Familiar Objects and their Shadows (Cambridge Studies in Philosophy)

FA M I L I A R O B J E C T S A N D T H E I R S H A D OW S Most contemporary metaphysicians are sceptical about the reality of familiar objects such as dogs and trees, people and desks, cells and stars. They prefer an ontology of the spatially tiny or temporally tiny. Tiny microparticles “dogwise arranged” explain the appearance, they say, that there are dogs; microparticles obeying microphysics collectively cause anything that a baseball appears to cause; temporal stages collectively sustain the illusion of enduring objects that persist across changes. Crawford L. Elder argues that all such attempts to “explain away” familiar objects project downwards, on to the tiny entities, structures and features of familiar objects themselves. He contends that sceptical metaphysicians are thus employing shadows of familiar objects, while denying that the entities which cast those shadows really exist. He argues that the shadows are indeed really there, because their sources – familiar objects – are mind-independently real. c r aw f o r d l . e l d e r is Professor and Head of the Philosophy Department at the University of Connecticut.

C A M B R I D G E S T U D I E S I N PH I L O S O PH Y General Editors Jonathan Lowe (University of Durham) NOAH LEMOS (College of William and Mary) Advisory Editors Jonathan Dancy (University of Texas, Austin) John Haldane (University of St. Andrews) Gilbert Harman (Princeton University) Frank Jackson (Australian National University) William G. Lycan (University of North Carolina, Chapel Hill) Sydney Shoemaker (Cornell University) Judith J. Thomson (Massachusetts Institute of Technology) Recent Titles David Lewis Papers on Ethics and Social Philosophy Fred Dretske Perception, Knowledge, and Belief Lynne Rudder Baker Persons and Bodies Rosanna Keefe T heories of Vagueness John Greco Putting Skeptics in T heir Place Ruth Garrett Millikan On Clear and Confused Ideas Derk Pereboom Living Without Free Will Brian Ellis Scientif ic Essentialism Alan H. Goldman Practical Rules Christopher Hill T hought and World Andrew Newman T he Correspondence T heory of Truth Ishtiyaque Haji Deontic Morality and Control Wayne A. Davis Meaning, Expression and T hought Peter Railton Facts, Values, and Norms Jane Heal Mind, Reason and Imagination Jonathan Kvanvig T he Value of Knowledge and the Pursuit of Understanding Andrew Melnyk A Physicalist Manifesto William S. Robinson Understanding Phenomenal Consciousness D. M. Armstrong Truth and Truthmakers Keith Frankish Mind and Supermind Michael Smith Ethics and the A Priori Noah Lemos Common Sense

Joshua Gert Brute Rationality Alexander R. Pruss T he Principle of Suff icient Reason Folke Tersman Moral Disagreement Joseph Mendola Goodness and Justice David Copp Morality in a Natural World Lynne Rudder Baker T he Metaphysics of Everyday Life Sanford Goldberg Anti-Individualism Michael J. Zimmerman Living with Uncertainty

Familiar Objects and their Shadows Crawford L. Elder University of Connecticut

CAMBRIDGE UNIVERSITY PRESS Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Dubai, Tokyo, Mexico City Cambridge University Press The Edinburgh Building, Cambridge CB2 8RU, UK Published in the United States of America by Cambridge University Press, New York www.cambridge.org Information on this title: www.cambridge.org/9781107003231 © Crawford L. Elder 2011 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2011 Printed in the United Kingdom at the University Press, Cambridge A catalogue record for this publication is available from the British Library Library of Congress Cataloguing in Publication data Elder, Crawford. Familiar Objects and their Shadows / by Crawford L. Elder. p. cm. – (Cambridge Studies in Philosophy) Includes bibliographical references and index. ISBN 978-1-107-00323-1 1. Reality. 2. Phenomenalism. 3. Knowledge, Theory of. 4. Metaphysics. 5. Ontology. I. Title. BD331.E43 2011 111–dc22 2010045715 ISBN 978-1-107-00323-1 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

To the memory of J. N. Findlay (1903–1987) my philosophical father

Contents Acknowledgments

xi

Introduction 1 Two false friends of an ontology of familiar objects 1.1 The point of affi rming kind-sameness 1.2 The point of affi rming numerical persistence 1.3 Pushmipullyu representations 1.4 Explosivism 1.5 Conventionalism 1.6 Summing up 2 Conventionalism as ontological relativism 2.1 Sameness and objects 2.2 The “softness” of sameness in kind and numerical sameness 2.3 Carving out strange kinds 2.4 Carving out strange individuals 2.5 The world onto which we project kind-sameness and persistence 2.6 We who project kind-sameness and persistence 3 Realism about material objects: persistence, persistence conditions, and natural kinds 3.1 Realism and mind-independent persistence conditions 3.2 Nonqualitative persistence conditions? 3.3 Qualitative persistence conditions 3.4 Apparent departures from a natural kind 3.5 Infi mae species 4 Ontological preference for the temporally small 4.1 An explosion of temporal counterpart relations? 4.2 Restricting the t-counterpart relations that really obtain: fi rst attempt

1 7 9 12 13 19 25 28 30 32

ix

36 39 44 48 52 54 55 57 60 62 66 70 72 76

Contents

5

6

7

8

4.3 Restricting the t-counterpart relations that really obtain: second attempt 4.4 “Nonsupervenient” t-counterpart relations 4.5 Motivations for exdurantism Ontological preference for microphysical causes 5.1 Three threats to mental causation 5.2 Causation and exceptionless laws 5.3 Causation, invariance, and beliefs and desires 5.4 Causal exclusion I: complex events at the level of the microparticles 5.5 Causal exclusion II: complex events at the level of neurochemistry 5.6 The problem of mental quausation Ontological preference for the spatially small 6.1 Just where does “dogwise arrangement” obtain? 6.2 Causal competition between dogs and microparticles dogwise arranged 6.3 Why “more and less different from” is not defi ned over the microphysical events that subvene dog behaviors 6.4 Chains of causation involving microparticle events 6.5 Summing up A third false friend of familiar objects: universal mereological composition 7.1 The Sider/Lewis argument for UMC 7.2 Two forms of UMC objects, two forms of strangeness 7.3 The properties that characterize UMC objects 7.4 Why contrariety fails to obtain 7.5 UMC and mereological essentialism Concluding Hegelian postscript 8.1 Hegel and contemporary scepticism about familiar objects 8.2 Hegel on property-identity: the crucial place of graded contrast with contraries 8.3 Objections to the Hegelian position

Appendix: “Mutually interfering” dimensions of difference References Index

x

79 81 83 88 88 90 93 100 107 112 114 118 124 128 131 137 139 142 148 151 158 162 166 166 168 181 190 195 200

Acknowledgments Large portions of chapter 6 appeared (under the title “On the Phenomenon of ‘Dog-Wise Arrangement’”) in Philosophy and Phenomenological Research (74 [2007], pp. 134–57); large portions of chapter 7 (under the title “Against Universal Mereological Composition”) in dialectica (62 [2008], pp. 433–54); and large portions of chapter 2 will appear (under the title “Carving Up a Reality in Which There Are No Joints”) in a collection forthcoming from Blackwell’s (Steven D. Hales, ed., A Companion to Relativism). I thank the editors of these journals, The European Society for Analytic Philosophy, and Blackwell’s for permission to use these passages. Early versions of several chapters were presented as papers: chapter 7, at Durham University; chapter 1, at University of Georgia and at Auburn University; chapter 5, at University of California at Irvine, at University of Calgary, and at my own department’s summer discussion group. I am sincerely grateful to these audiences: their questions made me understand better the arguments of those chapters. I am also very grateful to various individuals for conversation on the topics of particular chapters. Before I mention even one of them, I want to say that the people in question often disagreed with me; happily, in philosophy, one can learn as much from disagreement as from agreement. Chapter 1: Ruth Millikan, James Paton. Chapter 3: Stephen Schwartz, Randy Carter (= W. R. Carter). Chapter 4: Jesse Mulder. Chapter 5: David Slutsky, Michael Lynch. Chapter 6: James Woodward. Chapter 7: Henry Laycock, Achille Varzi. Chapter 8: Jonathan Lowe, Nancy Cartwright. Appendix: Jesse Mulder, B. J. Strawser, Mohan Matthen.

xi

Introduction In general, contemporary metaphysics is deeply sceptical of the familiar objects in which common sense believes. It is far more ready to attribute reality to entities that are much smaller – to the particles and wave packets and strings which microphysics treats as real, or to the “mereological simples” for which philosophical reflection provides some support. Any such view must fi nd some way of explaining why there appear to be familiar medium-sized objects in the world. Many metaphysicians suppose that we can do just that. We can explain why it appears that the microparticles of the world compose familiar objects, why it appears that these objects persist across careers in which they lose and gain component microparticles, and why it appears that these objects have and exercise causal powers. The main business of this book is to argue that leading examples of such reductive explanations fail. For time and again such explanations project downwards, onto the small entities of the preferred ontology, structures and relations and features that properly belong to familiar objects. Such projection is harmless so long as one allows that there also are, in addition to the small entities, the familiar objects that form the starting point of the projection. But if – as is generally the case – the aim is to expunge familiar objects from ontology, the invocation of such structures and relations and features is illegitimate. The opponents of familiar objects are then helping themselves to shadows cast downwards, onto the level of the preferred small entities, while denying that the sources of these shadows exist. The metaphysical position which this book is intended to support is that at least many of the familiar objects that common sense recognizes are mind-independently real. The book begins with a chapter that undertakes to refute two false friends of this commonsense ontology. The fi rst is the modal conventionalist, who holds 1

Introduction that the general ways in which nature’s kinds are marked out from one another, and the general ways in which persistences of members of those kinds are delimited, are fi xed by the “descriptive content” and “referential intentions” that we associate with our sortals and matter-names. Such a view fails to treat as mind-independently real the phenomena of sameness in kind and of persistence across time. It therefore falls short of realism about familiar objects (indeed about any objects) since it makes the natures that objects share with others of their kind, and the careers which each individually traces out, be functions of our cognitive and linguistic practices. The other false friend is the explosivist, who is happy to award mind-independent reality to the familiar objects of common sense, but who cheapens that award – indeed nullifies it – by awarding reality likewise to every imaginable crosscutting of the world’s individuals and kinds. The modal conventionalist thinks that nothing is required of the world in order for our general ways of tracing persistences and detecting kind-samenesses to be correct; the explosivist thinks that nothing special is required, since any general ways of doing this cannot fail to track real persistings and real samenesses in kind. Chapter 1 argues that neither conventionalism nor explosivism embodies an adequate understanding of the ways that our talk about sameness in kind, and about numerical persistence across time, functions in our cognitive economy. Chapter 2 deals further with modal conventionalism. Of the two false friends of common-sense ontology, conventionalism and explosivism, conventionalism has been the more influential, and so deserves the more protracted treatment. One form which conventionalism has assumed is a view that might aptly be labeled “ontological relativism.” Chapter 2 examines one arresting argument for ontological relativism, and contends that that argument fails. More radically, chapter 2 argues that any argument for ontological relativism must fail – that the view is conceptually untenable. Between them, chapters 1 and 2 raise serious objections against the main currently prevailing forms of antirealism about material objects. Chapter 3 then sets forth a realist position on material objects. It articulates the connections between realism about the existence of material objects and realism about the two forms of sameness discussed in chapter 1, and shows what a realist ontology that incorporates all three elements must look like. 2

Introduction The book then turns to the opponents of common-sense ontology. As puzzles such as the ship of Theseus have made vivid for millennia, familiar objects, if real at all, seem to survive across a messy variety of alterations – and not just of loss and gain of component particles, but of hard-to-delimit qualitative alteration. Here the desire for a cleaner ontology may motivate an ontological preference for entities that are temporally smaller than familiar objects appear to be – for temporal stages, each of which lasts no longer than the shortest possible physical change. Such temporally tiny entities can explain away the appearance that there are familiar objects that persist over long careers by serving as the truth-makers for claims expressing that appearance. If a temporal stage of the right qualitative character stands in temporal counterpart relations to other stages having the right qualitative character, then, say stage theorists, a sentence that asserts or presupposes the persisting of a familiar object can be rendered true. But I argue that temporal counterpart relations – if they are not going to saddle us with an explosivist account of the world’s persistences – constitute an illegitimate projection downward from the careers and powers of familiar objects. The second group of opponents are the causal exclusionists. From the time of Plato’s Eleatic stranger it has seemed plausible that familiar objects, if real, must be capable of bringing about effects. But any familiar object is wholly composed of entities that, spatially, are vastly smaller – in particular, the microparticles of physics. If physics is closed and complete, it may seem, then, that the several doings of the component microparticles must between them cause anything which the familiar object may be said to cause. We have apparent overdetermination, which is apparently intolerable, and the apparent victory goes to the microparticles. But to which microparticles? When, I shall argue, microparticles are grouped in the ways relevant for awarding them efficacy over the effects that common-sense attributes to familiar objects, they are grouped in ways that illegitimately project downwards, from the level of the familiar objects themselves. Causal exclusion arguments, focused on familiar objects, generally fail. They fail in particular for the apparent exclusion of mental causation. Mental causation is the case I shall use to focus the debate about whether a familiar object’s microparticles steal away the apparent efficacy of that familiar object. That is, I shall discuss whether beliefs and desires – states of that most familiar of familiar 3

Introduction objects, a person – genuinely cause behavioral outcomes. But I will indicate how the argument generalizes to the cases of other familiar objects. The third group of opponents are the sceptics about composition. It seems secure, to these opponents, that tiny entities entirely occupy any volume in which common sense supposes a familiar object to be present. But is there also the one large object which these entities seem to compose? These philosophers begin with uncertainty about what, in general, composition might amount to, and proceed to scepticism about whether there objectively is any such phenomenon at all. It unquestionably appears that microparticles compose such (relatively) large objects as dogs and trees and desks – but that appearance may amount to no more than that the microparticles themselves are “dogwise” (or “treewise” or “deskwise”) arranged. I argue that such adverbial arrangements are a projection downwards from familiar objects. They are, indeed, perfectly real if the objects from which the projection proceeds are themselves real. That is, there is a genuine phenomenon of microparticles’ being dogwise or treewise or deskwise arranged , if this phenomenon just amounts to the fact that those microparticles jointly occupy (and are confi ned to) the entire volume in which a dog or tree or desk exists. If that sort of fact obtains, indeed, it likewise provides the analysis of what it is for those microparticles to compose a dog or tree or desk. But the opponents in this third group want to explain away, rather than affi rm, the reality of such familiar objects. And if familiar objects do not really exist, the phenomena of dogwise or deskwise arrangedness are purely imaginary. A fourth group of opponents thinks that composition is not something which microparticles owe to the reality of the familiar objects within which they are found, but something that they possess in their own right. Indeed any plurality of entities whatever composes something, says this fourth group. Composition is a “free lunch” (in Armstrong’s phrase), which comes automatically with the bare existence of the components. This is the doctrine of universal mereological composition (UMC). The proponent of this doctrine qualifies as a third false friend of the ontology of common sense. The proponent appears to be a friend because, at any moment at which common sense supposes a familiar object to exist, she will fi nd a mereological sum of microparticles that occupies just that volume which common sense supposes the familiar object to occupy (with the small qualification 4

Introduction that the mereological sum will have relatively precise boundaries, while the familiar object apparently has vague boundaries). But the proponent of UMC is a false friend, not just because of her explosivist commitments, but because, as I shall argue, her stand-ins for familiar objects are compositionally brittle, while familiar objects themselves are compositionally flexible. That is, across the phases of its existence, a familiar object might have incorporated different microparticles from those that it did; not so, I argue, the mereological sum of microparticles that is located where that familiar object is. UMC is a false doctrine, I shall argue. The composed objects which it countenances would in general be characterized only by certain structural properties, properties that fail to contrast to greater and lesser degree with their own proper contraries. But determinate contrast-with-contraries is constitutive of the very identity of any genuine property. The argument of this book is defensive. The book identifies inadequacies in contemporary attempts to “explain away” familiar objects, attempts intended to show that no familiar objects really exist. Some readers may fi nd themselves wishing to see positive arguments in favor of familiar objects. As responses to the contemporary opponents of familiar objects, these would be arguments to the effect that some familiar objects are perfectly real – not necessarily that every familiar object posited by common sense, or by one of the special sciences, is real. I will offer no argument for that more limited conclusion, because I believe that any such argument would be question-begging: it would have to proceed from premises that assume that at least some familiar objects exist. For the proponent of familiar objects, as I see matters, the situation is exactly that of Neurath’s boat. We can suspect individual planks of rot, can remove them and examine them, and can even replace them if need be. But we do this while afloat on the boat – while standing on other planks. There can be no systematic justification for standing-on-planks-in-general-and-as-such. The history of post-Cartesian philosophy of course contains many efforts at establishing the reality of familiar objects – objects such as trees and dogs, stars and cells, perhaps even desks and pencils. These arguments all proceed from premises, allegedly more secure than the conclusions to be established: that we exist, and engage in various cognitive and perhaps practical activities. But if we ourselves are familiar objects – human organisms, say – and if our featured 5

Introduction activities engage and are directed at other familiar objects, then these arguments are question-begging in just the way I have indicated. The only alternative is to start from a picture of ourselves as transcendental egos, and to secure the existence of familiar objects by virtue of their relation to transcendental mental activity. This runs counter to the naturalist position, to which I subscribe, that we are objects in the world of familiar objects, distinguished mainly by the history of natural selection that has fashioned us. Beyond that, the only reality which such an argument can deduce for familiar objects is a mindconferred, mind-dependent reality. Not here. The thesis of this book is that familiar objects – at least some of them – are mind-independently real. Their detractors seek to impugn the ontological status of familiar objects by using shadows which those objects cast, while denying that the shadows have a source.

6

1 Two false friends of an ontology of familiar objects Judgements asserting one or another of two kinds of sameness are crucial, I shall argue, both for our practical mastery of the world and for our theoretical understanding of it. On the one hand, there are judgements saying that one object is the same in kind as other objects, or that some matter is the same in kind as matter found elsewhere. On the other hand, there are judgements saying that the object in front of us is numerically the same object, or that the matter is the very same matter, as we encountered earlier or will encounter later. In making these judgements we call upon observation and understanding. In order to affi rm sameness in kind, we must observe that various similarities obtain between one object and others, or between matter here and matter elsewhere. In order to affi rm persistence across a single episode of observation, we must observe that an object (or some matter) has moved continuously, while retaining largely the same features, and in order to affi rm persistence across separate episodes, we must observe that the object (or the matter) now before us presents features appropriately related to those observed in an object (or some matter) encountered at other times. But we must also understand which sorts of similarities indicate sameness in kind, which sorts of relations mark out persistences. We must understand, for example, that sameness in chemical microstructure indicates sameness in kind, as between two portions of matter, while sameness in color does not, and neither do sameness in heft or in location. We must understand that specific sorts of sameness or change are to be expected in a persisting object of the kind to which the object observed earlier and the object observed later belong, and specific spatiotemporal relations to the place of the earlier observation. To speak of a cognitive performance as a case of understanding is to say that it can fail as well as succeed: where understandings occur, misunderstandings are possible. Just what is required of the world, in 7

Familiar objects and their shadows order for the understandings that undergird our judgements of kindsameness and numerical persistence to succeed? What are the truth conditions for our implicit ideas about which sorts of properties are constitutive of kind-sameness and which sorts of relations indicative of numerical persistence? To these questions, prevailing philosophical opinion offers one or another of two markedly puzzling answers. One is that nothing is required of the world in order for our ideas about how nature’s kinds are delimited, and how nature’s persistences are marked out, to be true – our ideas are true, but they do not have truth conditions. For the sentences that express these ideas are analytic; they are not true in virtue of the world’s being one way or another.1 This is the claim of the “modal conventionalists” (or “modal conceptualists”), to whose views I return in section 1.5. The other prevailing opinion is that nothing special is required for these ideas of ours to be true. They are true in virtue of ways the world is, but the world is many ways at once. It contains crisscrossing kinds so numerous that any scheme for assigning objects to kinds is bound to capture a way the world is.2 It contains vastly many colocated objects which persist over quite different spans of time – objects which may even persist in quite different forms from one another, or at quite different locations – so that any scheme we might subscribe to, for tracing persistences, is bound to be right. 3 These are claims made by philosophers who might all appropriately be called “explosivists,” to whose views I return in section 1.4. Either sort of answer is markedly puzzling. If (as I shall argue) our judgements of kind-sameness and numerical persistence are crucial both to our practice and to our theory, it is nothing short of amazing that whatever general scheme we might embrace for making such judgements is bound to succeed – that the only error possible is error in executing that scheme, by making faulty observations. The business of this chapter is to articulate a realist picture of our implicit 1

2

3

Thomasson, Ordinary Objects, pp. 67–68; Sidelle, Necessity, Essence, and Individuation, p. 128. For an illustration, see Hirsch, Dividing Reality, pp. 24–25 – but concerning Hirsch’s own position, see note 21 of this chapter. Hawthorne, Metaphysical Essays, p. vii; Sosa, “Existential Relativity,” p. 142; Sider, Four-Dimensionalism, p. 133.

8

Two false friends of familiar objects ideas about how kind-samenesses are constituted and how persistences are marked out. On this realist picture, these ideas do indeed have truth conditions. The modal conventionalists are wrong. But not every such idea that we might embrace will be true – the truth conditions are sparse. Explosivists are wrong. I shall articulate a realist picture of our schemes for tracing kind-sameness and numerical persistence, but will not offer much in the way of positive argument for that picture. For, in my opinion, the presumption must be that some realist account of these schemes is correct; it is simply not credible that on matters so crucial for our survival and our understanding of the world, just any way that we might proceed cannot end up steering us wrong. But I shall take pains to explain how, if the realist picture I offer is correct, both conventionalism and explosivism misunderstand our judgements about kind-sameness and numerical persistence. 1.1

t h e poi n t of a f f i r m i ng k i n d -sa m e n ess

Why does it matter to us to note that two (or more) objects are the same in kind, or that some matter here is the same in kind as the matter over there? The standard answer is that such judgements guide inductive inferences. That is the answer I shall set forth in this section, adding some not entirely standard details. The kinds of the world typically count for us as falling into families: there are kinds of animals, kinds of food, kinds of fluids and plants and even artifacts. Typically, the kinds within any such family count for us as collectively characterized by certain sorts of properties, and as individually characterized by just one instance of these sorts. Thus we suppose that each kind of animal is characterized by a particular style of locomotion, a particular diet, and a particular (if roughly defi ned) body shape and size. Foods of the various kinds are, in general, each characterized by an aroma, a texture, a taste, and some even by a particular way we feel in response to eating them. Different kinds of fluids are, in general, each characterized by a particular color and scent and viscosity, and perhaps even by a particular boiling or freezing point. Each kind of plant has a characteristic pattern of growth, and a characteristic morphology when mature. The various kinds of artifacts typically count for us as each having a particular shape and design, a particular use, and as having 9

Familiar objects and their shadows a characteristic association with a setting in which or with which that use gets performed.4 To put it generally: the kinds in each family count for us as selecting just one out of the properties in each of several (or even many) ranges of properties. Each kind does this, we suppose, in member after member or portion after portion. For this reason, judgements that a plurality of objects (or of portions of matter) belong alike to a common kind have the function of directing certain kinds of inductions. The judgements say, so to speak: you need only discover which member out of each relevant property-range is present in a few observed members of a kind, in order to be sure that that same property is present in other members of that kind (or other portions, wherever located). Judgements of kind-sameness enable instantaneous discoveries about items that are remote. Once we have made such discoveries, moreover, judgements that newly encountered objects belong to that same natural kind enable us to say: these newly encountered objects have properties that may not yet have been revealed to our observation. We need only observe enough about the newly encountered objects to judge that they belong to that same kind, in order to be sure that these newly encountered objects have other properties beyond those we immediately observe – including properties that we cannot detect in the current observational setting. Judgements of kind-sameness then enable amplifi ed observation of items that are present. How do we know just which sorts of properties it is, from which each kind in a given family selects just one instance? Modal conventionalists maintain that this knowledge is a priori. It is analytic, they say, that the animals in a particular natural kind are all characterized by a particular body shape and a roughly defi ned body weight upon maturity,5 analytic that food of a given kind is everywhere characterized by a particular taste and texture and smell. I shall return to this position in section 1.5. For now I merely note that it seems implausible that knowledge of the sort in question is never empirical. It was an empirical achievement of some moment, one might well suppose, to learn that the various fluids and 4 5

Elder, “On the Place of Artifacts in Ontology.” Thomasson, Ordinary Objects, pp. 38–44 and 48–53; Sidelle, Necessity, Essence, and Individuation, p. 112.

10

Two false friends of familiar objects matters of nature are each characterized by a distinctive chemical microstructure. It was a very substantive empirical discovery that the various species of animals are each characterized by a roughly defi ned genome. Some of what we know, about the sorts of properties by which nature’s kinds are individuated, might well have been “hardwired” into us by natural selection. It seems plausible, for example, that we were “hardwired” to reidentify particular nutritious substances, and to differentiate them from one another, by tracking each through a characteristic combination of smell and taste and texture. In other cases, it seems easy to suppose that we might empirically have learned which sorts of properties individuate the kinds in a given range. This is all that need to have happened. We observed that certain properties cluster together in object after object, or sample after sample of matter, widely enough to warrant the suspicion that the clustering was no accident. We then gained evidence that the clustering was causally governed, by observing certain patterns of covariation – by observing that properties contrasting (whether mildly or sharply) with some of those in that recurrent cluster themselves reliably get accompanied with variants on other properties in the original cluster. We learned that one pattern of clustering was no accident, by observing that there are variant patterns of clustering.6 In learning this, we learned that each cluster marks out one kind in a common family; at the same time, we learned which sorts of properties it is, by which the kinds in that family are individuated. In any case, our understanding of which sorts of properties individuate the kinds in nature seems empirical in the sense of being empirically revisable. We may once have thought that foods of particular kinds are by nature found in particular narrowly defi ned environments, and might later have learned that the same foods can grow in different sorts of environments, or in different seasons. Once we did believe that each kind of metal by nature had a particular realworld connection with a bodily humor and with a particular planet; later we learned that this is false.

6

Elder, Real Natures and Familiar Objects, Ch. 2.

11

Familiar objects and their shadows

1. 2 t h e p o i n t o f a f f i r m i n g n um e r ica l pe rsist e nce Why do judgements of persistence matter to us? Why does it matter to note that numerically the same object has been present across some protracted episode of observation, or that numerically the same object is present to our observation now as was present at an earlier episode; why and when does it matter to us that the very matter (the very gold, clay, water) that we’re now dealing with is the same matter as we dealt with some (short or long) time ago? There are two main ways in which such a judgement can be of importance. First, it can be crucial in enabling us to learn something about what members (or samples) of a given kind in general are like. It can enable us to learn how members (or samples) all alike respond to particular kinds of circumstances or impingements – how they move (or stay still) in response to collisions, how they respond to being heated or poured out over a fl at surface, how they grow or ripen, how they react to the ingestion of particular foods, how they learn from particular sorts of experience. Put abstractly, keeping track of numerical persistence can be crucial in enabling us to detect reliably repeating patterns of “before and after” among the accidental states which members of a given kind can assume. Thereby we come to know more about the kind: the scope is expanded both for instantaneous knowledge of items that are remote, and for amplified observation of items that are present. But judgements of persistence are likewise crucial in enabling us to gain information about what one or another member (or sample) of a kind individually is like. They can enable us to know that this piece of copper, having been flexed repeatedly before, is likely to snap; that this dog is likely to bite; that this person will be able to translate a letter received from a German editor; that this car must have been in a particular sort of accident. Put abstractly, such judgements enable us to apply rough-and-ready “conservation principles” applicable to the kind to which it belongs – principles that say either that, for members of this kind, certain accidental features once acquired will tend to stay in place, or that certain accidental features once acquired will tend to mutate into particular other accidental features.7 These 7

Millikan, Language, Thought, and Other Biological Categories, pp. 291–92.

12

Two false friends of familiar objects principles enable us to predict and retrodict the career of an individual member of a kind, and to deal with that member in a way appropriate to it individually. How do we learn how to trace persistence across time for individual objects or portions belonging to a given kind – how do we learn which sorts of sameness, and which patterns of alteration, attest to numerical persistence? Again it seems plain that at least much of our knowledge here is gained empirically. Our hardwired ability at perceptual tracking enables us to detect short-term patterns of “before and after” in the accidental states that members of a given kind can assume. More sophisticated cognitive activities enable us to detect patterns that are more long-term: we observe enough about an object to identify it as a member of kind K , add to it knowledge about how members of K move (or do not move), sometimes together with partial knowledge as to how members of K alter qualitatively over time, and arrive at the knowledge that this member of K is the very member we earlier observed. Thus we use partial knowledge about patterns of continuity and alteration among K s to bootstrap our way to further such knowledge. 1.3

p u s h m i p u l ly u r e p r e s e n tat i o n s

The previous two sections give an account of the roles played in our cognitive economy by our judgements of kind-sameness and of numerical persistence. Does this account furnish reason for thinking that our schemes for making such judgements have truth conditions (that is, that the modal conventionalists are wrong) and that radically different schemes for making such judgements cannot be just as correct as our actual ways (that is, that the explosivists are wrong)? That is, does it provide the materials for a sparse realist understanding of our ideas about the general ways in which the world’s kinds are characterized and individuated, and the general ways in which the world’s persistences are limned? Here is what I will suggest. Our judgements of kind-sameness and of numerical persistence can play the roles that they do, in our cognitive economy, just because each is a species of what Ruth Millikan calls “pushmipullyu representations”8 – and because each sort of 8

Millikan, Language: A Biological Model, Ch. 9.

13

Familiar objects and their shadows sameness-judgement is a pushmipullyu representation, neither the conventionalist account of our schemes for making such judgements, nor the explosivist account, can be correct. This section will proceed in three stages. First, I will say what, for Millikan, a “representation” is. Then I will say what is distinctive about “pushmipullyu” representations. Last, I will show in what way judgements of kindsameness and of numerical persistence are themselves pushmipullyu representations. The subsequent two sections will then be able to point out aspects of these two kinds of sameness-judgements which conventionalism and explosivism fail to accommodate. A representation, for Millikan, is something that by nature stands between two devices or programs that are fashioned by natural selection or attuned by learning – one of which is the producer of the representation, the other of which is the consumer.9 The producer device may historically have been located in one organism, the consumer device in another, or the two may have historically been lodged in the same organism. In either case, the producer device must have been selected for, over the course of evolution or learning, as a causal consequence of its having produced various representations that often enough induced various specific responses in the consumer device, responses that often enough were useful for the consumer device – or more accurately, for the organism in which the consumer device was lodged. There are two main ways in which this can have happened. First, the representations may have built up parts of a cognitive map that the consumer device was assembling of its host’s surroundings – a map that would serve to steer and attune the host’s behavior to those surroundings, in ways that would lead to the satisfaction of the host’s desires. In this case we may think of the representations as indicatives – as signaling particular ambient states of affairs. Where this map was capable of being stored, for future enlistment in pursuit of a wide variety of purposes, we may think of the representations as producing beliefs.10 But second, the representations may have induced the consumer to launch and steer behaviors that were useful for the consumer and its host organism. In this case we may think of the representations as imperatives – as commanding particular behaviors. Where the imperatives were capable of being 9 10

Millikan, White Queen Psychology, Ch. 4. Millikan, Varieties of Meaning, Ch. 14.

14

Two false friends of familiar objects stored, for subsequent enactment once the host’s cognitive map registered circumstances in which enactment was particularly likely to succeed, we may think of the representations as being desires.11 But evolution fashions a more primitive sort of representation as well – one that is particularly important in guiding the behavior of animals simpler than us. This sort of representation is neither strictly indicative nor strictly imperative. Or better, it is from one vantage point indicative and from another imperative – just as the two-headed creatures that Dr. Doolittle called “pushmipullyus” had two faces.12 Such representations occur when, in the behavioral and cognitive repertoire of the consumer’s host, there is just one sort of response to a particular kind of state of affairs that the representation is designed to induce the host to make – and where that sort of response historically was useful for that host only when, and only because, it was geared to a state of affairs of that kind. In such a case we can, with equal justice, say either that it is the function of the representation to launch and guide just that sort of response, or that it is the function of the representation to signal just that one sort of state of aff airs. Consider, as an example, the famous neural signal S which the eye of a certain kind of frog (Rana pipiens) sends to that frog’s brain. The function of this signal, one might say, is to launch a tongue-snap at just the right angle to intercept a passing fly or other nutritious bit. For that is the only response to S that the producer of S can have been designed to induce in members of R. pipiens; members of this species do not keep notes, for later use, about likely trajectories and locations of fl ies. At the same time, the only sort of circumstance, in response to which it historically was useful for members of R. pipiens to launch a tongue-snap, was the passage nearby of a fly or other nutritious bit; tongue-snaps are useless unless they precisely correspond to the nearby passage of a fly. So we might equally well say that the function of S is to indicate, to represent, the nearby passage of a fly or nutritious bit. In one way, then, S is an imperative: it says, so to speak, “gulp thataway now!” In another, it is an indicative: it says, so to speak, “fly passing by thataway now.” In animals simpler than us, the indicative value of a pushmipullyu representation can only be represented by an occasion sentence, one 11 12

Millikan, Varieties of Meaning, Ch. 16. Millikan, Language: A Biological Model, Ch. 9.

15

Familiar objects and their shadows that reports a transient state of aff airs. But I shall presently suggest that this is due, not to the very nature of a pushmipullyu representation, but instead to the simplicity of the animals in which one most readily thinks of such representations as occurring. To gain a sense of the way simplicity constrains the indicative value of a pushmipullyu representation, consider another well-worn example from Millikan, namely the tail-slap of the beaver. When a beaver sees or hears a predator approaching, he will loudly slap his tail against the surface of the water. (A beaver will also sometimes do this when all that has happened is that a branch has fallen, or that human observers have spoken loudly to one another.) It seems extremely likely that natural selection fashioned the neural routines that produce the tail-slaps, and thereby the tail-slaps themselves. For the tailslaps have the effect of getting nearby beavers to dive beneath the water, thus escaping from would-be predators. The tail-slaps, in other words, serve to activate and guide a consumer device – a pattern of neural wiring that leads the beavers who hear the tail-slaps to dive as soon as possible after the slap occurs. It is because beavers historically made that one response to the tail-slaps, it seems likely, that the neural routines that produce the tail-slaps came to be replicated across the generations. Hence the tail-slaps can be viewed as imperatives that say, so to speak, “dive immediately, fellow lodge members!” But the only circumstance that historically rendered it useful for the lodgemates to make such a dive was the presence nearby of a predator. Hence the tail-slap can also be viewed as an indicative, the function of which is to say, so to speak, “a predator has just approached!” Now suppose that we were equipped with beaver bodies. Suppose too that our habits of tail-slapping could be attuned by ways that we learned to deal with our surroundings – so that tail-slaps might proliferate in part because of ways we learned to respond to them. We might learn to respond to tail-slaps in more ways than just diving. We would dive, of course – that would have to be our immediate response. But we could also quickly look in the direction of the slap, accumulate information about the area on the shore at which predators are most likely to approach, and undertake to obstruct access to that area by felling large trees over there. We could move our lodge to a place not visible from that part of the shore. The circumstance which rendered these sophisticated responses useful would be 16

Two false friends of familiar objects represented by a standing sentence: “Predators in general approach at thus-and-such area of the shore.” But actual beavers are quite different. They show notoriously little foresight in choosing which trees to fell, and they are highly inflexible in their lodge-building behaviors. Even if they were able to store information as to where, as a standing matter of fact, predators come to the water, they would not purposively obstruct access to that area, and would probably be unable to attune their lodge-building to that information. The only response that actual beavers can usefully make to tail-slaps is a response they make now, and that response is diving. Now for the suggestion that our two kinds of sameness-judgements are themselves pushmipullyu representations. The key to this suggestion is the idea that each type of sameness-judgement was selected for in a process of learning: we learned ways to use our linguistic and cognitive programs so as to produce these judgements in correspondence to particular circumstances in the world, and simultaneously learned ways that we could “consume” these judgements by making particular sorts of inferences, and these sameness-judgements got installed in our repertoire because they thus induced, usefully, just these sorts of inferences. To move to specifics, consider, fi rst, how our judgements of kind-sameness might have gotten established through such a pattern of learning. The role that such judgements play in our cognitive economy, I have claimed, is to get us to generate and to apply inductive inferences about the members of a kind – to infer that other members of a kind will prove to have the same properties, falling within certain sorts of properties, as the currently observed members display, and that currently observed members have such properties even though they are not presently displaying them. These are inferences that we learned to make not just at the time of hearing a judgement of kind-sameness, but at an unrestricted range of subsequent occasions. Thus judgements of kind-sameness may be seen as imperatives that say, so to speak, “on some occasions or many, whether now or in the future, judge that members of this kind are alike with respect to such-and-such sorts of properties.” But only one sort of circumstance can reliably have rendered it useful, on indefi nitely many occasions, for us to make such inductions. The particular properties of the relevant sorts, that we found to cluster together in observed members of the kind, must have clustered together for a reason: the occurrence of some of these particular properties must 17

Familiar objects and their shadows have causally controlled, or been controlled by, the occurrence of others in the cluster; and if such causal control obtained at all, it must also have been the case that the occurrence of variants on some of these properties would entrain, or be entrained by, the occurrence of variants on others of these properties.13 Thus our judgements of kind-sameness can with equal justice be viewed as signaling, as representing, just such causally controlled clustering. They can be viewed as indicatives which say, “the particular properties of such-and-such different sorts, which you will fi nd in these objects, cluster together for a causal reason – and variants on these properties will cluster in different objects in a causally controlled way.” Next consider our judgements of numerical sameness over time, of persistence. Such judgements matter in our cognitive economy because they tell us where to look to discern reliable patterns of “before and after” in the accidental states which the members (or samples) of a given kind assume. More fully, they direct us to discover such patterns, and also to apply these discoveries to individual members (or samples) of a kind – to predict or retrodict what one or another member of the kind individually is like. But we learned that inferring how in general the members of some kind respond to the accidental states that befall them, and applying what we infer to the careers of individual members, was useful not just at the particular time that we heard (or muttered to ourselves) a judgement of numerical persistence. We learned to respond to judgements of numerical persistence – whether by making general inferences or by applying the inferences to individuals – on indefi nitely many occasions, whether at the time of the judgement or later. Thus such judgements function for us as imperatives that are not time-restricted. They say, so to speak, “expect that the accidental state which you currently observe in this object was causally controlled by earlier accidental states of that object that you either did observe, or could have observed, in a pattern of causation that is typical of the kind to which this object belongs; expect too that any currently observed state in turn is tied, by kind-spanning patterns, to particular subsequent states.” At the same time, there was just one sort of state of affairs to which such an imperative could usefully have been geared – just one state of aff airs 13

Elder, Real Natures and Familiar Objects, pp. 89–92; Woodward , Making Things Happen.

