensory ability, having a traumatic accident, undergoing major surgery, winning an Olympic gold medal, participating in a revolution, having a religious conversion, having a child, experiencing the death of a parent, making a major scientific discovery, or experiencing the death of a child” (Paul 2015, p. 16). There are two ways of framing such personal transformations. One could take a third-​person approach to them, like Edna Ullmann-​Margalit (2006). She considers situations where an agent must decide amongst certain options, at least one of which, by her own lights, will change her personality, and calls decisions made in such scenarios opting. As Ullmann-​Margalit describes it, if the agent goes for the personality-​changing option in such scenarios, a New Person comes into existence. She says: New Person is now, by hypothesis, a transformed person. Opting transforms the sets of one’s core beliefs and desires. A significant personality shift takes place in our opter, a shift that alters his cognitive as well as evaluative systems. New Person’s new sets of beliefs and desires may well be internally consistent but the point about the transformation is that inconsistency now exists between New Person’s system of beliefs and desires, taken as a whole, and Old Person’s system taken as a whole. Ullmann-​Margalit 2006, p. 167 She goes on to illustrate the idea with the following example: I was told of a person who hesitated to have children because he did not want to become the ‘boring type’ that all his friends became after they had children. Finally, he did decide to have a child and, with time, he did adopt the boring characteristics of his parent friends—​but he was happy! I suppose second order preferences are crucial to the way we are to make sense of this story. As Old Person, he did not approve of the personality he knew he would become if he has children: his preferences were not to have New Person’s preferences. As New Person, however, not only did he acquire the predicted new set of preferences, he also seems to have approved of himself having them. Ibid. In this case, New Person differs from Old Person not only with respect to her first-​ order preferences about what to do in particular situations (e.g., about whether to stay out late or come home early), but also with respect to her second-​order preferences about which first-​order preferences to have. When an agent undergoes a personal transformation, both her first-​and second-​order preferences change in

Transformative choice  189

tandem. This is a purely third-​person way of characterizing personally transformative experiences. In contrast, Paul characterizes personally transformative experiences by highlighting first-​personally accessible changes. One of her examples is this. Imagine that you have the chance to become a vampire. With one swift, painless bite, you’ll be permanently transformed into an elegant and fabulous creature of the night. As a member of the undead, your life will be completely different. You’ll experience a range of intense, revelatory new sense experiences, you’ll gain immortal strength, speed and power, and you’ll look fantastic in everything you wear. You’ll also need to drink blood and avoid sunlight. Paul 2014, p. 1 On this view, when you become a vampire, the phenomenal character of your lived experiences, i.e., what it’s like for you to undergo those experiences, will change. These changes may be connected to the changes in one’s preferences: assuming that our preferences are sensitive to some extent to the phenomenal character of our experiences, radical changes in the phenomenal character of certain experiences could affect the preferences of the agent. In Paul’s example, you can know that your preferences will change, but you don’t have first person access to the way they’ll change. This affects the way you can think about the choice.You can know on the basis of the testimony of others how your preferences will change when you become a vampire. For instance, everyone you know may have already become a vampire, and may tell you that they love it. They might even tell you that if they were offered a pill to become human again, they would reject the offer without a moment’s thought. Despite having all this information, you still won’t know what it is like to be a vampire, since you can only come to know what it’s like to be a vampire by becoming one. Thus, since you lack access to the phenomenal character of a vampire’s lived experiences, you lack access to certain concepts, pieces of information, or cognitive skills necessary for understanding why the preferences that come with vampirehood are rational to maintain (or, at least make sense to have) for someone who is a vampire. The case is intended to illustrate ways in which radical changes in phenomenal character can lead to changes in the very core of our personality: since the beliefs, desires, and commitments that make us who we are can be shaped by what it is like to be us, certain types of radical changes in the phenomenal character of our lived experiences can personally transform us. Our focus in this chapter is on this latter point: on the way in which radical changes in our lived experience can lead to changes in who we are. Let an epistemically transformative act be an act that brings about an epistemically transformative experience. Let a personally transformative act be an act that brings about a personally transformative experience. Two features of personally transformative acts are worth getting clear on.

190  Nilanjan Das and L. A. Paul

First of all, acts that are epistemically and personally transformative raise problems for theories of practical rationality. Those problems are not our concern here. In what follows, we focus exclusively on personally transformative acts, without addressing the question of whether or how they are also epistemically transformative acts, or how epistemic transformation relates to practical decisionmaking. Second, some personally transformative acts might be preceded by a moment of conscious choice where the relevant agent decides to perform that act. But this may not always be the case. In particular, some transformative acts are not the product of conscious choice. Or, as Callard (2018) correctly observes, sometimes, we gradually transform our personalities without ever directly deciding to undergo that transformation. Yet, that doesn’t mean that we are thoughtlessly drifting into these new personalities. Rather, in many of these cases –​e.g., in a case where one becomes a mother, an artist, a lover of classical music etc. –​ one actively tries on new roles and activities which enable one to navigate the world using a new set of beliefs and preferences, or to experience the world in a new way. On Callard’s account, we can aspire to rational self-​transformation by aiming at it indirectly, guided by proleptic reasons. Our notion of a personally transformative act covers these kinds of cases (unchosen, unconsciously chosen, aspirational). Since the agent could still have preferences about such acts (whether or not they guide her actions), her act of undertaking the project counts as a personally transformative act, even if the agent never explicitly decided to perform a personally transformative act.

9.2  Self-​creation In this section, we defend the following claim. The Principle of Act-​ Dependence. If an agent performs a personally transformative act, then (keeping all else fixed), she thereby creates a new self that wouldn’t exist if she hadn’t performed that transformative act. To understand what this means, start with the notion of self that the principle invokes. For the purposes of this chapter, we’ll assume that an agent’s self at a certain time is constituted by the core values, beliefs, desires, commitments, ideals, character traits, etc., that make her who she is at that time, including the first person phenomenology that is realized by having these core values, etc. Now, a change in these needn’t be so dramatic that we are no longer inclined to think that the person who existed before the change exists any more. For instance, Sue may take a college course that changes her worldview radically, thus changing her plans about what to do with her life. But that needn’t mean that the person that Sue was before taking the college course has ceased to exist. Plausibly, that person still exists, but she’s no longer exactly who she was earlier. On this way of thinking, a person’s life may be partitioned into intervals, corresponding to a series of successive selves, which in

Transformative choice  191

turn are suitably extended (collections of) temporal parts. Here’s a more careful definition. Self. For any person S, a temporal part x of S that exists during an interval of time t counts as a self iff (i) Prudential Status. x has prudential status, i.e., it or its well-​being can be an object of prudential concern for that and other temporal parts of S5; (ii) Constancy. S’s self-​identity-​conferring mental states remain unchanged during t, i.e., during any two sub-​intervals t1 and t2 of t, the mental states that make S who she is at t1 are the same as those that make her who she is at t2. (iii) Maximality. There is no interval of time t* during which S exists, such that (a) t* is distinct from t, (b) t is a sub-​interval of t*, and (c) S’s self-​identity-​ conferring mental states remain (relevantly) unchanged during t*.6 Let’s explore this a bit. Start with the idea of a temporal part of a person. If Sue is a person who is now in her 40s, Sue-​in-​her-​20s is a temporal part of Sue. Why should we think that persons have such temporal parts? It follows from a view about persistence. According to endurantists, objects persist through time by being wholly present at each moment at which they exist. According to perdurantists, they persist by having temporal parts that are wholly present at each moment at which they exist. More precisely, let an object x be an instantaneous temporal part of an object y at t iff (i) x exists at, but only at, t; (ii) x is part of y at t; (iii) x overlaps at t everything that is part of y at t.7 Perdurantists just say that for any object x to persist, if t is a time at which x exists, there is an instantaneous temporal part of x at t. Endurantists deny this. To take a concrete case, suppose your desk has persisted through time. Perdurantists would say that it has done so by having instantaneous temporal parts at every moment at which it has existed. According to them, however, the desk was never wholly present at any of those moments. Endurantists would say that it persisted by being wholly present at every moment at which it has existed. Similarly, if Sue has persisted through time, then, according to perdurantists, she too has done so by having instantaneous temporal parts at every moment at which she has existed; endurantists would deny this. Perdurantism is preferable to endurantism as an account of objective persistence; for the former allows us to respond to certain puzzles satisfactorily.8 (For an account of the subjective persistence of selves that is consistent with this approach, see Paul 2017.) Typically, perdurantists claim that persons and material objects do not just have instantaneous temporal parts, but also have temporal parts that themselves are extended temporally.The most liberal version of this view would be one according to which a person or a material object has temporal parts corresponding to any period of time at which she exists.This would mean that just as Sue-​in-​her-​20s counts as a temporal part of Sue, so does Sue-​in-​October-​2018 or Sue-​at-​this-​moment. For the purposes of this chapter, we’ll take a temporal part of a person to count as a self just in case it satisfies three constraints. The first is Prudential Status: a self or its well-​being can be an object of prudential concern for that person (or other

192  Nilanjan Das and L. A. Paul

temporal parts of that person). This seems plausible:  the well-​being of different temporal parts of a person can be the object of prudential concern for that person. In fact, some people might say something even stronger: namely, that all instantaneous temporal parts of a person ought to be objects of equal prudential concern for that person.9 The second feature is Constancy: if t is the period of time during which a self exists, the agent’s identity-​conferring mental states—​the mental states that make her who she is—​cannot change.We may imagine a fictional case, where a person changes her mental states at t1 so that, immediately after t1, she is no longer who she was before t1, but then at t2, she reverts back to her old mental states before t1. In such cases, the same self doesn’t persist through the continuous interval of time that includes the time before t1 as well as the time after t2.The third feature is Maximality: a self is the largest temporal part of a person unified by sameness with respect to who she is. Here’s a way of illustrating this. Suppose Sue doesn’t undergo any personal transformation in her 30s: who she is at 31 is no different from who she is at 32, or 33, and so on. According to Maximality, this means that Sue-​at-​31, Sue-​at-​32, Sue-​at-​ 33, etc. are parts of the same self, but aren’t selves in their own right. This is because Sue-​at-​31 exists during an interval of time that has a non-​empty intersection with a larger interval of time, namely the period during which Sue is in her 30s exists, and Sue’s self-​identity during her 30s isn’t any different from her self-​identity at 31. How does this help us with the Principle of Act Dependence? We’ll assume that a person S at two times t and t* is realized by the same, temporally extended self if and only if the self that realizes her at t is no different from the self that realizes her at t*. We are interested in cases where, in virtue of S performing a transformative act at t, a new self comes into existence at t*. Such cases are those where, when we hold relevant background conditions fixed,10 the following counterfactual is true:  if, at time t, S hadn’t performed that transformative act, then the new self would not have been created at t*. (And, by extension, we assume that if the act had not been performed, the same self that realizes S at t* would persist from t to t*.) So we are focusing on cases for which the Principle of Act-​Dependence holds.

9.3  The non-​identity problem for transformative acts Parfit’s (1987) version of the non-​identity problem is motivated by two plausible claims. The Person-​Affecting Principle. An act can be morally wrong only if it makes things worse for some existing or future person. The Comparative Notion of Harm for Persons. Suppose an act or an event brings a person into existence such that (i) the person wouldn’t have existed in the absence of the act, (ii) the person’s existence is avoidably flawed, and (iii) the person’s existence is still worth having for the person. Then, this act or event does not make things worse for that person.11

Transformative choice  193

We can apply these two principles to the following scenario. The 14-​Year-​Old Girl. This girl chooses to have a child. Because she is so young, she gives her child a bad start in life. Though this will have bad effects throughout this child’s life, his life will, predictably, be worth living. If this girl had waited for several years, she would have had a different child, to whom she would have given a better start in life. Parfit 1987, p. 358 The intuition is supposed to be that the girl makes the morally wrong decision. How do we explain this? According to the Person-​Affecting Principle, this can be true only if her act makes things worse for, or harms, the child. According to the Comparative Notion of Harm, the girl’s act doesn’t make things worse for, or harm, the child. For, if the girl had waited, a different child would be born, and this child’s life (despite its flaws) is still worth living. So, either we must reject the intuition that the girl does something wrong, or we must give up the Comparative Notion of Harm or the Person-​Affecting Principle. We can create a similar problem with respect to transformative acts.

A.  Two principles Once again, we start with two principles. The first is an analogue of the Person-​ Affecting Principle. The Self-​Affecting Principle. An agent can rationally prefer not to perform an available act A only if there exist an available act B and a current or possible future self x of the agent such that the expected well-​being of x conditional on the agent’s performing B is greater than the expected well-​being of x conditional on the agent’s performing A.12 Consider the following pair of cases. Surgery I. On Monday, you have to schedule a surgery to have your wisdom teeth removed, painfully but safely, under a weak local anaesthetic.You are certain about the following facts. There are two surgeons who could perform the surgery.The surgery will begin exactly at the same time no matter who performs it. But the first surgeon will take more time to perform the operation, so you will be in pain for a longer period of time.Whom will you pick as your surgeon? Surgery II. On Monday, you have to schedule a surgery to have your wisdom teeth removed, painfully but safely, under a weak local anaesthetic. You are certain about the following facts. There are two surgeons who could perform the surgery. The surgery will begin exactly at the same time no matter who performs it. Moreover, both surgeons will take exactly the same amount of time to perform the operation. Whom will you pick as your surgeon?

194  Nilanjan Das and L. A. Paul

If you are rational, in Surgery I you will prefer not to pick the first surgeon, but in Surgery II you will remain indifferent between the two. The Self-​Affecting Principle explains why. Since you know that the first surgeon in Surgery I will subject you to pain for a longer period of time, you know that in Surgery I that there is at least one (existing or future) self that would be made worse off if you picked the first surgeon. By contrast, you know that in Surgery II things won’t be better or worse for your present or future selves in either of these scenarios. That is why it is irrational for you to prefer one surgeon to another. The second principle is an analogue of the Comparative Notion of Harm for Persons. The Comparative Notion of Harm for Selves. Suppose an act or event results in the existence of a future self such that (i) the future self wouldn’t have existed in the absence of the act or event, (ii) the future self ’s existence is unavoidably flawed, and (iii) the future self ’s existence is still worth having for that future self. Then, this act or event does not make things worse for that future self. Here is a way of motivating this principle. Many disabled people report that from their own first-​personal standpoint, they don’t experience their disability as a harm, since (a) despite its challenges, disability doesn’t make their lives unworthy of having, and (b) if they didn’t have that disability, they wouldn’t be who they actually are.13 Here is a revealing passage from Emily Ladau who writes: I can’t count how many times I’ve been asked variations of the question: “If there was a pill that could cure your disability, would you take it?” Though the short answer is a resounding “No!” I rarely get the chance to elaborate on the complex feelings and emotions that are behind my answer. I think “cure” is actually a rather loaded term in relation to my disability, because to cure something implies that you are returning the body to its normal state. My disability is my normal state.To cure me in accordance with the medical definition of the word would not only give me new abilities, but also essentially transform me into a whole new person. I can’t imagine myself as an able-​bodied person, because I  never was an able-​bodied person. I’ve embraced my disability as a huge facet of my identity, and I take pride in it. While I don’t define myself solely by my disability, having a disability has undeniably shaped who I  am. Without my lived experiences as a disabled person, I  would be a completely different Emily. And as tough as certain aspects of my life have been, and though I know I will continue to face disability-​related challenges throughout my life, I  wouldn’t trade my life for a minute. My disability has given me a place in a community and a culture; it has been the reason why I’ve had amazing adventures and unforgettable experiences. To walk freely up and down stairs for one day would never measure up to the things I’ve done because I have a disability. Ladau 2013

Transformative choice  195

Building on this idea, we can create a case like this. Disability I. Suppose you went blind as a child because of an accident. Since then, your blindness has given you the benefit of certain valuable experiences that a sighted person wouldn’t be able to have. For example, you hear and feel things that sighted individuals fail to notice.You experience the world differently and discover new ways to relate to your environment. It has also allowed you to become part of a community and a culture, which sighted people do not have access to. And you know that if the accident hadn’t happened and you retained your sight, you wouldn’t have enjoyed these experiences, or formed the close relationships with other members of the blind community, which have shaped the way you see yourself and have enriched your life. Thus, you have come to see your own blindness as a blessing in disguise. In this case, as Ladau puts it, perhaps you “wouldn’t trade” your life for the life of your non-​disabled counterpart who somehow avoided the accident.That is, the self that you are would not trade itself in for a new self. From your (self-​ish) perspective, your disabled life –​despite its challenges –​is fulfilling and is not to be traded for a scenario where you don’t exist at all. Unless we want to treat the selves realized by disabled people like Ladau as irrational, we should grant that this preference is rational. Moreover, as Barnes (2016) convincingly argues, we have very little reason to think that this is just some sort of adaptive preference or self-​deception that doesn’t reflect actual quality of life. So, assuming that your preference does provide evidence for your well-​being, we should conclude that your life as a disabled self is clearly worth living. Moreover, since you see your disability as a blessing in disguise and therefore don’t experience it as a harm, this suggests that the self you are now wasn’t harmed by the accident.14 This lends plausibility to the Comparative Notion of Harm for Selves.

B.  The Problem Consider a slightly different scenario. Disability II. You learn from your doctor that you will become blind soon if you don’t undergo cataract surgery. If you were to become blind, there wouldn’t be any serious transition costs: fortunately, your family is extremely supportive, and you live in a society that doesn’t treat blind people all that differently from the sighted. However, you will inescapably lose certain capacities: you won’t able to read very many books, you won’t be able to paint, and you won’t be able to play the sport you love the most. At the same time, your blindness will allow you to have certain experiences that a sighted person couldn’t have. And it will give you a place within a community and a culture that sighted people don’t have access to. Finally, you are certain that your blindness will eventually change how you see yourself: your identity will at least be partly shaped by your blindness, and you’ll be glad that you became blind.

196  Nilanjan Das and L. A. Paul

If you choose not to have cataract surgery, you will be performing a transformative act: you will be radically changing who you are. According to the Principle of Act-​Dependence, then, your decision to go blind will bring into existence a future self of yours that wouldn’t exist if you chose to have the surgery. From the perspective of your current self, the lived experience of this future self is unavoidably flawed.Yet, at the same time, there is no reason to think that the life your blind self would have wouldn’t be worth having. Here is the problem. On the one hand, it seems tempting in this case to say that you can rationally prefer the act of undergoing cataract surgery to doing nothing. You know that you won’t be able to do lots of things that you currently care about, enjoy, and excel at: your ability to play certain sports, to paint, to read a lot of books, etc. On the other hand, this claim is incompatible with the conjunction of the Self-​ Affecting Principle for Selves and the Comparative Notion of Harm for Selves. According to the Self-​Affecting Principle, you can rationally prefer not to perform an act A only if there is another available act that is expected to make things one of your present or future selves better than A does. This means that if an act maximizes the well-​being of your current and future selves, it is rationally impermissible for you not to prefer that act. But in this case, your blind future self wouldn’t exist if you underwent cataract surgery. Since that self ’s existence is worth having, according to the Comparative Notion of Harm for Selves, not having surgery does not harm your blind future self or make it worse off. Therefore, it is indeed irrational for you to prefer undergoing the surgery to not undergoing the surgery; in fact, you should be indifferent between the two options. The upshot is this. If we accept the Principle of Act-​Dependence, either we have to reject the intuition that it is practically rational for you to prefer to undergo the surgery in this case, or we have to give up one of the two principles we introduced (i.e., the Self-​Affecting Principle and the Comparative Notion of Harm for Selves). In this respect, the problem has a structure that is exactly analogous to the non-​identity problem. It’s worth noticing how this problem of transformative choice is different from other problems raised by scenarios of transformative choice. Two such problems ought to be mentioned. First of all, Paul (2014) argues that epistemically transformative experiences reveal a problem for standard decision theory. Consider an epistemically transformative experience like tasting the durian fruit for the first time. If an agent is given the choice of undergoing such an experience, can it be practically rational for her to take it? Paul’s claim is that since an agent cannot know what that experience will be like before undergoing the experience itself, she cannot accurately assign subjective value to the possibility where she undergoes that experience. But standard decision theory—​which requires us to maximize expected value of some kind—​can help us generate an ordering of preferences over options only if we can assign values to the different outcomes of taking each option. So, standard decision theory is silent about cases like this.15

Transformative choice  197

Personally transformative experiences, e.g., the experience of having a child, give rise to a different problem. In practical decision-​making contexts, they can change the relevant agent’s core values and preferences in unforeseeable ways. Writing about the transformative choice of becoming a vampire, Paul (2014) says: “So until you actually become a vampire, you cannot know if the values of any of the relevant outcomes will swamp the rest, or how to compare the subjective value of being a vampire to the subjective value of being human, or which preferences about the outcomes that you’d have as a vampire will be the same as the preferences you have as a human’’ (p. 44). The idea is this: when an agent makes a decision, she makes a decision on behalf of her current as well as her future selves. But if the agent can’t accurately forecast the preferences of her future selves (after performing an act), she cannot take those preferences into account while making her decision.16 The problem we are concerned with doesn’t depend on the agent’s prior ignorance about what experiences she will undergo as a result of her transformative act, or what her future preferences or values will be. As we noted in section 9.1, our focus is on a different issue: the way that radical changes in lived experience can create a new self. It’s a problem simply about which selves will exist as a result of an agent’s transformative act. For instance, in Disability II, we can assume that the agent somehow (by some feat of imagination or a blindness simulator) has full access to what it will be like to be her blind self, or what the values or preferences of that self will be. Still, the problem we have raised will persist. As long as the current self believes that the existence of her blind self is unavoidably flawed due to the lack of certain abilities that only the sighted possess, it will be rational for her current self to prefer undergoing the cataract surgery to not undergoing the surgery. And yet, given that her future blind self ’s existence is worth having, it will be hard to say that not undergoing the cataract surgery was bad for, or harmed, her blind self.

9.4  Possible responses Let us consider some responses to the problem posed in the last section.17

A.  Denying the intuition The first response is to reject the intuition that it is rational for the agent to prefer that she not go blind. If you aren’t convinced by the somewhat natural example described earlier, consider a variant of the same case. Disability III. Your ophthalmologist gives you two options. She could either painlessly blind you, or you can retain your imperfect, but fairly well-​ functioning, vision. The rest remains the same as in Disability II. Here, it certainly seems rational for you to prefer not to be blinded by your ophthalmologist. But note that this case isn’t all that different from Disability II. Just

198  Nilanjan Das and L. A. Paul

as you have two options in that scenario, so also you have two options in this case: either to go blind or not to. The consequences of these two options are the same in the two cases. If it is rational for you to disprefer blindness in this case, it should also be rational for you to disprefer blindness in Disability II. However, someone who rejects the intuition that it is rational for you to disprefer blindness in Disability II may still reject the same intuition with respect to Disability III. They may offer the following error theory. Error Theory I. We are conditioned to believe that disability is intrinsically bad (i.e., bad not only because of the social disadvantages that it gives rise to, but bad in itself). This belief is what explains our intuitions with respect to Disability II and III. But this belief is false.18 But the problem we are raising doesn’t, strictly speaking, have anything to do with disability. So, consider another example: Procreation. If you have children, you will realize a new self: your love for your children will change who you are. At the same time, your financial situation will also deteriorate: as a result, you will have to work longer hours, eat less healthy food, and will have much less time for the hobbies or pastimes that make your life slightly more exciting. But your life as a parent, despite being worse than your previous life, will be full of other valuable experiences that will make it worth having. This case is analogous to Disability I and II. On the one hand, we may be tempted to say that it is rational for you to strictly prefer not to have children in this scenario. On the other hand, this claim is incompatible with the conjunction of the Self-​ Affecting Principle and the Comparative Notion of Harm. According to the Self-​Affecting Principle, you can prefer not to perform an act A only if A makes at least one of your present or future selves worse off or harms them.This means that if an act maximizes the well-​being of your present and future selves, it is rationally permissible for you to prefer that act. But the future self that comes into being when you have children wouldn’t exist if you didn’t have children. Moreover, its existence is worth having. So, according to the Comparative Notion of Harm, your future self in the scenario where you perform the transformative act isn’t worse off, and isn’t harmed. Therefore, it cannot be rational for you to prefer not to have children in this case. There might be a different error theory that we could appeal to. Error Theory II. What explains our intuitions with respect to Disability II and III is a kind of irrational status quo bias, i.e., a preference for the current state of affairs to continue. This error theory is slightly more plausible than the previous one; for it can also explain why it might seem rational for the blind person in Disability I to prefer

Transformative choice  199

not to undergo the cataract surgery and gain sight.19 The general strategy behind this error theory is to maintain that what makes such preferences seem rational to us is that we often irrationally prefer that the current state of affairs continue as it is, and take ourselves to be rational to do so. However, this isn’t entirely obvious. Following Parfit (2011), we may distinguish two kinds of views about the rationality of preferences. According to objective theories, our reasons to prefer one outcome to another ultimately depend on the features of these outcomes. According to subjective theories, we have no reasons for our preferences, except in a derivative case where we prefer one outcome to another because the former helps us fulfil some other preference we have. Now, if we hold a subjective view about the rationality of preferences, then in a case like Disability II, a status quo bias may indeed be rational. After all, you might attach greater value to a situation where you retain your sight and thus are able to continue the activities that you currently value than to a situation where you lose your eyesight and are unable to continue those activities. If we are subjectivists, we cannot dismiss such preferences as irrational.20 Suppose we are objectivists about the rationality of preferences. Even then, it’s not obvious why we couldn’t vindicate the rationality of your preferences in Disability II. According to the description of Disability II, your blind self is worse off than your possible sighted future self:  the latter has certain capacities that the former lacks. These capacities, on an objective view, may contribute to the well-​being of the sighted self to an extent that cannot be compensated by the other benefits that the blind self could receive. If this description of the case is correct, then the blind self, even though its existence is worth having, may indeed be worse off than the sighted self. If the choice is between a situation where your future self is better off and a situation where your future self is worse off, then it does seem rational to prefer that you retain your eyesight. In this sense, this case doesn’t seem obviously like a case of irrational status quo bias.

