The Moon Points Back 0190226870, 9780190226879

Citation preview

The Moon Points Back

THE MOON POINTS BACK

z Edited by

KOJI TANAKA, YASUO DEGUCHI, JAY L. GARFIELD, AND GRAHAM PRIEST

1

1 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford New York Auckland  Cape Town  Dar es Salaam  Hong Kong  Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trademark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016

© Oxford University Press 2015 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. Library of Congress Cataloging-in-Publication Data The moon points back / edited by Koji Tanaka, Yasuo Deguchi, Jay Garfield, and Graham Priest. pages cm Includes bibliographical references and index. ISBN 978–0–19–022687–9 (paperback) — ISBN 978–0–19–022686–2 (cloth) 1. Sunyata.  2. Buddhist philosophy.  3. Analysis (Philosophy)  4. Philosophy, Comparative. BQ4275.M66 2015 294.3'42—dc23 2014039429

1 3 5 7 9 8 6 4 2 Printed in the United States of America on acid-free paper

Contents

Contributors 

vii

Introduction—Koji Tanaka, Yasuo Deguchi, Jay L. Garfield, and Graham Priest

xi

1. Persons Keeping Their Karma Together: The Reasons for the Pudgalavāda in Early Buddhism—Amber D. Carpenter 2. On Minds, Dharmakīrti and Madhyamaka—Tom J. F. Tillemans

1 45

3. On Being Humean about the Emptiness of Causation—Ricki Bliss

67

4. Essence and Emptiness—Roy W. Perrett

97

5. The Net of Indra—Graham Priest

113

6. Buddhist Reductionism and Emptiness in Huayan Perspective—Nicholaos Jones

128

7. Constructing a Logic of Emptiness: Nishitani, Jízāng, and Paraconsistency—Yasuo Deguchi

150

8. Nāgārjuna’s Logic—Aaron J. Cotnoir

176

9. Conventional Truth and Intentionality in the Work of Dharmakīrti—Laura Guerrero

189

vi

Contents

10. Relativism in Buddhist Philosophy: Candrakīrti on Mutual Dependence and the Basis of Convention—Elena Walsh

220

11. Two Truths and Method—Jay L. Garfield

245

References 

263

Index  279

Contributors

Ricki Bliss is an Assistant Professor at Lehigh University, Bethlehem, Pennsylvania. She wrote her doctoral dissertation under Graham Priest at the University of Melbourne, Australia, after which she studied with Yasuo Deguchi for two years at Kyoto University, Japan, as a JSPS postdoctoral research fellow. She works mainly in analytic metaphysics. Amber D. Carpenter studied Sanskrit at SOAS to keep her sane while writing a PhD on Plato’s Philebus at King’s College, London. She has published on topics in ancient Greek philosophy and, more recently, on Indian Buddhist philosophy (Indian Buddhist Philosophy, Routledge 2014). In both Greek and Indian philosophy, her main interest is in ethics, and particularly in how matters of metaphysics, epistemology, and philosophical psychology intersect with the questions of who we should be and how we should live. She is Associate Professor at Yale-NUS College. Aaron J.  Cotnoir received his PhD in philosophy from the University of Connecticut in 2010. He is currently a lecturer in the Department of Philosophy at the University of St Andrews. He works primarily in metaphysics and philosophical logic, and has strong research interests in philosophy of religion (both Western and non-Western). Yasuo Deguchi is Professor of Philosophy, Kyoto. His research interests include philosophy of science and analytic Asian philosophy. As for the latter subject, he coauthored several papers with Jay Garfield and Graham Priest, such as “The Way of Dialetheist:  Contradictions in Buddhism” (Philosophy East and West 2008), and contributed a chapter titled “Nishitani on Emptiness and Nothingness” to Nothingness in Asian Philosophy (Routledge 2014). Jay L. Garfield is Kwan Im Thong Hood Cho Temple Professor of Humanities and Head of Studies in Philosophy at Yale-NUS College, Professor of

viii

Contributors

Philosophy at the National University of Singapore, Recurrent Visiting Professor of Philosophy at Yale University, Doris Silbert Professor in the Humanities and Professor of Philosophy at Smith College, Professor of Philosophy at Melbourne University, and Adjunct Professor of Philosophy at the Central University of Tibetan Studies. Laura Guerrero is Assistant Professor of Philosophy at Utah Valley University. Her work focuses on drawing productive connections between contemporary Western and classical Indian theories of mind, metaphysics, and knowledge. Nicholaos Jones is an Associate Professor in the Department of Philosophy, and coordinator for the Program in Science and Technology Studies, at the University of Alabama in Huntsville. His research interests include Chinese Buddhism, with a focus on the Huayan School, as well as philosophy of science, with focuses on idealizations and visualizations. His writings have appeared in Asian Philosophy, Journal of Chinese Philosophy, Philosophy East and West, Philosophy & Biology, Erkenntnis, and Studies in History and Philosophy of Modern Physics. Roy W. Perrett is presently a Research Associate of the School of Historical and Philosophical Studies, University of Melbourne, and was formerly Professor of Philosophy at the University of Hawai‘i at Mānoa. He has published widely on both Indian and Western philosophy, including his books Death and Immortality (Springer 1986), Hindu Ethics: A Philosophical Study (University of Hawai‘i Press 1998), and the forthcoming An Introduction to Indian Philosophy (Cambridge University Press). Graham Priest is Distinguished Professor of Philosophy at the Graduate Center, City University of New York, and Boyce Gibson Professor Emeritus at the University of Melbourne. He is known for his work on nonclassical logic, particularly in connection with dialetheism, on the history of philosophy, and on Buddhist philosophy. He has published articles in nearly every major philosophy and logic journal. His books include:  In Contradiction:  A  Study of the Transconsistent (Martinus Nijhoff 1987; 2nd ed., Oxford University Press 2006); Beyond the Limits of Thought (Cambridge University Press 1995; 2nd ed., Oxford University Press 2002); Towards Non-Being: The Semantics and Metaphysics of Intentionality (Oxford University Press 2005); Doubt Truth to Be a Liar (Oxford University Press 2006); Introduction to Non-Classical Logic Form If to Is (Cambridge University Press 2008); One (Oxford University Press 2014).

Contributors

ix

Koji Tanaka is Lecturer in the School of Philosophy, RSSS, at the Australian National University. His research focuses on logic, history and philosophy of logic, philosophy of language, Buddhist philosophy, classical Chinese philosophy and Japanese philosophy. Tom J.  F. Tillemans is Professor Emeritus of Buddhist Studies at the University of Lausanne, Switzerland. His published work has focused largely on comparative philosophy, Buddhist logic, epistemology, and philosophy of the middle (Madhyamaka). His most recent book is How Do Mādhyamikas Think? (Studies in Indian and Tibetan Buddhism, Somerville, MA:  Wisdom). Tillemans was editor of the Journal of the International Association of Buddhist Studies and is presently editor in chief of the 84000, a project to translate Buddhist canonical literature from Tibetan and Sanskrit (http://84000.co). Elena Walsh is a PhD student in the philosophy department at the University of Sydney. She has an MA in philosophy from the University of Melbourne. Her research interests include Buddhist philosophy, moral psychology, and the philosophy and psychology of emotion.

Introduction Koji Tanaka, Yasuo Deguchi, Jay L. Garfield, and Graham Priest

It is apparent to those who have paid attention to developments in both traditions, that the contemporary Anglophone Western philosophical tradition and the Buddhist philosophical tradition have a lot to say to one another. Their central concerns overlap sufficiently to determine a domain of shared interest; their histories and idioms are distinct enough that each has something to learn from the other about that domain. The contributors to this volume are committed to advancing the dialogue between these two traditions that is now well underway. This volume is a successor to Pointing at the Moon (D’Amato, Garfield, and Tillemans 2009), which collects essays directed specifically at the interface of contemporary logic and analytic philosophy and Buddhist philosophy. That volume addresses a wide range of topics at this interface, deploying a common analytic methodology. The present volume, many of whose contributors also contributed to the previous volume, focuses more specifically on the Buddhist concept of emptiness—primarily, though not exclusively as it is deployed in the Madhyamaka tradition—exploring its implications for contemporary philosophy and the ways that contemporary philosophy can illuminate classical texts and problems in the Buddhist tradition. Emptiness—roughly, the lack of essence, or substance, in things—is one of the central conceptions in Mahāyāna Buddhist metaphysics, and represents one of its distinctive contributions to world philosophy. Following Nāgārjuna (second to third centuries ce), Mādhyamikas argue that the

xii

Introduction

idea that things have intrinsic nature, or any kind of independent existence, or the idea that reality has a privileged foundation, is necessarily false. Instead, they argue, we can make sense of reality only as a complex of impermanent, interdependent, conventionally existent entities, states, and processes. While this may be an unfamiliar way to think about ontology in the West, it has proven, in the last few decades of Western-Buddhist interaction, to be fecund. The analysis of emptiness and its implications, however, raises difficult and subtle questions. Many of these, of course, were addressed extensively in the classical Buddhist philosophical traditions of India, Tibet, and East Asia. However, contemporary logic and analytic philosophy can bring to the study of these issues a range of tools and techniques that were unavailable in the Asian Buddhist traditions. Just as these techniques have advanced debates in Western philosophy, they can advance debates in Buddhist philosophy. Using these analytical techniques, we can provide more precise accounts of the meaning of emptiness, and sometimes deeper explorations of the implications of this doctrine, hence contributing to the development of Buddhist philosophy in the contemporary world. Moreover, just as Buddhist analyses have advanced philosophical debates in Asia, they can advance such debates in the West. The chapters collected in this volume address three principal and interrelated domains in which emptiness matters: the nature of the person, the nature of impersonal reality, and the nature of truth and cognate notions. One of the best-known and most distinctive doctrines in Buddhist philosophy is the doctrine of anātman, or no-self. It is central to a Buddhist soteriology and metaphysics that a personal self—an independent, enduring subject of experience and owner of body and mind—is chimerical. On this view, a person is nothing more than an interconnected sequence of physical and psychological states and processes, with no core or owner. How to understand the details of this view and its implications is a matter of considerable debate in the Buddhist world, and a matter for debate between Buddhist and non-Buddhist interlocutors in India. One early Buddhist school—the Pudgalavādins, or proponents of the reality of the person—argued that while at an ultimate level of analysis all that we find are sequences of evanescent impersonal phenomena, nonetheless, supervenient on those is a real entity—the pudgala, or person—which has clear identity conditions and properties of its own. This view was subject to trenchant critique as a relapse into orthodox ātamavāda and

Introduction xiii

eventually fell out of favor in Buddhist India. Amber Carpenter, in her contribution to this volume (­chapter 1), argues that this rejection may have been premature, and that using contemporary metaphysical approaches to personhood, we can see that pudgalavāda has resources which the tradition may have underestimated. Persons may be empty, she argues, but are not, for all that, eliminable from our ontology. While Carpenter uses Western analyses to clarify and rehabilitate an early Buddhist idea regarding a kind of reality for persons, Tom Tillemans uses a Buddhist analysis to intervene in contemporary Western discussions of eliminative materialism (­chapter  2). He argues that modern eliminativism—appearances to the contrary notwithstanding—is inconsistent with the view of emptiness. This is because eliminative materialism privileges a physical level of description as having a convention-independent reality which denies a personal level of description. Tillemans argues, instead, that a Madhyamaka analysis demonstrates that these levels of analysis are interdependent and equally necessary. The emptiness of the person does not entail the nonemptiness of the physical. Just as Western metaphysics, in Carpenter’s hands, reinvigorates a classical Buddhist position, so does Buddhist metaphysics, as Tillemans advances it, provide resources to resist an influential Western position. In many ways, later Buddhist philosophy takes the earlier Buddhist position on the nature of persons, and generalizes it to all things. The selflessness of the person becomes the selflessness of phenomena: all things are empty of intrinsic nature; there is no ground of being. Ricki Bliss examines the implications of the doctrine of emptiness for the Madhyamaka analysis of the crucial notion of causation (­chapter 3). A number of recent scholars in philosophy and Buddhist studies have argued for a close analogy between Hume’s treatment of causality and a Buddhist understanding, trading on the emphasis by both Hume and Nāgārjuna on the emptiness and conventional nature of the causal relation. Bliss argues that this analogy is not as close as some have suggested, and that taking Buddhist theories of causation to be Humean may run afoul of the emptiness of the causal relata themselves. Her analysis evidences the utility of taking contemporary metaphysical approaches into Buddhist discussions and of bringing a Buddhist sensibility to contemporary metaphysics. Emptiness is often glossed in Western discussions as essencelessness. This reading raises the question of the relation between the Indian concept denoted by svabhāva and Western metaphysical notions such as essence, necessity, and ground. Nāgārjuna and Candrakīrti (sixth century

xiv

Introduction

ce) argue that all phenomena are empty of any intrinsic nature and that that is their intrinsic nature. Roy Perrett argues (­chapter  4) that this apparent paradox can be resolved by disambiguating svabhāva. Drawing on contemporary understandings of necessity and grounding, he argues that, on one reading, svabhāva is the metaphysical ground of a thing; on another it is its necessary quality. Hence, he argues, the Mādhyamika can maintain nonparadoxically that all things necessarily lack a metaphysical ground. Emptiness in China has a subtly different flavor from emptiness in India. When Buddhism was transmitted to China it was adopted into a culture already redolent with a metaphysical framework inherited from the Daoist and Confucian traditions. Chinese Buddhism is inflected by these ideas. The Huayan tradition universalizes the idea of interdependence. According to philosophers in this tradition, it is not merely that everything depends upon some other things, but that every thing depends upon all other things, and that each phenomenon interpenetrates every other phenomenon. This is represented in Huayan literature by the metaphor of the net of Indra, an infinite network of perfectly reflective jewels, each of which reflects the entire network. Graham Priest provides an interpretation of this doctrine in terms of modern graph theory (­chapter  5), demonstrating the utility of contemporary logical-mathematical tools for making precise classical Buddhist doctrines, and for defusing suspicions of mystical incoherence. Nicholaos Jones deploys the Huayan conception of interdependence, referred to in that tradition as round fusion, against the mereologically reductionist interpretation of emptiness developed by some recent analytic commentators on Buddhist metaphysics such as Mark Siderits. His essay (­chapter 6) is a nice example of how competing analytic interpretations of Buddhist doctrine find precise articulation using the tools of analytic Western metaphysics. It is one thing to talk about reality—whether personal or impersonal—but quite another to talk about talking about reality. And philosophy demands the latter as well. Buddhist philosophers as well as Western philosophers hence take logic, the philosophy of language, and epistemology as central areas of concern. Indeed, these are areas in which fruitful dialogue is possible as is shown by the next contributions. Emptiness plays a key role in the thinking of a number of members of the Kyoto School. Nishitani, in particular, took it that a certain understanding of logic was a key element in his views on emptiness. The logic was

Introduction xv

never developed in a formal way, however. Drawing further on ideas of Jízāng, one of the most important thinkers in the Chinese San-lun school of Buddhism, Yasuo Deguchi shows how one may do this, applying techniques of contemporary paraconsistent logic (­chapter  7), thus showing how contemporary developments in Western logic can be applied with illumination to more traditional Buddhist thought. The logic deployed by Deguchi is a three-valued logic. It can be seen as a special case of a four-valued logic which, arguably, characterizes the traditional form of argumentation deployed in Buddhist thought, for example, by Nāgārjuna in Mūlamadhyamakakārikā. This is the catuṣkoṭi, according to which, a sentence may be true, false, both, or neither. The four-valued structure provides the basis of the most fundamental relevant logic, called First Degree Entailment. Aaron Cotnoir (­chapter  8) asks whether this logic can in fact provide an adequate analysis of Nāgārjuna’s arguments and positions, and argues that it does not. He suggests a different way of proceeding, involving the distinction between conventional and ultimate truth (reflected in a distinction between two forms of the catuṣkoṭi), showing how reflection on Madhyamaka metaphysics can inform the philosophy of logic. Conventional truth describes things as delivered by ordinary experience; ultimate truth captures the way that things are independent of our interests, practices, and cognitive faculties. It is notoriously difficult to provide an adequate analysis of either conventional or ultimate truth, however. Indeed, Buddhist philosophers are at odds with one another on these matters. Laura Guerrero addresses the problem of making sense of conventional truth—in particular, in the context of the great Buddhist epistemologist Dharmakīrti’s (seventh century ce) philosophy (­chapter 9). She argues that the only way to make sense of his view is to take a pragmatist reading of a deflationist account of truth. We see here the virtues of bringing the most recent work in Western semantics and the philosophy of logic to bear on the philological project of understanding classical Indian Buddhist literature, and the virtue of taking Indian Buddhist epistemology seriously in a Western philosophical context. One might worry that any account of conventional truth that really represents it as conventional lapses into relativism. That is, if what we know and take to be true is unavoidably enmeshed with our interests and our ordinary epistemic practices, there seems to be no real sense in which knowledge can count as a norm that governs rational inquiry. Candrakīrti addresses the nature of conventional truth and defends its normative

xvi

Introduction

status more directly than any other Indian Mādhyamika. Elena Walsh (­chapter 10), drawing on recent work on epistemic norms, considers the plausible charge that even his account is, in the end, a mere relativism. She argues that it is not. Candrakīrti’s account of epistemic warrant, she argues, saves his account from arbitrariness and provides a coherent understanding of conventional truth as true. This discussion shows how a consideration of epistemological problems such as relativism can benefit from drawing on both Western and Buddhist literature. Collectively, these chapters demonstrate three things. First, they show the value of taking seriously a philosophical idea—in this case, that of emptiness—deriving from one tradition, to enrich the philosophical reflections of another. Second, these essays demonstrate the importance of bringing different traditions and their texts together for mutual enlightenment. Third, they show that the resulting dialogue benefits both partners. Buddhist philosophical problems and insights come into sharper focus through the lens of contemporary Western analytic techniques; Western philosophy acquires from Buddhist philosophy new problems, new directions of inquiry, and new conceptual resources for understanding its own problematic. Of course, how best to pursue this philosophical interaction is itself an interesting philosophical question—ideally, itself to be negotiated in that very interaction. The methodological issues raised by this project are explored by Jay Garfield in his essay (­chapter 11), with which we conclude the volume. Each of these chapters advances an important debate, and sheds light on a significant philosophical problem; each of these debates and problems is of interest to Western and to Buddhist philosophy. We offer them, however, not only to further current discussions, but principally in the hope that they will inspire and encourage other philosophers, Western and Buddhist, to join us in this exciting cross-cultural intellectual endeavor.

The Moon Points Back

1

Persons Keeping Their Karma Together The Reasons for the pudgalav ā da in Early Buddhism Amber D. Carpenter “Nāgasena is a name, O King, a convenient designation.” “What makes it convenient, Reverend Sir, to have such a designation?” “That so considering these aggregates together serves our purposes, your Majesty.” “But how is it that just this way of grouping the aggregates and not some other serves our purposes?” “Because our purposes are thus-and-so, and not such-and-such. If they were such-and-such, different ways of grouping and designating aggregates would be convenient, Sire.” “But what are these purposes, reverend Nāgasena, and how? My purpose was to come to this meeting. How can that—the purpose—be just a name for its constituent parts? For it was only this intention that moved me, and the chariot-parts (conveniently named), to get here.” “This must be granted, O King.” “ ‘And what, good Bhikkhu, is my purpose with you? My purpose may have brought you here. But neither my aims nor anyone else’s grew you up from a small boy into the monk you are today. Yet without that, how could I have formed the purpose of inviting you to this discussion?’

2

Amber D. Carpenter

‘Actions and their consequences, Venerable Sir, belong to me, King Milinda, not to the elements (though granted, they are not apart from their elements). Just so, the boy-Nāgasena-elements belong to each other and to the novice-Nāgasena-elements and to the Nāgasena before me now. And this is not because it is convenient to consider it so; rather, it is convenient and indeed possible so to consider it because it is in fact so.’ ‘This is no further fact, venerable Sir, no extra entity; this is just the very fact you already recognize, namely that Nāgasena is a convenient designation, not a random one—that is to say, some causal streams are really individuated.’ ”

You will not find this passage in the Milindapañha as we have it, because King Milinda was not a Pudgalavādin. Neither were the authors of the Dialogues with King Milinda Pudgalavādins, though there should have been plenty of them about to consult regarding improvements to the text, had the authors wished. From around the second century after the Buddha’s parinirvāṇa, there were Buddhist monks adhering to the “personalist” interpretation of the Buddha’s teachings on the self; eventually, at least four different schools included “personalism,” the pudgalavāda, in their view. The earliest espousers of this doctrine flourished,1 and in the seventh century ce, nearly a quarter of Buddhist monks in India were “Pudgalavādins.”2 So it was no negligible fringe group of Buddhists holding that something called the “person” really existed.

1.1 Pudgalavādins as Philosophers This may seem shocking at first. For surely what is distinctive of Buddhist religio-philosophy is precisely the denial of the existence of such a self. Call it a “person,” a “being,” a “living thing,” or whatever you like, that which answers to the sense of “I” is what the Buddha taught us is the source of suffering if “clung to”:  “I too do not see any doctrine of self that would not arouse sorrow, lamentation, pain, grief, and despair in one who clings to it.”3 One might be tempted to think that if people calling themselves



Persons Keeping Their Karma Together

3

Buddhists could deny the existence of “self,” while asserting the existence of a “person,” then the anattā or no-self position of the Buddhist must not have amounted to much. To deny the ātman was no more than a linguistic taboo4—it was not a serious philosophical position, well articulated and maintained for defensible reasons. I think such a social-functional explanation of the no-self position does an injustice both to the studiedly ambiguous position that emerges from the Buddha’s discourses and to the serious effort by the Buddha’s followers to interpret his teachings correctly. It is not altogether clear that the Buddha, as recorded in the discourses, categorically denies anything that might be called self, and every variant by another name.5 In fact what emerges in the centuries immediately following his parinirvāṇa is a dispute over exactly what the Buddha did mean to assert and deny on the matter. And when Buddhists divide over this issue, they do not quibble about words—pudgalavādins and their detractors refer equally to “beings,” “persons,” and “self” in such contexts, the attention remaining focused throughout on whether this supposed entity is or is not an unchanging, independent principle, such as the Brahmanical philosophers call ātman.6 Any distinct, therefore eternal and unaffected, principle with which to identify was certainly denied by the Buddha and specified as a wrong view, leading to suffering. At the same time, the Buddha himself used all the words cognate with self, and frequently reasoned as if a person could be reindentified as the same individual over time. And he was just as adamant that nihilism was not an option. So there was considerable work of interpretation to be done in order to determine what that crucial “right view” on the matter was. Since the best evidence available—namely, the discourses of the Buddha as they had been remembered by his followers in the years following his parinirvāṇa—was both ambiguous and conflicting, this meant that citing scripture was not enough to settle disagreements, though much scripture would certainly be cited in such disputes. Each party could as easily cite passages in support of their position as they could explain through reinterpretation the passages cited against them. Principles of interpretation were developed—in particular, the distinction between literal or definitive (nītārtha) and indirect or non-definitive teaching (neyārtha)—and their application was then in turn contested.7 Not only were the specific applications of interpretive principles contested, so too were the candidate texts to which they might apply. For at the time the pudgalavāda was being articulated, criticized, and defended—and

4

Amber D. Carpenter

persuading many committed Buddhists—the canon had not been fixed among the various parties to the debate. Even if disagreement began as the simple matter of trying most faithfully to interpret the teachings of the Buddha, these difficulties—together with the Buddha’s own admonishment that one appeal to one’s own experience and understanding in seeking the truth, rather than simply taking his word for it—meant that eventually appeal must also be made to what position is the most reasonable and defensible. While a dispute between the Pudgalavādins and their opponents was indeed an attempt to lay claim to “what the Buddha taught,” that dispute was nonetheless philosophically grounded and motivated as well—it was one that necessarily took place through arguments and critique, giving reasons why, among the possible interpretations consistent with scripture, this or that was to be taken as definitive.8 I shall be exploring what those distinctively philosophical pressures might have been that drove or inspired some Buddhists to claim that persons existed, and what theoretical work such a claim might have been doing in their overall view.9

1.2 The Pudgala: Ultimately Real, Substantial, Both, or Neither? There are essentially three substantial sources available to us in attempting to reconstruct the view of the Pudgalavādins, and the reasoning behind it, in its earliest formulations.10 Two of these, the first sections of the Kathāvatthu (KV), and the ninth chapter (or appendix) to Vasubandhu’s Abhidharmakośabhāṣya (AKBh), are written by opponents of the pudgalavāda. Less substantial, but also falling into this category are Harivarman’s Tattvasiddhiśāstra (or Satyasiddhiśāstra), and the Sarvāstivādin Vijñānakāya (which is largely similar to the Theravādin Kathāvatthu, though less elaborate).11 As hostile exegetes, we cannot be certain that they represent the pudgalavāda reliably, and these texts often present it elliptically. We cannot even be entirely certain that the view the Kathāvatthu argues against is the same as the view Vasubandhu argues against—for apart from questions of misrepresentation, there is the possibility that the pudgalavāda itself developed and grew more sophisticated over time, in the face of critique.12 Our third significant source for the pudgalavāda might be thought to operate as a check on such ambiguities, for they are the four texts



Persons Keeping Their Karma Together

5

written within schools espousing the pudgalavāda.13 These are preserved only in Chinese translations, and unfortunately often have to do with the pudgalavāda only in parts; largely they concern other views held by these schools.14 Of these four texts, the Sāṃmitīyanikāyaśāstra has the most useful material for reconstructing the claim that the person exists and the reasoning behind it.15

1.2.1  The Pudgala Exists, Really and Ultimately … According to the Kathāvatthu, the fundamental thesis under dispute is whether “The person is found to exist really and ultimately.”16 By “person” is meant indifferently, “self, being, vital principle.”17 The generous gloss on “person” makes it clear that we have to do here with a dispute over how things are, not a mere quibble over how things are to be called. If our third-century ce commentator, and indeed the Kathāvatthu itself, is not prejudicing the case irreparably against the Pudgalavādin, then it must be that what is meant by these various terms—person, self, being, vital principle—is still to some extent up for grabs. Though a contentious position, the Pudgalavādins were not heretics in virtue of this claim about the person.18 They were never actually treated as non-Buddhists, nor were the arguments against non-Buddhist adherents to the self thought to be sufficient against the Pudgalavādins. In fact, a number of different “self substitutes” developed within Buddhism over time and through argument and reflection.19 What we find is that the Pudgalavādins, like their Buddhist opponents, were very clear about their fundamental agreement with the anattā teachings of the Buddha: there is no substantial, simple, and eternal “self,” with identity and existence of its own, separate and separable from all other factors. Whatever the pudgala (sattva, ātman, call it what you will) may be, and whatever it means to assert its real existence, it is certainly not an endorsement of the Ātman of the eternalists. If the Pudgalavādin’s opponent could prove, as he wished, that adherence to the pudgalavāda could not in fact be distinguished from commitment to an eternal soul, the Pudgalavādin would acknowledge himself defeated and forced to abandon the view. So what should this claim that the person exists “really and ultimately” amount to, if not the assertion of a substantial, independent Self? According to the commentary, to exist really and ultimately means, “that which is not to be apprehended as not fact, like magic, a

6

Amber D. Carpenter

mirage and the like; actual” (real), and “that which is not to be accepted as hearsay; highest sense” (ultimate).20 This bland formulation is not very informative; it indicates only that “really existing” and “existing ultimately” are contrastive notions, contrasted namely with “illusionary” and “what is merely said to be so.” Mirages and magic are not downright and absolutely false; there is something or other really, truly going on when a mirage appears. But what is really there is not what appears to be there, as in the case of magic tricks. So likewise, what is accepted on hearsay need not be false; it is unverified, accepted on the basis of a conversational context. This insistence that the pudgala is ultimately real thus hints at one of the oldest epistemological-metaphysical distinctions in Buddhist philosophy, between that which is ultimately true or real (paramārthasat) and that which is conventionally true or real (saṃvṛtisat).21 There are different ways of cashing out this fundamental distinction between conventional and ultimate reality, and different interpretations of what those ways are.22 But there are two principles which must be observed by any interpretation, and carefully balanced. First, there is no escaping that the distinction is a hierarchical division—“ultimate” is better than “conventional,” more real, more true, more correct, accurate, or valid. Second, and in tension with this, this hierarchy must be maintained without losing sight of the fact that conventional truth must nevertheless be true. One way to think of how these two principles might be jointly respected, without committing to one interpretation over the others, is by appeal to the “in virtue of” relation. That is to say, it is because ultimate reality is as it is that particular conventions are indeed correct or true; they are well grounded.23 Conventional truths are true in virtue of how things stand ultimately, even when they do not actually describe or capture that ultimate reality accurately. This distinction makes extremely minimal claims about the respective natures of ultimate and conventional reality, or criteria for determining into which class a candidate truth might fall. There is primarily only the asymmetrical relation between them to tell the difference. Thus the much-used example of the chariot as a conventional, but not ultimate, reality: It is because of the pole, because of the axle, the wheels, the body of a chariot, the flag-staff of a chariot, the yoke, the reins, and because of the goad that “chariot” exists as a denotation, appellation, designation, as a current usage, as a name.24



Persons Keeping Their Karma Together

7

In addition to the dependency relation, what is evident is the close connection here between convention and convenience, and the way we talk—or conceptions. Conventional realities are those convenient ways of talking that may not be strictly accurate, but they are very effective in communicating efficiently, for pragmatic ends—so that it is no wonder that the later distinction, between substantial and conceptual reality (dravyasat and ­prajñaptisat25), becomes assimilated to the earlier distinction between ultimate and conventional existence.26 That conventional reality (saṃvṛtisat) should be coextensive with conceptual reality (prajñaptisat) may not be objectionable; all of our conventional truths will be conceptual.27 But the assimilation of ultimate and substantial reality is not obviously so harmless. For this introduces something decidedly new:  before, we could be open about what might turn out to be ultimately real; now, it is asserted that only substances are ultimately real. The only way to be ultimately, really real is to be a substance—paramārthasat is dravyasat.28 Any other sort of reality is derivative, based on substances and real in virtue of their reality.29 This is not an implication of the original distinction between saṃvṛtiand paramārthasat—it was not, for instance and as we saw above, any part of the later but essentially conservative gloss on existing “ultimately and really.” Yet the metaphysics within which the Pudgalavādins—like almost all Buddhists (Mādhyamikas excepted)—operated came to recognize only two ways in which something can be real:  either it is a substance or it is conceived based on substances, these conceptions being designated true (or realities) in virtue of being based on, or dependent upon, what is ultimately (or really) real—namely, substances. Indeed, it is probably more correct to say that the only mode of being recognized is substantial being, for the conceptions derived based on those have a similarly derivative existence—insofar as they are real, it is in virtue of the reality of the substances on the basis of which they are conceived. In this respect, my choice to render these ultimate realities by “substance” is deliberate: they are that which underlie phenomena as we experience them, and that which thus grounds or justifies those experiences.30 But this, note, goes one step further than the original distinction between ultimate and conventional reality. One important consequence of the assimilation of substantial to ultimate reality is that qualities, for instance—conceived as modifications of substances—are not really real. Quantity, position, relation, and all the other ways in which substances might be disposed, taken singly or

8

Amber D. Carpenter

in combination, are not ultimately real.31 Either these experienced phenomena are themselves substances—in the minimal sense of independently specifiable individuals32—or else the experience of them must be explained in some other way, via the imposition of our mental activity onto the really existing substances. The Abhidharmikas, that is to say, stoutly reject any categorial schema.33 Substances themselves may come in different kinds, of course, but it would be wrong to say that they differ from one another in their qualities, for this would imply a substance distinct from its qualities, and in which the qualities inhere—and this is precisely the sort of metaphysical picture that is rejected. It may be more correct to say that each substance, each ultimately existing individual, simply is the quality or factor which differentiates it from others. What exactly it means to be a substance, and therefore what substances there are, can be explained in different ways. Substance may not be “stuffy” or material in any way, for instance. For the Buddhists, substance centrally includes the notions of individuality and exclusion. We see already in the Kathāvatthu the expectation that if two entities are both ultimately real, then they must be distinct from each other (see Kathāvatthu 1.1.130 ff). If each individual obeys the relentless laws of homo-categoriality—particularly, exclusion of other—then both distinctness and identity between two entities do not admit of degrees.34 In the absence of any differentiation in modes of being, and in aspects of being, any two (purported) substances are either in fact the very same individual substance, unqualifiedly; or else they are utterly distinct from one another. Over and over again we see the Abhidharma philosophers—whether in the Kathāvatthu or Vasubandhu in the Abhidharmakośabhāṣya—pressing this point insistently against the Pudgalavādins: their “person” must be either identical with the substantial entities known to exist (in which case their claim collapses), or it must be utterly distinct and distinguishable from those, and so be a substantial, independent individual in its own right (the unacceptable “eternalist” position).35 If it is neither the same nor different, then it cannot be a substance at all—and this, in their view, is tantamount to acknowledging that it is not ultimately real.