18

Two false friends of familiar objects that could reliably have rendered it useful for us to expect kindspanning patterns of “before and after” among accidental states. The imperatives embodied in judgements of persistence must have corresponded to that state of aff airs, in order for our responses to those imperatives to be successful enough to have caused those judgements to get installed in our repertoire. That state of aff airs is that members of the kind which we are observing are disposed by their nature – by their essential properties – to respond to the advent of particular accidental states by retaining those states or by passing into specific other accidental states. Judgements of numerical persistence can thus be said to be indicatives which say, “the objects which you are dealing with are disposed, by virtue of the essential properties characteristic of the kind to which they belong, to pass from certain accidental states to particular later accidental states in causally governed patterns.” 1. 4

explosivism

If sameness-judgements of these two kinds are pushmipullyu representations, then explosivism fails to acknowledge their imperative aspect, and conventionalism fails to acknowledge their indicative aspect. In the present section, I undertake to demonstrate the fi rst of these claims. But fi rst let me say just which views I include under the rubric of “explosivism.” One form of explosivism focuses on numerical sameness across time; the other, on sameness in kind. The former has more proponents than the latter, and so I shall begin with it. The strongest version of explosivism about numerical persistence is Ted Sider’s position that every “assignment” picks out a perfectly real diachronic fusion – or, in Sider’s terms, a perfectly real minimal D-fusion.14 An “assignment” is a function from times to material entities existing at those times; a minimal D-fusion is something that exists at only the times in the assignment. One good example of a D-fusion is Lewis’s “trout–turkey.”15 This is an entity that for a certain period of time is a trout swimming in a brook, and is, for a range of times beginning immediately after this period, a turkey strutting in a pen. The claim that every minimal D-fusion is as real as any other 14 15

Sider, Four-Dimensionalism, p. 133. Lewis, Parts of Classes, pp. 79–81.

19

Familiar objects and their shadows entails that, during the fi rst period, many different objects exactly coincide with the trout–turkey. One is an object that goes on being a trout even after the period in question, indeed a trout spatiotemporally continuous with the one originally swimming in the brook; another is an object that is a desk across a range of times immediately succeeding on the fi rst period; etc. There is an “explosion” of temporarily coincident objects, differentiated by having quite different careers. This “explosion” is sizeable indeed. An assignment can map a single time onto material entities that are spatially removed from one another, and so a D-fusion can be, at many points in its career, spatially discontinuous. Sider also remarks that, barring special objections that he has not envisioned, an assignment may incorporate times that are not continuous with one another – which entails that a D-fusion may exist for a time, take a vacation from existence, and then return into existence at a later time.16 And even the temporally continuous D-fusions which at any time occupy only a single place are capable of suddenly jumping from one place to another, as Lewis’s trout–turkey does. One of them, for example, is located just where the trout is, during the initial period, and is located, a single second later, at a place 187,000 miles removed – thus appearing to have moved faster than the speed of light.17 Many D-fusions cease to exist at a particular time. Often, when a D-fusion ceases to exist, the world is qualitatively different from how it was during that D-fusion’s existence. Even so, the D-fusion does not cease to exist because some qualitative alteration occurs. It ceases to exist simply because “its time was up” – because the last time over which its assignment is defi ned has elapsed.18 Similarly, a D-fusion does not begin to exist when it does because certain qualitative alterations brought about its existence. Its coming-to-exist does not reflect or grow out of qualitative alterations that occurred immediately before, but only because “its time has come” – because the fi rst time over which its assignment is defi ned has arrived. D-fusions, in the phrase of one critic, come “into existence ex nihilo”19 – and their ceasing to exist can be just as sudden and qualitatively unprovoked. 16 17 18 19

Sider, Four-Dimensionalism, p. 136. Balashov, “On Vagueness,” pp. 527–28, and Hawthorne, Metaphysical Essays, Ch. 6. Koslicki, “Crooked Path,” p. 127. Thomson, “Parthood and Identity across Time,” p. 213.

20

Two false friends of familiar objects It follows that D-fusions do not, strictly speaking, have qualitative persistence conditions. A D-fusion ceases to exist not because of what, qualitatively, is happening to it in the last moment of its existence, but just because that moment is the last one over which its assignment is defi ned. But there are other versions of explosivism about numerical persistence that do treat the persisting objects that they recognize as having qualitative persistence conditions – as having certain properties essentially. Consider Sosa, who maintains that any combination of matter with form yields an object, defi ned essentially by just that combination.20 A particular snowball, for example, is the combination of some snow with a roughly spherical shape. And exactly where that snowball is located there is also an object that bears a more broadly defi ned shape: it is a “snowdiscall,” an object that essentially is shaped anywhere along a spectrum running from spherical to disk-shaped. Different shapes, different persistence conditions. A snowdiscall can survive being squashed, while a snowball cannot; snowdiscalls are more durable than snowballs. But Sosa’s position also entails that located exactly where the snowball is, there may be an object that is far more brittle than the snowball. If, that is, the snowball is perfectly spherical, there is an object having perfectly spherical shape, and it ceases to exist as soon as the snowball undergoes the loss of a tiny peripheral chunk. Along the same lines, Hirsch speaks of 21 objects that exactly coincide with cars, but that are far less durable than cars: an “incar” ceases 20 21

Sosa, “Existential Relativity,” p. 142. It would be more straightforward to say that Hirsch “affi rms the reality of ” incars and outcars , klables and cpersons , but that would oversimplify Hirsch’s actual position. His position is that no one has clearly shown why we would not fare just as well as we actually do if we spoke of such strangely persisting objects instead of the more familiar cars, tables, and persons. But it may really be that, mind-independently, only the cars and the tables and the persons are there. Even if the world were populated by objects that traced out such familiar courses of existence – and sparsely so populated, not also populated by incars and klables and cpersons – strangely individuating languages might, for all anyone has shown, work just as well as our familiar language (Hirsch, Dividing Reality, Chs. 3–4; Hirsch, “Rules for a Good Language,” pp. 699–700). Just so, Hirsch’s position on kinds is that even if the world were mind-independently sorted into such familiar kinds as apples and oceans (see below in the text), and not also into strange kinds such as “apceans” (a single kind which encompasses both apples and oceans), still strangely classifying languages might, for all anyone has shown, afford all the practical and cognitive mastery that we actually possess. So it is not accurate to say

21

Familiar objects and their shadows to exist when it ceases to be in a garage, and an “outcar” ceases to exist when it is pulled into a garage.22 Hirsch’s “klables” and “cper sons” move quite suddenly, just as trout–turkeys do: a klable located in the dining room will instantaneously move, at the very stroke of noon, into the living room.23 Mark Johnston 24 recognizes objects that are united only by the mutual gravitational attraction of their parts – objects which are even more durable than snowdiscalls. Thus there is a veritable explosion of views that are explosivist with respect to numerical persistence. Explosivism with respect to the kinds into which the world’s objects fall is less often explicitly endorsed. The best examples are provided by the extensions of kind-terms in Eli Hirsch’s “strange languages.” One familiar kind encompasses all and only the world’s apples. But this kind is overlapped by another which includes, in addition to the apples, all and only the world’s oceans; yet a third kind includes all the apples and all the cars.25 There are also, as we have just noted, strangely fi ne-grained kinds. The kind incar includes all the world’s cars located in a garage; cars located outside of garages belong to a different kind , outcars. Sosa’s strangely persisting objects , for their part, belong to kinds both coarse (e.g. snowdiscalls) and fi ne (e.g. snow-perfect-spheres). Thus we have an explosion both in instances of kind-sameness and in instances of numerical persistence. Do any consequences follow from the claim that there are these myriad unforeseen instances? Might it matter to us that, coinciding with the familiar objects recognized by common sense and empirical science, there are also many strangely persisting objects? Would it matter that familiar objects are joined by such strange instances of kind-sameness, as well as by the more familiar instances?

22 23 24 25

that Hirsch simply is an explosivist – and my word choice is meant to fall just shy of saying that. (Similarly, it is not right to say that Hirsch is a conventionalist – but that too is sometimes said, just as it is sometimes said that he is an explosivist.) What is accurate is the contention that Hirsch’s strange languages can be viewed as illustrating one way in which explosivism might be true. That is the contention I mean to be communicating in the text. Hirsch, Dividing Reality, p. 26. Hirsch, Dividing Reality, p. 26. Johnston, “Hylomorphism,” §17. Hirsch, Dividing Reality, pp. 24–25.

22

Two false friends of familiar objects These questions are best approached separately – and so let us begin by asking how it might benefit us to recognize strangely persisting individual entities. Explosivists about persistence are entirely willing to concede that the strangely persisting individual objects in which they believe might be of little interest from the standpoint of our familiar human concerns. They might be of no practical importance – no one has any need to keep track of individual trout–turkeys, and even the most avid participants in snowball fights have no use for knowing how many snowdiscalls are lying about. The strangely persisting objects may not even be objects for which empirical science need make allowance – certainly empirical science does not concern itself with objects that travel faster than light, or that suddenly undergo inexplicable increases in their mass. The judgements that trace the careers of strangely persisting objects might well play no role, then, in our actual cognitive economy. But those judgements would still be true – those strangely persisting objects still really are out there, say explosivists. If judgements of numerical persistence are pushmipullyu representations, however, then they essentially have an imperative function. They essentially function as directives, enjoining and guiding inductions that pinpoint causally governed patterns of continuity and alteration which the persisting objects can be counted on to display. To speak of judgements of numerical persistence as being “true,” under this imperative aspect, is to treat them as directives that are valid or well founded, directives that it would pay us to heed. But the judgements of persistence which explosivists offer could not usefully be treated as enjoining such inferences. They would, by the explosivists’ own admission, play no particular role in our actual cognitive economy at all. But then to the extent that they do essentially have the function of imperatives, they would be imperatives that are best ignored. They would not, in the relevant sense, be true. Next consider the various strange judgements about kind- sameness that, according to explosivists, we might truly make. Might we usefully judge that an apple is the same in kind as the oceans, or the same in kind as cars? Are there things to be learned, or ends to be gained, by lumping together masses of snow that are spherical or squashed, hand-sized or huge? If judgements of kind-sameness are pushmipullyu representations, they are partly imperatives, which direct us to expect that, for certain ranges of properties, whatever instance of 23

Familiar objects and their shadows each kind we discover to be present in some members of a kind will be present in all other members as well. For such directives to be “true” is for them to be well founded, directives which it would pay us to heed. Would it pay us to inspect apples and run inductions over cars or oceans? What might the study of a huge, round, fl attened mass of snow tell us about snowballs? Let me start fi rst with the Sosa case. Snowdiscalls that we have not observed can be counted on to be the same, with respect to a broadly defi ned and determinable shape, as the snowdiscalls that we have observed. But there is no other range of noteworthy properties from which snowdiscalls, as a class, all alike make a distinctive selection. That is, nothing interesting goes together with having just that broadly defi ned shape – any more than something interesting goes together with having the less broadly defi ned shape of being spherical-to-half-flattened. To be sure, any snowdiscall is some snow (as is any mass of snow that is spherical-to-half-flattened), and snow everywhere is characterized by particular instances of certain sorts of properties. Snow has a particular color, a temperature falling in a certain range, and a propensity for clumping together in various ways at various temperatures and densities. But these are reasons for endorsing the judgement that snow everywhere is matter of one particular kind – for thinking, in other words, that this judgement directs us usefully to expect kind-spanning samenesses of particular sorts. These are not reasons for thinking that snowdiscalls are objects belonging to a common kind. At the same time, if we say, of the original compact mass of snow, that it is the same in kind as other snowballs, we may arguably have succeeded in assigning that mass to a genuine kind. For snowballs are artifacts. New snowballs get fashioned on the model of previous snowballs, as a causal consequence of uses that previous snowballs served. They served this use – and the new snowballs will serve this use – in virtue of being sized just so the hand can grasp them, shaped just so that they can be thrown in straight lines, and compacted enough so as not to fall apart in fl ight.26 Sosa’s crisscrossing kinds seem to be too narrowly defi ned. Hirsch’s kinds seem to be too broad. How could we usefully frame an expectation that, for certain ranges of properties, whichever members we fi nd in observed apples can be counted on to show up in all oceans – or 26

Elder, “On the Place of Artifacts in Ontology.”

24

Two false friends of familiar objects all cars? Hirsch does have an answer: the property-ranges in question are populated by disjunctive properties. One such property might be represented by the predicate “contains seeds or contains fi sh”; another, by “has red skin or is salty”; etc. But it is natural to suspect that such so-called “properties” are concocted, and that drawing inferences as to where they are found is a useless undertaking. In the next chapter I undertake to confi rm this suspicion. 1.5

Con v entiona lism

Concerning conventionalists, my contention is that they fail to acknowledge the indicative aspect of judgements of kind-sameness and of judgements of numerical persistence. I will deal with kindsameness fi rst, numerical persistence second. But before beginning I will briefly sketch the conventionalist position. Conventionalists maintain that terms purporting to designate kinds in nature – and even terms purporting to name individual members of kinds – are never purely denoting expressions.27 Any such term is analytically associated with a certain descriptive content, which ties the reference of that term to one kind or another within a certain family of kinds. Thus it is analytic that if “tigers” refers at all, it refers to a kind of animal; that if “water” refers, it refers to a chemical kind; that “gold” refers, if to anything, to one of the elements; that if “beef ” names anything, it names a kind of food. The descriptive content analytically associated with a kind-term may leave much undisclosed about the essential nature of the particular kind which it designates. But it will at the least indicate which sorts of properties it is, by which that kind is characterized, and differentiated from other kinds within its family. That much is analytically packed into the name itself for that kind. My fi rst objection concerns judgements of kind-sameness: these are pushmipullyu representations, I contend, but conventionalism fails to acknowledge their indicative aspect. In consequence, conventionalism misses out on the truth conditions of these judgements. But to focus this objection, we must note that even for conventionalists, judgements of kind-sameness do have some truth conditions. 27

Sidelle, Necessity, Essence, and Individuation, pp. 67–68 and 85; Thomasson, Ordinary Objects, pp. 38–44 and 48–53.

25

Familiar objects and their shadows Any judgement of kind-sameness incorporates judgements that the objects (or portions of matter) so related are similar in several, even many, qualitative respects. These similarity-judgements are ordinary empirical judgements, and unproblematically have truth conditions. But judgements that two or more objects are the same in kind go beyond mere judgements of similarity. One way to put it is that judgements of kind-sameness assert that the objects are similar not just in casual ways, but in ways that encourage inductions over those objects. The traditional understanding of kind-membership – with which conventionalists agree, as we shall soon see – provides a more crisp formulation of this “going beyond”: it says that objects alike in their kind-membership are thereby alike in their persistence conditions. The crucial question, for the objection I mean to make, is how we go beyond ordinary empirical judgements of similarity to judgements of kind-sameness. Do we do this on the strength of additional empirical fi ndings? Conventionalists say No. Instead we do it, they say, by bringing to bear our conventions for characterizing and differentiating the kinds within the family to which the objects in question belong, conventions that count certain sorts of similarities and not others as constitutive of kind-sameness within this family. These conventions are incorporated in the descriptive content analytically associated with the kind-terms we are using. The statements that articulate these conventions are – just because they are mere articulations of conventions – analytically true. So their being true does not require that the world be one way or another; they are true regardless of how the world is. As one prominent conventionalist puts it, these statements “place no demands on the world”; they “do not require truth-makers at all”; concerning any one such statement, “nothing is required of the world to make it true.”28 But consider that judgements of kind-sameness are pushmipull yu representations. They direct us to do something, and they signal the occurrence of the very state of aff airs which historically has made doing that something useful. A judgement of kind-sameness directs us to expect, for an object that falls somewhere within a certain family of kinds, a distinctive connection between the ways that object selects from each of several ranges of properties. If the object 28

Thomasson, Ordinary Objects, pp. 67–68.

26

Two false friends of familiar objects selects property a from Range 1, we are directed to expect that there will be just one property that it selects from Ranges 2, 3, and 4. If we already have examined objects belonging to this kind-family that have property a , we are instructed to expect in advance of detailed observation that this a -bearing object will have the same properties as those already examined objects have, from out of Ranges 2, 3, and 4 – observing that this object has a will amount to observing that this object has those other properties. If we have not examined other a -bearing objects belonging within this general family of kinds, we are instructed that examining this object more closely will afford us instantaneous knowledge about other a -bearing objects belonging within this family. What sort of state of aff airs might render it useful for us to heed such directives? Only one. There must be a reason why the presence of property a in Range 1, in an object falling within this general family of kinds, clusters together with the presence in that object of particular other properties found in each of Ranges 2, 3, and 4. The clustering cannot be accidental, but must be causally underlain – the presence of some of the properties in this cluster must causally control the selections made from the other property-ranges. And if this is the case, there will be a pattern of covariation on that cluster of properties. The presence of individual variants on a, from out of Range 1, will have to entrain the presence of particular individual variants on the properties selected from each of the other ranges. The indicative value of a judgement of kind-sameness, then, is just that there is this causally regulated clustering of particular sorts of properties among the members of the kind. This is a state of aff airs out there in the world, entirely independent of our conventions. It is a truth condition for a claim of kind-sameness, and conventionalism ignores it. Consider, next, our judgements of numerical persistence. Conventionalists actually hold that two dimensions of individuation are strictly conventional – and that both dimensions are structured by conventions written into the descriptive content of the terms we use to designate the kinds within a particular family of kinds. First, as we have just seen, there is the way the kinds themselves within a family are individuated. But second, there is the way that individual objects (or samples of matter) that fall within a general family of kinds are individuated – in particular, what it takes for an object (or 27

Familiar objects and their shadows sample) that exists at one time to be numerically the same object (or the very matter) as exists at a subsequent time. This too is a strictly conventional matter, conventionalists say.29 One thing that our conventions require is that the object (or sample) at the later time belong to the same kind as the object (or sample) at the earlier time. But our kind-terms encapsulate additional conventions that constitute what it is for the later object to be, not just a K (or some M ), but the very K or M that existed earlier. These conventions fi x persistence conditions for the members (or samples) of each kind in the family. The statements that articulate these persistence conditions – just because they are mere expressions of our conventions – are true analytically. They do not have truth conditions. But, I say, judgements of numerical persistence are pushmipullyu representations. As imperatives, they direct us to discover particular patterns of continuity and alteration among the accidental states which the persisting objects or samples in a kind assume over time. As indicatives, they signal the occurrence of the one sort of state of aff airs that renders it useful to respond to these directives. That one state of aff airs is, again, a matter of causal control. It is the fact that the essential properties that members or samples of a particular kind have, by virtue of belonging to their kind, make it no accident that there are repeating patterns of continuity and alteration among accidental states. The essential properties ground fi xed dispositions by which particular newly arriving accidents cause particular accidental sequels. That such dispositions obtain, among the items which we judge to persist, is a wholly mind-independent matter. It is no mere artifact of our conventions. This is a truth condition for our judgements of numerical persistence, and conventionalism fails to acknowledge it. 1.6

Sum m i ng u p

Our judgements of kind-sameness and of numerical persistence have particular purposes – purposes having to do with inductions. Explosivism does not treat these purposes seriously enough. These two kinds of sameness-judgements often succeed in their purposes 29

Sidelle, Necessity, Essence, and Individuation, pp. 52–57; Thomasson, Ordinary Objects, pp. 56–59 and 157.

28

Two false friends of familiar objects just because they often are geared to world-given clusterings of properties – where the property clusters in question underlie reliably repeating patterns of continuity and alteration among accidental properties. Conventionalism does not treat seriously enough the ways in which these sameness-judgements are geared to such world-given phenomena. Now in calling these sameness-judgements “pushmipullyu representations,” I have been claiming that it is constitutive of their very nature that they serve inductive purposes, and do so by virtue of corresponding to world-given clusterings of properties. Thus my charge is that explosivism and conventionalism each ignores something constitutive of the very nature of our judgements about these two forms of sameness.

29

2 Conventionalism as ontological relativism In the previous chapter I defended a realist position on kind-sameness and on numerical persistence. I held that nature’s kinds typically fall into families; that properties of certain sorts are common to the members of each such kind and differentiate the kinds within each family; and I briefly sketched the ways we learn, empirically, which sorts of properties are, in this way, constitutive of kind-membership. I held also that retention of properties of these same sorts sets up persistence conditions for the members (and samples) of nature’s kinds – meaning that persistence conditions, too, are things we learn from nature. Both these claims are contested by the view I called “conventionalism.” Conventionalists, as I said, hold that descriptive content written into the very meanings of our sortals and matter-names fi xes which sorts of properties are determinative of kind-membership and of persistence. But this basic conventionalist message can be articulated in either of two quite different ways, and thus far I have examined only one. The articulation I have so far examined says: since it is the very meanings of our sortals and matter-names that fi x which sorts of properties determine kind-membership and mark out persistences, the truths as to which properties these are, in the case of each natural kind, are analytic, and they do not have truth conditions. They are true, but they are not true in virtue of anything. A different way of articulating the basic conventionalist message is to say: it is in virtue of our adhering to our conventions for individuating kinds, and for tracing persistence on the part of members (or samples) of those kinds, that certain sorts of properties and not others qualify as conditions on kind-membership and on persistence. The truths as to which properties these are are true in virtue of something. They are true in virtue of our conventions, our practices. These conventions are written into the descriptive meaning that attaches to our sortals and matter-names. 30

Conventionalism as ontological relativism The distinction between these articulations is subtle but significant. Take, for example, the claims “necessarily, water anywhere has molecular structure H 2O” and “any individual sample of water can persist only so long as it retains molecular structure H 2O.” All conventionalists think of such claims as combining two elements. The former, for example, combines “necessarily, water has whatever deep-lying explanatory structure empirical science may reveal it to have,” and “empirical science reveals the deep-lying explanatory structure of water to be the microstructure H 2O.” But for the version of conventionalism that we have so far examined, only the latter of these two claims has a truth condition at all. Hence the entire truth condition for, “necessarily, water anywhere has molecular structure H 2O,” lies in a mind-independent fact about water which chemistry discovers. This sort of conventionalism, therefore, can at least claim to have no quarrel with realism about objects and matters.1 Even if minds were altogether removed from the world’s history, the full truth-maker for that claim would remain in place. The other articulation of conventionalism, in contrast, is frankly antirealist. It holds that we (by our practices, our conventions) are responsible for the status, as essential, of whatever microstructure water may turn out to have – and for the status, as a persistence condition for particular samples of water, of that same microstructure. It is the frankly antirealist version of conventionalism that is more widely endorsed in recent philosophy. In particular, it is embodied in a widespread view that might appropriately be labeled “ontological relativism.” This version of conventionalism is even more obviously at odds with realism about the world’s objects and matters – and a fortiori, with realism about the world’s familiar objects and matters – than is the conventionalism discussed in chapter 1. This chapter therefore focuses on this widespread, alternative form of conventionalism, namely ontological relativism. This chapter examines the motivation for ontological relativism, and argues that the challenges which it faces are at least as weighty as the ideas that motivate it. “Ontological relativism,” as I am using the phrase, contends that which objects it is that populate the world is a relative matter – that different answers to which objects (and which 1

But see Elder, “Conventionalism and Realism-Imitating Counterfactuals,” and Sidelle, “Modality and Objects.”

31

Familiar objects and their shadows kinds of matter) there are in the world are true, relative to different conceptual schemes, and that there is no scheme-independent fact of the matter on this question. The central motivation for this view is the conventionalist thought that there is something particularly “soft,” particularly unobjective, about the two forms of sameness discussed in the previous chapter. The fi rst of these is sameness in natural kind, and the second is numerical sameness across time – the sort of sameness embodied in an object’s persisting. Proponents of ontological relativism argue that the facts as to where either sameness obtains are fi xed wholly by our “conceptual scheme.” More precisely, the facts as to where sameness in kind obtains are fi xed wholly by the application conditions for our sortals and our noncount nouns for kinds of matter.2 The facts as to where one and the same object persists (or where the same matter, e.g. the same gold, still persists) are fi xed wholly by the re application conditions for our sortals and our matter-names – that is, by the rules that govern our judgements of the form “the same K is still continuing to exist.”3 Kind-sameness, and numerical persistence, are in this sense regarded as projections of our conceptual scheme. 2.1

s a m e n e s s a n d o bj e c t s

But then why shouldn’t the view in question be called “samenessrelativism”; why does it merit the title “ontological relativism”? The answer to this question shows that there is an important difference between the present view and more familiar forms of relativism – relativism about ethics, for example, or about epistemic justification, or about history. The reasoning behind any form of relativism begins with the observation that incompatible properties, from within a certain range, appear to characterize one and the same item in the world. Depending upon one’s body temperature, one and the same object can seem both hot and cold; depending upon one’s moral standards, one and the same action can seem both permissible and blameworthy; depending upon the context, one and the same judgement, formed in response to the very same sort of evidence, can seem either 2 3

Much as in Thomasson, Ordinary Objects, pp. 39, 41, 56, 157. Much as in Thomasson, Ordinary Objects, pp. 40, 57–59.

32

Conventionalism as ontological relativism warranted or reckless. The relativist then argues that these relations to a point of view (or situation or context) enter into the very content of what is said when one ascribes the (seemingly) incompatible properties. That is, the relativist claims that statements attributing these properties to items in the world involve a suppressed (or at least easily overlooked) argument place; that they are truth-evaluable only when a value is supplied for that argument; and that a relevant value is supplied only by specifying a particular point of view or context or situation. Further, the relativist claims that, as a general matter, truth for ascriptions of the properties in any of the relevant ranges is determined by which perspective or context is assigned to the suppressed argument place – that different such ascriptions are rendered true by varying the perspective or context that is assigned. Thus if a given action is permissible relative to one moral standard, there must be another standard relative to which it is not permissible – and relative to that other standard, at least some actions not permissible relative to the fi rst must be permissible. What then of the position that sameness in kind, and numerical sameness across time, are fi xed wholly by the semantics of our sortals and our matter-names – what sort of relativism might follow from that position? Might the relativism that follows be about kind- sameness and persistence themselves? But here a disanalogy from familiar forms of relativism quickly shows itself. We can readily understand how one and the same object might be hot relative to one perspective and cold relative to another – or how one and the same action can be morally permissible relative to one point of view and shabby relative to another – because we can clearly understand what it is for that object (or action) to be the same object (or action) standing in both relations. It is not so easy to understand how one and the same object might, relative to different systems of sortals and matter-names, belong to different natural kinds – how it might, in other words, be characterized by different natures. Even less easy is it to understand how one and the same object might, relative to such different “conceptual schemes,” persist over different courses of existence. Suppose, to start with the second case, that relative to one system of sortals, a particular object is undergoing a change that will end its existence, while relative to another system that same object is undergoing a mere alteration in a continuing course of existence. Do we really understand how it can be that the same object is ceasing 33

Familiar objects and their shadows to exist, relative to one index, and simultaneously continuing to exist, relative to the other? Off hand, it seems that perishing is not something that you can do just from one perspective, or in some one respect, or from one point of view – it seems rather to be an all-ornothing aff air. That is, whether an object persists or perishes seems to enter too deeply into the very identity of that object, for it to be possible for the very same object to do both simultaneously, relative to different perspectives. For that matter, the natural kind to which an object belongs seems also to enter too deeply into that object’s identity, for one and the same object to belong to different kinds – to have different natures – relative to different perspectives. It seems hard to understand how an object can have a particular nature just in one respect or from one point of view; having a particular nature seems to be an all-or-nothing aff air. If the point seems unclear when put in these general terms, it may help to consider concrete examples. Eli Hirsch, as I noted in chapter 1 (note 21), speaks of a “strange language” in which a typical sortal is “apceans.”4 From the standpoint of our system of sortals, an apcean is either an apple or an ocean. Now, it is one thing to say that in the very areas of the world in which we suppose there are apples, there really are entities characterized by a radically disjunctive nature – so characterized, regardless of the point of view from which they are regarded. It is quite another to say that there are in the world apples – things that are by nature small, sweet fruits – but that, relative to a “strange” set of sortals, these very objects, apples, have a nature that encompasses an option of being vast, inedible, salty, and aqueous. In what way can it be the same objects that have both natures, even if one says that nature-having obtains only relative to a particular perspective – doesn’t an object’s nature enter too deeply into its very identity, for the claim that it is the same object that stands in both relations to be coherent? Or, consider another of Hirsch’s languages designed to illustrate not “classificatory strangeness” – strange ways of discerning where kind-sameness obtains – but rather “individuative sameness.” One sortal in this other language, as I remarked in chapter 1, is “incar.”5 From the standpoint of our actual system of sortals, an incar is a car that continues to exist only so long as it is in a 4 5

Hirsch, Dividing Reality, pp. 24–25. Hirsch, Dividing Reality, p. 26.

34

Conventionalism as ontological relativism garage: once backed out of a garage, an incar thereby ceases to exist. (If the resultant entity later enters a garage, it is a numerically distinct incar that does so.) It is one thing to say that the areas of the world that we take to be populated by cars are in fact populated by incars. It is quite another to say that there are in the world cars, and that, when backed out of garages, such objects in one respect continue to exist, and in another respect are destroyed. Being destroyed seems not to be something that can happen to an object just in one respect, or just from one perspective. Rather it seems that either destruction happens, or it doesn’t. The idea that an object’s identity is inseparable from its membership in a particular natural kind, and from its being such as to survive across certain sorts of changes and not others, must itself be kept separate from familiar thoughts about vagueness. It is a familiar thought that the precise point at which an object goes out of existence – or comes into existence in the fi rst place – may not be objectively fi xed, and that to the extent that the object’s existence does begin and end at precise points, it does so only relative to one or another out of a range of admissible precisifications. Thus there may seem to be no perspective-independent fact of the matter as to the precise moment at which a person begins to exist, or perishes, and no precise end points for the existence of a particular sunset. But these thoughts are persuasive just because it seems clear that a person or a sunset must, to go on existing, retain certain sorts of properties – and that the properties admit of borderline cases. It would be quite different to claim that which sorts of properties a particular object must retain to go on existing – which sorts of properties are encompassed in that object’s nature – is likewise not objectively fi xed, but obtains only relative to one or another “conceptual scheme.” It is this claim that makes one ask: in virtue of what is it one and the same object that has, relative to the different “conceptual schemes,” different persistence conditions? To see the force of the question, imagine – or try to imagine – that object o has undergone a change such that, relative to Conceptual Scheme 1, o has merely altered, while relative to Conceptual Scheme 2, o has ceased to exist. If o has undergone just that change – o itself, one and the same object as existed before the change – then o does still exist, and Conceptual Scheme 2 misrepresents what has happened. And if o has not itself survived that change, then Conceptual Scheme 1 misrepresents what has happened. 35

Familiar objects and their shadows It is unsurprising, then, that the “ontological relativism” which this chapter examines locates its claim of relativism “one level up”: not at the level of particular objects such as apples and cars, but at the level of the world. What our version of “ontological relativism” (henceforth OR) claims is that relative to one system of sortals and matter-names, the world is populated by objects (and portions of matter) that fall into such-and-such natural kinds, and have such-and-such persistence conditions; relative to another such system, the world is instead populated by objects (and portions) that fall into thus-and-so natural kinds, and have thus-and-so persistence conditions; and that there is no system-independent fact of the matter as to which objects it is, by which the world is populated. The relativism here is not asserted at the level of particular objects. That is, OR makes no claim that when we say, of a particular object, that it belongs to a particular natural kind, or that it persists only so long as particular conditions are satisfied, we say something that incorporates a suppressed argument place. What OR does say is that claims to the effect that the world (or a particular area of the world) is populated by just thus-and-so sorts of objects do incorporate a suppressed argument place. Such claims are true (or false) relative to just one system of sortals, among a plurality of possible systems. There is no system-independent fact of the matter as to how the world is populated. 2. 2

t h e “s o f t n e s s” o f s a m e n e s s i n k i n d a n d n um e r ica l sa m e n e ss

What reason is there for believing the motivating thought behind OR – that the phenomena of sameness in kind, and of numerical sameness across time, are fi xed wholly by the semantic conditions for application and reapplication of our sortals and matter-names? Why not suppose instead that kind-sameness and persistence are mind-independent phenomena? Many arguments for the motivating thought have been offered, both by philosophers who are proponents of OR and by philosophers who are not. Here there will be room for detailed consideration of only one such argument. But before identifying this argument, let me speak briefly about some others. Some philosophers consider it fairly obvious that sameness in kind, and numerical sameness across time, are language-dependent phenomena. These philosophers offer rather casual arguments for the 36

Conventionalism as ontological relativism thesis. Putnam, for example, speaks as if the thesis can be established by a simple reductio. Some philosophers, Putnam observes, think that if we can get a baptismal utterance of “horse” to refer to some individual horse – which perhaps can be done by ostension of the horse, or by trading upon a causal relation between that horse and that tokening of “horse” – we thereby get the sortal “horse” to refer to a particular natural kind. But what connects the ostended horse with all the others? These philosophers evidently think that the world throws a lasso that falls around all the horses, and only the horses.6 But the idea that the world throws such a lasso is simply naïve! It is equally naïve to suppose that the world presents us with individual objects that come “ready-made” – that come with courses of existence neatly packaged by end points supplied by the world.7 Alan Sidelle offers a more concerted argument for the motivating thought behind OR – and the argument is important enough that I must at least sketch it.8 Sidelle’s argument is epistemological. It can seem, Sidelle notes, that we discover empirically which features it is, that unite together the members of a natural kind. But in fact this discovery is never purely empirical. Instead it merely implements schematic semantic rules that we supply, rules that govern the use of sortals or matter-names that belong to a common family. Thus for the names we use for the world’s liquids, we may subscribe to a rule that says: it is correct to apply one such name to samples at different spatial locations, or across different possible worlds, just in case the same particular chemical composition is present. We then empirically examine some paradigm (or at least uncontested) samples of, say, water, and learn that they are all characterized by the microstructure H 2O – and it thereby appears that we have discovered, empirically, that all samples of the kind water are united by that microstructure. But the relevance of what we have discovered, for characterizing that natural kind, derives wholly from our schematic semantic rules. Our semantic rules for using names for the world’s liquids incorporate a “blank check” for the world to fi ll in; the world fi lls the check with “H 2O”; but the check has value only because we have signed it. In other words, what it is for two or more samples of liquid to be the 6 7 8

Putnam, Reason, Truth and History, pp. 53–54. Putnam, “Why There Isn’t a Ready-made World.” Sidelle, Necessity, Essence, and Individuation, Chs. 2–4.

37

Familiar objects and their shadows same in kind is fi xed by our semantic rules. It is even correct to say that it is fi xed wholly by our semantic rules: for precisely what those rules require is that we obtain an entry from empirical fi ndings. Might purely empirical research apprise us of where there is, in the world, sameness in kind – and where there occurs, in the world, numerical sameness across time? That question lies at the heart of yet another argument for the motivating thought behind OR, the argument on which I shall mainly focus. Suppose, for reductio, that the phenomena of kind-sameness and of persistence are truly mindindependent. Then either these phenomena are empirically detectable, or they are not. The latter idea raises obvious alarms – why is it that we seem to be so skilled at tracing the boundaries of nature’s kinds, and at tracking persistences? – and might on verificationist grounds be rejected as incoherent.9 But if these phenomena can be empirically detected, this argument says, we face the problem that they can be detected in too many places. Many different answers for where kind-sameness and persistence obtain might all be proposed, and each of them would be as predictive, and as empirically serviceable, as all the others. But these answers are incompatible. They cannot all correspond to mind-independent phenomena. Here then is what I shall treat as the “master argument” for OR. (1) All manner of nonstandard schemes for discerning sameness in natural kind, or numerical sameness across time, could be just as serviceable and just as empirically predictive as are our actual schemes. (2) There is no mind-independent phenomenon of kind-sameness, and no mind-independent phenomenon of persistence; these phenomena do obtain, but obtain only in virtue of one conceptual scheme or another [from (1)]. (3) There is no mind-independent fact of the matter as to which objects it is by which the world is populated; instead, true answers to this question are true only relative to one conceptual scheme or another [from (2)].