B.  Denying the comparative notion of harm for selves Another strategy involves denying the Comparative Notion of Harm for Selves. Someone who adopts this strategy would have to say that in a case like Disability I, the disabled future self of the agent is genuinely harmed by her accident, even though she isn’t in a position to recognize the harm itself. First of all, a response of this sort is only available to a person who accepts some form of welfare objectivism. Welfare objectivism is the view that there is a certain thing or certain things—​a set of freedoms, functions, or capabilities, a list of basic goods, etc. —​that constitute the good life, independently of whether these things are desired by the particular person who lives the relevant life or can be said to be happy without them. A person who wishes to say that you (as you exist years after your accident) in Disability I are in fact harmed by your accident cannot accept a theory of welfare on which welfare is a matter of happiness or desire satisfaction;

200  Nilanjan Das and L. A. Paul

for all your (actual or suitably idealized) preferences may indeed be satisfied in this case, and you may reasonably be called happy. However, welfare objectivism isn’t sufficient for us to deny the Comparative Notion of Harm for Selves. What we need is a notion of harm on which even though your later self couldn’t be better off in Disability I, that is, in a case where the accident that disables you still harms your later self insofar as it deprives your later self of certain freedoms, capabilities, or functions. Perhaps a defender of this position could come up with a list of goods, such that if an act or event prevents an agent from possessing certain items on that list at a certain time, the self that exists at that time is harmed. The problem is this. Even if we could come up with a list like that, an agent who finds herself in a situation of transformative choice would not necessarily know what that list of goods is, or which items on the list are such that lacking them would constitute a harm. For example, in Disability II, if you rationally take the testimony of disabled people seriously, as you should, you might be rationally extremely confident that their disability doesn’t actually constitute a harm. Thus, by your lights, the expected harm that disability causes to your present or future selves may indeed be negligible (or, at least, need not be greater than the harm that the surgery causes). Therefore, according to the Self-​ Affecting Principle, you cannot rationally prefer to undergo the cataract surgery; for the expected harm that going blind poses isn’t more than the expected harm posed by the cataract surgery. This, in turn, will conflict with the intuition that you are rational to prefer to undergo the cataract surgery. Thus, even if the Comparative Notion of Harm for Selves is false, the problem that we saw in Disability II can be raised here again.

C.  Denying the Self-​Affecting Principle This discussion makes salient a different strategy for solving the problem: rejecting the Self-​Affecting Principle. According to this principle, an agent can rationally disprefer an available act A only if, in comparison with other available acts, performing A is expected to make things worse for, or harms, her present self or one of her future selves. However, there are alternative, plausible, theories of prudence that are incompatible with this principle. Start with the following simple theory. The Presentist Theory of Prudence I. It is rationally permissible for an agent to prefer an available act A to an available act B iff the expected well-​ being of the agent’s current self conditional on her performing A, is greater than the expected well-​being of the agent’s current self conditional on her performing B. It’s unclear what this theory entails in cases like Disability II. In that scenario, there are two acts that are available to the agent: going blind or not going blind. While it

Transformative choice  201

is clear that these acts will affect the agent’s future self, it’s unclear whether they will affect the well-​being of the agent’s current self. At least, it seems coherent to say that they won’t affect the agent’s current well-​being at all. So, this version of the present theory won’t do at all. We may replace it with: The Presentist Theory of Prudence II. It is rationally permissible for an agent to prefer an available act A to an available act B iff the expected value of A is greater than the expected value of B, where the value of the different outcomes of A and B is fixed by the agent’s current actual preferences. This proposal is substantively different from the previous presentist theory, since it appeals not to the well-​being of the agent’s current self but rather to the actual preferences of the agent’s current self. It might solve the problem that we are dealing with. In Disability II, you might already have preferences about blindness:  you might actually prefer to have sight to going blind. Given these preferences, undergoing the cataract surgery may indeed uniquely maximize expected value. But note that this solution won’t always work. What if you haven’t given the matter of going blind any thought at all? In such a scenario, you may not have a preference either way, so no value can be assigned to the different outcomes of the available acts. As a result, the expected value of the options will be undefined.21 Moreover, we cannot solve the problem of value gap by appealing to your rational preferences. Since the theory is supposed to predict what your preferences about going blind and not going blind should be, the theory cannot generate this prediction by appealing back to your rational preferences about going blind and not going blind. Doing so would make the theory circular. What this shows is that the problem cannot solved by appealing to the agent’s current well-​being or preferences. We might hope that it can be solved by a theory of prudence that takes into account the agent’s future well-​being (or preferences). Consider the following theory of prudence. The Totalist Theory of Prudence. It is rationally permissible for an agent to prefer an available act A to an available act B iff the expected net well-​ being of the agent’s possible current and future selves conditional on her performing A, is greater than the expected net well-​being of the agent’s possible current and future selves conditional on her performing B.22 We can see how this theory easily takes care of cases like Disability II and Procreation. In those situations, the agent has no uncertainty: she knows exactly what will happen if she takes any of her options. In Disability II, she is certain that if she goes blind, she will bring into existence a future self, which will be worse off in comparison with the future self that will exist if she undergoes the surgery. So, given that the net well-​being of the possible current and future selves that will exist if she undergoes cataract surgery is greater than the net well-​being of those selves that will exist if she doesn’t undergo the surgery, it’s rationally permissible for the

202  Nilanjan Das and L. A. Paul

agent to prefer to undergo the surgery in this case. A similar diagnosis will allow us to address Procreation. However, this view faces the same problem that totalist theories, i.e., versions of utilitarianism, face in population ethics: namely, the repugnant conclusion.23 To see why, consider the following case. Pills. You are twenty-​five now, and in relatively good health.The bad news is that you have been diagnosed with a life-​threatening disease. There are two pills available to you. Pill A will allow you to live for about twenty-​five years in roughly the same physical conditions that you are now in. But if you take Pill B, your health will deteriorate radically and you will be bedridden for the remainder of your life. But you will be able to live for sixty more years, and your existence will be worth having. Suppose you will undergo personal transformations every year, so your self will change every year for the rest of your life. And you know this.You also know that taking Pill B will diminish your annual well-​being exactly by half in comparison with what it otherwise would be. So, you can be rationally certain that all else being equal, the net well-​being of your current and future selves conditional on your taking Pill A will be k + 25x, where k is the well-​being of your current self, while x is you’re the well-​being of each of your annual future selves (if things continue as they are now). By contrast, if you take Pill B, your net well-​being of your current and future selves will be k + 60 (x/​2). According to the Totalist Theory of Prudence, rationality requires you to prefer Pill B, and thus be bedridden for sixty years. This seems bad. We may try to fix this problem by adopting: The Averagist Theory of Prudence. It is rationally permissible for an agent to prefer an available act A to an available act B iff the expected average well-​being of the agent’s possible current and future selves conditional on her performing A, is greater than the expected average well-​ being of the agent’s possible current and future selves conditional on her performing B.24 This will avoid the problem that Pill-​style cases raise for the Totalist Theory of Prudence. Since the average well-​being of your future selves when you take Pill B is exactly half of the average well-​being of your future selves when you take Pill A, the average well-​being of your current and future selves in the former scenario is greater than the average well-​being of your current and future selves in the latter scenario. So, you are required by the Averagist Theory of Prudence to take Pill A. While this might be a solution to the problem raised above, it faces another problem. The Life-​Prolonging Drug. You know that the rest of your life will be exceptionally wonderful. Now, you are offered a drug that will allow you

Transformative choice  203

to live for one more day than you are supposed to: on that day, you will be slightly worse off than you were earlier, but things will still be quite nice. There are two possible situations: in one, you take the drug, and in the other, you don’t. Suppose you will undergo personal transformations every year, so your self will change every year for the rest of your life. And you know this. In the first situation, the average well-​being of your current and future selves is lower than it is in the second; for your well-​being on the last day of your life in the first situation is lower than what it was earlier. So, the Averagist is committed to saying that you are required not to take the drug in this case. Once again, this seems too strong: while you might be rationally permitted not to prolong your life, it doesn’t seem as if you are required by rationality not to do so. The challenge here is to come up with a principle of prudence that satisfies two desiderata. First of all, it should allow us to bring into existence a non-​existent future self that is better off in comparison with a distinct non-​existent self. Second, this principle shouldn’t allow us to prolong human lives in cases by creating additional future selves just because those future selves are guaranteed to have an existence with a very low level of well-being. Yet it should allow us to do when they have a reasonably high degree of well-​being. Similar attempts at finding similar theories have been unsuccessful in population ethics, and we expect the same problems to arise for proposed principles of prudence.25

9.5  Conclusion Let’s take stock. In this chapter, we began by showing that scenarios of transformative choice can create a problem that is exactly analogous to the non-​identity problem. We then went on to consider three possible responses to this problem and showed that none of them obviously succeed. This discussion has two consequences: the first for population ethics, and the second for theories of prudence. First, since we have shown that there is an intrapersonal analogue of the non-​ identity problem, we suggest that the non-​identity problem has nothing in particular to do with population ethics. It belongs to a more general class of problems that arise whenever an agent faces a choice of creating another agent (whether it’s a self or a person) whose existence would be unavoidably flawed but still worth living. Second, the discussion poses a challenge for existing theories of prudence or practical rationality. Standard theories of prudence require the practically rational agent to prefer acts that maximize expected value of some sort. Our intrapersonal analogue reveals a problem with almost all versions of this theory.

Notes 1 The earliest discussions of this problem occur in Narveson (1967), Kavka (1981), Woodward (1986), and Parfit (1987). For more recent work on the topic, see McMahan (1981, 2009, 2013), Bykvist (2007, 2015), Adler (2012), Arrhenius and Rabinowicz (2015), Rabinowicz (2009), and Fleurbaey and Voorhoeve (2015).

204  Nilanjan Das and L. A. Paul

2 In our discussion, we have attempted to preserve the structure of Parfit's original argument while discussing human lives in a way that does not make egregious assumptions about their value. 3 Ostrovsky et al. (2009). 4 Cases of this kind can be treated as analogous to the case of Mary in Jackson (1982).This in turn raises an interesting question of what the relevant kind of epistemic transformation consists in. Following the different views on the Mary example, one could flesh it out in terms of the acquisition of certain phenomenal concepts (Loar 1990), or in terms of the acquisition of certain cognitive skills (Lewis 1996), or in terms of the acquisition of self-​locating information (Egan 2006). 5 One worry about views like this is that assigning temporal parts of a person prudential or moral status makes it difficult to explain why practical rationality or morality seems to require a temporal part of a person to subject itself to a small harm to protect a future temporal part from a greater harm; for the later temporal part’s good fortune doesn’t really compensate the earlier temporal part for the harm that it is subjected to. See Miller (2015) and Johnston (2016) for different versions of this worry. As Kaiserman (fothcoming) correctly notes, such problems only arise for stage-​theoretic versions of perdurantism. 6 Note that this notion of self is importantly different from a notion that Parfit (1987, pp. 301–​306, pp. 326–​328) proposes and Shoemaker (1999) defends. For Parfit, selves are person-​stages united by strong psychological connectedness, e.g., direct memory connections. Since relations of strong psychological connectedness aren’t transitive, a person at a certain time may have two different selves that overlap with each other. However, on our view, selves don’t overlap with each other. Our conception is closer to a view of the self defended by Kristjansson (2010) and Strohminger and Nichols (2014), on which certain morally assessable beliefs, desires, dispositions, etc., of an agent constitute her self. However, for the sake of avoiding confusion, it’s worth pointing out that Strohminger and Nichols use the terms “self ” and “person” interchangeably, which, again, we don’t. 7 Sider (2001), ch. 3. 8 Ibid., chs 4–​5. 9 It’s worth addressing a worry here. Saying that selves have prudential status doesn’t commit us to a view on which separateness of persons doesn’t matter in prudential reasoning. For Brink (1997), the assumption that persons are metaphysically separate units allows us to preserve the hybrid structure of prudence, which permits an agent to be biased in favour of herself but does not permit her to be biased in favour of any of her temporal parts. As Brink (2011) acknowledges, the fact that an agent is required by rationality not to be biased in favour of any of her temporal parts only motivates the requirement of temporal impartiality, i.e., the requirement that the agent should have equal concern for all of her parts, and not the stronger requirement of temporal neutrality, i.e., the view that the agent should attach equal weight to the well-​being of her temporal parts. Even though Brink defends this latter requirement at a number of places, we think it is implausible: it leads to the same problems that the Totalist Theory of Prudence (discussed in Section 3) leads to. If only the requirement of temporal partiality is true, then there remains a non-​trivial question for theories of prudence to settle, i.e., the question of how to distribute benefits and harms across different temporal parts of the same person. So, if prudential reasoning is concerned with this question of distribution, then it indeed may be right to say that different temporal parts of an agent, e.g., selves, can be objects of prudential concern. 10 In particular, for simplicity, we are ruling out the possibility of overdetermination or preemption in transformative self-​creation: for discussion of problems with preemption and

Transformative choice  205

overdetermination for reductive accounts of dependence, see Paul and Hall 2013.We are helping ourselves to this simplification because we are interested in formulating a puzzle as it relates to Parfit’s identity problem, not in giving a fully reductive, independent account of transformative self-​creation. 11 Here, we are working with a notion of harm, according to which harming a person or a self involves making things for it. If we adopt a non-​comparative notion of harm, we may reject this view; for example, see Shiffrin (1999). However, we can formulate a version of the same puzzle by substituting every occurrence of “make things worse” with “harm.” Note, however, that Shiffrin will reject the relevant version of Comparative Notion of Harm for Persons, because she thinks that inflicting a harm to confer a pure benefit is morally wrong unless the subject of the harm consents to it. But her view has the implausible consequence that even when a person’s life goes extremely well but involves some unavoidable harm, even then it’s morally impermissible to bring that person into question. 12 Here, and everywhere else, the expected well-​being or harm (or any other kind of value) is calculated according to a credence function that is rational for the agent to currently adopt. 13 For discussion of this point, see Saigal et al. (1996), Albrecht and Devlieger (1999), Gill (2000) and Goering (2008), and Barnes (2016). 14 Harman (2009) argues that the preferences expressed by disabled people in these cases are strongly person-​affecting:  they are happy with their disabled lives, and don’t identify with the people they would have been had they not been disabled. But such preferences, according to Harman, shouldn’t give us reason to think that their lives are just as good as that of their non-​disabled counterparts. Even if Harman is right that a life of a non-​ disabled person is overall better than the life of a disabled person, it can still remain true that the disability in question doesn’t harm the disabled person: given that she wouldn’t exist without the disability, there’s no clear sense in which she would have been better off without her disability. 15 In subsequent discussion, some have asked whether this can be solved by appealing to knowledge norms of action:  for a sample of the literature, see Pettigrew (2015, 2016), Dougherty, Horowitz, and Sliwa (2015), Moss (2016), Fraser (2018), and Isaacs (forthcoming). 16 This problem should be distinguished from a different problem that doesn’t depend on ignorance in the same way: namely, the problem of making decisions in scenarios where the agent’s values change over time. For discussion of the problem raised by Paul, see Briggs (2015) and Pettigrew (ms). 17 An initially tempting response to the problem:  if psychological continuity is what matters for survival, then we might think that a person cannot survive her personal transformation (at least if her core beliefs, desires, commitments, etc., track psychological continuity). Parfit’s (1987, pp.  326–​328) example of the nineteenth-​century Russian nobleman seems like a good example of this. Since it is rationally permissible for a person not to prefer her own death, it is rationally permissible for her not to perform a personally transformative act. Two responses. First, we have already argued that a person can survive personal transformations. Second, even if this diagnosis were plausible in some cases, it doesn’t generalize well. It seems overly strong to claim that a person who is thinking of having a child or a similar personally transformative experience is contemplating death. Or consider examples where a personally transformative act brings about a gradual transformation. The person who exists at any stage of such a transformation may be strongly psychologically connected with the person who exists at any immediately preceding stage of that transformation, making the person who exists before the

206  Nilanjan Das and L. A. Paul

transformative act psychologically continuous with the one after the transformation is complete. Such a case shouldn’t therefore be treated as a case of death either. Thanks to Joe Horton for discussion. 18 This is connected to an error theory that Barnes (2016) defends in relation to Parfit’s ‘handicapped child case’. 19 Bostrom and Ord (2006) offer an error-​theory of this sort to explain why it seems wrong to enhance human intelligence by genetic engineering. For a related discussion of the rationality of status quo bias, see Nebel (2015). Some of our discussion is based on things Nebel says. 20 Perhaps, there’s room for rejecting this response. For example, the subjectivist about the rationality of preferences could impose a diachronic constraint whereby an agent is required to take into account the values or preferences of her future selves into account while making her decisions. So, in Disability II, given that the agent’s future blind self (if brought into existence) will prefer its own existence to its non-​existence, the agent’s current self might have some reason to prefer that that future self exists. This response doesn’t immediately convince us: we need to know more about what this diachronic constraint is. First, suppose the diachronic constraint is some sort of “I’ll be glad I did it” principle: if an agent is rationally certain that if she performs an act, she’ll be (rationally) glad she did it, then she is required by rationality to prefer to perform that act. This principle (as Harman (2009) has convincingly argued), is questionable especially in the kinds of cases we are discussing. Setting these cases aside, this principle also leads to inconsistent verdicts. Second, suppose the relevant diachronic constraint is some kind of a “linear pooling” principle:  namely, that the expected value of an act (which partially determines the agent’s preferences about it) must be calculated in light of a value function that is formed by aggregating the value functions of the agent’s present self and the future selves that will come to exist if the agent performs the relevant act. This principle is more plausible: as Pettigrew (2019) shows, if an agent doesn’t conform to such a constraint, she will be predictably exploitable. But this account fails to make any concrete prediction about Disability II: depending on the weight that each value function gets, it could still be rational for the agent to continue to prefer that she retain her sight. So, the challenge for the opponent would still be to come up with an appropriate method of aggregation that predicts that this preference isn’t rational. 21 This theory is terrible in other ways too. Consider a case of future-​discounting. Surgery III. I have been given the option of undergoing a painful surgery under a weak local anaesthetic either tomorrow or in a month. If I undergo the surgery tomorrow, it will last an hour. But if I undergo the surgery in a month, it will last four hours, so the pain will be four times as much. In this scenario, if it seems irrational for me to prefer to undergo the surgery in a month. But if I  am a future discounter and actually want pain to be further away in the future than nearby, then, given my current preferences, the Presentist Theory of Prudence II could entail that it’s rationally permissible for me to prefer to undergo the surgery in a month. This seems bad, or at least, needs more of a defense than we can muster. 22 This version of the totalist theory only takes into account the well-​being of the agent’s future selves. But we can offer a preference-​based analogue of this theory. The Preference-​Based Totalist Theory of Prudence. It is rationally permissible for an agent to prefer an act A to an B iff the expected value of A is greater than the expected value of B where the value of any outcome of A or B is just

Transformative choice  207

the sum of different values assigned by the agent’s current and future selves to that outcome. If an agent’s preferences over outcomes can be represented as cardinal utilities, this proposal can work. However, this proposal might raise intra-​personal analogues of the problem of interpersonal utility comparisons. Moreover, it will be subject to the same problem that the Totalist Theory of Prudence is subject to. 23 See Parfit (1987, ch. 17). 24 Once again, we can construct a preference-​based version of this theory as we did for the totalist theory, and it would be subject to the same problem that we raise below. 25 For a clear survey of these attempted solutions, see Arrhenius, Ryberg, and Tännsjö (2017).

References Adler, M. (2012). Well-​Being and Fair Distribution: Beyond Cost-​Benefit Analysis. Oxford: Oxford University Press. Albrecht, G.L. and Devlieger, G. (1999). The disability paradox: high quality of life against the odds. Social Science and Medicine, pp. 977–​988. Arrhenius, G. and Rabinowicz, W. (2015). The value of existence. In: I. Hirose and J. Olson (eds.) The Oxford Handbook of Value Theory. Oxford: Oxford University Press, pp. 424–​444. Arrhenius, G., Ryberg, J. and Tännsjö, T. (2017). The Repugnant Conclusion. The Stanford Encyclopedia of Philosophy (Spring 2017 Edition), Edward N. Zalta (ed.), URL = . Barnes, E. (2009). Disability and adaptive preference. Philosophical Perspectives 23(1): 1–​22. Barnes, E. (2016). The Minority Body: A Theory of Disability. Oxford: Oxford University Press. Bostrom, N. and Ord, T. (2006). The reversal test:  Eliminating status quo bias in applied ethics. Ethics 116(4): 656–​679. Briggs, R.A. (2015). Transformative experience and interpersonal utility comparisons. Res Philosophica 92(2): 189–​216. Brink, D.O. (1997). Rational egoism and the separateness of persons. In J. Dancy (ed.), Reading Parfit. Blackwell, pp. 96–​134. Brink, D.O. (2011). Prospects for temporal neutrality. In C. Callender (ed.), The Oxford Handbook of Philosophy of Time. Oxford: Oxford University Press. Bykvist, K. (2007).The benefits of coming into existence. Philosophical Studies, 135(3): 335–​362. Bykvist, K. (2015). Being and wellbeing. In I. Hirose and A. Reisner (eds.), Weighing and Reasoning. Oxford: Oxford University Press, pp. 87–​94. Callard, A. (2018). Aspiration: The Agency of Becoming. Oxford: Oxford University Press. Dougherty, T., Horowitz, S. and Sliwa, P. (2015). Expecting the unexpected. Res Philosophica 92(2): 301–​321. Egan, A. (2006). Secondary qualities and self-​location. Philosophy and Phenomenological Research 72(1): 97–​119. Fleurbaey, M. and Voorhoeve, A. (2015). On the social and personal value of existence. In I. Hirose and A. Reisner (eds.), Weighing and Reasoning. Oxford: Oxford University Press, pp. 95–​109. Fraser, R.E. (2018). Stakes sensitivity and transformative experience. Analysis 78(1): 34–​39. Gill, C.J. (2000). Health professionals, disability, and assisted suicide:  an examination of empirical evidence. Psychology, Public Policy, and Law, 6(2): 526–​545. Goering, S. (2008). ‘You say you’re happy, but …’: contested quality of life judgments in bioethics and disability studies. Journal of Bioethical Inquiry, 5: 125–​135.

208  Nilanjan Das and L. A. Paul

Harman, E. (2009). ‘I’ll be glad I  did it’:  reasoning and the significance of future desires. Philosophical Perspectives 23(1): 177–​199. Isaacs,Y. (forthcoming). The problems of transformative experience. Philosophical Studies. Jackson, F. (1982). Epiphenomenal qualia. Philosophical Quarterly 32: 127–​136. Johnston, M. (2016). The personite problem:  should practical reason be tabled? Noûs 50(4): 617–​644. Kaiserman, A. (forthcoming). Stage theory and the personite problem. Analysis. Kavka, G. (1981). The paradox of future individuals. Philosophy & Public Affairs, 11: 93–​112. Kristjánsson, K. (2010). The Self and its Emotions. Cambridge: Cambridge University Press. Lewis, D. (1990).What experience teaches. In W.G. Lycan (ed.), Mind and Cognition. Blackwell, pp.  29–​57. Ladau, E. (2013). The Complexities of Curing Disabilities. Blogpost on Words I Wheel By. URL = https://​wordsiwheelby.com/​2013/​08/​complexities-​of-​cures/​ Loar, B. (1990). Phenomenal states. Philosophical Perspectives 4: 81–​108. McMahan, J. (1981). Problems of population policy. Ethics, 92: 96–​127. McMahan, J. (2009). Asymmetries in the morality of causing people to exist. In M. Roberts and D.Wasserman(eds.) Harming Future Persons: Ethics, Genetics and the Nonidentity Problem. Berlin: Springer. McMahan, J. (2013). Causing people to exist and saving people’s lives. The Journal of Ethics, 17: 5–​35. Miller, K. (2015). Prudence and person-​stages. Inquiry: An Interdisciplinary Journal of Philosophy 58(5): 460–​476. Moss, S. (2016). Probabilistic Knowledge. Oxford: Oxford University Press. Nebel, J.M. (2015). Status quo bias, rationality, and conservatism about value. Ethics 125(2): 449–​476. Olson, E.T. (1997). The Human Animal: Personal Identity Without Psychology. Oxford: Oxford University Press. Ostrovsky, Y., Meyers, E., Ganesh, S., Mathur, U. and Sinha, P. (2009). Visual parsing after recovery from blindness. Psychological Science, 20(12): 1484–​1491. Parfit, D. (1987). Reasons and Persons. Oxford: Clarendon Press. Parfit, D. (2011). On What Matters.Volumes I and II. Oxford: Oxford University Press. Paul, L.A. (2014). Transformative Experience. Oxford: Oxford University Press. Paul, L.A. (2017). The subjectively enduring self, Routledge Handbook of the Philosophy of Temporal Experience, ed. I. Phillips. Routledge. Pettigrew, R. (2015). Transformative experience and decision theory. Philosophy and Phenomenological Research 91(3): 766–​774. Pettigrew, R. (2016). Transformative experience, L.A. Paul (ed.) Mind 125(499): 927–​935. Pettigrew, R. (ms.) Choosing for Changing Selves. Bristol University. Rabinowicz, W. (2009). Broome and the intuition of neutrality. Philosophical Issues 19(1): 389–​411. Saigal, S. and Rosenbaum, P. (1996). Health related quality of life considerations in the outcome of high-​r isk babies. Seminars in Fetal and Neonatol Medicine, 1(4): 305–​312. Shiffrin, S. (1999).Wrongful life, procreative responsibility, and the significance of harm. Legal Theory 5(2): 117–​148. Strohminger, N. and Nichols, S. (2014). The essential moral self. Cognition 131(1): 159–​171. Ullmann-​Margalit, E. (2006). Big decisions:  opting, converting, drifting. Royal Institute of Philosophy Supplements 58: 157–​172. Woodward, J. (1986). The non-​identity problem. Ethics, 96: 804–​831.