1.2.2  … but not in the way of a substance This final move is the one the Pudgalavādin resists. The person admittedly is not a substance; nevertheless, we cannot therefore conclude that the person does not exist ultimately. The pudgala might be really real, but



Persons Keeping Their Karma Together

9

“not in the way that (other) real and ultimate things are” (Kathāvatthu 1.1.1). By insisting that the person is ultimately real, the Pudgalavādin wants to reject the standard Abhidharma view that the person is merely prajñaptisat—a conceptual or conventional reality, whose reality is a borrowed one, grounded on the ultimate reality of its constituent parts. But they want to do this without rejecting the Abhidharma metaphysical framework altogether—so that, while their position does not seem an unreasonable or especially complicated one, it turns out to be an incredibly difficult position to articulate. So when we come to Vasubandhu’s fourth-century engagement with the pudgalavāda, he offers the Pudgalavādin the choice between the pudgala’s being a substance and its being conceptual—and he has the Pudgalavādin first consider what is at stake in being substantially real before choosing the “conceptual” alternative. This concession, however, should not be taken as a retraction of their insistence in the Kathāvatthu that the person is ultimately real; it is rather a recognition that “ultimately real” has come to mean something else, and is an attempt therefore to defend the same position in different language.36 Thus, the Pudgalavādin insists again, the person is not conceptual in the way of other conceptual realities. This special sense is then illustrated by an analogy with the relationship between fire and fuel.37 Although it is true that fire is conceived on the basis of fuel combusting, fire, the point seems to be, requires nothing else for its conception but this. The internal or necessary relation between the combustible and combustion suffices to ground the grasping of the aggregates as “fire”—unlike other conceptual realities, we do not require, in addition, that it be somehow conventional, convenient, or useful so to grasp the aggregates. So similarly, with the person: the aggregates themselves cause their being conceived as a whole; there is no appeal to usefulness or practice required in addition. This makes their collective being as a whole independent of any considerations beyond how they actually are. Thus the special sense in which the pudgala is uniquely “conceptual” is consistent with its also being ultimately real, but not in the way of a substance. “Ultimately real, but not in the way of a substance” and “conceptually real, but not in the way of other conceptual realities” come to much the same thing. The way in which the ultimately real dharmas exist together is itself ultimately real, and so suffices—without appeal to convention or purposes or use—to cause in us the conception of “person.” This is, however, not something readily expressed in either version of the conventional-ultimate distinction.

10

Amber D. Carpenter

Had the Pudgalavādins not wanted to retain Abhidharma metaphysics as a whole, their position would not be so difficult to articulate. For consider one obvious way to hold that the person is real, but not a substance: introduce a variety of categories, all of them various ways of being really real, only one of which is “substantially.” A  Naiyāyika, or an Aristotelian for example, might ask whether it is even coherent to claim that the only way to exist is substantially—even substances cannot be, if there are not other modes of being by which they are qualified and individuated.38 To be is to be a “this something,” and this requires that there be kinds and qualities and relations. If qualities, relations, and so on can be really real, just as real as the substances which they qualify and relate, then the person could be some such nonsubstantial, ultimately real entity.39 But the Pudgalavādins do not want to take that route. Their argument is not with the criteria of being—with what existence means, nor therefore with what existing things there are, in general. Their aim is not a wholesale rejection of the Abhidharma metaphysical project. Much more modestly, they want within that framework to articulate a single principle which struggles to find voice, and which they feel must nevertheless be accommodated. The Pudgalavādins want to insist that the person is really, ultimately real, without its thereby being a substance; and, conversely, they want to insist that though the person is not a substance, neither is it thereby only a conceptual entity (and so only derivatively real).40 And this is a real challenge to the Abhidharma metaphysics of their opponents. There can be real things that are not substances—at least, there must be this one, the pudgala. (Why they want to claim this will be investigated in sections 1.3 and 1.4.) And since being ultimately real means precisely not having derived or deceptive reality, as conventional realities have, there are nonsubstances that are not conceptual realities—the distinction between conceptual or conventional reality, and substantial reality is not exhaustive.41 Since the Pudgalavādins certainly do not want to assert that the self is a substance, we might put their view more circumspectly as the claim that “the person is not just a conceptual reality.” Although there is of course a concept of a person, the correct account of that concept will not follow the manner of other constructed concepts. In the ordinary case, a conceptual reality is grounded on its constituent elements, and borrows its reality from them; but it gets its identity as a whole from the efficacy of using them. The constituents, by contrast, are all different from each other, different from the complex reality conceived, and they are ultimately real. In the case of the person, however, conceiving it is grounded just on



Persons Keeping Their Karma Together

11

constituent elements. Something about them or how they are suffices for their identity as a person—there is no additional appeal to convenience and conceptualizing required. Their “being a person” is the ultimate reality on which conceiving a person here is grounded. Thus, it is in virtue of eyes, hands, volitions, cognitions, and the pudgala that one correctly conceives of “a person.” Since the challenge is to the exhaustiveness of the dravyasat-prajñaptisat distinction, or—to put the same thing another way—to the assimilation of paramārthasat and dravyasat, it is no surprise that in the Kathāvatthu these two different ways of formulating the pudgalavāda should come to the same:  (1)  The person is found as ultimately real, but not in the way other ultimate realities are;42 and (2) although there is indeed a conceptual reality to the “person,” the story of its reality is not the same as that given for other conceptual realities.43 My suggestion is this:  It was the attempt to assert the ultimate reality of something not a substance, without dropping the basic noncategorial schema, that drove the Pudgalavādins to express themselves in the neither-nor language for which they were viciously mocked. And it was the attempt to articulate a kind of reality for which there is no space within the metaphysical framework which drove the Pudgalavādins to assert that the person is avaktavya, literally “not to be said,” or ineffable, inexplicable.44 For what it positively is is literally—without abandoning the framework entirely—unsayable. And yet it is from within the framework that the need to posit some such reality is felt to arise. Unfortunately, both the “neither-nor” language and the avaktavya claim make it difficult to determine what exactly their position was and, more importantly, why they held it. In our efforts at what must then be a bit of philosophical imaginative reconstruction, we will discover that the “what” and the “why” mutually illuminate each other: as we uncover what the person is, so will the reasons for insisting upon its reality become clearer.

1.3  What is the Pudgala, and What is it For? What is the problem that positing a really existing person is supposed to solve? Why insist it really exists? And how is it not a substance?

1.3.1  Unity With, and Without, Selves The substantial-self hypothesis of Brahmanical thought serves three philosophical functions:  it provides a subject (enjoyer), an agent, and a

12

Amber D. Carpenter

unifier.45 The Self—the substantially existing individual, distinct from the transient physical and psychological events in a life—is that which has experiences, initiates action, and that in virtue of which temporally, numerically, and modally distinct experiences can belong to, or be parts of, the same whole. These phenomena of cross-modal unity and unity over time, and these logical principles of experience (agency and subjectivity) are what the Buddhist who wants to dispense with self must explain in some other way.46 Of these outstanding explanatory tasks, the Pudgalavādin is primarily concerned with problems of unity in multiplicity, and questions of individuation that attend them.47 This is not a new observation—those scholars who have concerned themselves with the pudgalavāda have generally observed that the central concern is something to do with unity.48 We witness this interest in complex unity—that is, unity-in-multiplicity—in the Pudgalavādins’ three ways of designating the person: namely, on the basis of the aggregates, on the basis of transition, and on the basis of cessation.49 It is important to be clear that these are grounds for designating a group of elements as a pudgala;50 although the pudgala is not without them, they are not what the pudgala is. In marking these out as the territory of concern, the Pudgalavādins are clearly interested in the unity consisting of constituent substances at a time (aggregates) and in the continuity and the distinctness of continua over time (transmigration and cessation); they are interested, that is to say, in individuation, which becomes especially clear in designating the person based on cessation, where questions of unity-in-multiplicity cannot arise, though continuity and individuation of continua can. In the matter of the person, or self, the Pudgalavādins position themselves as a third alternative—neither Brahmanical nor Abhidharmika. As is the way in such Buddhist debates, this will be articulated as the middle and moderate position between two unacceptable extremes—between the eternalism of Brahamanical thinking, which takes self to be an independently existing unity, and (on the other hand) the inadvertent nihilism of the other Abhidharma Buddhists, who have done away with a really existing self altogether.51 Naturally, the non-Pudgalavādin Buddhists see it otherwise. They present their own position as the middle between extremes—casting the Pudgalavādins as inadvertent eternalists—by acknowledging the conceptual reality of the self (thus avoiding nihilism), while denying its substantial reality.52 The self is, then, like every other purported complex whole,



Persons Keeping Their Karma Together

13

a supposed unity projected by the conceiving mind onto a multitude of really existing discrete substances. While there is basis in ultimate reality for each of the constituents, there is no basis in ultimate reality for considering that there is some one thing there; for ultimately, they are not one, but many.53 The convention that there is one thing here will be true, but only derivatively, in virtue of the reality of the many constituents thus conceived. Were the unity itself, as such, to have some basis in ultimate reality, the story goes, then this basis would have to be some new thing, distinct and existing separately from the other constituents. This is a consequence of the noncategorial metaphysics which recognizes no “modes” of being.54 Something existing in this way would of course be a substantial individual, on a par with its constituent substances. Such substantial unities in the case of individual living beings are Selves. The Brahmanical philosopher readily objects that this Buddhist nonself view cannot account for various experiences of unity—e.g., complex perception, memory, desire. If experiences are not anchored to individuated owners, the Naiyāyika objects,55 then they might arise in any old order, so that it would be possible for Devadatta to recall what Yajñadatta once experiences. But in fact this does not happen. Vasubandhu’s Abhidharmika response is simply:  “there is no connection [between causal continuum Devadatta and causal continuum Yajñadatta] since these two minds are not related as cause to effect within one continuum.”56 That is to say, it just happens, as a matter of fact, that experiences do not ever take that kind of “transpersonal” order. There need be no unitary owner to explain the fact that Devadatta “recalls” only his own previous experiences. The relations between elements are those of spatiotemporal proximity—contiguity (at a time), or succession (causal connectedness).57 There is no other way for elements to be connected, and there is no significant differentiation within this kind. Any two elements either are spatiotemporally connected, or they are not. These connections follow patterns, some of which we conveniently designate “persons.”

1.3.2  Inadequacy of Selfless Unity But when we consider how our bodies are in constant interaction with our physical environments, and indeed with other minds, it becomes clear that this answer will not do.58 Causal connectedness cannot be all there is to “belonging to,” or individuating groups of aggregates or causal streams;

14

Amber D. Carpenter

it does not suffice to decide the question of which elements belong to which other elements or to which causal continua.59 For causal continua do overlap and intersect. Consider:  My verbal expression may give rise to several instances of hearing; but only one of those is counted as being in the same continuum as the vocal expression. Your intention to help may cause innumerable consequences—movements, sounds, feelings, desires. But only some of these effects will be counted as being in the causal continuum conventionally designated “you”; some will not. If all we have is the one sort of relation between ultimately real substances, then we cannot appeal to this to explain the difference. Although this point is not explicitly stated, it is what makes the Nyāya challenge “Why doesn’t Devadatta remember Yajñadatta’s experiences?” a potent one. Individuating causal streams into distinct continua presupposes that some connections are special in some way—and our convenience, these just happening to matter to us more than those (or to be more useful60), won’t do, for it is the fact of these special kinds of connections really existing that makes such conceptions useful.61 The Buddhist might reply, as Vasubandhu does, that it is simply a fact that this experience-dharma here does not cause a subsequent experience-dharma there.62 But since this is not always the case, since sometimes some dharmas within one continuum-stream do cause dharmas to arise in other continua, this response does not meet the objection. If Devadatta’s vocal act causes a hearing-consciousness to arise “over there”—in what one would conventionally call Yajñadatta—why is that hearing-consciousness not part of Devadatta, especially if that hearing-consciousness is causally connected to the pain-consciousness that arises as “Yajñadatta’s fist” hits “Devadatta’s face”? Alternatively, the question is, why would it never be convenient to consider them so? The hearing-consciousness, like the speech-intention and the speech-act, are all causally connected to the aggregate-stream conventionally designated “Devadatta.” There are two reasons, I think, why appeal to the principle of convenience and efficacy governing concept formation will not suffice to individuate persons. The first is that making judgments of convenience relies on two factors, both of which are problematic in the case of persons. The first factor is purposes:  it is only convenient or efficacious to group aggregates one way rather than another provided we have some ends for which so grouping them is useful. The King wants to be able to get to his meeting with Nāgasena, and in general wants to be able to get to various places quickly and comfortably. And so it is convenient to think of a



Persons Keeping Their Karma Together

15

certain aggregation of elements as “a chariot.” But such ends are themselves constructed, arising only within conceptual reality. Conventional reality is thus presumed in making judgments of convenience. But this conventional reality is populated with persons, above all oneself—it is by thinking of Nāgasena as “whole” that King Milinda is able to conceive a desire to see him, and it is only presuming himself “the same person” at the beginning and end of any journey that the King might conceive a desire to go anywhere. The circularity seems to be this: it is convenient to grasp a group of aggregates together as a person only given some ends; and these ends are conceivable in the first place only by thinking in terms of persons.63 Having ends presupposes the individuation that “useful for some end” is supposed to engender. The only end which may not be thus dependent upon presuming persons would be aiming at nirvāṇa—and for this, thinking in terms of persons is a positive obstacle.64 The second factor on which judgments of convenience rely is that ultimate reality be such that constructing concepts this way rather than that way is effective.65 If, however, the appeal to efficacy is circular, and so empty, in the case of conceiving persons as unities, then this element of a supposed judgment of convenience drops out, and the claim must now rest entirely on an appeal to how things ultimately are. That is, it is convenient to grasp these many dharmas as a unity because these dharmas really are, in some significant sense, related exclusively to each other—and not to other dharmas with which they causally interact.66 But this, I suggest, just is the pudgalavāda. The second reason appeals to convenience are inadequate is similar to the problematic presumption of persons in order to have (most) ends at all. Given our shifting concerns, it seems likely that different attributions of belonging would be more convenient under different circumstances. Gathering one subset together as “the person” may be convenient under some circumstances, while grouping aggregates differently is convenient under other circumstances, or for other ends. What the Abhidharma Buddhist has not explained is how, if persons are purely “convenient designators,” they remain the same throughout changes in ends and purposes. If, however, it is overall convenient to have conceptual unities that do not so shift their boundaries, then this brings us swiftly to the deeper question of appeals to global unity, and the entitlement or conditions for making such an appeal. I will address this in section 1.4.2.2. The Nyāya-Vaiśeṣika opponent might think that the only way to explain this is to posit the real existence of two distinct individuals—Devadatta

16

Amber D. Carpenter

and Yajñadatta—to whom some but not other experiences belong. This is where the Pudgalavādins want to stake their middle ground. They are alert to the fact that the ultraminimalist Buddhist view inevitably presumes the individuation of person-constituting aggregates and person-constituting streams.67 The view must be able to discriminate between distinct persons—particularly, as we will discuss below, in order to articulate the phenomena of perception and karma; and it can do so only by in practice recognizing that some elements and streams belong together and not to others.68 But the Pudgalavādins want to grant the metaphysically most minimal claim possible; so they will insist on the reality of “these belonging together,” but without supposing the separate existence of someone or something to whom these elements belong.69 They belong to each other, not to some further substance.70

1.3.3  Pudgala-Unity Such a view recognizes that individuation of causal continua is presupposed, not projected. That some elements belong to each other, and not to any others, is really, ultimately true. This would mean that we are correct—getting rightly at the way things ultimately are—in regarding just these and not other constituents together, as being bound to one another in a way they are not bound to other ultimately real elements. But it is not so in the way that other ultimate truths are, for it is not a substance; it is not an individual with a separate, distinct identity of its own.71 The basis for regarding these aggregates as a person is not just the aggregates, but more precisely “the fact that these elements are thus-and-so disposed with respect to one another.”72 We have, then, two reasons why the pudgala will not conform even to the very early, canonical Abhidharma notion of a relation: First, the pudgala is not a single dyadic relation between two individuals, nor could it be conceived as a series of such relations without missing the point; second, the pudgala is not an isolatable, distinct individual with its own unchanging identity, as the paccayas are meant to be. Without simply being the aggregated elements (for the pudgala is rather the fact of their belonging together, not to others, and not others to them), still it is not different from or separable from these.73 So far from excluding other substances (as one substance excludes another), “persons” so understood would in fact have their particular identities determined by their constituent substances. Indeed, we can see that the pudgala described here would even



Persons Keeping Their Karma Together

17

necessarily reflect the particular nature and qualities of different elements taken together. Thus we see in the Kathāvatthu how the “person” takes the character of the aggregates, without becoming the same as them.74 According to the Kathāvatthu 1.1.171–181, the person, though conceived based on aggregates that are transient, does not share this transience with them (§171); yet it does share in other of their properties: the concept of good or bad person can be derived from good or bad feelings respectively (§174–175), and also from good or bad consciousness (§177–178); and even the concept of a seeing person can be based on sight (§180). The pudgala inherits the distinctive qualities of its aggregates, but it does not become two persons just because there are two elements (§183). And while each moment of consciousness has or belongs to a person, it is not the case that each new moment signifies or belongs to a new person (§186, 193). All of this is exactly how it should be, if the person is regarded as the reality that these and no other aggregates belong only to each other. This is one way to read the much discussed fuel-fire analogy, without overreading it.75 The Pudgalavādins liken the relation between pudgala and its aggregates to the relation between fire and fuel.76 A lot of energy can be put into trying to think exactly what the relation between fire and fuel is—how it should be different from the relation between chariots and their parts; and then further, how the philosophers of that time might have expected us to conceive of the fire-fuel relation. But the sūtras collected as the discourses of the Buddha give us on several occasions the observation that a fire is named according to its kind of fuel. Thus, for example Majjhima Nikāya, sutta 38: a fire is reckoned by the particular condition dependent on which it burns—when fire burns dependent on logs, it is reckoned as a log fire; when fire burns dependent on faggots, it is reckoned as a faggot fire; when fire burns dependent on grass, it is reckoned as a grass fire; when fire burns dependent on cowdung, it is reckoned as a cowdung fire.77 Different fires are distinguished in kind and named according to their kind of fuel. That this is the correct way of distinguishing fires suggests a particularly close relation between fuel and fire, without identifying them. It is precisely by not being able to exist separately from the fuel that the fire is rightly named according to its fuel, without being

18

Amber D. Carpenter

identified with that fuel. So similarly with the person:  without being identical to the aggregates, it borrows its qualities from them, and cannot be otherwise individuated.78 This point will hold just as much for complex unity at a time,79 as for unity through change over time.80 Just as it is ultimately true that these aggregates belong together, so is it ultimately true that some elements existing at different times genuinely belong to one another, and not with others. This is why the Pudgalavādins argue that to deny the santānin, “possessor of continuity” would indeed be to deny continuity (santāna).81 This need not be a rehash of the Nyāya-Vaiśeṣika claim that there are no experiences without an experiencer—indeed, there is reason to think that the Pudgalavādin would reject such a claim.82 If the santānin, “possessor of continuity,” is the pudgala, and the pudgala just is the fact that these and not other aggregates are connected to each other as they are not to others, then denying the pudgala will indeed be tantamount to denying continuity. In the metaphysics of nonsubstantial individuals conceived as “streams,” persons are “stream-individuators,”83 though not by being themselves individuals, nor by being agents holding or bringing dharmas together. Rather the pudgala is the fact that these dharmas do really belong together; it marks the recognition that this is an ultimate truth, really real, not a convenient convention or handy way of conceptualizing what is ultimately really there. On the other hand, it is not a substance; not dravya-sat, nor “haecceity” or “thisness.” Nor is the person the provider of individuating characteristics—all characteristics are provided by the constituent dharmas. Moreover, any set of aggregates would fail to capture the person unless it included not only every element that ever arises in this causal stream, but also somehow included just that cessation to which this stream led.84 The aggregates, and even the transition from one group and kind of aggregates to another, are the bases of the person; but they cannot exhaustively define the person. What the Pudgalavādin sees—and his Buddhist opponent would rather ignore—is that mere proximity at a time and mere causal connectedness over time are not sufficient to determine which are the correct and incorrect ways of grouping aggregates in our conceptualizations of “many as one.” Neither is “our convenience” the missing part of the explanation. It would be, at least sometimes (and perhaps generally), more convenient not to recognize certain causal connections as distinct, integrated, and excluding all others—for many aggregates rightly excluded from some unity might well be causally connected to it in relevant ways. Moreover,



Persons Keeping Their Karma Together

19

appeal to “our convenience” (or indeed, to “relevance”) presupposes purposes for which some, but not other ways of subdividing aggregates will be useful. But such aims can arise and be articulated only within conceptual reality. That is to say, having purposes at all presupposes subdivisions among generically similar causal connections which are supposedly only made based on the principle of convenience. The Pudgalavādin recognizes and avoids this circularity by acknowledging that, in one central case, our convenience does not determine but rather tracks the different ways in which aggregates are related—either as forming a unity, or as being collocated or causally connected without forming a unity. Taking certain collocations of elements to be a single thing is correct because they really do belong to each other in some way that they do not belong to the other elements.85 Do we want to say that the pudgala thus explains why persons do not intersect?86 Since it is itself “inexplicable,” we might rather say that pudgala names the fact that there are causal chains and connections which are special and different from others (the difference between my perception causing my subsequent perception, or my perception causing my desire, and my perception causing your perception, or my volition causing your desire). The rival Abhidharma account recognizes, and can recognize, no real difference here.87 But that these elements belong to each other, and not to other elements also proximate or causally connected, the Pudgalavadins contend, is a basic fact which we perceive; and they recognize that it is only by perceiving this that we can claim that certain associations of elements are significant, and that our discrimination of distinct causal continua as distinct are correct. This, I  suggest, is what it means to say that the Buddha, using his omniscience, “sees beings decreasing and being reborn,” as the scriptures record, and yet the person is not itself a visible object like color (Kathāvatthu 1.1.198–199). According to Vasubandhu, this point is generalized across all sense modalities, as indeed it should be if “perceiving the person” amounts to perceiving that certain elements or streams belong together. Vasubandhu presents the claim as a response to an epistemological objection, regarding how we come to know the person: They must state by which of the six consciousnesses a person is known to exist. They say that a person is known to exist by all six. They explain how by saying that if a consciousness is aware of a person in dependence upon a visible form known to exist by means

20

Amber D. Carpenter

of the eye, it is said that a person is known to exist by means of the eye; but it is not said that the person is or is not the visible form.88 Vasubandhu criticizes this claim, insisting that either one perceives only the aggregates, in which case “the same account can be given of milk or other such [complex] things”—so the person is every bit as conceptual as, and no more ultimate than, every other complex phenomenon—or else one must see something distinct from them, which would make the person a visible (or audible, etc.) phenomenon, and one separate from the rest of the visibles (audibles, etc.) constituting the basis on which we conceive the person. Even worse, Vasubandhu continues, since “each of the five organs encounters its own domain and objects; none encounters the domain and objects of another,” if the Pudgalavādins insist we might perceive the person equally with any of the sense modalities, then the person must be both essentially visible and essentially audible and essentially tactile, and so on. And this is clearly incoherent.89 But the Pudgalavādin’s position here is not as incoherent as Vasubandhu makes it appear.

1.4.  From the Epistemology of Persons to Metaphysics 1.4.1  Answering Vasubandhu There are two towers on the hill, at some distance from each other. There is no question but that you come to know the towers by perception, and this perception is visual perception. You see the towers, and this is your access to them. Now, what about the space between them, or the fact that they are at some distance from each other? Space, according to the Buddhists, is acausal, so the distance itself cannot give rise to the perception of distance.90 But this knowledge or awareness also cannot be identical simply to awareness of either or both of the two towers. They are perceptions of towers, and as such do not themselves indicate anything about distance, which is not of course a tower. This question is simpler than, but structurally similar to, the question of how we could come to know the pudgala, if the foregoing account of what the pudgala is meant to be is broadly correct. Within the Buddhist framework, there are two options:  you come to know this either by perception or by inference. It may feel like it is just perception. But of what, and via which mode of perception? If this seems



Persons Keeping Their Karma Together

21

impossible to answer, we might then have recourse to inference, as follows: I see one tower; I see a second tower; and I infer from their nonidentity that there must be a distance between them.91 But it is immediately clear this merely relocates the same problem. For if it is by their nonidentity that I draw this inference (and how else would I? What other sort of inference might it be?), then we must wonder how it is that the nonidentity was apparent to me. Did I  perceive that these two towers are not identical, or did I infer it? If an inference, on what basis? If a perception, what strange sort of perception is this? “Nonidentity” is not a visible quality, nor an audible one… and so on. Perhaps, then, we should return to our initial inclination to say that we perceive the distance between the towers, and face head-on the question: what sort of perception is this? It is not a visual perception of the space itself, since that causes no perception whatsoever to arise; and it is not the same as the visual perception of the two towers or of either one of them. In this case, however, one might say there is a positive perception of a grassy knoll, namely of precisely that grass knoll by which the towers are separated.92 It is by this that one infers the distance between the towers. This will not quite suffice—for now I perceive three things: two towers and a grassy knoll. And if it is by inference from these perceptions that I should come to know the distance at which the towers stand, then similar issues of ordering and relatedness arise: Is it by perception or inference that I recognize the difference between the two towers, and that the grassy knoll is between, not beside them? Since there is nothing for such an inference to be based on—except a perception of the very fact—it seems here too we must appeal to perception. This, I suggest, is the kind of issue the Pudgalavādins are trying to raise about the person. Just as, apparently, in seeing the two towers I thereby see they are at some distance from each other, so in perceiving some portion of perceptible person-constituting aggregates I  therewith perceive the person. And since person-constituting aggregates belong to various kinds, we might appeal to any sense modality in the same way as a route to come to know the person. It is not the distance between the aggregates, however, but their belonging together—belonging to each other—that one perceives through any perception of person-constituting aggregates. Thus the epistemological claim of how we come to know the pudgala crosses into metaphysical territory of what is there to be known.

22

Amber D. Carpenter

1.4.2  Persons v. Nonpersonal “Wholes” If this line of thought shows anything, there is a danger it may thereby have shown too much. For it looks now as if I perceive some whole—“two towers at some distance from each other”—every bit as ultimately real as the person. But the Pudgalavādins do not actually want to say that in every composite perception some whole is perceived. There is supposed to be something special about the aggregates comprising the person. Now in the case of the towers on the hill, it looked like there might be an alternative account—what feels like perception is in fact a subtle inference, so that there need be no “distance” as such perceived. Can this deflationary, inferential, account of perceiving the distance between the towers on the hill be made to work, after all? If so, does it work as well in the case of perceiving a person in the aggregates? If so, we may have lost one argument for supporting the pudgalavāda. If not, however, and if the inferential account does work for towers, then so much the better. We will have seen the kind of point the Pudgalavādins want to make, without making that point apply indiscriminately to everything.93 In the case of the towers on the hill, I might perceive a tower, perceive a grassy knoll, and perceive a tower; and then I have to put all these discrete moments of eye-consciousness together. This would be the story of how I perceive that there are two towers at some distance from each other, without having to grant what seems like “seeing the distance between the towers” or “seeing that they are at a distance.” Now there must arise here a question of composition: If these are discrete moments of eye-consciousness, how are they to be put together? What ordering principles are there? How do I know the difference between two moments of eye-consciousness of the same tower, plus a moment of grassy-knoll-eye-consciousness, on the one hand; and an eye-consciousness of a tower, one of a grassy knoll, and one of a second tower, on the other? (The tower-eye-consciousnesses need only be similar, not even identical, for this to be a problem.) Instead of appealing directly to perception, we might instead recall that I never have just three moments of visual consciousness at my disposal. There are innumerable mental impressions arising. These three visual impressions at issue must be so ordered as to make most sense of the rest of the information available to me, so that I am able to function in the world and navigate my way round it. This would be something like the principle of convenience: when confronted with a manifold of simples, the mind so orders these as to be successful and efficient in action.



Persons Keeping Their Karma Together

23

One might wonder whether this supposedly deflationary account can in fact dispense with assertions about reality genuinely being so ordered, and this order being fully real. But let us allow for the moment that it can, for there is another question more relevant to the intra-Buddhist dispute about the person: Supposing the inferential account succeeds for grassy knolls and towers, can it work just as well for persons and their aggregates? Or in that case is there yet something left unaccounted for? We will find, I  think, two reasons why the inferential account given of how two towers can be known to be at a distance from one another will not work when we turn to explaining the perception of the person.

1.4.2.1  What Is Perceived When There Is a Pudgala, and Otherwise Not? The first reason is the absence of a suitable correlate to play the role of the “grassy knoll” in our tower example. If I can just infer from seeing the grassy knoll that the towers stand at some distance from each other, this is on the basis of something I straightforwardly perceive: the grassy knoll. But in the case of the person, there is no such basis. Persons differ at least from the example given, if not from all complex unities, in having no possible basis for such an inference except the person itself. This we must then come to know by perception. The explanation unnecessary for towers on hills is unavoidable in the case of persons: Through perceiving the aggregates, any subset of them, through any modality, we perceive by that same sense-modality that just these aggregates belong together. And I am able to do this because there is really something there to perceive—but that can be nothing other than their belonging together, or to one another; that is, the person. What is this “belonging together”? What does one perceive in perceiving the aggregates that makes it correct to judge “this is a person”? It cannot simply be the recognition that these aggregates exist in close spatiotemporal proximity, for this is generally true of all aggregates that are merely imputed to be wholes; in fact it is also, as we saw above, true of elements both within and outwith the set of elements rightly judged to be constitutive of a particular person. We perceive “these belong to each other” only in some special cases. One plausible way of describing what is special about these cases—and which also captures the sorts of examples appealed to in our pudgalavāda texts—is that they involve the sort of groupings at a time and over time that are marked by new functionality, and by development.