Why doesn’t (2) say rather that kind-sameness and persistence obtain only relative to one conceptual scheme or another? The answer was given in section 2.1. The key idea is that an object’s kind-membership, 9

Cf. Putnam, “Realism and Reason.”

38

Conventionalism as ontological relativism and its persistence conditions, enter so deeply into that object’s very identity that it is not coherent to say that one and the same object is perishing, relative to one conceptual scheme, and continuing, relative to another – or that an object belongs to one natural kind, relative to one conceptual scheme, and to a different kind, relative to another. But this still leaves room for relativism at the level of “the world,” and that is what (3) asserts. Indeed (2) forces relativism at the level of the world (unless indeed there is something illegitimate in talking about “the world,” or about “objects” in general10). The world can be determinately populated by objects only if there are, in the world, determinate phenomena of kind-sameness and persistence; but kind-sameness and persistence obtain only in virtue of one conceptual scheme or another; so which objects there are in the world must depend on – be relative to – which conceptual scheme is chosen as reference point. But what reason is there for believing (1) – and should we really believe that (1), even if true, gives reason for believing (2)? Those are the topics that I shall examine in the next sections. 2.3

c a r v i n g ou t s t r a n g e k i n d s

Eli Hirsch, as we noted in chapter 1, invites us to consider various “strange languages.” Some of these languages employ sortals that seem not to “cut the world at the joints”: any one such sortal is satisfied by objects that belong to what seem, intuitively, to be quite disparate natural kinds. One such language features the sortals “apcean” and “carple.” An apcean, as I said before, is (in our terms) either an apple or an ocean; a carple is either a car or an apple.11 (For different examples of such “strange objects,” see Sosa or Davidson.12) Other strange languages feature predicates that seem to blend together properties that objectively are distinct. One such language calls certain objects “gricular” – they are either green or circular – and other objects “grincular” – they are either green or not circular.13 Hirsch argues that if there is something wrong with these languages, it is by no means easy to say what this is. It is not clear that they 10 11 12 13

Thomasson, Ordinary Objects, Ch. 6. Hirsch, Dividing Reality, pp. 24–25. Sosa, “Existential Relativity”; Davidson, “Emeroses by Other Names.” Hirsch, Dividing Reality, pp. 4–5.

39

Familiar objects and their shadows are cognitively off limits – not clear, in other words, that we mightn’t actually think in such a strange language. It is not clear that they misrepresent our immediate perceptual experience of the world. Above all, it is not clear that they would be disastrous in practice. On the contrary, Hirsch argues, it is hard to deny that use of such languages could afford us all the practical and cognitive mastery of the world that we in fact enjoy. Hirsch’s strange languages thus provide vivid illustrations of why (1), in the master argument for OR, seems plausible. Hirsch himself holds that (1) only seems plausible – that in fact, there is something wrong with his strange languages, but a “something” that neither he nor anyone else has adequately diagnosed. Nor is Hirsch a proponent of OR. On the contrary, he holds that for all his arguments show, the world really may be assorted into mind-independent natural kinds, the members of which trace out mind-independent courses of existence.14 But Hirsch nevertheless has, perhaps despite himself, provided the best illustrations extant of why the “master argument” can seem cogent. We could really make our way about the world just as well as we do, Hirsch argues, if we took it to be populated by such things as apceans and carples. In this section I argue that Hirsch is wrong about this. Before beginning, a preliminary note: there is (as Hirsch sees) one way in which successful employment of the strange languages might entirely fail to suggest that we could prosper as well as we do, even if we “divided reality” using quite different sortals and predicates. If we used Hirsch’s strange sortals only in conjoined forms – if we spoke only of apceans-which-are-carples, of apceans-which-are-not-carples, and of carples-which-are-not-apceans – then we would merely be speaking, in long-winded fashion, of apples and oceans and cars. Our strange language would then afford practical mastery for exactly the same reason, and in exactly the same way, as our actual language does. Hirsch’s strange languages are more interesting than this. Their speakers make uncompounded employment of the strange sortals (and of the seemingly disjunctive predicates).15

14 15

Hirsch, Dividing Reality, Chs. 3–4; “Rules for a Good Language,” pp. 699–700. Hirsch, Dividing Reality, p. 5.

40

Conventionalism as ontological relativism Imagine, then, a population of humans – call them “the Strangers” – whose scrambled system of sortals incorporates “apceans” and “carples,” and who commonly assert claims in which one of these sortals appears all by itself. Why might it be of use for Strangers to speak generally of apceans or of carples, and to treat them as natural kinds? The usual answer for why talk of natural kinds is useful is that members (or samples) of natural kinds share interesting characteristics in common, and that we can learn of these clustered characteristics by inductive extrapolation from sample groups. But apceans and carples seem not, at fi rst blush, to have interesting characteristics in common; indeed it seems that almost nothing is true of apceans generally or of carples generally, and that what little is true is of no practical importance. Apceans are not, in general, characterized by any particular shape or size or mass; they do not have a characteristic chemical composition or a particular origin; they are not in general edible, nor are they in general inedible. About the only characterization that holds true of apceans in general (or of carples) is that they are all composed of matter. But that, according to many philosophers, is true of everything. Yet Strangers could affi rm informative and useful claims about apceans and carples if they employed (what look to us to be) disjunctive predicates. Apceans can all be said to have a certain roughly defi ned chemical make-up, if we allow (what look to us to be) disjunctive specifications of chemical make-up. Carples can all be said to have a certain disjunctive sort of origin. Moreover Strangers could, by using disjunctive predicates, say useful and informative things not only about apceans and carples in general, but about apceans and carples that are in particular states, that have particular accidental features. They could say how, in general, carples respond and react to particular circumstances. But for Strangers to be as good at this latter job as we are – and this is the point that Hirsch neglects – they must wield disjunctive predicates for carples (or for apceans) that answer to two requirements. These disjunctive predicates must pick out distinct ways that carples can be – two such predicates must not be satisfiable by the same carple at the same time. And they must fall into families, such that, at least over broad ranges of circumstances, carples can be counted on to satisfy at least one predicate in the family. For these are the requirements met by the predicates on which we base many of the useful 41

Familiar objects and their shadows predictions we make, as to how the familiar objects (and matters) that we believe in will behave, and how they can best be employed by us. We judge that depending on which temperature a particular hot beverage (or a particular baked apple) selects, out of a particular range, the beverage will either burn the tongue, taste pleasantly warm, or be unpleasantly tepid. We judge the temperature that is present in a given ocean determines whether, and to what extent, that ocean will support the health of coral reefs. We judge that one and the same model of car will, depending on the number of speeds in its transmission, deliver different rates of acceleration and different figures for gas mileage. We judge that depending on the fi rmness found in a given batch of apples, different pressures will be required to press cider from the apples – and that depending on the sweetness of the apples, fermentation will proceed at different rates, and produce different levels of alcohol, so that greater or lesser amounts of sugar may have to be added to achieve successful fermentation. Our judgements here depend on identifying what causal systems theorists call “an invariance.” An invariance is a pattern of connection between two variables such that – to put it metaphorically – intervening on the value of one variable, across a certain range, enables one to manipulate the value taken by the other variable. The identification of an invariance, as recent insightful work by Woodward and Hitchcock has shown, is often crucial for detecting causation.16 (More on invariances in chapter 5.) If Strangers are (as advertised) to derive the same level of practical mastery from their strange system of sortals as we derive from ours, it seems overwhelmingly likely that they will have to be just as good as we are at detecting causation. But then the disjunctive predicates, by which they pick out the accidental states that characterize members (and samples) of their strange natural kinds, must meet the two requirements mentioned above. In different words, Strangers must in effect attribute to these members (and samples) accidental properties that fall into contrary ranges, just as do the properties in our predictions about baked apples and oceans and cars. The properties in such a range must be incompatible, for members of the kind to which they 16

Woodward and Hitchcock , “Explanatory Generalizations, Part I: a Counterfactual Account,” and “Explanatory Generalizations, Part II: Plumbing Explanatory Depth”; Woodward , Making Things Happen.

42

Conventionalism as ontological relativism are attributed, but must be such that, at least across many familiar circumstances, members of the kind can be counted on to select at least one. How might predicates depicting such contrary ranges be constructed? A simple recipe suggests itself. Assemble, using our predicates, lists of accidental properties that form contrary ranges from which cars make determinate selections, and lists of accidental properties that form contraries from which apples make determinate selections, and then connect, using disjunction, each member in the fi rst batch of lists with some member in the second batch. For cars, one such list will feature horsepower ratings; a second, number of cylinders; a third, number of speeds in the transmission. For apples, one such list will feature degrees of sweetness (perhaps vaguely designated as “very sweet,” “slightly sweet,” “tart,” or perhaps by numerical scores in the Brix scale); another, degrees of fi rmness; a third, particular colors. Then one accidental feature, that carples in general can select, will be expressible as “fourcylortart”: carples which are cars will select this feature by having four cylinders, and carples which are apples will select it by being tart. But two problems quickly confront the simple recipe. The fi rst arises if more lists of contraries characterize cars than characterize apples; the second arises if the same number of lists characterizes both carples-of-the-car-sort and carples-of-the-apple-sort, but the lists incorporate different numbers of entries. Both problems can be illustrated by focusing on yet another strange sortal, namely “appluark.” In our parlance, an appluark is either an apple or a quark. Apples, let us say, select from the contraries on each of five lists: fi rmness, sizewhen-mature, density-when-mature, sweetness-when-ripe, and color-when-mature. But quarks are characterized only by (electric) charge, color, spin, and mass. So how are we to devise predicates for the accidental properties in contrary ranges, such that appluarks in general can be counted on to select only one, but at least one, property in the range? There are many contrary properties on the sweetness list, but on the list for spin, there is only spin up and spin down. Do we pair several different properties on the sweetness list with the same spin – say, with spin up? But then we fail to pick out accidental properties that are incompatible, for appluarks in general. An appluark that is a quark with spin up will select several such disjunctive properties. Or do we pair only two properties on the sweetness 43

Familiar objects and their shadows list with properties on the spin list – e.g. tartness with spin up, and extreme sweetness with spin down – and leave the rest unpaired? But then we will have predicates for accidental properties, for which appluarks-which-are-quarks are not even eligible: these properties will not belong to ranges from which appluarks in general can be counted on to make determinate selections. Nor will it help if, by some miracle, we fi nd just as many entries on the spin list as on the sweetness list, and just as many on the (quark-) color list as on the size list, and just as many on the charge list as on the (apple-) color list. We will still have an entire list, namely density-when-mature, from which appluarks-which-are-apples make determinate selections, left over! Our attempt to construct ranges of disjunctive properties, from which appluarks both of the apple sort and of the quark sort can be counted on to select at least one and at most one, will have failed. I conclude that only in extraordinary cases will strange sortals such as “carple” afford us the same practical mastery as our actual sortals. For in countless standard cases, we will not be able to identify invariances involving the items designated by strange sortals, and hence will be unable to detect much causation. 2. 4

c a r v i n g ou t s t r a n g e i n d i v i dua l s

If there were something particularly “soft” or unobjective about sameness in natural kind, one might expect that we could prosper just as well as we do even if we wielded quite different sortals for nature’s kinds – for example, sortals that stitched together what appear, to our customary ways of thinking, to be disparate natural kinds. Just so, if there were something “soft” or unobjective about numerical sameness over time, one might expect that we could prosper as well as we do even if we attributed different courses of existence to such objects as we take to populate the world. Where now we take there to be an object that survives a particular sort of change, perhaps we could do equally well to suppose that there is instead an object that is thereby destroyed, and that gets replaced by a distinct object. Perhaps where we take a destruction to occur, we could do equally well to suppose that what has happened is a mere alteration. Perhaps indeed we could recognize objects that are capable of surviving across changes which we now consider it impossible for any object to undergo, whether that object thereby perishes or survives. 44

Conventionalism as ontological relativism Here are illustrations of these three switches in conceptual scheme. Perhaps where we currently take there to be cars we might instead take there to be incars and outcars. An incar is, as we would put it, a car that is located in a garage; backing an incar out of a garage results in its gradual destruction (as I noted in section 2.1), and in its replacement by a distinct object, an outcar; and driving an outcar into a garage destroys it in turn, leaving only a new incar. Or perhaps where we currently take there to be trees – objects that are destroyed if chopped up and passed through a woodchipper – we could instead take there to be treesorchips, objects for which the trip through the woodchipper marks a mere alteration in state. Or, fi nally, in at least many of the places where we take there to be tables, perhaps we could instead take there to be what Shoemaker calls “klables.”17 Klables are found only in houses in which, as we would put it, there is exactly one table in the living room and exactly one in the dining room. Any klable that, in the morning, is located in a living room, instantaneously moves at noon to the dining room, and instantaneously returns at midnight to the living room; a klable located in the morning in a dining room performs, at the same two times, the opposite instantaneous movements. (For other examples of strangely persisting objects, see Sidelle.18) The fi rst and third examples of objects with strange courses of existence are discussed by Hirsch.19 Hirsch argues that we could prosper just as well as we do if we took there to be just these strangely persisting objects. I contend that here again Hirsch is mistaken. The problems I shall raise against positing “klables” apply also to the strangely persisting “treesorchips,” which are my invention, rather than Hirsch’s. In the previous chapter I argued that the function that judgements of numerical persistence serve in our conceptual economy – the raison d’être of these judgements – is to induce us to discover and apply causally underlain patterns of continuity and alteration that span the members of a given kind. In the case of the strangely persisting objects that explosivists posit, such patterns will generally not obtain at all. The accidental states that characterize a particular trout will not 17 18 19

Hirsch, Dividing Reality, p. 26. Sidelle, “Sweater Unraveled,” §5. Hirsch, “Strange Thoughts of the Third Kind,” §§7–8.

45

Familiar objects and their shadows exercise any causal control at all over the accidental states later found in a particular turkey; similarly, a minimal D-fusion may incorporate, at earlier times, objects that have no causal connection whatever with those which that D-fusion incorporates at later times. But some of the strangely persisting objects that Hirsch envisions would provide for the discovery and application of patterns by which accidental properties earlier acquired would causally control particular accidental sequels. An outcar, once dented in a traffic accident, will tend to retain that dent, and an outcar which idles roughly at 11:00 am will tend to go on idling roughly at noon. At the same time, however, strangely short-lived objects such as incars and outcars seem to hinder and restrict the application of discoverable patterns of continuity, as much as they invite it. If an incar being backed out of the garage at 9:00 am displays a rough idle, or has a dented fender, we seemingly have no reason to expect that the rough idle or the dent will be present at 9:30 – after all, the very bearer of these accidental features will no longer exist! Similarly, discoverable patterns of alteration, that would span the kind cars, have little application to either incars or outcars: even if rusty holes have appeared in the underbody of an outcar, there is little reason to expect that very outcar to display yet bigger holes in the future. In a similar way, the accidental features acquired by a klable in the morning can be expected to stay in place only up until noon; moreover, they will exercise virtually no causal control over the accidental features of that same klable in the afternoon. Now once again, Hirsch does have a reply. He could say that we are thinking of the reliable patterns of continuity and alteration as being something like conservation principles, or rather conservation and/ or mutation principles. That is, we are thinking of them as patterns by which accidental states of a given object causally control accidental states later present in that very object. But the Strangers whose language recognizes strangely persisting objects might well rejoin that we must discover and apply transference principles. For example, we should think that the very same accidental properties, for which our principles predict conservation, will get transferred from any incar to the corresponding outcar, once the incar is backed out of a garage. But this reply deprives the “incar/outcar” language of its interest, from the standpoint of OR. Earlier we noted that if a strange language features the sortals “apcean” and “carple,” but employs them 46

Conventionalism as ontological relativism only in compounded form – “apceans which are carples,” etc. – then that language affords its users practical mastery for just the same reasons, and in just the same ways, as does our ordinary language. If this is all that step (1) in the “master argument” for OR amounts to, then (1) provides no support for (2). Just so here. Coupling the terms “incar” and “outcar” with transference principles, in place of conservation principles, merely reproduces, in long-winded form, our existing conceptual scheme. Parallel objections confront the suggestion that, in at least many of the places where we suppose there to be tables, we could just as well suppose that there are klables. Suppose that, of a given morning, Klable A in the dining room is deeply scratched, while Klable B in the living room is shiny and smooth. Being deeply scratched seems a paradigm case of the sort of accidental property for which our conservation principles predict conservation. But come 12:01 pm, Klable A will be shiny and smooth! Now, speakers of the “klable” language do have a reply. They can say that Klable A is steadfastly k-scratched. In the morning hours, what it is for a klable to be k-scratched is, as we would put it, for the table located where that klable is to be scratched; during the hours between noon and midnight, a klable’s being k-scratched amounts to the fact that the table in the other room (dining room if the klable is in the living room, living room if the klable is in the dining room) is scratched. But this reply raises troubles of its own. If the wrong shade of fi nish restorer is applied, to a badly scratched table, what results is an unsightly pattern of lines. Suppose, then, that in the afternoon, a fi nish restorer of the wrong shade is spread over Klable A. Does this accidental property – having a mismatched shade of fi nish restorer – interact causally with Klable A’s other accidental property, being k-scratched, to produce unsightly lines? If so, we have a bizarre case of remote causation, and it seems hard to believe that speakers of the “klable” language will be as well off, practically, as we are. Or should “klable” speakers insist that the only accidental properties that Klable A can acquire in the afternoon are k-accidental properties – ones that obtain strictly in virtue of how matters stand with the table in the dining room? But then “klable” speakers are saying that, from noon until midnight, the whole story on what happens to Klable A, and how A reacts, is found by attending to the table that remains in the dining room. The differences between their strange conceptual scheme and our ordinary one have 47

Familiar objects and their shadows been buried completely – just what happened when “incar” speakers elected to speak of transference principles, and when “apcean” speakers restricted themselves to compound constructions. 2.5

T h e wo r l d o n t o w h i c h w e p r oj e c t k i n d -sa m e n e ss a n d pe rsist e nce

Hirsch’s strange languages show vividly how (1) can seem to be true. Even so, it is highly questionable whether (1) is adequate to establish (2). Might the proponent of OR offer stronger reasons for believing (2)? In this section I argue that it is fruitless to pursue the question – that (2) is assuredly an untenable position. The reason, in a nutshell, is that (2) places insuperable obstacles against understanding our perceptual response to the world. At fi rst blush, it seems the proponent of (2) offers a picture of our perceptual experience that is entirely plausible and reasonable. The proponent will say that we deploy our system of sortals and predicates – our “conceptual scheme” – so unhesitatingly, so much in the manner of an unconscious reflex, that it seems to us that the world we encounter in experience is already populated, independently of our attending to it, by objects of various kinds, objects that survive across certain changes and cease to exist at others. But really, the proponent must add, in this seeming there is an element of illusion. Really, the kind-samenesses and individual persistences that we take ourselves to observe obtain only in virtue of our own system of sortals. Consequently the very objects that we take ourselves to observe obtain only in virtue of our own system of sortals. (See section 2.1 on the connection between an object’s very identity and its kind-membership and persistence conditions; see also chapter 3.) In perception we causally confront – even if we do not consciously confront – a world that is objectless. Exactly what is the world that we causally confront like? This question raises a problem for the proponent of (2) precisely because we do not, in perception, confront that world in the manner of a monolithic “group subject.” On the contrary, each of us confronts the world independently, apart from everyone else. To be sure, there may be room for some element of a monolithic “we” in the story of our perceptual encounter with the world. Perhaps we all deploy our system of sortals and predicates – our common “conceptual scheme” – because of a 48

Conventionalism as ontological relativism culture that we share, or because of an evolutionary history that we all have in common. (I will briefly return to this topic in the next section.) Even so, implementing that common system of sortals is something that each of us must do on his own. That is, the mere fact that I am trained or programmed so as to fi nd, somewhere, some instances of the same sortals as you are trained or programmed to apply does not, by itself, explain how it happens that I very often fi nd an instance of the same such sortal, in just the same spatiotemporal location, as you fi nd. Something independent both of my scheme-implementing mental activities and of your scheme-implementing mental activities must explain this convergence in implementation. There must be some mind-independent way the world is. Proponents of (2) are generally quite willing to characterize the world as it exists independently of our exercise of our conceptual scheme.20 It is simply so much stuff, proponents say, across which a play of properties occurs – one property giving place to another and that other to yet a third. But just which properties are they, that get instantiated in the world as it mind-independently exists? Proponents of (2) rarely address this question, but it is an important one.21 For if it should turn out that only very few properties, or only very thin properties, can occur in the world as it is independently of us, there may be no answer to the question bruited above. There may be no explanation of how and why the world nudges me in one direction, in the implementation of our shared system of sortals, and likewise nudges you in one direction, and the two directions turn out to match. Here is what gives the rarely addressed question its force: according to (2), there is no mind-independent phenomenon of persistence. Surprising consequences follow. For starters, it cannot happen, in the mind-independent world, that some entity persists from a moment at which it occupies place p1 to a moment at which it occupies place p2. But at least that is required for there to be motion: for it to be the case that something moves, that very something must at successive times occupy different places. (It does not matter whether we take persistence to be endurance or instead perdurance or exdurance; 20

21

Sidelle, Necessity, Essence, and Individuation, pp. 55n, 57; “Sweater Unraveled,” pp. 441–44. Elder, “Conventionalism and the World.”

49

Familiar objects and their shadows (2) denies that the phenomenon mind-independently occurs, however it is analyzed.) So no motions, and no rates of motion – no velocities. Might shapes and sizes obtain in the independent stuff ? If shapes and sizes can obtain “all by themselves,” without being shapes and sizes of something, then innumerable overlapping shapes and sizes will all obtain. But such self-standing shapes and sizes will be too numerous to nudge me, in my implementation of our conceptual scheme, in any one direction rather than another – and there will be no explanation of convergent implementation. Suppose we say, then, that in the independent stuff, shapes and sizes obtain only where they are the boundaries of spatially unitary entities. Then what is a spatially unitary entity – what is it for the same something to extend across, and be contained within, some volume of space? There seems no way to address the question except by asking when, as a matter of empirical psychology, we judge that some unitary something occupies (and is contained within) some volume of space. In the central case – as Hirsch 22 and others have shown – we make such a judgement just when we suppose that the matter throughout such a volume forms a maximal, cohesive, separately movable whole. Shall we say, then, that a size and a shape obtain in the independent stuff whenever it is true that a certain portion of stuff will all move “as a piece” if certain forces are applied to it? But no motion can occur in the independent stuff. There is, however, a different basis on which we sometimes judge that a volume of space contains, and is wholly occupied by, some one something. If we are confronted with something that seems just too big and massive to be moved, or that seems too fi rmly embedded in surrounding matter to be moved separately, our judgement that that one something occupies (and does not extend beyond) a particular volume seems to be guided by qualitative homogeneity within that volume – homogeneity that ends at the volume’s borders, and thus sets up “border contrast” with what lies beyond. But now we must ask: homogeneity with respect to which sorts of qualities, which sorts of properties? We want the judgement of homogeneity to ground a judgement on where borders lie, and not to presuppose such a judgement. So we must look for properties that 22

Hirsch, Concept of Identity, pp. 105–12.

50

Conventionalism as ontological relativism can characterize subregions within a given region of space, while at the same time characterizing that region as a whole. That is, we must look for “border-insensitive,” rather than for “border-sensitive,” properties. Shape, size, and total mass are border-sensitive properties. What we need are properties such as viscosity or density, or perhaps even melting point and index of refraction – properties that can be present throughout a given portion of matter, regardless of where the borders of that portion lie. But these standard examples of border-insensitive properties will not do; we will need others instead. The reason, again, is that according to (2), there simply is no such phenomenon as persistence, in the mind-independent stuff. So it cannot happen that the same portion of stuff passes from being in a solid state to being in a liquid state; melting, that is, cannot occur in the independent stuff; and so no melting point can characterize a portion of the independent stuff. Nor can it happen that the same portion of stuff, as was earlier contained in a beaker, now spreads widely over a surface on which it is poured, or instead remains in a compact mass. There can be no viscosities. Indeed these problems with standard examples of border-insensitive properties suggest a general point. They suggest that there can obtain, in the mind-independent stuff, no dispositional properties. For the manifestation of any disposition requires persistence: one and the same thing must fi rst be subjected to the relevant trigger, and must subsequently display the disposition. It is one thing to say that in isolated, “fi nkish” cases, a dispositional property can obtain, even though some factor will intrude, once the triggering event occurs, to prevent display of the disposition. It is quite another to say that in a world in which it is in principle impossible that any disposition ever be manifested, there can nevertheless occur dispositional properties. But the world as it exists independently of the exercise of conceptual schemes is just such a world, according to (2). So just how many properties can really occur in the mindindependent stuff ? It appears that the only remaining candidates may be immediate, phenomenal properties – properties reported by such predicates as “looks red now,” “smells like lavender now,” “feels fi rm now.” The world as it exists mind-independently may be only so much stuff across which a play of sense-data takes place. And it is notoriously hard to see how an array of sense-data can nudge my implementation of our conceptual scheme – our system of sortals 51

Familiar objects and their shadows and predicates – in one particular direction rather than another. The world as it exists independently of our system of sortals may afford no explanation at all of convergent implementation. (There is indeed a rather drastic way in which one could make the apparent need for such explanation simply disappear. I could argue that you – and all the others in “we” – exist only in virtue of my own mental activity. In that case your implementation of “our” system of sortals will itself be something constructed by me. Small wonder, then, that I take your implementation as converging with mine – compare Husserl.23 But it is questionable whether any less drastic position can make the need for explanation disappear.) 2.6

W e w h o p r oj e c t k i n d - s a m e n e s s a n d pe rsist e nce

But set aside, for a moment, questions about how the world engenders in us convergent implementation of our shared system of sortals. There are other questions, gestured at above. What explains our sharing of a common system of sortals, a common “conceptual scheme”? Indeed, who are “we” – what binds us together? And what are we like? For the proponent of (2), there is simply no phenomenon of sameness in kind – and likewise, no phenomenon of persistence across time – except in virtue of the exercise of some system of sortals. The particular kind-samenesses that we recognize, and likewise the particular courses of individual existence that we recognize, obtain only in virtue of our exercising our system of sortals. By exercising that system, we bring it about that there are the kind-samenesses and the persistences that we believe in. But then one naturally wants to ask: what are we like independently of – metaphysically “prior to” – our exercising that system? Are we all members of a common natural kind? Do we have determinate (and bounded) courses of existence? Proponents of (2) sometimes dismiss such questions as ill-formed. Putnam, for example, maintains that by our world-engendered exercise of our conceptual scheme, we bring about not just the facts concerning objects in the world, but even the facts that concern us. “[T]he mind and the world,” Putnam 23

Husserl, Cartesian Meditations, §§55–56.

52

Conventionalism as ontological relativism writes, “ jointly make up the mind and the world.”24 In other words, the world of objects is “a kind of play,” a series of stories, of which we are the authors; we do ourselves appear in the stories; but for all that, “the authors in the stories are the real authors.”25 Putnam does say that these utterances are metaphorical. For my own part, I am unable to see what they are metaphors for – to me, these utterances more closely resemble Zen koans than philosophical positions. In Cartesian Meditations, Husserl offered a nonmetaphorical response to the same sorts of challenges. Like the proponents of OR, Husserl maintained that the world of objects obtains only in virtue of our mental activities. He then had to confront the question whether we ourselves are objects in the world – and in particular, whether we trace out determinate (and bounded) courses of existence. His answer was that by virtue of a “mundanizing self-apperception” – a regarding of ourselves as objects in the world – we thereby do qualify as objects in the world.26 But who performs the mundanizing self-apperception? Are we, as the performers, bodily beings whose existence begins with birth and ends with death? Husserl admitted a difficulty in identifying the we who are world-constituting transcendental egos with the mundanized beings in the constituted world: he called this “the problem of birth and death.”27 No one can feel comfortable being faced with Husserl’s “problem.” It remains unclear how OR should respond to it. 24 25 26 27

Putnam, Reason, Truth and History, p. xi. Putnam, “Realism and Reason,” p. 496. Husserl, Cartesian Meditations, §45. Husserl, Cartesian Meditations, §61.

53

3 Realism about material objects: persistence, persistence conditions, and natural kinds In chapter 1 I argued that judgements of numerical persistence are imperatives, directing us to discover and draw inferences from causal patterns by which newly arriving accidental features will either lead to their own continued presence, or will give way to particular other accidental features, depending on the natural kind to which the persisting object (or portion of matter) belongs. Such causal patterns are clear enough that we can discover them, at least widely enough to have rendered it fruitful for us to heed the imperatives embodied in judgements of numerical persistence – widely enough, then, to have earned for those judgements a place in our conceptual/ linguistic repertoire. Where such patterns obtain, they are grounded in the properties essential to the natural kind in question. Thus in treating judgements about an object’s course of persistence as guiding us to such patterns, we are treating the events that may befall an object over its course of persistence as governed by the properties essential to the kind to which it belongs. In effect, then, we treat any course of persistence as delimited by some instance of kind-membership. We act as if any object (or sample of matter) can continue to exist only so long as it stays in its natural kind: we treat the conditions on kindmembership as being, at the same time, persistence conditions for the members of that kind. This position on persistence conditions is philosophically controversial. Nothing said in chapter 1 establishes that it is true. For all that is said in chapter 1, there may be phases in, or extensions to, a given object’s course of persistence that we neither note nor care about in making our judgements of persistence. We might have latched onto mind-independent courses of persistence because of the predictive and retrodictive knowledge that can be gained so long as those courses are bounded by membership in a single natural kind; but there may 54

Realism about material objects be other phases in these courses that play no role in explaining why and how we come to care about numerical persistence. The aim of this chapter is to show that the philosophically controversial position on persistence conditions, which in effect we treat as being true, really is true. This result is of considerable importance for this book since, as I shall argue, the controversial position has to be true in order for realism to be defensible about the existence of material objects – and not just familiar material objects such as common sense recognizes, but less familiar objects such as those recognized, say, by cell biology or plate tectonics or neuroanatomy. Realism about material objects, including the familiar ones, is a starting point for this book – not something for which this book argues. The task of this chapter is to show that if we begin with the premise that realism is true, we end up affi rming the controversial position that the objects of nature (and the samples of nature’s matters or stuff s) can persist only so long as they retain the properties necessary for membership in the natural kind to which they belong. The chapter begins by arguing that realism about objects requires realism about their courses of existence, their careers; then argues that realism about careers requires realism about persistence conditions; and fi nally argues that the only defensible realist position on persistence conditions is one that makes membership in a particular natural kind a life-and-death matter for the objects of nature. The last step is the hardest, and requires a special sharpening in the account of just what a natural kind is. 3.1

R e a lism a n d m i n d -i n de pe n den t pe rsist e nce con di t ions

Realism about objects, if it is to merit its name, must attribute mindindependent existence to objects that are neither abeternal nor eternal. These are objects that begin to exist at a certain time, that continue to exist across certain changes – even if only such trivial changes as change in location or change in age – and that pass out of existence at a later time. I maintain that in order to attribute mind-independent existence to such perishable objects, one must attribute to them mind-independent courses of existence, careers, that mind-independently begin and end where they do, and that mind-independently span such changes as the object undergoes. 55

Familiar objects and their shadows One can indeed imagine a version of realism that made no such claim. This version would say that each perishable object exists mind-independently, but that the beginning of its existence, the continuing of its existence across changes, and the ending of its existence are all phenomena that obtain only in virtue of our ways of thinking or talking. But such a position would seem miserably unstable. Each perishable object would enjoy mind-independent existence, but there would be no mind-independent fact of the matter as to when it exists, no mind-independent span that its existence takes up. As it exists mind-independently, any such perishable object would be without temporal location – and hence would be quite disturbingly incomplete. Meinong believed that there are – in some sense of “there are”! – just such “incomplete objects.”1 There are persons who were not born on any particular day, golden mountains that do not have any particular height or weight, ivory spheres that have no particular diameter. But it is nearly impossible not to regard such incompleteness as the mark, precisely, of a fictional object. Just so with perishable objects that, supposedly, exist mind-independently, but do not mind-independently exist somewhen. Let us take it as granted, then, that realism about the objects of nature (and the portions of nature’s matters or stuff s) requires realism about the courses of persistence which these objects trace out. Each mind-independently begins to exist at a certain point, persists across certain changes, and ceases to exist at another point. Need there be some mind-independent ground or principle that underlies the fact that each such object ceases to exist just when and where it does: need there be mind-independent persistence conditions for nature’s objects? Or might it be that each real object (or many of them, in any case) ceases to exist at a certain point, but not in virtue of any further state of affairs – not in virtue, for example, of any particular qualitative change which it undergoes? In that case ceasing-to-exist would be a “brute fact.” And for an object to have avoided ceasing-to-exist, up to a certain point in its career, would likewise be a “brute fact.” Persistence itself would be brute. The idea seems hard to entertain, but there actually is one argument in the literature that would support it, at least as it applies 1

Findlay, Meinong’s Theory of Objects and Values, Ch. 6.

56

Realism about material objects to composite objects. Ned Markosian has argued for the position that composition itself may be “brutal.”2 That is, there could be two spatially compact aggregations of microparticles, that were microparticle- for-microparticle duplicates of one another, and that featured just the same relations among their members, but such that the first composed a single large object, and the second did not. (There might be some reason to think that the history that lies behind the spatial compactness of the aggregation is relevant to whether the microparticles within it compose something; in this case, Markosian’s claim would be that the two aggregations could be exactly matched in the qualitative details of their histories.) Thus Markosian presents us with a synchronic case in which composition f loats free from any further facts. We need only imagine a diachronic variation of the same case to see how it might be thought that a composite object could exist at one moment, cease to exist the next, but not cease to exist in virtue of any further fact. Markosian’s discussion, then, helps us to understand the idea that (at least in the case of composite objects) persistence (and persistencefailure) might be brute. But it does not help us to believe it. If persistence were brute, it would be deeply puzzling why we care about it at all, or bother to keep track of it. Indeed it would be hard to see how we could ever have developed the concepts and vocabulary necessary for making claims about persistence.

3. 2

N o n qua l i tat i v e p e r s i s t e n c e c o n d i t i o n s ?

Persistence conditions, then – realism must extend this far. But we are still a long way from the claim that an object can go on existing only so long as it retains the properties by virtue of which it belongs to its natural kind. There are two positions on persistence which hold that an object can go on existing despite retaining no properties in particular – at least, no properties of the ordinary, qualitative sort. The fi rst of these, from Stephen Schwartz, maintains that an object need only retain its haecceity in order to go on existing. 3 The second, of 2 3

Markosian, “Brutal Composition.” Schwartz, “Essence of Essence.”

57

Familiar objects and their shadows older vintage, holds that an individual object goes on existing so long as the matter of which it is composed goes on existing – but not “the matter” in the usual sense. In the usual sense, “the matter” of which a statue is composed might be some gold or some clay. Intuitively it seems clear that there are properties which some gold must retain, in order to go on existing: these perhaps include ductility and malleability, and assuredly include being-composed-of-atoms-havingAN-79. The same is true of course for matter belonging to any other particular matter-kind. But when this second position on persistence refers to “the matter,” it claims to refer to a matter that may for a time belong to one matter-kind or another, but that need not belong to any one matter-kind in order to go on existing. This second position, in short, pins persistence to the “prime matter” to which Aristotle worries, in Metaphysics Z 1029a3–36, that he may be committed. Let me speak fi rst to Schwartz’s position. A “haecceity,” for the uninitiated (or uncorrupted?), is a property that necessarily is possessed by only one entity in the universe, and which that one entity necessarily possesses – where both occurrences of “necessarily” express metaphysical necessities. (Etymologically, a “haecceity” is the thisness of this object.) Schwartz argues that it is metaphysically possible that an object should retain its haecceity even while losing any other property that it may have, and concludes that any object need only retain its haecceity in order to go on being itself – in order, in other words, to go on existing. But are there such things as haecceities? On the view on property-identity that I shall employ in chapters 5 –7, and shall defend in chapter 8, a haecceity is not a property at all. Indeed, it is nothing; the very concept is confused. Happily, this objection can safely be postponed. For the older position on propertyless persistence – the one derived from Aristotle – faces an objection that also applies to Schwartz’s position. What this older position says is that any object (or sample of a particular kind of matter) is essentially just that parcel of prime matter; all else is just a matter of accidental states which that prime matter assumes. The objection is that as soon as we do entertain the thought that nature is really populated by parcels of prime matter, we are required to envision massive, widespread coincidence, for which there can be no explanation. Let me approach this point by speaking metaphorically. If nature is populated only by parcels of prime matter, it can address

58

Realism about material objects only hypothetical imperatives to the entities that populate it. It can say only, “Parcels of matter! To the extent that you elect to belong to one natural kind or another, you must display the properties characteristic of that kind, and you must respond to outside influences in the ways characteristic of that kind; in certain circumstances, you must depart from that kind altogether. If you elect to belong to the kind dog, you must appear furry and four-legged; you must respond to being kicked by yelping; if you should be fried by a bolt of lightning, you must depart from the kind dog altogether. If you elect to be some water, you must appear clear and potable; you must boil if heated to 212ºF or higher; and if subjected to hydrolysis, you must depart altogether from the kind water. But you are free at any time to rescind your membership in the kind to which you have elected to belong: it is not as if you must, on pain of death, continue to belong to your current natural kind. For this reason, if you currently belong to the kind dog, you are free suddenly to present cat-like features; or to respond to being kicked by emitting the hollow ringing of a gong, or by exploding; and you are free to respond to being struck by lightning by forming a pool of molten lava.” The problem, of course, is that nature appears to be populated by entities that take their membership in a particular natural kind quite seriously. Nature’s objects and portions of matter appear to treat the duties attendant upon their kind-membership – duties both as to the properties they present and the behavior they perform – as unconditionally binding. They do not respond to impinging influences in ways uncharacteristic of their kind, do not suddenly display properties that lie outside the nature of their kind, and do not appear to be destroyed when exposed to events that members of their kind characteristically survive. If nature really is populated by parcels of prime matter, it is an amazing coincidence that each such parcel behaves in the way one would expect of an entity permanently wedded to just one natural kind. To drop the metaphor: the hypothesis that nature is populated by parcels of prime matter is empirically indefensible. There is not a scrap of evidence for supposing it true, and supposing it true requires us to posit a standing and inexplicable coincidence. And the very same coincidence confronts us if we suppose that nature is populated by objects that need only retain, over the courses of their existence, their haecceities.