10 PRUDENCE AND SELF-​CONCERN Dale Dorsey

Derek Parfit, in Reasons and Persons, expresses skepticism about a theory of rationality he identifies as “S” or the “self-​interest theory.” Parfit puts it this way: “S gives to each person this aim:  the outcomes that would be best for himself, and that would make his life go, for him, as well as possible.”1 Parfit holds that S faces a “two-​ front war”:2 the S-​theorist must defend the claim that we are rationally required to promote the best life for ourselves overall, rather than any particular time-​indexed concern. (That is, I should equally weigh later and present benefits to myself.) But it also must say that it is not rationally permissible to promote the interests of others, or to conform to moral demands. Parfit holds that S cannot survive this two-​front war, and indeed loses to the critical present-​aim theory (CP), according to which one is rationally required to conform to reasons that derive from one’s present aims, along with any aims one at present is rationally required or permitted to desire. For the purposes of this chapter, I’m granting to Parfit that S loses to CP as a general theory of practical rationality. Perhaps—​for one thing—​we can treat our own interests and concerns as secondary to moral aims, the aims of our community, or family, etc. But I’m interested in a slightly different question here. Sometimes our choices affect ourselves. What reasons do we face insofar as they do? Could a theory that takes S’s shape, while it is not an acceptable theory of rationality writ large, be acceptable as a theory of normative self-​concern? To put this more precisely, could it be the case that, simply when it comes to ourselves, rationality requires us to promote “the outcomes that would be best for ourselves, that would make our lives go, for ourselves, as well as possible”? Call this, for the sake of having a label, a prudential theory of self-​concern—​ self-​concern, on this view, takes an explicitly prudential character. After preliminaries, I turn to what I take to be the most significant challenge for a prudential theory of normative self-​concern, viz., that because prudence is temporally neutral, i.e., requires us to promote welfare benefits without regard to

210  Dale Dorsey

their temporal location or the times of our lives that are benefited, prudence can (and does) result in intrapersonal alienation. However, or so I  also argue, there appears to be a puzzle concerning prudence and its plausibility as a theory of self-​ concern: prudence seems most problematic in cases in which we are asked to set aside our prudentially relevant commitments, projects, aims, and achievements for the sake of projects, etc., that we have abandoned or have not yet taken on. But prudence seems much more plausible, or so I argue, for other sorts of welfare benefits, benefits that do not derive from success in our, e.g., aims. I argue, however, that this puzzle can be solved by rejecting a prudential theory of self-​concern not because of its characteristic temporal neutrality, but because of its characteristic welfarism.

Prudence and self-​concern Before I begin in earnest, a few preliminaries are worth discussing right away. The topic of this chapter is self-​concern: what we owe, if I may say so, to ourselves. The right theory of self-​concern will be a theory of what we should care about insofar as we should care about ourselves, will be a theory of the sorts of practical reasons that concern ourselves, will be a theory of what we ought, all-​things-​considered, to do when no one else is around or affected. Though there may be a number of additional conceptual questions that surround the notion of self-​concern, I take the intuitive idea to be clear enough to work with here. A note is worth making, however. Plausibly, reasons and obligations of self-​ concern can come in two (at least) varieties: subjective and objective.3 Subjective reasons of self-​concern are those that arise and are weighed given the agent’s limited epistemic circumstances. If, for all the evidence available, it turns out that my decision to join the circus will lead to the best life for myself then, subjectively speaking, I have reason of self-​concern to do so. But if it is the case that, in point of fact, joining the circus will harm me in important ways, then though there may be subjective reason to do so, I will have substantially weaker objective reasons of self-​concern to do so. In looking for an appropriate account of self-​concern, I’m interested specifically in an account of objective reasons: reasons that are not drawn from the agent’s limited epistemic circumstances. With this being said, a natural theory of self-​concern is prudence. When we decide to act for our own sake, it seems quite natural to believe that what we ought to do is to promote our own good, to the greatest extent possible. However, there are two facts that I want to bring out about prudential concern here. First, as already suggested, prudence is temporally neutral. On this view, the time at which the benefit occurs, or the part of one’s life to which the benefit accrues (i.e., past, present, future, far future, etc.) plays no normative role in determining the weight of the reason of self-​concern generated by the benefit in question. Second, prudence is welfarist.The practical reasons accepted by a prudential theory of self-​concern are limited only to facts about welfare benefits to the agent in question. Generally, the larger the welfare benefit, the stronger the reason.

Prudence and self-concern  211

So, on the view we’re discussing here, self-​concern ought to take the shape of prudential concern: concern for the overall quality of one’s life. But stated this way, who could doubt it? What, after all, matters when we matter, but the quality of our life? And insofar as all parts of our life are parts of our life, why should we accept anything other than a temporally neutral, welfarist, prudential theory of self-​concern?

Temporal neutrality Well, maybe there is a problem after all. To get this problem off the ground, I’m going to make a (I hope) relatively uncontroversial axiological assumption. I’m going to assume that one is benefited when one’s life aims, projects, or long-​term achievements are successful. This proposal is broadly ecumenical across different theories of well-​being.4 Of course, one thing should be noted. As Parfit suggests,5 it’s not always clear when a person has a project or aim whether that project or aim has prudential significance. For instance, I might undertake the project of saving the rainforest, say, for the sake of the planet. Or I might undertake the project for my sake, as a life’s achievement, saving the rainforest. The difference here is subtle,6 but I  take it as an assumption that those projects I  undertake for my sake, when successful, will benefit me. However, once this axiological assumption is in place, a problem arises with prudence as a theory of self-​concern. David Brink puts it like this: Prudence, as traditionally conceived, is temporally neutral. It attaches no intrinsic significance to the temporal location of benefits or harms within the agent’s life; the prudent agent should be equally concerned about all parts of her life. But people’s values and ideals often change over time, sometimes in predictable ways, as when middle age and parenthood often temporize youthful radicalism or spontaneity with concerns for comfort, security, and predictability. In situations involving diachronic, intrapersonal conflicts of value, prudence—​in particular, temporal neutrality—​appears to require the agent to subordinate her current ideals to her future ones or at least to moderate pursuit of current ideals in light of future ones. But this demand may seem to sacrifice authenticity, if we suppose that authenticity requires acting on the ideals that the agent reflectively and sincerely accepts at the time of action.7 Typical counterexamples to prudence’s temporal neutrality might take a following shape: as a young man, I reject activity X, but that as I get older, I come to embrace it.8 Prudence, given its temporal neutrality, may command me to promote my participation in activity X, rather than to engage in the project I’m committed to now. Insofar as to do so would require me to give up on things to which I’m committed for the sake of later benefits I reject, prudence would seem to be, well, alienating. Brink is correct—​but it is worth noting that he understates the problem. It may very well be that a temporally neutral account of prudence requires us to temper

212  Dale Dorsey

the pursuit of our current aims and interests for the sake of future ones. But it also requires us to temper the pursuit of our current aims and interests for the sake of past ones. To see this, let’s consider some examples. First, take: George: George has spent the last thirty years writing and publishing a long-​ form fantasy epic that has drawn considerable readership and interest among the public. Originally conceived as a three-​novel cycle intended to surpass J.R.R. Tolkein’s classic Lord of the Rings trilogy, the length of his epic ballooned once to six, now to seven, novels each running well north of 1000 pages. After having completed the fifth novel of the series, George lost considerable interest in telling the story, and decided instead to focus on shorter projects to take place in the same universe, leaving the originally planned epic uncompleted. In this case, assuming the story isn’t embellished in irrelevant ways, a temporally neutral theory of prudence seems to indicate that George’s strongest prudential reasons are constituted by the intrinsic goods that are a result of his finishing his fantasy epic. Recall that we are making the general assumption that success in one’s projects or large-​scale aims is a prudential benefit. But it would seem quite strange to say that success in a large-​scale aim has no value for one’s whole life if one completes it successfully only at a time one doesn’t value it. In addition, it seems quite clear that it is better for one’s past self to have been engaged in a success rather than a failure. And so given the significance of this prior project to the entirety of George’s life, one can assume that its completion is a very large benefit, explaining substantial prudential reasons. But he doesn’t wish to do this—​he considers writing the novel a waste of his time, and (honestly) doesn’t care at all about its being completed.Though he still enjoys the characters and universe he created, he prefers to spend time working out smaller stories rather than finishing the final two major novels in the series. Assuming he obtains some benefit from this at all, this benefit will not accrue to his present or future selves.9 Take also: Jerome: Jerome, for many years of his life, sought to climb to the top of the world of high finance. After getting an education in economics and business, he put his skills to the test, working for a number of large financial institutions, eventually ending up as second-​in-​command of a large Wall Street hedge fund. But prior to being appointed as the leader of this fund, and enjoying the privileges that such a position entails, Jerome comes to view the Wall Street world and culture as corrupt—​he starts to avoid describing what he does to people he meets. Eventually he simply cannot take it anymore, and quits. He decides to put his talent to better (in his eyes) use as financial manager for a small rural coop.

Prudence and self-concern  213

Jerome had a goal:  to rise to the top of the world in high finance. But before he managed that, he had a change of heart—​he repudiated this goal as empty, and decided to work instead for something that gave him more fulfillment. Now, one can imagine this case in a number of ways that would not result in Jerome having strongest prudential reason to keep working as a master of the universe. For starters, it could be that his previous pro-​attitude did not rise to the level of a genuine valuing attitude, being insufficiently considered, informed, or perhaps incoherent, with other things he valued more strongly. But we can assume this away. Imagine that it really was his actual valuing attitude to rise to the top of Wall Street. Surely this counts as a project that Jerome could make successful, if only he were to stick with it. Prudence as I understand it would treat this project as of at least equal significance with goods that Jerome might obtain now, the good—​say—​of his working at a rural coop. But even if we believe that it may be plausible to hold that Jerome can and should treat his future projects as of equal significance with his present ones, it seems not a little absurd to suggest that he ought to continue to promote projects he has since abandoned. Temporal neutrality, therefore, seems to be an intuitively problematic feature of any account of the nature of self-​concern.

Parfit on projects Parfit holds that cases like those on display show that self-​concern needn’t be prudential. It is worth considering one argument for this claim. (Note:  this is not Parfit’s only argument against a temporally neutral form of self-​concern, but I’m going to focus on this point here.) For Parfit, some projects of the sort on display in Jerome and George’s cases needn’t be treated with temporally neutral concern, i.e., those that arise as a result of our “ideals” or “value judgments”: “According to S, I should give equal weight to all of my present and future desires.This claim applies even to those future desires that will depend on a change in my value-​judgments or ideals. When it is applied to these desires, this claim is indefensible. In the case of reasons for acting that are based on value-​judgments, or ideals, a rational agent must give priority to the values or ideals that he now accepts.”10 However, Parfit’s argument is too quick. Why should it be indefensible to give equal weight to the value judgments or ideals that one will have throughout one’s life? After all, as Parfit appears to be suggesting, the satisfaction of such ideals is a profound prudential benefit. (If it weren’t such a benefit, there would be no reason that S would counsel giving equal weight to past and future ideals.) Parfit says this: “On the Self-​Interest Theory, this young man must give the same weight to his present and his predicted future values and ideals. This would be giving the same weight to what he now believes to be justified and what he now believes to be worthless or contemptible. This is clearly irrational. It may even be logically impossible.”11 However, Parfit makes a mistake here, both involving a crucial ambiguity in the notion of “giving weight”. What does it mean to “give weight” to a particular ideal? One possible thing it might mean is to treat this ideal as justified. Of

214  Dale Dorsey

course, prudence would surely yield a problematic verdict if it required agents to treat ideals they have repudiated as just as justified as their current ideals (leaving aside, for the moment, the possibility that one’s repudiated ideals are not as justified). But this isn’t what prudence requires. Prudence does not require one to treat one’s future ideals as justified. First, there is nothing in a prudential theory of self-​ concern that requires a person to take any attitude whatsoever. After all, prudence simply says that we should act in such a way that maximizes our overall good. It doesn’t say anything about how we should treat our various ideals, i.e., whether we should treat them as justified or unjustified. Rather, the right action is simply the one that will lead to the greatest overall benefit, not the one that is, as it were, motivated by a concern for the justification of one’s ideals, or in any other way. But even if we ignore this point, there is a second problem. Even if prudence requires us to take particular attitudes, it does not require us to take any particular attitude toward our ideals. It requires us only to take positive attitudes toward our well-​being. But notice that we do this all the time, even if we refuse to treat other ideals or sets of ideals as justified. For instance, I do not believe that my daughter’s taste in music is at all justified. But I nevertheless think there is equal reason to satisfy her taste in music as there is to satisfy my taste in music because to do so will benefit her and I am, say, as a matter of being a good parent, neutral between benefits for myself and benefits for her. To put this in a slightly different way, Parfit seems to suggest that S requires one to take a particular attitude toward the content of the ideals of one’s future and past self. But this—​as one might call it—​de re attitude is not required by a prudential theory of self-​concern. First, no attitude whatsoever is required by prudence. But even if this is false, prudence could at most require only that one take a positive attitude toward future benefits—​to say that one’s future benefits are equally weighty as one’s current benefits. This does not require one to believe that one’s future ideals are justified, just that one hold that there is good reason to benefit one’s (benighted) future self or past self (by, presumably, advancing one’s future or past self ’s projects). When we put the actual commitments of a prudential theory of self-​concern this way, it does not appear as absurd as Parfit suggests.

Brink’s defense of temporal neutrality Even if Parfit’s argument does not succeed, temporal neutrality clearly counsels us to do things that run afoul of our current commitments. And this may very well be implausible as a theory of self-​concern. However, Brink12 puts forward a defense of a temporally neutral version of prudence against the charge that it interrupts an agent’s authenticity. His strategy is to show that there need not be as much conflict between prudence and authenticity—​ or, as I discuss in the previous section, as much alienation caused by a temporally neutral account of prudence—​as is previously thought. Any alienation that in fact remains can be explained by failures to take steps to avoid alienation. Given this, or so Brink suggests, in cases in which there is a conflict between prudence and

Prudence and self-concern  215

authenticity, or where prudence is alienating, it is plausible to hold that temporally neutral concern should override one’s particular, temporally indexed aims and ideals as a matter of normative self-​concern. How is this accomplished? According to Brink, “we need to ask about the status or credentials of the conflicting ideals.”13 According to Brink, “value judgments”—​ such as our ideals, projects, and so forth—​can be “fallible and can be more or less justified.”14 Now, it may be worth taking a brief moment to inquire as to what it might mean for a person’s “value judgments” to be “more or less justified” in the present context. Recall that we are explicitly discussing prudential benefit, and so I interpret Brink’s claim to be that the value judgments—​i.e., that this particular project is good for me, or my desire to undertake this aim, and so forth—​actually do, or do not, line up with value-​independent facts about prudential value. For instance, it may be that my project of climbing Mount Everest is justified insofar as actually achieving this goal would be a prudential benefit. But my project of systematically counting the blades of grass (which is not supported by value-​independent facts) in my lawn is not—​doing so is not (or one may suppose) a prudential benefit. Given the distinction between justified and unjustified projects, aims, etc., Brink separates cases of conflict between aims and ideals into symmetrical and asymmetrical cases. Symmetrical cases are those in which a person’s changing ideals do not differ, or at least differ to a very small degree, in their capacity to be justified. Asymmentrical cases are those in which one or the other ideal is clearly more reasonable or more justified than the other. Take asymmetrical cases first. Asymmetrical cases permit to two forms, first in which one’s current ideals are justified but one’s later ideals are not, and vice versa. If the first category holds, there’s no call for prudence to command inauthenticity or for prudence to be alienating. After all, one’s current ideals more accurately reflect the underlying evaluative facts, and given that prudence responds to those evaluative facts, one has every reason to pursue one’s current ideals wholeheartedly given that they are the ones to which prudential reasons—​even prudential reasons neutrally construed—​respond. However, in the other version of an asymmetrical case, where one’s ideals get more justified over time, then it’s hard to see how a concern about alienation or authenticity might plausibly tell against prudence. After all, one’s current ideals are not well-​justified. Satisfying them will, in such cases, not be a benefit, because one’s value judgments do not track (or track only imperfectly) facts about what actually benefits. What about symmetrical cases? Setting aside cases where neither one’s future or present ideals are well-​justified (in which Brink suggests, quite plausibly, that one ought to find some new ideals15), we can consider cases where one has, say, “familiar conflicts between success in professional and family life” or “conflict between excelling as a professional athlete early in life, which may require forgoing extended educational and professional training and may impose significant health costs later in life.”16 In this case, according to Brink,“temporal neutrality recognizes a conflict of objective reasons and counsels a kind of neutrality among the competing ideals. On reflection, this seems right. If the agent can pursue Before’s ideals

216  Dale Dorsey

unreservedly only by completely frustrating After’s ideals (and vice versa), then there seems something objective wrong with the unreserved pursuit of present ideals.”17 Here Brink is suggesting that temporal neutrality (and the accompanying alienation) is plausibly a feature of prudence, insofar as one’s ideals are neither more nor less justified. Can Brink’s defense allow a temporally neutral account of prudence to avoid concerns about alienation and authenticity?18 While there is much in what Brink says, I  am more inclined to skepticism than optimism. First, begin with one of Brink’s main assumptions, viz., that one’s prudential ideals can be more or less justified or reflective of the underlying value facts. Note that this is a substantive first-​order evaluative assumption, i.e., that when it comes to prudential benefits, success in one’s projects, etc., can benefit or not. (In other words, that there are ideal-​or valuing-​independent value facts.) And while I do not wish to litigate this assumption here, making this assumption in defense of a temporally neutral form of self-​concern violates a sensible principle I  call axiological neutrality. Generally speaking, principles of rationality or self-​concern ought not to stand or fall with substantive theories of what actually makes someone better-​off, or better-​off at particular times. In other words, whether we are licensed to prefer a near term good to a long term good, or that benefits be in the future rather than the past, and so on, should be independent of the various substantive claims about prudential value and what makes one better-​off—​what, in fact, constitutes those benefits. It would be a disappointing result for a temporally neutral account of prudential rationality were it to be the case that this account of prudential normativity is incompatible with, e.g., a pure desire-​satisfactionism or subjectivism, or any other account of what states of affairs benefit people at particular times.19 The second problem is that even if we allow that there may be cases of better-​ or worse-​justified prudential ideals, it is not at all clear that George or Jerome are instances of this. It’s not immediately clear to me, in other words, that a life as a shorter story author is worse than the life of an epic fantasy author (or vice versa) independently of that author’s goals or desires, nor is it immediately clear to me that being a manager of a large financial institution is worse or better as a prudential matter (though of course as a matter of morality there may be further things to discuss) than the manager of a small rural coop. Now, of course, Brink fully considers the possibility that there will be cases where neither one’s current belief nor one’s future belief will be better justified than the other. (That is, symmetrical cases.) For these cases, as I have already noted, he suggests temporal neutrality is plausible as an account of one’s objective reasons—​ that when considering strictly the self-​regarding authority of prudential demands, the exhortation to temporal neutrality is plausible. But as an intuitive matter, Brink is more sanguine about this than I  am. Temporal neutrality enjoins not just now-​for-​later sacrifice, but now-​for-​earlier sacrifice as well. As well as later-​for-​earlier, earlier-​for-​later, and any other combination that would be supported by the strongest normatively significant prudential goods. But it’s hard to see how it’s plausible to urge that, in making a

Prudence and self-concern  217

strictly self-​regarding decision, one should (all things considered) stick to the project one has since abandoned—​or, stronger, simply repudiated—​on grounds that one’s earlier self will be benefited to a greater extent than one’s present self will be harmed. In short, in cases like those on display in George and Jerome, it does in fact seem as though there is a conflict between the account of prudence I defend and authenticity—​prudence will in fact be alienating. But in these cases, it seems difficult to defend the temporal neutrality on offer here as being intuitively plausible, or to suggest that there is not substantial intuitive pull behind the thought that one’s current ideals, rather than past ideals, ought to have substantial self-​regarding normative force.

On the other hand So far we have been considering the prudential significance of our long-​term aims, ideals, projects, and life goals. But let’s forget about those for a moment. I claim, or so I shall argue in this section, that abstracting from those considerations yields a far less problematic picture of a temporally neutral account of prudence when it comes to the proper theory of self-​concern. Temporal neutrality, or so I claim here, has a substantial degree of intuitive power.20 To see this, I’m going to assume the absolute worst possible axiological view when it comes to such a picture. Let’s say that we adopt a desire-​satisfaction view, and we hold that satisfied desires benefit whenever the objects of those desires occur.21 Now, with these assumptions, many have been tempted to criticize a temporally neutral picture. For instance, in discussing the merits and demerits of a desire-​satisfaction theory of well-​being, Richard Brandt writes: The desire theory holds, then, that greater welfare corresponds to greater satisfaction of desire, and that a benevolent person, in deciding what to do, does or at least ought to perform that act among the options open to him which will maximize desire-​satisfaction. The idea seems to be that we consider all the desires a person has… at some time or other, or many times, over a lifetime, and what that person… should aim at is to maximize the satisfaction of these desires. This conception is unintelligible. That there is a problem begins to appear when we reflect that some desires need not count. Suppose my six-​year-​old son has decided he would like to celebrate his fiftieth birthday by taking a roller-​coaster ride. This desire now is hardly one we think we need to attend to in planning to maximize his lifetime well-​being. Notice that we pay no attention to our own past desires.22 I’m going to leave aside here Brandt’s mistake of running together an axiological theory (the “desire theory”) with a normative account of self-​concern (“ought to perform the act among the options open to him which …”). With that out of the

218  Dale Dorsey

way, Brandt does slip in a pretty nifty zinger against temporal neutrality. A  temporally neutral form of prudence, especially one that is conjoined to a desire-​ satisfaction theory of well-​being of the sort I assume here, is committed to holding that we ought to treat our past desire to ride a roller-​coaster on our 50th birthday as what we ought to do as a matter of self-​concern. But is this verdict really as problematic as Brandt says? To begin it would do to look at the specific verdicts my view might offer about Brandt’s case and what assumptions we must make about the case itself such that my view will in fact deliver these verdicts. Begin with: Reason:  The fifty-​ year-​ old has a reason of self-​ concern to ride the roller-​coaster. Assume that it can in principle be a prudential benefit to the six-​year-​old self that I ride a roller-​coaster when I turn 50. Of course, this would require certain assumptions about the desire in question. In particular, it cannot be understood as a valuing attitude that is “conditional on its own persistence”.23 In other words, it cannot be that the six-​year-​old desires to ride the roller-​coaster only if he continues to wish to do so. Indeed, commonsensically, such desires seem to be understood conditionally in just this way. While I may want to, e.g., ride a roller-​ coaster later in life, this is likely to be colored by the thought that I’m interested not so much in the roller-​coaster itself, but rather in being the sort of person who would want to do so, and hence any value I place in riding the roller-​coaster will be conditional on my actually having the relevant valuing attitude.24 But let’s leave this point aside for the moment. Assume that the desire is not conditional on its own persistence. Given a prudential theory of self-​concern, this entails reason for me to do so. So let’s move on. What about: Equally Weighty Reason: the fifty-​year-​old has just as strong a reason of self-​ concern to ride the roller-​coaster as he does to do what he currently prefers, viz., relax in an easy chair with a glass of bourbon. Whether my view delivers Equally Weighty Reason obviously depends on our understanding of the case at hand. In particular, the case must be interpreted such that the benefits are of equivalent size. But this requires that we make some stipulations about the mental states of the six-​year-​old and 50-​year-​old. However, it must be more than this. It must be that, for instance, there is no further ancillary disvalue to be had for the 50-​year-​old self in going on the roller-​coaster. It must be, for instance, that the extent to which one disvalues (at 50) going on a roller-​coaster ride (and the necessary features of doing so: waiting in line at the amusement park, seasickness immediately following, and so on), and any potential opportunity costs

Prudence and self-concern  219

(such as the inability to relax with a glass of bourbon) are made up for by the relative size of the benefit to the six-​year-​old. What about: Obligation:  the fifty-​year-​old has an obligation of self-​concern to ride a roller-​coaster. Under what conditions might Obligation be correct? To begin, we need only consider cases in which the psychological states imagined to be in equipoise sufficient to guarantee Equally Weighty Reason are tilted, if ever so slightly, toward taking the roller-​coaster ride. In other words, in terms of the relative size of the benefit, the benefit for one’s six-​year-​old self to be had by taking the roller-​coaster ride must be larger than the benefit to be had to one’s 50-​year-​old self in relaxing with a glass of bourbon.25 I think this about sums it up. Now, under the relevant conditions required for a prudential theory of self-​concern to deliver each of the particular verdicts, is this a plausible verdict to deliver, independent of the arguments that I’ve offered for temporal neutrality so far? Take Reason. It seems to me that Brandt is incorrect when he says that we give, or perhaps should give, our past valuing attitudes no weight at all. Indeed, one might imagine that with no cost at all, I could make it the case that something to which I am now indifferent, but that mattered a great deal to me in the past, would occur. I find it quite implausible to believe that there’s nothing to be said for doing it. Indeed, or so it strikes me at least in this case, one should do it—​after all, it’s of no cost. Why not make the wishes of your past self come true? Under, then, the psychological assumptions required to deliver Reason on the current view it seems precisely the correct answer to deliver. What about Equally Weighty Reason? Note that Equally Weighty Reason does not imply that one is prudentially obligated to take the roller-​coaster ride. On the other hand, because there are, well, equally weighty reasons do either, this entails that one is prudentially justified in doing either. But for my money, once again, this seems precisely the right verdict. Imagine that my 50th birthday rolls around, and you know that I would prefer to spend it relaxing with a nice glass of bourbon. But you learn that instead of doing this, I drove to a local amusement park, and instead rode a roller-​coaster. “Why on Earth would you have done that?” you ask, “I thought you wanted to relax with a nice bourbon!” “I did,” I reply, “but then I happened to come across my old grade-​school diary. I saw that, written right there—​I must have been no more than six or seven—​was my wish that when I turned 50 I would ride a roller coaster. And I thought to myself, why not make that wish come true? I sure did miss the bourbon, though!” I find the thought that my response would be insufficient to prudentially justify myself very hard to believe. Your response to me would likely be a mild chuckle, but it would not be the thought that I haven’t conformed to proper standards of self-​concern.