24

Amber D. Carpenter

Thích Thiên Châu cites an analogy drawn between the person and an eye: Just as, through a reunion of the four great elements (caturmahābhūtasaṃyoga) there is a dharma “eye” (cakṣus), so, through the reunion of the five aggregates (pañcaskandhasaṃyoga), there is a dharma “individual” (pudgala).94 Functional unity certainly is something special about persons. Because they are just this sort of bundle of matter, conception and intention, they can do particular kinds of things, like commit murder (something with specific karmic consequences), rather than just disturb other agglomerations of elements in various ways. But if the self were absolutely non-existent, then there could not be the killing of beings nor would the killer have killed anything. There would be nothing like theft and robbery, heresy and lewdness, ­telling lies and drinking wine… 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 nor any result thereof.95 Murder cannot be done by aggregates taken as such, for all they do is to rearrange other aggregates. The killing disappears. Killing is something that only certain aggregations can do, because they are together, and taken as a whole. And indeed, it is something about functional unity that explains why one can be guilty of killing a cow, but not of “killing” a clay cow. “If living beings are only names, then just as killing a clay ox does not entail the guilt of killing, so if one kills a real ox there should be no guilt.”96 There is an important difference between living and nonliving beings, which means that living beings can do things and have things done to them that inanimate beings cannot. It is tempting, then, to think of this special coherence generally in terms of the emergent characteristics such bundles have in virtue of being bundled precisely with each other and not with others. The trouble is, mere functionality does not distinguish persons from chariots. The very reason it is convenient to designate certain bundles as “chariots” is, presumably, the new or emergent functionality of chariot-constituting aggregates—a phenomenon noticed as an “emergent functionality” only because of our interests in moving easily from place



Persons Keeping Their Karma Together

25

to place. Only such and such a configuration of such and such aggregates gives rise to a vehicle-capacity, and this capacity is what interests us about this particular conjunction of aggregates, just so (and not otherwise) delineated. True, chariots cannot be killed, but they can be destroyed. Yet persons are ultimately real, and chariots are not. Evidently the new functionality present when person-constituting aggregates are present is thought to differ significantly in kind from the new functionality that emerges from the aggregation of chariot-constituting elements.97 And indeed, one might well think there is something distinctive about organic functioning—something reflected in the difference between being dismantled and being killed. But pointing to functionality itself does not sufficiently get at this difference. Perhaps focusing on notions of “development” will be more illuminating. To illustrate development, the Kathāvatthu offers the example of a girl becoming a woman (KV 1.1.193), and Vasubandhu supplements the point with examples of a person becoming a grammarian, or a priest (AKBh §3.4).98 Such aggregations display characteristics markedly different from other aggregations, such as chariots.99 When we speak of a girl becoming a woman, the very notion of development picks out an internal coherence through change, independently specifiable and indeed making sense as development only by reference to the previous stages.100 In contrast to the case of the chariot, it is precisely by disregarding considerations of usefulness, and looking only at the different parts and their relations at different times, that we can make sense of the notion of development, and growth.101 The different stages in the succession and the different elements at a time have a coherence that cannot be explained in the same way as the unity of chariot parts over time—as our convenience, our aims, and our mental activity simplifying the manifold disparate elements and their conjunctions. What we acknowledge in the very conceiving of “becoming a grammarian” or “growing into a woman” is the fact that certain elements really belong only to each other, and that it is only in recognition of this that we conceive of the girl growing into a woman in the first place. On my hypothesis that the pudgala amounts to the fact of certain elements really belonging only to each other, the claim is that this “belonging to” is basic—ultimately true, and not explicable in any other terms, or by reference to anything else. We may now have a criterion, or joint pair of criteria, for distinguishing which bundles imply persons. And we have thereby the beginnings of an explanation why there must be ultimately real persons for any

26

Amber D. Carpenter

person-constituting aggregates, but not for all bundles. In the first place, person-constituting bundles have emergent functionality—can do new things only when they and their activities are taken as wholes. Secondly, and relatedly, this new functionality is not one that answers to our needs, but one to which we are responsive. This emerges from a consideration of the very notion of “development” or growth, which is conceivable precisely because it is perceived to be there—and not because we have a standing interest in the end-product. These joint criteria are satisfied only in the case of living beings. I think we can make the line of thought more precise, and more compelling, by considering just what is special about the acquired characteristics belonging only to whole person-constituting bundles—our example of “new functionality” already gave us a hint of this. We can do this by returning to the towers on the hill. I said above that there are two divergences between the tower-case and the person-case, and looking at the second now might help to clarify what is distinctive of aggregations which are, by their interrelations, persons.

1.4.2.2  Persons as Sources of Unity The second disanalogy between the towers-on-the-hill and the aggregates-and-their-person cuts deeper than the first. The “inference” explanation as described above was crucially incomplete. For without acknowledgement, it relied on the unity of the cognizer to provide coherence to the sense impressions. In a way, this is no more than what the standard Abhidharma account grants—minds supply the perception of unity when confronted with what is in fact a manifold. But what such an account overlooks is that a unified cognizer is thereby smuggled in. If the preferred Abhidharma account of causal succession of moments of consciousness with no real fact about interrelatedness were true, then, the Pudgalavādin claims, “the Buddha could not be omniscient, because a mind with its mental factors is momentary, it cannot know all things. But a person may know all things.”102 But forget knowing all things—even knowing those two towers on the hill are standing at some distance from each other would be impossible. In our earlier story, we generously allowed a mind that “so orders the manifold” as to make most sense, or be most convenient and coherent.103 The principle of convenience is thus allowed its due. But, whether in a single complex case or in the vastly complex case of omniscience, such cognitions require that there be some kind of unity in discerning the manifold



Persons Keeping Their Karma Together

27

together to determine which groupings will be convenient and coherent. This source of unity might be an extra element, as a Brahmanical soul or, more minimally, as a Kantian agent; or it might be that we could give a still more minimal account and simply say that the fact of the making-coherent, the fact of mental organization, is basic. It is a bare, brute—but inescapable—fact of the matter that certain groups of cognitions display a categorically different sort of integrity from the run-of-the-mill causal connectedness between aggregates in general. Complex cognition is possible because the perception-dharmas and consciousness-dharmas here just are related to one another in a distinctive way: in a person-constituting way.104 This is the most minimal explanation possible, and I  suggest is the one the Pudgalavādins are insisting must be accepted as the only way to do justice to the phenomena without adopting a substantialist theory of self.105 Unfortunately, admitting the ultimate reality of complex facts, rather than simple substances—of relations whose character is determined and indeed constituted by their relata—is not expressible within the framework of Buddhist metaphysics—and comes to be positively rejected.106 The second “disanalogy” between our two cases thus shows that what is special about persons is that they are unity-conferring in virtue of being organized in special ways distinguishing them from other sorts of continua. This is the distinctive emergent functionality which cannot be explained by appeal to convenience, because it grounds the very possibility of any such appeal. There are not “stream-individuators” for all complex phenomena because persons are the individuators for all other continua, through the sort of activity that constitutes them as the distinctive sort of unity they are. Moreover, the “new functionality” specific to persons is not just that they grow and reproduce—plants do that;107 it is rather that their growing and their emergent functioning indicate a unity between a variety of different sorts of elements, specifically including perceptual elements and cognitions, as well as desires—and only thereby are they able to be sources of the sorts of cognitions in virtue of which prajñaptisat, conceptual reality, can arise. And only then can karma, and the specific responsibility that comes from realized intentions, arise.

1.5  Karma and Development We saw that the texts suggest that person-constituting bundles are distinct in kind because of (1) their growth or development—a sort of development

28

Amber D. Carpenter

in which the end is not given by any interests or desires of ours; and (2) their emergent functionality. The emergent functionality of particular interest is the ability to conceive of aggregates as unities, the experience of “many” as “one.”108 It is from this that the distinctively person-related functions emerge, most centrally action. For karma is not just any knocking together of discrete substances. Only a certain subset of such interactions qualifies as that peculiar phenomenon “action.” Having its source in the unity-conferring configuration of intentions, perceptions, consciousnesses… that is a person enables action to have moral and not just material characteristics, which therefore attract moral and not just material consequences. We can see this confirmed if we return to the three bases of designation of the person—the aggregates, transmigration, and cessation. In the Saṃmitīyanikāyaśāstra (Sns), considering especially the designation of the person based on transmigration, we see that what most frequently emerges as the point of concern is not the theoretical question about whether the person reborn is the same person as the one whose incarnation immediately preceded,109 but concerns about karma. As to the deeds, one does one’s own deeds. But what does self-done mean? A. It means that one receives (the fruits of one’s own deeds). What does one’s own deed mean? It means (to make) a distinction (between the deeds of oneself and those of another). But why? Because (the results of one’s deed) do not go to another. (Sns 1.165) We saw above the claim in the Saṃmitīyanikāyaśāstra that “there could not be the killing of beings nor would the killer have killed anything… good and bad would yield neither freedom nor bondage. . . . There would be neither the doer nor the deed nor any result thereof.” This claim that denying the person is tantamount to denying the possibility of ethical goodness is prominent also in the Kathāvatthu (KV 1.1.200–216), where the Pudgalavādins are pressed to answer questions such as: “Are ethically good and bad acts (karmas) known to exist? And the doer of them also? And the instigator also? And the enjoyer of the effect—is he also known to exist?”—and to demand answers from their opponents in return.110 In some way, the “person” signifies the connection between this act and that rebirth, or between this tṛṣṇā and this rebirth.111 But why did the Pudgalavādins insist that there really be some “person,” in order individuate tṛṇṣas and correlate them correctly to appropriately individuated rebirths?



Persons Keeping Their Karma Together

29

Were the Pudgalavādins the unwitting victims of Brahmanical thinking, supposing that moral phenomena—unlike all other phenomena—require some substrate to inhere in, if they are to exist at all? I think not. A more compelling explanation of why the Pudgalavādins saw the annihilation of moral phenomena in the elimination of the person is that these moral phenomena require that distinct kinds of elements relate to each other in specific ways not accounted for in the patterns of any one sort of element taken on its own. Within a kind, the laws or patterns of succession are fixed relative to one another: “There is a fixed order about rūpa” (Sns 3.201). But in questions of karma what becomes inescapably an issue is the appropriate fitting together of a variety of kinds of elements, typically the physical and the “mental” types of elements—patterns of succession not determined or exhausted either by the rules of succession of matter, or by the rules of succession of moments of consciousness, or intention, and so on. It is through intention, a mental factor, that a distinctive sort of cause is set into effect; and it is usually through body or form (rūpa) that these effects arise, generally as suffering. Because of belief in “formless” realms, where karma can mark the connection between action and result, this may not necessarily have to do with dualistic concerns about the integration of mind and matter, but rather with the integration of the five main sorts of dharmas more generally. Only this sort of interaction between kinds of elements gives rise to the distinctive order of succession singled out and designated karma. In fact, we might think of karma as the name for the patterns by which different types of dharma, each with its own logic of succession internal to its kind, are organized with respect to one another. So in the Kathāvatthu we see that we cannot say that the maker of karma is distinct from karma made (KV 1.1.214), since the person just is the fact that these aggregates so belong to each other as to constitute morally inflected actions and results. Likewise, the fruit and experiencer thereof cannot be said to be two different things (KV 1.1.203). Only because these aggregates do belong together can this later event be the “fruit” of an earlier event.112

1.6 Conclusion We have highlighted now two ways in which the collection of elements constituting persons differs from those constituting nonpersonal complex

30

Amber D. Carpenter

unities. In the first place, persons develop over time, and do not just change; and they do not develop for the purpose of creating a previously desired and conceived of functionality (as happens when a chariot is built). Rather, it is because this order of changes really belongs together in a distinctive way that we are able to conceive of the organism as growing. In the second place, persons are not plants. It may be that all organisms develop; but that anything can be conceived of as developing, or any complexity conceived as such at all, requires the real unity of complex persons. That is to say, the very possibility of conceptual reality—and also of the categorically complex phenomena constitutive of karma and its links to its appropriate fruit—rests on the real reality of just those interactions of matter, intention, perception, cognition, and consciousness rightly perceived as belonging to each other, and not to other elements. The pudgala is, on this view, the distinctive mutual-relatedness of diverse kinds of elements at a time and over time. It is just that belonging together which constitutes the possibility for moral agency, development, and unifying conceptualized reality. The capacities to serve as a moral agent and to develop are not reducible to the aggregates themselves, for all these presuppose the reality of some but not other elements’ belonging together. None of this, note, implies the existence of an immaterial, substantial, independent substratum for the aggregates. The noncategorial metaphysics to which the Buddhists were committed, however, prevented there being any intelligible way of expressing this position. The pudgala, taken as a unique sort of fact about belonging together, was sui generis—a fifth kind, as they maintained. But it is the nature of this kind of thing not to be separable from its constitutive elements, and to gain its specific character through the history of those varied and changing constitutive elements. It is thus no substance, no “underlying subject” of properties, nor an active agent of unity distinct from the unique unity it is (though of course it is the source of unified conceptions, prajñaptisat). But that these very elements belong exactly to each other, that they really do interact with each other in ways different in kind from the interactions of elements generally—this truth cannot be eliminated, nor be merely conceptual.

Notes 1. See Thích Thiên Châu (1999, 10), according to whom the personalist Vātsiputrīya school was “one of the most powerful and flourishing of early Buddhism.”



Persons Keeping Their Karma Together

31

2. See Ganeri (2007, 162–163); and Priestley (1999, 2), who observes that at this time just one of the pudgalavāda schools, the Sāṃmitīya, “was second only to the Mahāyāna in the number of its adherents.” 3. Majjhima Nikāya 1.137, sutta 22: Anlagaddūpama Sutta. 4. The phrase is from Steven Collins (1982, 77). The denial of self amounted, says Collins, to a “linguistic taboo,” whose function was to distinguish Buddhism from Brahmanical outlooks. 5. Leonard Priestley (1999) brings this out nicely in the first chapter of his excellent monograph on the pudgalavāda. See also Schmithausen (1973, 161–186), and Frauwallner (1953, 217–225), both modern scholars who deny that outright rejection of self was an original part of the Buddha’s teachings, rather than a later accretion or interpretation of those teachings. For a more recent and sophisticated attempt to grasp the ambiguity of no-self in earliest Buddhist, see Hamilton (2000). 6. Cf. Priestley (1999, 81–82):  “In the Vijñānakāya, one of our earliest sources, the Pudgalavādin is represented as saying, ‘There is a self (ātman), a sentient being (sattva), a living being (jīva), a being who is born (jantu), a being who is nourished (poṣa), an individual (puruṣa), a pudgala’ (T 1539, 542c24 f.).” 7. The distinction has authority in the Nikāyas (see, for instance, Aṅguttara Nikāya 1.60: “There are these two who misrepresent the Tathāgata. Which two? The one who represents a Sutta of definitive meaning as a Sutta of non-definitive meaning, and the one who represents a Sutta of non-definitive meaning as a Sutta of definitive meaning”). Unfortunately, there is no authoritative sūtra indicating which sayings fall into which category, so that intra-Buddhist dispute often involved debate over whether a saying agreed to be the Buddha’s is to be taken literally or not. This is particularly so in the dispute over the person, since the Buddha-dharma—spoken in ordinary language often for ordinary listeners—invariably makes reference, implicitly or explicitly, to persons, beings, living things, and so on. 8. Priestley and Châu, who have done some of the most comprehensive recent work on the pudgalavāda, as well as Venkataramanan (1953), explore and expound the pudgalavāda as if it were adhered to only because some happened to think this the best interpretation of the Buddhist sūtras. According to Châu, the Pudgalavādins got this view from a thorough study of the scriptures:  “It is probable that this very specific, difficult to express, position was established after the Pudgalavādins had fully studied the attitude and findings of the Buddha and his disciples on false views concerning the world and self” (Châu 1999, 150). Priestley is even more explicit that regard for right interpretation was the motivating factor in those espousing the pudgalavāda:  “To the Pudgalavādins themselves the whole matter must have appeared otherwise. We can guess from their patient insistence on the actual words of the Buddha, whatever their apparent inconsistencies, that they believed that they

32

Amber D. Carpenter

were simply presenting what the Buddha himself had taught, without presuming interpretation” (Priestley 1999, 118). Since, however, the Pudgalavādins apparently laid claim to a nonstandard canon of sūtras in order to support their view (Vasubandhu, Abhidharmakośabhāṣya, §2.6.1, in Duerlinger 2003, 84), this seems an inadequate motivation for insisting on a position which often attracted virulent hostility from fellow Buddhists. 9. Among recent scholars, I  think Duerlinger (2003) does the most to take the pudgalavāda philosophically seriously. 10. Although later Buddhist philosophers (including Candrakīrti, Bhāviveka, Śāntarakṣita, and Kamalaśīla in his commentary on Śāntarakṣita) continued to engage with the pudgalavāda, my own discussion will remain with the earlier formulations of the view, and responses to it. 11. The Satyasiddhiśāstra is translated into English as Satyasiddhiśāstra of Harivarman (Harivarman 1978, 296). There are several more minor sources, earlier as well as later; see Priestley’s broad and detailed treatment of the pudgalavāda literature remaining to us (Priestley 1999, chap. 2). 12. Priestley offers an account of how the pudgalavāda may have shifted over time (see especially Priestley 1999, 71, 87–88). However, the specific change of position that he recommends—namely, from declaring the person “not a substance” to declaring the person “neither substance nor non-substance,” to finally agreeing that their pudgala is, after all, a substance—is not, as will become clear below, one that I think did happen. 13. These are the Sānmídǐbù lùn (or Saṃmitīyanikāyaśāstra), the Lü èrshíèr míngliǎo lùn (or Vinayadvāviṃśativyakti), the Sānfǎdù lùn, and the Sì ēhánmù chāojiě. These are counted as three by Priestley, the last two evidently being translations of the same text, the Tridharmakhaṇḍaka (or Tridharmakaśāstra, as Châu has it; the Chinese transliterations I am using are those used by Priestley, as are the reconstructed Sanskrit titles). These two translations diverge substantially, however, at certain points, so that one may actually get information in one not found in the other. Indeed, Priestley (1999, 71) argues that “the disagreement between the Sānfǎdủ lủn and the Sỉ ēhánmủ chāojiě in their versions of the Tridharmakhaṇḍaka’s account of the second kind of conception is so extensive that the versions must surely represent different stages not only in the development of the text, but also in the elaboration of the doctrine which it presents.” 14. Châu (1999) has detailed information about the history and content of these treatises in particular. 15. The Sāṃmitīyanikāyaśāstra has been translated into English by K. Venkataramanan (1953). 16. Since Buddhist schools generally agree there to be some conventional or constructed sense in which person-talk and self-talk is not just permissible but



Persons Keeping Their Karma Together

33

correct, I  find Duerlinger’s exposition of the Pudgalavāda simply confusing. See remarks below, especially note 53. 17. Attā satto jīvo:  The Debates Commentary (Kathāvatthuppakaraṇa-Aṭṭhakathā) (Law [1940] 1989), n. 5 to section 1.1.[8]‌. This commentary on the Kathāvatthu is traditionally ascribed to Moggaliputta-Tissa. 18. Contra Duerlinger (2003) most strongly, who describes Vasubandhu as trying to “wipe out” this heretical view. But also Châu (1999) uses the notion of “heretical” as if this were the significance of the view and objections to it. Heresy in any case seems a most inapt notion to bring to the discussion. “Excommunicated” (literally, being thrown out of the community) generally resulted from failure of discipline, not failure of doctrine or “view.” 19. Châu (1999, 138–141) has a list of five non-pudgala self substitutes of various early Buddhist schools; see also Collins (1982, 230–244) and, for withering remarks as well, Conze (1962, 132–133). 20. The Debates Commentary (Kathāvatthuppakaraṇa-Aṭṭhakathā) (Law [1940] 1989), 1.1.[8]‌. 21. See Priestley (1999, 84), for the antiquity of this distinction, particularly relative to the familiar distinction between substantial and conceptual reality, discussed below. 22. Vasubandhu’s account of saṁvṛtisat as opposed to paramārthasat at AKBh 6.4 is well known (that which remains after analysis, even mental analysis, is paramārthasat; what does not resist analysis is saṁvṛtisat); Dharmakīrti (PV 2.3, 3.3) makes the distinction according to whether or not something is capable of causal influence. For Candrakīrti’s definition, see Prasannapadā 24.8; for Śāntideva’s, see Bodhicaryāvatāra 9.2. For general discussion, see Sprung (1973). 23. This would be respected by, for instance, Vasubandhu’s definition (though this may not do to capture the Madhyamaka form of the distinction). “Being useful” plays a significant role in Mark Siderits’s account of what makes a conventional truth well grounded, but the structural point is essentially the same: “standing behind every conventionally true statement is some (much longer) ultimately true statement that explains why accepting the conventionally true statement leads to successful practice” (Siderits 2007, 58; emphasis mine). Compare also, 62: “for every statement that is conventionally true, there is some (much longer) ultimately true statement that explains why it works.” Presumably, ultimate reality does not actually explain why “it works” or is useful, but rather describes how things ultimately are such that, given certain ends, such conceptions are useful. I return to this below. As usual, Mādhyamikas may be outliers here, precisely aiming to reject the purported one-way dependence of conventional on ultimate reality (see Dan Arnold, e.g., Arnold 2012a, for such an interpretation of Madhyamaka).

34

Amber D. Carpenter

24. Horner (1996, 27). The example is drawn originally from SN 1.135, the verses of the bhikkhunī Vajirā to Māra, to which Nāgasena goes on to refer. 25. I  will not try to argue for one interpretation rather another of prajñapti and prajñaptisat. It has to do with conceiving and mental activity; but see important arguments by Ganeri for why we should not take it to mean “nominal,” or assume automatically that “conceptually real” means illusory or fictitious. However, when contrasted with substantial reality, it is inferior, as conventional is inferior to ultimate reality. 26. See Paul Williams (1981, 237) here on the equivalence; see also Priestley (1999, 94), “The ‘true and ultimate’ seems to be what is later defined as the substantial (dravya) in contrast to the conceptual entity (prajñapti).” 27. Though one might wonder whether all conceptual truths are thereby merely conventional, and not ultimate truths. If reality is intelligibly structured, as Plato, Aristotle, or the Nyāyaikas held, then some concepts could get it right and so be ultimately true—not merely convenient conventions. 28. See, e.g., Paul Williams (1981). I shall have more to say about the translation of dravyasat as “substantial reality” below. Briefly, in philosophy, substance does duty as “individuator,” “essence,” “that which underlies,” and “that which persists through change.” The Buddhist “ultimately realities” will not turn out to play this lattermost role—nothing does; but as the first three familiar senses of substance are relevant to that which is dravyasat, I have chosen to stick with this familiar term of art, for this I believe is where the discussion is to be properly located. 29. Paul Williams describes the situation thus: “The secondary existent is dependent upon mental and linguistic construction, synthesis out of more fundamental and indubitable elements. These primary existents, on the other hand, are those elements which make up the synthesised existents and therefore cannot themselves have the sort of dependence which the latter enjoy. . . . A primary existent is primary and thus independent because it is the basis upon which rests secondary existents such as most of the objects of our everyday world” (P. Williams 1981, 240). 30. Collett Cox’s discussion of dravya and its various usages (Cox 2004, esp. 547–549) brings out how the term does seem persistently to pick up on different sorts of things that might well belong to “substance” in its classical philosophical usage: individuation, bearer of properties, being distinct from all others. 31. One is tempted to call such a conception of being “impoverished” but for the tenacity and ingenuity required to maintain it. 32. Whether conceived of as color tropes, pain tropes (cf. Ganeri 2001, chap. 4), or in some other way; the Buddhists call them dharmas (or, in Pāli, dhammas). The Abhidharma canon in fact has an entire treatise on “conditional relations” (paccaya), the exhaustive Paṭṭāna. What becomes clear is that conceiving a paccaya as a dhamma amounts to making relations into discrete individuals, just



Persons Keeping Their Karma Together

35

like every other dhamma. While such a dhamma might “link” two other dhammas, its doing so involves no alteration to its specific character, or to that of the “linked” dhammas—we have simply a series of dhammas. It is no wonder that later Buddhist philosophers asked when such a dhamma was supposed to arise (with the previous or the subsequent dhamma in a series, if they are temporally related); and that much later Prabhācandra, commenting on Dharmakīrti’s On Relations, says: “In the presence of something, something comes into existence and in the absence of that it does not come into existence.  .  . . If this is so, then let those very two aspects (bhāvābhāva) be taken as the cause-and-effect relationship; what is the need of postulating an unreal relation?” (The comment is on kārikā 11–12 of Dharmakīrti’s Sambandha-parīkṣa:  “If [you say that] cause-and-effect relationship is the relation which is of the nature of anvaya-vyatireka [i.e. in the presence of a cause an effect is seen and in the absence of it no effect is seen], then why not the same two [bhāvābhāva, i.e., anvayavyatireka] be taken as the cause-and-effect relationship?”). 33. And according to Paul Williams, for instance, rightly so:  “Indeed it is by no means clear by what criteria we can talk of different sorts of existence . . . but the notion that two entities exist in different ways is itself paradoxical, for existence would not seem to be the sort of thing which could be divided into different types” (P. Williams 1981, 228). Collett Cox (2004, 569) argues that the Sarvāstivādins did indeed recognize “at least four separate abstract typologies of existence … the very presence of these typologies demonstrates that existence was not considered to be uniform; experienced objects do not simply exist or not exist, but rather can exist in different ways.” However, it is not clear to me that the three she identifies actually are different modes or ways of being (in the sense of a categorical schema), rather than derivations from the one primary mode of being. 34. See Duerlinger (2003, 265 ff, esp. 278) for discussion of view that all phenomena are substances. 35. For example, KV 1.1.17: “Is the person known in the sense of a real and ultimate fact, and is material quality also known in the sense of a real and ultimate fact? [P:  Yes.] Is material quality one thing and the person another? [P:  Nay, that cannot truly be said.] Acknowledge the refutation: If the person and material quality be each known in the sense of real and ultimate facts, then indeed, good sir, you should also have admitted that they are distinct things.” In the next sections (§18–73), the same argument is made concerning fifty-five other acknowledged ultimately real entities, such as feeling, perception. 36. The concession is the basis for the current consensus that, in spite of appearances in the Kathāvatthu, the Pudgalavādin is not defending the ultimate reality of the person. Duerlinger, Châu, and Priestley all at some point take it that the position defended is that the person is a conceptual reality in some aberrant sense rejected by other Buddhists, who acknowledge the conceptual reality of

36

Amber D. Carpenter

persons. In taking the Pudgalavādin at his word, and granting that he means the person to be ultimately real—and specifically not just conceptually real—I thus go against the current grain. 37. I  will have more to say about the Pudgalavādin’s use of the fire-fuel analogy below. 38. Prabhācandra, commenting on Dharmakīrti’s Sambandha-parīkṣa, kārikā 3, credits his opponent with thinking that real individuals require relations:  “A positive entity requires that (relation, this is what you say).” 39. This, of course, is not the way Aristotle went. For an attempt to bring Aristotelian and Buddhist metaphysics of substance and persons more directly into constructive contact, see Carpenter (2014). 40. I  suspect that missing this challenge—that is, assuming the Pudgalavādins agree that “substantial” and “conceptual” are exhaustive options—explains why Châu, for instance, credits the Pudgalavādins with the view that the person is simply a conventional or conceptual reality; their denial that the self is a substance is taken immediately for the assertion that it is both conceptual and not ultimate. Priestley also assumes the Pudgalavādins accept the ­exhaustiveness of the substantial-conceptual distinction: “Either their belief was that the pudgala could be indeterminate and yet substantial, or else they held it to be conceptual and yet ‘true and ultimate’ ” (Priestley 1999, 84); but of course there is a third option, namely, that the self is ultimate but not substantial. See also Duerlinger (2003, 132). 41. Vasubandhu has the Pudgalavādins hold that “a person is not substantially real or real by way of a conception” (AKBh §2.1.1, in Duerlinger 2003, 73; AKBh 9; DS ed. 1192.7, “it exists neither as substantial nor as conceptual,” naiva hi dravyato-sti, nāpi prajñaptataḥ). Priestley does not so much ignore this possibility as find it frankly unintelligible: “Certainly the Vātsīputrīya’s contention that the pudgala is neither conceptual nor substantial seems absurd” (Priestley 1999, 99); like Williams, he is persuaded of the fundamentals of Abhidharma ontology. 42. KV 1.1.1. 43. KV 1.1.171–192. 44. Duerlinger (2003, 61n29, 66n66) strongly favors “inexplicable” over “ineffable,” rejecting the common view that the Pudgalavādins are claiming that the person cannot be spoken of or conceived. But of course one can say a lot, especially negatively, about something, without having the vocabulary or conceptual means of articulating what something is. I think it is this latter inability which leads the Pudgalavādins to say of the person that it is avaktavya. Duerlinger’s objection to “ineffable”—which, incidentally, shares its construction with avaktavya—stems from his insistence that the best formulation of the pudgalavāda is in terms of the person as a conceptual reality, which its “inconceivability” would seem to preclude.



Persons Keeping Their Karma Together

37

45. Cf. Vasubandhu (1984, 74), §5. 46. For ongoing disputes between Buddhists and Naiyāyikas over whether any nonsubstantialist (Buddhist) account could be adequate to this task, see Arindam Chakrabarti (1992, 103–117) and Ganeri (2000, 639–657; 2007, chap. 7). More contemporary still, and involving phenomenologists in the dispute, are the excellent essays collected in Siderits, Thompson, and Zahavi (2011). 47. Priestley also finds concerns with agency and subjectivity; but these may well be closely related to the more prominent interest in unity. 48. Conze (1962, 126–128); Châu (1999, 161). 49. See especially the Saṃmitītyanikāyaśātra for the three bases of the pudgala. The Tridharmakhaṇḍaka does not use the word “transition” to describe the second basis, but both versions we have do point to unity over time, whether indicating just the past (Sānfǎdù lùn) or bringing in both past and future (Sì ēhánmù chāojiě). 50. Or, in the case of cessation, that there has been a person and whether that person exists or not is now indeterminate. 51. In this early period Nāgārjuna and his followers were often taken by Buddhists to have fallen into the nihilistic extreme—and Vasubandhu, for instance, does not waste many words saying so (AKBh §3.10, Duerlinger 2003, 236–237). 52. See Harivarman as a representative spokesman of this view: “To say that the soul exists in the ultimate sense is satkāyadṛṣṭi, and to say that the soul does not exist in the empirical sense is the wrong view, mithyā-dṛṣṭi. But it is right view (samyak-dṛṣṭi) when we say that the soul exists empirically and does not however exist ultimately. . . . Thus is to be understood the statement that soul exists and that it does not exist” (as freely translated by N. Aiyaswami Sastri; Harivarman 1978, 2.295). 53. This is the point I  think Duerlinger badly confuses in his account of Vasubandhu and his debate with the Pudgalavādins. To agree that something is ultimately real, when one conceives of person or any other unity, is not to concede that the person is ultimately real. (See, e.g., Duerlinger 2003, 11–12: “these conventional realities ultimately exist” is his view of the pudgalavāda thesis in the Kathāvatthu, indeed of the Theravāda view. Similar remarks are found on pp.  10–12, 20, and elsewhere.) See Harivarman’s reply to the naïf who supposes “There may not be any harm in calling the five aggregates of elements as the soul”: on the contrary, “The advocate of the soul thinks that the soul is one. Then the five aggregates would be one. This is faulty” (Satyasiddhiśāstra 292–92); cf. Harivarman (1978, 73):  “The aggregates are other than pudgala because the former is five whereas the latter is one. If the soul exists at all it ought to be other than the aggregates.” 54. It is curious that Paul Williams readily endorses the noncategorial metaphysics of the Buddhists (in P. Williams 1981, 228; note 33 above), but in section 4 of “The Absence of Self and the Removal of Pain” (in P. Williams 1998, 115–119),

38

Amber D. Carpenter

he strongly resists the mode of argument which goes on to insist that if the whole is real, it must be some distinct individual in addition to the parts. Of course, as he says, this is not what we mean when we say the whole is “more than” the sum of its parts; and this is correct. But “we” probably do not ordinarily work within a default metaphysics that is so minimal as the Buddhist one (if, indeed, we are able to make good sense of what we do mean at all). As he recognizes in section 6, “For there to be a whole other elements from different categories need to be involved” (P. Williams 1998, 129). 55. Uddyotakara, for instance, commenting on Nyāyavārttika ad Nyāyasūtra 1.1.10. This can be found translated by Matthew Kapstein as an appendix to Kapstein (2001), where one can also find a translation of Vasubandhu’s AKBh 9 rather more readable than Duerlinger’s. 56. AKBh 9, §4.1.1 (Duerlinger 2003, 97). It is, of course, an anachronism to suppose Vasubandhu responds to Uddyotakara; but the Nyāyasūtra itself already contains the seeds of Uddyotakara’s later objection. 57. “Causal” connections are not robust, on the Buddhist account. 58. Duerlinger rightly notes in his commentary (2003, 242) that it will not: “Minds in different causal continua of aggregates, [Vasubandhu] assumes, cannot be appropriately causally related, but he does not explain what constitutes aggregates being appropriately causally related or what constitutes aggregates belonging to different causal continua of aggregates. He cannot distinguish [as the Nyāya-Vaiśeṣikas do] … Unfortunately, this view does not explain how different causal continua of aggregates are to be distinguished from one another” (emphasis mine). 59. This resembles the “sophisticated” version of Siderits’s second “circularity” objection (Siderits 2003, 45). 60. A  criterion emphasized by Mark Siderits’s account (Siderits 2007, chaps. 3 and 4). 61. I  think this gets at Paul Williams’s objection, posed especially in section 7 of “The Absence of Self and the Removal of Pain”:  “It is difficult to see how the continuant-collective model is going to be able to explain that unity in a manner which will not beg the question” (Williams 1998, 130). Essentially the same objection is made to Parfit’s attempt to do without any mechanism for stream-individuation (see discussion in Ganeri 2007, 187). Siderits offers difference in quantity in place of difference in quality, and proposes to meet the objection in terms of “maximal causal connectedness,” and an appeal to consequentialism (Siderits 2003, 45–50). One peculiarity of such a defense is that, to the extent that it succeeds, it undermines any motive we might have had to abandon notions of the self, or to suppose that it was holding these very notions which led inevitably to suffering. 62. Compare Parfit (1986, e.g., 232–233; 285–286):  when you know all the facts of the sequence of the aggregates there is not some further fact that you still do not know. Whether in addition one wants to designate a situation as “this (same) person” is a matter of decision.