59

Familiar objects and their shadows

3.3

Qua l i tat i v e p e r s i s t e n c e c o n d i t i o n s

Realism about perishable objects, then, requires the position that I will call “the qualitative principle.” This principle says that when an object (or a sample of matter) ceases to exist, it does so in virtue of losing certain properties; that the object’s (or the sample’s) persistence over its actual course of existence was owed, at least in part, to its steadfastly possessing those same properties; and that its existence began when and where it did in virtue of those same properties’ coming to be jointly instantiated just then and there. This principle neither asserts nor denies that an object’s existing across some span of time, along some path through space, just consists in the coinstantiation, along this space–time worm, of those properties. All that it says is that ceasing-to-be, persisting, and coming-to-be are not brute: these phenomena obtain in virtue of something, and that “something” is at least in part a matter of the advent, continued presence, and departure of certain properties. Just which properties is it, the continued presence of which accompanies an object’s persistence, and the departure of which marks the object’s ceasing to exist? Traditionally, “the qualitative principle” gets articulated as the very claim about kind-membership for which I will be arguing. That is, the properties which an object (or a sample of matter) must retain in order to go on existing are, on the traditional understanding, precisely those by virtue of which that object (or sample) belongs to its natural kind. I shall treat this articulation as occupying the status of a “default assumption”: special argument is required to dislodge it, I shall assume, and no special argument is needed to uphold its claim to plausibility. This may initially seem like a dialectical ruse. It might initially seem that “the qualitative principle” could plausibly be articulated as a more modest claim – as merely requiring, say, that an object retain over the course of its existence some of the properties by virtue of which it belongs to its natural kind, but not necessarily all of those properties. But plausible illustrations of this more modest articulation all end up affi rming some kind of connection, after all, between numerical persistence and kind-membership – even if this ends up being some higherlevel kind. Suppose, for example, someone surmised that an electron could go on existing so long as it retains just its characteristic rest mass, and that it need not also retain its characteristic charge. The 60

Realism about material objects surmise seems worth entertaining just because, in this case, the electron would still go on being a lepton of some variety. It would be far less plausible to surmise that an electron could go on existing provided it merely retained properties so “thin” that retention of them would be compatible with an electron’s becoming an animal (e.g. a dog) or a plant (e.g. an oak tree) – properties such as being physical, or having some mass or other, or having a location in space and time. One could perhaps plausibly surmise that a squirrel could survive a transformation in which it “turned into” a rabbit; it is not at all plausible, intuitively, that a squirrel could “turn into” a star or a neuron. But this very defense of the presumption that the qualitative principle must tie persistence to kind-membership – thus turning the conditions on kind-membership into persistence conditions – also shows us what the main threat to that presumption is. It appears that there are higher-level natural kinds – e.g. leptons, animals – that are superordinate to lower-level, but perfectly genuine natural kinds (e.g. electrons, squirrels). Provided there are many such hierarchical arrangements among the natural kinds that populate the world, it can seem entirely possible, intuitively, that an object (or sample of matter) should depart from a natural kind to which it belongs, and yet continue to exist. And in that case membership in a natural kind does not, in and of itself, fi x persistence conditions. We need not focus on exotic cases in which, supposedly, an electron “turns into” a positron, or a squirrel “turns into” a rabbit. For many philosophers would be inclined to say that the following are perfectly genuine natural kinds: ice, adolescents, and diabetes sufferers. It seems entirely plausible, intuitively, that when a chunk of ice melts, nothing has really been destroyed; it is just that some H 2O has assumed a new form. It just seems true that when an adolescent ceases to be an adolescent, nothing ceases to exist; it is just that the adolescent turns into an adult. If a cure for diabetes were invented, administration of it to diabetes sufferers would not end their existences. And the threat that I have just identified, to the traditional articulation of the qualitative principle, does not just suggest that membership in a natural kind may not always fi x persistence conditions for the members (or samples) of that natural kind. It suggests that it is self-contradictory to hold that membership in a natural kind always, in and of itself, fi xes persistence conditions. No object, and no sample of matter, can at the same time both cease to exist and continue 61

Familiar objects and their shadows to exist. But if one and the same chunk of matter can both belong to the natural kind ice and to the natural kind H2O, then exactly such ceasing-to-exist-while-continuing-to-exist can happen – if membership in a natural kind, just in and of itself, fi xes persistence conditions. All that we need is some melting. If one and the same object can belong both to adolescents and to human beings, then all we need is some growing-up. How can we retain the qualitative principle – in at least some version – and yet avoid self-contradiction? There are only three options. (1) We could hold that membership in any natural kind, whether higher or lower, does by itself fi x persistence conditions – but that in the cases that seem to saddle us with self-contradiction, we really begin with two distinct objects that occupy exactly the same volume of space. (2) We could hold that membership only in some natural kinds fi xes persistence conditions for the members or samples of that kind; kind-membership is not always a life-and-death matter. (3) We could hold that the putative lower-level kinds, from which it seems so obvious that a member or sample can depart without ceasing to exist, are not true natural kinds at all. I shall argue, over the course of the next two sections, that the third option is the right one. Here is where I will present the sharpening in the account of what a natural kind is that I promised at the outset of this chapter. 3. 4

A p pa r e n t d e pa r t u r e s f r o m a n at u r a l k i n d

First let me discuss problems with the other two options. The idea that distinct objects can, at a time, occupy exactly the same volume of space seems to many philosophers hard to credit. The standard sort of example here is the statue of Atlas made entirely of clay. There is a statue there, but there is also, apparently, a parcel or a portion of clay. The two objects differ in their modal profi les and so, by Leibniz’s law, must be distinct. But how – the sceptical philosophers ask – can two distinct objects fit into the very same volume? And why is it that when the 100-pound statue and the 100-pound parcel of clay are both placed on a scale, the scale reads “100” rather than “200”? Perhaps there are ways of removing these worries. Perhaps, for example, we can speak of a special relation between entities that 62

Realism about material objects not only permits, but requires, co-occupancy of a single volume of space – a relation that requires that if a statue occupies a particular volume of space, some parcel of matter must occupy that volume as well. (This relation, if it exists, is called “constitution.”4) But it seems harder to believe that when we look at a human adolescent, we are actually seeing two distinct objects, co-occupying a single volume. It seems that a human being can be there before us only by there being, before us, a child or an adolescent or an adult: it seems to lie in the very nature of a human being that, at any time, it is either child, adolescent, or adult. The connection between the statue and the matter that (at any particular time) composes it seems, intuitively, to be looser. It seems that that very statue could have been there before us, at this very time, even though a cunning vandal had replaced some of the clay with some putty; or had replaced half of the clay with putty; or had even replaced all of the clay with putty (cf. Zangwill’s plastic David ).5 Indeed a wide range of variation seems to be possible in the material composition which that very statue could have had – and the range of variation seems not to be crisply delimited. At the least, no delimitation seems to follow from what it is for a statue to be a statue. But it does follow, from what it is for a human to be a human, that a human must at any time exist in one life-phase or another. Modulo quibbles about our customary taxonomy of life-phases, this does entail – as I said just above – that a human can be there before us only by there being before us a child, or an adolescent, or an adult. For that matter, it follows from what it is for a human to be a human that any human is, at any time, either free from disease, or affl icted by one or more diseases – which makes it equally awkward to suppose that there might be two distinct objects before us, namely a human and a human diabetic. Certainly it seems plausible that some H 2O can be present before us only by there being some ice or some liquid water or some steam before us. In sum, the fi rst option for avoiding self-contradiction seems strained and ad hoc. Might one hold – as Option 2 does – that the membership conditions for some natural kinds double as persistence conditions for the members of those kinds, while other natural kinds have membership conditions that do not amount to persistence conditions for 4 5

Baker, Metaphysics of Everyday Life. Zangwill, “Aesthetic Functionalism.”

63

Familiar objects and their shadows their members? The suggestion has a certain appeal. Option 2 might enable us to claim that a human adolescent could cease to belong to the kind adolescent without ceasing to exist, while also claiming that once that same human adolescent ceases to belong to the kind human being, he really does exist no longer. But there is a serious problem with Option 2. If it has traditionally seemed plausible that an object could not lose the properties by virtue of which it belongs to its natural kind without ceasing to exist, that is not because it has seemed plausible that there is a mere extensional equivalence between the properties that are membership conditions and the properties that are the object’s persistence conditions. Rather it has seemed plausible that an object must retain certain of its properties to go on existing just on account of the fact that it needs those properties to belong to its natural kind. To put it differently, it has seemed plausible that, for any object in nature, membership in the natural kind to which it belongs is a life-and-death matter. Option 2 needs to say that the properties required for membership in certain natural kinds (but not others) are, additionally, existence-requirements for the members of those kinds – but it cannot say that they are existence-requirements just in virtue of being required for those objects to belong to a natural kind that they do belong to. For in that case, whenever certain properties are required for any object to belong to a natural kind that it does belong to, those same properties would have to amount to persistence conditions for that object. No, something more must be added, to the mere fact that certain properties are required for an object to belong to a given natural kind, for those properties to qualify, additionally, as persistence conditions for the object. Certain natural kinds (and not others) have somehow to be dignified as “career- defi ning” – as being such that membership in them is a life-and-death matter. What extra factor could confer this dignity – could make membership conditions amount also to persistence conditions? The danger for Option 2 is that the extra factor it identifies will be one that is supplied by us. Option 2 might say, for example, that it is our practices or conventions that render membership in certain natural kinds, and not others, career-defi ning. Then the courses of existence that do obtain in nature – the careers, in other words – would not be mindindependent. This danger can be underestimated. It might seem, for example, that a realist about persistence could avoid it by enlisting Baker’s 64

Realism about material objects distinction between “primary” and nonprimary kinds.6 The “primary kind” to which an object belongs, in Baker’s parlance, is the kind one would mention in order to say “what that object most fundamentally is”;7 a nonprimary kind to which that same object belongs is a kind such that membership in it reflects merely the object’s condition or situation or circumstances.8 Thus the primary kind to which an individual person belongs might be person or human being; the same person might belong to the nonprimary kind husband or painter. If one held – as Baker does hold9 – that it is membership in only its primary kind that fi xes an object’s persistence conditions, and not its membership in any nonprimary kinds, then it seems, intuitively, that we get just the results which Option 2 seeks to secure. For intuitively, it seems that, for any adolescent human diabetic, the primary kind is human being, and that human adolescent or human diabetic is only a nonprimary kind. Perhaps one could extend Baker’s distinction to matter-names: perhaps there is a distinction between matter-names that tell us, in regard to some particular matter now before us, what that matter most fundamentally is, and those that tell us merely of the condition or circumstances in which that matter is found. In that case we might get the sort of result that Option 2 seeks to secure in the case of ice and H 2O. For it does seem intuitively plausible that when we confront some frozen water, the right answer to the question of “what that matter most fundamentally is” is “some H 2O,” rather than “some ice.” But unless we can point to more than mere intuitions to ground the distinction between the natural kind human beings and the putative natural kind human adolescents – or between the natural kind H2O and the putative natural kind ice – then the distinction between “primary kind s” and “nonprimary kind s” does not yield an implementation of Option 2 that is compatible with realism

6

7

8 9

Baker, Persons and Bodies, pp. 39–40; cf. Baker, Metaphysics of Everyday Life, pp. 33–37. Baker does not offer a term for the kinds that are not primary, though she believes there are many; so “nonprimary kind” is my own coinage. The same idea occurs in Wiggins , Sameness and Substance Renewed. All primary kinds that our language equips us to pick out are picked out by what Wiggins calls “substance sortals,” and “substance sortals” (as opposed to “phased sortals”) are those that serve to give “the most fundamental kind of answer to the question ‘what is x?’ ” (p. 30). Baker, Persons and Bodies, pp. 39–40; cf. Baker, Metaphysics of Everyday Life, pp. 33–37. Baker, Persons and Bodies, p. 39.

65

Familiar objects and their shadows about persistence.10 For it seems plain that our intuitions as to which kinds it is, membership in which is constitutive of what an object (or a parcel of matter) “most fundamentally is”, are connected with our intuitive judgements as to which sorts of alterations the object (or matter) in question could survive. But if so, a proponent of Option 2 who enlists Baker’s distinction will be telling us: “Only some natural kinds are such that membership in them is career- defi ning. The natural kinds in question are just those that we intuitively judge to be career-defi ning.” The most natural way of interpreting such a claim is to read it as saying that membership in a particular natural kind is career-defi ning just in virtue of our considering it to be career-defi ning – which of course is miles away from realism about the courses of persistence traced out by nature’s objects and parcels of matter. Alternatively, the claim could be read as saying that what it is for some natural kind to be career-defi ning is one thing, what it is for us to consider it career-defi ning is another, but that all natural kinds that meet the latter qualification also meet the former. But this sounds like a marvelously convenient coincidence, and in any case leaves unanswered the question as to what it is for a natural kind to be career-defi ning. 3.5

Infimae species

Now for Option 3. I hold that what we said about natural kinds, in chapter 1, motivates an account of natural kinds on which the putative lower-level kinds such as ice, adolescents, and diabetes sufferers are not true natural kinds at all. If so, we are free to embrace the traditional idea that kind-membership furnishes persistence conditions without fear of self-contradiction. A main topic in chapter 1 was the role played in our conceptual economy by natural kinds. I said that judgements of kind-sameness 10

Baker does not claim to give a systematic, more than merely intuitive, explication of the distinction. But she does (in Persons and Bodies, p. 41) “suggest” one relevant “consideration”: an object’s membership in its primary kind “brings about instantiation of whole classes of properties” that its membership in a nonprimary kind does not bring about. Unfortunately, this does not work for the case of human being vs. human diabetic : membership in the latter kind brings about instantiation of all the properties that membership in the former brings about, and a rich array of further properties as well.

66

Realism about material objects enjoin inductive inferences of particular sorts, and that the injunctions have been fruitful often enough, for judgements of kind- sameness to get established in our conceptual-cum-linguistic repertoire, just because they have often enough corresponded to the presence of properties that cluster together in ways anchored in the causal workings of the world. In just which ways are the properties comprised within the nature characteristic of a natural kind causally connected to one another? On the account of causation that I shall endorse in chapter 5, and that I tacitly relied on in chapter 1, the connection can be said to be a matter of causal control. Property A causally controls coinstantiated property B just in case hypothetical interventions on that instance of A, that replaced A by a distinct but comparable property – one that differed from A to greater or lesser degree – would entrain a replacement of the co-occurring instance of B by a corresponding variant on B – by some property that differed from B to a corresponding degree. By “intervening” on the instance of A, one could “manipulate” the property that replaces the coinstantiated instance of B. (Of course, since we are speaking here of properties essential to the original kind, the outcome of this “intervention” would be found in a distinct, though kindred, natural kind.) Given this understanding of causal connection, the picture that emerges seems to be this: any property incorporated in the nature characteristic of a natural kind must either control, or be controlled by, one or more other properties incorporated in that nature. Now let us compare the credentials of the putative natural kind ice with those of the natural kind water. If ice is a natural kind in its own right, one of the properties by which it is essentially characterized is being at a temperature of 32° F or lower. If ice is merely one among several variant forms of the natural kind water, then any sample of ice has the property of being at a temperature of 32° F or lower only accidentally; the main property by which it is essentially characterized is having molecular structure H 2O. Interestingly, these two putative essential properties are alike in one way. Each would not be controlled by any other property in the essential cluster to which each belongs. Each would be causally integrated into that cluster only by itself controlling other properties within it. It is worthwhile to dwell for a moment on the negative point here. If you sought to alter the molecular structure present in some sample of water – if you sought to remove the structure H 2O from that sample 67

Familiar objects and their shadows of matter, and to replace it by some variant molecular structure – you would not go about the task by seeking to alter the specific gravity of that sample of matter, or its index of refraction, or its melting point or freezing point. For intuitively, the direction of control seems to go in precisely the opposite direction: the molecular structure underlies, and is not shaped by, the “physical chemistry” properties characteristic of water. If you sought to alter the temperature-range property present in some piece of ice – and if ice is a natural kind in its own right, this would amount to destroying the ice, and replacing it with a sample of matter belonging to some other natural kind – you would not go about the project by seeking to alter the color of the ice, or its specific density, or its tactile properties. Instead, you would bring the ice close to a source of heat! And in the case of ice, the obverse point is even more obvious. On a hot summer day, you would not suppose that you can prevent a sample of ice from being destroyed – through the process we call “melting” – merely by leaving undisturbed the color characteristic of ice, or the characteristic specific density, or the characteristic hardness. But each of our two putative essential properties does seem to control others within its putative kind-nature. Molecular structure H 2O controls freezing point, boiling point, index of refraction, specific gravity, and so on through a long list of surface properties. Being at a temperature of 32° F or lower controls the color, hardness, and specific density characteristic of ice. Do the two properties, then, really have equally strong claims to being essential properties, incorporated in a nature characteristic of a natural kind? Intuitively, the answer seems to be No, and the reason is intuitively salient. Intuitively, it seems that being at a temperature of 32° F or lower does not all by itself ensure the color, hardness, and specific density characteristic of ice; rather, it does this only by enlisting the dispositions conferred on any chunk of ice by the molecular structure H 2O. The presence of being at a temperature of 32° F or lower seems only to be an occasion on which molecular structure H 2O exercises its causal control over such surface-level features as color and specific gravity and degree of tactile resistance. And having molecular structure H2O does seem, all by itself, to underlie the dispositions which the advent of being at a temperature of 32° F or lower actualizes. But are we sure this intuitive answer should be trusted? Can we point to a reason not to say, instead, that having molecular structure H 2O 68

Realism about material objects is only an occasion on which being at a temperature of 32° F or lower exercises its control over surface-level features of just the sorts we have mentioned? In other words, can we point to a reason for ruling that the deeper causal explanation is provided by the microstructure property, rather than by the temperature property? Yes, we can. Molecular structure controls a vast range of surfacelevel properties, and does so in a highly articulated way – across a vast range of circumstances. So deep is its control that we can be sure that the molecular structure found in water can occur in instances of no other natural kind. The temperature property characteristic of ice controls only relatively few surface-level properties, and its control is exercised only across a narrow range of circumstances – indeed, at the single nodal point of temperature 32° F itself. Moreover, the control exercised by this temperature-range property, in the case of ice, is confi ned to the putative kind ice itself; there is nothing distinctive about iron found at 32° F or lower, say, or gold. So weak is the control exercised by the temperature-range property, that we can fi nd that very property occurring in the instances of other natural kinds. This, then, is the recipe I offer for weeding out impostor natural kinds from genuine ones – and thus preserving the traditional claim that an object (or a sample of matter) cannot exit the natural kind to which it belongs without ceasing to exist. The properties comprised in the nature characteristic of any genuine natural kind must each either be controlled by other properties within that nature, or must, each in its own right, control other properties within that nature.

69

4 Ontological preference for the temporally small Common sense and learned science alike suppose that the objects that they believe in – or the overwhelming majority of them, at least – persist across changes, exchanging one intrinsic property for another. A point familiar to those who have followed the recent literature on persistence, and amazing to those who have not, is that defending this commonplace supposition takes considerable philosophical effort. The fi rst message of this chapter is that it takes even more effort than has yet been recognized – that a widely favored way of dealing with the philosophical problems that arise here, namely “stage theory” or “exdurantism,” as much betrays the convictions of common sense and learned science as it honors them. For exdurantism, unless supplemented, lands us in the “explosivism” scouted in chapter 1; and the only supplementations that will avert explosivism require that the temporal stages, in which exdurantism believes, be tied together in ways that exdurantism cannot explain – tied together by jointly occupying shadows cast by entities to which exdurantism cannot appeal. The reason why exdurantism is widely favored is that it appears to combine two virtues, of which each rival view can claim only one. Perdurantism has the virtue of ruling, along with our commonplace supposition, that no contradiction is involved in one and the same object’s possessing contrary properties – in the same poker’s being now hot and later cold, or the same person’s being now seated and later standing upright. But it achieves this virtue by holding that familiar objects are temporally extended as well as spatially extended, and that it is only one temporal part of a poker or a person that exists either now or later. Endurantism holds that any real object is present in its entirety at each moment that it exists, which counts as a virtue from our commonplace way of thinking, but it achieves this virtue by holding either that what seem to be intrinsic properties are covertly 70

Ontological preference for the temporally small relational, or that the having of them is adverbially modified, relativized to a time. Only exdurantism holds that a poker or a person or a tree is wholly present at each moment that it exists, and possesses, in a nonrelativized way, straightforwardly intrinsic properties.1 There is of course a cost to achieving these two virtues together, just as there are costs to achieving either singly. Exdurantism provides a surprising reading for “at each moment that it exists”: a dog or a poker or a person is extremely short-lived, says exdurantism; indeed any dog or poker or person lasts for only a single moment.2 The persistence of such objects across changes, then, gets treated as “vicarious persistence”: each persists by virtue of having, at other moments of time, temporal counterparts, dog-stages or poker-stages or person-stages located at those times. This treatment of persistence gains plausibility from its close analogy with Lewis’s treatment of modality, on which (for example) the fact that I could have been thinner today, than I actually am, amounts to my having a modal counterpart, in some other possible world, that is just like me except thinner. 3 But if we really are willing to accept that the objects recognized by common sense and learned science persist in this “vicarious” way, how might it be that exdurantism fails to wholly honor our commonplace suppositions about persistence? Here is what I will argue. Perdurantism pictures familiar and scientific objects as space–time “worms,” spread out across many distinct moments. Exdurantism identifies familiar and scientific objects with momentary slices of these “worms.” But in doing so, I will argue, it slices the worms so fi nely as to sever the ligaments that hold them together in nature. It leaves us – unless it is specially supplemented – with no grounds for ruling that those same slices, those same temporal stages, do not just as much combine with slices found in what common sense and empirical science would regard as quite different space–time “worms.” 1

2 3

For excellent overviews of these three positions, see Balashov, Persistence and Spacetime, Ch. 2, or Haslanger, “Persistence through Time.” “Exdurantism” is named after the phenomenon in which it believes, namely exdurance – a form of duration that an entity enjoys by virtue of what is, strictly, temporally outside it, or in other words “vicarious persistence” (see next paragraph of text). The term was coined by Haslanger in “Persistence through Time.” Hawley, How Things Persist, pp. 48–52; Sider, Four-Dimensionalism, pp. 60, 188–208. Sider, “All the World’s a Stage,” p. 437.

71

Familiar objects and their shadows Before us, say common sense and biology, there is a trout swimming in a brook; what we’re seeing is actually a momentary trout-stage, says exdurantism; but this same stage can claim as temporal counterparts, not just subsequent trout-stages, but subsequent turkey-stages (as in the Lewis example given in chapter 1), and even immediately subsequent argon-atom-stages located well outside the light cone of the current trout. Exdurantism, I will argue, gives us exactly the strangely persisting continuants in which explosivists believe. At least, it does so unless supplemented with special premises. These premises would have to assign ontological weight to some temporal counterpart relations – those, for example, that tie a present troutstage to later and earlier trout -stages – and would ban, as not really there in the world, the counterpart relations that would give us the explosivist’s strange continuants. But from whence could the privilege derive? Only, I shall argue, from a projection downward onto temporal stages of courses of persistence that obtain independently of temporal stages and their counterpart relations. Let me add one word about the motivations behind our three rival positions, namely endurantism, perdurantism, and exdurantism. Thus far I have mentioned only motivations drawn from commonplace suppositions about persistence. There are others, having to do with the ways these views respond to distinctly philosophical problem cases. I will give a fuller discussion of motivations in the last section of this chapter. Only then will it have become clear that remarkably powerful considerations are needed to motivate exdurantism. None is available, I shall argue. 4 .1

A n e x plosion of t e m por a l c ou n t e r pa r t r e l at i o n s ?

The core of my objection against exdurantism will be that, barring special supplementation, it must recognize, for any individual temporal stage, too many temporal counterpart relations. Some of these will secure for the temporal stage “vicarious” persistence across the sort of career that common sense and empirical science consider typical for the members of the natural kind to which that stage seems to belong; but others will secure for it “vicarious” persistence across bizarre sorts of careers. But the problem is not that exdurantism recognizes, for any individual temporal stage, more than one 72

Ontological preference for the temporally small sort of temporal counterpart relation. “Too many” here does not mean “more than one.” It is, on the contrary, an attractive feature of exdurantism that for some individual temporal stages, it recognizes more than one sort of temporal counterpart (“t-counterpart,” let us say) relations. For it is in this way that exdurantism provides a neat solution to standard puzzle cases that appear to involve “coinciding objects.” Suppose, for example, that before me now is a statue of Atlas made entirely of gold . Suppose too that over the next half-hour a vandal wielding a blowtorch will leave, in place of the statue, a mere pool of melted gold. Then how many objects are before me now? One natural-seeming answer is that two objects are there, occupying exactly the same volume of space. For the statue will be destroyed a halfhour from now, while the sample of gold (sometimes called “the parcel of gold” or “the mass of gold”) will still exist; by Leibniz’s law, these must be distinct objects. At the same time, however, this answer can seem incredible, as we noted in the previous chapter. How can two distinct objects manage to fit into the same volume? Why is it that when we place the 100-pound statue on a scale, and also (thereby!) place the 100-pound parcel of gold on the scale, we get a reading of “100” rather than “200”? Exdurantism supplies a ready answer. What is before me now is a single object-stage, but it is tied by t-counterpart relations of two different types to stages both subsequent and previous. It is tied by the counterpart relations characteristic of statues to a relatively limited range of future stages (and of past stages), and by the counterpart relations characteristic of samples of particular kinds of matter to a relatively wider range of future (and past) stages.4 If I were to point to that statue and say, “well, but how long will that object last?,” the answer would depend on which sortal I supply in explaining just which object I mean. Both t-counterpart relations characteristic of statues and t-counterpart relations characteristic of lumps of gold are “out there,” prior to my supplying a sortal; what my employment of a sortal does is to make salient which t-counterpart relations are relevant to semantic assessment of a claim about persistence. The same point explains why, depending on the sortal employed, the attribution to an individual temporal stage of a particular 4

Sider, Four-Dimensionalism, Ch. 5; cf. Hawley, How Things Persist, pp. 156–58, 183–89.

73

Familiar objects and their shadows “historical property” or “lingering property” can switch from being true to being false. “Historical” properties and “lingering” properties are properties that no individual stage can instantiate by virtue of what it is like intrinsically over the meager course of its own existence: any stage can possess such a property only “vicariously”; and possession of a historical property will depend on how matters stand with t-counterparts to that stage that are temporally prior to it, while possession of a lingering property depends on how matters stand both with t-counterparts prior and subsequent. Thus consider “Ted was once a young boy.” Since “Ted” counts for us as a person-name, according to exdurantists, semantic evaluation of that attribution of a historical property will depend on how matters stand with earlierexisting counterparts to which the stage picked out by “Ted” stands in the particular t-counterpart relations proper to persons. Person counterpart-hood can tie a present stage only to earlier-existing person -stages. The stage called “Ted” is spatiotemporally continuous with a whole series of earlier person-stages, and some of these qualify as young. So “Ted was once a young boy” is true.5 But “Ted was once a fertilized ovum” might well be false if, as seems plausible, a fertilized ovum is not in its own right a person. At the same time, “that living creature began as a fertilized ovum,” said while pointing to Ted, might well be true. Or consider Yuri Balashov’s example of a “lingering property.”6 “In the summer of 1860, Lev Nikolaevich Tolstoy was at Yasnaya Polyana, writing War and Peace” should come out as true; and if we imagine ourselves back in Tolstoy’s study in 1860, pointing to him writing at his desk, “Graf Tolstoy is writing a novel titled War and Peace” should also turn out to have been true, even though its truth may not have been something of which we could be certain at the time we said it.7 But the property attributed by “is writing a novel” (or “was writing a novel”) cannot be instantiated by any temporal stage by virtue of what it is like, intrinsically, over the meager span of its own existence – and assuredly not the property attributed by 5

6

7

Sider, “All the World’s a Stage,” pp. 437–38 and 446–47; Sider, Four-Dimensionalism, pp. 193–96. Balashov, “About Stage Universalism,” p. 24; cf. Hawley, How Things Persist, p. 65, or Sider, Four-Dimensionalism, pp. 197–98. Quite un certain, in fact. Tolstoy’s original working title was 1805.

74

Ontological preference for the temporally small “is writing War and Peace”! Any stage must earn that attribution “vicariously,” by virtue of how matters stand with a vast host of temporal stages both previous and subsequent. Which sorts of temporal stages? “Tolstoy,” like “Ted,” is a person-name. The t-counterpart relations appropriate for evaluating a sentence that features that name as subject are person-counterpart-relations, and these can obtain only between a present stage that is a person and stages earlier and later that themselves are persons. But the referent of our token of “Tolstoy” was spatiotemporally continuous with just such a host of other person stages, and each of these was doing something that could help earn for that referent the property “was writing War and Peace.” The attribution is true. The same semantic story shows why “that person is writing War and Peace” would have been true. But now consider, Balashov says, the sortal “writer–cucumber.”8 A “writer–cucumber” is an entity that is a writer up to a certain point in time, and after that time is a cucumber (cf. “trout–turkey”). Imagine again that we are back in Tolstoy’s study; that satisfaction at the present time of “writer–cucumber” requires that a person exist now and that a cucumber exist five minutes from now; and that we point towards Tolstoy’s desk and say, “that writer–cucumber is writing a novel titled War and Peace.” Here the attribution of our lingering property is false. At the same time, “that writer–cucumber would provide a delicious crunch in the salad at our dinner tonight” may be true. If nonstandard sortals such as “trout–turkey” and “writer– cucumber” make nonstandard sorts of t-counterpart relations semantically relevant, then all manner of bizarre claims about courses of persistence, traced out by objects in our familiar environment, will turn out true. We will have exactly the “explosion” in courses of persistence that explosivism affi rms. But chapter 1 has supplied a reason why the exdurantist should seek to avoid explosivism. How might she do so? One obvious route would be to fi nd fault with nonstandard sortals such as “trout–turkey” and “writer–cucumber.” These are bizarre philosophical inventions, the exdurantist could say. They appear to make nonstandard t-counterpart relations relevant to the semantic evaluation of attributions of lingering properties, and of other claims about courses of persistence. But 8

Balashov, “About Stage Universalism,” p. 30.

75

Familiar objects and their shadows really, no such t-counterpart relations are semantically relevant. The only t-counterpart relations that matter, semantically, are those associated with the standard sortals that we actually employ. But this response just invites a further question: what is special or privileged about the sortals that we actually employ? Is it merely that we actually employ them? Unless we can secure for our familiar sortals some further credential, beyond the mere fact that they are emplaced in our actual linguistic and conceptual practices, we will be in danger of the antirealism about persistence scouted in chapter 3. For unless we do this, which claims are true, about the courses of persistence that are traced out by the world’s objects, will be a function of ourselves and our conceptual/linguistic practices; and then, by disquotation of “is true,” which courses of persistence are traced out by the world’s objects will be a function of ourselves and our conceptual/linguistic practices. What the exdurantist really must say, if she is to avoid explosivism, is that the sortals we actually employ have – for the most part – the further credential of being associated with t-counterpart relations that really obtain, out there in the world. Bizarre sortals, for the most part, make us think of t-counterpart relations that do not obtain at all. Temporal-counterpart-hood is a language-independent, mindindependent phenomenon. As the point is put by the exdurantist Katherine Hawley – to whose views I will return in section 4.4 – “there are objective differences between series of stages which correspond to ordinary objects, and those which do not.”9 We are familiar with instances of this phenomenon, the exdurantist can now say: instances of it are what bind together the momentary stages that together make up the careers of the familiar objects in which common sense and much empirical science believe. But this still leaves us with a philosophical question. What is the nature of this phenomenon – what is its correct analysis? 4.2

R e s t r i c t i n g t h e t- c ou n t e r pa r t r e l at i o n s t h at R E A L L Y o b ta i n: f i r s t at t e m p t

Or, to put this philosophical question differently, what is distinctive about the courses of persistence traced out by familiar objects? 9

Hawley, How Things Persist, p. 90.

76

Ontological preference for the temporally small The answer, I argued in chapter 1 (pp. 19, 28), is that they are axes of causal control. Newly arriving accidental states cause, by way of regularities characteristic of the natural kind to which the object belongs, either the continued presence of those very accidental states, or replacement of them by particular other accidental states. This phenomenon of causal control is indeed the reason why we even have, in our conceptual/linguistic repertoire, judgements of numerical persistence. But what creates the raison d’être for those judgements is not just a matter of the ways in which newly arriving accidents actually, at a present moment, shape particular accidental sequels. It is also a matter of ways newly arriving accidents would shape, or will shape, or have shaped particular accidental sequels. What is crucial are the dispositions, characteristic of any natural kind, to respond to the advent of accidental properties in a particular range by passing into particular other accidental states. Might exdurantism use this account to answer the remaining philosophical question – might it avoid explosivism by maintaining that the t-counterpart relations that really obtain, out there in the world, are grounded in these accident-governing dispositions? But for exdurantism, accommodating dispositions (and dispositional properties) is not a straightforward matter. Any view must concede that a given entity can possess a particular dispositional property only if it is at least metaphysically possible that that entity should manifest the disposition in question. “Finkish” intrusions may make it causally impossible, in a given case, for an entity to manifest a dispositional property that it really has; but at least the metaphysical possibility of manifestation is there, and it is the role of the fi nkish factor to occlude it. For an entity to manifest a disposition, however, that very entity must fi rst undergo the “triggering” event that is characteristic of the disposition, and then undergo the “response” event that manifests the disposition. But no individual temporal stage lasts long enough to undergo both events: any individual temporal stage lasts no longer than the shortest physically possible alteration.10 Thus 10

Hawley, How Things Persist, pp. 48–50. Many philosophers hold that the laws of nature do not hold in all metaphysically possible worlds – not even all worlds in which the subject matters of the laws obtain – and consequently hold that events that are physically impossible, at a given world, may nevertheless be metaphysically possible at that world. They might then argue that even if, in the actual world, it is physically impossible that a temporal stage should last from the “triggerin g” event for a dispositional

77

Familiar objects and their shadows no temporal stage can instantiate a dispositional property by virtue of what it is intrinsically like, over the meager span of its own existence. If a temporal stage is to have a dispositional property, it must have it “vicariously” – in much the way that a temporal stage has historical properties or lingering properties. But thus far we have only shown that the exdurantist has extra work to do – not that he cannot hold that the t-counterpart relations that really obtain in the world are grounded in the accident-governing dispositions that tie together the courses of persistence characteristic of familiar objects. Thus far, all that the exdurantist has to do is to make out the claim that any dog or person or tree does have the dispositions that tie newly arriving accidents to particular accidental sequels. When we speak of “a dog” or “a person” or “a tree,” the exdurantist says, we really are speaking of a dog-stage or person-stage or tree-stage. So what the exdurantist needs to do is to say how such a stage can have, albeit “vicariously,” the standard accident-governing dispositions. In general, this is the sort of “vicarious” account that an exdurantist can offer for dispositional properties. A given temporal stage can, in the simplest case, undergo the “triggering” event for a particular disposition, even within the meager course of its own existence; then all that is needed is that some subsequent temporal stage should figure in the “display” event – where that subsequent temporal stage is a t-counterpart of the earlier one. Or, if the “triggering” event takes too long to be consummated in the meager span of one stage’s existence, what is needed is that one stage should undergo the beginning phase of the “triggering” event; have t-counterparts that successively undergo the rest of the stages in the “triggering” event; and have yet later counterparts that, successively, undergo the various stages in the “display” event. property which that stage really has, through the “display” event that manifests that disposition, this is still metaphysically possible, and hence that any individual stage can in its own right possess a dispositional property – that it need not have such a property just “vicariously.” What this response overlooks is that it is supposed to lie in the very nature of a temporal stage that it lasts for less time than any physically possible change. It follows that in any possible world in which a given temporal stage exists (or: has a modal counterpart), it is, at that world, physically impossible that that stage (or modal counterpart) should last from “trigger” to “display.” Thus there is no possible world in which a particular temporal stage itself lasts over the course of a display of one of its dispositions.