220  Dale Dorsey

Obligation, admittedly, is trickier. Just as we wouldn’t want to say (under the right suppositions) that you aren’t justified in taking the roller-​coaster ride, it would seem implausible to say that you aren’t justified, even if prudential value mildly tilts in favor of one’s six-​year-​old self, relaxing instead with a nice glass of Kentucky’s finest. However, it’s not entirely clear to me that—​on further reflection, anyway—​ this isn’t precisely the verdict we should accept, at least under the conditions, which would make it a result of the view to which I subscribe here. To begin, consider the possibility of future desires about what I  do on my 50th birthday. Perhaps, on my 80th (I should be so lucky!) I come to think back on my 50th birthday party, and value its being the case that I spent that birthday with a person with whom I’m not particularly acquainted now, but whom I later come to be quite close to. I  prefer relaxing with bourbon, but my older self prefers that I spend it with this other person. And imagine that to do so would be better for my 80-​year-​old self than to drink the bourbon would be for my 50-​year-​old  self. Am I required to spend the day with this person, on grounds that my 80-​year-​ old self desires it (in the relevant way)? To my ears, this sounds no more intuitive than the suggestion that I should spend the day on roller-​coasters. However, this creates a puzzle. Most hold that it is intuitive to engage in now-​for-​later sacrifice, of which this case is an example. And hence we would treat this particular judgment as not indicative of the real demands of prudence. After all, despite my current recalcitrance, it seems a paradigmatic instance of prudential rationality that I go to the dentist to avoid a much more painful toothache later. To put this another way, the judgment that we are not prudentially obligated to look after the interests of our 80-​year-​old self in the relevant way is outweighed, on balance, by the reasons we have to reject temporal biases. Our later interests are just as weighty as our present interests. But if we’re willing to overrule our intuitive reactions in the case of now-​ for-​later sacrifice, it’s not clear why we aren’t willing to do so in the case of a bias against now-​for-​earlier sacrifice. Indeed, once we recognize that going on the roller-​coaster is, in fact, a benefit to one’s early self, that it promotes a good for oneself, it seems entirely strange to say that we could lack an obligation of self-​concern to promote it, once we recognize that doing so makes our lives go as well as possible. I suspect that recalcitrance on this matter is due to the general thought that we cannot influence the well-​being of our past selves. But this thought, or so I  have assumed for the sake of argument here, is mistaken. And recognizing that we can do so should come along with no less significant normative implications than the recognition that we can affect our welfare years in the future. If circumstances are that the goods promoted by my going on the roller-​ coaster are weightier on balance than the goods promoted by my not doing so, I should go on the roller-​coaster. When we abstract, then, from our professed ideals or long-​term projects, achievements, goals, and so forth, it seems right to hold that self-​concern ought to be temporally neutral.

Prudence and self-concern  221

But this just illustrates how paradoxical an inquiry into the nature of self-​concern can be. It seems on the one hand (i.e., in the cases of George and Jerome) that temporal neutrality should simply be rejected. It seems on the other hand (i.e., in the case of my six-​year-​old desire) that temporal neutrality can be sensibly defended. How to proceed?

Is self-​concern welfarist?26 If—​a big “if ”, admittedly—​everything I’ve said so far is right, it would appear that we treat welfare benefits that depend on our long-​term aims, ideals, projects, commitments, and so on very differently from those (such as my six-​year old desire) that do not. Temporal neutrality seems defensible for the latter, far less defensible for the former. The key, then, is to somehow differentiate, as a matter of self-​concern, the significance of our projects, ideals, and so on, and our more momentary desires, such as those on display in the case of the roller-​coaster ride. But how to do this? One might hold, for instance, that different welfare benefits permit of different pictures of self-​concern:  one temporally neutral (momentary desires), the other temporally biased (ideals, projects, etc.). But this seems, initially, unpromising. In particular it would seem that though there are differences in the prudential upshot of such goods, it’s hard to see why such a difference would entail that we are licensed or not to be temporally biased.27 How to square this circle? The thesis I would like to suggest, somewhat “suggestively”, I suppose, is that when it comes to our projects, self-​concern involves reason to promote their success even absent a specific focus on the welfarist upshot that the success of those projects involves. Projects, in other words, have a kind of per se significance for normative self-​concern—​normative significance, in other words, that is independent of their particular prudential significance. And if this is right, while the demands of prudence are certainly a source of reasons of self-​concern, the projects in which we are engaged are also sources of reasons of self-​concern, independently of their prudential signficance. To see this more clearly, imagine that I take on a project that is not, if completed, a prudential benefit. Perhaps I  value, e.g., saving the rainforest not as a matter of prudence (I do not desire to pursue it for my sake) but rather because I  am committed to preserving species biodiversity on the planet. Now, surely there are moral reasons to do this. But it seems like one reason for me to help to save the rainforest in particular (rather than, e.g., helping to reduce pollution in the oceans or helping to reduce the spread of the Gobi Desert or by helping to cure malaria in the Carribean) is that doing so is my project, a project I value and have undertaken. If this is correct, then this should apply not just to other-​regarding projects, but to projects we undertake for our own sake, as well. This thesis is broadly ecumenical, and fits a number of views concerning the nature of practical normativity. Take, for instance, the following passage from Korsgaard:

222  Dale Dorsey

An agent might think of herself as a Citizen of the Kingdom of Ends. Or she might think of herself as someone’s friend or lover, or as a member of a family or an ethnic group or a nation. She might think of herself as the steward of her own interests, and then she will be an egoist. Or she might think of herself as the slave of her passions, and then she will be a wanton. And how she thinks of herself will determine whether it is the law of the Kingdom of Ends, or the law of some smaller group, or the law of egoism, or the law of the wanton that will be the law that she is to herself… Practical identity is a complex matter and for the average person there will be a jumble of such conceptions.You are a human being, a woman or a man, an adherent of a certain religion, a member of an ethnic group, a member of a certain profession, someone’s lover or friend, and so on. And all of these identities give rise to reasons and obligations.Your reasons express your identity, your nature; your obligations spring from what that identity forbids.28 Korsgaard here pinpoints the source of normativity as a person’s practical identities—​ the general evaluative outlook a person has. Notice that Korsgaard does not treat a person’s practical identities temporally neutrally:  it’s not as if one’s future practical identities are just as normatively significant as one’s present practical identities. Rather, it’s the practical identities one has now, the practical identities that matter to you now that provide you reasons for action. One does not have to commit to any particular full-​ blown account of normativity to hold that granting commitments, projects, and the like special normative significance is a plausible feature of practical rationality (whether or not all of practical normativity is constructed by such commitments, aims, or interests). For instance, Ruth Chang holds that while our commitments do not constitute the complete picture of practical reasons, our commitments can have the result of giving rise to new reasons to act. For Chang, our commitments involve a willing that there be a reason to conform or follow through on the commitment. My commitment, say, to finishing my book, or climbing Everest, etc., involves willing that the completion of the book be a state of affairs with normative significance.29 Indeed, the plausibility of the independent normative significance of our temporally indexed commitments seems entrenched. For instance, consider George. Initially, it is plausible to say that George, right here and right now, can justify his decision to work on shorter projects rather than his epic given the commitments he takes toward that project, given his valuing attitudes as they are now. While George may be prudentially obligated to complete his past project, and while this surely generates some practical reason for him to do so, it is also plausible to hold that the fact that he is committed to his present project is also a practical reason for him to avoid the epic and complete other aspects of his universe—​the fact that this is his project provides justificatory force that is independent of its prudential significance, its significance qua benefit.

Prudence and self-concern  223

This claim seems to me straightforward and to comport with commonsense. I may, as Chang suggests, commit to a certain person, commit to a particular relationship. Plausibly, this fact (my commitment, evaluative attitudes, and so forth) entails that I have special practical reason to follow through on this relationship or cause, etc., that would be absent were my commitment not present. And this fact seems to have very little to do with the prudential significance of this relationship. I  may commit to, say, la revolución. While there may be very good moral reasons to advance this cause, and perhaps even prudential reasons, the fact that I  have committed to this cause seems itself a normative fact, something that strengthens the normative force that stands behind actions to advance this political end. And, likewise, nothing about this fact seems to depend on the extent to which the success of this project is a benefit, though benefit it may be.30 But if our commitments can give rise to reasons, or stronger reasons, to forward my political causes or to advance our relationship in a way that is independent of the particular prudential significance of such projects, the natural result would be that this normative fact applies to any sort of commitment, even when such commitments have prudential significance. If this is correct, then while prudence may command that we respect temporal neutrality when deciding whether to complete our projects, the fact that we have committed to particular projects can itself have per se self-​regarding significance that will compete with, and will occasionally outweigh, the temporally neutral welfarist significance of such projects. Generally, then, the supposition that normative self-​concern is temporally neutral is false. Our commitments are themselves normatively significant independent of the benefits they yield. Sometimes these commitments will be to projects that have prudential significance, and hence we can be all-​things-​considered justified in fulfilling our present projects even if to do so would entail living a worse life. This proposal entails a theory of self-​concern that is not temporally neutral. But the way in which it rejects temporal neutrality is not by holding that the significance of facts about well-​being can change depending on the temporal location of benefits. Rather, it rejects temporal neutrality by rejecting welfarism, viz., the claim that reasons of self-​concern are constituted only by facts about the prudential good. According to the view I suggest, reasons of self-​concern are not entirely constituted by facts about a person’s well-​being. Rather, reasons of self-​concern can also extend to facts about the existence of projects, commitments, and ideals. Now, this does not imply that as a matter of self-​concern, one has practical reasons to take up those projects one endorses non-​prudentially, or in a way that will not result in a benefit. After all, if I endorse saving the rainforest strictly for the sake of the welfare of future generations, it is clear that this is a project that I would not undertake as a matter of self-​concern. It follows, then, that reasons of self-​concern will correlate with welfare benefits. But this does not entail that the theory of self-​concern is, ultimately, welfarist. But even though the view is non-​welfarist, it does entail that self-​concern can make a principled distinction between welfare goods that derive from the success of our projects, ideals, aims, and achievements, and those goods

224  Dale Dorsey

that do not. In the case of the former, we have stronger reason to act in accordance with those projects we adopt now —​such as George’s project of confining himself to short tales of his fantasy universe—​than projects we may have abandoned, despite their significance as welfare achievements.

Conclusion Parfit, in Reasons and Persons, suggests that a theory that simply enjoins prudential concern cannot be the whole story of practical rationality. On this, I’m tempted to agree. But what if we abstract from considerations that involve others, such as moral considerations or other forms of other-​regarding concern? What theory should we adopt of normative self-​concern? Here prudence becomes a natural answer. But prudence as a theory of normative self-​concern is paradoxical.While it seems plausible to hold that temporal neutrality is intuitive for some benefits, it seems plausible to believe that we can at least grant some preference to our current projects, aims, and commitments insofar as these have prudential significance. However, or so I argue here, it is not prudence’s temporal neutrality per se that is the root of the problem. Rather, or so I suggest, it is prudence’s welfarism. If we reject welfarism, and hold that self-​concern can sometimes be shaped by our commitments directly (even if those commitments have prudential significance), this allows us to escape what paradox may arise when considering prudence as a substantive theory of self-​concern.

Notes 1 Derek Parfit, Reasons and Persons (Oxford: Oxford University Press, 1984), 3. 2 Parfit, 126. 3 For instance, see David Brink, “Prudence and Authenticity: Intrapersonal Conflicts of Value” in Philosophical Review 112 (2003), 219. 4 For instance, one could accept such a view on a broad variety of so-​called “objective list” theories, along with perfectionist theories, as well as certain forms of subjectivism. (See, for instance, Dale Dorsey, “The Significance of a Life’s Shape” in Ethics 125 (2015).) 5 Reasons and Persons, 494. 6 Cf. Dale Dorsey, “Subjectivism without Desire” in Philosophical Review 121 (2012), 419–​422. 7 Brink, “Prudence and Authenticity: Intrapersonal Conflicts of Value”, 215. 8 Parfit’s own infamous case of the Russian nobleman fits this bill. See Reasons and Persons, 327. 9 Worth stipulating is that George’s previous desire to finish his fantasy epic was not “conditional on its own persistence”: he (correctly) predicted that the project might alienate his future self at certain points, but was nevertheless persistent in his desire that the project be completed despite this. 10 Parfit, 155. 11 Parfit, 155. 12 Brink, “Prudence and Authenticity”, 215. 13 Brink, 236.

Prudence and self-concern  225

14 Brink, 236. 15 Brink, 237–​238. 16 Brink, 238. 17 Brink, 238. 18 Note:  part of Brink’s defense of temporal neutrality appeals to the understanding of authenticity as “being true to oneself ”, which, on Brink’s view, means being true to all the parts of oneself. Thus, on this account of authenticity, there is no genuine conflict between authenticity and temporally neutral versions of prudential rationality. And he may be right about this, but if so the concerns about alienation separate from concerns about authenticity. Alienation seems explicitly tied to one’s ideals at a particular time. 19 Worth noting here is that Brink is a steadfast critic of desiderative and subjectivist views of prudential value. Cf. “The Significance of Desire” in Oxford Studies in Metaethics, ed. Shafer-​Landau, v. 3 (Oxford: Oxford University Press, 2008). However, this should make no difference in the present context. If temporal neutrality is a justified approach to prudential rationality, it should be justified regardless of the facts of prudential benefit. 20 Further, and more developed, arguments for temporal neutrality are offered in Dale Dorsey,“A Near-​Term Bias Reconsidered” forthcoming in Philosophy and Phenomenological Research. 21 This view is defended in, e.g., Dale Dorsey, “Desire-​satisfaction and Welfare as Temporal” in Ethical Theory and Moral Practice 16 (2013). 22 Richard Brandt, A Theory of the Good and the Right (Oxford: Oxford University Press, 1979), 249. Indeed, while Parfit himself rejects temporal neutrality for goods of this kind (Reasons and Persons, 165–​166), the case under consideration here seems to me far more indicative of a true case of present-​for-​past sacrifice. 23 Parfit, 150. 24 Compare the lyrics to the classic Beach Boys song, “When I Grow Up to Be a Man”: Will I dig the same things that turn me on as a kid? Will I look back and say that I wish I hadn’t done what I did? Will I joke around and still dig those sounds When I grow up to be a man? The Beach Boys, “When I Grow Up to Be a Man” by Mike Love and Brian Wilson (The Beach Boys Today!: Capitol Records, 1965). 25 This follows given the traditional assumption that prudence is a maximizing doctrine: that one ought, prudentially, to conform to the strongest balance of prudential reasons. 26 The argument of this section is given in much more detail in “The Normative Significance of Self ” in Journal of Ethics and Social Philosophy, and The Limits of Moral Authority, ch. 6. 27 I try to make this argument in “Prudence and Past Selves”, forthcoming in Philosophical Studies. However, the argument I offer there would not be successful at differentiating projects and, e.g., the roller-​coaster ride, insofar as the latter is a “cooperative good,” as I define it there. 28 Korsgaard, 101. 29 Chang, “Commitment, Reasons, and the Will” in Oxford Studies in Metaethics, v. 8, ed. Shafer-​Landau (Oxford: Oxford University Press, 2013). 30 An important note seems worth making here.You might wonder, metaethically speaking, how our commitments give rise to new normative facts—​how, in other words, the fact that I  have committed to something makes it the case that actions that promote that thing have greater justificatory significance than they otherwise would have had.

226  Dale Dorsey

Of course, there are options here. For Chang and Korsgaard, commitments, practical identities, projects, and the like give rise to new reasons. For Korsgaard, our practical identities are the source of normativity. For Chang, while our commitments do not determine all of the normative landscape, we create new reasons by committing in cases in which reasons would otherwise be in “equipoise” (Ruth Chang, “Grounding Practical Normativity: Going Hybrid” in Philosophical Studies 164 (2013), 178). For my money, the suggestion that we create new reasons by committing seems wrong, in part because the typical examples of the normative significance of commitment, i.e., to a political cause, to the well-​being of a particular person, and so on seem to be cases in which there already are reasons—​the fact that my friend will be benefited by some action φ is reason independent of my commitment. Rather, it seems to me sensible to hold not that we create new reasons in committing, or taking on projects or practical identities, but rather we strengthen reasons that already exist. For more on this, see “The Normative Significance of Self,” op. cit.

11 REDUCTIONISM, SELF-​ CONSTITUTION, AND THE MORAL SIGNIFICANCE OF PERSONAL IDENTITY Carol Rovane

In Reasons and Persons, Derek Parfit offered a psychological reductionist account of personal identity, according to which the existence of a person consists in nothing but certain sorts of events standing in certain sorts of relations.1 He took this account to show that Locke was right to distinguish personal identity from animal identity. But more significantly, he took the account to show that the distinction between one person and another is far less deep than we ordinarily take it to be, and he argued that psychological reductionism therefore deprives so-​called “common sense morality” of crucial metaphysical support, and the same goes for any morality that elevates the moral importance of the individual by elevating such values as individual interest, individual autonomy, individual responsibility, individual rights, and so on, as we find in the broader liberal tradition. It was largely on this ground that he recommended a shift towards consequentialist morality. In a critical notice on Reasons and Persons, Christine Korsgaard correctly pointed out that Parfit failed to take due account of the fact that persons are agents, and she went on to argue that taking this fact duly into account will lead us towards a normative conception of agency, which in turn will turn us away from consequentialism to Kantian morality2 –​which, in effect, will lead us back to the liberal moral tradition that has so often claimed commonsense status for itself, and for which Kant himself claimed that status. In subsequent work, Korsgaard sought to further consolidate her defense of Kant with an account of self-​constitution that she situated within a constructivist meta-​ethical framework.3 In this chapter I will revisit this early debate between Parfit and Korsgaard, and argue for different conclusions from theirs on both the metaphysical and the moral fronts. With respect to reductionism: Korsgaard was right to object that Parfit’s psychological reductionist account of personal identity does not take due account of the fact that persons are agents. She

228  Carol Rovane

was also right to claim that the particular kind of agency that persons possess is irreducibly normative. But this irreducibility of the normative to the non-​normative does not rule out the very different sort of metaphysical reduction for which Parfit argued, according to which the existence of a person consists in nothing but certain sorts of events standing in certain sorts of relations. We can carry out the latter sort of reduction while respecting the irreducibility of the normative, by construing the events and relations in question in suitably agential and normative terms. The upshot is a reductionist account of personal agency, which says that the existence of a person consists in nothing but certain sorts of intentional activities that stand in certain sorts of normative relations. I have argued elsewhere that this account entails a novel interpretation of Locke’s distinction between personal and human (animal) identity, because it entails that the condition of individual agency can be realized in different ways with respect to individual human lives. It can be realized in a one-​to-​ one way, so that there is an agent of roughly human size; but it can also be realized in a many-​to-​one way so that there are multiple agents within a single human life; and it can also be realized in a one-​to-​many way so that there are group agents that span many human lives.4 With respect to self-​constitution: Korsgaard was right to claim that persons are self-​constituting. I take this to mean that the existence of a person is a product of effort and will, and that the efforts through which a person comes to exist are its own efforts. In other words, I take this to mean that an individual person makes itself. Yet prima facie, this might seem to be impossible. For prima facie, it would appear that making requires a maker, and so no self making could possibly go on unless a self was already there to do it. If this appearance were veridical, then we would have to construe all talk of self-​ making as merely loose talk, which really enjoins persons to make something of themselves –​that is, of selves whose existence is already metaphysically given. But the reductionist account of personal agency construes such talk literally. It says: the existence of a person consists in nothing but the intentional activities that constitute it; and these activities make a person; and since a person is nothing but these activities, the person makes itself. Furthermore, the account claims that this is true of all agents no matter what their size –​that is, it is true of agents of human size, as well as multiple and group agents. With respect to the moral significance of personal identity: To be frank, I have not yet fully worked out to my satisfaction what the larger moral significance the reductionist account of personal agency is. But my arguments in this chapter do make a start in working this out. Korsgaard was right that Parfit’s reductionist account of personal identity does not take due account of personal agency. But it turns out that when we do take due account of this, we do not end up where Korsgaard expects us to be. Her own thesis of self-​constitution actually supports the central plank of the reductionist account of personal agency, and this account is in deep tension with her Kantian project. It does not support her rationalist ambition to derive the categorical imperative of Kantian morality from the conditions of rational agency alone. In addition, it

Moral significance of personal identity  229

completely undermines the Kantian conception of the person, qua agent, as an end in itself. In these respects, and more, the reductionist account of personal agency converges with Parfit’s sense of the moral significance of his own proposed reduction of personal identity. He thought that his reduction should help us to see that the individual person cannot hold the kind of moral significance that commonsense morality attaches to it. The reductionist account of personal agency agrees with him.Yet it does so without turning its back on the central importance of agency in our moral thinking, or elevating consequentialist moral considerations in the way he did.

The reductionist account of personal agency Irreducible normativity Here are some claims about persons that the reductionist account of personal agency aims to incorporate: persons exercise their agency intentionally, and therefore knowingly; persons recognize both themselves and one another as agents; persons recognize both themselves and one another as capable of apprehending and responding to the normative requirements of rationality; persons can engage one another in distinctively interpersonal ways, such as conversation, argument, criticism, etc.; distinctively interpersonal engagement takes place within the space of reasons, as Sellars so aptly characterized it, which is a normative space in which persons apprehend, and draw attention to, and respond to, the normative force of the considerations from which they deliberate and act. The sense in which personal agency is irreducibly normative goes beyond the fact that persons apprehend and respond to the requirements of rationality. It pervades the entire realm of their intentionality, including thought as well as action. To say that thoughts and actions are irreducibly normative is to say that they cannot be fully captured within the sort of functionalist framework that has been widely regarded as affording the best hope for a naturalistic and scientific account of the mind.5 Such a naturalistic functionalism would reduce intentional attitudes (sometimes called propositional attitudes) to causal dispositions to reason and act in certain law-​governed ways.6 In contrast, the reductionist account of personal agency construes intentional attitudes as commitments in the normative sense, according to which they constitute the normative basis from which a person’s deliberations and actions ought to proceed, as opposed to the causal basis from which they actually do proceed. So for example, to construe a belief as a commitment is to construe it as an attitude of holding something true, where to hold it true is to be committed to deliberating and acting in the light of its truth. Similarly, to construe an evaluative attitude as a commitment is to construe it as an attitude of holding something to be good, where to hold it to be good is to be committed to deliberating and acting in the light of its worth.7 When I say that commitments of these kinds are irreducibly normative, I mean that they are not reducible to causal dispositions to deliberate and act in accord with them. A person may fail to live up to their commitments

230  Carol Rovane

(by failing to deliberate and act in accord with them) and yet still possess them, so long as they hold that they ought to live up to them; and they hold that they ought to live up to them so long as they take their perceived failures to live up to them as grounds for self-​criticism.8 When the reductionist account of personal agency invokes this irreducibly normative conception of intentional attitudes, as commitments, it affirms the sharpest of distinctions between ought and is –​which is to say, it does not try to make sense of how we might close, or even cross, the metaphysical gap that the distinction registers.9 On this matter, it affirms a dual standpoint theory of freedom, versions of which can be found in the work of Kant and Spinoza, and in more recent work as well.10 A dual standpoint theory posits two separate domains. One domain consists of matters of fact, which are appropriate targets for scientific inquiries that would be carried out from what might be called the observer’s standpoint, from which the aim would be to explain and predict on the basis of causal laws. The other domain consists of matters of value which are shot through with normativity; these matters are appropriate targets for deliberation about what to do, and other forms of practical inquiry as well, all of which would be carried out from the agential standpoint. From the agential standpoint, the following sorts of evaluative questions can be raised and answered: What is good and bad? What is better and worse than what? What would it be best for me to do, all things considered? And when we say that agency itself is irreducibly normative, we are saying that questions such as these are very much at issue when we recognize the agency of ourselves and others. That is, when we discern others’ actions, and the intentional attitudes (i.e., commitments) in the light of which they are undertaken, we are not discerning matters of value-​neutral fact but are normatively engaged; and that is why we are in a position to engage others in distinctively interpersonal ways such as conversation and argument, as opposed to regarding them as mere things to be observed, predicted, causally explained, and manipulated.11

The requirements of rationality The argument for the reductionist account of personal agency exploits a conceptual tie between the concept of an individual agent and certain normative requirements that are constitutive of individual rationality, in the sense that they define what it is for an individual agent to be fully, or ideally, rational.12 My arguments in this chapter do not require me to settle on a definitive list of what rationality specifically requires of agents. But here are some candidate requirements:  consistency (an agent ought to resolve any conflicts among their beliefs), closure (an agent ought to accept the implications of their attitudes), and transitivity (an agent ought to achieve a transitive ordering of its options and values).13 In addition to such specific requirements, there is a more general one, which is an overarching requirement on an individual to achieve overall rational unity within oneself, by arriving at and acting upon all-​things-​considered judgments. The things-​to-​be-​considered in such judgments are all that the agent holds to

Moral significance of personal identity  231

be true and good, where holding things to be true and good is the same thing as embracing commitments to deliberate and act accordingly. As I explained above, these commitments are not to be regarded as causal dispositions, naturalistically construed, in the light of which an agent’s behavior could be predicted; these commitments are normative, in the sense that they register considerations that the agent ought, by their own lights, to take into account when they deliberate and act. Thus, what an all-​things-​considered judgment aims to register is the joint normative significance of all of an agent’s commitments taken together, with regard to the question, what would it be best for me to do in the light of all that I think? The way to arrive at such all-​things-​considered judgments is by meeting the other, more specific, requirements of rationality such as the ones I mentioned above. But as I’ve said, it does not matter for my purposes here exactly what these more specific requirements are. All that matters is that meeting them contributes to the aim of meeting the overarching requirement to achieve overall rational unity within oneself, by arriving at and acting upon all-​things-​considered judgments. I should clarify that I  take the requirements of rationality to be internalist in two related senses, both of which reflect the fact that deliberation and action must always proceed from a first person point of view. First, when I deliberate, rationality does not require me to take into account matters of fact and value concerning which I  do not have any commitments already, for such matters are external to my point of view; second, rationality does not require me to take into account the commitments of others, except insofar as my own commitments dictate that I should do so, for otherwise those other-​regarding considerations would remain external to my point of view. The second aspect of internalism highlights that while it might be a moral failing on my part to disregard others’ commitments, it is not necessarily a rational failing. It also highlights a deep point about the first personal nature of commitments: it is impossible to deliberate from anyone else’s commitments because, in the very act of taking any consideration as a basis for one’s own deliberations, one thereby undertakes a commitment of one’s own. As a result, internalism brings with it an unavoidable individualism, which returns us to the conceptual point from which the reductionist account of personal agency proceeds, namely, that the requirements of rationality are constitutive, in the sense that they define what it is for an individual agent to be fully or ideally rational.