Persons Keeping Their Karma Together

39

63. Siderits (2003, chap. 3) suggests addressing this part of the objection by appeal to consequentialist principles, arguing that at least one end can be impersonally conceived. It seems the point is that there is some impersonal end, reduction of suffering, best attained by conceiving of ourselves as persons with local aims and projects. I am not convinced this solution works—see note 64. 64. Siderits seems to acknowledge this difficulty (Siderits 2007, 83). The end that he proposes, with respect to which conceiving of aggregates as persons is convenient, is “reducing suffering”; but it is precisely for this quite general end that it is uncertain whether thinking in terms of persons is a help or a positive hindrance. 65. Compare Siderits (2007, 75): “As Nāgasena said, the conventional truth is that we are persons. This is conventionally true because it is ultimately true that these present skandhas are the cause of the future skandhas in this series.” 66. “Might this not explain why we conveniently designate the one series as ‘me’ and the other as ‘you’?” Siderits asks (2007, 83). “Might this not be the ultimate truth that makes it conventionally true that we are distinct persons?” 67. Note the regularity with which reductionist responses to concerns about continuity, the sameness of the person over time, presume that we have already individuated a distinct causal stream. 68. Parfit’s reductionist reply to essentially the same objection requires appeal to “in the particular life that contains [x]‌…”; Parfit glosses this by referring “this” life to the one that “is now directly causally dependent on body A,” grounding the individuation of persons on the individuation of lives, and grounding this in turn on the individuation of bodies (Parfit 1986, 226). But of course, with their belief in reincarnation—that “the same” causal stream extends through several different bodies and types of body—such a strategy, even if it works in principle, is not available to the Buddhist. 69. Because this introduces no “separately existing entities” (Parfit 1986, 275), and yet also insists on the reality of this fact of belonging together—which might consist in the aggregates being thus and so disposed (if you had to point to where it happens), but which goes beyond the mere recognition of the existence of each individual element—I am not certain whether this view should be considered reductionist on Parfit’s terms, or as a variant of the “further fact view” (Parfit 1986, 210). I  am tempted to think that the reductionist/nonreductionist positions as Parfit describes them—with the “further fact view” falling among the nonreductionists (e.g., 1986, 279)—are not the exclusive and exhaustive options he supposes (Parfit 1986, 216, 273: “a fundamental choice between two views”). At any rate, for what it is worth, if reductionists, then the Pudgalavādins are of the sort that accept Parfit’s theses (3) and (5), but not (4), the person is its constituents, or (9) a complete description of reality could be impersonal (Parfit 1986, 211–212). 70. So we should be cautious about describing the Pudgalavādin’s view as that the “pudgala is something more than the reunion of its constituents,” as, for

40

Amber D. Carpenter

instance, Châu does (1999, 161). In his concern to insist on “the irreducible difference between one stream of consciousness and another,” without positing a substantial self, Dan Zahavi (2011, 67–68; see also 69, 74) comes across strongly as a modern-day Pudgalavādin. 71. Thus I think Duerlinger is correct in describing the pudgalavāda as asserting that “persons” do not have identity of their own. His account, however, leaves it unclear why they would assert such a thing, or what they might mean by it. 72. “We, however, plead that the five aggregates put together form the soul” (Harivarman 1978, 70; emphasis mine). Against this we must set a remark from Sns: “Now we say, where (the Buddha) has spoken (of it to) the Srāvakas he has done so (only) as based on the elements, that is all. There is no other self. . . . Now we say, the Buddha said that the person exists as conditionally cognized. Therefore that which is (here referred to) is only the (objective) counterpart of this (cognition) and not any real self” (Venkataramanan 1953, 177). Since the Saṃmītīyas have just asserted that “that the self exists etc. can be said” (175), we might suppose that these arguments (Venkataramanan 1953, 177)  are aimed at those who mean by asserting the real existence of the self to assert the separate existence of a distinct individual, something the Saṃmītīyas, and all Pudgalavādins, would reject. 73. See especially KV 1.1.130–145, where the pudgala cannot be said to be different from any of the skandhas, the āyatanas, the dhāthus, the īndriyas, or any of the other dharmas; and yet neither is it simply identical with any or all of them together (§138–141). Compare also what we see in the pudgalavāda texts, that it is impossible to say whether person is same or different (Ssu 4c24–25), separate or not separate (Tds 19a26) from dharmas existing at the three times (see Châu 1999, 91–92; the texts cited are the two translations of the Tridharmakhaṇḍaka, the Sānfǎdù lùn and the Sì ēhánmù chāojiě, respectively. See also Châu 1999, 156–157). 74. See Graham Priest’s (2012) argument that this will be a feature of whatever it is in virtue of which a manifold constitute a genuine unity. 75. See Duerlinger’s concern (2003, 64n48), that concentrating exclusively on this one analogy for how to think of the relation of pudgala and aggregates distorts our interpretation. His “Vasubandhu on the Vātsīputrīyas’ Fire-Fuel Analogy” (Duerlinger 1982, 151–158) has detailed discussion of the Pudgalavādin use of the analogy, as that is found in the Abhidharmakośabhāṣya. 76. See AKBh 9, §2.1.3 (Duerlinger 2003, 74). 77. Mahātaṇhāsankhaya Sutta, in the Majjhima Nikāya (MN) 1.260, translated by Bhikkhu Ñāṇamoli and Bhikkhu Bodhi; see also, for instance MN 2.182, sutta 96: Esukāri Sutta, where the analogy is specifically used to model the relation of rebirth to circumstances.



Persons Keeping Their Karma Together

41

78. I do not want to pretend that this exhausts what might be got from reflection on the fire-fuel analogy, which does indeed warrant reflection. Among others, it is worth attending to another use of the fire-fuel analogy to person-aggregates in the context of the question “Where does the liberated person go upon ­entering final nirvāṇa?” as found, for example in the Aggivacchagotta Sutta (MN 1.486–488, sūtra 72). And there is a rather different, also well-known use of the analogy to illustrate someone who has “burned up” sensual attachments (see, e.g., Majjhima Nikāya 2.203–204 sutta 99: Subha Sutta; “like a fire that burns independent of fuel such as grass and wood”). 79. The person designated by way of the aggregates: “a person is not substantially real or real by way of a conception, since he is conceived in reliance upon aggregates which pertain to himself, are acquired, and exist in the present” (ABK 9, §2.1.1; Duerlinger 2003, 73). 80. The person designated by way of transmigration, concerned primarily with karma, with the appropriateness of morally inflected causes and consequences. This focus on karma may be significant, and will be discussed below. 81. Though perhaps not in the way Châu intended it:  “The specific relationship between the pudgala and the supports is explained by the continuity of a single individual independent of others. There is continuity (santāna), there is therefore a possessor of continuity (santānin). According to the Pudgalavādins, to deny the possessor of continuity is to deny continuity” (Châu 1999, 161). 82. The experiencer or subject of experiences and the agent of action are two traditional roles of “self” that the Pudgalavādins do not seem especially concerned to revive. Note, for example, that they deny that the pudgala is the experiencer of pleasure, according to KV 1.1.154; and further, the fruit and the experiencer of the fruit cannot be said to be two different things, at KV 1.1.203. 83. Châu (1999, 135) describes the person as individuator, or devised for “enhancing the principle of individuation.” 84. E. M. Forster writes, in Howards End (1910, chap. 40), “People have their own deaths as well as their own lives, and even if there is nothing beyond death, we shall differ in our nothingness.” 85. Situating these views within Parfit’s discussion, we might think of Pudgalavādins as reductionists who insist that elements require “the right sort” of connectedness in order to constitute persons, while the non-Pudgalavādins must insist that any sort of connectedness will do—since, after all, there is only the one sort, viz. proximity-succession, undifferentiated. 86. As Châu (1999, 161) writes. 87. It simply happens that some subdivisions of interconnected dharmas are usefully grouped together, given our purposes. Siderits’s account of the Abhidharma position actually acknowledges this need, in a footnote: “Where there are the right kinds of causal connections, it is conventionally true that punishment is

42

Amber D. Carpenter

deserved even when ultimately distinct skandhas are involved” (Siderits 2007, 67n16). Defending the ultraminimalist account, Siderits would no doubt explain “right kind” by reference to efficacy. But since we seek here “whatever is the responsible agent,” the only thing that could make these connections “the right ones” is that those previous skandhas really are responsible, in some way the others are not. 88. AKBh 9, §2.5 (Duerlinger 2003, 77–78); the quotations in the remainder of the paragraph are also from Duerlinger. 89. We might think that this would not be true of the consciousness “sense-modality,” and that it—like Plato’s psyche in the Theaetetus—could be called in to access equally any of the modes of sensory input. Consciousness, however, on the Abhidharma account is not that sort of thing, for it too is just a skandha: a heap of distinct kinds of consciousnesses, eye-consciousness, ear-consciousness, etc. Vasubandhu clarifies that the “mind-organ” does not perceive objects, for it is incapable of grasping anything as an object. It is “mental-consciousness” that grasps mental objects—but presumably, if the Pudgalavādins adopt this explanation of (now merely apparent) cross-modal perception of the person, then they have granted the whole point, namely, that the person as such is prajñaptisat, conceptual. 90. This is the standard Buddhist account, but there were dissenters. The Kathāvatthu records briefly a debate between the Theravādins and Buddhists who argued for the ultimate reality and indeed the perceptibility of space (KV 6.6–7). And according to Vasubandhu, the space between objects is an unconditioned but perceptible dharma (AKBh 1.5, 1.28). 91. According to Buddhaghosa, “In ‘the interval between two trees,’ here, the shape of the trees is seen with the eye, but as to the interval, there being no shape to it, it is space’; that which appears is an act of ideation, not of sense-cognition” (Pañcappakaranatthakatha 6.7). Cox 2004, 557–558, has a useful discussion of the treatment of the reality of space in the Mahāvibāṣā, according to which space is acausal but can be known via prasaṅga-type arguments, and transcendental-type arguments. 92. Or, according to Vasubandhu, it is light and darkness that is perceived, “a certain type of colour” (AKBh 1.28). 93. Priestley’s first (and ultimately rejected) interpretation, and those of many other commentators who focus on the “wholeness” and “unity” factors of the pudgala, do not present views that enable us to make a principled distinction between the wholeness or unity of the pudgala, and that of other nonpersonal complex wholes, such as mountains, plants, and chariots (artifacts). See, for instance, the following description of the view ascribed to the Pudgalavādins (Priestley 1999, 101): “They maintain that it is ‘true and ultimate’ in the special sense that its existence and functions are not reducible to those of its constituents; its relation to the five aggregates is indeterminate because, although it is not reducible to them, neither is it independent of them.”



Persons Keeping Their Karma Together

43

94. Châu (1999, 159). The passage comes from the Mahāprajñāpāramitāśāstra, vol. 1, p. 43. The passage goes on to say that, according to the Vātsīputryas, “the pudgala is a fifth category, an ineffable (avaktavya) dharma, contained in the basket of texts (piṭaka)”; presumably, the eye does not also belong to this fifth category, so the analogy must be only a suggestion of a direction of thought, rather than a very close similarity. 95. Sāṃmitīyanikāyaśāstra, Venkataramanan (1953, 177), with slight modifications. 96. Harivarman’s Satyasiddhiśāstra, as translated by Priestley (1999, 96). 97. I think in general not enough attention has been paid to the fact that it is only the pudgala that is supposed to be ultimately real—not just any functional unity; our explanations of why the Pudgalavādins held the view must recognize some difference between these sorts of unities, otherwise their arguments for persons turn out to be arguments for holism in general, which is certainly not a position they would endorse. 98. The Vijñānakāya contains a longer treatment of such “developments,” in the fourth section of the second chapter, as translated by Watanabe (1983, 187–189). 99. Châu (1999, 146) reminds us that “man is something different from a chariot.” 100. Reinforcing the impression that a special kind of continuity is at issue, is KV 1.1.183–188, where each moment of consciousness, of seeing, and so on has person, but not each new moment is a new person. Milindapañha 40 attempts to account for such notions by way of dependence; but since any given state of affairs arises in dependence on a wide variety of conditions, this cannot pick out the developmental trajectory. There must be some implicit recognition of some “special kind” of dependence relating posterior to prior states. 101. What Aristotle recognized must be an “internal principle of change,” only more minimally construed. 102. AKBh 9, §3.1; Duerlinger (2003, 86). 103. Śrīdhara later makes this, by his time rather familiar, point. 104. The alternative of course is to drop the inference-account of complex cognition in all cases in favor of the perception account—but this would only have the unfortunate consequence of making all complex wholes (and even complex states of affairs) really, ultimately there to be perceived and not merely imputed by the cognizer. 105. It might be worth asking why the Kantian view would not be accepted. Was it not conceived of? Or is it rather that it shows up in the current metaphysical framework as simply a variant of the substantial self? 106. See Dharmakīrti’s Sambandha-parīkṣa, where the primary objections rely implicitly on the presumption that anything really existing must be a substance. 107. If plants don’t pose a problem (and there is no evidence that they did) then this means it is not organic unity that is at issue, any more than structure or function:  it is karmic (i.e., specifically mind-body, thought-action) unity. See Lambert Schmithausen’s discussion of the variety of Buddhist attitudes toward

44

Amber D. Carpenter

the question of plant life, in Schmidthausen (1994) and at greater length Schmidthausen (1991). Ellison Banks Findlay (2002, 252–263) argues that in early Buddhism, “plants have one sense, that of ‘touch,’ and that plants comprise the ‘stable, or non-moving’ category of living being.” If correct this would support the general idea that the special connection is between different kinds of dharma such that specifically karmic effects may arise. 108. This might be an alternative explanation of the close connection Priestley explores between conceptual reality and the pudgala. 109. The Kathāvatthu focuses a lot on this, to the detriment of the Pudgalavādin. 110. KV 1.1.212, as translated by Shwe Zan Aung and Mrs. Rhys Davids (Kathāvatthu 1915). 111. See the concern for linking one rebirth to another in the Kathāvatthu, 1.1.158–170, particularly §166–169 where it is acknowledged that the answer cannot be “sameness of any of the aggregates.” For discussion of the need for some such link, see also Châu (1999, 166); and Châu (1999, 174) for the discussion of the problem of karma. Duerlinger also links concern with agency and morally apt subjectivity to the Pudgalavādins’ use of the fire-fuel analogy (Duerlinger 1982, 152–153). 112. The pudgalavāda does not, I  think, go so far as what Jaegwon Kim (2006, 547–559) characterizes as “classical metaphysical emergentism,” even though some of the language and areas of concern are similar. As Kim characterizes it, “the demands of emergentism make the supervenience relation involved in emergence necessarily unexplainable; we cannot know what kind of dependence grounds and explains the supervenience relation involved in emergence… the supervenience condition on emergence simply amounts to the assertion that there is an in-principle unexplainable covariation between the putatively emergent properties and their base properties” (Kim 2006, 556). But in spite of the reminiscent “inexplicable,” the Pudgalavādins were not in the business of even trying to explain “emergent” (presumably “conventional”) properties in terms of “base” (presumably “ultimate”) properties. The fact of action and development requires that this “belonging” be ultimately real—that is all. What is “inexplicable”—or unsayable—for the Pudgalavādin is the “belonging to each other” of the person-constituting aggregates itself—not any relation they might have to the conventional realities we speak of in everyday life.

2

On Minds, Dharmakīrti and Madhyamaka Tom J. F. Tillemans

2.1  The Problem Do Buddhists have a view of mind they can plausibly defend in current philosophical debates often dominated by forms of materialism?1 In recent years the 14th Dalai Lama and a number of scientists have been meeting regularly in Dharamsala, India, to discuss and compare Buddhist and modern psychological or cognitive science, approaches to the nature of mind. And no doubt the most difficult problem they have taken up—one which potentially significantly divides Buddhists from many major contemporary Analytic philosophers and cognitive scientists—is the question of physicalism, that is, the view that all that there is physical in nature or can be thoroughly explained in terms of the physical sciences.2 The rather extreme version of physicalism that I  will take up here is that of the Canadian philosophers Paul and Patricia Churchland. They do not try to show that mind and brain are somehow identical entities, but argue instead that mind and the mental are just pseudoentities accepted in “folk theories,” pseudoentities that can and (at least in the Churchlands’ estimation of things) probably will end up being eliminated by better science. This stark version of the position that all is only material has a certain type of rigor appealing to the tough-minded who advocate facing facts without old notions that obfuscate. It is in one way or another taken seriously by many who would invoke science as the best or only source of knowledge. It merits attention in an East-West discussion, even if there are arguably

46

Tom J. F. Tillemans

other more sophisticated physicalisms on the market. Last but not least, it is the version of physicalism that figured in a round of the Dharamsala discussions. As Patricia Churchland participated in those discussions, what I have to say can be seen as a continuation of a contemporary comparative philosophy dialogue that has already begun and in which positions have already been staked out.3 It’s time to reexamine how the Buddhists could best proceed in that debate. They may well have been betting on the wrong arguments and have more promising ways to defend mind in their philosophy.

2.2 Dharmakīrti Let us first briefly go back to the Indian canonical sources to be clear on the main historical antecedents for a contemporary East-West debate on physicalism. For the Buddhist side, the physicalist school against which they (and other Indians) argued vociferously was the Cārvāka, a school that no doubt was perceived as threatening, although we have no clear image of its actual situation in India, or its institutions and the number of its adherents. Indeed we have almost nothing remaining of the Cārvāka’s own writings, apart from a few fragments from its founder, Bṛhaspati (second century bce?), and a much later text, the Tattvopaplāvasiṃha of Jayarāśi (eighth century ce).4 Nonetheless we do have in-depth refutations of their positions by their Buddhist and non-Buddhist opponents, and it is these polemical treatments which enable us to form something of an image of the broad outlines of the Cārvāka physicalist position. It held that only the material, specifically the four material elements, existed and that the mental was a type of insubstantial phenomenon, a kind of epiphenomenon, that reduced to (or perhaps was to be eliminated in favor of) the material in the same way as facts about the intoxicating potential of beer were nothing other than facts about the physical composition of beer; an account of beer’s make-up in terms of its material elements adequately explains its intoxicating powers. The underlying motivation: to deny reincarnation and hence the law of karma (retribution), for Cārvākas maintained that because physicalism is so, life and consciousness must end with the destruction of one’s body. Now, historically, the usual Buddhist way to refute physicalism was that of the seventh-century ce philosopher Dharmakīrti and his followers. Indeed, the defense of mind fell almost exclusively to the so-called pramāṇavāda school of Buddhist logic and epistemology, a school which



On Minds, Dharmakīrti and Madhyamaka

47

advocated a type of Buddhist metaphysical realism, albeit one with a subtler ontology than most.5 Pramāṇavādins, as well as the other schools, which largely borrowed the antiphysicalist ideas of these logicians or epistemologists, put forth a series of arguments of uneven quality essentially based on the second chapter of Dharmakīrti’s major work, the Pramāṇavārttika (PV). The central point was that the four elements (earth, air, fire, and water) and their combinations and configurations could not be the primary cause of the mental, in that there are mental properties, such as ways of thinking, character traits, and so on, that regularly vary even though the material elements remain the same.6 For example, Pramāṇavādins sometimes argued that newly dead corpses have all the same matter as living bodies, but obviously don’t think, and so on (see Pramāṇavārttika 2.51) Identical twins can have the same matter making up their bodies, because they come from the same parental sperm and blood, but they have different character traits. We can reconstruct and abridge Dharmakīrti and his followers’ basic argumentation in three steps, all controversial: (a) Minds and physical things, such as brains, are distinct kinds of real stuff or substances (dravya) because of having clearly different properties. (b) Like must cause like; cause and effect are between fundamentally similar kinds of things.7 Thus: (c) Because mind and the physical are distinct kinds of stuff, and only like can cause like, the brain and other physical things cannot be the cause of the mental—they can at most be secondary influences, auxiliary causes (pratyaya). A previous mind is the substantial substratum (upādāna) of the present mind, much like previously existing clay is the substantial substratum for the presently existing pot. This three-step argument is not just supposed to convince the adversary that there are in fact minds: it is above all to be a proof of reincarnation. Indeed Pramāṇavārttika arguments become the basis for a genre of Buddhist scholastic literature known as paralokasiddhi “proof of other lives,” or more literally, “proof of other worlds,” seeking to establish that for each sentient being there is a continuum of mental states that spans

48

Tom J. F. Tillemans

past lives and will (with some rare exceptions) also extend to future lives. To put things in the formal terms adopted by the later pramāṇavāda scholastic:  because mind and brain are distinct kinds of stuff and only like causes like, the first moment of a newborn person’s mind would have to be caused by a mind, and that mind would have to be in her immediately preceding previous life.8 The fact that it is reincarnation that is argued for, and not just the distinct existence of mind, brings supplementary difficulties. In spite of the enormous respect that I, like many other orientalists and philosophers, have for Dharmakīrti, I think someone has to say honestly and clearly that steps (b) and (c) in the above reasoning—and indeed all his supporting arguments for the principle that it is mental states that primarily cause other mental states—are very difficult to defend nowadays. For example, the series of involved arguments concerning the complete physical sameness of a live and dead brain do not make much sense in the twenty-first century, when we have sophisticated detection of neural activity. The account of what happens to the brain and body at death has thus evolved, so that it would surely be unacceptable now to say that brains before and after death are physically exactly the same. Moreover, the “like causes like” principle in (b) may well not prove what the Buddhist wants it to prove. If it is somehow used to show that only mind is sufficiently like mind to be its main cause, an adversary might reverse the argument and say that if that is what follows, so much the worse for “like causes like” and its seemingly intuitive appeal. Worse yet for the Buddhist, many physicalists would accept the “like causes like” principle and argue that it is precisely the reason why mind must either be identical to, or must be eliminated in favor of, the brain. The Dharmakīrtian argument could thus be turned on its head as follows:  since the brain does very clearly influence our so-called mental states, those states must actually be physical in nature, like the brain—anything else would mean that we would have to accept a strange causality between very different sorts of stuff, so different that causality would become unintelligible. This is, pertinently, pretty much the exact argument that a physicalist like Daniel Dennett makes, calling it the “standard objection to dualism.”9 Many of Dharmakīrti’s arguments in the philosophy of mind may have made good sense, and indeed elicited rationally founded conviction, in a time when neurology and genetics were nonexistent and the physical theory of the time was rudimentary. But they are unconvincing now. It should not be surprising that his metaphysics, logic, and philosophy of



On Minds, Dharmakīrti and Madhyamaka

49

language fare much better than his antiphysicalist arguments. Logic and metaphysics depend very little on the science of the day; Dharmakīrti’s antiphysicalist arguments do depend on the science of his day, and that is a major reason why many of them look dated and implausible. Naturally, not everyone would agree with that assessment. Let me, before going on, take up two alternative accounts of what Dharmakīrti’s philosophy of mind, as represented in (b) and (c), is all about. First, Alan Wallace also seems to acknowledge that Dharmakīrti’s arguments on mind, mental causation, and rebirth will not convince many. But, for him, this is because the nature of the mind is supposedly a fact directly accessible only to advanced contemplatives, while to others it is a “hidden” and inaccessible phenomenon (parokṣa). Those without considerable proficiency in “contemplative science” therefore cannot validly deliberate about what mind is. In a book with Brian Hodel, he writes: But the origins, nature and role of consciousness in nature presently remain hidden from the investigations of scientists and philosophers. And without comprehending the nature of the mind and consciousness, one cannot make valid inferences on the basis of Dharmakīrti’s philosophy. … Dharmakīrti’s reasoning here requires not the unquestioning assumption of his assertions about the nature of consciousness, but valid knowledge that his hypotheses are true. However, unless one is an advanced contemplative who has experientially put those hypotheses to the test, such knowledge may remain inaccessible. (Wallace 2007, 86–87) I suspect that Wallace is staking out his own position about contemplative knowledge and who can validly debate what. But, in any event, Dharmakīrti didn’t place those preconditions on a debate about the existence of mind as distinct from matter. Indeed, there is, as far as I  can see in reading his works, no suggestion that he regarded everyone but advanced meditators as somehow incompetent to deliberate about the basic facts concerning mind.10 If he had, his own anti-Cārvāka arguments would have become superfluous debates with an unqualified adversary. Note that Dharmakīrti did place entry requirements on certain debates about ethical matters—such as that acceptance of Buddhist scripture is necessary to deliberate about the details of the law of karma11—but he didn’t do that in his defense of mind. He and his school do indeed hold that mind is parokṣa (literally, “outside [the range of ] the senses”), but that is only to say

50

Tom J. F. Tillemans

that it is something that needs to be understood by inferences (anumāna), not that it is unknowable by ordinary thinkers. Moreover, his general positions on meditative understanding and yogic perception (yogipratyakṣa) do not price philosophical debate on important subjects, like mind, the four noble truths, and others, outside the competence of nonmeditators, or even non-Buddhists.12 His arguments for mind, dualism, and reincarnation stand or fall on their own, without the entry ticket of proficiency in “contemplative science.” Second, John Taber has reexamined many of those anti-Cārvāka arguments in Dharmakīrti, and while he too concludes that they would not provide convincing arguments to establish dualism or rebirth nowadays, he suggests that such is not actually how they need to be taken. They are arguments showing that mind primarily causing mind and the resultant corollaries about reincarnation that supposedly follow from this are not impossible states of affairs. In short, instead of a proof that mind is the primary cause of mind, and so forth, we supposedly have the much weaker demonstration that it is not impossible that it is. Taber writes: In the end, I  do not really think that Dharmakīrti offers us anything we can use to decide the debate about consciousness. … Dharmakīrti has only shown that the materialist has not established that it is impossible that consciousness could exist independently of the body. … However, we shall see presently that this is really all he needs for the purposes of the main argument of the Pramāṇasiddhi [i.e., the argument in the second chapter of Pramāṇavārttika]. (Taber 2003, 495–496) And about reincarnation, Taber says: [A]‌ll Dharmakīrti has to prove in regard to transmigration in order to establish his main thesis that the Buddha is “one who has become a means of knowledge” is that it is possible. (Taber 2003, 497) Now, Taber’s argument is tied in with an elaborate exegesis of the second chapter of Pramāṇavārttika, one that largely follows Eli Franco, and I obviously cannot go into that here. One would, however, hardly need forty verses of elaborate arguments to show that such minds and their reincarnations are “logically possible.” Most of us, even the Churchlands, I  would imagine, might well agree to that right off the bat, even if we



On Minds, Dharmakīrti and Madhyamaka

51

were convinced physicalists about the actual world. There are certainly weirder things in possible worlds than minds and reincarnation. The upshot is that this reading in terms of possibility appears uncharitable, as it would leave Dharmakīrti arguing for what seems to be relatively obvious. Proving mere possibility, understood stricto sensu, that is, in the sense of modal logic, was hardly a major preoccupation of Dharmakīrti and his successors. Indeed, I don’t know of any other major arguments in pramāṇavāda philosophy that were clearly designed to show only that something could be the case in some possible world, in other words, that it is simply logically possible in the sense of modal logic. In Pramāṇavārttika 1’s discussions about establishing entailment (vyāpti) by not seeing counterexamples, Dharmakīrti repeatedly objects that though we might not see them, “there could be/are (sambhava, srid pa) such counterexamples”. However, here it is clear that he did not mean logically possible states that were known to be nonactual, like skyflowers and turtle hairs. Rather he was just talking about things in the actual world that people hadn’t yet discovered, such as (to take his example in 1.13) when one reasons wrongly that all the rice in a pot has been cooked because one has seen that quite a lot is, but hasn’t yet happened to see the rice that isn’t. The point of his arguments in Pramāṇavārttika 1.13–15 against Īśvarasena’s idea of establishing entailment by merely not seeing (adarśanamātra) counterexamples is predominantly epistemological, namely, that we are not sure that there aren’t such things after all in the actual world, for adarśanamātra methods of investigation are epistemically flawed—they are incomplete and undefinitive.13 By far the most straightforward and natural interpretation of what Dharmakīrti intended is that he saw himself as providing the best account of mind and reincarnation, indeed the true account, and not merely one, amongst others, that was logically possible and could be true in some or another consistent world. Perhaps Taber did not use “possibility” and “impossibility” in that strict modal sense, but meant something looser. Is what we can glean from Taber’s discussion just that he thinks Dharmakīrti’s arguments should be at the table in the modern debate as a philosophically plausible alternative to physicalism? That would indeed be debatable, and it is in fact what I think Buddhists should be discussing, although I would maintain that Dharmakīrti will fare less well than many seem to think he will. In any case, this is obviously a different matter than the question of possibility.