78

Ontological preference for the temporally small But now we do see a genuine obstacle for the exdurantist. In order for it to be the case that any given stage has accident-governing dispositions – or any particular dispositions – there must independently be a fact of the matter as to which subsequent temporal stages are its t-counterparts. Correct attribution of a dispositional property presupposes an answer on where the t-counterpart relations lie. Hence a stage’s possession of a dispositional property cannot fi x the facts as to which subsequent stages are its counterparts. The exdurantist cannot, without circularity, use accident-governing dispositions to privilege a particular distribution of t-counterpart relations. 4 .3

R e s t r i c t i n g t h e t- c ou n t e r pa r t r e l at i o n s t h at R E A L L Y o b ta i n: s e c o n d at t e m p t

Thus far we have been supposing that the exdurantist who seeks to avoid explosivism would want to say that the world’s t-counterpart relations are fi xed by the very phenomenon that makes it useful for us to assert claims of numerical persistence, according to the argument of chapter 1. This phenomenon is the causal control exercised over accidental states of an object by earlier accidental states of that same object. But the nonexplosivist endurantist could give an alternative answer that is nearly as good. He could hold that the world’s t-counterpart relations are fi xed by a phenomenon coextensive with causal-control-of-accidents. This is the transmission of essential properties from one temporal stage to the next. This answer too would keep exdurantism free from explosivist consequences: the world’s t-counterpart relations would then obtain where and only where common sense (and empirical science) are, by their lights, entitled to claim that enduring objects trace out their careers. And exdurantists do have reason to claim that just such transmission of essential properties occurs, from one temporal stage to the next. That is, they have reason to suppose that very widely, the presence of essential properties p1, …, pn in a temporal stage at t causes the presence, in some spatially adjacent temporal stage at t+moment, of p1, …, pn. For exdurantists must hold that it is no accident that the world sustains the illusion of enduring objects that retain their properties across time. Some philosophers explain this appearance by positing “immanent causality”: they say that an object’s having 79

Familiar objects and their shadows p1, …, pn at t causes the having of p1, …, pn, at t+moment, by that very object. For exdurantists, the real explanation for the endurantist’s “illusion” must lie in facts about temporal stages, and none of these is present at more than one moment. So exdurantists have reason to assert an analogue of immanent causation, namely causal transmission of essential properties. But just what account of causation can exdurantists invoke, in asserting this transmission? In chapter 5, I shall endorse the “invariance” account of causation advocated by Woodward and Hitchcock. On this account – to put it roughly for now – a putative cause c really does cause event e just in case variants on c, more and more different from c itself, would get followed by variants on e, more and more different from e itself. But this model has no clear application to the causal transmission of essential properties. A dog at t has the properties essential to dogs, and an electron at t has the properties essential to electrons. But it makes no clear sense to ask, “what if that very dog [or electron]” – which, for exdurantists, is a question about a dog-stage [or electron-stage] at t – “had had properties other than those essential to a dog [or to an electron]?” But other accounts of causation are available. Most of them make counterfactual dependence the core of causation. Suppose, for discussion, we pick the Mackie/Bennett version of such an account.11 Suppose we say that a cause of e is a Necessary part of an actually occurring s ufficient condition for e – or, to say it Mackie’s way, that a cause of e is an Insufficient but Necessary part of an Unnecessary but Sufficient condition for e, an INUS condition for e. (I have elsewhere12 argued that this model of causation, at the level of type causation, actually amounts to something much like the invariance account; but they sound different, and so let us treat them as different.) In this way or another, then, the exdurantist might claim that there is a temporal stage at t+moment having the properties essential to dogs as a causal consequence of the fact that there is a temporal stage at t having the properties essential to dogs. But why and how will this claim seem plausible? Why should the way some entity is at t exactly mold the way some distinct existence is at t+moment? Why wouldn’t such “exact molding” amount to a case of magical influence, a strange temporal 11 12

Mackie, “Causes and Conditions”; Bennett, Events and Their Names. Elder, Real Natures and Familiar Objects, Ch. 2.

80

Ontological preference for the temporally small version of “action at a distance”? Certainly it would seem puzzling that the presence of the properties essential to dogs, in a temporal stage at t, should instantaneously cause those same properties to be present in spatially adjacent temporal stages also located at t. Why is it any easier to understand how that same presence should cause the presence of those same properties at a temporally adjacent point – in a temporal stage located at t+moment? And if a temporal stage at t, bearing the properties essential to dogs, can exactly mold (at least in essential respects) a temporal stage located at t+moment, why might it not mold many such stages – or why not stages located at t+10 minutes? The answer seems clear enough: we have been thinking all along that the temporal stage at t+moment, located spatially just next to the stage we began with, is the t-counterpart of that original temporal stage. It is the temporal stage that “stands in,” at t+moment, for that original stage; it is the temporal stage that we must be talking about, when we make claims of the form “the very dog, present at t+moment, that existed at t.” But then the claim that a given temporal stage causally transmits its essential properties to an immediately subsequent temporal stage relies, for its plausibility, on the presupposed idea that the immediately subsequent temporal stage is its t-counterpart. Temporalcounterpart-hood is presupposed for the plausibility of the causal claim. Hence the causal claim cannot be held to analyze what t-counterpart-hood consists in. A putative analysans that isn’t believable, unless we already have warrant for believing in the analysandum, is not fit for its assigned role. 4.4 “Nonsu pe rv e n i e n t” t- c ou n t e r pa r t r e l at i o n s To avoid explosivism, exdurantists must privilege just those t-counterpart relations that would assemble stages into the courses of persistence that common sense and empirical science attribute to familiar objects. They could do so, I have argued, neither by grounding the privilege on the dispositional control of accidental states, nor by grounding it on the causal transmission of essential properties. Could they do so by saying that the privilege is “brute” – not something that is founded on the intrinsic character of the temporal stages that are related by the privileged t-counterpart relations? That is 81

Familiar objects and their shadows exactly the position affi rmed by Katherine Hawley. Stages are joined by “nonsupervenient” t-counterpart relations. Hawley writes, I claim that there are relations between the distinct stages of a persisting object which are not determined by the intrinsic properties of those stages. … What are these non-supervenient relations? They are the relations, whatever they are, which underpin the relation of “immanent causation” which holds between stages of the same object.13 The later stage depends for its state upon that of the earlier because they are stages of the same object; because, according to me, they stand in a non-supervenient relation to one another.14

One certainly can see Hawley’s motivation for assigning special weight to such nonsupervenient t-counterpart relations. If they really were out there in the world, they would explain why the world appears to be populated by entities that respond to accidental impingements by passing into particular accidental sequels – while appearing to persist across the whole response – in ways governed by the essential properties characteristic of the kinds to which they belong. They would also explain the appearance of “immanent causation” that transmits essential properties across moments, if explanation indeed is needed. The endurantist could say that none is needed. He could claim that the object that exists at t+moment, bearing thus-and-so essential properties, is numerically the same entity as the object that existed, at t, bearing just those essential properties – and so no explanation is needed for why it, this very entity, has the same essential properties at t+moment as at t. But an exdurantist might well want an explanation of why the slice of the space–time worm that exists at t+moment appears to have been exactly molded by that distinct existence, the slice that existed at t, and Hawley’s nonsupervenient t-counterpart relations would supply the explanation. But of course Hawley’s position here appears to be deeply ad hoc. One needs to explain why it appears that familiar objects persist across causally regulated accidental episodes, and so one “discovers” relations binding nonpersisting, momentary entities that will enable the momentary entities to mimic this appearance. Why is it not more ontologically parsimonious to say that there are the very entities that 13 14

Hawley, How Things Persist, pp. 85–86. Hawley, How Things Persist, p. 87.

82

Ontological preference for the temporally small there appear to be – entities that persist across such episodes, in whole courses of persistence? One needs – if committed to exdurantism – to explain why essential properties get transmitted across the successive slices of space–time worms in just such a way as to sustain the illusion that enduring objects retain their essential properties across their careers. Why not just say that there is a cross-temporal unity to the worms that need not be reassembled at the level of stages – or that, indeed, there really are enduring objects? Hawley’s nonsupervenient t-counterpart relations are a textbook case of what I call “a shadow.” 4 .5

M o t i vat i o n s f o r e x du r a n t i s m

On the face of it, exdurantism is not a particularly attractive position. Indeed it almost appears to be paradoxical, as I shall explain in a moment. So there is a serious question as to what explains its widespread popularity. In chapter 8 I shall offer some nonstandard speculation on this question. Here, I shall review two of the motivations for exdurantism brought forth by its proponents – one from Ted Sider, and one from Katherine Hawley. I shall suggest that these motivations have little persuasive force. If I am right about this, it provides some warrant for my speculations in chapter 8. But fi rst for the appearance of paradox that exdurantism presents. The view claims that whenever we make a claim about a dog or a tree or a person – with the isolated exception of “diachronic counting”15 – we really are saying something about an individual dog-stage or treestage or person-stage. An individual temporal stage, proponents of this view agree, is extremely short-lived; Hawley argues persuasively that we should consider every temporal stage to last for a shorter time than any physically possible change. But claims such as these, says the view, are commonly true: “this dog has been alive for seven years”; 15

Suppose that a bunch of bananas has, all morning long, lain undisturbed in a bowl on the kitchen counter. Then to get the right answer to the question “How many bananas have been present in my kitchen over the course of the morning?,” exdurantists say we must not count individual banana-stages. This is the one sort of case in which, when speaking of “a banana,” we are not speaking of an individual banana-stage. Instead we must construe the question “How many bananas?” as asking about spatiotemporally continuous series of banana-stages. (See Sider, “All the World’s a Stage,” p. 448; Hawley, How Things Persist, pp. 63–64.)

83

Familiar objects and their shadows “Ted will one day be an old man”; “that person is writing a novel with the title War and Peace.” The appearance of inconsistency here is strong. One way to dispel it would be to ban, in ordinary contexts, disquotation from sentences that end “… is true”: that is, to hold that in saying “ ‘this dog has been alive for seven years’ is true,” one is not committing oneself to saying that some dog has been alive for seven years. Another would be to say there is no contradiction between saying that some dog lasts only for a moment, and saying that that very dog has been in existence for seven years. The third way would be to claim that some of the statements we affi rm must be taken as having serious ontological import, and others must not. “That dog has been in existence for seven years” does not assert that some entity genuinely exists, which has existed for seven years; “that dog is a temporal stage” does assert that some entity genuinely exists, but over only a momentary span of time. Actual exdurantists in effect elect this last position, as do many philosophers who think that the apparent ontological commitments of common talk can and should be “paraphrased away.” But there is always the thorny issue of how, in a principled way, to segregate the things we say that do have genuine ontological import from those that do not. Here is a motivation for exdurantism offered by Sider in the very article in which he introduced the view into contemporary debate.16 The motivation presupposes that perdurantism is a more defensible position on persistence than endurantism, and seeks to show that exdurantism has an advantage over perdurantism. The sort of case in which this advantage emerges, Sider suggests, is typified by a Parfitlike split-brain scenario.17 Ted undergoes some ghastly accident, and doctors can save Ted only by transplanting his brain. Two potential recipients are available, each intact except for having an empty cranium. The doctors transplant one of Ted’s hemispheres into the fi rst recipient, the other hemisphere into the second. Both transplants succeed, it seems: both Recipient 1 and Recipient 2 seem to have personalities much like Ted’s, and appear to recall many events that Ted experienced. But Recipient 1 (“Ed”) is distinct from Recipient 2 16

17

Sider, “All the World’s a Stage,” pp. 437–42; cf. Sider, Four-Dimensionalism, pp. 188–208. Parfit, Reasons and Persons, pp. 254–66.

84

Ontological preference for the temporally small (“Fred”), by Leibniz’s law. It cannot be that both Ed and Fred are Ted, still continuing to exist, since that would violate the transitivity of identity. Yet if neither is Ted still continuing to exist, the doctors have, through their zeal to save Ted, killed him after all. David Lewis proposes a solution, from the perdurantist standpoint.18 Ed and Fred are temporally extended “worms,” and they have in common a long temporal part – the part that we picked out by our use of the name “Ted.” The idea that temporally extended entities can share a part is, after all, entirely acceptable from the perdurantist standpoint. It is how one explains the protracted (apparent) colocation of what appear to be two distinct objects, the statue of Atlas and the parcel of gold. Really, after Atlas is formed and before the vandal arrives, there is just one temporally extended entity there. But we can think of this one entity as fl anked by further temporal parts, both earlier and later – as we do when we call it “a parcel of gold” – or can think of it as having no further temporal parts outside the period in question (as we do when we call it “a statue”). Just so, the temporally extended entity called “Ted” is a proper part of the longer temporally extended entity called “Ed,” and a proper part of the longer temporally extended entity called “Fred.” But if we ask what this entails about the relation between Ed and Fred, Sider says, we fi nd that we have a result that is very hard to swallow. Ed and Fred, it turns out, for years occupied one and the same human body. Rather than say that Ed and Fred are each temporally extended entities, which for a long time spatially coincided, it would be better to say that neither Ed nor Fred is a temporally extended entity at all. On any occasion when we say anything about Ed or about Fred, we are really just talking about a momentary temporal stage. Is this really an improvement, though? Now it will be true to say that, at many times, there is a person who is both Ed and Fred – and, what is worse, Ted as well! It is true that we will be saying this only of a momentary person-stage – and perhaps it may not seem so bad if two (or three) persons spatially coincide for just an instant. But an instant is as long as any person lasts, we must remember! No person does last, or could last, longer than a single instant. Given that, spatial overlap “for just a mere instant” is not so “mere.”

18

Lewis, “Survival and Identity.”

85

Familiar objects and their shadows Hawley, for her part, points out that standard positions on persistence tend to face unpleasant problems concerning the end points of a familiar object’s existence; in particular, perdurantism must provide a principled answer as to when a plurality of temporal parts does and does not have a fusion .19 Here is an illustration of my own of this challenge. Suppose that some cat languishes and, as the saying goes, “dies a slow death.” In such cases it can seem hard to maintain that there is a point at which the cat exists no longer – so hard that some philosophers maintain that even after death has arrived, the same entity is still there, but is now a corpse.20 One option, unpopular with all but epistemicists, is to say that the problem here is merely epistemological: there is a precise point at which the cat exists no longer, but we are unable to determine where that point lies. But it would be even worse, many philosophers suppose, to say that there is some entity such that, at various times, its existing or not existing is a vague matter. Existence cannot be vague. (For more on this, see chapter 7, pp. 143-45.) It is far more acceptable to hold that certain properties can be vague – or, to put it as a point about language, that it is vague whether some predicate is satisfied. Perhaps, for example, there really is richness, even though there are cases in which there is no fact of the matter as to whether S is rich. And if vagueness is acceptable for properties, it can be acceptable for t-counterpart relations.21 It can be vague whether a barely breathing cat-stage stands in the cat-appropriate t-counterpart relation to stages found in that same cat’s healthy youth. The only entities are cat-stages, and they are wholly precise. It is the t-counterpart relations alone that are vague. If we had to rule on when a plurality of cat-stages has a fusion – on how many such stages are incorporated in a cat’s life, say – then we would be strongly tempted to recognize vaguely bordered courses of existence. But if we avoid perdurantism, Hawley is thinking, we only have vague cases of the cat-appropriate t-counterpart relations. But this raises anew the question about how we know which statements do and which do not carry ontological weight. Hawley may have given us the means to claim that at certain times, “ ‘Tibbles still 19 20 21

Hawley, How Things Persist, p. 52. Carter, “Will I Be a Dead Person?” Cf. Hawley, How Things Persist, p. 47, on “the inflexibility of the identity relation.”

86

Ontological preference for the temporally small exists’ is not quite true and not quite false.” But – barring a ban on disquotation – this means that at certain times, the right thing to say is “Tibbles does not quite exist still, and does not quite not exist.” We will after all be back at vagueness concerning the existence of an entity, unless we can rule that “Tibbles” does not pick out an entity. But it will be hard to identify a principle behind such a ruling.

87

5 Ontological preference for microphysical causes Might it be held that the microparticles that occupy the volume in which common sense supposes a familiar object to be present really cause anything that the familiar object appears to cause? That is what causal exclusion arguments contend. In this chapter I focus on what most philosophers would consider the most eminently excludable case of putative causation by a familiar object, namely mental causation – the case in which it appears that a person’s beliefs and desires bring about some behavioral outcome. I argue that mental causation appears to be genuine causation, and that no there is no cause at all, at the level of small entities within the person, that can be held to bring about the same outcome. The argument against causal exclusion readily generalizes to other cases of apparent causation by familiar objects, in ways that I shall indicate, both here and in chapter 6. 5.1

T h r e e t h r e at s t o m e n ta l c au s at i o n

Can we vindicate the common-sense conviction that a person’s mental states, and in particular her beliefs and desires, are causally efficacious – that they shape that person’s behavior? Three obstacles appear to stand in the way. The fi rst is the idea that any true causing must instantiate some exceptionless law.1 For it seems clear that the closest we can come to laws that link beliefs and desires to behavioral outcomes are mere generalizations that must be hedged by ceteris paribus clauses. The second obstacle comes from the thought that a person’s mental states supervene upon particular neurochemical states characterizing his brain – and perhaps ultimately supervene upon extremely complex states involving microparticles within his brain. The subvening states of affairs, at either level, may seem to have at 1

Davidson, “Mental Events.”

88

Ontological preference for microphysical causes least as strong a claim to being the causes of those behavioral outcomes that the person’s beliefs and desires seem to cause, as do those beliefs and desires themselves. Since it is hard to believe in systematic causal overdetermination of those behavioral outcomes, the subvening states of affairs thus appear to exclude the putative causation by the beliefs and desires. And even if we can dispel the threat of causal exclusion, we still face a third obstacle. Even if beliefs and desires can somehow be said to cause (aspects of ) a person’s behavior, there is still a question whether they can be said to do so quâ beliefs and desires – in virtue of having the particular contents they do. This third obstacle is an aspect of “the problem of mental quausation.”2 In this chapter I concentrate on the fi rst two obstacles, and then briefly sketch a reply to the third. I begin by arguing that, to judge by recent work by Woodward and Hitchcock, many genuine cases of causation instantiate only some explanatory invariance – where an “invariance” is a generalization that need not be exceptionless. 3 I next argue – in a way not anticipated by Woodward or Hitchcock – that there is typically no invariance, of the sort crucial for a claim of causation, linking a complex microparticle state or event in the brain to those behavioral outcomes that common sense credits beliefs and desires with producing. So the same reasoning that defends mental causation against the fi rst of our obstacles defeats the second obstacle as well: it shows that at the level of the microparticles, there is no causation at all that threatens causal exclusion. Nor will it help the causal exclusionist to move up one level, to neurochemical states in the brain: here too no invariance links the would-be excluding states to behavioral outcomes, except insofar as we take those states as subvening causally efficacious beliefs and desires. Finally, I argue briefly that “the problem of mental quausation,” for beliefs and desires, rests upon a confusion. It supposes that what it is for beliefs and desires to have their characteristic contents is one thing; that their causing – or putative causing – of behavioral outcomes is another; and that we can coherently ask whether it is in virtue of the fi rst thing that the second thing obtains. On the 2 3

Horgan, “Mental Quausation”; cf. McLaughlin, “On Davidson’s Response.” Woodward and Hitchcock , “Explanatory Generalizations, Part I: A Counterfactual Account”; Woodward, Making Things Happen; Hitchcock , “Intransitivity of Causation.”

89

Familiar objects and their shadows view which I will sketch – a view put forth over many years by Ruth Millikan4 – what it is for beliefs and desires to have their contents just is for them to cause, across “historically normal” circumstances, particular shapings of behavior. The having of content includes causal shaping of behavior, and so one cannot properly ask whether it underlies shaping of behavior. Millikan’s view pictures beliefs and desires as states that naturally selected programs or devices in the brain assume, by virtue of their design. Thus beliefs and desires are themselves biological products. And biological products are par excellence entities that produce outcomes not in accordance with exceptionless laws, but only in accordance with explanatory invariances. 5. 2

C au s at i o n a n d e xc e p t i o n l e s s l aw s

A line of thought that has been influential in philosophy of mind ever since Davidson’s paper “Mental Events” maintains that every true causing must instantiate some exceptionless law. This line of thought furnishes the fi rst of my three obstacles to vindicating mental causation. For it does seem clear that any so-called law linking beliefs and desires to behavioral outcomes would have to be hedged by ceteris paribus clauses, clauses that have the effect of saying “except when this ‘law’ fails to hold.” But recent work by Hitchcock and Woodward has put in question the general view on causation that this line of thought defends. The application of that general view to beliefs and desires seems particularly questionable if we suppose – as I shall be assuming – that beliefs and desires are biological products. For biological products – the trait s and devices that natural selection fashions – seem to be characterized by a kind of causality that the insistence on exceptionless laws would rule out. Biological products are therefore the fi rst place we should look, if we want to see reason for suspecting that there is something wrong with the “exceptionless laws” view on causation. Evolutionary biology claims to identify and study traits and devices that are characterized, not just by a recurrent morphology, but also by a shared history of natural selection. Tokens of these trait s proliferate across the generations of a lineage, the thinking runs, as a 4

Millikan, Language, Thought, and Other Biological Categories; White Queen Psychology ; On Clear and Confused Ideas.

90

Ontological preference for microphysical causes causal consequence of effects that ancestor tokens produced in earlier generations – effects that increased the reproductive fitness of their host organisms. Evolutionary biologists commonly take this sort of shared history to be a necessary condition for two individual trait s to be tokens of the same type: the same trait occurs in two different organisms, evolutionary biologists commonly judge, only if the two tokens share not just a common morphology, but also stem from the same sort of selectional history. Thus evolutionary biologists commonly take it as essential to the individual traits and devices belonging to a common type, that they are replicated from ancestor tokens that produced a particular fitness-enhancing effect. But the crucial point is that these ancestor tokens need not have invariably – exceptionlessly – produced that beneficial effect. Natural selection does not insist on perfection. All that is needed is that the ancestor tokens have often enough produced the beneficial effect – often enough to give the genes that code for them selectional advantage over their alleles. Tokens of a given biological trait or device are, essentially, such as often enough to cause a particular kind of effect.5 They almost never are such as exceptionlessly to cause that effect. How might this image of biological design be reconciled with the thesis that all genuine causing is exceptionless causing – causation that can be subsumed under some exceptionless law? Davidson’s own way of handling the apparent causal efficacy of mental events provides the fi rst and most salient suggestion. Exceptionless laws do obtain at the level of fundamental microphysics, Davidson held, and so one need only view any individual mental event as just being some highly complex microphysical event, and the way is cleared for seeing how that mental event might produce a behavioral outcome in accordance with some exceptionless law – provided, of course, one likewise views the behavioral outcome itself as just being some further, equally complex, microphysical event. In parallel fashion, one could argue that whenever a particular biological trait or device produces the sort of outcome for which its biological lineage was naturally 5

Some readers will be willing to grant me that this is true: “necessarily, the tokens of a given biological trait or device are such as – sometimes, but perhaps not always – to cause a distinctive effect.” But they will suppose that the statement expresses a merely de dicto necessity. I argue that it expresses a genuinely de re necessity; see “Different Kind of Natural Kind” and “On the Place of Artifacts in Ontology.”

91

Familiar objects and their shadows selected, what has really happened is that some massively complex event involving a vast plurality of microparticles has, in accordance with the laws of microphysics, produced some further and equally complex event involving microparticles. But this response incurs a cost. It suggests that the taxonomy proper to evolutionary biology does not carve events in the world along causally significant lines. For if the instances of a given type of biological device are viewed as enormous aggregations of microparticles, instances of the same biological type turn out to be instances of highly divergent (and highly complex) microphysical types. One preserves the idea that instances of a given biological type often enough produce a characteristic sort of outcome, but preserves it by representing the different instances of that outcome as being wildly diverse outcomes of a microphysical character. Indeed one reduces away the central biological idea as to what causes instances of the given biological trait or device to get replicated in generation after generation after generation. The central biological idea is that ancestor devices in such a lineage often enough caused the same sort of outcome, that this common causal history itself caused the replication of the lineage, and that this common causal history is part of what holds current instances of the lineage together as a kind. If real causation is causation at the level of microphysics, not one element of this idea remains in place. Evolutionary biology, we must rather judge, studies a strangely sustained illusion: really, highly divergent aggregations of microparticles, located at different points in time, cause divergent microphysical outcomes but sustain, at the level of biological description, the appearance that the same sort of causing has – not invariably, but often enough – not only occurred at these different points in time, but caused fresh instances of such causing. There is a second and less radical way in which one might try to reconcile the standard image of biological design with the idea that all true causing fits under exceptionless laws. One might argue that there is after all exceptionless causing at the level of biological description itself. The idea would be that, for any (type of ) biological trait or device, there are “historically normal circumstances,” of a particular qualitatively specifi able type, in which tokens of the type invariably produce the characteristic effect. These would be just those circumstances of some one particular biologically specifiable type that obtained on all the propitious historical occasions, on which 92

Ontological preference for microphysical causes ancestor tokens actually accomplished the trick that caused them to be replicated into the next generation. Really, however, nothing in actual biological practice warrants the supposition that there are such infallibly supportive circumstances: a device might perfectly well get selected for even if, under the very best of circumstances, it performs only often enough the trick that gave it selectional advantage over whatever devices were coded for by the relevant alleles. Evolutionary biology, in sum, rests on the idea that natural selection is shaped by histories of causation that obtain often enough but not invariably. Before we dismiss this idea, we should ask very seriously whether there might not be a rival account of causation – one on which true causing does not require exceptionless laws. And in fact there is one very plausible rival, put forth over recent years by Woodward and Hitchcock.6 In the next section I will sketch this rival account of causation, and will show that it is compatible with the idea that beliefs and desires cause shapings of the host’s behavior. In the section after that I shall use the same account of causation to confront the specter of overdetermination. The argument there will allow that whenever an individual belief or desire gets tokened in a person’s brain, a vast plurality of microparticles in that person’s brain comes to instantiate some vastly complex microphysical property, but will show that, on the Woodward /Hitchcock account of causation, there is no clear content to the claim that the occurrence of this complex microphysical property causes any shaping of the host’s behavior. The section following will ask whether overdetermination gets entailed by a different route – whether overdetermination follows from the efficacy of physical properties instantiated at a level above that of the microparticles, namely that of neurons and their parts. 5.3

C au s at i o n, i n va r i a n c e , a n d b e l i e f s a n d desi r es

The account of causation put forth by Woodward and Hitchcock takes its cue from what events actually get treated as causes in actual 6

Woodward and Hitchcock , “Explanatory Generalizations, Part I: A Counterfactual Account,” and “Explanatory Generalizations, Part II: Plumbing Explanatory Depth”; Woodward , Making Things Happen; Hitchcock , “Intransitivity of Causation”; Hitchcock , “Tale of Two Effects.”

93

Familiar objects and their shadows sciences. Sciences as diverse as economics, medicine, evolutionary biology, mechanics, and agronomy take themselves to have found a causal connection between two types of events just when they have identified an invariance between events of the two types. An invariance is a relation such that to variations on a putative cause there correspond, or there would correspond, variations on the putative effect. There is a variable that subsumes the putative cause – typically a multivalued variable – and alterations on the value of that variable produce answering alterations in the variable that subsumes the putative effect. (“Produce” is of course a causal locution, but Woodward and Hitchcock do not purport to offer a reduction of causation to noncausal phenomena.) To put it metaphorically, by intervening on the cause variable (call it “Vc”) one can manipulate the effect variable (call it “Ve”). But the manipulation need not always work: the function from values of Vc to values of Ve need not be exceptionless.7 In particular, the dependence of Ve on Vc may itself depend on other factors in the causal system, factors whose character is not fi xed by the value taken by Vc: such other factors will be values of “exogenously determined” variables, and for some settings of these further variables, the dependence of Ve upon Vc may break down. So long as there is some range of settings of the exogenously determined variables such that interventions on Vc will reproducibly and reliably be accompanied by corresponding variations on Ve, there is a true case of causation. But it does matter how wide the range of settings is, across which an intervention on Vc will produce an answering alteration in Ve. The wider the range – the more robust and widely obtaining the invariance – the deeper and more powerful will be the causal explanation yielded by the invariance of the actual value of Ve.8 This account of causation captures much of the intuitive appeal of Lewis’s counterfactual dependence account of causation,9 but in other ways does far better. For example, Woodward and Hitchcock note that the question “what if things had been different with the putative cause?” can have consequences along several different routes for 7

8 9

Woodward and Hitchcock , “Explanatory Generalizations, Part I: A Counterfactual Account”; Woodward , Making Things Happen , pp. 65–70. Hitchcock , “Tale of Two Effects.” Lewis, “Causation.”

94

Ontological preference for microphysical causes the value of Ve.10 Thus they are able to acknowledge that use of birth control pills, by its own direct effect on women’s physiology, has a mild tendency to increase the rate of thrombosis, but also, by virtually eliminating pregnancy – itself a quite robust cause of thrombosis – can have a stronger tendency to reduce the rate of thrombosis. They are able to acknowledge this because they take the question “what if the rate of use of birth control pills had been different?” as an “explicitly non-foretracking” counterfactual.11 That is, they are willing to ask, “what if in this population pregnancy were avoided at the same near-uniform rate as is, in fact, achieved by the use of birth control pills, but the use of birth control pills had been far less than the actual use among those in this population: what effects would this then have had upon the incidence of thrombosis?” This is how they are able to acknowledge that directly, the use of birth control pills mildly increases the incidence of thrombosis, but through its influence on pregnancy rates (and on balance) reduces the incidence of thrombosis. For connected reasons, the Woodward–Hitchcock account deals more elegantly than Lewis’s account does with cases of preemption and late cutting,12 and avoids implausible claims of transitivity in causation that Lewis is hard pressed to avoid.13 Are beliefs and desires tied by invariances to particular shapings of behavior? I now argue that the answer may well be Yes – and that to see this, one must consider systems in which a fair number of exogenously determined variables have just the right settings. My argument rests on two fairly plausible assumptions: that folk psychology is roughly right about the roles that beliefs and desires play in the brain’s operations, and that beliefs and desires are physical states produced by naturally selected programs or devices in the brain. The detailed ramifications of these twin assumptions are much explored in Ruth Millikan’s work, and little explored elsewhere. But the basic picture is pretty much common coin – it is not specifically Millikanian. The idea that beliefs and desires are the product of naturally selected devices entails that beliefs and desires have their own proper 10

11 12

13

Hitchcock , “Intransitivity of Causation,” and “Tale of Two Effects”; Woodward, Making Things Happen , Ch. 2. Hitchcock, “Intransitivity of Causation,” p. 275ff. Hitchcock , “Intransitivity of Causation,” §§1–2, 5–6, 8; Woodward , Making Things Happen, pp. 74–86. Hall, “Two Concepts of Causation”; McDermott, “Redundant Causation.”

95

Familiar objects and their shadows functions.14 The proper function of beliefs is different from the proper function of desires; so I will discuss these in turn. Desires – as folk psychology seems to suggest – have it as their proper function to target various goals for action. If the program that tokens desires was fashioned by natural selection, we can be sure that at least many of the goals that desires historically have targeted have been ones the attainment of which was useful or beneficial for the host. But there may also have been quite a few ill-formed desires, targeting goals the attainment of which would have been useless or outright harmful for their host. There have no doubt been many unrealistic desires – ones that targeted goals that their host could never be in a position to attain or even pursue – and many unactionable desires, desires such that, as things happened to work out, their host was never actually in a position to act on them. Even those desires that targeted useful goals which their hosts actually could and did pursue are not likely to have brought about right away the useful results of goal-attainment. Most desires are probably stored for a fair length of time before their host receives the information that here, now, there is a situation in which the targeted goal can really be attained. So it is not realistic to say that any individual desire all by itself brings about the attainment of some goal that is beneficial to its host. The right circumstances have to be in place, and have to be represented as being in place by the host, before this can happen. (Those are the “exogenously determined variables,” as I shall say presently.) But we can be sure that if the desire-fashioning program is the result of natural selection at all, desires do often enough come to be accompanied by inner registerings that the right circumstances indeed are in place, and do then bring about actions that result in beneficial outcomes. These inner registerings are beliefs. The proper function of beliefs is to represent particular states of affairs as obtaining in the world, states of aff airs that may prove relevant to the enactment of those desires that the host comes to form. If the program that fashions beliefs is itself the product of natural selection, we can be sure that at least often, the states of aff airs that historically have been represented as obtaining really have obtained – the beliefs have been true – and have sooner or later been crucial to the attainment of the goals 14

Millikan, Language, Thought, and Other Biological Categories, Ch. 2, and White Queen Psychology, Ch. 1.

96

Ontological preference for microphysical causes that the host’s desires came to target. More fully, we can be sure that often, beliefs have attuned or adapted the host’s goal-pursuing behaviors to really obtaining environing circumstances, in just such a way that those goal-pursuing behaviors succeeded. But here too we can assume that natural selection left much room for error and misfi ring. The belief-fashioning program may have fashioned a fair number of false beliefs. It may also have fashioned irrelevant beliefs – ones that would never have been relevant to any desires that the host would, in the normal course of things, come to form – and many nonuseful beliefs, beliefs that, as things happened to work out, represented states of aff airs that never came to be relevant to enactment of the host’s actual desires. Even those beliefs that accurately represented states of affairs, to which the host at one point had to attune his goal-pursuing behavior s in order for those behaviors to succeed, are not likely to have brought about their useful attunement right away. Most beliefs are stored for a fair length of time before they can play a role in successfully aiming or adjusting the host’s pursuit of his goals. So it is not right to say that any belief can all by itself cause such attunement-to-circumstances of goal-directed behaviors that the host undertakes. The host has to have formed desires to which the represented states of aff airs are, in the circumstances which the host occupies at a particular time, relevant. (Again, these are the “exogenously determined variables.”) But we can be sure that if the program that fashions beliefs was indeed a product of natural selection at all, historical beliefs often enough were true, came to be relevant to actually present desires, and actually attuned the pursuit of those desires in ways that intersected with the states of aff airs that they represented, so as to result in successful pursuit. In this way, it seems, we can credit particular desires and particular beliefs with causing particular shapings of behavior. Even so, I have not yet squarely faced the question about invariances. Is it really true that – when the exogenously determined variable s are set in just the right ways – variants on a given belief would have gotten accompanied by corresponding variants on particular attunements of behavior? Or that – again fi xing the variables in the right way – variants on a particular desire would have engendered pursuit of a correspondingly variant goal? But the answer does seem to be Yes. We are supposing that natural selection somehow hit upon a device that could produce all manner of useful goal-targeting states, even 97

Familiar objects and their shadows ones the actual enactment of which would depend upon the host’s eventual acquisition of all manner of new beliefs. Similarly, we are supposing that natural selection hit upon a device that could produce all manner of beliefs, including ones that could attune and guide the host’s enactment of all manner of desires that the host might eventually form, but did not yet have at the time the beliefs were produced. In order for the fi rst device to be able to produce so rich an array of desires, and for the latter device to be able to produce so rich an array of beliefs, there would have to be some system by which the devices fashioned new desires and new beliefs.15 That is, the particular states that are the host’s goal-targeting states – its desires – would have to display a certain compositionality; so too would the host’s potentially action-attuning states, its beliefs. Put differently, the devices that produce desires and beliefs would have to have systematic ways of varying the desires they produce, and the beliefs they produce, so as to target variant goals and represent variant states of aff airs. For it is unrealistic to suppose that the belief-producing device, for example, could have been rigged up to produce noncomposite and unrelated brain states for each surrounding circumstance that it is the job of beliefs to represent; there are just too many such states – and so, mutatis mutandis, for the desire-producing device. In the case of either device, the only design that is simple enough for nature to have been able to hit upon it, while still being powerful enough to do the job we are supposing it to have been selected for, would be one that allows systematic variations and similarities among the beliefs and desires that are produced. To illustrate the point I am making here, let me offer some speculations as to specific ways in which the belief-producing device might be equipped – whether by natural selection or by individual learning – to induce systematic variations on the individual belief states that it produces. I will start with the assumption that the simplest sort of belief is supposed to do its action-guiding job by corresponding to the possession, by some object or substance, of some property. One such belief, let us say, is supposed to correspond to the berries’ being sweet. Then it is plausible to surmise that the belief-producing device has ways of varying the brain state that is this belief so as to produce 15

Millikan, White Queen Psychology, pp. 80–82, 89–91, 98–99.