Reductionism, self-​constitution, and Locke’s distinction According to a Parfit-​style reduction, the existence of a person consists in nothing but certain sorts of events standing in certain sorts of relations. According to Korsgaard’s thesis of self-​constitution, the life of a person, qua agent, does not begin unless and until agency is actually exercised. The reductionist account of personal agency combines these two claims. It says that the existence of a person, qua agent, consists in nothing but certain sorts of intentional activities, which are directed at meeting the constitutive requirements of individual rationality, and through these intentional activities an individual person constitutes itself.

232  Carol Rovane

If the thesis self-​constitution seems to be paradoxical, it is only because we implicitly reject the sort of Parfit-​style reduction of personal agency that I am proposing. That is, the thesis appears paradoxical only when we take it for granted that there must be some further thing that is the person, a thing whose existence is already given, who then does the thinking and acting through which the person is supposed to constitute itself –​which then appears to contradict the thesis that the person really is constituting itself. However, even philosophers who would have us suppose that there must be some such further thing, which is the person, must grant that this thing cannot do the thinking or acting in question unless they have a first person point of view from which to do so.This would be a point of view from which the person can recognize its identity in a first person way –​in Locke words, a person “can think itself as itself in different times and places.” This would also be a point of view from which the person can recognize its thoughts and actions in a first person way, as its own –​it can think its thoughts as “one and all mine,” as Kant put it. Finally, this would be a point of view from which the person can apprehend the normative force of the constitutive requirements that define individual rationality, and thereby come to hold that it ought to achieve rational unity among all and only its own thoughts and actions, which is to say, within its own first person point of view, which is to say, within itself. I think most philosophers assume, with Locke that such a first person point of view must be the point of view of a single consciousness –​what from here on I will refer to as a phenomenological point of view. In fact, this assumption has generally been shared on both sides of the dispute about personal identity that Locke inaugurated. Animalists who reject his distinction, and who affirm that a person just is a human being (or other suitably endowed animal), take it for granted that each human being (or animal) is the site of a single consciousness, and that the phenomenological point of view of that single consciousness is the point of view from which the agent, now conceived as a human being, deliberates and acts. The reductionist account of personal agency rejects this nearly universal assumption about the first person point of view. It makes a quite different assumption, which attends to the fact that the requirements of rationality are both internalist and individualist in the sense I explained above, namely, that what rationality requires of an individual agent is that it ought to deliberate from all and only its own thoughts, the thoughts that figure in its own point of view. The account denies that the first person relation of ownership, by virtue of which different thoughts are recognized to belong to one’s own point of view, can be equated with the phenomenological relation of having direct access to them within a single consciousness. It affirms instead that the first person relation of ownership must be understood in directly normative and agential terms, in terms of embracing a commitment. In other words, what one ought to take into account, to deliberate and act from, is not just any old item to which one might have direct conscious access, but rather, what one holds to be true and good, which is just the same thing as what one is committed to taking into account. What unites different such commitments within the same

Moral significance of personal identity  233

point of view, so as to make them one and all mine, is to be understood in directly normative terms too, in terms of a further commitment –​a commitment to taking them all as a common basis for my deliberations and actions, as all falling within the scope of the “all” of my all-​things-​considered judgments. According to the reductionist account of personal agency, it is through this active embrace that different commitments come to figure in the same first person point of view from which an individual agent deliberates and acts as one. From here on, I will refer to an agent’s first person point of view, so conceived, as a rational point of view. This talk of a further commitment, to embracing different commitments together, might seem to suggest that there must be a further fact involved in the existence and identity of an individual person  –​something beyond intentional activities and their relations, a great uniter who functions as the executive self who is doing all of this embracing of commitments. The suggestion would be, then, that this further thing is what recognizes its various commitments as one and all mine. But if this further thing is to accomplish all this uniting, it must recognize the scope of its own commitment to achieving overall rational unity. This is impossible without recognizing which intentional activities belong together as the commitments among which such rational unity ought to be achieved; and this in turn is impossible unless the intentional activities themselves incorporate this recognition as part of their contents, for it is only through intentional activities of these kinds, that any such imagined further-​thing-​that-​is-​supposed-​to-​be-​the-​ agent could possibly conceive and implement its commitment to overall rational unity. We may safely infer that it is the commitments themselves that are doing all of the uniting work, by recognizing and responding to one another as forming a common basis from which deliberation and action ought to proceed. And this is how they come to figure in the same first person point of view, and to constitute the life of a single person. I have just described the process of self-​constitution.A person, qua agent, constitutes itself by engaging in certain sorts of intentional activities, namely, embracing various commitments with the aim of achieving the sort of overall rational unity among them that is constitutive of individual rationality. These activities, so directed, constitute the person. Since, according to reductionism, the person is nothing but these activities so directed, these activities just are the person. It follows that the person literally constitutes itself. And there is no whiff of paradox here, about how the person must already be there in order to do the constituting, for the reasons I just explained in the foregoing paragraphs. The process of self-​constitution can be carried out within different boundaries. Intentional activities can be directed at achieving overall rational unity across different human lives, and in such a case there will be a group agent who qualifies as an individual agent in its own right. Intentional activities can be directed at achieving overall rational unity within different portions of a single human life, and in such cases there will be multiple agents within that life. The cases of group and multiple agency vividly illustrate that the rational point of view from which an agent deliberates and acts is not the same thing as the

234  Carol Rovane

phenomenological point of view of a single human being’s consciousness. But it is important to appreciate the underlying reason why these are not the same. The reason why is it does not matter how the sort of overall rational unity that rationality requires of an individual agent is achieved  –​it matters only that there is a commitment to achieving it through some means or other, along with efforts to fulfill that commitment.14 A group agent deliberates and acts from a single rational point of view, which gets forged when the intentional activities through which it constitutes itself recognize and respond to another as forming the common basis of its deliberations. This will typically require speech-​based communication across human lives. Through such speech-​based communication, various commitments that are scattered across the group are able to recognize one another as a common basis for deliberation about what the group should do, and in this way they actually forge the group agent’s rational point of view. It matters not that the unification of these different commitments within this group point of view requires speech-​based communication, just as it matters not whether they all figure in a single unified consciousness. All that matters is that they have the combined aim of achieving overall rational unity among them, by arriving at and acting upon all-​things-​considered judgments that work out their combined normative significance for the question, what would it be best to do in the light of them all? When this happens, there is a single group agent who is literally deliberating and acting as one, and this single group agent has its own group point of view that can be addressed and engaged by others in distinctively interpersonal ways. In contrast, multiple agents within the same human life deliberate and act separately from distinct rational points of view, even though all of the thoughts that figure in their separate rational points of view are accessible from within the same unified consciousness –​from the human being’s phenomenological point of view.This situation would in some ways resemble what used to be called “multiple personality disorder” and now goes by the name of “dissociative identity disorder.” Those who suffer from these disorders exhibit alternate ‘personalities’ who deliberate and act separately, and who can therefore be separately engaged in interpersonal relations. But in the context of these dissociative disorders, the separate functioning of the alternate personalities is facilitated by so-​called “amnesia barriers” –​the personalities do not have direct access through consciousness to one another’s thoughts. Whereas, when multiple agents deliberately constitute themselves within a single human life, they do retain conscious access to one another’s thoughts, because the human being is the natural site of a unified consciousness. But this does not prevent them from forging separate rational points of view from which to deliberate and act. The standard view would seem to be that this must still be a kind of disorder, a kind of pathology, or at least a rational failing. But the standard view overlooks the fact that there is a great deal of recognizably rational activity in the case. Why does it suddenly no longer count as rational simply because it is directed at meeting the requirements of individual rationality within different boundaries than a whole human life? The reductionist account of personal agency insists that it does count as

Moral significance of personal identity  235

such –​hence the claim that there can be multiple agents within a single human life, even though they constitute themselves within a single consciousness. Of course, there can be a commitment to achieving overall rational unity within the whole of a single human life, and then there will be an agent of roughly human size. But note that even in this case, there is a distinction between the agent of roughly human size and the human being, and also between their points of view. The agent has a rational point of view from which it deliberates and acts, whereas the human being has a phenomenological point of view from which various episodes are ‘co-​conscious’. In general, the agent’s rational point of view includes only the normative basis of its deliberations –​only what the agent ought, by their own lights, to take into account in its all-​things-​considered judgments so as to achieve overall rational unity within itself. Even when an agent is committed to achieving such overall rational unity with the boundaries of a whole human life, its rational point of view must be actively forged by embracing various other commitments together as a common basis of deliberation, whereas, the human being’s phenomenological point of view naturally arises with normal biological existence and development. Moreover, there will not be a perfect coincidence of these two points of view: the agent’s rational point of view will exclude many of the conscious episodes that figure in the human being’s phenomenological point of view  –​such as desires and urges for things that the agent does not hold to be good (think of the case of addiction); the agent’s rational point of view will also include some commitments that do not figure within the human being’s phenomenological point of view –​ these would be commitments that are retained and kept track of by other means, such as writing them down.15 The question arises, why should efforts ever be directed at achieving overall rational unity within certain boundaries rather than others? The answer is that there must be something worth doing whose pursuit requires striving to achieve such unity within particular boundaries –​this would be a unifying project. It is not hard to see that some unifying projects would require nearly the whole of a human life –​ these are what moral philosophers generally refer to as “life projects,” examples of which would be long-​term personal relationships, lifelong careers, ongoing institutional, and political affiliations. But other unifying projects might require that a group agent constitute itself –​I think this may have happened in the Manhattan Project, and it certainly can happen in a philosophy department.16 Still other unifying projects might require less than a whole human life, and then different agents might constitute themselves in order to pursue them side-​by-​side within the same life. This may be harder to imagine. Long ago, I suggested that a single human life could harbor a philosopher, a teacher, and a musician, each of whom deliberates and acts only from considerations that relate to its own project.17 My suggestion was not that it is impossible to pursue such projects together by a single agent of human size. My suggestion was that this is possible only when there is a larger unifying project to lead a unified human life. But this latter project really is a project, and therefore it is optional. I myself do not think that the project of leading a unified human life is always feasible, especially in fragmented social conditions. Consider an immigrant

236  Carol Rovane

child who is raised in a very traditional family, but who then gains entry into one of New York City’s elite public high schools. It may well be that such a child will become the site of two very different points of view, one of which is organized around traditional values and activities of the family, the other of which is organized around the liberal values and individualist pursuits that are encouraged at school. The unifying projects for the sake of which group agents constitute themselves do not typically require the resources of the whole of the lives of their human ‘members’. As a result, these lives will be marked by the same kind of rational fragmentation that we find in cases of multiple agency. Some of the intentional activities that occur within each of the human member’s lives will be directed at forging the rational point of view of, and constituting the life of, a group agent, while other intentional activities that occur within that human life are directed at forging the point of view of, and constituting the life of, an agent somewhat smaller than human size. If this seems impossible, just think of what it is like to encounter your friend when they are behind the desk of a bureaucratic institution, and finding that everything that is said and done in that encounter reflects and implements the bureaucracy’s point of view, but then after office closing time, you manage to engage the familiar point of view of your friend over drinks. Again, I am claiming that such rational fragmentation within a human life is not necessarily a rational failing –​not in cases in which multiple agents constitute themselves within a single human life, and not in cases in which group agents draw on parts of many human lives but not the whole of them. Many philosophers are prepared to affirm the possibility of group agents, but not the possibility of multiple agents. This combination of attitudes would make much more sense if we rejected Locke’s distinction and embraced the animalist view of personal identity according to which a human being is a natural person –​not something that constitutes itself, but something whose existence would be biologically given. Then we might allow that such natural persons can together constitute group agents who qualify as merely artificial persons.18 This view would assimilate group agency to collective agency, which is a social phenomenon, the agency of many. Although there may be some room in this view to hold groups –​i.e., collectives –​ responsible, such responsibility could not be assimilated to individual responsibility; moreover, the view leaves room to regard real individuals  –​i.e., the individual human beings who are supposed to be natural persons  –​as the ultimate loci of responsibility for what they collectively do, when they create artificial group agents. In this way, the view manages to acknowledge the possibility of group agency while also retaining some of the special moral significance that commonsense morality and liberal culture attach to human individuals. The aspect of the reductionist account of personal agency that is particularly hard to digest is that it really does reject the view that human beings are natural persons.The account takes both reductionism, and its corollary about self-​constitution, that seriously. It claims that all agents constitute themselves through their intentional activities, and that they all qualify as individuals in exactly the same sense, no matter what their size –​regardless of whether they are group agents, multiple

Moral significance of personal identity  237

agents or agents of human size. In making these claims, it clearly departs from some of our commonsense views about personhood and personal identity. But this does not mean that it is mistaken. It only means that if we become convinced that it is true, then we will have to revise various aspects of common sense –​which is exactly what Parfit maintained in connection with the reduction that he offered. The revisions in the offing will not be confined to the metaphysical dimensions of personhood and personal identity, for there will have to be moral revisions as well.

An initial moral implication We have seen that, in each and every case, an individual person requires reasons to constitute itself as the size agent that it is –​that is, it needs reasons to constitute itself as the particular agent that it is –​and these reasons are supplied by a unifying project, which is the end for the sake of which the agent exists. It directly follows that individual agents are not ends in themselves in the sense that Kantians claim. Of course, agents can still be treated with the sort of respect that Kantians associate with the conception of persons as ends in themselves. But at the same time, they could not coherently view one another as possessing the kind of immeasurable worth that is supposed to attach to an end in itself. Individual agents themselves know that their very existence is for the sake of the ends that their existence makes it possible to pursue.

Comparison with Korsgaard’s view In this section, I will undertake a compare and contrast exercise, between the reductionist account of personal agency and certain aspects of Korsgaard’s views. The point of this exercise is not scholarship, and so I will not quote chapter and verse.19 Likewise, the point is not refutation, but rather to register various points of divergence and disagreement with her, and to begin the process of navigating them.

Irreducible normativity: Laws and constructivism In Korsgaard’s view, the most fundamental act is that of laying down a law for oneself –​an act through which one binds oneself to treat like cases alike. There are several points of convergence between Korsgaard’s account of this activity and the reductionist account of personal agency. Like the act of embracing a commitment, the act of laying down a law for oneself is an irreducibly normative act. Thus the effect of laying down a law for oneself is not that one comes to have a causal disposition to reason and act in accord with that law –​that is, the effect is not that one will, inevitably and predictably, treat like cases alike in exactly the way that the law requires. Rather, the effect is that one holds that one ought to do so. When agents lay down laws for themselves, they do so from the same agential standpoint that they occupy when they embrace commitments. This is the standpoint from which freedom and value are intelligible.

238  Carol Rovane

Korsgaard situates all these claims within a constructivist metaphysics of value, according to which agents are in some real sense a source of normativity. Her suggestion seems to be that the very reality of matters of value depends upon the activities of agents, in which they lay down laws for themselves. Thus in her view, without such activities, there would only be matters of fact –​there wouldn’t be any matters of value because there wouldn’t be anything with irreducibly normative force. Others have raised concerns about whether Korsgaard’s (or indeed any variety of) constructivism can deliver any form of objectivity within the domain of values.20 I  will not rehearse those concerns here. Instead I  will simply reiterate that the reductionist account of personal agency is informed by a naïve realism about matters of value. This naïve realism sees an analogy between our epistemic relations to matters of fact and to matters of value, even though these two domains are accessible from two utterly distinct standpoints. In both domains, to apprehend any such matters is to embrace a commitment. In the domain of facts, to embrace a commitment amounts to holding something to be true; and in the domain of values, to embrace a commitment amounts to holding something to be good and worthy of pursuit. The important point is that, in both domains, there is something that we can get right or wrong, concerning what really is true, and concerning what is really good and worthy of pursuit.21

The requirements of rationality: Their content and force If Korsgaard sees the most fundamental act as laying down a law for oneself, she sees the most fundamental requirement of rationality as the general requirement to lay down laws for oneself. It is very much part of her view that, by laying down laws for oneself, and striving for the consistency that they demand, one achieves a kind of rational and practical unity within one’s whole self. This is not exactly the same thing as the sort of overall rational unity that I claim rationality requires of individuals, which would be achieved by arriving at and acting upon all-​things-​considered judgments. But I don’t think this difference matters very much to the arguments to come. Korsgaard agrees that rational requirements are both individualist and internalist in the senses that I elucidated earlier: they speak to individual agents, demanding of them they achieve unity within themselves; and individuals are subject to them only insofar as they actually recognize and embrace them. Korsgaard also agrees that the requirements of rationality are constitutive, but she goes on to emphasize that this makes them inescapable. The reductionist account of personal agency can accommodate her point about their inescapability to some extent, by reasoning as follows: to be an agent is to occupy the agential standpoint from which the normative force of the constitutive requirements of rationality is apprehended; to apprehend them just is to regard them as binding; if an agent were to cease to regard them as binding, the agent would cease to be an agent, and hence it would cease to be at all.

Moral significance of personal identity  239

But Korsgaard takes the inescapability of the constitutive requirements of rationality to show that their normative force is unconditional, or categorical in Kant’s sense. I simply don’t see why this further conclusion is forced on us. We may allow that the requirements of rationality are inescapable for agents because they are constitutive, and yet we may also regard them as holding with merely conditional force, telling an agent what they must do if they are to be rational. When we regard them as merely conditional, we allow that there could be reasons to cease striving to meet them, and that these reasons could be recognized by an agent while occupying the agential standpoint –​though of course the agent would also have to recognize that acting upon those reasons would necessarily amount to suicide.22 When we view the constitutive requirements of rationality in this conditional way, we are making room for the possibility that nothing in the domain of value carries an unconditional normative force –​that everything of value can be evaluated and re-​evaluated together with, and in the light of, other values. Korsgaard takes a different view. According to her, if there is to be anything of value at all, then there must be something whose value is unconditional. There are several reasons a philosopher might agree with her. They might think that if all values were conditional, this would lead to an unacceptable regress, which can only be ended by positing unconditional value. Relatedly, they might think that values require an ultimate foundation, and that what is foundational cannot be conditional on anything else, but must be the unconditioned condition on which everything else is conditional. They might also agree with Korsgaard that there must be an ultimate source of normativity –​which she herself locates in the self-​constituting activities of agents. In her view, the threat of regress is to be ended, and a foundation for value is to be supplied, via the unconditional value that agents unavoidably place on their own agency in the course of exercising their agency. It is also her view that when agents value their own agency, they take the constitutive requirements of rationality to be unconditionally binding on them. Foundationalism seems to be one of those dividing issues in philosophy that does not admit of a non-​ question-​ begging resolution. Unless one is already a foundationalist, one is not likely to be moved by the sorts of arguments that foundationalists offer in favor of foundationalism. So I’ll just repeat that the reductionist account of personal agency proceeds without recognizing the need for a foundation, and it construes the requirements of individual rationality accordingly, as constitutive but not unconditional. No doubt, the worries that foundationalists have, about how we can possibly navigate a sphere of value that has no ultimate foundation, will continue to worry them. Korsgaard locates the motivation for, and power of, her constructivist metaphysics of value in the hope that it holds out, for supplying the foundation that she seeks. According to constructivism, there would not be any matters of value at all unless there were self-​constituting agents who lay down laws for themselves, and thereby become bound by the normative force of those very laws. The issue of constructivism vs. realism is another deep dividing issue that may not admit of a non-​question-​begging resolution. Even if it does admit of one, it is not on

240  Carol Rovane

offer here. I will simply register that the reductionist account of personal agency is governed by much more modest ambitions than Korsgaard’s. It sees no need to seek an ultimate foundation for value. And it portrays agency not as a source of value, but as a way of responding to value. There is only one last point to make about the content and force of the normative requirements of rationality as Korsgaard understands them. She takes the (allegedly) unconditional requirement to lay down laws for oneself to have a moral corollary, in the form of a Kantian universalizability requirement to lay down laws for oneself only if those laws can consistently be laid down and acted upon by everyone. It might seem that this further moral requirement cannot really be a requirement of rationality per se, because rational requirements are supposed to be internalist and individualist, whereas the moral requirement makes a social reference to all rational agents. But there is no conflict here. The Kantian moral requirement to lay down universalizable laws speaks to individual agents, and it instructs each of them about what they must individually do in order to be fully rational. As far as I can see, this individualism is central to the Kantian moral outlook: so long as one obeys the categorical imperative oneself, there is no rational or moral failing on one’s own part, and this would be so even if it turned out that no one else ever obeyed it. The reductionist account of personal agency does of course recognize a rational requirement of consistency; and it can allow that this requirement provides agents with resources with which to conceive Kant’s universalizability test, and to conceive the particular form of respect that the test helps to isolate. But it does not agree with Kant and Korsgaard that agents have an unconditional moral duty to obey this requirement, in part because it does not agree with them that the requirement of rationality from which it is supposed to issue is unconditional –​it is merely constitutive.This point dovetails in interesting ways with the point I made earlier, that the reductionist account of personal agency undermines the Kantian conception of the person, qua agent, as an end in itself.

Reductionism, self-​constitution, and Locke’s distinction revisited In Korsgaard’s view, agents constitute themselves by laying down laws for themselves. But not any old law will do. It must be a law that can give rise to what she calls a practical identity, examples of which include being a parent, being a teacher, being a citizen. Such an identity-​conferring law would organize the practical life of an individual agent in ways that are quite similar to the ways in which a unifying project organizes the practical life of an individual agent according to the reductionist account of personal agency. These are important points of similarity between Korsgaard’s view of personhood and personal identity, and the reductionist account of personal identity. But the connection goes deeper: Insofar as she intends her claim about self-​constitution literally, this commits her to accepting the central plank of the reductionist account of personal agency, that the existence of a person consists in nothing but certain sorts

Moral significance of personal identity  241

of intentional activities in the right sorts of normative relations. For the only way in which an agent can be self-​constituting is by being identical with the activities through which it constitutes itself, and if this is so, then the existence of that agent consists in nothing but these activities. It is clear that Korsgaard’s preferred way of understanding the distinction between the person, qua self-​constituting agent, and the human being, qua animal, aims to deprive it of any real significance. Indeed, the upshot is almost as if she endorsed the animalist view on which human beings are natural persons. For although she allows that the life of a person does not begin with the biological birth of a human being, but begins only when the human being becomes the site of self-​constituting activities, she nevertheless takes it for granted that whenever these self-​constituting activities occur, they necessarily issue in an individual agent of roughly human size. That is, she does not see any room for the possibility that multiple agents can constitute themselves within a single life.23 Korgaard’s dismissal of the possibility of such multiple agents allows her to overlook the ways in which the thesis of self-​constitution is at odds with any morality that attributes the kind of moral significance to the individual that we find in both Kantian morality and the commonsense morality of liberal culture. The presumption of any such individualist morality is that this moral significance attaches to human individuals. This presumption is easier to vindicate when we embrace the animalist view on which human beings are natural persons. But Korsgaard helps herself to it anyway, by taking for granted that whenever a person constitutes themself, it will have to do so in such a way that its life will coincide with a given human life. I want to pause, now, to explain why I think we should not be persuaded by the sorts of considerations that move Korsgaard to dismiss the possibility of multiple agency. Her view is that the human organism places various constraints on self-​constituting activities. She declares for example, that we have only one body with which to act. In addition, she claims that the individual human being is a site of various kinds of functional unity that work against the possibility of multiplicity –​in particular, the unity of a human being’s perceptual-​motor organization, which is bound up with the unity of a human being’s consciousness. Let me address these claims in turn. It is unclear what is supposed to follow from the claim that we have only one body with which to act. If the implication is supposed to be that only one thing can be done with a given human body at a given time, it is clearly false. I myself have often stood at the stove stirring the pot while at the same time speaking on the phone to a coworker and watching over a child at play just outside the kitchen window. But even if the implication were true, it wouldn’t join the issue, because the possibility of multiple agents does not depend upon the possibility that they could act simultaneously. In fact, it is far more likely that they would take turns, in something like the way Locke imagined when he raised the possibility of a day person and a night person alternating in the same human life.What multiple agents need isn’t separate bodies in the biological sense. What they need is separate points

242  Carol Rovane

of view from which to deliberate and act. And this brings me to Korsgaard’s point that the human organism is the site of natural functional unities. Each human being is naturally the site of a single phenomenological point of view. I have already discussed at length how, and why, the rational point of view of an agent need not coincide with the phenomenological point of view of a single human being’s unified consciousness. But let me clarify further what it means to say that direct awareness in consciousness does not suffice for the normative relation that an agent bears to all and only its own thoughts –​which is to say, the commitments that the agent ought, by its own lights, to deliberate from. Direct awareness through consciousness is really a form of epistemic access, and mere knowledge of a thought does not make it one’s own in the normative sense. One agent may have knowledge of another agent’s thoughts, without thereby coming to regard those thoughts as a normative basis for their own deliberations and actions. This would be a case of social knowledge. And according to the reductionist account of personal agency, such social knowledge is possible within a single unified consciousness. That is, it is possible that within the unified consciousness of a single human organism, multiple agents can forge and maintain separate deliberative points of view, for the sake of pursuing distinct unifying projects that take up less than a whole human life. It is no threat to the separateness and distinctness of their rational points of view, that they enjoy direct epistemic access to one another’s commitments through consciousness. Knowing about one another’s commitments in this intimate way does not necessitate that they should embrace those commitments as a normative basis for their own deliberations and actions –​they can still deliberate and act separately from separate commitments within the same unified consciousness, and these separate commitments will respectively constitute their separate rational points of view. Although Korsgaard dismisses the possibility that there can be multiple agents within a single human life, she does allow that there can be group agents spanning many human lives. This combination of attitudes puts her very much in league with the philosophers who view human beings as natural persons and who view group agents as artificial agents who are created through social means, by natural persons. For even though she does not view human beings as natural persons, she does nevertheless think that insofar as any human being is a site of personhood at all, it will be the site of an individual agent of roughly human size, and furthermore, this agent of human size will necessarily remain intact as the individual that it is, even when it participates in a collective effort with other such agents to form a group agent. The upshot is that even though Korsgaard holds that agents are self-​constituting, she regards the boundaries that mark one individual agent of human size off from another as firm  –​they cannot be renegotiated in the ways that the reductionist account of personal agency claims.This is what enables Korsgaard to view agents of human size as holding the kind of moral significance that individuals are supposed to have according to both Kantian morality and the commonsense morality of liberal culture. In the context of group agency, such agents of human size appear to be available as ultimate loci of responsibility for what group agents do. They also

Moral significance of personal identity  243

appear to be more appropriate targets of Kantian respect than group agents are, for they are portrayed as somehow more basic and more real –​for they constitute themselves, whereas group agents are constituted by them. In contrast, the reductionist account of agency does not see the boundaries that mark one human life off from one another as setting any systematic or insuperable constraints on how individual agents can constitute themselves with respect to those boundaries. It simply isn’t true that each human life must be the site of one and only one self-​constituting agent of human size. Multiple agents can constitute themselves within a single human life; and group agents can constitute themselves across human lives; and all of these individuals are individuals in exactly the same sense; none is more basic or more real or less artificial than any other; and all of them exist for the sake of things worth doing.