52

Tom J. F. Tillemans

There may well be other defenses and reinterpretations of Dharmakīrti’s positions, especially (b) and (c), but we’ll have to stop here. Let’s close this aside and go back to the starting point of the three-point Dharmakīrtian reasoning. Even if mind primarily causing mind, and thus steps (b) and (c), are problematic, it has to be said that we do find in the first step a general Buddhist strategy that is recognizable in many defenses of mind-body dualism:  mind just doesn’t appear to be the kind of thing that is anything like brain processes; they seem to have very different properties. Indeed, minds, for Buddhists like Dharmakīrti and many others, are inherently luminous. For example, Pramāṇavārttika 2.208 famously states prabhāsvaram idaṃ cittaṃ prakṛtyā (“Luminous is this mind by its nature”); brains and other material things are not “luminous.” Minds are unobstructed in their ability to evolve and develop ever greater virtues, such as compassion and wisdom; material things and the processes they undergo are invariably limited—one can learn more and more without limit, but one cannot heat water infinitely, for at a particular point in time it just boils away. For Dharmakīrti and his followers, the two types of entities manifestly appear to be—and it is thought that they thus are—very different kinds of things. The later Tibetan tradition, for example, in its primers on Dharmakīrti’s philosophy of mind, that is, their so-called blo rig(s) literature, would characterize mind as “clear and knowing” (gsal zhing rig pa), it being understood that “clear” highlights the manifestly “immaterial” (gzugs can ma yin pa) character of mind and that “knowing” is in opposition to the manifestly insentient character of matter (bem po). And an appeal to phenomenal, or manifest, dissimilarity of the mental and the physical properties has also been a fundamental theme in many Western philosophers’ defense of dualism: there would seem to be a whole lot of properties we could ascribe to mind that we couldn’t easily ascribe to brains and vice versa. To put it roughly, minds have experiential contents, subjective aspects, qualia, private aspects, and so forth, whereas brains do not. The question then arises:  if we strip away the rather hopeless project of proving reincarnation and providing proofs for mind causing mind—in short, steps (b)  and (c)—is the remaining scaled-down Dharmakīrtian argument for mind in step (a)  philosophically plausible in the current debate? The short answer:  it is perhaps plausible when pitted against some forms of physicalism, but it will fare much less well against the radical versions we are considering. Indeed, an appeal to the fact that mind and matter have different properties may



On Minds, Dharmakīrti and Madhyamaka

53

plausibly work against those Western physicalists who hold that the two are contingently identical entities, but it works much less well against the Churchlands’ approach, that is, what is termed “eliminativism.” The Churchlands, Steven Stich, and before them Richard Rorty, with his “Disappearance Theory of Mind,” do not maintain that there are minds and that they also are identical with something material, but rather that mind is most likely a pseudophenomenon and hence eliminable, much like a whole mass of other pseudophenomena that have been eliminated by not figuring in better replacement theories of what there actually is in reality. If there is a scientifically adequate physicalist account of all that had previously seemed to require explanation in terms of mental entities, then—so argue the Churchlands—those mental entities will fall by the wayside, just as demons, occult forces, humours, astrological forces and other postulates of superstition and outdated medicine fell by the wayside with the advent of a better-informed modern medical science. The going common-sense views about the existence and nature of mind would turn out to be an outdated and false folk theory. Now, the Churchlands’ eliminativist position specifically targets propositional attitudes—that is, believing that such and such a proposition is the case: knowing that …, hoping that …, wishing that …, thinking that …, intending that …, fearing that …, and so on and so forth. According to our common-sense ideas and parlance, we do, of course, hold that we believe, think, fear, and so on that various things are thus and so, and we maintain we should believe that such and such is so if something else is the case. Such beliefs, thoughts, and so on, are individuated by their propositional content, and involve intentionality (i.e., they are directed to something) and normative concepts (e.g., what follows from what; what we should think). Many philosophers recognize that those features and others are extremely difficult to explain in a usual physicalistic description of what there is. The Churchland-style eliminativist physicalists too hold that those features very likely won’t be vindicated by a seamless reduction to brain states. But they go further: it’s actually not even worth trying to do so. So much the worse for propositional attitudes and so much the worse for the folk theory. As Paul Churchland puts it, “[F]‌olk psychology is a radically inadequate account of our internal activities. … On his [the eliminative materialist’s] view it will simply be displaced by a better theory of those activities” (Churchland 1981, 72). Strong stuff. I  think the Buddhist may be able reply to such radical eliminativism in a number of ways, but I also think that those ways, if they

54

Tom J. F. Tillemans

are to have a chance of being taken seriously, will not be the approach that Dharmakīrti and his followers put forward in step (a). The problem with the general Dharmakīrtian strategy is this: it doesn’t matter to an eliminativist that mind and brain don’t appear to be the same sorts of things, or that they have very different types of phenomenal features, or that minds would involve propositional attitudes and brains wouldn’t. After all, eliminativists will retort that it doesn’t matter all that much to them that demons, phlogiston, and other pseudoentities might seem to have such and such properties, while the entities figuring in scientific theories explaining how diseases, combustion, and so forth come about don’t have those properties. The real point for the eliminativist is that citing the distinct properties of demons, what they look like and so forth, is just useless deliberation about pseudoentities; all that is worth talking about and all that is real is what figures in the physicalist account—that is all there actually is. Where Dharmakīrti was convinced that the two entities looked too different to be the same, the modern-day eliminative physicalist accepts one type of real entity and waves away all dualist intuitions about the manifest distinctness of the mental from the physical as irrelevant. The primary issue at stake in this contemporary debate over eliminativism, then, is not whether mind does or doesn’t have such and such distinct properties, but whether we could just drop the folk theory pretty much all together. How far can ordinary worldly beliefs and linguistic practices be reformed and replaced, for obviously eliminating talk of the propositional attitudes—beliefs, thoughts, feelings, intentions, and so on—is an enormous rupture? The Churchlands’ view is indeed an extreme one. They, in effect, propose a wholesale error theory about our intellectual life, and advocate that those errors would not figure at all when we finally have a first-class scientific upgrade.

2.3 Madhyamaka It is a fact of history that the burden of defending mind against forms of physicalism has, over time, been delegated almost exclusively to the Buddhist epistemologists, the pramāṇavāda school, so that we find few arguments on these issues other than those (more or less) in keeping with the second chapter of Dharmakīrti’s Pramāṇavārttika. And in the contemporary East-West dialogues on these subjects, whether in Mind-Life discussions14 or in the Buddhist teachings that Westerners follow in Dharma



On Minds, Dharmakīrti and Madhyamaka

55

centers, Dharmakīrti’s thought is overwhelmingly present. But couldn’t there be a more promising Buddhist approach to the mind-body problem? There is another great Buddhist philosophy that needs to be examined here, one that has a radically different stance on metaphysics from the pramāṇavāda school in that it does not recognize any entities whatsoever as fully or ultimately real—in effect it shuns ontology all together. This is the Madhyamaka, the Philosophy of the Middle, that stems from Nāgārjuna (second to third centuries ce), Bhāviveka (sixth century), Candrakīrti (sixth century), and other luminaries in India and Tibet, a philosophy that holds that everything, be it mental or physical, is empty of intrinsic nature (svabhāva).15 Could there be a defense of mind following Madhyamaka philosophy instead of Buddhist epistemology? Indeed, I  think that there is a Madhyamaka case to be made in the antiphysicalist debate, one that has a significantly better chance of handling radical eliminativism than the Pramāṇavārttika perspective. Whether it would tell against other contemporary physicalisms is a more ambitious matter. I’m convinced that a broadly Madhyamaka-style quietistic rejection of ontology is important there too, but I cannot go into that here. In any case, radical eliminativism is one of the most uncompromisingly stark varieties of physicalism circulating these days so it is first potentially important to see if the Buddhists have something to say to it besides rather dated Dharmakīrtian arguments. After that, one would have to take up the question of whether other prominent physicalisms somehow collapse into a type of eliminativism and how uncompromising they would be. A Madhyamaka case for mind will largely turn on its position on the two truths, customary (saṃvṛti) and ultimate (paramārtha).16 The particularity of its stance on the customary is that it is global: nothing is ultimately established; there are only customary things, things that are the way they are because of human contributions, like designations (prajñapti), conceptions (kalpanā), linguistic labeling and conventions (saṃketa), and customary transactions (vyavahāra). This is sometimes characterized as a type of global antirealism,17 although I think it can also be convincingly regarded as a sort of quietism holding that the whole question of the reality or unreality of a world outside human transactions is misguided and one on which we should not take positions or hold theses (pakṣa). No doubt the hardest question in interpreting Madhyamaka is how to take its idea that all things are what they are because of thought, language, and their place in human transactions. The force of that “because” will

56

Tom J. F. Tillemans

vary considerably according to different Buddhist writers, as we can see by contrasting two major Tibetan Madhyamaka approaches. One of the more extreme Tibetan schools, the Jo nang pa, will in effect offer a kind of global error-theory, a strong antirealism: we mistakenly think and say that there are such and such things, but they are only “existent from the point of view of mistaken minds” (blo’i ‘khrul ba’i ngor yod pa); customary truths “exist” because of our errors. This is starkly formulated, but it is not far from what many Indo-Tibetan Mādhyamikas held: it is what I have called “typical Prāsaṅgika.”18 Others, such as the Gelukba (dge lugs pa), will say that there are right answers and sources of knowledge (tshad ma = pramāṇa) about how customary things are. Their point seems to be that customary things simply have no way they are (sdod lugs) intrinsically (rang bzhin gyis = svabhāvena) that could offer a deeper justification for why we talk and think of them in the ways we do; if we persist in demanding something real and fundamental that makes us at least sometimes right, we risk unintelligibility—the idea of a metaphysical undergirding for our ordinary practices turns out to be incoherent.19 Such a quietistic interpretation is nicely put by saying that there is no other way of establishing things (‘jog byed gzhan med pa) besides, or beyond, our various customary transactions (tha snyad  =  vyavahāra).20 Antirealism and quietism are two different interpretations of the Madhyamaka, and I strongly prefer the latter, but it may be that for the present discussion we can more or less work with either. So can customary truth, for a Mādhyamika, be reformed and replaced, and if so how much of it? As we saw, a Churchland-style physicalist would indeed advocate very radically changing our folk theories in favor of a first-class upgrade. Do the Mādhyamikas have arguments, turning on features of customary truth, to show that this is neither desirable nor possible? I think they do, especially the Mādhyamikas of Candrakīrti’s school do, and that this leads to a potentially more promising Buddhist defense of mind from radical eliminativism. True, historically some Mādhyamikas, influenced by Buddhist Logic and of a so-called Svātantrika persuasion,21 did seem to maintain that large-scale reforms in what is recognized as customary truth were desirable—for example, Śāntarakṣita (eighth century) and others argued that idealism and nominalism represented needed upgrades of lower-level views on customary truth. Candrakīrti and Prāsaṅgika-Madhyamaka, on the other hand, were adamantly opposed to all such global reforms proposed by philosophers. In accepting only “what the world recognizes” (lokaprasiddha),



On Minds, Dharmakīrti and Madhyamaka

57

Candrakīrti may perhaps accept piecemeal changes in accordance with the standards and procedures of the world, but it is quite clear that he has no use for across-the-board reforms, such as idealism, nominalism, the logicians’ ontology, their account of sources of knowledge (pramāṇa), and other such matters. Philosophical attempts to do better than the world in the name of metaphysics and epistemology are dismissed by him as a type of “intoxication” (smyos pa).22 We can confidently suppose that he, or someone carrying on his views, would have a similar reaction if confronted with a modern-day physicalist who proposed a large-scale upgrade of ontology in the name of science.

2.4  The Arguments One can find two types of Prāsaṅgika-Madhyamaka arguments against large-scale reforms of customary truth, the first showing the enterprise to be pointless, the second that elimination of basic features of the customary is incoherent. The first is the type of argumentation that we find in Madhyamakāvatāra 6.48, where Candrakīrti seeks to show that a large-scale reform (notably the Yogācāra rejection of the external world) is pointless because nothing more real is gained in this way. This seems to be a generalizable “by default” argument against large-scale reform: there is nothing ultimately real to be gained by such reforms of customary truths (given that everything is empty and thus of equal status), so why not just leave them (largely) as they are? The argument, however, is not likely to be persuasive to anyone other than a convinced “propounder of emptiness” (śūnyavādin). While it might work for someone who held a global view that nothing whatsoever can be intelligibly taken as fully real, it would not for a person, Buddhist or otherwise, who held that there are some real things and that it is our duty as philosophers to come up with an ontology that reflects that fact. We will concentrate on the second type of argument, that is, the incoherence of across-the-board reform. I think that the most plausible reconstruction of a general Madhyamaka case against large-scale upgrades is one that invokes “indispensability arguments,” or, in other terms, “transcendental arguments,” that show that such and such a proposed reform would lead to self-contradiction because if it were true, it would undercut the conditions indispensable to its truth or, at least, to our sincerely asserting it to be true. That strategy seems to figure quite regularly in Madhyamaka

58

Tom J. F. Tillemans

thought and indeed in other areas of Indian philosophy. Dan Arnold has maintained that this type of argument is what is at stake in Nāgārjuna’s defense in Mūlamadhyamakakārikā (MMK) 24 of emptiness (and hence dependent arising) as being indispensable to any discussion whatsoever about things in the world, including whether there are Four Noble Truths or not. Nāgārjuna’s point in MMK 24.20 is that if things were not empty of intrinsic existence and thus were not dependent upon causes and conditions, there could not be anything that ever arose: a precondition therefore for any discussion about things and their status is that they be dependently arisen, that is, empty of intrinsic existence.23 We can also see a type of transcendental argument behind the rejection of large-scale reforms of entrenched features of common sense or customary truth, but it is one that goes in the opposite direction of MMK 24.20. Instead of arguing that emptiness is a precondition for any discussion, it holds that leaving the customary largely intact is also a precondition for understanding emptiness. Such seems to be a key part of the argument against large-scale reform that is to be found in MMK 24.10 and Candrakīrti’s Prasannapadā thereupon: But, unless one accepts what is customary in the world, i.e., what is expressed, expressions, consciousness and objects of consciousness, one cannot teach the ultimate truth [i.e., emptiness]. . . . Thus, to show this [Nāgārjuna] states: “The ultimate is not taught unless one bases oneself upon the customary [MMK 24.10].”24 The Prasannapadā gloss of this passage contains two separable ideas. The first is obvious: arriving at a nonconceptual and nondiscursive realization of an ultimate truth beyond language is not practically, or pedagogically, possible without an initial explanation based upon language and concepts of things. But there is also a generalizable principle implicit in MMK 24.10 that can be formulated as follows: so long as we are in fact thinking discursively about a subject or teaching it—whatever it might be and whatever we might think about it—we must accept and rely largely upon the basic linguistic conventions and conceptual distinctions of the world. They are presupposed in the very fact of there being such thought or discussion. Whether on emptiness or another subject, the argument cannot then result in the outright elimination of great swaths of customary truth, on



On Minds, Dharmakīrti and Madhyamaka

59

pain of undercutting the preconditions for that argument to occur at all. If that is right, then there could not be a wholesale upgrade where all of customary truth, or its essential structures, was simply replaced by something better. This is, I would maintain, a transcendental argument: we cannot rationally advocate wholesale elimination of the most basic features of our ordinary conceptual scheme without a type of self-contradiction, more specifically a type of pragmatic contradiction. In the case of attempts to abandon talk of universal properties, negative states, persons, and so forth, it is indeed difficult to see how one can discuss such matters without at least recognizing that there are in some way or another properties that pertain to several particulars, states that do or do not obtain, and people who talk, deliberate, and so forth. Equally, we cannot even discuss the merits of eliminative physicalism without recognizing that there are thoughts about the subject, and good reasons for us to think in favor or against. Nor can we accept, or assert with conviction, that something is the right position if we don’t allow that we can, in a meaningful sense, come to believe it to be right. The incoherence, a kind of practical self-defeat, is that if such eliminativist positions were true, they would themselves all be unbelievable and unassertable. Ever since Lynne Rudder Baker’s book Saving Belief (Baker 1987), this argument for the indispensability of the mental—belief is an exemplary case—has often been called the argument against “cognitive suicide.”25 A decade later, after Saving Belief, Baker explains the basic idea as follows: To deny the commonsense conception of the mental is to abandon all our familiar resources for making sense of any claim, including the denial of the commonsense conception. (Baker 1998, 12) The claim denying the commonsense conception, if true, undermines extant ideas of language and understanding; so without a new account of how language can be meaningful in the absence of belief and intention, we have no way to interpret the claim denying the commonsense conception. (Baker 1998, 15) Is such a radically new physicalist account of our intellectual life without propositional attitudes forthcoming? It does not exist now, nor does it even

60

Tom J. F. Tillemans

appear to be on the horizon. Eliminative physicalists like Paul and Patricia Churchland—who are of course inclined to the extreme idea that we have never actually had propositional beliefs, thoughts, and so on, in the first place and need to finally come to grips with that uncomfortable fact—give tentative scenarios of a weird future world in which such confusions are thoroughly eliminated from our going theory in favor of a new propositionless form of “understanding.”26 There have been enough critiques of those Brave New World scenarios, and it is all too easy to pile scorn on failed attempts. A more interesting debate centers on the question: What would the failures show, if anything, about the truth of the eliminativist position? It is sometimes said that eliminativists could simply acknowledge that they cannot as yet explain how we make assertions in the beliefless world, but that in any case people don’t generally need to give a theoretically satisfying account of what assertions are in order to simply make them.27 Baker replies that in the beliefless world of the eliminativist there can be no account of how assertions could be true or false, sincere and insincere, and so on, nor even of the difference between assertions and mere emissions of sound. (In our ordinary concept of an assertion, for example, the idea that such and such is a sincere assertion of p presupposes that the utterance of p is somehow conditioned by the belief that p.) Hence her point that, as far as we have any reason to think, the eliminativists have no recognizable concept of a sincere, meaningful, or true assertion.28 The burden of proof would fall on them to provide more than mere promissory notes. Still, it has to be said that these debates will continue, and it should be no surprise that cognitive-suicide arguments do not just simply carry the day unproblematically. Physicalism certainly doesn’t just lie down and die:  it rolls with the punch. So, obviously much more can be said and would need to be said to flesh out this approach. In the last section of this chapter, we’ll sketch out a sample of potential Buddhist moves in some of the further debates that fairly naturally arise.

2.5  Some Remaining Issues First, the radical eliminativist might perhaps retort that she could change any individual element of our folk psychology and thus arrive finally at a large-scale upgrade, but that suicide comes in only if we were to try to change them all at once. She could in effect appeal to Otto Neurath’s



On Minds, Dharmakīrti and Madhyamaka

61

famous analogy of mariners at sea who can repair their boat only plank by plank, because they need to use the boat while they fix it. Improvements to conceptual schemes would be like that boat repair. The Nāgārjunian would have to reply that that is not the point. Rather, the point would be that several such individual elements—beliefs, thoughts, minds, and likely universals, persons, negative states of affairs, and others—cannot, or will not, be replaced at all. Second, let us grant, as I think we should, that it should be very difficult for the Churchlands, or anyone else, to picture convincingly a human life without thoughts, beliefs, and all the other propositional attitudes. And let us suppose, as I think is the case, that this extreme conceptual difficulty of doing without them supports the view that (as far as one can reasonably see) they are here to stay. But would the Madhyamaka position be that radical elimination of our folk psychology is incoherent (or “pragmatically self-contradictory”) and thus cannot happen (i.e., à la Lynne Rudder Baker) or simply that it de facto will not happen because the most fundamental folk psychological ideas are not in any case theoretical sorts of views? Indeed, Charles Chastain (1998) has argued that the folk psychology with its various ways of speaking, thinking, and so on, is not a theory at all and is thus not going to be challenged or replaced as a bad theory by better science. Is MMK’s defense of the customary then a bona fide transcendental argument or is it a just a reasonable prediction that fundamental customary truths will largely remain unchanged whatever our theoretical deliberations? Exegetically, the former seems more likely, as there does seem to be, at both crucial points in MMK 24.20 and 10, a strong appeal to necessity and indispensability. These passages are not just arguing for a de facto immunity of the customary. Third, one can imagine an intra-Buddhist objection:  doesn’t the Mādhyamika herself in some sense eliminate the customary when she arrives at the path of seeing (darśanamārga) and directly perceives only the lack of intrinsic natures, that is, the emptiness of all things? The point is tricky. There is no doubt that Mādhyamikas maintain that customary things do not appear to the āryas (Noble Ones) in their meditative absorption on the ultimate truth, emptiness. Nonetheless, it is quite another thing to say that Mādhyamikas eliminate customary truth or refute (bādhā, gnod pa) its existence. As I have argued elsewhere, later Mādhyamikas, like Candrakīrti, Kamalaśīla (eighth century), and others, including the fourteenth-century luminary Tsong kha pa, have quite clearly maintained that the perspective of emptiness does not actually refute customary truth but only false projections,

62

Tom J. F. Tillemans

or superimpositions (sgro ‘dogs = samāropa), upon it: the customary is unassailed by any ultimate perspective, be it an analysis or the resulting direct perception.29 In fact, we can go further: it looks like this is nothing less than the mainstream position of Indian Madhyamaka from the sixth century on, be it Svātantrika or Prāsaṅgika. As, for example, Jñānagarbha (eighth century) and Śāntarakṣita argue in Satyadvayavibhaṅgavṛtti and Satyadva yavibhaṅgapañjikā, if someone philosophically analyzes customary truth, then it is not customary truth, but something else (don gzhan) that they are deliberating about, and thus they would end up completely refuted (gnod pa ‘ba’ zhig tu zad do) for having strayed from the main subject of debate!30 In short the Mādhyamika targets a superimposed nature of customary truths, namely that they are established intrinsically and independently, whereas the customary truth per se is never attacked. Finally, would these transcendental arguments, if successful, prove any truths about what there really is and isn’t in the world, or does that matter all that much to a Buddhist? They might show that there are pragmatic self-contradictions in denying the customary, but would they establish any genuine truths? I would say—perhaps surprisingly but not cavalierly—that the issue is indeed not pressing for a Mādhyamika. Transcendental arguments are used in many areas of philosophy, and the recurring problem is invariably to evaluate what, if anything, they prove or if they could ever answer a skeptic. Ever since the pioneering articles of Barry Stroud on the logic of such arguments, analytic philosophers have recognized that while transcendental arguments may well show the indispensable need for us to believe in such and such things—for example, persons, properties, minds, propositional attitudes, and so on—a skeptic may still retort that none of them are thereby proven to really exist.31 As Stroud emphasizes, there seems to be an unbridgeable gap between us having to think that p is the case, on the one hand, and p actually being the case, on the other.32 Now, it has become routine in the growing philosophical literature on transcendental arguments to distinguish between strong versions that would seek to answer a skeptic head-on and prove some fact in the world that he doubted, and weaker or more moderate versions that are belief-oriented rather than fact- or truth-oriented, and thus would seek to show no more than indispensability to our conceptual scheme.33 Reading MMK 24.10 as a transcendental argument would make it come squarely down on the side of moderation, one that may show that we, by our own standards, are reasonable to hold such beliefs, even if it does not refute a persistent skeptic who holds that our standards may all be wrong and



On Minds, Dharmakīrti and Madhyamaka

63

that things as they actually are could be very different from all our beliefs about them. Is this moderate stance good enough to establish mind? Probably not for a metaphysical realist, who accepts that there is a completely belief-independent way in which things are as they are anyway. Suffice it to point out here that while Stroud’s gap may thus well be unbridgeable for such a realist, it is difficult to see that it could tell against a Mādhyamika, who espouses the idea that how things are cannot be separated from our practices of how we conceive of them, talk of them, and act upon them. In short, when realists hold that real facts are as they are anyway, completely independent of all our beliefs, conceptions, and so on, then the move from how our beliefs invariably are to how the facts must be becomes very problematic. This would indeed be Stroud’s gap. But, for better or for worse (depending upon one’s attraction to metaphysical realism), it would seem that a gap like that can hardly exist or even be intelligible to an antirealist or quietist, East or West. The Mādhyamika approach to the realist’s persistent skepticism about transcendental arguments might thus be a wise refusal to pursue the debate. And this, I should think, would not be cavalier at all, but would be a significant, reasoned, Buddhist stance: mind should essentially be left “fine as it is unanalyzed” (avicāraramaṇīya), with no deeper justificatory account needed or possible. This is not still yet another theory of mind; it is a Buddhist quietism about the philosophy of mind.

Notes 1. The present chapter, in different forms, was presented at the Center for Buddhist Studies of the University of Kathmandu and the Department of Philosophy of the University of Kyoto. It was subsequently the keynote address of the Sixteenth Congress of the International Association of Buddhist Studies, Taiwan, 2011. Thanks for helpful remarks along the way to Jay Garfield and Koji Tanaka. A somewhat shortened version is appearing in my book How do Mādhyamikas Think?, forthcoming in the series Studies in Indian and Tibetan Buddhism (Somerville, MA: Wisdom). 2. Cf. Taber (2003, 481): “[Physicalism is] the view that all phenomena including consciousness either are physical in nature or at least can be explained physically, in terms of the laws and principles of the physical sciences.” 3. See Houshmand et al. (1999). 4. The text is studied and translated in Franco (1994). 5. On Dharmakīrti’s life, works and philosophy, see Tillemans (2011a).

64

Tom J. F. Tillemans

6. For a translation and study of the relevant section in Pramāṇavārttika, i.e., ­chapter 2, verses 34–72 (PV 2:34–72), plus Prajñākaragupta’s commentary, see Franco (1997). 7. Cf. Taber (2003, 490): “It would appear to be a basic principle of causation that like causes like: trees give rise to other trees, horses give birth to horses, milk yields yoghurt, and clay is made into pots. … If the body did cause cognitions, if something could give rise to something else completely different in nature, then, says Dharmakīrti, there would be ‘unwanted consequences’ (atiprasaṅgāt, P[ramāṇa]V[ārttika] 2.35–36a). He doesn’t tell us what these unwanted consequences are, but he may have had in mind the same possibility the Sāṃkhya believed would result from the truth of the asatkāryavāda, that ‘anything could come from anything.’ ” I think that is exactly right. “Anything coming from anything” is a frequent worry in discussions of causation, Madhyamaka included. 8. See Tillemans (1993). 9. See “Why Dualism is Forlorn,” Dennett (1991, 33–39). 10. If he had, his own anti-Cārvāka arguments would have become superfluous debates with an unqualified adversary. Note that Dharmakīrti did place entry requirements on certain debates about ethical matters—e.g., acceptance of Buddhist scripture is necessary to deliberate about the details of the law of karma—but he didn’t do that in his defense of mind. 11. See Tillemans (2010). 12. On Dharmakīrti’s and Kamalaśīla’s theories of meditation and yogic perception, see Tillemans (2013). 13. See Tillemans (2011a). 14. See, for example, http://www.mindandlife.org (accessed Nov. 30, 2013). See also Flanagan (2011, chap.  3) on his arguments with the Dalai Lama about physicalism. 15. For the history, texts, and philosophy of this school, see, for example, Seyfort Ruegg (1981); see also Seyfort Ruegg (2010). 16. To get some idea of what this is about, one can consult Newland and Tillemans 2011. 17. See, for example, Siderits (1988); Westerhoff (2011). 18. See the introduction and c­ hapter 2 in Tillemans (forthcoming). Note that there are no doubt better global antirealisms on the market than a wholesale error theory such as that of the Jo nang pa. A potentially seductive interpretation of Madhyamaka and its denial of customary things being what they are “intrinsically” might be to see it in terms of Michael Dummett’s global antirealism. One might then attempt a rational reconstruction of Buddhist realisms and antirealisms by saying that the former accepts evidence-transcendent ascriptions of truth and the latter does not. This might well yield a palatable global anti-realism which consists in saying that there never are evidentially unconstrained truth-conditions. See Hale (1999, 283ff). It appears that Madhyamaka antirealism is largely taken in this way in Siderits 2003.



On Minds, Dharmakīrti and Madhyamaka

65

19. This is not far from the interpretation developed in Perrett (2002). It bears resemblance to a type of minimalism advocated by Mark Johnston that holds that metaphysical pictures of what there is do not (or, what is stronger, cannot) provide the justificatory undergirdings of our ordinary practices. 20. See the summary by Go rams pa bSod nams seng ge of Tsong kha pa’s views in the former’s lTa ba’i shan ‘byed (Sa skya Students’ Union, Sarnath edition, 14): dbu ma rang rgyud pa rnams kyis tha snyad du rang gi mtshan nyid kyis grub pa’i chos khas len gyi / ‘dir de khas mi len pas gang zag dang chos tha snyad du ‘jog pa’i tshul ni / ‘di ni lha sbyin no ‘di ni lha sbyin gyi rna ba’o zhes sogs tha snyad btags pa’i tshe na / tha snyad de’i dbang gis lha sbyin dang lha sbyin gyi rna ba la sogs pa yod par ‘jog gi de las gzhan pa’i ‘jog byed med pa ni tha snyad du yod pa’i don no (“The Svātantrika-Mādhyamikas accept that on the level of the customary, things [i.e., dharmas] are established by their own intrinsic characteristics. However, here [in the Prāsaṅgika system] this is not accepted. Thus, the way they establish people and things customarily is that when we apply linguistic conventions such as ‘This is Devadatta,’ ‘This is Devadatta’s ear,’ we establish customarily via linguistic conventions that Devadatta, Devadatta’s ear, and so forth exist. But there is no other way than that to establish them—this is what is meant by customary existence.”). 21. On Svātantrika-Mādhyamikas and Prāsaṅgika-Mādhyamikas, respectively— “The Ones who Proceed by Autonomous Reasonings” and “The Ones who Proceed by Logical Consequences [of other’s views]”—see Dreyfus and McClintock 2003. Note that the terms “Svātantrika” and “Prāsaṅgika” are modern Sanskritizations of Tibetan terms for a distinction between Indian Madhyamaka subschools. The distinction is not labeled as such, nor probably even explicitly made, in Indian texts, but it is nonetheless arguably of very great importance in understanding Indian Madhyamaka. 22. See Tillemans (1990, 1:177, 179), §8, §17. 23. See Arnold (2005b). 24. Sanskrit in Louis de La Vallée Poussin’s edition (La Vallée Poussin 1913, 494). French translation in May (1959, 229). 25. See also Garfield (1988). 26. Indeed the prospects for the beliefless world are more than a little weird: e.g., we may simply be without language (Churchland 1981, 88: “[S]‌poken language of any kind might well disappear completely, a victim of the ‘why crawl when you can fly?’ principle”), have bookless “neural” libraries (88: “Libraries become filled not with books, but with long recordings of exemplary bouts of neural activity”), and, last but not least, we may “understand” other people only in the nonpropositional, purely causal, way in which one brain hemisphere “understands” or “conceives of” the other (88). 27. See Chastain (1998). 28. Cf. Baker (1998, 27): “Thus, as far as we know and as far as we have any reason to think, eliminative materialism is assertible only if it is false.”

66

Tom J. F. Tillemans

29. See Tillemans (2004). 30. See Jñānagarbha’s Satyadvayavibhaṅgavṛtti, sDe dge edition, f.10a–b: ci ste kun rdzob ni ji ltar snang ba yin te / de la ni ji skad bshad pa’i dpyad pa’i gnas med pa nyid do // ‘di ltar / rnam par dpyod pa byed na don gzhan du song bas gnod par ‘gyur // kho bo cag ni ‘di la dpyod par mi byed kyi / dpyod par byed pa la ni ‘gog par byed do // gal te dpyad par byas te ma rung na ma rung du zad do // ji ltar snang ba’i ngo bo kun rdzob pa la brten nas de dpyod pa byed pa ni don gzhan du ‘gro ba’i phyir gnod pa ‘ba’ zhig tu zad do (“But the customary is just as it appears. There is absolutely no place for analyzing it in the [philosophical] way just explained. The point is as follows. If [you] do analyze, then [the object] is something else, and so [you] would be refuted. We [Mādhyamikas] do not analyze this [customary truth]. But, if [you] analyze it, [you] are refuted. After [you] have engaged in analysis, then if something is impossible, it is nothing other than impossible. An analysis of that [object], after one has relied on the customary nature just as it appears, concerns something else, and that is why [such an analysis] is completely refuted.”) Śāntarakṣita, in his Satyadvayavibhaṅgapañjikā, accepts this view of Jñānagarbha and elaborates upon it; see D. f.38b–39a: ‘di ltar ji skad bshad pa’i rnam par dpyod par byed pa yin na ni // gnod pa de tshar gcad pa’i gnas su ‘gyur ro // ci’i phyir zhe na / don gzhan du song bas te / skabs yin pa’i don bor na rtsod pa gzhan la brten pa’i phyir zhes bya ba’i tha tshig go // ‘di rnam par ‘grel bar byed de / dbu ma pa kho bo cag ni kun rdzob ‘di la dpyod pa sngar ma bstan gyi / khyed kyis byed na ‘gog par byed do / gal te dpyad pa byas te / de dag gi de ma rung na ma rung du zad do // kho cag la ci gnod (“If one were to engage in analysis of the sort just explained, then that refutation would be a point of defeat [tshar gcad pa’i gnas = nigrahasthāna]. Why? Because it would concern something else. In other words, this is because when one abandons the object that is actually being discussed, one is then reliant upon a different debate. This is commented as follows: We [kho cag] Mādhyamikas have not as yet formulated any analysis of this customary [reality], but if you [khyed] did so, then you would be refuted. After [you] have engaged in analysis, then if something about these [objects] is impossible, it is nothing other than impossible. What refutation would this be for us [Mādhyamikas]?”) Translations are my own; cf. Eckel (1992, 89). 31. See Stroud (1968; 1994). 32. Stroud (1994, 234) put the matter as follows: “Even if we can allow that we can come to see how our thinking in certain ways necessarily requires that we also think in certain other ways… how can truths about the world which appear to say or imply nothing about human thought or experience be shown to be genuinely necessary conditions of such psychological facts as that we think and experience things in certain ways, from which the proofs begin?” 33. See Stern (1999, 51).