98

Ontological preference for microphysical causes beliefs that attribute different properties – in the fi rst instance, contraries of the given property, and hence properties that contrast with sweetness to varying degrees (extreme sweetness, sourness, insipidness, etc.). It is also plausible that the belief-producing device deploys an axis of variation such that variants on the given belief will be supposed to represent how matters stand with members of different kinds in the same family of kinds – how matters stand, in this instance, with different kinds of edible items. Perhaps the beliefproducing device even deploys an axis of variation that marks differences in the kind of property, the instantiation of which the belief is supposed to correspond to – perhaps either surface-level observable properties, or dispositional properties and powers, or social properties. The important point, in any case, is that we have no reason to doubt that the belief-producing device is equipped to produce belief states that can diverge from one another on a number of different dimensions, where differences from a given belief (e.g. “the cherries are sweet”) as measured along any one of these axes are independent of differences as measured along the other axes. We have no reason to suspect, in other words, that the dimensions on which belief-states diverge from one another are “mutually interfering” dimensions of difference – in a sense that I shall introduce in the next section, and shall discuss at greater length in the appendix. For now, however, the important point is one that does not depend on any details about the specific ways in which the beliefproducing and desire-producing devices can induce variations on their products. For now, the important point is just that it is plausible to assume that, when the belief-producing device was functioning in the ways that won the favor of natural selection, there would have been variants on any actually produced belief, that would have led to variant attunements of behavior – provided, that is, we set the exogenously determined variables so that the variant beliefs would all have been relevant to one or another actually present desire. And it is plausible that, when the desire-producing device was functioning in the ways that won the favor of natural selection, systematic variants on a given desire would have led to actions that achieved variant goals – provided we set the exogenous variable for beliefs so that beliefs are present that could attune the variant actions in ways that lead to success.

99

Familiar objects and their shadows

5. 4

C au s a l e xc l u s i o n I : c o m p l e x e v e n t s at t h e l e v e l o f t h e m i c r o pa r t i c l e s

Suppose that whenever a particular belief gets tokened in a person’s neural hardware, a shadow event occurs at the level of the microparticles in the person’s brain. Billions of microparticles come to instantiate, between themselves, some massively complex structural property, describable only by some massive conjunctive predicate, each conjunct of which spotlights a property or interaction found in an individual microparticle or a small group of microparticles. The relation between the tokening of the belief, and the occurrence of this complex microparticle event, is not simply simultaneity and spatial coincidence, let us further suppose. Let us say that the tokening of the belief (call it b, for “belief ”) supervenes, at least weakly, on the complex microparticle event (call it c, for “cause” – though the very point of the argument here will be that we have no grounds for thinking that c really causes anything). Then the question opens up whether c might perhaps overdetermine any effect that we credit b with causing. Suppose b does indeed eventually cause some attunement in the host’s behavior: suppose that the state of aff airs that b represents does at some point become relevant to the way the host can go about pursuing one or another of his desires. This attunement will have a description at the level of muscle contractions and nerve impulses, and perhaps also at the level of inferences and practical planning (and let us call the event so described a, for “attunement”). But that attunement, so described, will itself be accompanied by another shadow event at the level of other microparticles. Let us further suppose that a supervenes on this other complex microparticle event (which we will call e, for “effect”). The question opened up is this. Does c cause e – and by causing e, does c thereby cause a? If so, the shaping of the host’s behavior which b brings about, namely a, will be causally overdetermined. From the viewpoint of this chapter, this question amounts to asking: is there a variable that subsumes c, and a variable that subsumes e, such that there is an invariance joining values of the one variable to values of the other? The contention that b causes a is thus far safe, at least until overdetermination rears its ugly head and gets transformed into causal exclusion. For as we noted in the previous section, we should expect that the design of the belief-producing device ensures 100

Ontological preference for microphysical causes that variants on b would – when all goes according to design – be accompanied by variants on a. Is there an invariance that links a similarly multivalued variable that subsumes c to a multivalued variable that subsumes e? Is there even an invariance linking a single -valued variable subsuming c to a single-valued variable subsuming e? In this section I argue that there is no clear case for answering Yes to either question. Before I begin the argument, let me point out that it will be crucial for us to consider c strictly as a complex event involving billions of microparticles, and not also as an event upon which b supervenes – and likewise to consider e strictly as a complex microparticle event, regardless of whether it subvenes a. For the suggestion that we will be trying out is that c causes e in a way that simply bypasses the apparent causation that runs from b to a. Only if this suggestion is vindicated can it emerge that c causes a in a way that renders b ’s apparent efficacy redundant – a way that therefore excludes b ’s apparent causal efficacy. Maintaining this focus is harder than it may appear to be, as I shall point out in the last paragraph of the following section, section 5.5. There is an invariance between two multivalued variables, to use the metaphor favored by Woodward and Hitchcock, just in case intervening on the value of the putative Vc (cause variable) would produce answering variations in the value of the putative Ve (effect variable). The precise function that takes values of Vc over into values of Ve can vary. It can be continuous, so that every small alteration in Vc produces an equally small alteration in Ve. It can be discontinuous, so that before any alteration occurs in Ve one must run through a number of small variations in Vc.16 It might even be nonlinear. But this much seems clear: if there is to be any function at all, by which alterations in Ve are tied to alterations in Vc, there must be a fact of the matter as to which variations in Vc are smaller and which are greater. More precisely, there must be an ordering of possible alterations in Vc, along which greater and greater alterations make for values of Ve that are correspondingly different from the value with which we began. The event c which, we are now suspecting, may causally overdetermine the attunement-to-circumstance (namely, a) of the host’s behavior consists in the instantiation of a massively complex structural property at the level of microparticles. Can we really think of a 16

Woodward , Making Things Happen , p. 66.

101

Familiar objects and their shadows variable that will subsume c, such that different values of this variable will amount to variants on c that line up in an ordering of greater and greater difference from c itself ? Let us fi rst address a prior question: do we even understand what it would be for complex microparticle events that are variants on c to differ from c to differing, but commensurable, degrees – to differ from c itself more and less? What makes this question seem formidable is that the property that c instantiates is so complex, being specifi able only by a conjunctive predicate involving thousands and thousands of conjuncts. It seems to follow that the complex microparticle properties, which variants on c would instantiate, could differ from the property which c instantiates along thousands and thousands of dimensions of difference. Now there can indeed be properties that are variants on one another, and that differ from one another on several different dimensions, for which there is nevertheless a clear-cut phenomenon of “more and less different.” Colors differ from one another in saturation, hue, and brightness, and yet the difference between any one color and another is comparable to – commensurable with – the difference between the fi rst color and a third (at least provided we are willing to regard the differences as distances within the Munsell color solid). But that is because the different dimensions on which colors differ from one another are mutually independent, and the relation between them is merely additive. In contrast, when one looks to the dimensions on which c differs from alternative complex microparticle events, one fi nds that the dimensions are not independent of one another. A particular distance on one dimension of difference, for two such complex microparticle events, can amount to a massive difference or a trivial difference between the overall events themselves, depending on the distance between the two events on other dimensions of difference. This claim will be illustrated in a moment. The upshot is that while any one such complex microparticle event may lie at different distances from each of two others, on any dimension that appears to permit a comparison of degrees of difference from the fi rst event, those different distances never in and of themselves amount to true cases of “more and less different” relating the overall events concerned. Indeed those different distances do not even make stable and fi xed contributions to a phenomenon of greater and lesser difference. So while there are countless (literally!) variants on c, which differ from c itself in countless 102

Ontological preference for microphysical causes different ways, there is in general no phenomenon of “more and less” that orders those differences. Our intuitions may, it is true, seem to suggest that we can think of some variants on c that differ from c itself far less than many, many other variants. We will return to this point shortly, and will see why we perhaps should not trust these intuitions. But even if we give evidential weight to these intuitions, and rule that there are some defi nite cases of “____ is less different from c than _____,” among complex microparticle event s that are variants on c, it will remain unbelievable that there is any general ordering of relative differences from c, such that any two randomly selected variants on c will occupy determinate positions in that ordering. Consider, then, nine possible variants on c. Event c1. It instantiates a structural property, the description of which differs from the description of c’s structural property with respect to 1,000 conjuncts belonging in the latter description. Event c2 . It instantiates a structural property which differs with respect to 5,000 conjuncts. Event c3. With respect to every conjunct where its description differs from c’s description, a microparticle bears a markedly different charge or energy-state from that of the microparticle described in the corresponding conjunct of c’s description. Event c4. With respect to every conjunct where its description differs from c’s description, a microparticle bears only a slightly different charge from that of the microparticle described in the corresponding conjunct of c’s description. Event c5. Its description requires only 1,100 conjuncts – that is, c5 is a less complex microparticle event than c – but differs at 1,000 of those conjuncts from c’s description. Event c6. Its description is the same, conjunct for conjunct, as c’s, and requires no more conjuncts than c’s description does, but its conjuncts are satisfied at regions far more spread out from one another, spatially, than are c’s conjuncts. Event c7. Its description incorporates every conjunct in c’s description, but requires 18,000 more conjuncts as well – c7 is a far more complex microparticle event than c. Event c8. Thirty-six conjuncts within its description depict the microparticle s which they describe as lying within a volume which common sense would suppose to be occupied by a dog.

103

Familiar objects and their shadows Event c9. Three hundred conjuncts depict the microparticles which they describe as lying within a volume which common sense would suppose to be occupied by a pond.

Event c1 seems, at fi rst blush, to differ from c itself less than c2 does. (One thousand differences are fewer than 5,000.) But if c1 also differs from c in the way that c5 differs from c, c1 may be more different from c than c2 is. (For then almost all of c1 will be different from c.) This will especially be true if c1 further differs from c in the way that c3 does. (Almost all of c1 will differ a lot from c.) But if c2 likewise differs from c in the way that c3 does, c1 may after all be less different from c than c2 is. (Five thousand big differences are worse than 1,000 big differences.) Of course, that will all depend on whether c1 further differs from c in the way that c6 does. (If c6 is spread out across two galaxies, it is going to be very different from c.) And what about c7 – is it more different from c than c2 is, if c2 itself differs from c in the way that c3 does? (If c7 is massively complex, is it more different from c than a variant that is lots different from c at 5,000 places?) The point that these questions illustrate is that there is systematic interference among the various dimensions along which variants on c can differ from c itself. The dimensions do not relate additively to one another. The consequence is that relative distances from c, along any dimension that affords an ordering of “more and less different” for complex microparticle events that are variants on c, will not in general amount, in and of themselves, to degrees of difference for those complex microparticle events themselves from c itself. But then in general, for pairs of variants on c that differ differently from c on any one dimension of difference – that can, in respect of that dimension, be said to be “more and less” different from c itself – there will simply be no fact of the matter as to which member of that pair is as a whole more different from c, and which less. But then even if we could think of a multivalued variable that subsumes c and all its complex microphysical variants, there would be no prospect of claiming that values of that variable that are more and more different from the value found at c bring about variants on a that are correspondingly different from a itself. In the relevant sense, there is no such phenomenon as “more and more” different from c – no neat ordering of the microphysical variants on c in terms of relative difference from c

104

Ontological preference for microphysical causes itself. So there is no prospect of an invariance that links a multivalued variable subsuming c to a multivalued variable subsuming a – or even to a multivalued variable subsuming e. But if it has no clear sense to ask how matters would have gone if complex microevents more and more diff erent from c had occurred in c ’s place, it still might make perfectly good sense to ask how matters would have gone if c had not occurred at all. Philosophers inclined to think that microparticles found where b occurs between them cause e, and thus overdetermine a, might therefore suppose that there is an invariance between a two-valued variable C that subsumes c – the values of C being {c occurs, c does not occur} – and a two-valued variable E (namely {e occurs, e does not occur}) that subsumes e. So let us ask: what really would have occurred, if c had not occurred? Most philosophers will agree that to ask this question is to ask about the closest possible worlds to the world in which c does occur, i.e. closest to the actual world. These will be worlds in which some complex microparticle event very much like c, and differing from c just slightly, occurs in place of c itself. Our intuitions may, as we noted several paragraphs above (p. 103), suggest that we can fairly readily identify at least some such variants on c. The recipe we will follow, if we heed these intuitions, will be to envision a variant event that is “ just like c on every dimension but one,” and that involves some minimal divergence from c on the remaining dimension. Thus we might suppose that a variant that is “otherwise just like c ” except that one of the microparticles, spotlighted by one of the conjuncts in the specification of c, is placed at a slightly greater spatial remove from its nearby neighbors than c depicts it as being. But this overall variant event may be less like c, in respect of its causal powers, than a variant that differs from c on two dimensions – a variant that features stronger charges or higher energy states obtaining in that spatially removed microparticle and in its neighbors. In general, as we have seen, there are countless ways in which a small divergence from c along one dimension of difference can be made to amount to no overall difference from c, or to sizeable overall difference from c, by divergences (or samenesses) that obtain along other dimensions. We must be very hesitant, therefore, to claim that we can identify any variants on c that differ “only minimally” from c itself. But even if there objectively are such variants, we must remind ourselves that

105

Familiar objects and their shadows if there are any, there are many. For possible variants on c diverge from c on thousands and thousands of different dimensions of difference. Even if we trust our intuitions to have identified three or four candidates for the title “minimally different variant on c,” there will be thousands and thousands of candidates that we have not thought of. And then the crucial question is this: if we take c together with these thousands and thousands of other variant events, is it plausible that c alone was causally suffi cient to produce e – in other words, that c itself was not just causally sufficient for e, but causally necessary? If c had not occurred, are we really sure that e would not have gotten produced anyway? It seems out of the question that we should really have warrant for answering Yes to these questions. But then we have no warrant for believing in a variance linking even singlevalued variables that subsume c and e. We have no reason for thinking, therefore, that c causes e. Consequently, we have no reason for thinking that c causes a. The claim of b to have caused a remains unrivalled. A fi nal note on the argument of this section. One might worry that the argument “proves too much.” For if the phenomenon of “more and less different from” is ill-defi ned for the variables that might subsume the microparticles involved in c, it is ill-defi ned in many other cases as well. The complex microphysical event c attracted our attention because, we supposed, it subvened the occurrence of mental state b. But similarly complex microphysical events appear to subvene all manner of macrolevel states involving macroscopic objects: such events appear to subvene a baseball’s flying towards a window, for example, or a car horn’s blowing loudly. Thus the argument of this section may seem, if sound, to show that there is no real worry about causal exclusion for even such events as these, involving nonconscious macroscopic objects. To this misgiving I plead guilty as charged. Indeed my aim in this chapter, as I indicated at the outset, is to provide reason for thinking that all causal exclusion arguments, directed at the apparent efficacy of familiar objects, fail. I focus here on the causation of behavioral outcomes, which people apparently exercise by virtue of their beliefs and desires, merely because this is reckoned to be the most eminently excludable instance of causation by a familiar object. I intend the argument to be general. In the next chapter, I shall show how basically the same argument defends the causal efficacy of a dog’s bark. 106

Ontological preference for microphysical causes

5.5

C au s a l e xc l u s i o n I I : c o m p l e x e v e n t s at t h e l e v e l o f n e u r o c h e m i s t r y

Whenever a particular belief or a particular desire gets tokened in a person’s neural hardware, an event occurs at the level of the neurons in the person’s brain. Certain axons and dendrites, say, come to be more ready or less ready to fi re, or a change occurs in the distribution and movement of neurotransmitter molecules within those axons and dendrites. The threat which so far we have been considering is that vastly complex microphysical events in the cranium might exclude the causal efficacy of beliefs and desires. The new threat is that complex events at the fi ne-grained level of individual neurochemical molecules and parts of neurons might likewise exclude the causal efficacy of beliefs and desires. But let me be clear. Beliefs and desires, on the view I am putting forth, really are brain events. But more particularly, they are brain events of the sort that the neuroanatomist recognizes. The new worry is that their apparent efficacy is excluded by brain events of a different sort, the brain events that the neurochemist studies. Here is how the neuroanatomist – as I am using the term – views the brain. The brain is populated by devices and systems that were fashioned by natural selection. One example might be the so-called “language organ.” Other examples, I say, are the desire-tokening and belief-tokening devices. These have historically defi ned proper functions. They are designed to produce individual beliefs and desires, brain states that inherit, from that same history, proper functions of their own. Any one such state may get realized in separate, even disparate, neurons and subneuronal elements. It may continue to get realized across molecule-level changes in its realizers. The neurochemist, in contrast, focuses on molecules, and interactions among molecules, that are salient strictly because of the biochemical properties that they now possess and produce. The neurochemist’s judgements as to how long “the same” neuron-level event is going on, or over how wide an area it occurs, will not trade on suppositions about how natural selection has structured the brain. That is, even if it is true that neurochemically different configurations, occurring at different times and places in the brain, all realize a common stage in some designed way of functioning, the neurochemist will nevertheless rule that these are different neuron-level 107

Familiar objects and their shadows events. Thus even if it is true that natural selection has fashioned a “language organ,” that “language organ” is likely to appear, from the standpoint of the neurochemist, as a disparate plurality of separate neurons, neuron-parts, and neurotransmitter molecules. Indeed it may appear as a shifting disparate plurality, one that incorporates numerically different neuronal and subneuronal elements on different episodes of its operations. Just as with language, so too – we must be prepared to discover – with beliefs and desires. From the standpoint of the neuroanatomist, it may well turn out that there resides within the brain something that can properly be called “the belief-tokening module” (or “the desire-tokening module”). Even if so, we must be prepared to learn that from the standpoint of the neurochemist, this unitary “module” is a hodgepodge of disparate and spatially separated neurons, neuron-parts, and neurotransmitter molecules. Indeed it may be a shifting hodgepodge, incorporating numerically different parts on different occasions. From the standpoint of the neuroanatomist, the tokening of an individual belief in this region will count as a single (if not simple) event. But from the standpoint of the neurochemist, we must be prepared to learn, the tokening of a belief will count as a plurality of independent and diverse events, which may not even occur at the same time as one another, and which will assuredly occur in different subjects of change – in different, indeed scattered and diverse, neuron-parts and neurotransmitter molecules and neurons. So here is the remaining worry about overdetermination. If a belief that is present across some expanse in the brain comes to cause some attunement in the host’s behavior, can it equally be said that what the neurochemist sees in that same expanse causes that same attunement? What the neurochemist sees is a disparate plurality of scattered, perhaps nonsimultaneous, small neurochemical states and events involving separate neurons, neuron-parts, and neurotransmitter molecules. Can such a plurality of scattered small events be credited with causing a single outcome – namely the same behavioral attunement as we suppose the belief to cause? To address this question, we must face a more general one. This is the question of when, whether, and why it may be legitimate to think of many events as collectively causing a single outcome. Many philosophers speak as if this question has an obvious answer. They 108

Ontological preference for microphysical causes speak as if it is obvious that if each element of such a “many” causes a single small outcome, it automatically follows that the “many” itself causes the collection of those small outcomes. But on the face of it, this simple inference seems merely to be a case of “the fallacy of composition.”17 In any case, the idea that any true causing must instantiate some explanatory invariance entails that this simple inference is too quick. That is what the argument of the preceding section in effect showed. It showed that to defend the claim that many small events, each involving a separate small object, between them cause some single outcome, we must defend the claim that the instantiation of a single complex structural property, by that plurality of small objects, is linked by an explanatory invariance to that single large outcome (or to something single and large upon which that outcome supervenes). Just so here. The threat of overdetermination stands only if the neurochemist can argue that what happens from his perspective, when (as we are supposing) an individual belief causes an attunement of behavior, falls under an explanatory invariance. The neurochemist must argue that the many neurons, neuron-parts, and neurotransmitter molecules between them instantiate some complex structural property, such that the right sort of invariance holds – such that variants on that structural property, more and more different from the actually instantiated complex property, would be accompanied by behavioral attunements correspondingly different from the attunement actually produced (or would be accompanied by large-scale events on which such variant attunements would supervene). I contend that “more and more different from” is not defi ned over the fi ne-grained events that would be variants on the actually occurring event. That is, if we view the neurotransmitter molecules, neuron-parts, and neurons on which the putatively efficacious belief supervenes in the abstract – putting out of our mind the thought that they subvene this belief – the contrasting structural properties, which would qualify as variants on the structural property actually instantiated, seem to contrast along many different dimensions. Suppose, for example, that the structural property actually instantiated is given by the following description.

17

Elder, “Physicalism and the Fallacy of Composition.”

109

Familiar objects and their shadows Incorporates a neurotransmitter of type α at location l1 moving across a synapse at location l 2; and incorporates an axon of type β at location l 3 spiking at such and such a frequency; and incorporates a pyramidal cell at location l4 rigged to attune its outputs to six other pyramidal cells, at locations …,” and so on, through twenty-three conjuncts.

Then these are some variant structural properties to the one that is actually instantiated. Variant Structural Property 1. Its description incorporates twenty-two conjuncts that are the same as those in the actually instantiated property, and seventy more conjuncts, of which this is a typical example: “… and incorporates a third baseman who is batting .267.” Variant Structural Property 2. Its description incorporates the same conjuncts as the fi rst eighteen in the actually instantiated property, but no others. Variant Structural Property 3. Its description incorporates pretty much the same conjuncts as the original twenty-three, and no others, but specifies that each conjunct is satisfied at a location 500 miles or greater from any location at which another conjunct is satisfied.

Which of these is more like the structural property actually instantiated, and which less? I contend that the question has no answer – that there is simply no “fact of the matter.” Above (pp. 103–4) I used nine variant properties to make a parallel point. Since it is now clear how the current list can be extended, I trust that merely listing four variant properties is sufficient to make the point. “But this is unfair!,” I expect the neurochemist to reply. “These are ridiculous, made-up structural properties – they aren’t really relevant alternatives to the structural property that is instantiated in the brain, from my point of view, where the actual belief is present!” Precisely. But ask why these are not relevant alternatives. The answer must be: “these aren’t alternative structural properties which the many neurons, neuron-parts, and neurotransmitter molecules might between them instantiate, while that same brain is still doing its usual business!” But the only explication of “doing its usual business” that will help the neurochemist here is this: “while still composing into beliefs that are variants on the one that actually causes the actual attunement of behavior.” That is, the neurochemist has to impose 110

Ontological preference for microphysical causes some discipline on which dimensions of difference, from the actually instantiated structural property, will count as relevant. And the only way to impose this discipline is to keep constant the assumption that some belief or other is present, that is efficacious in causing the actual behavioral attunement or some variant on it. In sum: the neurochemist may, in a way, be able to assemble a competing cause of the behavioral outcome, at the level of the small objects he recognizes, that will be sufficient to cause the behavioral attunement – but it will qualify as sufficient to cause this behavioral outcome only by composing into the belief which, we are supposing, causes that behavioral attunement. If this is causal overdetermination, it is entirely nonworrying causal overdetermination. (For a closely similar form of nonworrying causal overdetermination, see claim “(5)” in Merricks.18) A parallel point applies to the argument of the previous section, the section that considered microphysical overdetermination of a given shaping of behavior. That argument asked whether a genuine invariance might link c, the enormously complex event involving billions of microparticles, with an equally complex microphysical event e that subvened the given shaping of behavior. That argument contended that variants on c diverge from c on so many different, sprawling dimensions, that the very phenomenon of “more and less different from c” was undefi ned. But here friends of the microparticles might be tempted to make much the same sort of protest as I have attributed to friends of the neurochemist’s standpoint. “These so-called variants on c,” friends of the microparticles might protest, “feature ridiculous, made-up structural properties! Surely the relevant alternatives to c are ones that involve just as many microparticles as c itself does, not more or fewer, and involve only microparticles, and only microparticles that are just as close to one another in space and time as are the microparticles in c itself!” But ask why these might be the only relevant alternatives – just what the rationale would be for confi ning divergence from c to just these dimensions. The answer will have to be: “because only so will we be looking at microphysical events, involving billions of microparticles, that might subvene the occurrence of a mental event!”19 But then we will have been treating c as 18 19

Merricks, Objects and Persons, pp. 100–1. Cf. Sturgeon, “Physicalism and Overdetermination,” §6.

111

Familiar objects and their shadows essentially a subvener of b. Yet the whole point was to ask whether c might cause e (and thereby a) in a way that simply bypasses the apparent causation that runs from b to a – in a way that has nothing to do with c’s subvening b. If we save the claim that an invariance really does tie c to e, but does so only insofar as c subvenes the mental event b, we will not have attributed to c a causal efficacy that bypasses b ’s claim to causal efficacy. 5.6

T h e p r o b l e m o f m e n ta l quau s at i o n

Now a brief comment on “the problem of mental quausation.” When individual beliefs perform in the way they did perform, on the occasions when the device that produces them was winning the favor of natural selection, they attune the host’s pursuit of its desires to the presence in the world of particular states of aff airs. That is, an individual belief – when operating in the way that got the beliefmaking device to be selected for – initiates, steers, and modulates the host’s enactment of one desire or another, in such a way that some environing state of aff airs gets enlisted in making the host’s behavior, thus steered, turn out to be successful. More briefly: the individual belief causes the host’s goal-pursuing behavior to be caused to be successful by some particular environing state of aff airs. Now what Ruth Millikan has argued is that this two-step causal connection to an environing state of aff airs is what it is for the individual belief to have that state of aff airs as its semantic value, its content.20 If that is correct, a belief ’s having the content that it does is a function of its being such as to cause – when all proceeds according to design – the behavioral outcome that it does cause. Having content presupposes efficacy in shaping behavior. It is therefore incoherent to ask whether or not it is in virtue of having its content, that an individual belief has efficacy in shaping behavior. When individual desires operate in the way they did perform, on the occasions when the device that produces them was winning the favor of natural selection, they target the host’s behaviors towards goals which actually get attained, thanks to steering of those behaviors by the host’s beliefs. That is, an individual desire – when operating

20

Millikan, White Queen Psychology, Chs. 3 and 4.

112

Ontological preference for microphysical causes in the way that got the desire-producing device to be selected for – causes the attainment of some goal. Millikan 21 has argued that this belief-steered causing of a particular outcome is what it is for the desire to have that outcome as its content – to be a desire that that outcome obtain. If this is correct, then “the problem of mental quausation” is as confused in the case of desires as it is in the case of beliefs. A desire’s having the content it does is its being such as to cause – when all proceeds according to design – a particular outcome. Content presupposes causal efficacy in shaping behavior. One cannot coherently ask whether it underlies or yields causal efficacy in shaping behavior. 21

Millikan, White Queen Psychology, Chs. 3 and 4.

113

6 Ontological preference for the spatially small Here is a line of reflection that has dominated much recent metaphysics.1 Science gives us good reason to suppose that the familiar objects that common sense believes in are, if real at all, composed wholly of tiny microparticles – either of the microparticles that current microphysics regards as fundamental, or of other microparticles, perhaps smaller or stranger, that a perfected future microphysics will recognize. Yet as philosophers we must entertain seriously the possibility that some of these familiar objects do not really exist. Where common sense supposes there to be a tree or a baseball or a dog, we must be willing to say, there may just be many, many microparticles. 1

This line of reflection begins with van Inwagen’s powerfully argued Material Beings and extends through Merricks’s Objects and Persons and Rosen and Dorr’s “Composition as a Fiction”; it also makes an appearance in Rea’s World without Design. But within these works, the explanatory success of contemporary physics is not the only motive for taking seriously the idea that where common sense recognizes a dog or a tree or a baseball, there may really be just many very small things arranged in a distinctive fashion. A separate motive is provided by a priori arguments to the effect that the world may be populated, widely or throughout, by items so small that they have no proper parts. Whether motivated by science or by a priori arguments, these philosophers say that where common sense recognizes a dog or a tree, we must take seriously the suggestion that there may just be many simples. Van Inwagen indicates that the “simples” he means to discuss are for the most part the fundamental particles, the physical simples, that physical theory posits: see Material Beings, pp. 99 and 158. Merricks, whose term for the many small things is “atoms,” likewise explains that his discussions about “atoms” are discussions about “whatever microscopic entities are actually down there” (Merricks, Objects and Persons, p. 3). Thus van Inwagen and Merricks both appear to enlist the success of physics to give weight to the speculation that where common sense discerns a single dog or baseball, there may really just be many small things. But matters may be different with Rosen and Dorr (“Composition as a Fiction”). Rosen and Dorr contend that where common sense discerns a single dog or baseball, there defi nitely are only many little things: there are no instances of composition, on their view. But though Rosen and

114

Ontological preference for the spatially small It still would be possible to explain why common sense supposes, and is rewarded for supposing, that in just these regions there is a tree or a baseball or a dog. The explanation would be that the microparticles in these regions are arranged towards one another in particular ways, ways that reward the common-sense belief in a single object of a particular kind. Thus where common sense supposes that there is a dog, there might really be just microparticles dogwise arranged. We need not pause long to consider what dogwise (or treewise or baseballwise) arrangement consists in or amounts to.2 For, since it is certain

2

Dorr cite van Inwagen as having given plausibility to the idea that composition occurs far less widely than common sense supposes (“Composition as a Fiction,” p. 152), it is not clear whether their belief in “simples” is motivated by anything more than a priori mereology. Rea’s position is likewise not entirely clear. He sometimes indicates that the suggestion we must take seriously is that where common sense recognizes a tree, there may really just be matter arranged treewise (Rea, World without Design , pp. 97ff.); at other times, he says that we must take seriously the idea that where common sense discerns a tree, or where science discerns an H 2O molecule, there may only be “a collection of subatomic particles” (Rea, World without Design , p. 132). Van Inwagen offers a brief discussion (Material Beings, p. 105) of what really obtains in some region of space, when the folk would judge that the contents of that region compose a chair, and this discussion provides as detailed a recipe as one can fi nd in the literature for determining what chairwise arrangement would amount to – or what treewise or baseballwise or dogwise arrangement would amount to. Merricks’s account of what it is for fundamental particles to be statuewise (or baseballwise, etc.) arranged is particularly exiguous (Merricks, Objects and Persons, pp. 4–7). Merricks says that this is for the particles to be so characterized, and so arranged towards one another, that between them they would compose a statue – would, if it weren’t for the fact that it is impossible that there should be any statues. This is impossible, on Merricks’s view, because for any macroscopic object, to be is to have causal powers (Objects and Persons, p. 81), and yet anything a statue (or baseball) may seem to cause can and should be said to be caused by the fundamental particles within the statue instead (Objects and Persons, Ch. 3). Merricks thus speaks as if, if only we would abstract from worries about causal exclusion, we could fairly readily give a nonvacuous answer to van Inwagen’s “special composition question” both in the case of statues and in that of baseballs (van Inwagen , Material Beings, Ch. 2) – that is, we could identify the conditions under which composes a statue and composes a baseball occur. Van Inwagen, for his part, would say that in these cases the “special composition question” has no nonvacuous answer – that is, the right answer is that under no conditions do composes a statue and composes a baseball occur. If it really is as easy as Merricks seems to think it is to say what the conditions are under which fundamental particles would compose a statue or a baseball, perhaps we should take this as casting doubt on the premise that statues and baseballs cannot really exist unless they have causal powers of their own.

115

Familiar objects and their shadows that the world rewards the common-sense belief in dogs (and trees and baseballs), it is certain that that arrangement, whatever exactly it is, obtains among the microparticles in various regions. What it is important to consider is what more it would take for microparticles dogwise (etc.) arranged to compose a dog (etc.). The general question of composition, of what it is for many microparticles to compose an object, is the key to determining how many (if any) of the familiar objects that common sense believes in are really there. In this chapter I argue that this reflection has set the wrong philosophical agenda. The phenomena of treewise and baseballwise and dogwise arrangement are by no means to be taken lightly, I shall argue: there are very serious questions as to where these arrangements obtain, and what their bearers are. Put neutrally, dogwise arrangement is that by virtue of which the contents of some region of the world look and act like a dog. That is, it is that by virtue of which the contents of some region cause (or would cause) in conscious subjects certain sensory experiences, and cause (or would cause) in surrounding regions the sorts of observable effects that we expect dogs to cause. When dogwise arrangement (or treewise or baseballwise arrangement) is construed in this way as constituted by a causal power, it is indeed uncontroversial that the contents of some regions of the actual world possess it. But it is important to leave open, pending argument, the question of which contents of those regions bear that power, are “dogwise arranged.” If it should turn out that only dogs themselves can cause in observers doggish experiences, and in their surroundings dog-like observable effects, we shall have to say that it is dogs that are dogwise arranged. We must be prepared to say this even if it is more customary to speak only of a plurality of entities as being “arranged” one way or another: that is the cost of keeping the starting idea, that dogwise arrangement really obtains in some regions of the world, uncontroversial and neutral in its implications for whether there really are in the world dogs. If on the other hand it should turn out that many, many microparticles may, between them, bear this causal power, then we will say that it is these microparticles that are dogwise arranged. The latter is the position that animates the dominant metaphysical agenda, and I shall begin by allowing, for the sake of argument, that it may be right. But if it is right, I shall argue, then it is just not plausible that the many microparticles, that do bear any one instance of dogwise arrangement, are located just where 116

Ontological preference for the spatially small common sense supposes there is a dog. At any given time, on any given occasion, the microparticles that are dogwise arranged will be neither only nor all the microparticles that lie within the borders that common sense attributes to some dog. This is so, that is, if we think of the phenomenon of dogwise arrangement as one that obtains independently of whether there really are in the world dogs. That is what the dominant metaphysical agenda thinks: it thinks that there is a fact of the matter as to where dogwise arrangement obtains, even while it is an open question whether in any of these regions there are dogs. But suppose one were to allow that there really are in the world dogs. Then indeed one could construe dogwise arrangement, I shall say, as something that characterizes only the contents of those regions of the world occupied by dogs. But then what would the bearers of this arrangement be? The many microparticles within any such region; the dog within any such region (since now we have allowed that there is a dog there); or both the many microparticles and the dog? The answer “both” will seem to many to be out of the question: dogwise arrangement is that in virtue of which the contents of some region cause certain effects, and many will judge that we cannot allow, on pain of intolerable overdetermination, that both the dog and the dog’s component microparticles cause the same effects (see the parallel argument about the baseball and the baseball’s component microparticles in Merricks 3). The bearer of dogwise arrangement, many will infer, must be either the dog himself or the dog’s component microparticles. And there is a long line of thought, beginning with Davidson’s paper “Mental Events” and continuing through Merricks’s Objects and Persons, that says that in any such competition, the microparticles’ claim of causal efficacy trumps the dog’s. This would be an ugly result. We would have circumscribed the right microparticles to which to attribute dogwise arrangement by affi rming the reality of dogs, but then would have violated Alexander’s dictum4 by declaring the dogs causally inefficacious. But the true result is far prettier, I argue. Davidson’s anomalous monism was founded on the premise that any true causing must instance some exceptionless law of nature. But Woodward and 3

4

Merricks, Objects and Persons, Ch. 3; but cf. Sider, “What’s So Bad about Overdetermination?” Kim, Supervenience and Mind, p. 348.

117

Familiar objects and their shadows Hitchcock 5 have shown, as I argued in the previous chapter, that a far more defensible position on token causation is that it must be underwritten by invariances in nature. An invariance (to repeat) is typically a correlation, to which there can be exceptions, between a multivalued variable that subsumes the putative cause and a multivalued variable that subsumes the putative effect. This position can readily be used to show that events involving a dog can figure as causes of doggish sensory experiences in observers and as causes of gnaw-marks on bones or holes beneath fences. With some extra considerations not offered by Woodward and Hitchcock, I shall argue, their position can be used to show that no complex state of aff airs, involving the many microparticles within the dog, can figure as the cause of any of these effects. The dog himself emerges, then, as the uncontested bearer of dogwise arrangement. The existing metaphysical agenda treats as obvious and uninteresting the claim that, in various regions of the world, there are such phenomena as treewise arrangement, dogwise arrangement, and baseballwise arrangement. It treats the question whether the contents of these regions actually compose trees and dogs and baseballs as deep and nearly intractable – or, on some variants, as (surprisingly) merely verbal.6 This chapter argues that it is not at all obvious, yet it is the consequence of a sound theory of causation, that there are, in these same regions of the world, treewise and dogwise and baseballwise arrangement. In the process this chapter defends the reality of trees, dogs, and baseballs. What it treats as uninteresting is the question of composition. What it is for microparticles to compose a dog, on the position of this chapter, is just for them to be all and only the microparticles found within the borders of a dog. 6.1 J u s t w h e r e d o e s “ d o g w i s e a r r a n g e m e n t ” o b ta i n ? The hypothesis up for consideration is that just where the folk suppose that there is in the world a dog, or a baseball or a tree, there 5

6

Woodward and Hitchcock , “Explanatory Generalizations, Part I: A Counterfactual Account,” and “Explanatory Generalizations, Part II: Plumbing Explanatory Depth.” Hirsch, “Physical-Object Ontology, Verbal Disputes, and Common Sense”; cf. Rosen and Dorr, “Composition as a Fiction,” pp. 166–68.