Moral dimensions of a metaphysical disagreement Parfit saw his reductionist account of personal identity as inviting a consequentialist moral orientation. One important insight of consequentialism is that what actually happens when we exercise our agency is just as important as the principles on which we act. Another important insight is that we cannot always set aside considerations to do with the sum total of goods and harms when we are solving the moral problems that arise for us. But of course, it is only agents who can apprehend these consequentialist considerations, just as it is only agents who can apprehend the considerations that individualist morality puts at the center. For these reasons, and more, the reductionist account of personal agency is prepared to accord a distinctive kind of moral importance to agents, and more specifically, to individual agents. The account fully acknowledges that it is only individual agents, who possess their own rational points of view, who can raise moral questions at all, or engage in moral reflection. And it acknowledges that if individual agents can do these things, it is because they recognize that there are other points of view from which things matter besides their own. And it acknowledges that, insofar as individual agents recognize one another’s rational points of view, they can praise and blame one another and hold one another morally responsible, and more generally, they can engage one another for all sorts of purposes, including the purposes of moral inquiry and moral argument. Yet although the reductionist account of personal agency does allow that individual agents have a distinctive kind of moral importance, we have seen that it cannot accommodate what might be called the individualism of Kantian morality. So far I have only described one respect in which Kantian morality is individualist, which is that it portrays the categorical imperative as speaking to individual agents in the way that all rational requirements do. But here is another, much more important, respect in which Kantian morality is individualist: it portrays each individual agent as possessing a kind of moral worth all on its own, conceived apart from any comparative or aggregative evaluation. It is this second respect in which Kantian morality is individualist that makes it is so hard to reconcile Kantian moral

244  Carol Rovane

insights with the moral insights of consequentialism. And it is also the respect that the reductionist account of personal agency cannot accommodate. But first let me elaborate it more fully. The Kantian conception of an individual’s worth is most explicit in the second formulation of the categorical imperative, which instructs us never to treat agents as mere means, but always to treat them as ends in themselves. Owing to the centrality of laws in Korsgaard’s larger normative project, she is naturally much more interested in developing the moral insights that emerge with the first and third formulations of the categorical imperative, concerning universalizability and autonomy. Yet she surely does attach importance to what I’m calling the individualism of Kantian morality. This was already apparent in her early resistance to the consequentialist implications of Parfit’s psychological reductionist account of personal identity. But it is also implicit in her continued focus on the Kantian theme of laws, and its connection to the theme of universality. For I think the individualism of Kantian morality is really the flip side of its universalism, as I shall now explain. Kant’s categorical imperative is universal in more than one sense. First, it has a universal force –​ i.e., everyone is subject to it. But second, it also has a universal content –​ i.e., it requires each to take everyone into account in the very same way.This is not the aggregative way that consequentialist morality demands, but a way that preserves the separate moral importance and worth of each individual agent, taken in itself, apart from all else, and in a way that is not derivative of or conditional upon the particular ends that an agent might or might not pursue. This abstracting away from the particular ends that particular agents might embrace and pursue is accomplished by focusing on the general idea of a universal law. So long as each agent embraces and pursues ends with the thought, or perhaps I should say (if it were coherent to say this), the intention, that all others should do the same, then its pursuits will never be intended to come at the expense of, or to hinder, the pursuits of any other agent. This attitude amounts to a form of universal respect that is distinctively Kantian. And my point is, this form of universality recognizes and preserves the individual importance of each agent. The moral preoccupation with this distinctively Kantian form of respect helps to explain why it is so important for Kantians to show that Kantian-​style moral obligations are unconditional. It is not only because they presume, with Kant, that anything deserving to be called a moral value should command in an unconditional way, so that it trumps all other values; and it is not only because they might wish to demonstrate to the moral skeptic, and to Hume’s Sensible Knave, that they cannot escape the normative force of moral demands. It is also because they wish to preserve the distinctive content of Kantian moral obligations. As I’ve just explained, Kantian respect abstracts away from the particular merits of the particular ends that agents may or may not pursue. This abstraction is essential for valuing agents as ends in themselves; and once the abstraction is made, it would seem that there is nothing for our moral responses to agents, now conceived as ends in themselves, to be conditional upon.

Moral significance of personal identity  245

At many turns in this chapter I have registered that the reductionist account of personal agency takes it for granted that no matters of value present us with unconditional normative demands. I have also registered that it leaves no room at all for the Kantian conception of the agent as an end in itself. For when it says that the existence of an agent is always a product of effort and will, it also says that an agent exists for the sake of the ends that its existence makes it possible to pursue, and that this is true no matter what size the agent may be –​whether it be an agent of human size, or smaller or larger than that as we find in the case of multiple and group agents. When we conceive the very existence of an agent in this way, then the Kantian abstraction that would focus moral attention on respect for individual agents as an end in themselves simply doesn’t make any sense. This upshot might seem to be morally troubling. But I  think it would be quite wrong to see it as ushering in a dark moral vision. It is a vision that sees all exercises of agency as normative responses to matters of value that are in some sense objective, including the exercises of agency through which agents themselves are constituted –​they literally constitute themselves in order to realize ends that they apprehend to be worthy of pursuit. So you might say that on this metaphysical vision, the world calls forth agents, and this is possible because the world contains within it matters of value, along with the agential capacities through which matters of value can be apprehended and responded to. And I think it is worth emphasizing that this metaphysical picture gives a kind of primacy to personal pursuits that can never entirely give way to the demands of consequentialist morality, even though the counter-​balancing force that it puts in place against the aggregated weight of those demands is quite different from the force of a Kantian categorical imperative, and related conceptions of liberal rights and duties. Korsgaard’s constructivist metaphysics of value portrays things the other way round from the naïve realism that informs the reductionist account of personal agency. It is agential capacities that bring forth value into the world, because it is through their exercise that laws are laid down –​and without the laying down of laws there would be no normativity, but only a causal flow of events. I am truly doubtful that this vision is morally brighter, for there is a danger that the resulting freedom of the agential standpoint is not subject to any real normative constraints (I’m hardly the first to point this out). It is a surprising outcome, that an exploration of the idea that Korsgaard puts at the center of her normative project  –​namely, the idea that agents are self-​ constituting –​should ultimately serve to speak against, rather than in support of, her Kantian moral vision. Yet I don’t regard what I’ve offered here as an out-​and-​ out refutation of that vision. As far as I can see, nothing bars Korsgaard from pursuing a Kantian project that begins from a Kantian premise. That is, nothing bars her from asking, what must agents be like if they are to be subject to unconditional Kantian-​style moral obligations? In fact, I think that this is more or less how she generally does proceed. And it may well be that her constructivist metaphysics of value supplies a more compelling answer to her guiding question than any other

246  Carol Rovane

alternative. But even if that there were so, it would not suffice to refute whether the opposed metaphysics of value informs the reductionist account of personal agency, or its negative implications for Kantian morality.

Acknowledgments I would like to thank Akeel Bilgrami, Michael Della Rocca, Samuel Preston and Richard Strier for reading this paper at various stages, and for hearing me out on the issues in conversation too. And I would like to thank the anonymous reviewers for their helpful comments, and Andrea Sauchelli for bringing the volume together.

Notes 1 The fullest, and canonical, development of Parfit’s early view is to be found in Parfit [1984]. 2 See Korsgaard [1989]. 3 See Korsgaard [2009]. 4 The fullest elaboration of this account is in Rovane [1998].There, I call it the “normative analysis of personal identity.” 5 This naturalistic attitude is typified by Jerry Fodor. See Fodor [1990]. 6 I should clarify that functionalists like Fodor reject all-​out physicalist reduction, on the ground that the same functions can in principle have different physical realizations. Nevertheless, they see functionalism as affording a robustly scientific account of the mind, replete with the sorts of laws of the mind that Hume actively thought to discover, when he set out to do for the study of mind what Newton had done for the study of motion. For such scientists of the mind, the point of portraying intentional attitudes as causal dispositions is precisely to portray them as falling under causal laws. 7 The conception of commitments, as irreducibly normative, was elaborated for the purposes of decision theory by Isaac Levi. See, for example, Levi [1984] and Levi [1990]. It also figures centrally in Akeel Bilgrami’s account of knowledge in Bilgrami [1997], and in my own account of personal identity in Rovane [1998]. 8 It might be protested that the mere fact that we can fail to live up to our commitments doesn’t necessarily suffice to show that they are not reducible to certain sorts of causal dispositions –​after all, there are many reasons why a given causal disposition might fail to be manifested. But when we find that a causal disposition goes unmanifested, we do not suppose that it ought to be manifested, and we do not criticize the thing that possesses it for its failure to be manifested –​we simply look for the causal explanation of why it is not manifested. I thank Michael Della Rocca, Samuel Preston, Richard Strier, and an anonymous reviewer for pressing this point. 9 Although Hume famously argued that we can never derive an ‘ought’ from an ‘is’, he did not posit the kind of metaphysical gap I’m referring to here. What he meant was that ‘oughts’ only arise with desires, which is to say, they are not settled by our beliefs about how things ‘are’. All the same, Hume regarded desires as part of the course of nature that science studies, and this means that he did not posit the kind of metaphysical gap that I have in mind, that goes together with irreducible normativity. On the contrary, the ultimate implication of the Humean view is that morality emerges to be a branch of psychology, and hence reducible to something non-​normative. See Bilgrami [2006] for a particularly radical account of what it means to say that we cannot cross the metaphysical

Moral significance of personal identity  247

gap between the domain of values and the domain of facts. There he argues that we cannot even make sense of a supervenience relation between them. 10 See Hampshire [1975], Strawson [1962] and Bilgrami [2006]. 11 One reason why it can be so difficult to appreciate the fact, and also the significance, of irreducible normativity is that we can and do attribute intentional attitudes in the spirit of Dennett’s “intentional stance,” where the point is to causally explain, and to predict, someone’s (or something’s) behavior. See his Dennet [1990]. But then we are not viewing others as agents in the sense under discussion, but just as parts of the larger course and flow of nature. That is very different from attributing commitments to others for the purpose of engaging their points of view in conversation, argument, etc. My point here is that it is only in the context of such engagements that their agency is fully present to us, and fully apprehended by us. See Rovane [1994] for further discussion of Dennett, and Bilgrami [2006] for a rich discussion of this evaluative conception of agency. 12 The fullest version of this argument appears in Rovane [1998], but I have rehearsed its main points in many articles as well –​most recently in Rovane [2017]. 13 The requirement of closure is particularly controversial, because the implications of one’s attitudes are, presumably, infinite, and it would seem to follow that the requirement violates Kant’s dictum that ought implies can.The reason I nevertheless include it as a candidate requirement is that it seems to articulate a rational ideal that we comprehend, and I think we do implicitly appeal to this ideal as an open-​ended critical standard whenever we fail to accept the implications of our attitudes. I thank an anonymous reviewer for pressing me on this point. 14 These efforts need not be successful, and generally can’t be entirely successful because the ideal of overall rational unity is too demanding. But in accord with my discussion above of the irreducibly normative character of commitments, an agent possesses a commitment even if it doesn’t entirely live up to it, so long as it takes its perceived failures to do so as grounds for self-​criticism. 15 See Rovane [1993] for further discussion. 16 See Rovane [2004] for further discussion of the Manhattan Project, and Rovane [1998] for further discussion of the philosophy department. 17 See Rovane [1998]. 18 This view is defended by List and Pettit [2011]. See Rovane [2014] for a critical discussion of them. See Rovane [2017] for a more developed account of the difficulties that stand in the way of conceiving group agency as a social phenomenon. 19 My main sources are Korgaard [1989], Korsgaard [1996] and Korsgaard [2009]. 20 For early efforts see the commentaries included in Korsgaard [1996]. See also Gibbarrd [1999]. 21 In McDowell [1985] John McDowell assimilates values to secondary qualifies. In doing so, he retreats from a naïve realism about value to a slightly more subjectivist view, which allows that the very idea of a value might make an essential reference to the kind of thing that can apprehend matters of value, in much the same way that the very idea of a color might make an essential reference to the kind of thing that can perceive colors. Prima facie, this view coheres well with the claim that values can be apprehended only from the agential point of view. My own reservation about it –​my own preference for naïve realism, at least at this point in my own meta-​ethical inquiries –​concerns the extent to which it paints the domain of value with the wrong kind of subjectivity. When we agree with Locke, we locate the perceived nature of secondary qualities entirely in the mind of the perceiver –​this is so even though he defines such qualities in terms of a disposition to cause such subjective perceptions. The fact that certain aspects of reality cannot be apprehended, except by the right kind of knower –​the fact that it takes a certain kind

248  Carol Rovane

of knower to know certain things –​doesn’t by itself suffice to show that those things are subjective. It only shows that not every kind of knower can know it. Think, for example, of chicken sexers. 22 It may seem paradoxical –​or somehow impossible –​that one could recognize reasons to leave the agential standpoint from within the agential standpoint. I myself do not really see the problem here. The situation is akin to the one Parfit envisaged when he argued that if one wishes to be immune from the rational force of coercive threats one may have reason to take an irrationality pill. According to a self-​constitution theorist taking such a pill would, in effect, be a temporary and reversible suicide (see Parfit [1984]). I thank an anonymous reviewer for flagging this issue. 23 She explicitly mentions and dismisses the possibility in Korsgaard [1989]. But throughout most of her work, what we find are descriptions of the process of self-​constitution as taking place within a human life that is characterized by a great deal of natural unity. Such descriptions are particularly prominent in Korsgaard [2014].

Bibliography Bilgrami, A. [2006] Self-​Knowledge and Resentment, Harvard University Press, Cambridge, MA. Dennett, D. [1987] The Intentional Stance, MIT Press, Cambridge, MA. Fodor, J. [1990] A Theory of Content and Other Essays, MIT Press, Cambridge, MA. Hampshire, S. [1975] Freedom of the Individual, Princeton University Press, Princeton, NJ. Korsgaard, C. [1989] “Personal Identity and the Unity of Agency: A Kantian Response to Parfit,” Philosophy and Public Affairs 18(2): 103–​31. Korsgaard, C. [1996] The Sources of Normativity, Cambridge University Press, Cambridge, UK. Korsgaard, C. [2009] Self-​Constitution: Agency, Identity, and Integrity, Oxford University Press, New York, NY. Korsgaard, C. [2014] “The Normative Constitution of Agency,” in M. Vargas and G. Yaffe, eds., Rational and Social Agency:  The Philosophy of Michael Bratman, Oxford University Press, New York, NY. Levi, I. [1984] Decisions and Revisions: Philosophical Essays on Knowledge and Value, Cambridge University Press, Cambridge, UK. Levi, I. [1990] Hard Choices: Decision Making Under Uncertainty, Cambridge University Press, Cambridge, UK. List, C. and Pettit, P. [2011] Group Agency: The Possibility, Design and Status of Corporate Agents, Oxford University Press, New York, NY. McDowell, J. [1985] “Values and Secondary Qualities,” in D. Honderich, ed., Moral Objectivity, Routledge Press, London, UK. Parfit, D. [1984] Reasons and Persons, Oxford University Press, Oxford, UK. Rovane, C. [1993] “Self-​Reference:  The Radicalization of Locke,” Journal of Philosophy 90(2): 73–​97. Rovane, C. [1994] “The Personal Stance,” Philosohpical Topics 22(1/​2): 351–​96. Rovane, C. [1998] Bounds of Agency: An Essay in Revisionary Metaphysics, Princeton University Press, Princeton, NJ. Rovane, C. [2013] The Metaphysics and Ethics of Relativism, Harvard University Press, Cambridge, MA. Rovane, C. [2014] “Group Agency and Individualism,” Erkenntniss 79(S9): 1663–​84. Rovane, C. [2017] “Is Group Agency a Social Phenomenon?,” Synthese https://​doi.org/​ 10.1007/​s11229-​017-​1384-​1. Strawson, P. F. [1962] “Freedom and Resentment,” Proceedings of the British Academy 48: 1–​25.

12 NON-​IDENTICAL AND IMPERSONAL1 J. David Velleman

We have an obligation to ensure that when children are born, they are born into good enough circumstances. In many cases, though, we cannot significantly improve the circumstances into which people will be born in the future without changing who gets born. Obviously, if a financially strapped couple puts off child-​bearing until they are better situated, they are acting in the interests of their future children only de dicto —​the interests, that is, of whoever ends up satisfying the description “their children” —​since the children who are born well-​off will not be the ones who would otherwise have been born poor. Less obvious but equally true is that our efforts to slow climate change will benefit future generations only de dicto as well, since reducing our use of fossil fuels for transportation, heating, and cooling will affect who copulates with whom and when, which eggs get fertilized by which sperm, and consequently who is born. After several generations, there may be no one on Earth who would have incurred the costs of climate change if we had let it take its course; and if we do let it take its course, there may be no one who would have been better off if we hadn’t. So why not let climate change take its course? By the time any significant costs accrue, the world will be populated by people who wouldn’t have benefited from our preventive measures anyway —​people, indeed, who will owe their very existence to our not having taken those measures, which in preventing the costs from accruing would also have prevented those people from being born. No one will be worse off for our having shirked our supposed obligations to future generations. Why, then, should we suppose that we have any such obligations, in the first place? That’s the non-​identity problem, made famous by Derek Parfit, who named it for the non-​identity between the people who would exist if we took action and the people who would exist if we didn’t, neither of whom would be benefited or harmed by our choice.2

250  J. David Velleman

One way around the problem would be to adopt a non-​comparative conception of harm and benefit. Such a conception might be illustrated with cases like that of the mountain-​climber Aron Ralston, who amputated his own arm in order to free himself from a fallen boulder under which he would otherwise have died of starvation. Amputating his arm didn’t make Ralston worse off than he otherwise would have been, since he otherwise would have died a painful death, and yet we want to say that it harmed him. Maybe an amputation is harmful per se, irrespective of the total difference it makes to the victim’s well-​being.3 Still, a more plausible explanation for the intuition that Ralston was harmed is that the amputation harmed him in one respect, as to his arm, while benefiting him in another. Although the amputation didn’t make him worse off on balance, because the harm was less than the benefit, it did make him worse off with respect to his arm than he otherwise would have been —​a counterfactual contrast that would seem to underlie the intuition that it was a harm. So this case doesn’t support the non-​comparative conception, after all. There is much more to be said about the concepts of harm and benefit —​I’ll say some of it in a moment —​but it looks like we cannot use those concepts to evaluate the effects of our choices on future people. The concepts of harm and benefit attach to the difference our choices would make for particular people (they are “person-​affecting” concepts, as Parfit put it), and so they cannot be used to evaluate choices that might lead to our creating different people. Despite the problems inherent in person-​affecting moral concepts, there is a trend in moral philosophy toward “relational” theories, which are fundamentally person-​affecting. Stephen Darwall’s theory of the “second-​person standpoint” is one example. According to Darwall, our moral obligations arise from demands that others, singly and as a “moral community”, can make upon us and for which we are answerable to them.4 The trend goes back at least as far as P.F. Strawson’s “Freedom and Resentment”, which left in its wake the widespread assumption that being eligible as a target of inter-​personal “reactive” attitudes —​attitudes of the sort that are responsive to harm and benefit —​is the fundamental criterion of moral agency.5 In contrast to these views, I believe that morality is no respecter of persons, though it is indeed a respecter of personhood. Classical utilitarianism takes no account of particular people: more utility is more utility, less is less, no matter whose utility it is. In assessing an action’s impact on everyone’s welfare, utilitarianism interprets ‘everyone’ de dicto, judging an action to be permissible so long as the population of the possible world where we take the action is at least as happy as the population of the world where we don’t, a comparison that ignores the personnel of those populations. Since the people we would bring into existence if we did nothing about climate change would be less happy than the people we would bring into existence if we did something about it, the Principle of Utility yields an obligation for us to do so. Non-​identity is no problem. Unfortunately, the Principle of Utility also yields some very implausible results when applied to population ethics. The version of the principle that calculates the general welfare as a sum requires us to make as many babies as we can, so long as the

Non-identical and impersonal  251

next baby would enjoy some positive amount of utility, however small, and would not detract as much or more from other people’s utility as it would add to the total with its own. The thought that we are morally obligated to bring about this result is what Parfit called the Repugnant Conclusion. The alternative version of the principle, which calculates the general utility as an average, forbids us to create any people with lower-​than-​average utility. Maybe that’s not a problem in the current state of the world, given how many people live in desperate circumstances: the average is a dismally low bar below which the population probably shouldn’t be increased, by anyone’s lights. Yet if the average level of utility in the world somehow rose, the Principle of Average Utility would begin to forbid the creation of happy people simply because their happiness would be below average. Many philosophers find it implausible to prefer a small but deliriously happy population over a larger population whose members are merely ecstatically happy. At this point, however, my moral intuitions give out, because I don’t understand how or why to evaluate population size per se. In my view, the solution to the non-​identity problem is to give up the Principle of Utility in favor of the Categorical Imperative. Now, one might worry that Kant’s principle cannot handle population ethics any better than Bentham and Mill’s, because the formulation most suited to the case —​the Formula of Humanity —​ requires us to respect people as ends-​in-​themselves as well as means. The apparent problem is that the agent of procreation must put his or her maxim to the test of the Imperative before there is any person to respect, so that trying to conceive a child under the wrong circumstances cannot represent a lack of respect for that as-​ yet non-​existent person. This worry is misplaced, because Kantian respect is not in the first instance an attitude toward individuals. To take a mundane example, not flushing the toilet is disrespectful to the next person who will use it, even if the toilet is in a public restroom and the identity of the next user is not only unknown but indeterminate. Clearly, the obligation to the next person is de dicto, and so the immediate target of respect must be a description, not an individual:  it’s an obligation to (quote, unquote) ‘the next person’, and only by way of that description to the person who ends up being next. This interpretation is supported by Kant’s language in stating this formula of the Categorical Imperative. What Kant says is that the respect owed to a person is respect for humanity in that person. I think that this form of respect can also be paid to humanity in the abstract without being targeted at any particular instance. An individual can therefore intercept moral respect or disrespect that wasn’t aimed specifically at him. The next user of the toilet cannot be personally disrespected, but personhood can be disrespected, and the next user will subsequently end up at the receiving end of that disrespect. Similarly, future generations may find themselves at the receiving end of disrespect we show for personhood by letting climate change take its course. Failing to prevent wide-​scale flooding, famine, forced migration, and social unrest is disrespectful of human aims, which will be thwarted, human capacities, which will be