3

On Being Humean about the Emptiness of Causation Ricki Bliss

According to the Madhyamaka doctrine of emptiness, as set out by Nāgārjuna in the Mūlamadhyamikakārikā (MMK), everything is empty: everything is causally, mereologically, and, most crucially, conceptually dependent upon something else. In the particular case of causation, cause and effect are dependent upon one another for their existence; and dependent upon us as well. Our cognizing activity, explanatory practices, desires, and interests inform how we carve up the world, and how the causal nexus emerges. Causation is as dependent upon us as it is any actual features of reality.1 In an attempt to capture this aspect of the emptiness of causation, philosophers have tended to deny the obtaining of necessary connections between cause and effect. In particular, these philosophers follow Hume in claiming that any perceived necessary connection between cause and effect is an artifact of the mind: it is the ability of a cause to excite in us the idea of its effect that leads us to expect the one to follow upon the other. Or something like this. The causal relation is empty in virtue of being a mental artifact; and the causal relata are empty in virtue of being interdependent. Causation is a feature of conventional reality. Jay Garfield (2002, 24–45), for example, is quite explicit that what the emptiness of causation is suggestive of is something like the regularity account of causation as espoused by Hume. And Jan Westerhoff (personal correspondence) is also of the view that the Humean psychological account of necessary connections is in step with the emptiness of causation.

68

Ricki Bl iss

In advancing his regularity account of causation, however, Hume was advancing a principle that pertained not simply to the nature of the causal relation but also to the causal relata. According to Hume ([1738] 1985, bk. 1, pt. 3, sec. XIV), when we contemplate objects in themselves, we cannot locate the source of any necessary existential connection between them. According to Hume, necessary connections fail to obtain between existents that are distinct from one another. In its contemporary guise, Hume’s Dictum says that there are no metaphysically necessary connections between wholly distinct, intrinsically typed existents.2 Any principle that makes reference to “wholly distinct, intrinsically typed existents” surely ought to ring alarm bells in the head of any good Mādhyamika. For cause and effect to be wholly distinct from one another, and intrinsically typed, just sounds like what it is for cause and effect to fail to be empty. The worry that this chapter is concerned with, then, is that employing Hume’s Dictum as a means of emptying out the relation comes at the unfortunate cost of filling up, so to speak, the relata. If there are no necessary existential connections between cause and effect, then either there are no existential connections between them at all, or they are merely contingent. But if there are no existential connections between cause and effect, in what sense do cause and effect depend upon one another? And if the existential connections between cause and effect are contingent, how can cause and effect be both empty and yet able to exist without one another? It is not at all clear how cause and effect can be both empty and not necessarily existentially connected; not least when we understand emptiness in terms of existential dependence relations. I argue that denying the obtaining of necessary existential connections between cause and effect entails the distinctness, and thus non-emptiness, of the causal relata. In section 3.1, I introduce the Madhyamaka doctrine of emptiness, and discuss the proposed Humean interpretation of the doctrine. In section 3.2, I offer an overview of the regularity account of causation. In section 3.3, I offer an extended discussion of the dictum itself. In this section I consider analytic versions of the dictum in section 3.3.1, and synthetic versions in section 3.3.2. I conclude that neither analytic nor synthetic versions of the dictum yield a regularity account of causation that is compatible with its emptiness. In section 3.4, I offer a defense of my analysis in anticipation of possible objections.



On Being Humean about the Emptiness of Causation

69

3.1  The Madhyamaka Doctrine of Emptiness The doctrine of emptiness as it is set out in the writings of Nāgārjuna says that absolutely everything is purely dependently existent. Being true of everything, this is also true of emptiness itself. Everything is causally, mereologically, and conceptually dependent upon something else: effects depend upon their causes, wholes depend upon their parts, and everything depends upon the activity of cognizing minds. To exist in this way is to exist conventionally. Absolutely nothing exists independently. How exactly to cash out these senses of dependence is one of the focal points of the contemporary Asian-analytic literature. The relevant technical term in operation in the Sanskrit literature is svabhāva. To say of something that it is nonempty is to say that it possesses svabhāva. This is sometimes rendered as something like own-being or self-nature; also as intrinsic existence or intrinsic nature. How, exactly, to render this notion into the contemporary analytic vernacular, however, is exceedingly difficult, and the source of much debate. It is not uncommon to see technical terms such as essence and substance employed, along with the aforementioned nontechnical terms. Garfield (2002, 25), for example, prefers to understand the notion of essence as what is being denied by the doctrine of emptiness. On this he states: when a mādhyamika philosopher says of a table that it is empty, that assertion by itself is incomplete. It invites the question, “Empty of what?” And the answer is, “Empty of inherent existence, or… self-nature, or, in more Western terms, essence.” Elsewhere (Garfield 2002, 29), on the relationship between inherent existence, essence, and independence, Garfield states: keep in mind for the present that for a thing to exist inherently is for it to exist in virtue of possessing an essence; for it to exist independently of other entities, and independently of convention. Things here are already tricky. One issue with employing the technical term “essence” is that there is an important sense of independence that it doesn’t seem to capture. Things with essences can depend upon their parts, for example. There is nothing about my being essentially human

70

Ricki Bl iss

that entails that I am capable of, or indeed do, exist independently of everything else. Even as one who is essentially human, I still depend upon my parts, I once depended upon my parents, and my being the person that I am depends in large measure upon the people around me, their ideas concerning me, and so on. I have an essence and yet fail to be independent in certain very important ways. Although, in virtue of possessing an essence, my nature is, in some sense, fixed. Westerhoff (2007), on the other hand, suggests that the notion of svabhāva encompasses two important aspects:  an ontological aspect and a cognitive aspect, both of which cut across different concepts as they exist in the Western tradition. It is when using terms such as “essence” and “substance” that we are picking out the distinctively ontological sense of svabhāva. The sense of svabhāva best picked out by the term essence- svabhāva is the sense in which svabhāva is associated with the natures of phenomena. In particular, it is associated with the idea not only that things have properties that individuate them from other phenomena—so, in this sense, they are independent of other phenomena, our conventions, and so on—but, also, that some amongst those individuating properties are such that their bearers cannot exist without them. Or, at least, the bearers cannot exist as the things that they are without bearing the relevant properties. In other words, if something has svabhāva in this sense, it possesses some of its properties in such a way that were it to lose those properties, it would cease to exist as the thing that it is.3 But the rejection of essence-svabhāva is not, according to Westerhoff, the focus of the critique of Nāgārjuna’s MMK; instead, it is the rejection of substance-svabhāva. Anything that possesses substance-svabhāva exists independently of everything else. Things with substance-svabhāva are causally, mereologically, and conceptually independent. Good examples of entities that possess substance-svabhāva from the Western tradition are the space-time points of the physicalist and the simples of the Tractatus. The simples qua substances exist independently of any parts, and are thought to be real features of an objective reality, which also makes them independent of our concepts. It is these kinds of posits that are the real target of the Mādhyamika analysis. I do not wish to adjudicate on these matters. Whether or not Garfield or Westerhoff is correct in his understanding of the MMK is not my concern.4 Rather, I  am interested in taking seriously the claims made in the contemporary commentarial literature and exploring some of their implications. For this reason, then, I suggest we take seriously the thoughts both that what Nāgārjuna is rejecting is the idea that things have their natures



On Being Humean about the Emptiness of Causation

71

from their own sides vis-à-vis the possession of an essence, and that things can exist independently of other things both categorically and internal to certain specified relation kinds, such as causation. What this latter point means is that not only does Nāgārjuna want to deny that there is a fundamental level of ontologically independent phenomena, but also that cause and effect are independent of one another, for example. I will argue that regardless of which of the two accounts proposed here that we adopt, the problems faced by assimilating the Humean Dictum will be the same.

3.1.1  Dependence Relations The kind of dependence relations at issue in the MMK, it has been suggested, can be understood in terms of relations of existential and nominal dependence. Following Westerhoff (2009a, 26), we can define existential dependence as follows: existential dependence: An object a existentially depends on objects falling under the property F iff [if and only if ] necessarily, if a exists there exists something falling under F. This definition of existential dependence presumably subsumes both temporal and atemporal relations of dependence. The relationship between cause and effect will involve a temporal, existential connection: causes precede their effects in time. We can say that whenever some a exists—and is the kind of thing that has a cause—then there must be some cause of a that exists (or has existed): a existentially depends upon its cause because whenever a exists there must be something that falls under the property “the cause of a.” So too for relations of atemporal existential dependence. Take the relationship between my chair and its parts. If my chair exists, then necessarily, there also exist things—or a collection of things—that fall under the property “parts of my chair.” The second notion of dependence in operation in the Madhyamaka literature is that of nominal or notional dependence. I follow Westerhoff (2009a, 26) and define this as follows: notional dependence: Objects falling under the property F are notionally dependent on objects falling under the property G iff necessarily, if some object x falls under F there will be a distinct object y falling under G.

72

Ricki Bl iss

Think of the notions of husband and wife in communities that do not permit same-sex marriages. A  husband is notionally dependent on his wife—and she on him—because whenever something exists, and falls under the property “husband,” necessarily something must also exist that falls under the property “wife.” The relationship that the individuals who happen to be the husband and wife under consideration here bear to one another is not one of existential dependence, for if one of these individuals passes out of existence, the other shall not cease to exist. They will simply cease to fall under the property of “husband” or “wife.” There are cases, however, in which the relations of notional and existential dependence intersect. If the properties under which the relata of the relation of notional dependence fall are essential properties of the entities under consideration, then the relation of notional dependence is also one of existential dependence. Consider the North and South poles of the earth. In ordinary cases the relationship between North and South is purely nominal. If South America were to be swallowed up by an enormous wave, North America would not similarly cease to exist. In the case of the North and South poles the situation is different. The relationship between the North and South poles is not merely nominal. Both poles are essentially such that the other pole must exist. If the South Pole qua magnetic pole ceases to exist, then so too will the North Pole. Magnetic poles exist essentially in this symmetric relation of existential dependence with poles of the opposite charge. With this background in place, we can now attend more closely to the specific doctrine of causation adumbrated in the Madhyamaka literature; in addition to setting out Garfield’s proposed Humean interpretation of the account.

3.1.2  The Emptiness of Causation To claim that causation is empty is to claim that cause and effect are empty of svabhāva. To think that cause and effect possess svabhāva is to think: cause and effect are qualitatively and therefore also quantitatively distinct objects. They do not require one another… Second, the existence of cause and effect as “independent objects” or as “existing from their own side” refers not just to their mutually independent existence, but also to their independence of a cognizing subject. According to such an objectivist understanding of causation, the



On Being Humean about the Emptiness of Causation

73

interlocking chain of causes and effects is something that exists in the world independent of any observers.5 This understanding, I  suspect, Garfield would also agree with. To deny that cause and effect possess svābhāva is to think that they are dependent upon one another and dependent upon us as well. There is an obvious sense in which cause and effect are notionally dependent upon one another: causes elicit effects and effects have causes. Wherever something falls under the property “being a cause,” necessarily, there is some distinct object that falls under the property “being an effect.” And vice versa. The terms “cause” and “effect” are notionally, and symmetrically, dependent upon one another. This much seems uncontroversial. It also seems uncontroversial to claim that effects are, in some sense of existential dependence, existentially dependent upon their causes.6 After all, exactly what it is to be the cause of an effect is to bring that effect into existence. The relation of dependence that holds between the two is existential. What would seem controversial, however, is the thought that causes are existentially dependent upon their effects. And this, Nāgārjuna wants to claim as well. One way in which we can make sense of this claim is by recalling, as mentioned above, the relationship between notional and existential dependence in cases in which the property under which some phenomenon falls—where it notionally depends upon something else—is an essential property. In these cases, notional dependence entails existential dependence; and the symmetry of the notional relation is inherited by the existential relation. Consider the properties “being a cause” and “being an effect” as essential properties. We already know that anything that falls under one of these two properties will notionally depend upon the other. But where these properties are essential properties, neither property bearer can lose whichever of these properties they possess without passing out of existence. This means that were whatever falls under the property “being a cause” to go out of existence, so too would whatever falls under the property “being an effect.” And vice versa.7 The existential dependence between cause and effect is symmetric. A second way Westerhoff suggests we might understand this seemingly bizarre claim is in terms of dependence between properties: it’s not the case that the phenomena falling under the properties symmetrically depend upon one another but that the properties themselves are symmetrically existentially dependent upon one another. Recalling, again, the

74

Ricki Bl iss

definition of notional dependence above, if something that falls under the property “is a cause” is notionally dependent on something falling under the property “is an effect”, then the property “is a cause” existentially depends on the property “is an effect”. For the property “is a cause” can only exist if the property “is an effect” does as well.8 A third and final way we can make sense of the symmetric dependence between cause and effect is by considering causation in terms of what Westerhoff refers to as causal fields. These causal fields contain both causes and the myriad background conditions that must be in place in order that causation is successful: the illumination of my lamp as a result of my flicking the switch requires that the wires are in working order and that the bulb is good, for instance. It also requires an absence of negative conditions:  maskers and finks. A  causal field is composed of the phenomena we consider to be explanatorily relevant to a given effect. A causal field is, thus, something of a construct, or an artifact of human engagement with the world.9 In this sense, then, causal fields are dependent upon their effects. Were there to be no effect, there would be nothing to explain and thus no reason to carve out a causal field in the first place.10 The upshot of the emptiness of causation is that cause and effect are interdependent. This means that neither cause nor effect is independent of one another qua cause and effect. And if Nāgārjuna’s analyses of the possible alternatives to this view are correct, then cause and effect are necessarily interdependent. We turn now to Garfield’s discussion of the emptiness of causation and his proposed Humean account of emptiness. Changing gears somewhat, we also return to Garfield’s characterization in terms of essencelessness. According to Garfield (2002, 25): And this analysis in terms of emptiness—an analysis refusing to characterize the nature of any thing, precisely because it denies that we can make sense of the idea of a thing’s nature—proceeding by the relentless refutation of any attempt to provide such a positive analysis is applied by Nāgārjuna to all phenomena. The upshot of the Mādhyamika doctrine of emptiness, on Garfield’s account, is that it is a mistake to impute intrinsic natures to phenomena; it is a mistake to think that they exist from their own side. In terms



On Being Humean about the Emptiness of Causation

75

of causation, it is a mistake to think that phenomena qua causes contain within them the power to bring about their effects. What the dialectic of the MMK is aimed at showing us is that phenomena do not have it within their natures to be the cause or effect of anything because no phenomenon has any kind of fixed nature whatsoever. This is what it is to be empty. If this view is correct, what are we to say of the perceived experience of causation? Instead of understanding reality in terms of some real—in the most metaphysically robust sense—causal structure, we should understand our experience of causation in terms of conditions. What is it then, about some sets of event pairs, but not others, that makes them dependently related, if not some causal link present in some cases but not in others? Nāgārjuna replies (1:5) that it is the regularities that count. Flickings give rise to illuminations. So they are conditions of them. If they did not, they would not be. Period.11 The flicking of the light switch, my desire for light, the switch being in good working order, and so on, on this view, are the conditions that bring about the light’s being on (these are just the causal fields to which Westerhoff refers). What we are seeking, in positing causes, are explanations for events. And appeal to conditions, as opposed to occult metaphysical forces, allows us to explain events perfectly well. Garfield (2002, 29) continues: Explanation relies on regularities. Regularities are explained by reference to further regularities. Adding active forces or potentials contributes nothing of explanatory utility to the picture. What Garfield is advancing here is a regularity view of causation, according to which our explanatory practices are guided by our explanatory interests, desires and so on. And this kind of approach to explanation is commonly coupled with a denial of real causal structure in the world. Again Garfield (2002, 29): we can see Nāgārjuna as urging that even distinguishing clearly between explanans and explanandum as distinct entities, with the former containing potentially what the latter has actually, is

76

Ricki Bl iss

problematic. What we are typically confronted with in nature is a vast network of interdependent and continuous processes, and carving out particular phenomena for explanation or for use in explanations depends more on our explanatory interests and language than on joints nature presents to us. Here Garfield is suggesting that we understand Nāgārjuna in terms of, something very much like, a Humean regularity account of causation. In fact, this is exactly what Garfield (2002, 72) suggests: Nāgārjuna argues that the midpoint between reification of causation, the adoption of a realistic view with respect to causal powers, and nihilism, the view of a random and inexplicable universe of independent events, is the acceptance of the reality of conditions, and a regularist account of explanation. On such a view, what counts as explanans and as explanandum depends on explanatory interests and conventions for individuation and classification. Hume is often read (properly in my view) in roughly this way. Such a view, hence far from nihilism, is instead a moderate, sensible approach to explanation and to understanding. On Garfield’s suggested interpretation of the doctrine of emptiness, as espoused in the Prāsangika Madhyamaka literature, we are to understand the Mādhyamika as anticipating, and advancing, a regularity account of causation. On this kind of account, our experience of causation is nothing more than our experience of the constant conjunction (regularity) of events. Any necessary connection that links cause and effect is introduced by us as a result of expectations, explanatory praxis, and our joint carving activity.12

3.2  Hume and the Regularity Account of Causation In A Treatise of Human Nature, Hume argues that the idea that there is a necessary connection between cause and effect—that causes contain within them a power to bring about their effects—is just that, an idea. Nowhere, in the consideration of cause or effect can we uncover the source



On Being Humean about the Emptiness of Causation

77

or location of this necessary connection. Hume (bk. 1, pt. 3, sec. XIV) puts this as follows: Suppose two objects to be presented to us, of which the one is the cause and the other the effect; it is plain, that from the simple consideration of one, or both these objects we never shall perceive the tie by which they are united, or be able certainly to pronounce, that there is a connection betwixt them… . Did we never see any but particular conjunctions of objects, entirely different from each other, we should never be able to form any such ideas. Later in the same section, Hume presents two definitions of the causal relation: We may define a CAUSE to be An [sic] object precedent and contiguous to another, and where all the objects resembling the former are placed in like relations of precedency and contiguity to those objects that resemble the latter… . A CAUSE is an object precedent and contiguous to another, and so united with it, that the idea, of the one determines the mind to form the idea of the other, and the impression of the one to form a more lively idea of the other. And Hume concludes that we should understand the notion of causal power, or necessary connections between cause and effect as follows: Upon the whole, necessity is something, that exists in the mind, not in objects; nor is it possible for us ever to form the most distant idea of it, considered as a quality in bodies. Either we have no idea of necessity, or necessity is nothing but that determination of thought to pass from causes to effects, and from effects to causes, according to their experienced union. There is limited consensus on how we are to understand the regularity view of causation. Broadly, interpretations of the account fall into three camps—with variations within the camps. According to traditional views, Hume is espousing a reductive theory of causation—causation can be defined in terms of temporal priority, contiguity, and constant conjunction. Amongst those who hold this traditional interpretation, there is disagreement over what this means for real causal relations.13

78

Ricki Bl iss

A more modest version of the traditional interpretation takes it that we are epistemically precluded from knowing whether there are real causal relations that hold between events in the external world or not.14 On this interpretation, as we have no impression of a necessary causal connection, and impressions are the only means by which we have access to the world, it makes no sense to speculate either way as to how reality actually is. Some advocates of this approach even go so far as to claim that Hume himself was completely amenable to the idea that there may be objective causal relations; we just don’t have access to them. Another interpretive view is that of skeptical realism.15 The skeptical realist thinks there really are real causal powers. Advocates of this view take it that what Hume taught us was that whilst we can speak correctly of causal relations—correctly in the sense that our causal talk picks out these relations—we cannot know anything about the nature of the relations. What generates the impression in us of causal power is, in fact, the felt transition in our minds from cause to effect. In short, although there are real causal powers in the world, our ideas of them are defective. Finally, there is the projectivist interpretation.16 According to this view, our causal talk simply expresses our inferential behavior. It is completely silent on whether causation is contiguity, there are real causal powers, and so on. This approach differs from the traditional, most significantly, on the nature of the semantics of our causal talk.17 Hume interpretation is no straightforward matter. Although it will matter, in some sense of matter, which Hume the contemporary friends of emptiness wish to endorse, there is no escaping the relationship between the causal relation and the causal relata that underwrites the regularity account. Whether we take Hume to be espousing a thesis on which necessary connections are a matter of the mind involving relata that are conceptual constructs, or a matter of the world involving real events, the relationship between the necessary connections and the relata between which they fail to obtain, in the way that is relevant here, will be the same on all accounts. Interpretive worries aside, it is important to recognize, and attend to, the principle that we can glean from—that underwrites—the Regularity Account of Causation. This principle we can call Hume’s Dictum. Cursorily put, Hume’s Dictum says that there are no necessary connections between distinct existents. Whatever additional story we might tell—a story which Hume does tell—as to how it is that we have ideas about necessary



On Being Humean about the Emptiness of Causation

79

connections proceeds upon the more basic assumption that is expressed by the dictum. In Hume’s own words (bk. 1, pt. 3, sec. VI): There is no object, which implies the existence of any other if we consider these objects in themselves

3.3  Necessary Connections, Distinct Existents, and Emptiness In the contemporary analytic literature, Hume’s Dictum is both made more technically precise and applied to a variety of relations and phenomena of different categories: it is extended beyond the causal context. Precisifications of the dictum, we shall come to in a moment. Appeal is made to the regularity account of causation by the friend of emptiness in order to exploit its underlying commitment to the idea that there are no necessary connections between cause and effect: whatever necessity is involved in our understanding of causation, it is a product of our own habits of mind. The causal relation is empty in virtue of being dependent upon us. Hume arrived at his conclusion by noting that no matter how intensely we contemplate cause and effect, nowhere in them can we find any indication of the existence of, or connection to, the other. Necessary (causal) connections are something that we impute onto the world. Garfield (2002, 29) has suggested that “we can see Nāgārjuna as urging that even distinguishing clearly between explanans and explanandum as distinct entities, with the former containing potentially what the latter has actually, is problematic.” Yet, as is evidenced by the quotes above, exactly what Hume asks us to do is to consider objects in themselves that are distinct from each other. Shouldn’t the call to consider things in themselves—or the idea that there are things in themselves to be considered—register with concern to the friend of emptiness? Isn’t this just exactly what Nāgārjuna is aiming to deny? Moreover, what if the relationship between the denial of necessary connections between things in themselves entails the existence of, or the idea of the existence of, things in themselves? After all, Hume seems to be telling us something about the nature of necessary connections involving assumptions about the relata between which they would hold.

80

Ricki Bl iss

In this section, I set out some of the connections between the notions invoked in contemporary versions of the dictum. I choose to engage with contemporary formulations of the dictum for the reason that they employ philosophically substantive concepts whose analyses are available to us. It is too easy, where vague notions are employed, to make of a thesis what we will.18 I begin by introducing and explaining the precisified version of the dictum I will be employing. I then discuss what appear to be already some consequences of running together a regularity account of causation underwritten by this version of the dictum with the emptiness of causation. These thoughts aside, along with the brush clearing carried out in the previous section, will allow us to turn to a more careful consideration of the expressions in operation within the dictum. First, I will discuss versions of the dictum that are analytic; and what these mean for the proposed regularity understanding of the emptiness of causation. This will be followed by a discussion of several synthetic versions of the dictum, and their consequences. I begin by introducing Hume’s Dictum: Hume’s Dictum: There are no metaphysically necessary connections between wholly distinct, intrinsically typed existents.19 In contemporary analytic formulations of the dictum, the necessary connections under issue are taken to hold of metaphysical necessity. In every possible world in which one, or both—we’ll come to this in a moment—of the relata exist, they are connected by some specified relation. Next, the notion of an “object in itself” is recast in terms of intrinsic typing. This notion of intrinsic typing is, in its turn, cashed out in terms of intrinsic properties. Roughly, then, to speak of an object in itself is to speak of an object as the thing that it is independently of anything else. Intrinsic properties are the properties a thing has as an object in itself. Finally, the notion of “any other object” at work in the original version of the dictum is expressed in terms of objects being wholly distinct from one another. Each of these modifications is supposed to render the dictum into something meaningful in the contemporary idiolect, whilst retaining the spirit of its original formulation. The suggestion that the relata are wholly distinct requires further consideration. As I will not take up this particular aspect of the notion again, I will address it immediately. The dictum could employ the term “distinct” instead of “wholly distinct”. The strengthening of “distinct” to “wholly distinct” is required to rule out a cluster of widely accepted relations as falling prey to the dictum.



On Being Humean about the Emptiness of Causation

81

Sets and their members, wholes and their parts, are distinct without being wholly distinct; they are metaphysically necessarily connected. Although numerically distinct, sets cannot exist without their members, for example, whereas the members of a set can exist without the set. Understanding “distinct” to mean “numerically distinct” or “distinct such that one can exist without the other” yields a version of the dictum that is far too strong; too strong because it counts sets and their members as being distinct and thus stipulates that there should not be any necessary connections between them.20 Non-Humeans and Humeans alike agree that there are a variety of cases in which necessary connections do hold between existents that are distinct from one another on several important senses of distinct. I suggest that we also preserve the necessary connections that obtain between distinct, but not wholly distinct phenomena. My main reason for doing so is that, in the context of the present discussion, I prefer to leave these noncausal cases alone.21 And as Garfield suggests that we be Humeans only about causation, I suggest that we stick to these cases. I do suspect, however, that Garfield and company would wish to deny all the necessary connections between the distinct phenomena that I have committed to preserving. The consequences of this categorical denial of necessary connections is a discussion for elsewhere. Preliminaries aside, we can see that the dictum, so formulated, is ambiguous regarding all of its key terms. We will come to these as required. We can begin by considering the kinds of metaphysically necessary connections at issue. The denial of the obtaining of a metaphysically necessary connection between relata could be aimed at demonstrating one of two things: First, it could be aimed at showing that one or either entity can exist without the other. Second, it could be aimed at showing that both entities could exist without being connected to one another for some specified relation, R. In the first case, we have a denial of a metaphysically necessary existential connection, and in the second case, a denial of a metaphysically necessary relational connection.22 Hume appears to be interested in the existential connection.23 And discussions of emptiness seem to be very much focused upon the existence of phenomena. So, I will take it that the kinds of connections we are interested in are metaphysically necessary existential connections. Without saying anything further as to the definitions of the, as yet undefined, terms within the dictum, we are already in a position to say something significant as to the compatibility of a regularity account of causation with the doctrine of emptiness.

82

Ricki Bl iss

Hume’s dictum expresses a denial of the obtaining of metaphysically necessary existential connections between cause and effect. Standardly, this denial, in the case of causal existential connections, is commonly taken to entail the contingency of the relations under consideration. If in this world, the Fs cause the Gs, the contingency of the causal existential connection between the Fs and the Gs entails that there are worlds in which the Fs exist without the Gs, and worlds in which the Gs exist without the Fs. For any relation, however, that necessarily holds of necessity, Hume’s Dictum entails the denial of the obtaining of that relation altogether. Relations of existential dependence, it is commonly thought in the Western tradition, hold of metaphysical necessity if they hold at all. Indeed, Westerhoff’s definition of existential dependence has the necessity of the connection built into the definition. If the Gs depend upon the Fs, then in every world in which the Gs exist so do the Fs. One reason to suppose that relations of existential dependence hold of metaphysical necessity is that they are metaphysical relations; and the laws of metaphysics hold in every possible world. I suspect this has important connections to identity. If something is the thing it is in virtue of the things it depends upon, and identity is not contingent, then in every world in which it exists, so too will the things upon which it existentially depends. Another reason to suppose that relations of existential dependence hold of metaphysical necessity is because of their relationship to essences. It is quite common for philosophers to be of the view that the things upon which something existentially depends are the things upon which that thing is essentially such as to depend. I am essentially such as to be existentially dependent upon having kidneys and a heart, for example. In every world in which I exist so too must my kidneys and heart. Modal-existential analyses of essential properties tell us that a thing’s essential properties are just the properties it has to have, if it exists. Relations of existential dependence cannot be contingent where essences are involved. Returning to our discussion, if relations of existential dependence hold of necessity, and there are no necessary existential connections between wholly distinct existents, then there are no necessary existential connections between cause and effect. But if it is in virtue of existentially depending upon one another that cause and effect are empty, then cause and effect are necessarily existentially connected. We have a contradiction. To deny that cause and effect are necessarily existentially connected is to deny that they are empty. The principle underwriting the regularity account of causation renders it incompatible with the doctrine of emptiness.



On Being Humean about the Emptiness of Causation

83

Hume’s genius was (amongst other things) to suggest that causal relations are contingent instead of necessary. Perhaps, similarly, Hume’s Dictum could be used to show us that relations of existential dependence are contingent instead of necessary. This option will not be open to anyone who employs the notion of essential existential dependence. But perhaps this—the contingency of the connection—is exactly what Garfield and company would like to suggest. I prefer to steer clear of this possibility for the reason that it involves revising our understanding of the slew of dependence relations mentioned in connection with merely distinct phenomena—part–whole dependencies, the membership relation, and so on. Rather than insisting upon a revision of the standard (Western) view of existential dependence—according to which the relation holds of metaphysical necessity—I will speak instead simply of existential connections more generally; although, for the sake of clarity, I will continue to mention some of the consequences of the view under consideration for anyone who thinks that relations of necessary existential dependence obtain between cause and effect. We can say, then, that the upshot of the application of Hume’s Dictum is that the existential connections between cause and effect are contingent.24 With this possibility in place, we are able to move forward, for we have not been forced to deny that any relation of existential connection holds between cause and effect; therewith demonstrating the incompatibility of the theses under consideration. Nor have we yet found fault with Garfield’s proposed account of emptiness: Garfield does not invoke relations of existential dependence, and he eschews the notion of essence. Garfield’s account of emptiness does not involve relations of existential dependence that necessarily hold of necessity. Contained within the regularity account is reference to the terms “wholly distinct” and “intrinsically typed.” As things stand, these terms are ambiguous. Filling in various possible and, indeed, plausible definitions of the aforementioned terms will yield both analytic and synthetic versions of the dictum. We turn, now, to a discussion of these. I will argue that neither analytic nor synthetic versions of the dictum are compatible with the emptiness of causation.