118

Ontological preference for the spatially small are really just many, many microparticles. It would nevertheless be unsurprising and unremarkable, says the line of reflection rehearsed above, that the folk suppose that a unitary dog (or baseball or tree) is there, and that that supposition provides predictive and practical benefits. For (i) in any such region there would be an instance of dogwise arrangement, or of baseballwise arrangement, or of treewise arrangement. And (ii) the bearer of this arrangement – what stands in dogwise arrangement, or baseballwise or treewise – would be many, many microparticles. In this section I argue that there is a tension between claim (i) and claim (ii). If dogwise arrangement is to explain what it is supposed to, and if its bearer is many microparticles, those microparticles will not, it seems, be confi ned to the region where common sense supposes there is a dog, nor will they occupy all of that region. But there is one suggestion as to what dogwise arrangement consists in that would stop this objection before it gets under way, and it is best to begin by considering it. This is the “fictionalist” suggestion.7 It claims that what it is for many microparticles to be dogwise arranged just is for them jointly to occupy a region in which “folk ontology” supposes there to be a dog – or in which adherents of “folk ontology” would suppose there to be a dog, were any of those adherents present. We must so far remain neutral, the thinking would go, as to whether folk ontology would rightly suppose there is a dog there. Whether the dog is real or a mere fiction, the mere fact that adherents of folk ontology would believe a dog exists in a region amounts to the microparticles in that region being dogwise arranged. But this position on what dogwise arrangement consists in confl icts with the role that dogwise arrangement (or baseballwise or treewise arrangement) is supposed to play in the reflection that sets up the dominant agenda. That the contents of a given region are dogwise arranged is supposed to explain why the folk suppose that in that 7

Merricks, Objects and Persons, pp. 4–7; cf. van Inwagen , Material Beings, p. 105. Van Inwagen , in Material Beings, in effect provides a statement of what it is for simples to be chairwise arranged (see note 2 of this chapter), and part of what is involved, according to his statement, is that the simples jointly occupy a region (called “receptacle R”) in which the folk suppose a chair to be present. So the problem identified in the text occurs here too. If the fact that simples in a region are chairwise arranged is supposed causally to explain why the folk suppose there to be in that region a chair, chairwise arrangement cannot consist (even partly) in the folk’s making that supposition.

119

Familiar objects and their shadows region there is a dog, and the explanation is supposed to be causal: in virtue of being dogwise arranged, the contents of such a region are supposed to be such as to cause in folk observers doggish sensory experiences, and to cause in surrounding regions events that will look to the folk like the sorts of effects we expect to see dogs produce. So that the contents of a region are dogwise arranged must be a state of affairs distinct from the fact that the folk would suppose that there exists in that region a dog. The fictionalist position makes dogwise arrangement be just equivalent to the belief that it is supposed causally to explain. Suppose then that dogwise arrangement is causally efficacious, and assuredly not the figment of a fiction. If the bearer of any instance of this arrangement is many microparticles, the microparticles in question will, it seems, be all and only those that between them are such as to cause in folk observers (should folk observers be present) the judgement that a dog is present. Suppose, to begin with, that I make such a judgement on the basis of (what I call) “seeing a dog.” Just which microparticles participate in causing in me this experience? For me to have this experience, there must arrive at (what I call) my eyes a structured array of light that reflects off (what I call) the dog’s surface. Innumerable microparticles in the dog’s surface – at least, in that portion of the dog’s surface visible from where I am standing – must participate in reflecting photons in the ambient light in my direction. (For ease of expression I shall now drop the locutions that indicate we must remain neutral so far on whether there are dogs, sides of dogs, etc.) But many other microparticles between the dog and me also participate in causing the arrival of the structured array of light. Microparticles in the molecules of air must between them act in such a way as not to disperse the structured array, as a thick fog would do, or to occlude it, as a solid screen would do. Apparently, then, all these latter microparticles are included in those that are dogwise arranged, every bit as much as are those on the near side of the dog’s surface. Moreover, innumerable microparticles in the dog’s interior, and on his far side, are not included in those that bear this instance of dogwise arrangement. Suppose, secondly, that I judge that a dog is present on the basis of (what I call) “hearing a dog barking just around the corner of the house.” Involved in producing this experience are microparticles in the dog’s voice box and lungs and diaphragm, but not on the dog’s 120

Ontological preference for the spatially small exterior; microparticles in the molecules of air that propagate the sound waves emanating from the dog; microparticles in the surfaces of the trees and the shed, off which those sound waves are reflected. This second instance of dogwise arrangement apparently spreads very far indeed! But we need not confi ne ourselves to experiences of observing a dog within hearing or within sight. That the contents of some region are dogwise arranged is supposed causally to explain not only that I have doggish sensory experiences, but also that I have experiences of seeing (what I call) the sorts of effects one expects to see a dog produce. So suppose, third, that I have the experience of seeing a freshly dug hole at the base of my fence. I may very well judge that my dog must recently have been digging there at the fence. What microparticles participate in producing in me this experience? The case is one in which there arrives at my sense organs a very slow echo of causal chains initiated by microparticles in my dog, but not an auditory echo, rather a visual one. Microparticles in my dog’s paws and legs and muscles participated in causing the echo and hence the judgement it produced, but so too did microparticles in the soil and in the air. Most of the microparticles involved lie far outside the region in which common sense supposes there to be a dog. No matter; if what it is for microparticles to be dogwise arranged just is for them to be, between them, such as to cause in me a judgement that a dog is present, the microparticles that bear this instance of dogwise arrangement will be found in the soil and air as much as in the dog. In fact, on the present understanding of what dogwise arrangement is, the microparticles arranged in a single instance of it may not just extend beyond the borders of a dog, and occupy only part of the dog’s volume; they may even be found in the wrong places altogether. One vocalization of the barred owl sounds just like the bark of a dog. Suppose that the echo that led me to judge that a dog was just around the corner of my house emanated instead from an owl. Or suppose that the hole was really dug by a fox. Then the microparticles arranged in some instances of dogwise arrangement might include none that are located in a dog. It may seem that I am dragging out this objection to great and uncharitable lengths. Perhaps the more charitable way of construing the idea that there surely are in the world the phenomena of dogwise and baseballwise and treewise arrangement, even though it is far from 121

Familiar objects and their shadows sure that there are any objects in the world larger than microparticles, is to say that for microparticles to be dogwise arranged, more is required than just that those microparticles between them be such as to trigger and to reward the common-sense judgement that a dog is present (should adherents of common sense be on the scene). Perhaps microparticles must be among those that (between them) trigger and reward these judgements, to qualify as being dogwise arranged, but must meet some further requirement as well – a further requirement that will after all keep the microparticles involved in any one instance of dogwise arrangement within a region where common sense espies a dog. One such further requirement is suggested by comments of van Inwagen’s that bear on what it is for particles to be chairwise arranged: perhaps for microparticles to be chairwise or dogwise or other-object-wise arranged, they must between them form a maximal, cohesive, separately movable mass of matter.8 (The forces binding such particles, van Inwagen says, will be greater than those that would be brought to bear upon them by typical human efforts to move them.9) This further requirement would entail that microparticles in the air, through which the structured array of light is conveyed from the dog to my eyes, do not count among those that are dogwise arranged. For these microparticles do not figure in a cohesive, separately movable chunk of matter; a volume of air does not move as one piece. But will the extra requirement help ensure that the microparticles in the 8

9

That the matter in some region forms a maximal separately movable parcel is something that commonly is both necessary and sufficient for us to judge that a unitary object occupies that region. This is noted both by Hirsch (Concept of Identity, pp. 105–12), who suspects that there may be no mind-independent fact of the matter as to how far unitary objects such as chairs and trees spread across space and time, and by Grandy (“Artifacts”), who holds that such familiar objects have mind-independent borders and careers. This is why, when van Inwagen (in Material Beings) speaks of the forces that bind the particles that are chairwise arranged as being stronger than the forces that humans bring to bear in “moving a chair,” I take him to be invoking the idea of a maximal separately movable object. For that would be a good idea for him to invoke. Physical simples that are (familiar object)-wise arranged are such as, between them, to look like they form a unitary familiar object, and Hirsch and Grandy show that a standard way of looking that way is looking like a maximal separately movable portion of matter. See also Spelke, “Principles of Object Perception,” and “Initial Knowledge.” Van Inwagen, Material Beings, p. 105.

122

Ontological preference for the spatially small dog’s surface, that participate in launching the structured array of light, do qualify as dogwise arranged? They too do not form a maximal, separately movable mass of matter; the only way to move them is to move the rest of the dog along with them. Perhaps the next step in the direction of charity would be to loosen the extra requirement. Perhaps we should say that microparticles that are dogwise arranged (i) must participate in causing and rewarding judgements that a dog is present, and (ii) must be among a plurality of microparticles that form a maximal, cohesive, separately movable mass. But problems remain. One is that the microparticles on the surfaces of the trees and of the shed, that participate in bouncing the sound waves from the barking dog in my direction, will now qualify as themselves being dogwise arranged; in fact, even microparticles within the owl and within the fox will now count as dogwise arranged. The far more serious problem is the question of what right we have to be speaking of cohesive, separately movable masses of matter at all. At the present stage, where we are affi rming only that there are in the world the phenomena of dogwise and chairwise arrangement, we are supposed to be treating as an unsettled question what it would take for many microparticles to compose a single object, and even whether composition ever occurs. For all we affi rm at present, there are in the world only microparticles. Yet a cohesive, separately movable mass of matter would not be a microparticle; if real at all, it would be composed of microparticles. Now there is indeed one form of composition which some would say we can take for granted even at this stage in the dominant agenda. Those who believe in UMC (universal mereological composition) would say that for any plurality of microparticles, there is the further object that is the mereological sum of these microparticles.10 But a maximal, cohesive, separately movable mass of matter is by no means just the mereological sum of the microparticles within it. For that mereological sum continues to exist even after those microparticles are spatially disbanded, and existed even before they were spatially assembled; not so, the cohesive mass. Moreover, if (what common sense calls) chairs and dogs are examples of cohesive masses, a cohesive mass can continue to exist even if one of the microparticles within it is annihilated; not so, the mereological sum of those microparticles. 10

E.g. Armstrong, World of States of Affairs, pp. 12–13 and 185.

123

Familiar objects and their shadows Can any way be found of confi ning the microparticles that together bear a particular instance of dogwise arrangement, to the region in which common sense believes there to be a dog? Well, there is one unwelcome way of doing this. We could rule that microparticles are dogwise arranged just in case (i) they are among a plurality of microparticles that between them are such as to cause the folk to judge that a dog is present, and (ii) they lie within the region occupied by a dog. In this way we could maintain that the phenomenon of dogwise arrangement commonly obtains in the world; that it really obtains just where the dominant agenda says it does, namely where common sense supposes there are dogs; and that its bearer is always many microparticles. But the cost of course would be huge. The cost would be saying that there really are dogs in the world, and that common sense (at least usually) is right about where dogs are present. The claim that there is dogwise arrangement in the world would leave open no further question as to whether the microparticles so arranged compose dogs. 6. 2 C au s a l c o m p e t i t i o n b e t w e e n d o g s a n d m i c r o pa r t i c l e s d o g w i s e a r r a n g e d Adherents of the dominant agenda believe that, in many regions of the world, there are microparticles dogwise arranged, microparticles baseballwise arranged, and microparticles treewise arranged. These microparticles are in a way the darlings of the dominant agenda: they fi ll a very important explanatory role. But at the present point in our proceedings, we are observing the microparticles dogwise arranged held captive in the regions where common sense believes there are dogs – held by captors in whom adherents of the dominant agenda are not yet ready to believe, namely dogs themselves. In one way, however, these philosophers should be glad about their darlings’ captivity. At least this ensures that microparticles dogwise arranged will be there when you need them. What I mean by this is that adherents of the dominant agenda can now claim to fi nd warrant for inferences that we often make, trust in, and profit from. When we observationally detect a dog – to put it in the way common sense would put it – we often infer that over the near term, we will or could, in that same vicinity, observe other indications that a dog is present. We might hear a dog, or smell a dog, or 124

Ontological preference for the spatially small see what looks like the same dog but from a different vantage point, or might see marks in the soil of the sort that a digging dog would leave. Adherents of the dominant agenda must construe this as an inference from an observation that microparticles are dogwise arranged, in this vicinity, to a conclusion that microparticles will prove to be dogwise arranged, in that same vicinity. But concerning which microparticles are we predicting continued dogwise arrangement? It is easy to see warrant in an inference that proceeds from “these entities are now displaying such-and-such features, or such and such an arrangement” to the conclusion “these entities will, over the near term, still be displaying such-and-such features, or such and such an arrangement.” But thus far we have been allowing that over an interval during which common sense supposes itself to be continuing to observe the presence of some one dog, it may be radically different pluralities of microparticles that are instantiating dogwise arrangement. For thus far we have allowed the bearers of any one instance of dogwise arrangement to spread far beyond the boundaries that common sense attributes to dogs. On such a picture, it is hard to see why the fact that there now are microparticles dogwise arranged, producing in me the experience of “seeing my dog,” in any way predicts that there will or would be microparticles dogwise arranged that would produce in me the further experience of “hearing my dog barking,” or that there would be microparticles dogwise arranged that would produce in me the experience of “seeing a hole that my dog must have dug.” But now that the microparticles that qualify as dogwise arranged are held captive within the borders of dogs – in the sense that they are, at any moment, just those microparticles found within a dog – we can warrantedly claim greater confidence about our observational future. For we are now making inferences from a premise about the microparticles that were collected together by the borders of some dog, to conclusions about whichever microparticles will be collected together by those same borders. Some of the microparticles originally within those borders may indeed depart, and others may arrive. But at least all such pluralities of microparticles will have this in common, that they occupy the volume of a dog. Even so, the idea that in the regions where common sense believes there to be dogs there are both many microparticles dogwise arranged and unitary dogs raises urgent questions about causal competition. Are our observations of dogs, and the effects that dogs characteristically 125

Familiar objects and their shadows produce in their surroundings, caused both by the dogs themselves and by the many microparticles within them? The question would have no sting if one could say that what it is for a dog to cause doggish experiences in an observer just is for the many microparticles within the dog between them to cause the experiences. But that is not what adherents of the dominant agenda would say:11 they say that it is certain that there are many microparticles in such a region, between them causing observations, but entirely unsettled whether dogs exist or cause anything. The question would again have no sting if one could hold that dogs and the many microparticles within them causally overdetermine all observations of dogs and all effects that dogs produce in their surroundings. But many philosophers would fi nd such overdetermination to be as hard to believe in here as they fi nd it to be in discussions of mental causation.12 There, the idea that the complex physical event on which Sam’s decision to drive to the supermarket supervenes causes the complex physical event on which Sam’s arrival at the supermarket supervenes seems to many to show that Sam’s decision itself has no causal efficacy. Some philosophers are willing to believe in a systematic causal overdetermination of all human behavior 13 – but not many. Just so, few philosophers would fi nd it believable that doggish observations are caused both by dogs and by vast pluralities of microparticles. So which of our two competitors is really doing the causing – the dog himself, or the many microparticles? Most philosophers would not hesitate in saying that it is the many microparticles that, between them, do the real causing. The main reason for thinking this comes, by an indirect route, from Davidson.14 Davidson argued, as I noted in the previous chapter, that any true case of causing had to instance some law of nature, and that any true law is exceptionless. But belief–desire psychology is incapable of formulating exceptionless laws, Davidson held; the generalizations of (what we now call) folk psychology will inevitably be hedged with ceteris paribus clauses. From here Davidson reasoned that since mental events do 11 12 13

14

Cf. Merricks, Objects and Persons, pp. 68–69. See e.g. Heil and Mele, eds., Mental Causation. Mills, “Interactionism and Overdetermination”; cf. Sider, “What’s So Bad about Over determination?” Davidson, “Mental Events.”

126

Ontological preference for the spatially small cause human behavior, they are identical with complex neural events, and describing each mental event as the complex neural event that it is will eventually enable us to subsume its effect upon behavior under an exceptionless law. Contemporary discussion of mental causation takes the relation between mental events and complex physical events in the brain to be supervenience, rather than identity, and takes Davidson’s point about exceptionless laws to show that these complex physical events do cause behavior while mental events probably do not. Many philosophers would not hesitate to draw, from Davidson’s point about exceptionless laws, a parallel conclusion concerning the competition between the dog himself and his component microparticles. We say that the dog’s barking causes me to hear him, and that his digging caused the hole. Just so we say that the baseball’s fl ight caused the window to shatter, and that the tree’s growth eventually put the shed into shade and contributed to mildew on the shed. But there is simply no prospect of identifying exceptionless laws at the level of “folk physics ,” that is, at the level of talk about familiar objects such as dogs and baseballs and trees. Yet things are quite different with the complex physical event – ultimately, the extremely complex event featuring microparticles within the region of the dog – on which this barking or this digging supervenes. Microphysics can reasonably aspire, many philosophers would say, to identifying exceptionless laws – or at least to identifying generalizations that are as close to exceptionlessness as nature will allow, and close enough to qualify as true laws. The event involving many, many microparticles in the dog, on which the dog’s barking supervenes, truly causes a similarly complex event involving microparticles in my ears and brain, and thereby causes my hearing of the dog. The dog himself causes nothing. The popularity of such a verdict is in one way puzzling. Any law of nature that tied the microphysical event that subvenes the dog’s barking to the microphysical event that subvenes my hearing the dog would have an antecedent that is trillions and trillions of conjuncts in length, and a similarly enormous consequent. It seems very likely that such a law would be instanced exactly once in the history of the world. So while such a law would be exceptionless, its exceptionlessness would be an empty technicality. I shall return in section 6.4 to consider ways in which philosophers inclined to use causal exclusion arguments against dogs might respond to this worry. 127

Familiar objects and their shadows

6.3 W h y “m o r e a n d l e s s d i f f e r e n t f r o m ” i s n o t d e f i n e d ov e r t h e m i c r o p h y s i c a l e v e n t s t h at s u b v e n e d o g b e h av i o r s In any case, Woodward and Hitchcock have, both together 15 and separately,16 put into serious question the premise that any true causing must instance some exceptionless generalization – as I indicated in chapter 5. I contend that if, following Woodward and Hitchcock, we see causation as underwritten by invariances, rather than by exceptionless laws, we not only vindicate the claim that dogs often cause doggish observations and dog-like effects in their surroundings; we also undermine the claim that the many microparticles found within the dogs’ borders cause these same things. Let me begin with the easier claim, namely that invariances license familiar claims that events involving dogs cause familiar effects. Consider, for example, the claim that a dog’s barking will often cause in nearby persons the sensation of hearing a dog bark. Putative causes of that type can be subsumed under the two-value variable B {a dog barks, no dog barks}. Then if we set the exogenously determined variables at appropriate levels – if we set “level of background noise” at less than 4 decibels, “presence or absence of other dogs barking” at absence, “distance between dog and nearby person” at less than 40 feet, etc. – it surely is true that by intervening on the value of B, we can manipulate the value of H {nearby person has the sensation of hearing a dog bark, nearby person does not have this sensation}. But there are also richer and more interesting invariances linking a dog’s barking and a nearby person’s hearing. For there are multivalued variables under which a dog’s barking can be subsumed, and multivalued variables that are instanced by a person’s sensation of hearing, such that invariances link values of the former to values of the latter. For example, provided we keep the exogenously determined variables at the same appropriate settings, different values of how loudly a dog barks will pair with different values of how clearly and distinctly a person auditorily senses a dog to be barking. Different values of the pitch of the bark will pair with different values 15

16

Woodward and Hitchcock , “Explanatory Generalizations, Part I: A Counterfactual Account.” Woodward , Making Things Happen; Hitchcock , “Intransitivity of Causation,” and “Tale of Two Effects.”

128

Ontological preference for the spatially small of the pitch at which a person senses a dog to be barking. Similar points can be made about a case of putative causation that figures prominently in Merricks’s Objects and Persons.17 This is the claim that a baseball’s fl ight can cause a window to shatter. Across many settings of exogenously determined variables, controlling whether the ball fl ies onto the window enables control of whether the window shatters – there is an invariance between {a baseball fl ies onto window, no baseball fl ies onto window} and {window shatters, window does not shatter}. More importantly there are invariances that link multivalued variables subsuming the fl ight of a baseball, to the shattering of a window. To some degree one can control how violently the window shatters, and how large the hole is, by adjusting how rapidly the baseball travels.18 But consider, now, the extremely complex physical event, featuring trillions upon trillions of microparticles within the dog’s borders, that subvenes the dog’s barking. Does it cause the equally complex physical event, featuring trillions of microparticles in my ears and brain, that subvenes my sensation of hearing a dog barking? An initial question is what form we should attribute to the invariance or invariances that would link the bark-subvening event to the hearingsubvening event. Different answers are possible. But suppose – to begin with what seems to me the clearest case – we ask whether there is a single invariance, linking a multivalued variable that subsumes the bark-subvening event, to a multivalued variable that subsumes the hearing-subvening event. For such an invariance to obtain, as I argued in chapter 5, there must be a fact of the matter as to which variations in Vc are small and which are great. But do we even understand what it would be for variants, on the complex microphysical event that subvenes the actual barking, to differ from that event to greater and lesser degree? But the question which I mean to raise here is not just epistemological. Call the complex microphysical event that subvenes the actual bark b. Consider other complex microphysical events that would be variants on b. Is there a phenomenon of differing to greater or lesser extent from b, that is defi ned over these variant events?

17 18

Merricks, Objects and Persons, Ch. 3. Cf. Woodward , Making Things Happen , p. 218.

129

Familiar objects and their shadows The problem, as in chapter 5, is that candidate variants on b seem to differ from b on many, many different dimensions of difference – so many that the relevant dimensions of difference are “mutually interfering.” That is, b3’s differing more from b itself than b2 does, along any such dimension of difference, may amount to b3’s being less different from b itself than b2 is, depending on how b3 and b2 line up on yet other dimensions of difference. To see the parallels here with the argument in chapter 5 (pp. 103–4), consider these variants on b itself. Event b1. It instantiates a structural property, the description of which differs from the description of b’s structural property with respect to 1,000 conjuncts belonging in the latter description. Event b2 . It instantiates a structural property which differs with respect to 5,000 conjuncts. Event b3. With respect to every conjunct where its description differs from b ’s description, a microparticle bears a markedly different charge or energy-state from that of the microparticle described in the corresponding conjunct of b ’s description. Event b4. With respect to every conjunct where its description differs from b’s description, a microparticle bears only a slightly different charge from that of the microparticle described in the corresponding conjunct of b ’s description. Event b5. Its description requires only 1,100 conjuncts – that is, b5 is a less complex microparticle event than b – but differs at 1,000 of those conjuncts from b’s description. Event b 6. Its description is the same, conjunct for conjunct, as b’s, and requires no more conjuncts than b ’s description does, but its conjuncts are satisfied at regions far more spread out from one another, spatially, than are b ’s conjuncts. Event b7. Its description incorporates every conjunct in b ’s description, but requires 18,000 more conjuncts as well – b7 is a far more complex microparticle event than b.

The upshot is – just as in chapter 5 – that there is simply no prospect that a systematic ranking obtains, by which variants on b itself count as “more and less different from b.” That is, for pairs of variants on b that occupy positions on any dimension of difference from b itself, there simply is no fact of the matter as to which member of such a pair is more, and which less, different from b itself. But then it is empty 130

Ontological preference for the spatially small to speak of greater and greater differences from b, with respect to some multivalued variable that subsumes b, as producing, as a general matter, corresponding or answering variant values of the variable that subsumes the sensation of hearing the dog. Even so, we must also ask whether there may be a single -valued invariance that links b to the microphysical event that subsumes the sensation of hearing the dog. For even if it has no clear sense to ask how matters would have gone if complex microevents more and more different from b had occurred in b ’s place, it still might make perfectly good sense to ask how matters would have gone if b had not occurred at all. Hence again we must consider the closest possible worlds in which b does not occur – worlds in which some microphysical event or other occurs in b ’s place and “differs only minimally” from b itself. Again we must be leary of the intuitive idea that there even is a welldefi ned class of such “only minimally different” variants on b. But even if there objectively are such variants, we must remind ourselves that if there are any, there are many. For possible variants on b diverge from b on thousands and thousands of dimensions of difference. If we compare b with these thousands and thousands of other variant events, is it plausible that b alone was causally suffi cient to produce the hearing-subvening event – that if precisely b, that very event, had not occurred, the hearing-subvening event would not have? Again, it seems out of the question that we should really have warrant for answering Yes. 6. 4

C h a i n s o f c au s at i o n i n vo lv i n g m i c r o pa r t i c l e e v e n t s

But so far I have ignored what probably is the most promising strategy for arguing that the complex microevent that subvenes the dog’s barking causes the complex microevent that subvenes the nearby person’s experience of hearing the dog. Philosophers persuaded by Davidson that any individual causing must instance some exceptionless law, I said at the close of section 6.2 , would fi nd it embarrassing to hold that a single exceptionless law links the bark-subvening event b as a whole to the hearing-subvening event h: the “exceptionlessness” of such a law would be empty. So what such philosophers really are thinking must instead be that b is linked by many different exceptionless laws to h. Each individual microparticle movement or state-change, 131

Familiar objects and their shadows spotlighted by a single conjunct in the complex description of b, is linked by some exceptionless law to another microparticle event a few nanoseconds later, and that event by a different law to yet another microparticle a few nanoseconds later still, yielding at length one of the microparticle events that is featured in a conjunct in the description of h. Between them, by way of many laws indeed, the components of b would then be said to cause the components of h. Now suppose I am right in holding that we should accept the Woodward /Hitchcock claim that what really underwrites token causation is always some invariance in nature. Then the philosopher who thinks that the dog’s microparticles do all the real causing should probably say that what links b to h are many different invariances. If we look at each individual microparticle involved in b, and fi x the exogenously determined variables, we fi nd that an intervention on that microparticle’s activity would produce an alteration in the activity of some other microparticle, nanometers away, nanoseconds later. A chain could perhaps then be traced that would eventually lead from the activity spotlighted by one conjunct in the description of b to a microparticle event that is one of the components of h. But such a chain would involve trillions of steps, and tracing it would require holding fi xed the values of trillions of exogenously determined variables – variables determined not just exogenously but independently of one another. Would a connection so indirect, between a given component of b and a given component of h, really count as a case of token causation? Woodward and Hitchcock consider no causal chains as long as the one we are envisioning: such chains are the concern of metaphysicians, not actual working scientists, and Woodward and Hitchcock are concerned to make sense of actual scientific practice. But Woodward’s schema for token causation19 appears to suggest that he would say about such highly mediated and indirect causal chains just what Lewis said about them:20 the fi rst event in such a chain does really cause the last. This verdict, I shall argue, is contrived and misleading, if not indeed outright false. (And Woodward in the end says the same; see below.) But let us begin by letting Lewis put the claim forward in a plausible light.

19 20

Woodward , Making Things Happen , pp. 77 and 84. Lewis, “Causation,” p. 167.

132

Ontological preference for the spatially small Lewis says21 that he has, by his activity as a philosophy professor, caused countless deaths. He says this because it is a consequence of his view that causation is the ancestral of counterfactual dependence.22 Lewis has, over his career, written many letters of recommendation for graduate students of his. In many cases, had Lewis not written the letter, the graduate student would not have been offered the job that he or she was offered. If the graduate student had not been offered that job, she would not have taken that job. If she had not taken the job, she would not have moved to that city; hence she would not have met the person whom in fact she married; hence she would not have had the children whom in fact she had.23 These children would not have lived, and so they would not have died. So Lewis caused their deaths, and indeed the deaths of their off spring. To put this case in the terms of the Woodward/Hitchcock account, we are looking here at an extremely complex causal system, in which there are very many exogenously determined variables (presence of the eventual spouse in the relevant city, health of the parents good enough to enable conception, health of the mother good enough to enable gestation, etc.). We hold all such variables fi xed at their actual values, and ask how matters would have gone if there had been an intervention on solely the value of W {Lewis writes on behalf of the student, Lewis does not}. A letter from David Lewis did (rightly) carry a lot of weight. It seems true that an intervention of W would eventuate in the actual children never having been born, and hence never having died. But does it really square with our pretheoretic intuitions to say that Lewis’s letter-writing activities caused the deaths of all these future people? At the least, I would suggest, it is unclear whether this is the intuitive thing to say. The same is true of many “puzzle cases,” as both Lewis and Woodward admit; Lewis, for example, thinks it is intuitively unclear whether there are any actual cases of symmetrical 21 22 23

Lewis, “Causation,” p. 184. Lewis, “Causation,” p. 167. I am simplifying just slightly the chain of counterfactual dependence that Lewis actually narrates: in Lewis’s text, the recommended job candidate displaces a competitor, who then takes a different job, displacing a third candidate who had been considered for that job, and the third candidate goes on to have the children (Lewis, “Causation,” p. 184).

133

Familiar objects and their shadows overdetermination.24 Woodward, for his part, thinks it is not the primary business of a theory of causation to codify our intuitions.25 This seems a defensible position. Our intuitions are simply too plastic; they can be pushed one way or another by the influence of one or another “intuition pump.”26 The real objective phenomena, that our causal talk responds to, are the invariances that obtain in nature, says Woodward.27 Once we are clear on where the invariances are, we in a way know all the facts. For the rest, it is just a question of how we choose to talk. Yet there is, at the level of token causation, an invariance tying the value of W to the value of the variables that subsume the deaths of just those children. Just so, in the case of (what common sense calls) the barking dog. If we hold fi xed the values of trillions of exogenously determined variables – variables determined not just exogenously but independently of one another – then an intervention on one of the microparticle movements that is a component of b might very well produce an alteration in one of the microparticle movements that is a component of h. The question is, is such an invariance robust enough to yield a substantive, explanatory claim of token causation? This is not exactly to ask whether, across a wide range of circumstances, an individual microparticle movement that is a component of b would eventuate in the very same component of h as it actually did. For it probably does not even make sense to speak of that very microparticle movement, which figured as one among trillions of conjuncts in b, as itself occurring in some different setting or at some different time and place. But it is reasonable to ask whether it is in virtue of that movement’s being the sort of event that it is, that it eventuates in just such a microparticle movement as is included in h. Is there a connection of any robustness at all between that sort of a movement’s occurring, in that sort of a microparticle, and there following, many milliseconds later, that sort of an event within h that that movement is supposed to have caused? The suggestion I make is this. Invariance is the heart of the phenomenon of causation. So it is appropriate to ask, even concerning a 24 25 26 27

Lewis, “Causation,” Postscript E. Woodward , Making Things Happen , pp. 85 and 7–9. Cf. Dennett, “Quining Qualia.” Woodward , Making Things Happen , pp. 85–86.

134

Ontological preference for the spatially small putative case of token causation, whether there is some reproducible correlation at the level of type causation, some invariance, between a variable that captures the sort of event to which the putative cause event belongs and a variable that captures the sort to which the putative effect event belongs. If there is, the claim of token causation can be informative and explanatory. If there is not, the claim of token causation will seem contrived and even amusing, perhaps not strictly false but certainly not informative. The best cases of (apparent) token causation where there is no such invariance at all are provided by “Rube Goldberg machines.” These fictional causal systems, drawn by the cartoonist after whom they are named, featured mechanisms arranged in series, so that activation of the fi rst would result in activation of a mechanism far removed – and many crucial connections between the mechanisms relied on actions by animals or people. Thus a squirrel’s seizing an acorn might release a lever that would open the door to a cage containing a hungry dog, who would seize a nearby bone, thus pulling a string that releases a weight that causes a mechanical leg to kick a man reading a newspaper on a park bench. (Actual Rube Goldberg machines involved many more linkages.) The amusing suggestion presented by such a machine is that the squirrel’s seizing an acorn caused the man to get kicked in the face by a boot. What makes the suggestion amusing is, roughly, that there is in general no connection at all between whether and how squirrels gather food, and whether and how people suffer bodily violence. More precisely, there is an invariance that ties values of {squirrel seizes acorn, squirrel does not} to values of {nearby man gets kicked in face, nearby man does not}, but it is an invariance that obtains over only a point-sized range of cases – a range entirely populated by a “machine” apparently put together by bizarre coincidence. The more sober reading of the “machine” is that the squirrel’s action causes the cage door to be opened, this in turn causes the bone to be picked up, etc. – but that transitivity fails. My claim is that standard, explanatory claims of token causation require that some invariance obtain between the sort of event to which the putative cause belongs and the sort of event to which the putative effect belongs. To put it in Woodward’s terms, explanatory and informative claims of singular causation must instance a “possible-cause” generalization. “[A]s a general rule,” Woodward writes, “we are more willing to regard a singular-causal claim as a 135

Familiar objects and their shadows satisfying explanation if the putative cause and effect are described in such a way that they fall under a relatively invariant possible-cause generalization.”28 To put it differently, an informative claim of singular causation must indicate ways that interventions on or alterations in the putative cause event would enable manipulation of the putative effect event. “Many possible-cause generalizations should be understood as claims that this manipulability relationship … will hold in many different circumstances or for many different interventions – as claims that the relationship is in this sense robust or invariant across such changes.”29 Return, then, to the claim that Lewis’s writing a certain recommendation causes the death, eighty years later, of a child born ten years after the recommendation. There is virtually no correlation between the sort of event to which the putative cause belongs – perhaps “recommendation-writings by academic advisors” – and the sort of event to which the putative effect belongs – perhaps “deaths of seventy-year-old women.” There is a connection between Lewis’s writing the letter and the eventual off spring’s dying, but it obtains only because a host of exogenously determined variables assume just the right values, and do so independently of one another. In consequence, no one would ever dream that intervening on Lewis’s recommendation-writing activities is a way of manipulating the incidence of death in children born ten years later. Consider, next, the suggestion that an individual microparticle movement, that is a component in the bark-subvening event b, causes some individual microparticle movement that is a component in the hearing-subvening event h. Here too the connection between the putative cause and the putative effect is highly mediated. It depends on trillions of exogenously determined variables assuming just the right values. The connection does not really instance a possiblecause generalization at all – or if it does, it instances a generalization that obtains over only a point-sized range of cases. So the claim of token causation is utterly unexplanatory and uninformative, if not indeed outright false. It is contrived in the same way that a parallel claim about a Rube Goldberg machine is contrived – in the same way,

28 29

Woodward , Making Things Happen , p. 217, my italics. Woodward , Making Things Happen , p. 215.

136

Ontological preference for the spatially small but to a vastly greater degree. Again, the more sober reading is to deny transitivity. Before closing this section, I should fend off one objection to the claim that a substantive, explanatory claim of token causation must instance some invariance at the level of type causation. The objection is that there are perfectly genuine cases of mediated causation, both at the type level and at the token level, and that in these cases, the connection between cause and effect is underwritten not by one but by two or more invariances. At the type level, it is perfectly true that use of birth control pills causes a steep reduction in the rate of pregnancy; that avoidance of pregnancy greatly reduces the incidence of thrombosis; hence that use of birth control pills causes a reduction in the incidence of thrombosis. At the token level, one could plausibly say that this woman’s taking birth control pills caused her to avoid thrombosis. In both cases, the objection runs, the claim of causation is underwritten by two invariances: the invariance by which ingestion of birth control pills influences the rate of pregnancy, and the further invariance by which the avoidance of pregnancy influences the incidence of thrombosis. My response is that even so there is an invariance, albeit weaker than the two just mentioned, between whether women ingest birth control pills and whether they avoid thrombosis. Yes, there is a more robust invariance between the administration of birth control pills and the nonoccurrence of thrombosis across the population of women who would not otherwise avoid pregnancy, than across the population of women as a whole. But even across the population as a whole, administering birth control enables some manipulation of the rate at which thrombosis is avoided. That one invariance is what makes the claim of token causation plausible. 6.5

Sum m i ng u p

The widely prevailing agenda in metaphysics takes it as uncontroversial that the contents of many regions of the world are dogwise arranged or treewise arranged or baseballwise arranged. I agree. But we will be able to identify the right regions, I have argued, only if we say that within them there are not just many microparticles but also dogs or trees or baseballs. And the real bearers of dogwise and treewise and baseballwise arrangement – since these are fundamentally 137

Familiar objects and their shadows causal phenomena – are dogs and trees and baseballs themselves. Once one has affi rmed that there are in the world the phenomena of dogwise and treewise and baseballwise arrangement, therefore, it is far, far too late to raise questions about whether there really are in the world dogs and trees and baseballs.

138

7 A third false friend of familiar objects: universal mereological composition Metaphysicians who are sceptical about familiar objects for the most part suppose that the world is really populated by the fundamental particles of completed microphysics, or by mereological simples – or by both, should the fundamental particles turn out to be mereological simples. They then must explain why the world at least appears to contain objects that are so much larger, objects such as dogs and trees and boulders. The metaphysicians considered in the previous chapter suppose that they can explain this appearance by holding that adverbially identified “arrangements” obtain among the microparticles, such as dogwise and treewise arrangement. But other proponents of an ontology of the small respond differently. They hold that there really are unitary large objects, in all of the volumes in which common sense supposes a familiar object to be present, and that this follows merely from the premise that each of those volumes is occupied by microparticles or simples. These philosophers are adherents of UMC (universal mereological composition). In this chapter I argue that UMC is both unmotivated and untenable; the only way to have unitary, relatively large, objects in all the volumes where common sense supposes familiar objects to be present is to have familiar objects themselves. But even apart from its relation to the ontology of familiar objects, UMC is of considerable intellectual interest. Few theses widely accepted among philosophers who work in metaphysics seem more outlandish, to people who work in other areas, than UMC. What this doctrine says is that for any objects whatever – however arbitrary it may seem to consider them together – there is a further object of which those objects are parts. This further object is sometimes said to be “an ontological free lunch”: its existence, its defenders say, just is the existence of its parts.1 It is natural, when thinking along these 1

Armstrong, World of States of Affairs, pp. 12–13 and 185.