252  J. David Velleman

stunted, and human lives, which will be lost. It’s irrelevant that the subjects of those thwarted aims and stunted capacities won’t be the same as the people whose aims and capacities would have flourished in the alternative. The impersonality of Kantian respect accounts for the possibility of Kantian duties to oneself. Humanity commands my respect wherever it may exist, in myself no less than in anyone else, though of course also no more. Philosophers’ reluctance to countenance duties to oneself rests on the sense that they are self-​regarding, whereas morality is supposed to be a constraint on regard for self. But Kantian duties to self are not self-​regarding in any untoward sense, because they are not person-​regarding at all; they are personhood-​regarding, as moral duties ought to be. There is no favoritism toward oneself in respecting personhood in all its instances including one’s own. We can now think about our obligations to future generations without reaching untoward conclusions, repugnant or otherwise. The question becomes: What kind of circumstances does the dignity of personhood require us to provide for whatever instances of personhood we will end up creating? I won’t try to answer that question here. I’ll merely say that it strikes me as a morally sound way of thinking about future generations. Notice that non-​identity has turned out to be no problem for either utilitarian or Kantian ethics. The reason why non-​identity seemed to be a problem, in the first place, was that we tried to understand our obligations to future generations in terms of benefit and harm, whereas those terms play no role in either of these moral theories —​nor, for that matter, in any moral theory that I know of —​despite playing such a large role in everyday moral discourse. The reason why harm and benefit play no role in moral theory, I assume, is that their application is problematic across the board, not just in special cases such as population ethics.6 There are bad things that supplant or prevent even worse things, as in the case of Aron Ralston; there are also good things that supplant or prevent even better things; and it is therefore difficult to identify the appropriate baseline for the comparisons implicit in the concepts of harm and benefit. I’ve read that winners of Olympic bronze medals are much happier than winners of silver medals, because they compare their results to different baselines; but is there a fact of the matter as to which group has benefitted more? I now want to turn to a more serious problem with the concepts of harm and benefit —​more serious, that is, than the specialized problem of non-​identity or the general problem of the baseline. In his discussion of “what matters in survival” —​or as I will call it, selfhood —​ Derek Parfit considered the possibility of undergoing fission by being split into two qualitatively identical copies.7 He asked not just how his present, unitary self should regard the future products of fission but also how they should regard one another. Parfit judged that immediately after the split, he (either one of him) would regard his relation to his twin as almost as good as selfhood —​not numerical identity, of course, but a relation sufficient for personal survival. Immediately after the fission,

Non-identical and impersonal  253

Parfit thought, a twin should not be troubled by the news that he is about to die, provided that the other will survive, because the survival of his twin will be almost as good as his own. His twin has all the same attitudes and traits as he does and could therefore be counted on to complete his projects, continue his friendships, strive toward his ideals, and so on. In effect, his life would get lived by proxy. The consequences of this view for the case of fission strike me as a reductio of the view. A future person-​stage cannot be me by virtue of sharing my traits and attitudes or even my past. Someone who is merely like me isn’t the genuine article: at best, he’s a gifted impostor. What I want in wanting to survive is a future I can look forward to —​“can” in the sense that I am entitled to look forward to it, because my looking forward to it will be correct in some respect. I don’t mean that I  want to be clairvoyant. Mere survival is compatible with being completely mistaken about what the future will be like. In looking forward to the future, though, I seem to be somehow looking into it, and a future to look forward to must be a future into which I can look, whether or not I see it correctly. My forward view must be correct at least insofar as it seems to have access to the future, however obscure or distorted. The question is what sort of access it seems to have. When I look forward to the future, I don’t just imagine experiencing it; I imagine the experience as including a sense of the future’s being as I had imagined.8 This imaginative maneuver sounds difficult at first, but the difficulty is rather to avoid it. I can easily imagine being surprised by the next person who appears at my office door, but I cannot imagine being surprised by it’s being a particular person, because my image of that person’s appearance naturally includes my remembering having hereby imagined it and hence being prepared for it to be him. I thus imagine that this very image will have survived in my memory to meet its fulfillment in experience. I may of course turn out to have been wrong about who it would be; but even in that case, I may still have been right about remembering who I imagined it would be. And that is the access to the future I must have in order to be entitled to look forward to it: I must be able to send my image of future experience into future experience, via memory, so that I can experience the future with that image at the back of my mind, so that I can be justified in imagining it now as having the survival of this image in the background. I must be able to picture the experience “from the inside” not just by picturing it from the perspective of the subject but by picturing it from a perspective containing the picture itself, superimposed on the experience for comparison. Many of our forward-​looking attitudes tend to include, in their intentional objects, their own survival in this sense. We desire things envisioned as palpably gratifying this very desire, which thus envisions surviving to be gratified. In hoping for things, we hope to receive them with the open arms of this hope; in fearing things, we fear meeting them on legs trembling with this fear; and in planning things, we plan to do them on the basis of this plan. My mental life is a hubbub of antiphonal voices, calling in anticipation of a response that will be heard as answering that remembered call, responding in remembrance of a call issued in anticipation of that response. These attitudes are

254  J. David Velleman

investments in the future, in the sense that they anticipate surviving until they can be discharged, resolved, paid off. And what we want, in wanting to survive, is to be in a position to make such investments, in the first place.The lack of a future would impoverish our present, not just by ruling out future plans, but also by reducing us to desires whose intentional objects shouldn’t include their own gratification, hopes whose objects shouldn’t include their own vindication, and so on —​attitudes that are not entitled to anticipate a resolution. The phenomenon I’ve been describing also provides our sense of having numerical identity through time, though I agree with Parfit that there is no numerical identity to be had. Like Parfit, I am a four-​dimensionalist about identity through time, including the identity of people. That is, I believe that I’m not an enduring thing, wholly present at every moment of my existence (as the endurantists say), but rather a succession of momentary subject-​stages in a causally connected series of mental states. Nevertheless, I can still think of myself as an enduring thing because of the way that first-​personal reference is transmitted through time. Suppose I am working at the library and I think, “I’ll work for another hour and then I’ll call it quits for the day.” The second ‘I’ in this statement refers, not to me as I plan, but to a person in the future who quits work an hour from now. So how do I manage to think of that future person in the first-​person? What makes him “me” to me? In order for this plan to take effect, it has to be deposited in my memory now, and later retrieved by that future person. It cannot just sit in memory unchanged, however: over the next hour, an occasional glance at my watch will occasion an update to my plan, so that in half an hour it will be, roughly, “I’ll quit in half an hour”; and after another 20 minutes, “I’ll quit in 10 minutes”; and then “I’ll quit in just a moment.” These updates will not be new plans, fresh decisions to quit at a particular time; they will merely be revisions in my original plan. Over the course of the hour, the plan will have maintained ongoing reference to a future moment, until it arrives at the reference that makes it ripe for execution, namely, the indexical ‘now’, as in “Now I’ll quit.” Only at that point will it be actionable. Also necessary for the plan’s execution, of course, is the indexical reference to the agent who is to execute it. “Now David Velleman will quit” is not actionable, either, unless it is transposed into the first-​person. (“David Velleman will quit? Oh –​ that’s me.”) Unlike the reference to the time, however, the reference to the agent will not need to be transposed, because the plan will have specified from the beginning of the hour what “I” would do at the end. But why? Why will the plan have referred to its executor in the first person all along? I suppose the plan could initially have specified what “he” would do in an hour —​“he” meaning that future subject who will find himself with a plan to stop working at the moment that, from his perspective, will be “now”. If the plan had referred to him in the third person, however, it would have needed not only that final temporal update, from “in a moment” to “now”, but also a personal update,

Non-identical and impersonal  255

from “he” to “I”. And although the process of temporal updating is indispensable for marking out the interval from the time of decision to the time of execution, a process of personal updating would be pointless. I might as well install the first-​ person pronoun in the plan at the start, confident that when “in an hour” has turned to “now”, the pronoun ‘I’ will refer to the subject in possession of the plan, who will of necessity be the subject to whom I must delegate its execution. I can refer to that future person in the first-​person because I am framing a plan for execution at a time when he will have become the referent of ‘I’. That future subject is someone I can think of as “me” because I can frame first-​personal thoughts of which he will become the subject when it matters. This use of the first person is like my use of the second person in the voicemail message on my phone.9 The message “You have reached the mobile phone of David Velleman,” is addressed to whoever ends up hearing it and thus occupies the position of its audience, addressable as “you”. I suppose I could likewise frame my plan in the second person: “If it’s noon and you’re still working, then quit.” I might even write an abbreviated reminder on a slip of paper, using the imperative mood, which is implicitly second-​personal: “Take your lunch break at noon.” But a plan is not exactly like a written reminder. The future person who was intended to act on the written reminder would be the one reading it at noon, it’s audience, hence addressed by the present writer in the implicit second person; the future person who is intended to act on my plan will be the one thinking it at noon, its subject, addressable by that thought in the first person. Being able to refer to present and future subjects as “me” makes it seem that they are one and the same. Hence the illusion that I am an enduring thing. This illusion is reinforced by the way visual perspectives are superimposed when I  imaginatively anticipate a future occasion, such as my quitting work in an hour, as planned. I imagine that occasion from the first-​person point-​of-​view. And why? Conceivably, I could imagine the occasion from a bird’s-​eye perspective that includes the top of all the heads in the reading room where I  am working. But one of those heads will retain a remnant of this image, which will have prepared it for the occasion, and prepared it well —​or so I imagine, since I necessarily imagine the occasion as corresponding to the way I imagine it. For example, I may imagine the scene from a perspective in which the door is to the left, so that I will find myself with an image of the door as being to the left, so that I will turn in that direction when executing my plan. I thereby project myself into that future perspective, as if I —​the subject projecting it —​will be the one to occupy it rather than being the one depositing an image of it in memory to be retrieved by that future subject. Perceptual perspective is similarly preserved in memory. Although what was experienced as present is remembered as past, what was experienced from the first-​person perspective is also remembered from the first-​person perspective. Now, memories would be misleading if they represented past events from the temporal perspective from which they were experienced —​represented them, that is, as seen currently rather than having been seen previously. But because they represent events as having been previously seen, they can represent them as having been seen here

256  J. David Velleman

before my eyes without conflicting with present experience, in which something else is being seen now. And because they represent events as here before my eyes (albeit previously), their subject is remembered in first-​person position, as “me”, simply because of my having inherited temporally updated visual images from him. For what it’s worth, introspection on my own anticipatory imaginings reveals that I don’t distinguish between the present image and the future experience that it prefigures. It is as if I see the occasion in anticipation, just as I see past occasions in memory —​this despite the fact that anticipatory images, unlike memorial images, are not stored impressions of the occasions they represent. Undue faith in the mechanisms of memory might lead me to say that I literally “see” past occasions, as if the impulses from my optic nerves had merely taken a time-​out in memory before returning to impinge on my consciousness again. But no amount of faith in my powers of foresight could lead me to say that I literally see future occasions; and yet I seem to be literally fore-​seeing them, just as I literally re-​view occasions from the past. In both experiential memory and experiential anticipation, I regard a mental image as lining up with a similar experience, either because it was, so to speak, peeled off a past experience or because it will be overlaid on a future experience for comparison. I seem to be living in a tunnel, with a past visible at one end, in memory, and a future visible at the other end, in imagination.10 It’s as if I have gone from viewing up-​close what I am now re-​viewing at a distance in memory; and as if I am going to see up-​close what I am now fore-​seeing at a distance in imagination. In short, I seem to be moving through the tunnel. In fact, however, this whole experience may be happening to the subject of a single moment. He is retrieving a scene experienced and deposited in memory by the subject of an earlier moment; and he is imagining a scene and depositing it in memory to be retrieved by the subject of a later moment. And none of these subjects is moving at all: the movement is an illusion. Parfit’s fission twins are not on first-​personal terms in this sense. Neither of them can resolve the mental phrases started by the other, and so if either of them foresees dying in the near future, he cannot start any mental phrases of significance. He’s like a composer who can start composing a symphony only by writing “Unfinished Symphony” at the top of page one. With this account of selfhood in hand, let me return to the concepts of harm and benefit. There are many counterfactual David Vellemans whose lives diverged from mine at various points in the past, where their worlds diverged from the actual world. Most of them aren’t philosophers; some of them are rock stars; others are bums. Although each of them is David Velleman, however, I cannot frame first-​personal thoughts of which they will become the subjects, nor can I become the subject of their first-​personal thoughts. I can think of them as me only in the way that I can imagine being anyone else —​or, if you like, only in the way that one Parfit twin can imagine being the other. More importantly, I can have no first-​personal investments in their lives, or discharge any of their first-​personal investments, and so their good or bad fortune is no more to me than that of a twin. Even if I want or hope or fear

Non-identical and impersonal  257

things on their behalf, those emotions will never be resolved by how life turns out in their worlds, hence would counterfactually have turned out in mine. So although I care about them more than I care about a stranger, given that we share a name and a family of origin, I cannot care about them first-​personally, any more than the Parfit twins could look forward to one another’s future. It therefore seems to me that although the good or bad fortune of those other David Vellemans can give me grounds for envy or gloating, as anyone’s good or bad fortune might do, it cannot give me reason for first-​personal regret or relief. I could not have had what they have, nor lacked what they lack; for they aren’t me in the sense that transmits my first-​personal concern. Any trans-​world comparisons that I would presuppose in applying the concepts of harm and benefit are therefore inapplicable, not because those people aren’t David Velleman, but because despite being David Velleman, grown from the same sperm and egg, they aren’t and couldn’t have been me, accessible as referents for my first-​personal thoughts in points-​of-​ view accessible to mine. So whatever happened to divide their possible worlds from mine, it did me no harm or benefit. Here is a reason for not regretting what might have been  —​a reason other than the ordinary, pragmatic reason that nothing can be done about it. It is not just practically useless to have regrets about what might have been; it’s metaphysically confused, because the world as it might have been does not include anyone for whom I can have first-​personal concern. I can perhaps envy the people for whom things might have gone differently, as I envy any other person. But I cannot think of them as my more fortunate self, because they aren’t selves of mine in the relevant sense, and so I cannot regard what they have as something that I myself might have had. At issue here is only a particular kind of regret, namely, regret about what might have been. Another kind of regret, often called agent regret, is about what my actual selves did in the past. There is no confusion in regret over past mistakes, since I am first-​personally related to the agents who made them. The confusion begins when I regret not having today what I might have had if I hadn’t made those mistakes back then. As it turns out, then, the first person is not transitive: a future self of my past self is not necessarily a self of mine.The fate of a merely possible self of mine is no more pertinent to me than anyone else’s, since I can only imagine and never remember or anticipate undergoing that fate. But suppose that someone stole my identity and emptied my bank account. I am now reduced to poverty, daydreaming about the comfortable life I would have had if only my passwords had been longer. Surely, the thief harmed me, and we are tempted to say that the harm consists precisely in my being poorer than I might have been. But according to my view, it seems, I have no grounds for resenting the harm, because I am not first-​personally related to the David Velleman who would have been better-​off. My view still leaves me grounds for resentment, however —​namely, that the thief harmed my past self by depriving him of a comfortable future. When I complain,

258  J. David Velleman

“I could have been better-​off,” I  don’t mean, “I have a better-​off possible self ”; I mean, “My past self could have had a better-​off future.” More significantly, my past self is the one who was wronged by the theft. He is the one entitled to resent it as a violation of his privacy and property rights; I today can be indignant about that violation on behalf of my past self. But as Parfit explained —​and here I embrace his view whole-​heartedly —​my psychological connections to my past self grow weaker and weaker with time. The moral upshot is that resentment should fade as time passes. If, on the contrary, my resentment should be proportional to the difference in well-​being between my actual self and that other David Velleman, then it should increase as time passes, since that other David Velleman would have had capital to invest that I didn’t have. The difference between my actual well-​being and his will have grown, and so would my resentment if it were based on that difference. But rational resentment and rational regret are rather vicarious emotions felt on behalf of my past self, who recedes further and further into the past, hence further and further from my first-​ personal concern. In short, rationality favors letting bygones be bygones, in proportion to how far bygone they are. Let me conclude with a moral for moral psychology. Ever since the publication of Strawson’s “Freedom and Resentment”, the phrase “reactive attitude” has enjoyed unquestioned prestige in our field as a term for something that all of us can be expected to celebrate. The cause for celebration, supposedly, is that being a possible target for reactive attitudes, especially the negative attitudes, is a criterion, perhaps the criterion, of being a responsible agent rather than an amoral brute.We are supposed to be proud of our eligibility for resentment. Myself, I don’t go in much for resenting people. On the few occasions when people have done me a wrong, I have felt hurt, but beyond that my reaction has been simply to avoid them thereafter. If forced to deal with such people, or even think about them, I find them distasteful, but I don’t make the emotional investment that would be required to resent them. Resenting people is a way of maintaining an emotional relationship with them, at least in one’s own mind, and I don’t care to have an emotional relationship with people I cannot trust. Life is too short. Well, you might say, maybe you don’t resent people who wrong you, but surely you blame them. And I do blame them to the extent of interpreting their behavior as indicative of some flaw that makes them undesirable company. But not actively feeling blame, the emotion, is for me part and parcel of avoiding them, keeping them out of my life insofar as I can. Now, I confess that for most of my life I have been exquisitely susceptible to moral anger at figures of authority, especially when I feel they are being unfair. And I regret to say that, although any unjust exercise of authority makes me angry, my anger has on occasion mounted to the level of fury when I have felt myself to be the victim. Certainly, my anger is a Strawsonian reactive attitude, in that I held its targets responsible for wrongdoing. But I’m not especially proud of having felt that anger, nor to my knowledge did its targets take it as a compliment to their humanity. Fortunately, age

Non-identical and impersonal  259

has moderated the emotion, perhaps because I have begun to feel, with age, that the world will have to fight The Man without much further help from me. Moreover, there has always been an element of the impersonal in my anger, making it more like indignation than the sort of hostility that leaves the residue of a grudge. Thinking myself the victim of injustice heightened my sensitivity to it but didn’t lead me to regard it as being “about me”. Naturally, my introspection on this point is tainted by years of philosophy, and yet I cannot help thinking that what inflamed me is not disrespect for me so much as disrespect for something in me. And, more to the point, I am fairly certain that the true target of my indignation was injustice itself, not its agent —​injustice against which I felt powerless, but not the individual in whom I perceived, or thought I perceived, the power. I don’t offer these reflections in a spirit of complacency, much less pride. My tendency toward indignation rather than personal hostility has often alienated others to whom it seemed distastefully moralistic. I  know that colleagues have occasionally felt oppressed by my appeals to principle in department meetings. But in my more sanguine moments, I am inclined to think that their feelings arose from a misunderstanding, namely, that they were being accused of wrongdoing whenever I suggested that some decision would be less than ingenuous or would amount to using someone merely as a means. In fact, I never thought of myself as pointing the finger at anyone but only as pointing it at unfairness itself, and hence at something at which, I assumed, all of us would be indignant.Then again, maybe it was the evidence of this very attitude that made me seem objectionably high-​minded. What I am trying to describe here, in any case, is a moral sensibility that does not privilege reactive attitudes, that positively tries to avoid reactive attitudes in favor of impersonal moral emotions. I offer this description at the end of my arguments on non-​identity because those arguments have attempted to problematize our ordinary concepts of harm and benefit, which are central to the personal reactive attitudes such as resentment. Insofar as we moderate our sense of being harmed by people, we can moderate our resentment of them, and I believe we can do both without giving up our moral sensibility.

Notes 1 This chapter started life as a paper presented to a conference at Juan Carlos III University, Madrid; the Princeton Center for Human Values; and the Department of Philosophy at the University of Nebraska, Lincoln. Thanks to audiences at those presentations for comments and suggestions. 2 Parfit first discussed the problem in “On Doing the Best for Our Children,” in Ethics and Population, ed. Michael D. Bayles (Cambridge, MA: Schenkman, 1976), pp. 100–​115. See also Robert Merrihew Adams, “Existence, Self-​Interest, and the Problem of Evil,” Noûs 13 (1979):  53–​65; Gregory S.  Kavka, “The Paradox of Future Individuals,” Philosophy & Public Affairs 11 (1982): 93–​112; and Thomas Schwartz, “Obligations to Posterity,” in Obligations to Future Generations, ed. R. I. Sikora and Brian Barry (Philadelphia: Temple University Press, 1978), pp.  3–​13; Caspar Hare, “Voices from Another World:  Must We Respect the Interests of People Who Do Not, and Will Never, Exist?”, 117 Ethics

260  J. David Velleman

498–​523 (2007). I will deal primarily with Parfit’s discussion of the problem in Part Four of Reasons and Persons (Oxford: Oxford University Press, 1984). Rivka Weinberg interprets the problem differently in her book The Risk of a Lifetime: How, When, and Why Procreation May Be Permissible (Oxford: Oxford University Press, 2016). According to Weinberg, future people will have no grounds for complaint about climate change because they will have been compensated by the benefit of existing. She notes that Parfit and others tend to frame the answer to any complaint like this: “If we had addressed climate change, you wouldn’t have existed —​and your life is worth living.” In my view, the last clause is meant, not to tout the value of existence, but to forestall the further complaint “Yes, but I wish I hadn’t been born.” 3 Elizabeth Harman proposes a non-​comparative conception as a solution to the problem (“Harming as Causing Harm,” in Harming Future Persons, ed. Melinda A. Roberts and David T. Wasserman [New York: Springer Verlag, 2009], pp. 137–​154]. A similar conception of harm figures in Seanna Shiffrin’s “Wrongful Life, Procreative Responsibility, and the Significance of Harm” (5 Legal Theory 117–​148 [1999]). 4 Stephen Darwall, The Second-​ Person Standpoint; Morality, Respect, and Accountability (Cambridge, MA: Harvard University Press, 2006). 5 P.F. Strawson, “Freedom and Resentment,” 48 Proceedings of the British Academy (Oxford:  Oxford University Press, 1962), 1–​25; reprinted in Gary Watson, ed., Free Will, Second Edition (New York: Oxford University Press, 2003), pp. 72–​93. For some recent work on reactive attitudes, see Shaun Nichols, Bound:  Essays on Free Will and Responsibility (Oxford: Oxford University Press, 2015), Chapter 7; Per-​Erik Milam, “In Defense of Non-​Reactive Attitudes,” 20 Philosophical Explorations: 294–​307 (2017). 6 For a recent review of the problems with the concept of harm, see Ben Bradley, “Doing Away With Harm,” 85 Philosophy and Phenomenological Research (2012): 390–​412. For a reply to some problems, see Nathan Hanna, “Harm: Omission, Preemption, Freedom,” 93 Philosophy and Phenomenological Research (2016): 251–​273. 7 Reasons and Persons (Oxford: Oxford University Press, 1984), Part Three. My critique of Parfit’s view is developed more fully in “Persons in Prospect I: The Identity Problem,” in Beyond Price: Essays on Birth and Death (Cambridge, UK: Open Book Publishers, 2015), www.openbookpublishers.com/​reader.php/​349?349#page/​86/​mode/​1up; and in On Being Me (Princeton University Press, forthcoming). 8 In what follows, I elaborate on arguments that I originally made in “Self to Self,” 105 The Philosophical Review 39–​76; reprinted in Self to Self: Selected Essays (New York: Cambridge University Press, 2006); I expand on those arguments in On Being Me. 9 See Stefano Predelli, “I Am Not Here Now,” 58 Analysis 107–​115 (1998), and “I Am Still Not Here Now,” 74 Erkenntnis 289–​303 (2011). 10 Here I am echoing Parfit, Reasons and Persons, 280.