3.3.1  Analytic Versions of the Dictum There are several somewhat obvious candidate definitions of the terms “wholly distinct” and “intrinsically typed.” Some of these definitions are

84

Ricki Bl iss

nonmodal:  lack of mereological overlap, lack of spatiotemporal overlap. And some of them are explicitly modal. I begin by considering an explicitly modal sense of “wholly distinct”. Two things are wholly distinct just in case either can exist without the other. This is just to say: Wholly distinct (modal): Two existents are wholly distinct just in case they are not necessarily existentially connected.25 Plugging this into the dictum gives us the following: Hume’s Dictum (modal):  There are no metaphysically necessary existential connections between intrinsically typed existents that are not necessarily existentially connected.26 Assuming that the necessity at issue is metaphysical, we can remove this specification from the dictum. Assuming also that the very phenomena under consideration are those which are intrinsically typed, then we can remove this specification as well. What remains is an analytic statement: Hume’s Dictum (modal)*:  There are no necessary existential connections between phenomena that are not necessarily existentially connected. Hume’s Dictum (modal) is analytic. To be wholly distinct from something is to be not necessarily existentially connected to that thing, in which case, it is necessarily true that where the relata are wholly distinct from one another they are not necessarily existentially connected. Whether the denial of necessary existential connection is taken as a denial of any existential connection whatever, or simply as an affirmation of its contingency, on this version of the dictum the distinctness of the relata is in virtue of their not being necessarily existentially connected. Furthermore, and importantly, this analytic version of the dictum entails that there are, and cannot fail to be, necessary existential connections between non–wholly distinct existents; that’s just what it is for the relata to fail to be wholly distinct from one another. If cause and effect are empty in virtue of being existentially dependent upon one another, not only does the application of Hume’s Dictum seem problematic because it denies the very connections by way of which their emptiness



On Being Humean about the Emptiness of Causation

85

is achieved. But the analyticity of the dictum guarantees the nonemptiness of the relata. An analytic version of Hume’s Dictum fails the friend of emptiness both in terms of its consequences for the relations involved and in terms of its consequences for the relata. As Hume’s Dictum underwrites the regularity account of causation, the regularity account of causation and the doctrine of emptiness as it pertains to causation are incompatible. This incompatibility seems straightforward where the emptiness of causation is understood in terms of the obtaining of necessary existential dependence relations. The incompatibility is perhaps less obvious where the emptiness of causation is understood in terms of merely contingent existential connections.27 Recall that if the necessary existential connections under consideration are not necessarily necessary, then they are contingent. Causal relata are contingently connected just in case either can exist without the other. Therefore the denial of a necessary connection between cause and effect entails that the relata are wholly distinct from one another whether they are not existentially connected at all or merely contingently connected. Why? Because in both cases either relatum can exist without the other. This is just what it is to fail to be necessarily existentially connected. How can cause and effect be such that they could exist independently of one another and yet be empty? As things stand, the suggestion that there are no necessary existential connections between cause and effect, where causation is empty, appears to be contradictory and incoherent. If causation is empty in virtue of cause and effect being existentially dependent upon one another, then where there are no necessary connections between wholly distinct intrinsically typed existents, there are no relations of existential dependence between cause and effect. But if this is the case, then cause and effect aren’t empty. Moreover, where cause and effect are wholly distinct in virtue of not being necessarily existentially connected, they are necessarily nonempty. An analytic rendering of the dictum ensures the nonemptiness of the relatum—even where we can secure the emptiness of the relation. Absent the necessary existential connections and the proposed accounts of emptiness fall into incoherence. It is worth persevering, for there is an additional, as yet undefined, term within the dictum. The expression “intrinsically typed,” recall, is introduced into contemporary formulations of the dictum to precisify Hume’s notion of “a thing in itself”. Intrinsic typing is itself cashed out in terms of intrinsic properties. So, how we understand the notion of intrinsic typing

86

Ricki Bl iss

in the dictum will turn out to be dependent upon how we understand the notion of an intrinsic property. Discussions of intrinsic properties often begin with the reader being told that they have an intuitive understanding of which properties are intrinsic. We are told that an intrinsic property is a property a thing has independently of what is going on outside of itself; or a property a thing would have were it to exist alone in a world, for example. Accounts of intrinsic properties typically invoke the notion of independence. Not complete and utter independence—such that the only things that possess intrinsic properties are simples—but the sort of independence that some unified object has from everything outside of itself. Combinatorial accounts of intrinsic properties capture the notion of an intrinsic property in terms of the ability of some property bearer to exist alone in a world. We can put this as follows: Combinatorial intrinsicness: A property F is intrinsic only if it is possible that an entity x having F exist unaccompanied.28 This account of intrinsic properties is understood in terms of the notion of free modal recombinability. Something being such that it can exist unaccompanied just means that it is such as to not be necessarily existentially connected to anything else.29 Plugging this into Hume’s Dictum yields: Hume’s Dictum (combinatorial): There are no metaphysically necessary existential connections between wholly distinct entities that are not necessarily existentially connected.30 As with the previous case, if we assume that the necessity under question is metaphysical necessity, we can strip off that specification. We can also assume that what are under consideration are necessary connections between wholly distinct existents, so we can get rid of that specification as well, giving us: Hume’s Dictum (combinatorial)*: There are no necessary existential connections between entities that are not necessarily existentially connected. Once again, we have an analytic version of the dictum, the problems with which we are already familiar. Cause and effect being such that we are able



On Being Humean about the Emptiness of Causation

87

to consider them in and of themselves is just what it is for cause and effect to be intrinsically typed. And for cause and effect to fail to be necessarily existentially connected is exactly what is required such that they are intrinsically typed. Where there are no existential connections whatever between cause and effect, or merely contingent existential connections between cause and effect, they are wholly distinct from one another and have a way of being internal to themselves. Once again, either cause and effect fail to be empty because there are no relations of existential dependence that obtain between them; or they fail to be empty because the failure of necessary existential connections obtaining between them entails that they are wholly distinct and intrinsically typed. Moreover, as standard analyses make recourse to the notion of independence in some from or other, most, if not all, definitions of intrinsic typing will yield analytic versions of the dictum. I leave off discussion of analytic versions of the dictum. I have argued that they entail the nonemptiness of the causal relata, and therefore, that the regularity account of causation is incompatible with its emptiness. I turn to a discussion of synthetic versions of the dictum.

3.3.2  Synthetic Versions of the Dictum We can make relatively quick work of synthetic versions of the dictum. It is important, however, to discuss some possible synthetic renderings, for one might wonder if employing a nonanalytic version of the dictum might dissolve some, if not all, of the aforementioned difficulties. My discussion here is not exhaustive. Our first candidate synthetic version of the dictum employs a spatiotemporal sense of “wholly distinct”: Wholly distinct (spatiotemporal): Existents are wholly distinct just in case they do not spatiotemporally overlap.31 Plugging this definition of “wholly distinct” into the dictum we get: Hume’s Dictum (spatiotemporal): There are no necessary existential connections between spatiotemporally distinct, intrinsically typed existents.32 A second candidate synthetic version of the dictum employs a mereological sense of “wholly distinct.”

88

Ricki Bl iss

Wholly distinct (mereological): Existents are wholly distinct just in case they do not share a mereological part.33 Plugging this definition of “wholly distinct” into the dictum we get: Hume’s Dictum (mereological):  There are no necessary existential connections between mereologically distinct, intrinsically typed existents.34 These synthetic versions of the dictum have the advantage that the causal relata are not wholly distinct in virtue of not being necessarily existentially connected. For synthetic versions of the dictum, the denial of the necessary connection between cause and effect is not an affirmation of the nonemptiness of the relata. These versions of the dictum do not, however, have the regularity account of causation out of the woods. The first, and most important, reason for this is that Nāgārjuna wants to argue that cause and effect are neither spatiotemporally distinct, nor mereologically distinct: he also denies their converse.35 If the emptiness of causation entails that cause and effect are neither spatiotemporally nor mereologically distinct, then on these versions of the dictum there are no relata between which the obtaining of necessary existential connections can be denied. The second reason for thinking synthetic versions of the dictum are unacceptable is that they still have recourse to the notion of cause and effect as intrinsically typed. And recall that this notion of intrinsic typing is introduced as a technical way of cashing out the call to consider things as objects in themselves. As all senses of “intrinsically typed” will render the dictum analytic, all permutations of the dictum will carry with them the inbuilt assumption that cause and effect are such that either one can exist without the other. Both synthetic and analytic versions of the dictum appear to render the regularity account of causation they underwrite hopelessly incompatible with the doctrine of emptiness.

3.4  Final Remarks The advocate of the compatibility of the Humean regularity account of causation with the emptiness of causation wants to draw upon the Humean insight that any necessary connection that obtains between cause and



On Being Humean about the Emptiness of Causation

89

effect is an artifact of the mind. Causation is empty, thinks the friend of emptiness, because how we carve up the world causally is inextricably tied up in our interests, projects, concepts, and explanatory activity. There are no causal powers. If this is the point that is being made by reference to the regularity account, it might appear as though my critique has entirely missed its target. Or rather, it has hit a target that was never within our scope in the first place. I now offer several remarks in defense of the foregoing discussion. First, I would like to deal with what I take to be the most substantive objection to the foregoing analysis: no Mādhyamika will accept an ontology of wholly distinct, intrinsically typed causal relata. The version(s) of Hume’s Dictum that I  employ here is not one that the friend of emptiness would be willing to endorse. Problem solved. There are several points I would like to make in response to this claim. The first, and least serious, is that I am not sure how far we can deviate from the letter, or spirit, of Humeanism and still consider ourselves Humeans. No doubt there is room to move, but it seems right to think that there is some point where, although adhering to some aspect or other of Humeanism, we simply stop being Humeans. We wouldn’t call a position according to which there are no instantiation relations between wholly distinct intrinsically typed existents Humeanism just because the position posits wholly distinct, intrinsically typed existents. And whilst I can imagine several names we might use to refer to someone who thinks there are no necessary connections between the number 7 and everything that is green, I doubt “Humean” would be amongst them. My point just is that if the friend of emptiness wants a radical redefinition of Hume’s Dictum, it is not clear why we should call the position to which they are referring Humeanism. But I don’t think the issues here are merely terminological. Relations are either necessary or they are contingent. The friend of emptiness tells us that the connection between cause and effect is not a necessary connection. This means the connection must be contingent. But as I have argued, the contingency of the connection entails that the relata are able to exist independently of one another. This just means that where a is existentially connected to something that falls under the property F but the connection is contingent, we can have a worlds that are F worlds; and a worlds that are G worlds. But causal relata being such that they are able to exist independently of one another seems to be exactly what Nāgārjuna’s analysis is aiming to deny. Whether the denial of a necessary

90

Ricki Bl iss

connection entails the denial of the existential connection altogether, or its mere contingency, it is not clear how this does not yield the independence of cause and effect from one another. Let’s consider a possible Madhyamaka version of Hume’s Dictum: Hume’s Dictum (conventional): There are no metaphysically necessary existential connections between conventional existents.36 This version of the dictum contains an undefined term, “conventional existents.”37 To be conventionally existent is to be empty. And recall that to be empty is to be causally, mereologically, and conceptually dependent. It is to be lacking in svābhāva. That which is empty has no self-being or being from its own side. In keeping with Westerhoff’s suggestion, we have been understanding the emptiness of causation (at least partially) in terms of existential dependence:  the emptiness of causation entails that cause and effect are existentially dependent upon one another. The notion of self-being is also associated with essence possession and intrinsic nature. Owing to the difficulties associated with capturing this sense of svābhāva, and the context created by the present discussion, I’m going to define “conventional existent” as follows: Conventional existent: Existents are conventional just in case they are existentially connected and not intrinsically typed. Plugging this definition into the proposed dictum yields: Hume’s Dictum (conventional): There are no metaphysically necessary existential connections between nonintrinsically typed existents that are existentially connected. Let’s assume that the phenomena we are considering are not intrinsically typed allowing us to remove this specification. Let’s also assume, in the first instance, that relations of existential dependence necessarily hold of necessity. This leaves us with a flagrantly contradictory version of the dictum: Hume’s Dictum (conventional)*:  There are no existential connections between existents that are existentially connected.



On Being Humean about the Emptiness of Causation

91

If the kinds of existential connections here do not hold of necessity, then they are merely contingent. As the friend of emptiness wants to deny necessary existential connections, let’s also assume that between conventional existents the connections are contingent: Hume’s Dictum (conventional)**:  There are contingent existential connections between existents that are contingently existentially connected. On the face of it, this version of the dictum looks much better, as it does not yield a contradiction. It is, instead, analytic. This being the case, we could take it to entail that for causation to be conventional—empty—is for the causal connection to be contingent. But once we consider that these existents are not intrinsically typed, the dictum is once again in trouble. Recall from the previous discussion that the notion of intrinsic typing (introduced to precisify the Humean notion of “objects in themselves”) was cashed out in terms of intrinsic property possession analyzed in terms of aloneness at a world. According to combinatorial accounts of intrinsicality, a property is intrinsic just in case it can be had by its bearer unaccompanied. Which is just to say that intrinsic typing is understood in terms of existential independence. I suggest we reverse engineer this case to give us a plausible sense of not being intrinsically typed. Something is not intrinsically typed if it cannot exist unaccompanied. This is just to say that something is not intrinsically typed—cannot be considered as a thing in itself—just in case it is metaphysically necessarily existentially connected to something else. The relation of existential dependence must be metaphysically necessary to ensure that in every world in which something exists it is accompanied. Plugging this into the dictum gives us: Hume’s Dictum (conventional)***: There are no metaphysically necessary existential connections between existents that are metaphysically necessarily existentially connected. I have cast this version of the dictum in terms of the original denial of metaphysically necessary existential connections, rather than in terms of the assertion of their contingency. A version of the dictum that replaces the original reference to “metaphysically necessary existential connections” with “contingent existential connections” is still unacceptable, as

92

Ricki Bl iss

it involves reference to necessary connections between existents nonetheless. Our rejigged versions of the dictum do no better than the versions in the previous sections. One might wish to argue that the reason for this is that the versions of the dictum I have supplied here rely upon definitions that are infected with the same problems present in the preceding versions, namely, problems related to the nature of existential connections and its connection to emptiness. To an extent this is true. Recall that the modal definition of “distinctness” was originally selected as it allowed us to preserve a large number of cases of necessary connections between distinct but not wholly distinct existents. Recall, also, that I made this choice in order to isolate the causal cases for consideration. Perhaps many of the difficulties could be avoided by sacrificing all necessary existential connections and employing nonmodal definitions of “distinctness” and “dependence.” That said, at the very least, what this discussion ought to draw to our attention is the role denials of necessary existential connections play, in the Western tradition, in establishing varying kinds of independence. Just as various kinds of dependence are secured by way of necessary existential connections. Indeed, Westerhoff not only exploits the notion of existential dependence but builds the necessity of the connection into its definition. If the analysis provided here turns out to be incorrect, what it would seem to bring into focus is the need to clarify the relationships between existential dependence, necessary connections, emptiness, causal regularity, and independence. For, as things stand, it appears as though Humeanism about necessary connections secures the independence of the causal relata whose emptiness we are aiming to ensure. It may just be the case that the friend of emptiness means something different by the terms they employ. But it escapes me how we can cash out the notion of emptiness in terms of existential dependence relations according to which, if some a depends on F, if a exists then necessarily F exists, but deny either that these relations obtain at all, or that they involve a necessary connection. It is not at all clear how, if the denial of necessary existential connections between cause and effect entails their distinctness, we can still say that they are empty. And it is not at all clear how we can hold a regularity view of causation but then deny that there are distinct causal relata. To the final point. In section 3.2, I presented brief outlines of the several different interpretive frameworks in terms of which Hume’s insights are understood. On one of those frameworks, Hume is thought to be making a point about reality—there really is nothing more to causation than constant conjunction, regularity, and contiguity. On another approach, Hume



On Being Humean about the Emptiness of Causation

93

is thought to be making a point regarding our epistemic access to the world—we have no experience with actual necessary connections. If Hume’s Dictum is taken to be making a statement regarding reality—a robust metaphysical claim—then, as I have shown, the upshot of the dictum is that the world is populated by a series of wholly distinct, intrinsically typed existents. If, on the other hand, we take Hume to be merely making a statement about our epistemic access to the world and the psychological nature of necessary connections, the dictum nonetheless yields the result that cause and effect are wholly distinct from one another. Regardless of the ontological status of the relata involved in our account of causation, the application of Hume’s Dictum forces consequences both as regards the relation and as regards the relata. This being the case renders the application of Hume’s Dictum by the friend of emptiness thoroughly unacceptable. If recourse to Hume’s Dictum is supposed to secure the emptiness of the causal relation by pulling it into the mind—into the realm of the conventional—it seems to pull with it wholly distinct causal relata. Even if this does not commit us to a metaphysical thesis regarding the population of an external reality with independent existents, it surely commits us to conceiving of distinct causal relata. Not only does Hume urge us to consider things in themselves, the application of the dictum returns as cognitive artifacts causes and effects that are wholly distinct from one another. Such an error, in this context, is the gravest of all. Although Garfield does not employ relations of existential dependence in his account of emptiness, and he eschews the notion of an essence, his account is still lumbered by the consequences for the causal relata of the application of the dictum. A consequence it is not at all clear to me how to escape. Although my investigation has not been exhaustive, I  have demonstrated that on a number of very plausible versions of Hume’s Dictum—including Madhyamaka versions—in combination with assumptions held by the friends of the doctrine of emptiness, the regularity account of causation is incompatible with the Madhyamaka doctrine of emptiness.

Notes 1. Thank you to Yasuo Deguchi, Masaki Ichinose, Graham Priest, and Jan Westerhoff for helpful comments and discussion. 2. The discussion on Hume’s Dictum in these terms is heavily indebted to the work of Jessica Wilson in Wilson (2010).

94

Ricki Bl iss

3. This being the case would seem to entail that existents are dependent for their existence upon their essential properties—if they have them. 4. Westerhoff and Garfield might be seen to disagree on whether the MMK is intended as an attack on the existence of substances, or as an attack on the notion of an essence. More significantly, they might also be seen to disagree on how we should understand the doctrine of emptiness. Or perhaps the seeming differences are purely terminological, and thus not real differences at all. 5. Westerhoff (2009a, 94). 6. I am hedgy in my reference to relations of existential dependence here for the reason that what these relations are like turns out to be exactly what is at issue. 7. Westerhoff (2009a, 95–96). 8. Westerhoff (2009a, 96). Here Westerhoff himself references Nāgārjuna of the Ratnāvalī 1:49. 9. I  have not supplied an argument in defense of this claim regarding the concept-relative nature of causal fields. 10. Westerhoff (2009a, 96–97). Note also that the claim here is not that the individual constituents of the causal field are existentially dependent upon the effect, but that the causal field qua conceptual artifact is. 11. Garfield (2002, 29). 12. Although he doesn’t explicitly state anywhere that he endorses a Humean regularity account of causation, Westerhoff (personal correspondence) does agree with this diagnosis of the nature of the perceived necessary connection between cause and effect. 13. Advocates of the traditional view include Galen Strawson. 14. Advocates of this view include Don Garrett. 15. Wright (1999) endorses this approach. 16. One example of an advocate of this view is Blackburn and Simmons (1999). 17. Beebee (2007, 414–416). 18. This is not to say that I think the terms or concepts I invoke here are obviously the correct ones to use in an elucidation of emptiness, but we must start somewhere. And starting with concepts that we have a relatively good grip on will, at least, allow us to become clearer on what is at stake, what the consequences are and possibly even which concepts we shouldn’t try and employ. And where certain Western technical terms are advanced in the contemporary commentarial literature we must take them seriously. 19. Wilson (2010, 604). 20. See Wilson (2010, 600–603) for a more detailed discussion of these points. The kind of modal distinctness I refer to here, Wilson calls “weak modal distinctness.” When two things are weakly modally distinct from one another, one can exist without the other. Sets and their members, amongst other things, count as weakly modally distinct.



On Being Humean about the Emptiness of Causation

95

21. Siderits (2004) consider the relation of material constitution to be a kind of causal relation in keeping with Aristotle (see p. 399). Here I wish to avoid embarking upon an argument to this conclusion. Not because I disagree but because it is beyond the scope of this chapter. 22. Wilson (2010, 598–599). 23. Hume [1738] 1985, bk 1, pt 3, sc XIV. 24. Alternatively, one might accuse Westerhoff of equivocating importantly distinct senses of existential connection under the notion of existential dependence. Philosophers commonly refer to relations of existential dependence as “metaphysical” relations. Whereas the connection between cause and effect, whilst existential, is not normally subsumed under the notion of an existential dependence relation. Distinguishing between metaphysical relations of existential dependence and causal existential connections is the norm rather than the exception. 25. Wilson (2010, 607) refers to this definition as that of strong modal distinctness. 26. Wilson (2010, 60). 27. The definition of existential dependence offered by Westerhoff is generic. This means that a existentially depends upon something or other that falls under the property F. Between a and some b, which falls under the property F, the connection is contingent; a could just as well depend on some c that falls under F. But where a exists and existentially depends upon something that falls under the property F, necessarily something that falls under the property F exists (on Westerhoff’s definition). We can contrast this with a case in which the existential connection between a and something’s falling under the property F is contingent. In this case, if a merely contingently depends upon something’s falling under the property F, then something that falls under the property F needs not exist if a exists. In some worlds, a will depend upon something that falls under the properly F; in others it will depend upon something that falls under the property G; and so on. To assert the contingency of the existential connection between cause and effect, as understood here, is to endorse the latter view and not the former. We should not mistake the difference between a generic relation of existential dependence and a rigid one for the difference between a contingent relation of existential dependence and a necessary one. 28. Wilson (2010, 616). 29. Weatherson (2001, 366) defines an object as being lonely where “there are no wholly distinct contingent things in its world.” 30. Wilson (2010, 617). 31. Wilson (2010, 605). 32. Wilson (2010, 605). 33. Wilson (2010, 605). 34. Wilson (2010, 605).

96

Ricki Bl iss

35. For a discussion of these arguments, and the arguments against causation more generally, see Westerhoff (2009a, 92–93 and 99–109). 36. This version of the dictum was suggested by Jan Westerhoff. 37. Formulating a Madhyamaka version of the dictum works only if we have an ­independent understanding of emptiness and conventional existence. If recourse to Humeanism is supposed to help us understand the emptiness of phenomena, then it is question begging to formulate a version of the dictum employing the notion of conventional existence if we are attempting to understand it by way of the dictum in the first place.

4

Essence and Emptiness Roy W. Perrett

Madhyamaka Buddhism is famously centered on the doctrine of emptiness (śūnyatā), often glossed as the view that there are no essences. This chapter addresses two interrelated questions about that doctrine. First, is the Madhyamaka doctrine of essencelessness more plausibly to be regarded as a necessary or a contingent truth? Second, is the doctrine of essencelessness in contradiction with the views of those prominent Mādhyamikas who also claim that essencelessness is the essence of all things? Since the views of the Indian Mādhyamikas on some of the issues I shall be discussing are seriously underdetermined by the extant texts, I  shall often be freely moving between the “external history” of Madhyamaka philosophy (i.e., what was actually said by the Indian Mādhyamikas) and its “internal history” (roughly, in this case, what a modern Mādhyamika conversant with modern analytic philosophy might most plausibly say on the issues, where what is said has also to be compatible with the external history of Madhyamaka).1

4.1  Concrete Particulars Let us approach the issues I want to discuss by what might at first seem a somewhat circuitous route, beginning with a familiar topic in Western metaphysics: namely, the nature of concrete particulars.2 Notoriously, most Western metaphysicians have considered concrete particulars (the ordinary objects of our experience like tables and chairs, cats and dogs, and so on) as not, strictly speaking, elements of our most fundamental ontology.

98

Roy W. Perrett

Instead, ordinary objects are taken to be constructions out of ontologically more basic entities. Which entities? Two competing types of account have proved to be most popular: bundle theory and substratum theory. According to bundle theory, ordinary things are nothing but bundles or clusters of attributes. Some bundle theorists take the relevant attributes here to be (repeatable) properties; others take them to be property particulars (tropes). Either way, the crucial idea is that concrete particulars are all reducible to bundles of more basic entities (properties or property-instances). According to substratum theory, ordinary things are not to be identified with their attributes in this fashion. Instead, we need to distinguish between a thing’s attributes and the thing that has those attributes. A concrete particular is, then, a whole made up of the various properties we associate with the particular, together with an independent underlying subject or substratum that exemplifies those properties. What is the modal status of the central claims associated with these rival theories? Well, in the first place, the theories are presumably all theories about the categorical structure of the world and hence if true, they are supposed to be necessarily true: surely according to the bundle or substratum theories concrete particulars do not just happen to be bundles of attributes or wholes composed of a substratum plus attributes. We can say even more about the modal implications of the theories. Take first bundle theory. If concrete particulars are just bundles of attributes, then all of a concrete particular’s attributes are essential or necessary to it: that is, if any of those attributes did not enter into the composition of the particular, then that particular would not exist. (This is because if a concrete particular is just a set of attributes then its identity conditions are the same as a set’s, and all of a set’s members are essential to the existence of that set.) Substratum theory, in contrast, denies that all of a concrete particular’s attributes are essential to its existence because the real possessors of the attributes are held to be the underlying substrata and these are bare of attributes. An object’s identity is thus supposedly independent of any of its attributes: it is a contingent matter whether an object has any particular attribute. The worry with this move, however, is that the notion of a genuinely bare particular that is the substratum of a concrete particular seems problematic. Not only is there an epistemic concern about the empirical legitimacy of the concept of a bare substratum, but there seems to be a



99

Essence and Emptiness

contradiction involved in the supposition that a substratum (i.e., the thing that possesses attributes) in fact possesses no attributes. A more plausible version of substratum theory says that substrata are not really bare of all attributes, but just bare of all essential attributes. Is this supposed to be a necessary or a contingent truth? Presumably it has to be a necessary truth about the categorical structure of the world. But then it would seem that the bare substrata have no essential attributes essentially. The upshot of this brief survey (diagrammed in figure 4.1) is this. All of the reviewed positions about the composition of concrete particulars plausibly involve us in some sort of commitment to essences, given a familiar type of account of essence. And that account is roughly this: [E*]: O is essentially F ≡ Necessarily (If O exists, then O is F)

Concrete particulars

Substratum

Bundle

Bare? Bare of essential properties

Necessary

Contingent?

Essentially bare of essential properties?

Figure 4.1  The Structure of Concrete Particulars

Properties

Tropes

Essential

Essential

100

Roy W. Perrett

4.2  The Indian Context How does all this relate to Indian philosophy? In the first place, Indian metaphysicians also developed various theories of concrete particulars that paralleled the bundle and substratum theories of Western metaphysics. Thus, Buddhist philosophers in the various Abhidharma traditions all espoused some type of bundle theory according to which the concrete particulars of our ordinary experience are reducible to bundles of (momentary) impartite simples called dharmas. Moreover, these dharmas (of which there were competing classifications) are arguably all some variety of tropes.3 In contrast, Hindu and Jaina philosophers tended to favor some variety of substratum theory, though there was keen intramural dispute about the nature of the most plausible version of this. Nyāya-Vaiśeṣika is a good example here, offering a richly developed ontology that includes properties, tropes, and particulars.4 For Naiyāyikas an ordinary concrete particular is a construction out of these more ontologically basic types of entities: a substance that is an underlying substratum wherein properties and qualities inhere. What is the modal status of the central claims associated with these rival Indian theories? From the point of view of the “external history” of Indian philosophy (i.e., what was actually said by the classical Indian philosophers), the question is anachronistic. Modality was not an issue explicitly theorized by the Indian philosophers and there was no explicit distinction made between necessary and contingent truths. Indian logicians discussing nonexistent entities, for instance, never distinguished between logically impossible entities (like the son of a barren woman) and merely unexemplified entities (like a hare’s horn). This is not to say, of course, there was no kind of modal discourse in India: modality can be expressed in Sanskrit through the use of the gerundive and the Indian logicians discussed extensively a form of counterfactual reasoning called tarka, the employment of which implicitly requires an appreciation of a connection between two unrealized (logical) possibilities. But it has to be conceded that the historical Indian texts underdetermine any explicit answer to the question of whether the Indian versions of bundle theory and substratum theory are supposed to be necessarily true. Correspondingly, there is also no explicit answer to be found there as to whether the Indian theorists are committed to particulars having all their attributes essentially or substrata being essentially bare of essential properties.



Essence and Emptiness

101

What might we say, however, from the point of view of “internal history”? Well, surely much the same as we said earlier about the Western versions of bundle theory and substratum theory. First, the Indian theories are also all theories about the categorical structure of the world, and hence, if true, they too are presumably necessarily true: concrete particulars do not just happen to be bundles of attributes or wholes composed of a substratum plus attributes. Second, for much the same reasons as their Western counterparts, the Indian theorists are apparently also logically committed either to all of a concrete particular’s attributes being essential to it, or to a substratum being essentially bare of essential properties. The upshot is, then, that in India as in the West both types of position about the composition of concrete particulars plausibly involve us in some sort of commitment to essences. The apparent exception to this general Indian essentialism is, of course, Madhyamaka Buddhism, which is often described as holding the view that there are no essences. But this description needs much closer examination. Uncontroversially, Mādhyamika Buddhists are committed to the doctrine of emptiness (śūnyatā), that is, the doctrine that all entities (bhāva) are empty (śūnya). Three questions about this doctrine of emptiness immediately present themselves: (1) Empty of what? (2) Who is committed to entities that are not empty? (3) Why should we believe all entities are empty? To which the classical Prāsaṅgika-Mādhyamika answers are: (1*) Empty of svabhāva or essence. (2*) Everyone except the Prāsaṅgika-Mādhyamikas. (3*) Because supposing that there are essences entails contradictions (the prasaṅga method). A further word or two about each of these answers is in place. First, the notion of svabhāva. For the Ābhidharmikas who are Nāgārjuna’s primary targets in the Mūlamadhyamakakārikā, the fundamental constituents of the world are impartite dharmas, and a dharma is definable as what has intrinsic nature (svabhāva). Only dharmas are ultimately real (paramārthasat), but wholes constructed out of dharmas are conventionally real (saṃvṛtisat). Since for the Ābhidharmikas a dharma has but a single nature and has

102

Roy W. Perrett

that nature intrinsically, then a dharma could not lack that nature and still exist: that is, a dharma possesses its svabhāva essentially. Hence to deny, as Mādhyamikas do, that anything has svabhāva is tantamount to denying that anything has an essence. This is why the doctrine of emptiness is often glossed as the view that there are no essences.5 Second, we have the question of who is actually committed to the existence of entities that are not empty. Some modern scholars are skeptical that any of the Indian schools were really so committed.6 However, Nāgārjuna himself clearly holds that at least the Ābhidharmikas and Nyāya are so committed, and later Indian Mādhyamikas like Āryadeva and Candrakīrti add Vaiśeṣika, Sāṃkhya, Yogācara, and even Svātantrika-Madhyamaka to the list of culprits. The subsequent Tibetan Madhyamaka tradition tends to generalize the charge still further so as to include as culprits not only all philosophical schools except Prāsaṅgika-Madhyamaka, but all unenlightened beings. (The earlier arguments of this chapter, of course, lend further support to Madhyamaka claims by showing that in both India and the West the most popular types of position about the composition of concrete particulars apparently involve us in some sort of commitment to essences.) Finally, the favored Prāsaṅgika-Mādhyamika method for establishing the truth of emptiness is by the use of prasaṅga or reductio ad absurdum: that is, we assume for the sake of argument the truth of some claim of our opponent’s and then logically derive an absurd consequence from that assumption, thus showing the assumption to be untenable. As already mentioned, the Madhyamaka doctrine of emptiness is often glossed as the view that there are no essences. But this claim seems in tension with what seems to be the right answer to the modal question: “Is the doctrine of essencelessness more plausibly to be regarded as a necessary or a contingent truth?” While this modal question is certainly anachronistic from the point of view of the external history of Indian Madhyamaka, the most plausible answer from the point of view of the internal history seems to be that the doctrine of emptiness is a necessary truth. This is so for at least two reasons. First, like the rival views about the categorical structure of the world we have already considered, the Madhyamaka doctrine of emptiness is supposed to be ultimately true and hence presumably also if true, necessarily true: surely things do not just happen to be empty in the way that my car happens to be painted silver. Rather, to be is to be empty.7 Second, the favorite Madhyamaka method involves the use of prasaṅga or reductio ad absurdum arguments. Now an Indian prasaṅga argument,



Essence and Emptiness

103

like a Western reductio, can come in strong or weak forms. The strong form is a method of argument whereby a premise is reduced to absurdity by demonstrating that it entails a contradiction. The weak form is a method of argument where what is entailed is merely a falsehood, rather than a contradiction. But if any of the Madhyamaka prasaṅga arguments do establish that the assumption of svabhāva actually entails a contradiction, then necessarily everything is empty and nothing has an essence. Suppose all this is right and that the most plausible construal of the doctrine of emptiness is indeed that it is necessarily true that all entities are empty of essence. Assume further the already mentioned modal account of essence [E*]. Then it follows that emptiness (i.e., essencelessness) is the essence of all entities. Or in other words, all things are essentially essenceless! But that is a contradiction.

4.3  Essential Essencelessness? The first point I  want to make about this apparent contradiction (i.e., emptiness as essential essencelessness) is that I take it to be distinct from the ontological paradox that Jay Garfield and Graham Priest have named “Nāgārjuna’s Paradox,” and which they state as follows: All phenomena, Nāgārjuna argues, are empty, and so ultimately have no nature. But emptiness is, therefore, the ultimate nature of things. So they both have and lack an ultimate nature.8 This contradiction is, of course, just the conjunction of two claims arguably attributable to Nāgārjuna:  namely, “All things have no ultimate nature” and “Emptiness is the ultimate nature of all things.” The modal status of these two claims plays no crucial part, however, in the nature of the paradox. In contrast, the apparent contradiction of essential essencelessness that I  have unearthed is about the simultaneous absence and presence of essence, construed as a modal notion. Moreover, the arguments I presented for a Madhyamaka commitment to essential essencelessness both crucially involved the modal status of the doctrine of emptiness and the modal account of essence. These arguments are quite different from the arguments Garfield and Priest present in support of the ontological paradox they attribute to Nāgārjuna and which they take to be a true contradiction or dialetheia.