INDEX

Note: In this index, endnotes are indicated by “n” and the endnote number after the page number e.g., 182n17 is endnote number 17 on page 182. Graphs are indicated by bold type. Abhidharma, Buddhist 164, 167, 176, 177 accomplishment 41, 42–43 Act-Dependence, The Principle of 190, 192, 196 actual preferences, and counter-factual well-being 99–102 Actual World Subjectivism 100 addictive drug thought experiment 103–105 adjacent worlds thought experiment 119–120 advanced teletransportation (ATT) machine 163–164 agency: collective 236, 242; group 228, 233, 234, 235, 236–237, 242–243, 245, 247n18; individual see individual agents; moral 163, 174, 250; multiple 228, 233–235, 236–237, 241–242, 243; personal 7, 228–248 agent causation 174–175 agent-neutrality 23, 27, 34, 35, 37, 62, 174 agent-relativity 2, 3, 23, 30, 31, 34, 37 aggregates 170–171, 173, 174, 175, 176, 179, 182n17 aim-achievement 32, 110n5 alienation 210, 214–215, 216, 225n18 amnesia barriers 234 amputation 164, 172, 250 animalism 227, 232, 236, 241

Annie thought experiment 130 anti-essentialism 171 Appeal to Full Relativity 30, 31, 34–37, 45n6, 45n7 apperception 175–176 appraisal responses 122, 127n8 artificial persons 236 asymmetrical cases, of conflict between aims and ideals 215 A-theory of time 40 ATT (advanced teletransportation) machine 163–164 authenticity 211, 214–215, 216, 217, 225n18 Average Utility, Principle of/averagism 63, 74–75, 79, 90n17, 90n18, 251 Averagist Theory of Prudence 202–203 axiology 211, 216, 217 Bad Lives Case 87, 87 beliefs: false 14–15, 32–33, 41; moral 22, 62 beneficence, agent-neutral principle of 62 benevolence 63 best-case scenarios 41–42, 46 Better Chance Claim 80–85 betterness 117–118, 120, 121–122, 123, 125–126, 127n7, 127n13 betterness cycles 6, 127n8, 127n14

262 Index

bias: future 38–39, 40–44, 45–46n10; near 37–38, 39, 40, 43, 107; next year 33; one’s own favour 33–34, 44; past 40; present 40; present desires 33–34; status quo 198, 199, 206n18; time 3, 37–44, 45–46n10, 220, 221 block universe 40 brain/mind identity theory 162 Brand-New Life thought experiment 100–102 Brandt, Richard 217–218, 219 Brink, David O. 35–36, 36–37, 63, 204n8, 211–212 Brink’s defense, of temporal neutrality 214–217, 225n18 Broome, John 76, 77, 89n10, 90n26, 90–91n29 Buddhist Abhidharma 164, 167, 176, 177 Buddhist metaphysics 167, 171, 182n17 Buddhist ontology 168, 170, 171, 172, 173, 175, 180 Buddhist Personalism 170, 173–174, 175, 180, 182n11, 182n17 Buddhist Reductionism 6, 160–161, 166–174, 175, 176, 177, 180 Buddhist scholasticism 167, 174, 175 business school and philosophy PhD thought experiment 31–32 Canberra Plan 136 candidate properties of conation (CPCs) 136, 139, 140–142, 145, 146–150, 151–152, 154 Cartesian Ego/Cartesian Ego View 4, 48, 52, 53–54, 161, 164, 180, 181n3 categorical appraisal responses 122, 127n8 categorical imperative, Kant’s 228, 240, 243, 244 causal dispositions 229–230, 231, 237, 246n6, 246n8 causal overdetermination 172, 173 centred worlds 144–145 centred worm view/centred worms 144–146 chains, of psychological connections 4, 50, 51, 53, 141, 165, 166 Chang, Ruth 222–223, 225–226n30 Charlize Theron’s replica thought experiment 55–56 circularity 51, 53 climate change 69, 70, 82, 249, 250, 251–252, 259–260n2 “climbing the hill on the other side” metaphor 27–28

co-consciousness 58, 235 cognitive awareness 177, 178 collective agency 236, 242 collective level 17, 20, 23 collective self-defeat 2, 17, 19, 20, 23, 24, 25, 26, 28n11 combination of actions 44 combined spectrum 55–56, 181n4 commissurotomy thought experiment 56–58 common-sense morality 3, 13, 14, 22–27, 34–35, 229, 236; of liberal culture 241, 242 communication 26, 234 Comparative Notion of Harm 192–193, 194–195, 196, 198, 199–200, 205n10 compatibilism 174, 182n16 compensation 35–37, 45n8, 63, 64 conations/conativism, about personal identity 129–157 conceivability principle 171, 175 conceptual analysis 130, 136–137, 150, 155n2, 155n3, 157n30 conflict between aims and ideals 215 connectedness, psychological see psychological connectedness conscience 176 conscious experience 167, 174–179 consciousness 56, 57–58, 232, 234, 242, 256; and reasons and conscious persons 160–181, 183n27, 183n29; self- 6, 174, 175, 176, 178, 179, 181 consequentialism 2, 3, 7, 62, 227, 243, 244, 245; and rationality and time 34–35, 37; and unified theory of morality 13, 14, 17, 18–20, 21–22, 23, 26–28 constancy 191, 192 Constitutive Reductionism 3–4, 65n11 constructivism 227, 237–238, 239–240, 245–246 continuity, psychological see psychological continuity Continuum Argument for Intransitivity 120–121 Continuum Argument for the Repugnant Conclusion 107–108, 119, 120 contractualism 28 conventionalism 129, 131, 135, 138, 156n21, 156–157n23, 157n26 conventions: inter-personal 152; public 138–139; social 131, 138, 139, 152 counter-factual well-being, and actual preferences 99–102 CP (critical present-aim theory) 31–33, 209

Index  263

CPCs (candidate properties of conation) 136, 139, 140–142, 145, 146–150, 151–152, 154 critical present-aim theory (CP) 31–33, 209 decision theory 42–43, 196 decision-making 15, 20, 41, 57, 197 desire frustration 103, 105, 106 desire-first accounts 131, 157n26 Desire-Fulfilment Theory 30, 36, 40–41, 42–43, 94, 95–96, 101, 108; Global 102–105; Summative 102–105, 111n19; Unrestricted 97–99, 102 desires: fundamental 149; global 94, 102, 103, 104, 105, 106–107, 111n20; immoral 109; instrumental 149; intrinsically irrational 31–33; moral 33, 34 dessert treats thought experiment 124 diachronic conativism 129–155 diachronic inconsistency 43, 45–46n10 Dignāga 177, 178, 180, 183n25 direct awareness, through consciousness 242 direct collective self-defeat 20, 24, 25, 26 direct private conativism 140–150 direct self-defeat 23–25 disability thought experiments 194–202, 205n13, 206n19 dispositions 6, 139–140, 160, 174, 175, 176; causal 229–230, 231, 237, 246n6, 246n8; and unified theory of morality 15, 16–17, 18–19, 27 dissociative disorders 234 Distinctive Feeling Theory 96, 97, 110n7 distribution/distributive justice 4, 63–64, 66n33, 119, 204n8 divided minds 56–61 do-gooders, pure 18–19, 26–27 doubly-embodied divided mind 58–59 drug-addiction thought experiment 103–105 dual-aspect theory of mind 177 dualism 168, 169, 173 dumplings thought experiment 124 duties to self 252 dynamic model, of personal identity 176 Early Course, of surgery 41–42, 45–46n10 egological accounts, of self-consciousness 177 eligibility, and meaning 136, 139 eliminative materialism 163, 172 Eliminative Reductionism 162 eliminativism 163, 172 embodied experience 161

emergentism 170, 173 empty concepts, objection from 150–151 endurantism 134, 191, 254 episodic memory 52–53, 256 epistemic transformation 190, 196 epistemology 39, 49, 175, 177, 180–181 equality 4, 74–75, 79, 124 Equally Weighty Reason 218–219 error theory 198–199, 206n17 ethics, population 68–93 evidential decision theory 42 exdurantism 130, 131 Expansive Very Narrow Person Based Intuition (EVNPBI) 78–80, 82, 83–88 expected harm 200 expected utility 42 expected well-being 193, 200, 205n11 experiential anticipation 256 experiential memory 52–53, 256 Extreme Claim, for rationality 61, 62 false beliefs 14–15, 32–33, 41 fantasy epic thought experiment 212, 224n9, 216, 217, 220–221 faultless disagreement desideratum 6, 134, 135, 139, 142, 143, 149, 156n21 Featureless Cartesian View 51–52 Felicity and Freddie thought experiment 148–149, 150, 151 first-order metaphysics 129 foundationalism 239 Freddie and Felicity thought experiment 148–149, 150, 151 Full Relativity 30, 31, 34–37, 45n6, 45n7 fully neutral theory 35 fully relative theory 34 functionalism 173, 229, 246n6 fundamental desires 149 Further Fact View 164, 179 Future Tuesday Indifference thought experiment 32–33, 45n5 future-bias 38–39, 40–44, 45–46n10 George thought experiment 212, 224n9, 216, 217, 220–221 Global Desire-Fulfilment Theory/global desires 94, 102, 103, 104, 105, 106–107, 111n20 goods 7, 15, 24, 36, 148–149, 199, 200, 212, 213, 216, 220, 221, 223, 243 grading systems 122–123, 124, 125, 127n9

264 Index

group agency 228, 233, 234, 235, 236–237, 242–243, 245, 247n18 growing block view 40 happiness/Happiness Theory 35, 41, 63, 71, 89n11, 199–200, 251; and unified theory of morality 15–16, 19; and well-being 108, 110n5 Harmless Torturers thought experiment 21 harms 24–25, 193, 198, 200, 204n8, 211, 243 Hedonic Tone Theory 110n7 hedonism 42, 63; and well-being 95, 96–97, 98, 99, 102, 110n6, 111n22 hedonistic theories 30, 94, 99, 110n5, 110n6 high finance thought experiment 212–213, 216, 217, 220–221 Hume, David 32, 33, 44n2, 149, 151, 179, 244, 246n6, 246n9 hybrid theories 34, 35, 94, 96, 108–110, 111n26 identical twin thought experiment 58–59 identity: numerical 49, 130, 132, 133–134, 252–253, 254; over time see identity over time; personal see personal identity; qualitative 49 identity over time 6, 35, 44, 48, 50, 51, 53, 62; and personal identity 160, 162, 164, 165, 180 identity relation 130–131, 132, 133–134, 152, 171 identity theory/identity view 162 identity-conferring mental states 192 immersive subjectivity 177 immoral desires 109 imperceptible effects 21 impersonal description thesis 162–163; see also simple teletransportation (STT) machine thought experiment Impersonal Total Principle 118–119 implausible consequences, objection from 147–150 imprudence 62 incompatibilism 182n16 incompletely relative theory 34, 35, 94, 96, 108–110, 111n26 Independence of Irrelevant Alternatives 79–80, 91n34, 91n35 indirect conative theory 137–138 indirect self-defeat 16, 17, 18–19, 26 individual agents 230, 232, 233, 234, 238, 241, 242, 243–244; and the Kantian moral requirement 237, 240, 245

individual level 17, 20, 23, 25 individual rationality 2, 25–26, 234–235, 239 individual self-defeat 16, 17 individualism 231, 240, 243, 244 inequality 75 instrumental desires 149 intentionality 177, 180, 229 inter-personal conventions 152 intransitivity 6, 117, 120–124, 125, 127n13 intrapersonal tradeoffs 36 intrinsic causal powers/intrinsic properties 172, 173 irrational desires 31–33, 105 irreducible normativity 229–230, 237–238, 246n9, 246–247n11 I-thought 135, 137–138, 139–140 Jerome thought experiment 212–213, 216, 217, 220–221 Kantian morality 227, 228–229, 240, 241, 242, 243–244, 245–246 Kantian universalizability requirement 240 Kant’s categorical imperative 228, 240, 243, 244 King Lear thought experiment 99–100, 101 knowledge, social 242 Korsgaard, Christine 7, 221–222, 227–229, 231, 237–243, 244, 245–246 Ladau, Emily 194, 195 Late Course 41, 45–46n10 league-based satisficing 6, 117–118, 121–126 liberal culture, common-sense morality of 241, 242 Life-Prolonging Drug thought experiment 202–203 Locke, John 50, 179, 183n29, 227, 228, 231–232, 236, 240–243 McTaggart, J.M.E. 107 Malfunctioning Teletransporter Case thought experiment 49 many-person dilemmas 17, 24–25 MAP (Mere Addition Principle) 69, 73–77, 80, 81, 83, 87 marginal utility 42, 46 Martha and Phyllida thought experiment 131, 134–135, 142–143 maximality 191, 192 meaning 136

Index  265

memory 165, 253, 254, 255, 256; and personal identity 51, 52–53, 54, 56, 57, 65n12; semantic 182n7 mental states 163, 170, 173, 178, 188, 218, 254; conscious 178, 179; and habitual tendencies 175; identity-conferring 192; occurrent 177, 178, 183n29; and personal identity 49, 53, 58, 64n5; self-identityconferring 191; self-referential 175, 177 mental streams 180 Mere Addition 68, 69–70, 71–77, 74, 79, 80, 81, 83, 87 mereological nihilism 168, 172, 173 mereological realism 173 metalinguistic analysis solution 169 metaphysics 150; Buddhist 167, 171, 182n17; of content 64n5; first-order 129; Parfit’s 134; of personal identity 44, 64n2, 130, 144; of time 40; of value 238, 239, 245–246 micropsychism 183n27 Milinda, Questions of/Milindapañha 167, 174 Mill, John Stuart 27 mind: divided 56–61; philosophy of 161, 181n1; theory of 177 mind/brain identity theory 162 Moderate Claim, for rationality 61 momentary time-slices 36–37 moral agency 163, 174, 250 moral beliefs 22, 62 moral desires 33, 34 moral mathematics, mistakes in 20–22 moral obligations 3 moral psychology 174, 258 moral responsibility 59, 62, 174–175 moral theory 3, 4, 5, 13–28, 70, 252 morality: common-sense 3, 13, 14, 22–27, 34–35, 229, 236; consequentialist 227, 243, 244, 245; Kantian 227, 228–229, 240, 241, 242, 243–244, 245–246; of liberal culture 241, 242; and personal identity 62–64; Unified Theory of 13–28 multiple agency 228, 233–235, 236–237, 241–242, 243 My Past or Future Operations thought experiment 38 naïve realism 238, 245, 247n21 Narrow Hedonism 96, 97 Narrow Psychological Criterion 51 neutral theory 23, 35, 212 neutrality intuition 76, 90–91n29

New Person thought experiment 188, 190, 194 non-comparative notion of harm 205n10, 250, 260n3 non-conativism 129, 131 non-egological accounts, of selfconsciousness 177 non-identity 6–7, 7–8, 187–207, 249–260 non-reductionism 3, 48, 51–53, 54, 62, 132; and reasons and conscious persons 161, 164, 168, 173, 180, 181n3 non-risk-weighted causal decision theory 43 normative self-concern 7, 209–210, 215, 221, 223, 224 normativity 222, 230, 239, 245; irreducible 229–230, 237–238, 246n9, 246–247n11; practical 221–222; prudential 216 no-self view, of Buddhist Abhidharma 164, 166–174, 176–177, 180 numerical identity 49, 130, 132, 133–134, 252–253, 254 objection from empty concepts 150–151 objection from implausible consequences 147–150 Objective List Theory/Objective Lists 41, 42–43, 96, 108, 109, 110n5 object-oriented aspect, of theory of mind 177 obligations 222, 244, 245, 249, 250, 251, 252; agent-relative 23; and CommonSense Morality 22; to future self 62; moral 3; and Parent’s Dilemmas 24, 25; and self-concern 210, 219–220 occurrent mental states 177, 178, 183n29 one-level theories of consciousness 178 ontology 64, 145, 181n1; Buddhist 168, 170, 171, 172, 173, 175, 180 orthodox decision theory 43 other-reading practices 153 other-regarding conations 135, 149, 152, 153, 154–155, 221, 224, 231 overdetermination, causal 172, 173 overlapping chains 50, 53, 141, 165, 166 Overvold, Mark Carl 99 ownership view 162, 232 pain and pleasure, theory of 96–97 pain and pleasure thought experiment 42–43 pairwise comparison 122, 123, 124–125 panpsychism 183n27 paracognitive conativism 139

266 Index

paracognitive theory 137–138 paradox of happiness 15–16 Parent’s Dilemmas 24–25, 26 Pareto minus 80, 83, 87 Pareto plus 68–93 partite entities 167–168 PBI (person based intuition) 72–73, 78–80, 82, 83–88 perceptible effects, of decisions 21–22 perdurantism 130, 191, 204n4 persistence conditions of persons 130, 134; see also diachronic conativism person based intuition (PBI) 72–73, 78–81, 82, 83–88 person pluralism 135, 152, 153 Person-Affecting Intuition/PersonAffecting Principle 6, 192, 193 personal agency 7, 228–248 personal good 89n10 personal identity 48–66, 160–181; conativism about 129–157; determinate 53; dynamic model of 176; moral significance of 7, 62, 228–248; over time 6, 35, 44, 51, 53, 62, 162, 164, 165, 180; problem of 6, 161, 174, 180; psychological criterion for 6, 51, 52–53, 164, 165–166, 167, 175, 180; and rationality 61–62; reductionism and 161–166; theory of 52, 161, 163, 166, 175 personal transformation 6, 187–190, 197 Personalism, Buddhist 170, 173–174, 175, 180, 182n11, 182n17 person-based theory 81, 83 person-directed practices 139, 156n21 personhood 7, 34, 35, 62, 237, 240, 242, 250, 251–252 person-phases 129–157 phenomenological point of view 232, 234, 235, 242 philosophy of mind 161, 181n1 philosophy PhD and business school thought experiment 31–32 Phyllida and Martha thought experiment 131, 134–135, 142–143 physical continuity 50, 60; and conativism about personal identity 131, 136, 139, 142, 143, 144, 147, 150; see also sameperson relation Physical Criterion 51, 53–54, 55, 58, 164–165, 169–170 Physical Spectrum 55 physical states 49, 166, 173 Physical View 50, 55, 58, 164–165, 169–170

pills thought experiment 202 pleasure and pain, theory of 96–97 pleasure and pain thought experiment 42–43 Pleasure Pill Case thought experiment 82–83, 82, 85–87, 86 pluralism 135, 151–155 population ethics 68–93 practical concerns 70, 133, 149, 150, 152, 153–154, 160 practical dilemmas 17, 19, 20, 21, 24 practical identities 222, 240 practical normativity 221–222 practical rationality 7, 150, 190, 203, 204n4, 209, 222, 224 practice dependent accounts 131, 157n26 practice-based conativism 138, 139 Preference-Hedonism 111n20, 111n22 Present Aim Theory 31–33, 34, 209 presentism 40 Presentist Theories of Prudence 200–201, 206n20 primitive conations 141 Principle of Act-Dependence 190, 192, 196 Principles of Utility 63, 250–251 prisoner’s dilemmas 17, 23–25, 44 private conations 138, 139, 140, 147, 152 procreation thought experiment 198, 201–202, 251 procreative privacy 69, 70, 88–89n8 prudence, and self-concern 209–226 prudential benefits 212, 213, 215, 216, 218, 221, 225n19 prudential concerns 6, 139, 154, 160, 191–192, 210–211, 224; and personal identity 48, 54, 59, 60, 61–62 prudential ideals 216 prudential normativity 216 prudential rationality 216, 220, 225n18, 225n19 Prudential Status 191–192, 204n8 prudential theory, of self-concern 209–210, 210–211, 214, 218, 219 psychological connectedness 4, 141, 165, 204n5, 258; and personal identity 48, 50–51, 52–53, 59, 60–63, 64n3 psychological continuity 4, 162, 164, 165, 166, 180, 205n16; and conativism about personal identity 136, 140–141, 142, 143, 144; and personal identity 50, 54, 55, 60–61, 62; see also same-person relation psychological criterion, for personal identity 6, 51, 52–53, 164, 165–166, 167, 175, 180 psychological profile 54, 55, 56, 60, 162

Index  267

psychological reductionism 175, 180, 227–228, 244 psychological spectrum 54, 169 Psychological View 4, 48, 50, 53, 55, 58; objection to 52–53 public conations 138, 147, 151–152, 154 public conventions 138–139 punishment 62–63, 127n10 pure do-gooders 18–19, 26–27 pure theory 35 qualitative aspect, of theory of mind 175, 177, 178 qualitative identity 49 quasi-memory 53, 65n14, 182n7 Ralston, Aron 250, 252 rational activity 108, 109 rational aims 14, 48, 63 rational fragmentation 236 rational preferences 201 rational self-transformation 190 rationality: individual 2, 25–26, 234–235, 239; and personal identity 61–62; practical 7, 150, 190, 203, 204n4, 209, 222, 224; and time 30–46; theory of 7, 34, 48, 102, 209 Rawls, John 33, 35, 63, 109 R-branching individuals 60 Reason 219 reductionism 4, 6, 7; Buddhist 6, 160–161, 166–174, 175, 176, 177, 180; Constitutive 3–4, 65n11; eliminativist 163; and personal identity 48, 51, 52, 53, 61, 62–63, 64, 161–166, 175; and philosophy of mind 181n1; psychological 175, 180, 227–228, 244; and self-constitution 231–237; about the SP-relation 132 reductive mereological nihilism 173 reductive physicalism 173 reflexive self-consciousness/reflexivism 178–179 reflexivity 174–179, 180 reification 168 relational appraisal responses 122, 127n8 “relational” theories 250 relative theory 23, 34 relativity: agent- 2, 3, 23, 30, 31, 34, 37; full 30, 31, 34–37, 45n6, 45n7; time- 32, 34 remote desires 97–99 Repugnant Conclusion 5, 6, 73, 77, 117, 202, 251; and the Continuum Argument for Intransitivity 120–121; Parfit’s evasion of the 118–120, 124–126; single-life 105–108

Requirement of Equal Concern 61 responsibility, moral 59, 62, 174–175 risk aversion 41–42, 43, 46 Risk-Weighted Expected Utility Theory 42, 43, 46 roller-coaster thought experiment 217–220, 221 Russell, John 27 Same World Subjectivism 100 Same-People Pareto 76, 77, 78, 90n28 same-person relation 129–154, 155n8, 157n26, 157n27 second-person standpoint, theory of the 250, 255 self, duties to 252 self-acquaintance 177–178 Self-Affecting Principle, The 193, 194, 196, 198, 200–203 self-concern: normative 7, 209–210, 215, 221, 223, 224; and prudence 209–226; and welfarism 221–224 self-condemnation 14 self-consciousness 6, 174, 175, 176, 177, 178, 179, 180, 181, 183n29 self-constitution 7, 227, 228–229, 231–237, 239, 240–243, 245 self-creation 190–192, 204–205n9 self-criticism 230, 247n14 self-defeat 22, 26, 125; collective 2, 17, 19, 20, 23, 24, 25, 26, 28n11; direct 20, 23–25, 26; indirect 16, 17, 18–19, 26; individual 16, 17 self-defeating theories 2–3, 13–16 self-denial 15, 16, 19 selfhood 252–253, 256 self-identity-conferring mental states 191 self-interest theory 2, 7, 48, 60, 101–102, 209, 213; and rationality and time 30, 33, 38, 39, 40, 41, 44; and unified theory of morality 14–17, 18, 19, 20, 23, 24, 25–26 self-referential mental states 175, 177 self-reflexivity 180 self-regarding practices 135, 152, 153, 154–155 self-transformation, rational 190 semantic insulation 168–169 semantic memory 182n7 Sensible Knave, Hume’s 244 separateness of persons 35–36, 37, 63, 64, 204n8 shrinking block view 40 Siderits, Mark 167–169, 170, 172, 173, 174, 177 Sidgwick, Henry 1, 45n6, 63, 96

268 Index

simple teletransportation (STT) machine thought experiment 162–163, 166 single-life repugnant conclusion 105–108 small effects, of decisions 20–21 social conventions 131, 138, 139, 152 social entities 35, 37 social knowledge 242 social theory 138 sorites paradox 169 SP-relation 129–154, 155n8, 157n26, 157n27 stage theory 130, 131 status quo bias 198, 199, 206n18 S-theory see self-interest theory streams of consciousness 56, 57, 58, 178, 180 STT (simple teletransportation) machine thought experiment 162–163, 166 subject naturalistic account 137 subjective aspect, of theory of mind 175, 177, 178 subjective persistence of selves 191 subjectivist theories 100, 108, 109 subjectivity 177, 179, 180 sub-personal perspective 36 Success Theory 95–96, 98–99, 102 suicide 239, 248n22 Summative Desire-Fulfilment Theory 102–105, 111n19 Summative Theories 103, 105–106 supremely rational ultimate aim 14, 18, 33, 61, 101–102 surgery thought experiments 41–42, 193–194 survival 59–60, 205–206n16, 252–253; and conativism about personal identity 133, 134–135, 142, 149, 150, 153 symmetrical cases, of conflict between aims and ideals 215 taxonomy, of well-being theory 94, 95, 110n5 Teletransporter Case thought experiment 49–50, 65n17, 131, 142–143 Temkin, Larry 118, 120–121, 122, 124–125 temporal neutrality 30, 31, 32, 34, 37, 38–39, 209–223, 225n18 temporal-relativity 32, 34 theories/theory: of consciousness 178; critical present-aim 31–33, 209; decision 42–43, 196; Desire-Fulfilment see Desire-Fulfilment Theory; distinctive feeling 96, 97, 110n7; error 198–199, 206n17; evidential decision 42; fully neutral 35; fully relative 34; Global

Desire-Fulfilment 94, 102, 103, 104, 105, 106–107, 111n20; happiness see happiness/Happiness Theory; Hedonic Tone 110n7; hedonistic 30, 94, 99, 110n5, 110n6; hybrid 34, 35, 94, 96, 108–110, 111n26; identity 162; incompletely relative 34, 35, 94, 96, 108–110, 111n26; indirect conative 137–138; of mind 177; mind/brain identity 162; of morality 3, 4, 5, 13–28, 70, 252; neutral 23, 35, 212; non-riskweighted causal decision 43; Objective List 41, 42–43, 96, 108, 109, 110n5; orthodox decision 43; of pain and pleasure 96–97; of personal identity 52, 161, 163, 166, 175; of pleasure and pain 96–97; Present Aim 31–33, 34, 209; of prudence 200–203, 206n20; pure 35; of rationality 7, 34, 48, 102, 209; “relational” 250; relative 23, 34; Risk-Weighted Expected Utility 42, 43, 46; second-person standpoint 250, 255; of self-concern 209–210, 210–211, 214, 218, 219; self-defeating 2–3, 13–16; self-interest see self-interest theory; social 138; stage 130, 131; subjectivist 100, 108, 109; success 95–96, 98–99, 102; summative 102–106, 111n19; of time 40; Unrestricted Desire-Fulfilment 97–99, 102; value 6, 117; of well-being 94, 95, 110n5 third-personal practices 154 Three Outcome Case thought experiment 75–80, 76 threshold, and the SP-relation 141 time, and rationality 30–46 time-bias 3, 37–44, 45–46n10, 220, 221 time-neutrality 30, 31, 32, 34, 37, 38–39, 209–223, 225n18 time-relativity 32, 34 time-slices 31, 36–37, 38–39, 40 totalism 88n1, 90n17 Totalist Theory of Prudence 201–202 transformative choice 6–7, 187–207 trans-temporal numerical identity relations 133, 134 Two Outcome Case thought experiment 71–72, 71, 73, 75, 76, 79–80 two truths framework 167–168, 169 two-person dilemmas 17, 24 Ullmann-Margalit, Edna 188 ultimate moral aim 18, 63 uncentred worms 146 unified human life 235–236

Index  269

Unified Theory, of morality 13–28 unifying project 235–236, 237, 240, 242 unity of conscious experience 57–58, 174–179, 180 universalizability requirement, Kantian 240 Unrestricted Desire-Fulfilment Theory 97–99, 102 use 136 utilitarianism/utility 35, 42, 63, 64, 88n1, 202, 250–251 value 6, 117, 213, 215, 238, 239, 245–246 vampire thought experiment 189, 197 Vasubandhu 171 Very Narrow Person Based Intuition (VNPBI) 72–73, 75, 77–78 volitions 176 welfarism 7, 209–210, 211, 221–224 well-being 3, 4, 94–111; and Averagist Theory of Prudence 202–203; counterfactual 99–102; desire-satisfaction

theory of 217, 218; lifetime 44–45n4; maximization of 32; and prudence 191–192, 200, 214; and rationality and time 30–31, 32, 36, 38–39, 40–41, 42, 44n3; and Self-Affecting Principle 193; theories of 36, 40–41, 43, 94, 95, 96, 109–110, 110n5, 211 “What Makes Someone’s Life Go Best” 94–111 Wide Psychological Criterion 51 Wiggins, David 1, 4, 58 Williams, Bernard 1, 53–55, 65n16, 162, 169 wisdom teeth surgery thought experiments 193–194 Within-A-Mile Altruism 33 world centres 145 world-neutrality 35 worms 130, 144–146 worst-case scenario 41–42, 45–46n10 worth-having existence 68–69, 71, 76, 80, 85