104

Roy W. Perrett

I do agree with Garfield and Priest, however, that it will not do to dismiss out of hand the suggestion that Madhyamaka is committed to a significant but contradictory doctrine of emptiness simply on the grounds that contradictions supposedly entail everything (the doctrine the Western medievals called ex contradictione sequitur quodlibet). Here is an elegant statement of an argument for that very doctrine by the thirteenth-century philosopher Alexander Neckam: I am amazed also at those criticising the claim that from the impossible in itself anything whatever follows. This may be established in many ways, but few will show more clearly. Is it not the case that if Socrates is a man and Socrates is not a man, then Socrates is a man? But if Socrates is a man, Socrates is a man or a stone. So if Socrates is a man and Socrates is not a man, Socrates is a man or a stone. But if Socrates is a man and Socrates is not a man, Socrates is not a man. So if Socrates is a man and Socrates is not a man, Socrates is a stone. By a similar deduction, it may be proved that if Socrates is a man and Socrates is not a man, Socrates is a goat, and so on for any other thing, such as a rose, a lily and so on. Don’t you therefore see that in this way from this impossibility, that Socrates is a man and Socrates is not a man, anything follows?9 The logical structure of the argument is clear enough: (1) p & ~p [assumption] (2) p [1, simplification] (3) p ˅ q [2, addition] (4) ~p [1, simplification] (5) q [3, 4, disjunctive syllogism] But once set out in this fashion it is also clear that the argument can be resisted formally by denying the validity of disjunctive syllogism, or by adopting a variety of paraconsistent logics on offer in which the inference from p & ~p to an arbitrary conclusion q is not valid. Moreover, there are arguably a number of independent reasons for doubting whether contradictions entail everything.10 Nor is it clear that it will help to appeal to passages in Madhyamaka texts which apparently affirm the law of noncontradiction.11 I have in mind passages like these:



Essence and Emptiness

105

(1) For how could the real and the nonreal, being mutually contradictory, occur in the same thing? (parasparaviruddhaṃ hi sac cāsac caikataḥ kutaḥ). (Nāgārjuna, Mūlamadhyamakakārikā 8.7) (2) Since being and nonbeing are mutually contradictory, they cannot both exist in the same place in nirvāṇa (bhāvābhāvayor api parasparaviruddhayor ekatra nirvāṇe nāsti sambhava iti). (Candrakīrti, Prasannapadā 532 [on MMK 25.14])

In fact, Garfield and Priest freely admit that Nāgārjuna is committed to the law of noncontradiction in the domain of conventional truth.12 They hold, however, that “Nāgārjuna’s Paradox” is a profound limit contradiction of the sort that philosophers are driven to in charting the limits of thought, precisely because such limits are themselves contradictory. It is thus, they claim, a very special creature: a true contradiction or dialetheia. (For the purposes of this chapter I  remain officially agnostic about whether Mādhyamikas are really committed to such a dialetheic paradox.) Of course, nobody wants to claim that every apparent paradox is a real dialetheia. Indeed, the Madhyamaka commitment to the law of noncontradiction in the domain of conventional truth should remind us that most contradictions are simply false. But if most contradictions are false, then it is presumably desirable to avoid contradictions if trying to increase the probability of holding only true beliefs. Is there any way for a modern Mādhyamika to avoid commitment to the apparent contradiction of essential essencelessness I have argued for?

4.4  Two Senses of “Essence”? It has to be admitted right away that this task is not made any easier by some of the explicit utterances of certain prominent later Indian Mādhyamikas. Consider, for instance, this striking passage from the Svātantrika-Mādhyamika philosopher Bhāviveka (sixth century ce): Our position is that the essence [ngo bo nyid  =  svabhāva] is emptiness, since that is the essence of existents. (Tarkajvālā [on Madhyamakahṛdayakārikā 3.26])13 Nor can this apparent contradiction be brushed aside as merely some kind of Svātantrika confusion. Consider also this famous passage from

106

Roy W. Perrett

the paradigmatic Prasaṅgika-Mādhyamika, the great Candrakīrti (seventh century): That very thing which is called the nature of existents is their own form. Then what is that nature of existents? The essence [svabhāva] of existents. What is this essence? Nature. And what is this nature? Emptiness. What is emptiness? Essencelessness [naiḥsvabhāvāvyam]. (Prasannapadā 264–265 [on Mūlamadhyamakakārikā 15.2c–d])14 Such passages look disturbingly like explicit endorsements of a paradoxical doctrine of the essential essencelessness of entities! There is, however, a different passage from Candrakīrti that is interesting to consider here. It is from the Yuktiṣaṣṭikāvṛtti, a commentary on Nāgārjuna’s Yuktiṣaṣṭikā, and the text is now extant only in a Tibetan translation from the lost Sanskrit original: The essence of things is their not having any essence (dngos po’i rang bzhin ni ngo bo nyid med pa yin).15 The Tibetan here probably translates the original Sanskrit locution bhāvasvabhāvo hy asvabhāvaḥ, an allusion to a passage in the Aṣṭasāshasrikā-prajñāpāramitā that is overtly paradoxical in the manner of much Perfection of Wisdom literature. But the Tibetan translators seem to have deliberately refrained from preserving such an apparent oxymoron, preferring instead to attempt a distinction between the essence (ngo bo nyid) that entities do not have and the essence (rang bzhin) that they do have.16 Thus the essence (rang bzhin) of things is their not having any essence (ngo bo nyid). To be clear:  I  am not suggesting that any of this provides an overwhelming philological argument for there being two senses of “essence” in the original text. Indeed, as already mentioned, there is no such distinction marked in what was probably the original Sanskrit locution; and in other contexts the Tibetan terms rang bzhin and ngo bo nyid are typically treated as synonyms. But it does seem worth exploring a little further the suggestion that some such disambiguation would dissolve the apparent paradoxicality of the fact that at least some Mādhyamikas declare entities to be essentially essenceless. Given my original modal arguments for essential essencelessness, however, this suggestion does not seem to help. Suppose we do say that



Essence and Emptiness

107

the Madhyamaka position is that there are two senses of “essence” here such that both of the following claims are true: (a) All things are empty of essence1. (b) Emptiness of essence1 is the essence2 of all things. But now the modal problem returns: are (a) and (b) necessary or contingent truths? For much the same sorts of reasons given earlier, the answer is presumably that both (a) and (b) are necessary truths. Assume further that the two types of essence distinguished are both still in some sense essences (hence the use of svabhāva for both in the Sanskrit) and thus both still satisfy the modal account of essence [E*]. Then from (a)  it follows that since all things are necessarily empty of essence1, all things are also essentially empty of essence1. And from (b)  it follows that all things do indeed necessarily have an essence (essence2) and hence they essentially lack essence1 and essentially have essence2. In other words, all things are still just as paradoxically essentially essenceless!

4.5  Essence and Modality So far I have been assuming what I claimed to be a familiar modal account of essence: [E*]: O is essentially F ≡ Necessarily (If O exists, then O is F) That this is a familiar account of essence is indubitable; that it is the correct account is more controversial. In Western philosophy the seminal account of essence is to be found in Aristotle. Aristotle says that “definition is the formula of essence” (Metaphysics 1031a12) and, though he does not directly offer a modal account of essence, he gives modal accounts of two cognate notions. An accident is defined as “something which may either belong or not belong to some self-same thing” (Topics 102b6–7). He also explains that things are “prior and posterior… in respect of nature and substance [when the priors] can be without the other things, while the others cannot be without them” (Metaphysics 1019a1–4). In the subsequent Western tradition Locke picks up on the definitional element: “[The essence of a thing is] the Being of any thing, whereby it is

108

Roy W. Perrett

what it is” ([1690] 1999, 402: bk. 3, chap. III, sec. 2). J. S. Mill favors the modal element: “[The essence of a thing is] that without which the thing could neither be, nor be conceived to be” ([1843] 1974, 110: bk. 1, chap. VI, sec. 2). In the first half of the twentieth century G. E. Moore gives a formal characterization of the modal approach: [A property P is internal, i.e., essential, to A  just in case] (x=A) entails xP… [or, equivalently, just in case the material implication] (x=A) → xP is a necessary truth.17 And clearly the modal account of essence [E*] mentioned above is in this spirit. More recently, however, Kit Fine has presented a pair of powerful arguments designed to call into question the adequacy of such modal accounts of essence.18 Here they are: Argument I: 1. Necessarily (If Socrates exists, then there is a singleton set {Socrates}). 2. Necessarily (If there is a singleton set {Socrates}, then Socrates is a member of the singleton set {Socrates}). 3. Necessarily (If Socrates exists, then Socrates is a member of the singleton set {Socrates}). 4. So, being a member of the singleton set {Socrates} is part of Socrates’ essence. Argument II: 1. Necessarily (Socrates ≠ Eiffel Tower). 2. Necessarily (If Socrates exists, then Socrates ≠ Eiffel Tower). 3. So, being non-identical with the Eiffel Tower is part of Socrates’ essence. The first argument says that, following modal set theory, it is necessary that Socrates belongs to the singleton Socrates if he exists. Hence, applying the modal account [E*], it must be that part of Socrates’s essence is being a member of the singleton Socrates! The second argument asks us to consider any two unconnected nonidentical objects, say Socrates and the Eiffel Tower. Necessarily, if Socrates exists, he must be distinct from the Eiffel Tower. So [E*] tells us that part of Socrates’s essence is being nonidentical with the Eiffel Tower!



Essence and Emptiness

109

In each case, of course, it is intuitively false that the specified property is an essential one for that object. So surely something is wrong with [E*]. But what? A plausible diagnosis is this: an object’s essential properties are those that make it the object it is, but modal accounts of essence cannot distinguish the conditions of an object’s identity from the consequences of its identity. Can anything be salvaged from the original modal account [E*]? Well, since it is a biconditional, we can disassemble it into these two conditional claims: [E*1]: O is essentially F → Necessarily (If O exists, then O is F) [E*2]: Necessarily (If O exists, then O is F) → O is essentially F And perhaps what we can now say is that [E*1] is true but [E*2] is false.19 This would explain, for example, why it is intuitive to say that Socrates necessarily has the property of belonging to the singleton Socrates, but it is not true that Socrates essentially belongs to the singleton Socrates. Similarly, it explains why it is necessary that Socrates and the Eiffel Tower be distinct, but it is not essential to Socrates that he be distinct from the Eiffel Tower.

4.6 Conclusion How does all of this relate to the modal arguments presented earlier for a Madhyamaka paradox of essential essencelessness? And what might a modern Mādhyamika conversant with modern analytic philosophy most plausibly say on the issues? Here I  want put aside matters of external history:  specifically, the question of exactly what Bhāviveka and Candrakīrti actually meant by their claims that emptiness is the essence of all things (a thorny interpretive issue on which there is no scholarly consensus anyway).20 Instead, I  want to focus on internal history:  specifically, the question of whether the modal arguments for essential essencelessness I introduced do in fact commit Madhyamaka to a paradoxical doctrine of emptiness as essential essencelessness. The two arguments from Kit Fine that we considered earlier challenged the modal account of essence [E*]. In particular, they challenged the validity of the inference from a property’s being necessary to its being essential,

110

Roy W. Perrett

though not the reverse inference. The modal argument I  presented for the doctrine of essential emptiness required the truth of the biconditional [E*], including the disputed move from necessity to essence. So that argument fails. We need to distinguish those properties that an object has necessarily from those properties it has essentially, even if an object has all its essential properties necessarily. What might be a better account of essence? That is a large and controversial question, but consider the following suggestion:  think of an object’s essential properties as those that underlie and explain the object’s other properties.21 Essences are not merely those properties that an object has necessarily, but those that ground all its other properties. Exactly what “ground” means here is obviously open to further debate, but the leading idea is this: metaphysics is not about what exists, but about what grounds what. In other words, it is about what is fundamental.22 Clearly, for the Ābhidharmikas who are Nāgārjuna’s primary targets in the Mūlamadhyamakakārikā the fundamental constituents of the world are dharmas definable in terms of their intrinsic natures (svabhāva). These dharmas are the primary existents (dravyasat) and the composite entities constructed out of primary existents are secondary or “conceptual” existents (prajñaptisat). Both exist, but in different ways insofar as the former ground the latter. The logical structure of other Indian metaphysical systems is similar. To be sure, they disagree on the nature and number of fundamental entities, but they all believe some fundamental entities ground others. And, of course, much the same is true of most Western metaphysical systems. Madhyamaka, in contrast, famously rejects this whole project of searching for grounds, urging us instead (in the words of the Vimalakīrtinirdeśa Sūtra) to overcome “the habit of clinging to an ultimate ground.”23 This is why we are warned not to grasp at emptiness itself as grounding things, an error as dangerous as grasping a snake wrongly (MMK 24.11). Having now abandoned [E*], perhaps we can profitably revisit the earlier suggestion that the Madhyamaka position is that there are two senses of “essence” such that both of the following claims are true: (a) All things are empty of essence1. (b) Emptiness of essence1 is the essence2 of all things. So what are the two senses of “essence” here? I suggested earlier that we need to distinguish those properties that an object has necessarily from those properties it has essentially, even if an object has all its essential properties



Essence and Emptiness

111

necessarily. This might seem a bit tendentious: after all, a thing’s essential properties are those that make it the thing it is, and is this not also true of the properties a thing has necessarily? A more irenic proposal is to say instead that we at least need to distinguish between two senses of “essence”: Essence1: Those properties of an object that ground (i.e., underlie and explain) the object’s other properties. Essence2: Those properties of an object that hold necessarily of it. Moreover, the presence of essence1 entails, but is not entailed by, the presence of essence2. Adopting this suggestion, a modern Mādhyamika can safely affirm without contradiction the apparently paradoxical claim that emptiness (i.e., essencelessness) is the essence of all entities. All that is meant by that is that both of the following claims are true: (a*) All things are empty of essence1 (= all things are groundless). (b*) Emptiness of essence1 is the essence2 of all things (= all things are necessarily groundless). And there is no contradiction in this, for that an object has a property necessarily does not entail that the property is a grounding property of that object (even if an object has all its grounding properties necessarily). Of course, such a doctrine will certainly run counter to the grounding project common to most Indian and Western metaphysicians. But then, as Nāgārjuna reminds us: Beyond good and evil, profound and liberating, this [doctrine of emptiness] has not been tasted by those who fear what is entirely groundless. (Ratnāvalī 1.79)24

Notes 1. Cf. Lakatos (1971). 2. The relevant literature on this topic is enormous. For a particularly lucid overview of some of the main lines of the debate, see Loux (2006, 84–120).

112

Roy W. Perrett

3. See Ganeri (2001, 101–102), Goodman (2004). 4. For the details see further Potter (1977). 5. Note, however, the timely warning in Garfield (1999, 89n): “I have translated the Tibetan ‘rang bzhin’ (Skt:  svabhāva) with the English philosophical term ‘essence’… [But] there will be resonances of the original terms that are not captured by the translation and new resonances introduced that will be foreign to the original text.” 6. Robinson (1972), Hayes (1994). 7. Cf. Garfield and Priest (2003, 15). 8. Garfield and Priest (2003, 16). Note that Arnold (2005b, 187)  misdescribes “Nāgārjuna’s Paradox” by conflating it with Candrakīrti’s claim that “essencelessness” (naiḥsvabhāvyaṃ) is to be understood as the essence (svabhāva) of all things, a view Nāgārjuna never explicitly espoused. 9. Read (1988, 31). 10. See Priest and Routley (1989, 483–498). 11. Robinson (1967, 50–51). 12. Garfield and Priest (2003, 92–96). 13. Iida (1980, 88), Ames (2003, 46). 14. Arnold (2005b, 186). (Skt: yā sā dharmānāṃ dharmatā nāma saiva tatsvarūpaṃ. atha keyaṃ dharmānāṃ dharmatā? dharmāṇāṃ svabhāvaḥ. ko ‘yam svabhāvaḥ? prakṛtiḥ. kā ceyam prakṛtiḥ? yeyaṃ śūnyatā. keyaṃ śūnyatā? naiḥsvābhāvayam.) On this passage and its subsequent Tibetan exegesis see further Magee (1999). 15. Scherrer-Schraub (1991, 64, 218). 16. Cf. Burton (1999, 214). 17. Moore (1922, 293, 302). 18. Fine (1994, 4–5). 19. Note that Garfield and Priest (2003, 15) apparently affirm [E*2]: “Since emptiness is a necessary characteristic of things, it belongs to them essentially.” It seems doubtful, however, that the presence of this slip could alone undermine their reasoning for “Nāgārjuna’s Paradox.” 20. See inter alia Ames (1982; 2003), Magee (1999), Burton (1999), Arnold (2005b), Westerhoff (2009a). 21. Cf. Nozick (2001, 126). 22. See further Schaffer (2009), Correia and Schnieder (2012). 23. Thurman (1976, 99). 24. Huntington and Wangchen (1989, 26).

5

The Net of Indra Graham Priest Because they have no Selfhood, the large and the small can mutually contain each other. . . . Since the very small is very large Mount Sumeru is contained in a mustard seed; and since the very large is the very small, the ocean is included in a hair. Chengguan

5.1 Introduction Huayan, or, as it is called in Japanese, Kegon, is one of the most intriguing forms of Mahāyāna Buddhism.1 It paints a view of all things as interdependent and interpenetrating. Its most dominant metaphor, alluded to many times in the Huayan (Skrt. Avataṃska; Eng. “Flower Garland”) Sūtra, the key sūtra for this form of Buddhism, is the Net of Indra.2 This is described by one commentator as follows:3 Far away in the heavenly abode of the great god Indra, there is a wonderful net which has been hung by some cunning artificer in such a manner that it stretches out indefinitely in all directions. In accordance with the extravagant tastes of deities, the artificer has hung a single glittering jewel at the net’s every node, and since the net itself is infinite in all dimensions, the jewels are infinite in number. There hang the jewels, glittering like stars of the first magnitude, a wonderful sight to behold. If we now arbitrarily select one of the jewels for inspection and look closely at it, we will discover that in its polished surface there are reflected all the other jewels

114

Gr aham Priest

in the net, infinite in number. Not only that, but each of the jewels reflected in this one jewel is also reflecting all the other jewels, so that the process of reflection is infinite. In the metaphor, the jewels are the objects of reality—mountains and mustard seeds, oceans and hairs. Each encodes all the others. All interpenetrate in the great Dharmadhātu, the totality of all interrelated things. This is a beautiful metaphor. But what, exactly, does it mean? In what follows, I  will provide an answer. I  will do this in a perhaps unlikely way: with the help of a little contemporary mathematics—though as I will show, this does nothing more that bring out the content of Huayan views and give it a precision it could not otherwise aspire to.4

5.2 Identity The relationship between the elements of reality, as Huayan conceives it, is often described as one of mutual identification, mutual containment, interconnectedness, interpenetration, nonobstruction, noninterference. (I will usually stick with “interpenetration.”) This may help a little, but does not get us very far. The Chinese character often used for the relation is ji (Jap., soku), 即, a term which can bear many meanings, dependent on the context; and in any case, its use here is clearly a term of art. So this does not help much either. A somewhat flat-footed interpretation of the notion is simply as one of numerical identity. Fazang (643–712 ce) is officially the third Patriarch of Huayan, but is regarded by many as the effective founder of the school. And we find him saying—or translated as saying:5 If we take ten coins as symbolizing the totality of existence, and examine the relationship of existence amongst them, then, according to Huayan teaching, coin one will be seen as identical with the other nine coins. And his commentator, Cook, says, concerning another passage by Fazang:6 This passage makes it clear that Fazang does in fact assert the identity of the rafter and the building, or, in other words, the part and the whole, or the particular and the universal.



The Net of Indra

115

Now, a feature of identity is that it is normally taken as supporting the substitution of identicals:  if x and y are identical, then anything true of x is true of y. And we do indeed find Fazang reasoning sometimes in a way that suggests this:7 Question:  Since the building is identical with the rafter, then the remaining planks, tiles, and so on, must be identical with the rafter, aren’t they? Answer: Generally speaking, they are all identical with the rafter. The reasoning here would appear to be that if r is any rafter, p is any plank, and b is the whole building, then since r = b and p = b, r = p. This is an instance of the substitution of identicals. Note that Fazang endorses the interpenetration between the part and the whole. This, itself, would seem to rule out the interpretation of interpenetration as identity. A car is not identical with its steering wheel. But worse follows. The interpretations quickly collapse into what amounts, effectively, to trivialism: nearly everything is true. Consider me. Take any property, P, and some object with that property. Then since I  am that thing, I have the property P. This seems far too strong. There is clearly some sense in which I am not a slice of toast; and for all that George Bush and I may be one, I am not responsible for the invasion of Iraq and its consequences in the way that he is. Indeed, the passage just quoted does not, in fact, require the relation to be identity. Any transitive relation will do. So it would seem that whatever the notion is, it is not one of literal identity.8

5.3 Emptiness Cook himself gestures at another understanding of what the relation might be. We often say that things are the same, meaning by this that they have a property in common. The ripe strawberry and the top traffic light are the same in that they are both red. In the present case, this seems far too weak an understanding, however. Any two things have some properties in common—for example, the property of being one thing. So this is to reduce the relation to banality. But maybe, says Cook, to say that things are related in the way we seek, is to say that they have some really important property in common. What? Says Cook:9

116

Gr aham Priest

First, we must accept the basic concept of emptiness itself. Second, we must consider emptiness to be so fundamental to the being of things that despite their obvious and real differences, they are alike in a more essential way in being empty. If we can accept these premises, then the claim that all things are identical does not seem quite so improbable, because identity is claimed on the basis of this common emptiness. Emptiness, that is, being empty (śūnya), is the central metaphysical notion of Mahāyāna Buddhism. Its importance was first articulated by Nāgārjuna, the founder of the Madhyamaka school. To be empty is to have no svabhāva. This is a difficult term to translate; it is often translated as self-being, or intrinsic nature. Roughly, a svabhāvic entity is one that is a metaphysical atom (like Hume’s distinct existences, or the simples of the Tractatus). It is one whose nature is independent of all other things. A non-svabhāvic entity, by contrast, is one whose nature is secundum quid: the entity possesses it only in virtue of its relationship to other things.10 And, crucially, all things are empty. This indeed is their fundamental nature. As Nāgārjuna puts it in his autocommentary to the Vigrahvyāvartanī, quoting the Aṣṭasāhasrikikā Prajñāpārāmita Sūtra:11 By their nature, things are not a determinate entity. Their nature is a non-nature; it is their non-nature that is their nature. For they have only one nature, i.e., no nature. Indeed, the “everything” is to be taken very seriously: even emptiness itself is empty. The view was taken over, endorsed, and developed, by Chinese Buddhism in general, and Huayan in particular. As Fazang himself puts it:12 The all is the one, for both are similar in being non-existent in Nature . are produced by the mind and have no self-nature at all. This is called the absence of characters. The scripture says, “All dharmas are originally empty in their nature and have not the least character.” Now, according to Cook, to say that all things are one is to say that they share this one (non)nature. They are the same in this most fundamental



The Net of Indra

117

of ways. That all things are one, in this sense, is stating an important truth of Buddhism, and one of the appropriate profundity. However, it really doesn’t seem to do justice to the metaphor of the Net of Indra. According to the metaphor, all things interpenetrate with all other things. To grasp one is, in some sense, to grasp all. This seems to go a lot further than the simple claim that all things share the same fundamental property. You can know everything about a and know that a and b share this most fundamental property without knowing much else about b. The Indranet13 must be at least powerful as the internet!

5.4  … and Its Structure But Cook’s insight, that emptiness is important to what is going on, at least takes us in the right direction. To understand how, we need to look more closely at the nature of emptiness. If something is empty, its identity is determined by its locus (location) in a certain network of relations. A thing’s being what it is, is its bearing a bunch of relations to other things. A perhaps helpful way of understanding this, especially for those coming from a background of Western philosophy, is by thinking of Leibniz’s notion of time. What is it to be a particular time, say 1066? According to Newton, times are things existing in and of themselves. They would have been there, even had the universe been empty of all matter and events. By contrast, Leibniz gave a relational account of time. To be 1066 is to be before Britain’s colonization of Australia, after Caesar’s invasion of Britain, contemporary with the Norman invasion of Britain, and so on. 1066 is simply a locus in a set of temporal relations between events. Does that mean that it does not exist? In a sense, yes: the time does not exist in the way that Newton took it to; it has no intrinsic existence. In a sense no: Leibniz would not deny that there was a 1066; but it exists and is what it is only as this locus in a bunch of relations.14 Similarly, what is it to be Graham Priest? My being is constituted by having been born in London in 1948, being the child of George and Laura, residing most of my adult life in Australia, being the father of Marcus and Annika, dying in ?, and so on. Anything that related to those things in those ways would be me; there is no ding an sich Graham Priest. I am essentially constituted by my place in that web of relations.15 We may represent this graphically. Suppose, for the sake of illustration, that my being is constituted by relating to three objects, a, b, and c. Then we might depict the situation thus (figure 5.1):

118

Gr aham Priest

Figure 5.1  The Locus of a Network of Relationships

where the labels on the arrows indicate the relations in question. Anything at the spot marked would, ipso facto, be me.16 Note that relations have a direction, marked by the direction of the arrow. (Brutus killing Caesar is not the same as Caesar killing Brutus.) It will be helpful in what follows to arrange for all the arrows to point in the same direction.17 One can do this because every relation, ρ, has a converse, ρc. yρx iff xρcy; and both of these say the same thing. So we can o

always replace an arrow pointing in the wrong direction, x ← y , with one ρc

pointing in the right direction,  x → y .18 Of course, since the a, b, and c in fi ­ gure 5.1 are themselves empty, the same analysis must be applied to them. They themselves are nothing more than loci in a field of relations. We might depict this as follows (using the magic number three again, and ignoring the superscripts on the arrows for perspicuity) (figure 5.2): a0 a1 a2

b0 b1 b2

c0 c1 c2

Figure 5.2  The Network Expanded



The Net of Indra

119

Each of the ais, bis, and cis, must be treated in the same way, of course. On ultimate analysis, then, the original object turns out to be the locus which is the root of a tree which is infinite along every branch, of the following kind (figure 5.3):

Figure 5.3  A Fully Expanded Network

In the end, all content has disappeared. What is left is pure form—structure. Indeed, what the tree gives is exactly the ontological structure of the object.19

5.5 Interpenetration And now, something interesting can happen. Consider, say, the north pole of a magnet n. This is a north pole only because it relates to the south pole, s, in a certain way, ρ. The north pole could not be what it is if there were no south pole. Hence, its structural tree will look like this (figure 5.4). (I record which object is at each node, so that one may keep track of matters, but label only the relation ρ to avoid clutter.)

n

ρ

...

...

s

...

...

...

Figure 5.4  A North Pole

As one might expect, the tree of s is a part of the tree of n. But of course, s is symmetrically related to n. So if we fill out the tree a bit further, we will get the following (figure 5.5):

120

Gr aham Priest ... n

ρ

s ...

... ρ

n ...

... ρ ... ...

Figure 5.5  A North and South Pole

The tree of n is part of the tree of s!20 The two objects intermingle in the most intimate way. The structure of each is literally a part of the other. Note that this can happen only because the trees of both n and s are infinite. If either were finite, the situation could not arise. At any rate, one could not hope for a nicer representation of the idea that the two objects “interpenetrate.” The ontological structure of each contains (encodes) the ontological structure of the other. The relation of two trees each being a subtree of the other is obviously a symmetric and transitive relation.21 The Huayan notion of interpenetration may be taken to be exactly this.

5.6  Li and Shi We are not finished with the story yet, though. There is a standard distinction in Mahāyāna Buddhism between two realities, conventional and ultimate. Conventional reality is constituted by the phenomenological objects of common experience, such as Mount Sumeru and a hair. These are empty. Ultimate reality is the reality that appears once one strips away the reification of conventional thought: emptiness itself (śūnyata, kong 空). As already noted, though, it would be a serious mistake to reify this. Like everything else, it is empty. Since this ultimate reality is empty, it has only relational nature, and it is at the root of a structural tree of its own. What does it relate to? The objects of conventional reality. It relates to these in the profoundest of ways. For a start, each phenomenal object could not be what it is if it were not a manifestation of this ultimate reality. But conversely, ultimate reality could not exist if it did not manifest itself through these phenomena.22 In his Treatise on the Golden Lion, Fazang explains the relationship between the two using the metaphor of a statue of a lion made of gold. Ultimate reality is the gold; the phenomenal world is the lion-appearance (so that its elements are the eye, the ear). And one cannot have the one without the other.



The Net of Indra

121

Thus, ultimate reality and the objects of phenomenal reality are like two sides of the same coin. Each can be what it is only given the other. In other words, they interpenetrate. They play north and south to each other. Let us write e for ultimate reality, and a, b, c,… for the objects of phenomenal reality. Then what we have seen is that the trees of e and a (and b, c) are as follows (figure 5.6): a e

...

e

b c

...

...

...

a ... b ... c ...

Figure 5.6  The Interpenetration of Li and Shi

And what fi ­ gure  5.6 shows is that emptiness interpenetrates with every object. Neither is this simply an artifact of the diagram: it is a core Huayan thesis. In Chinese Buddhism, ultimate reality goes by many names (emptiness, Buddha Nature). In Huayan, it is standardly called li 理 (principle).23 The objects of conventional (phenomenal) reality are referred to as shi 事 (event, fact). Huayen isolates four features of the dharmadhātu.24 The first is the existence of the objects of phenomenal reality, shi. The second is the existence of ultimate reality, li. The third is exactly that li and shi interpenetrate, lishi wuai, 理事無礙法. (The fourth we will come back to in a moment.) This interpenetration is expressed by Dushun (557–640 ce), the First Patriarch of Huayan, in his Meditation on the Dharmadhātu, as follows:25 Li, the law that extends everywhere, has no boundaries or limitations, but shi, the objects that are embraced by li, have limitations and boundaries. In each and every shi, the li spreads all over without omission or deficiency. Why? Because the truth of li is indivisible. Thus, each and every minute atom absorbs and embraces the infinite truth of li in a perfect and complete manner.   Shi, the matter that embraces, has boundaries and limitations, and li, the truth that is embraced [by things], has no boundaries or limitations. Yet this limited shi is completely identical, not partially identical, with li. Why? Because shi has no substance — it is the selfsame li. Therefore, without causing the slightest damage

122

Gr aham Priest

to itself, an atom can embrace the whole universe. If one atom is so, all other dharmas should also be so. Contemplate on this. Before we leave the matter, one further comment. The relationship between conventional and ultimate reality, shi and li, is a somewhat vexed one in Madhyamaka Buddhism.26 In some sense, they are distinct; in some sense, they are one. The graphical representation of the relation provides a very clear articulation of the matter. First, emptiness, e, is empty. That is, it itself is the root of a non-well-founded tree. Next, it and the elements of conventional reality are distinct: they have different trees. But, finally, e is identical with conventional reality, in the sense that it interpenetrates every one of its objects. The tree for each such object is both a part of, and has as a part, the tree for e.

5.7  Shi and Shi We are still not finished. If a and b are shi, phenomena, then a interpenetrates with e, and e interpenetrates with b. But interpenetration is, as I noted, transitive. So a interpenetrates with b. One can see this diagrammatically. Come back to our fi ­ gure  5.6. As this shows, the tree of c is a subtree of the tree of b. Symmetrically, of course, the tree of b is a subtree of c—though this goes off the diagram. (The tree for b is a subtree of the tree for e, which is a subtree of the tree for c.) Hence, all shi interpenetrate. This is, in fact, the fourth Huayan principle: the interpenetration of shi and shi: shishi wuai, 事事無礙法界. All phenomena interpenetrate. Chengguan (738–839? ce), the Fourth Patriarch of Huayan, puts it thus:27 Because they have no Selfhood , the large and the small can mutually contain each other.  .  . . Since the very small is very large, Mount Sumeru is contained in a mustard seed; and since the very large is the very small, the ocean is included in a hair. And Fazang thus:28 has the characters of roundness and smallness. This is a fact . Its nature is empty and non-existent. This is principle