139

Familiar objects and their shadows lines, to suppose that the identity of this mereological objec t is given by the identity of its parts – that this object could not have had different parts from its actual ones, and that necessarily this object exists exactly where and when its actual parts exist.2 Some philosophers, it is true, use the phrase “mereological sum” (or “mereological sum of …”) to designate objects for which these modal claims do not hold. 3 But the mereological objects on which this chapter focuses are objects to which the mereological essentialism , implicit in the slogan “ontological free lunch,” does apply (as do the associated claims about spatial and temporal location). It is these objects that most philosophers have in mind when they speak of “mereological sums,” and these objects that most philosophers take the thesis of UMC to affi rm.4 In the last section I will defend this restricted focus. The thesis that there is a mereological object, for any arbitrary plurality of objects, is treated as being neutral on the question of which objects are there to be welded into mereological sums. Perhaps these objects include only the microparticles of physics, or only these together with persons, or perhaps they also include the familiar inanimate objects in which common sense believes. The thesis is just that mereological summing always yields a real object, provided it starts with real objects. This neutrality about fused objects explains why the examples usually given of these mereological objects typically have, as their parts, such familiar objects as stars and tennis shoes and the Eiffel Tower 5 – even though very, very few philosophers 2

3

4

5

Is numerical sameness in the parts not only necessary but also sufficient for numerical sameness in the mereological sum – that is, is it a necessary truth that so long as numerically the same objects compose some mereological sum, they compose numerically the same mereological sum? “Yes” is the answer that best harmonizes with the idea that any mereological sum is an “ontological free lunch ,” and that is the answer that van Inwagen gives in Material Beings (pp. 75–78). But Michael Rea gives the opposite answer, even while claiming to be discussing mereological sums: Rea, “In Defense of Mereological Universalism,” pp. 349–51. For further discussion of the sufficiency claim, see section 7.5. Van Inwagen, “Can Mereological Sums Change Their Parts?”; Saucedo, “Parthood and Location.” Baker, Persons and Bodies, pp. 179–85; Lowe, Kinds of Being, Ch. 6; Wiggins, Sameness and Substance, pp. 30–34; Sidelle, “Sweater Unraveled,” pp. 430–33; Sanford, “Fusion Confusion.” Rea, “In Defense of Mereological Universalism,” p. 248; Merricks, Objects and Persons, p. 51.

140

A third false friend of familiar objects: UMC currently working in metaphysics defend the reality of such familiar objects as these. This chapter opposes UMC. I begin by examining the most appealing argument for it in the recent literature. This is the argument, taken by Theodore Sider from materials in David Lewis,6 that forms the basis for Sider’s argument for four-dimensionalism.7 Assessment of this argument is thus of interest beyond the topic of UMC. I argue, amplifying an objection by Kathrin Koslicki,8 that Sider’s argument rests on a claim that would be rejected by those whom the argument seeks to persuade (as Koslicki partly points out) and that is strangely unmotivated (since the claim forgets that, at various places in the world, there are stuff s and matters such as butter or coffee or petroleum). I then turn from arguing that UMC is unsupported to arguing that it is false. UMC is sometimes treated by its own advocates as sounding odd and counterintuitive;9 nevertheless, few concerted objections have been offered against it. The objections that have been offered generally10 focus on a special extension of UMC – namely the thesis, discussed in section 7.2 , that there are in the world diachronic mereological objects – and present one version or other of the idea that these objects are profoundly unlike the Aristotelian substances in which common sense believes, since the later stages in the careers of these objects typically do not causally reflect, or intelligibly grow out of, the earlier stages.11 The motivation here appears to be to get adherents of UMC to take more seriously the sorts of features that common sense (and perhaps empirical science as well) looks for, in judging about which objects populate the world. The objection that I shall offer here is different. I think it is true and important that diachronic mereological objects are deeply unlike the objects familiar to common sense and to empirical science – but I question whether that point is likely to carry much weight in the 6 7 8 9 10 11

Lewis, “Causation,” pp. 212–13. Sider, Four-Dimensionalism, pp. 120–39. Koslicki, “Crooked Path.” Lewis, Parts of Classes, pp. 79–81. One interesting exception is Laycock , Words without Objects, pp. 95–96. Koslicki, “Crooked Path,” pp. 125–28; Thomson, “Parthood and Identity across Time,” p. 213.

141

Familiar objects and their shadows contemporary climate. For it is now fairly common for metaphysicians to endorse ontologies that differ sharply from the ontological commitments – at least, the apparent ontological commitments – of common sense and many empirical sciences. The objection that I shall offer is that the “structural” properties, which are the only properties that would characterize the typical UMC object, are not genuine properties at all. Typical UMC objects, then, simply have no properties at all. This objection does not say: we should not believe in such objects, since they fail to be like Aristotelian substances. Rather it says: we should not believe in such objects, since they fail to be like anything – not even themselves. 7.1

T h e S i d e r /L e w i s a r g u m e n t f o r U M C

Sider’s argument for UMC12 has the form of a reductio. The hypothesis that UMC is false amounts to saying that it is not the case that for just any and every class of objects, there exists a mereological object of which those objects are all parts (and such that any part of the mereological object overlaps one or more of the objects in the class): it denies that every class has a fusion. What it asserts, in the phrase that Sider borrows from Lewis, is that composition is restricted. The argument itself has the form of a sorites. One envisions a continuous series of cases, stretching from an extreme at which composition defi nitely does not occur – there is only a plurality of objects, and not some single object which the members of the plurality jointly compose – to a case in which composition defi nitely does occur. There is, across the series, a progressive increase in whatever features they are, that the opponent of UMC thinks of as making for composition – perhaps spatial proximity, perhaps causal integration, perhaps qualitative homogeneity, perhaps a combination. This picture gets stated in the fi rst premise (P1) of Sider’s argument, and the following two premises do the real work. P1. “If not every class has a fusion, then there must be a pair of cases connected by a continuous series such that in one, composition occurs, but in the other, composition does not occur.”

12

Sider, Four-Dimensionalism, pp. 120–39.

142

A third false friend of familiar objects: UMC P2. “In no continuous series is there a sharp cut-off in whether composition occurs.” P3. “In any case of composition, either composition defi nitely occurs, or composition defi nitely does not occur.”13

As Sider rightly points out,14 it is the last premise that is the most controversial. A philosopher who thinks that baldness is “restricted” – that not just every person is bald, though some are – would probably deny that baldness is crisply restricted. She would more likely claim that there are borderline cases. Just so, the philosopher who thinks that composition is restricted might very well think that there are cases in which the objects in some class neither defi nitely compose a larger object, nor defi nitely fail to compose such an object. P3 needs defense, and Sider undertakes to defend it.15 The defense proceeds from this thought: if in some class there are n objects, and those objects together compose an object – if, that is, the class has a fusion – then there exist, with respect to that segment of the world, n+1 objects. If on the other hand there can be borderline cases of composition, then with respect to the segment of the world containing some class of n objects, it will be neither determinately true that there exist n+1 objects, nor determinately false. Thus we should believe the following conditional: (N) If there can be borderline cases of composition, then, for some fi nite and nonempty world, there is a numerical sentence that is indeterminate in truth-value.

“Numerical sentences” are ones that “contain only logical terms and the predicate ‘C’ for concreteness.”16 Thus a numerical sentence that says that there are in the world three objects reads as follows: ∃w∃x ∃y [Cw & Cx & Cy & w ≠ x & w ≠ y & x ≠ y & (∀z) (Cz → [z = w ∨ z = x ∨ z = y])]. Soon I will argue that (N) is in fact unmotivated. But at fi rst blush it certainly seems eminently reasonable, and so let us ask how the opponent of UMC might reconcile himself with its apparent truth. 13 14 15 16

Sider, Four-Dimensionalism, pp. 123–25. Sider, Four-Dimensionalism, p. 125. Sider, Four-Dimensionalism, pp. 125–26. Sider, Four-Dimensionalism, p. 127.

143

Familiar objects and their shadows The opponent affi rms the antecedent. But how – asks Sider, echoing Lewis – could the consequent possibly be true? There is nothing ambiguous or vague about “(∃x) … x …”

and

“(∀x) … x …,”

provided we are considering – as we should – unrestricted quantification. There is nothing ambiguous about “… = …” or about “… ≠ …” A numerical sentence says nothing vague. So how could it possibly be indeterminate in truth-value? Koslicki offers this response: it all depends on what sort of a world it is, that the sentence is quantifying over.17 If there really can occur in the world borderline cases of composition, then there will be numerical sentences that speak precisely about a vague situation, and these will be numerical sentences whose truth is indeterminate. This response presupposes that not all vagueness need be vagueness in speaking: it rejects “the linguistic theory of vagueness,” and maintains that vagueness can obtain out there in the world. Sider does consider this sort of response. He writes, “I mention this position only to set it aside; as I said above, I simply assume that this theory of vagueness is not correct.”18 But this renders Sider’s “defense” of P3 inconclusive. The “defense” rests on (N), and yet it simply refuses to speak to those philosophers who are unpersuaded that (N) always sets up a modus tollens. Here is a less confrontational version of the same sort of objection. Consider a fi nite world in which there are some banyan trees. A banyan tree is a most remarkable organism. As the branches grow they send out tendrils that reach downwards. These eventually reach into the earth, establishing roots; they can grow to look like nonprimary trunks; they can even grow to be as thick as the original trunk. Where there are well-established banyan trees, there seems to be no clear answer to how many banyan trees there are – and the problem seems to lie not with us and our ways of knowing, but with the world. I hasten to add that banyan trees are not a flukish “trick example.” Aspen trees are connected at the roots, and it can happen, in autumn, that a stand of aspens divides into large segments, each uniformly bearing its own distinctive coloration. There is a fungus 17 18

Koslicki, “Crooked Path,” pp. 118–19. Sider, Four-Dimensionalism, p. 129.

144

A third false friend of familiar objects: UMC below ground in northern Michigan which some count as being “the largest organism” in the world, though here too the count seems inherently contestable. Hegel found such examples to be of philosophical interest: he maintained that the plant kingdom was the area of nature in which individuality itself was attenuated.19 Which numerical sentence will be determinately true of a fi nite world containing mature banyan trees? None, it seems. If we opt for a numerical sentence that features relatively many existential quantifiers, we run the risk that some of the clauses asserting nonidentity are false. If we opt for a numerical sentence that features relatively few existential quantifiers, we run the risk that the universally quantified conjunct is false. In a world containing mature banyan trees, identity and distinctness seem to be vague. Sider considers this response as well. He writes, “I fi nd this doctrine obscure but have nothing to add to the extensive literature on this topic; here I must presuppose it false.”20 Again, Sider’s “defense” of P3, by way of (N), is inconclusive: it does not address a reason for thinking that certain numerical sentences might fail to be determinately true. But the challenge I have lodged against (N) invites a response to which Koslicki’s original challenge is immune. The proponent of UMC could simply deny that there are banyan trees out there in the world, to be quantified over. The proponent might elect to maintain that, apart from mereological fusions, there are only crisply individuated objects, such as the microparticles of physics. But there are two problems with this rejoinder, a greater and a lesser. The lesser problem is that at least some microparticles fail to be as crisply individuated as one might hope. An electron captured by an ionized helium atom becomes superposed with the one electron that is already there, with the apparent consequence that, if the atom again undergoes ionization, there is no fact of the matter as to whether the electron later stripped off was or was not the electron that was earlier gained.21 The greater problem is that, as I noted at the outset of this chapter, UMC is treated as neutral with respect to which objects there are in the world, to be welded into mereological sums: it says only that mereological summation always yields a real object, provided it starts 19 20 21

Hegel, Philosophy of Nature, §§343 and 347–48. Sider, Four-Dimensionalism, p. 130. Lowe, Kinds of Being, p. 62.

145

Familiar objects and their shadows with real objects, whatever these may be. Thus the defense of UMC should not presuppose a particular ontology, a particular roster of which objects are really there. In any case, there is room to wonder whether, as a general matter, (N) is a well-motivated claim at all. That is, there is room to wonder whether the question of how much composition occurs in the world has, in general, anything to do with how many objects there are in the world. For suppose that the n microparticles in some class between them compose some butter, or some wine, or some coffee. Which numerical sentence is rendered true by this instance of composition? To answer that question, it seems, we would have to say how many butters there are, or how many coffees the n microparticles have composed, or how many wines. But these questions are ill-formed: “butter,” “wine,” and “coffee” are noncount nouns. We might also ask – for anything Sider has shown – what happens if there is, on the part of the n microparticles, only a borderline instance of composing-butter or composing-coffee. Has the borderline character of the composition robbed determinate truth from some numerical sentence? Of course there are, in the philosophical lexicon, a number of count nouns that can be used to talk about what is there, where there is some butter or some wine or some coffee. One can speak of a “parcel” or a “sample” or an “aggregate,” and one can set about counting “parcels” and “samples” and “aggregates.” There is an ancient tendency among philosophers, recently well documented by Henry Laycock, to discern unitary objects when dealing with the parts of the world that contain butter or wine or coffee.22 It is connected with a tendency to ignore the fact that some noncount nouns are, semantically, nonatomic: that is, that they do not, as a function of their meaning, divide their reference over discrete units.23 Before these tendencies again spring into play, let me say some things about the coffee that is there before us, on an occasion that I shall describe, and say it using only a philosophically innocent count noun that occurs in ordinary usage, namely “portion.”

22 23

Laycock, Words without Objects, Chs. 2–4. Laycock, Words without Objects, pp. 135–39; cf. Lowe, Kinds of Being, pp. 161 and 72–74.

146

A third false friend of familiar objects: UMC The office assistant brings into the business meeting a large, full carafe of coffee. From it he pours coffee for each of the eight people participating in the meeting.24 The fi rst five people receive full cups; the sixth through eighth, a cup only half full. But Person 7 is assertive, and asks the assistant for more coffee. The office assistant obliges, and the carafe is now empty. How many portions of coffee are there in the conference room? One could say that the very large portion of coffee that the assistant brought in, in the carafe, is still there, but is now spatially discontinuous: it is separated across the eight cups. One could say that there are eight portions of coffee in the room. But what about Person 7 – didn’t he receive a second portion? Yet Person 1 has no more coffee in his cup than Person 7 now has. So does Person 1 have two portions? And what about the coffee in the bottom two inches of Person 2’s cup? It is coffee, and so it is some coffee, and so it is a portion of coffee. There is simply no fact of the matter, I suggest, as to how many portions of coffee there are in the conference room. There is much coffee in the room, to be sure. But there being much coffee there makes no contribution to the truth conditions for any numerical sentence. Nor would the truth of any numerical sentence be rendered indeterminate if the coffee had been quite badly made, and were only barely coffee at all. If the n microparticles in some class of microparticles only borderline-compose coffee, it does not follow that some numerical sentence is only borderline true. That is, it could perfectly well be indeterminate whether anything in the conference room composes coffee – and indeterminate whether any things ( plural) in the conference room compose coffee – but true that the whole scene occurs in a fi nite world, in which every well-formed numerical sentence is determinately true or false. Thus one might not only question – as Koslicki does – whether the consequent in (N) need in all cases be false; one might also question whether the truth of the antecedent in (N) suffices, in all cases, for the truth of the consequent. To put it differently: how widely composition occurs in the world does not, as a general matter, have bearing on how many objects there are in the world – as (N) alleges. The challenge to (N) that I am drawing, from the existence in the world of coffee and butter and wine, is in one way like the challenge 24

Cf. Laycock, Words without Objects, p. 49.

147

Familiar objects and their shadows to (N) that I draw from the existence of banyan trees. The proponent of UMC can defeat it by electing a particular ontology. She can deny that there is such a thing as butter, or such a thing as coffee, at all: instead, she can claim, there are only the many microparticles that, according to me, here and there and there again compose some butter or some coffee (together with the myriad mereological sums that incorporate those microparticles as parts). But again, this way of preserving the argument for UMC threatens the ontological neutrality that the doctrine, once established, is generally treated as having. If we do allow that there may be in the world such things as coffee and lemonade, there is indeed one other way in which we might wonder why there need be any connection at all between how much composition occurs in the world, and how many objects there are in the world. The “cases of composition” that Sider discusses are all cases in which some nonfuzzy class of objects – some class containing exactly n elements – has (or fails to have) a fusion. But it is not clear that all composition is like this. “If you take some water and some lemon juice,” Granny might have said, “and some sugar and a little salt, and if you mix them thoroughly together in just the right proportions, you will end up with some lemonade.” It would be a mistake for a proponent of UMC to argue for her view by denying that what Granny said is true. But if Granny’s utterance can be formalized, it must be formalized using the plural quantification that Boolos developed.25 7. 2

T wo f o r m s o f U M C o bj e c t s , t wo f o r m s of str a nge n ess

UMC objects, if there are such things, are deeply strange from the standpoint of common sense and even from that of empirical science. Does their strangeness furnish materials for arguing that such objects do not exist? So some philosophers have suggested.26 I myself am inclined to agree, but this is not the sort of objection I will emphasize; my objection focuses rather on the “structural properties” that would characterize typical UMC objects. Even so, I will in this section look at some of the ways in which typical UMC objects do seem strange. 25 26

See Laycock, Words without Objects, Ch. 4. Koslicki, “Crooked Path,” pp. 125–28; Thomson, “Parthood and Identity across Time,” p. 213.

148

A third false friend of familiar objects: UMC For this will help us to focus on interesting examples of the “structural properties” by which typical UMC objects are characterized. One preliminary observation: by “typical” UMC objects I do not mean “statistically normal.” UMC objects, if real, are fathomlessly numerous, and there is no prospect of counting them. Rather what I mean by “typical” UMC objects are ones that stand out as outlandish, by virtue of contrasting so greatly with the familiar objects of common sense and empirical science. Almost always these will be UMC objects that comprise very many parts, and often the parts will be considerably separate from one another spatially. Some UMC objects derive their appearance of strangeness from comprising parts that intuitively seem to belong to different ontological categories – e.g. microparticles together with baseball stadiums together with parts of noses – and others are strange in ways that rest on their comprising parts that all belong to the same ontological category. So I shall include UMC objects of both these sorts among “typical” ones. One main way in which typical UMC objects seem strange is that they are extraordinarily durable. You can pulverize them and scatter their parts to the winds, and yet they will still exist. UMC objects composed wholly of individual atoms are indeed nearly indestructible, and have a past that is incalculably ancient. At the same time, many UMC objects are breathtakingly fragile. In the volume where common sense discerns a tree, for example, numerous UMC objects are spatially present – partially spatially present, to speak precisely – that will cease to exist at an objectively unpredictable point in time over the next few seconds. For these objects comprise, as a part, a hadron that will very soon undergo beta decay. These strangenesses may not show that UMC objects do not exist. But many suppose that they at least create obstacles for the claim that familiar objects can be “reduced” to UMC objects – the claim, in other words, that UMC objects are what the folk and the scientists “really are talking about” when they talk about dogs and desks and cells. For in that case the folk and the scientists would hold wildly inaccurate beliefs about the spatiotemporal careers, and the modal properties, of the objects they “really are talking about” – too wildly inaccurate for the claim of “reduction” to be plausible.27 If UMC 27

Wiggins, Sameness and Substance, pp. 30–34; Sidelle, “Sweater Unraveled,” pp. 430–33; Sanford, “Fusion Confusion.”

149

Familiar objects and their shadows objects do exist, these philosophers would hold, we must either be eliminativists with respect to familiar objects, or must hold that there is massive Coincidence between familiar objects and UMC objects (though see section 7.5). But eliminativism is widely considered to be a tenable view.28 And if the prospect of Coincidence raises difficult philosophical questions, UMC objects are not solely responsible for them: even if the tree before me does not coincide with a tree-plussoon-to-decay-nucleus, it still coincides (so many say) with a large “parcel” of cellulose, which again differs in its career and its modal properties from the tree.29 Many proponents of UMC do respond to the worry that UMC objects seem breathtakingly fragile: they recognize UMC objects that can continue to exist even if some of their current parts go out of existence. At the same time, they respond to the complaint that UMC objects are eerily durable: they recognize UMC objects that may cease to exist, even though their current parts go on existing. The response consists in recognizing diachronic mereological fusions, in addition to synchronic ones. These are the objects that I identified in chapter 1 as furnishing the most striking illustrations of explosivism with respect to persistence. Each such object is (to repeat) constituted by “an assignment,” a function from times to objects existing at that time. 30 A diachronic mereological fusion (“D-fusion,” to use Sider’s term) is then said to have, as its parts at each of these times, all and only those objects onto which the assignment maps at that time. Since the assignment can map different times onto different pluralities of 28

29 30

One family of eliminativist views holds that there are no objects in the world (Hawthorne and Cortens, “Towards Ontological Nihilism”), and that the world is “thingless” (Sidelle, “Sweater Unraveled”), though it does contain primordial “stuff ” (Sidelle, Necessity, Essence, and Individuation; Jubien, Ontology, Modality, and the Fallacy of Reference). Another family comprises views to the effect that there are only mereological simples or physical simples, and that composition does not occur (Unger, “I Do Not Exist,” p. 234; Wheeler, “On That Which Is Not,” p. 166; cf. Rosen and Dorr, “Composition as a Fiction”). Closely related is the view that composition does not occur because there is only one thing, “the Blobject” (Horgan and Potr č, “Blobjectivism and Indirect Correspondence”). More distantly related are views to the effect that there are only simples together with animals (van Inwagen, Material Beings), or only simples together with conscious agents (Merricks, Objects and Persons). Rea, ed., Material Constitution. Sider, Four-Dimensionalism, p. 133.

150

A third false friend of familiar objects: UMC objects, it follows that a D-fusion can have, at different times, different parts; since the assignment can map different times onto greater and lesser pluralities, the D-fusion can be said to comprise more parts at one time, fewer at another. The D-fusions of greatest interest for proponents of UMC are “minimal D-fusions,” ones that exist only at the times over which the assignment is defi ned. 31 For the natural way to extend the idea of universal mereological composition to diachronic fusions is to say: any temporal crosscutting yields an object; more precisely, corresponding to any assignment there really exists a minimal D-fusion. So far as the defi nition of a minimal D-fusion goes, and barring special argument to the contrary, a minimal D-fusion can even exist at discontinuous times. 32 The UMC objects we considered before can be spatially discontinuous, but not temporally discontinuous; these further UMC objects can be both spatially and temporally discontinuous. As we noted in chapter 1, diachronic UMC objects trace out strange courses of persistence. Consider (for a fi nal time) Lewis’s trout–turkey:33 the qualitative character of the later phases in this object’s existence appears neither to be causally regulated by, nor in any comprehensible way to grow out of, the qualitative character of the object’s earlier phases. The same is true of Balashov’s “writer– cucumber.”34 Most disturbingly, some diachronic UMC objects violate laws of physics. 35 7.3

T h e p r o p e r t i e s t h at c h a r ac t e r i z e U M C o bj e c t s

If UMC objects really exist, however strange they may be, what are they like – by what properties are they characterized? I begin this section by arguing that in “typical” circumstances – circumstances that bring out what is distinctive and strange about UMC objects – UMC objects are characterized only by so-called “structural properties.”36 31 32 33 34 35 36

Sider, Four-Dimensionalism, p. 133. Sider, Four-Dimensionalism, p. 136. Lewis, Parts of Classes, pp. 79–81. Balashov, “About Stage Universalism,” p. 30. Balashov, “On Vagueness,” pp. 527–28. Armstrong, World of States of Affairs, pp. 32–38.

151

Familiar objects and their shadows I then introduce the reasons for thinking that, while some “structural properties” are properties of a perfectly genuine sort, the “structural properties” that would characterize typical UMC objects are not properties at all. In the next section I offer a fuller and more sophisticated version of the case against the “structural properties” by which typical UMC objects would typically be characterized. We can readily imagine circumstances in which UMC objects would be characterized by a number of familiar properties – but the objects would have these properties only accidentally, and the circumstances would be decidedly atypical. Thus consider, to begin with, a UMC object of our original sort – a synchronic mereological fusion – and suppose that the parts of this object are individual atoms. If it just so happens that these atoms are spatially close to one another, so that collectively they form a compact, separately movable mass of matter, then the fusion of these atoms can be said to have a certain mass, a certain velocity, perhaps even a certain color and shape and elasticity – the fusion might be said to have just the properties of an orange toy ball. But these will all, except the fi rst, be only accidental properties of that synchronic fusion (see section 7.5, on “UMC and Mereological Essentialism,” the paragraph beginning “Too strong: …”). That very fusion can go on being itself even if the atoms become widely dispersed, and could even now have been itself if the atoms had before now been widely dispersed. Yet if the atoms – the parts of this UMC object – were widely dispersed, this UMC object would not be a possible object of sight or of touch, and so would have neither a color nor a shape nor an elasticity. Nor can it be said that a widely scattered plurality of atoms, each moving in its own way or standing at rest, collectively has a velocity. 37 Only aggregate mass would remain. And certainly dispersal is the more “typical” case. That is, what is distinctive and strange about synchronic UMC objects is, in large measure, that so often they are spatially discontinuous objects. Or consider a diachronic mereological fusion, all of whose elements again are atoms, that just happens to take up the very fourdimensional volume in which common sense would take there to be a dog. Then this diachronic UMC object might be said to have a particular gait or odor or even a particular bark or disposition. But 37

Lowe, “In Defense of Moderate-Sized Specimens of Dry Goods,” p. 707.

152

A third false friend of familiar objects: UMC all that is essential to this diachronic UMC object is that, at each of the moments it exists, it comprises just those atoms that it actually comprises, and hence just that those atoms exist. It does not matter where those atoms are, nor whether they are spatially associated with one another – that is the force of saying that the object is a diachronic mereological fusion (see section 7.5). So at each of the moments over which the assignment for this diachronic UMC object is defi ned, the atoms that compose it could have been widely dispersed. The gait, the odor, and the bark are accidental properties of the diachronic UMC object. They obtain, moreover, only thanks to a spatial association that is decidedly “atypical.” That is, what is distinctive and strange about mereological objects of any sort is precisely that they are mereological: their parts need not be arranged one way or another, but need only be, in order for the object to exist. Is it in fact right to say that the whole nature of a mereological object is exhausted by the mere existence of its individual parts – that in typical circumstances, a mereological object can claim, as a unitary whole, no qualitative character of its own? Apart from the single property of aggregate mass – more precisely, of aggregatemass-at-a-time – this very nearly is right. If mereological objects have, as wholes, distinctive qualitative characters at all – characters that can set one such object off from others – those characters will consist in having one or another “structural property.” A “structural property” is one that can be specified only by a conjunction, each conjunct of which depicts the property-bearer as having a part that meets a certain description. The standard example is the structural property that is characteristic of methane molecules. 38 A methane molecule is characterized by: having one part that is a carbon atom; having another part that is a hydrogen atom and that shares an electron with a carbon atom; having another part that is a hydrogen atom that shares an electron with a carbon atom; having yet another hydrogen-atom part that shares an electron with a carbon atom; and having yet another again. This same structural property can be specified in a slightly different way, namely by spotlighting, in each of five conjuncts, just the intrinsic character of each of the five atoms in the methane molecule, then using a sixth conjunct to indicate the relations obtaining among the five atoms. The alternative formulation 38

Armstrong, World of States of Affairs, pp. 32–38.

153

Familiar objects and their shadows is important, since it provides a template for identifying structural properties that will characterize mereological objects across typical circumstances as well as atypical ones. For the nature of a mereological object requires no particular relations among that object’s parts: the object’s existence requires only that each of the individual parts exists; hence the specification of a structural property that reliably characterizes a mereological object will be like the second formulation of “the methane property,” but without the sixth conjunct. Every conjunct in this specification will say only what some one part of the mereological object is like intrinsically. Mereological objects may accidentally bear richer structural properties, such that specifying these properties requires indicating, in one or more of the conjuncts, relations among the parts. But the structural property essential to any UMC object – the property that characterizes that UMC object’s very nature – will be one that incorporates no particular relations among the parts of the object. My contention, as I indicated in the introduction to this chapter, is that some so-called “structural properties” are not genuine properties at all. One might wonder what I could possibly mean by such a claim – how could “a so-called property” fail to be a property? But on almost everyone’s intuitions, there are some locutions of the form “the property of being …” that fail to pick out a genuine property. “The property of being such that Des Moines is in Iowa” is a fairly uncontroversial example. What I am claiming is that properties that are genuine in the sense of being ways for things to be – of constituting answers to the question of what something is like – do not include certain so-called “properties,” and in particular not the “structural properties” that characterize typical mereological objects. I do not rest this claim on intuition. Rather I rest it on the position, found both in Aristotle and Hegel, that any genuine property contrasts with its own proper contraries. 39 One way of putting this position is that the very identity of a genuine property consists (at least partly) in its contrasting, sharply or mildly, with certain other properties. (For Hegel’s defense of this position, see chapter 8.) Another formulation focuses on property-bearers: it says that what it is for a thing to have a certain genuine property is (at least in part) for it thereby to 39

Aristotle, Physics Bk. i , Ch. 5; Bk. v, Ch. 1 and Ch. 5; Hegel, Logic of Hegel, §§89–98; Hegel, Science of Logic , pp. 109–37 and 600–22.

154

A third false friend of familiar objects: UMC differ, to greater and lesser degrees, from things bearing certain other properties. The key idea, on either formulation, is that for there even to be properties, there must be contrariety. Contrariety is like incompatibility, but involves more. The property of having an adult weight of 25 to 30 pounds is incompatible with the property of having an adult weight of 3–5 pounds, but is also incompatible with the property of having an index of refraction of 1.414; only an animal can have an adult weight of 25 to 30 pounds, and no animal can be a liquid (or a portion of a liquid). Yet it seems to lie in the very nature of having an adult weight of 25 to 30 pounds that it excludes, and contrasts with, having an adult weight of 3–5 pounds. It seems not to lie in the very nature of that property that it excludes having an index of refraction of 1.414. One way of unpacking this thought, as Bigelow and Pargetter have shown,40 is that having an adult weight of 25 to 30 pounds contrasts to different but commensurable degrees with other adult body weights: having an adult weight of 25 to 30 pounds contrasts more sharply with having an adult weight of 3–5 pounds than with having an adult weight of 10–15 pounds. But the difference between having an adult weight of 25 to 30 pounds and having an index of refraction of 1.414 is neither greater nor lesser than either of these differences; it is incommensurable with those differences, being a complete lack of connection. That is what it means to say that for the property of having an adult weight of 25 to 30 pounds, other adult body weights (or weight-ranges) count as “its own proper contraries.” The question whether “structural properties” are genuine properties is, from this standpoint, the question whether a typical “structural property” has its own proper contraries – whether there is a plurality of other properties, with which it contrasts to greater and lesser (but commensurable) extents. The answer pretty clearly seems to be Yes, at least in the case of some structural properties. Consider, for example, “the methane property.” Whether the methane property is instantiated depends, not just on the intrinsic character of each of five atoms, but also on their being atomically bonded so as to form a molecule. So long as we help ourselves to the assumption that variants on the methane property must likewise be properties of unitary molecules, it will seem easy to envision graded contrasts with the 40

Bigelow and Pargetter, Science and Necessity, pp. 53–62.

155

Familiar objects and their shadows methane property itself. The methane property will differ somewhat from the property that characterizes a hypothetical molecule having a silicon atom in place of one of the hydrogen atoms, and will differ more from the property that characterizes a variant molecule with one germanium atom at that spot instead. A molecule that replaces one hydrogen atom with a lithium atom will differ somewhat from a molecule bearing the methane property, and a molecule that replaces two hydrogen atoms with two lithium atoms will differ more.41 So long as we regiment the available dimensions of difference by considering only alternative forms of molecular composition, there may seem to be a well-defi ned phenomenon of “greater and lesser difference from” the methane property. (In so regimenting the dimensions, we will be heeding shadows, cast onto the domain in which the methane property is instantiated, by the objects in which molecular chemistry believes.) There will be questions as to how to sum together differences from the methane property on the dimension “number of nonhydrogen parts” with differences on the dimension “how different from hydrogen a given nonhydrogen part is.” But there is no principled obstacle to measuring overall difference between properties that depart from one another along more than one dimension. The earlier example of colors demonstrates this. Colors contrast with one another on three different dimensions, namely hue, saturation, and brightness. Yet for all that there is a fairly clear measure of resultant, overall difference between colors as unitary wholes: the measure is distance within the Munsell color solid. But now note that the structural properties that would essentially characterize typical UMC objects would not be like colors, and would not even be quite like the methane property. They would be like the methane property on the second formulation, but without the sixth conjunct – mere lists of parts, with no requirement on how the parts are related. What form would contrasts or differences between such structural properties assume? We can certainly envision cases in which one such property features an A-like part at its fi fth conjunct, and another such property, just like the fi rst one up through the fi rst four conjuncts, features a B-like part at its fi fth conjunct. But would the contrast here be just between the fi fth conjuncts themselves, or would it amount to an overall contrast between the 41

Wells, Structural Inorganic Chemistry.

156

A third false friend of familiar objects: UMC “host” structural properties as unitary wholes? One way to see the force of this question is to move just slightly in the direction of the “mere list” structural properties appropriate to UMC objects: consider an amplified version of the methane property, as it appears on the fi rst formulation, to which we have added a sixth conjunct that reads, “and contains the Eiffel Tower.” And now consider a number of variant structural properties that differ from “amplified methane” strictly in respect of this sixth conjunct. One reads “… and contains the Panthéon,” another “… and contains the Hôtel Biron,” another “… and contains a worm in Danielson, Connecticut,” a fourth “… and contains a hamburger in Weehawken.” With respect to the sixth conjunct, we seem to have greater contrasts, as we get to the end of this list, than we had at the start of the list. But are these greater contrasts contrasts between whole unitary structural properties, or just between sixth conjuncts? Certainly no difference whatever in how something acts is entailed by whether the structural property with “… and contains the Panthéon” is instantiated, or whether the one with “… and contains a hamburger in Weehawken” is instantiated instead: in either case you have a methane molecule acting like a methane molecule, and both the Panthéon and the hamburger sitting there like themselves. Differences with respect to color are differences with respect to how some unitary thing looks. Differences with respect to these “mere list” structural properties are not differences in how any unitary thing acts, not even some unitary but very large and spatially discontinuous thing. The problem with “amplified methane” is that the sixth conjunct has nothing to do with the fi rst five. That is why divergence with respect to the sixth conjunct amounts to divergence only with respect to the sixth conjunct, and not divergence from the structural property as a whole. One could replicate the very same problem with colors, by constructing a suitable conjunctive – though non structural – property. Thus one might start with Red 16, and specify it as a threefold conjunctive property: being located at such and such a point on the hue axis of the Munsell color solid , at such and such a point on the brightness axis, and at such and such a point on the saturation axis. Then one would add an unrelated fourth conjunct, such as “is located at longitude 10°43′0″ East.” Divergences from an object bearing this conjunctive property, with respect to any of the fi rst three conjuncts, would make for an overall difference in how the 157

Familiar objects and their shadows bearer of that divergent property looks. Divergences with respect to the fourth conjunct would not make for any difference in how anything looks or in how anything acts. But in the “mere list” structural properties that essentially characterize typical UMC objects, none of the conjuncts has anything to do with any of the others. Consequently there is little prospect of making out the claim that “mere list” structural properties can, as unitary wholes, differ to greater and lesser degrees from other structural properties. And if they cannot, it at least appears to follow that such properties do not, in their own right, have contraries of their own. It would follow that they are not really properties at all: that they are, precisely, mere lists of separate objects. 7. 4

W h y c o n t r a r i e t y fa i l s t o o b ta i n

But someone could charge that the above discussion “stacks the deck” against the structural properties that would characterize typical UMC objects, and that in two ways. For one thing, we have so far considered only the structural properties that would essentially characterize UMC objects, and these do indeed incorporate no requirements on relations among the parts of the UMC object. But even typical UMC objects – ones different from a synchronic fusion of atoms which just happens to have the color and shape of an orange ball, or from a synchronic fusion of atoms which just happens to be found exactly where common sense discerns a dog – will, at any given time, accidentally be such that their parts stand in some spatiotemporal relations to each other. Typical UMC objects will be characterized, albeit accidentally, by structural properties that are not “mere lists.” Second, there may be something biased in the suggestion that if the structural properties that characterize UMC objects (whether essentially or accidentally) are to contrast with their own proper contraries, they must so contrast as unitary wholes. Perhaps it is just in the nature of these properties that they are sums of disjointed ways that disjointed things independently are. If so, contrasts obtaining between such properties are bound to be contrasts that obtain only between isolated conjuncts. In this section, then, we will consider properties that characterize UMC objects both essentially and merely accidentally. We will focus on the properties of typical UMC objects – objects that incorporate a 158

A third false friend of familiar objects: UMC vast number of widely scattered parts, in some cases parts that intuitively seem to belong to a diverse array of different ontological categories. But we will allow that one such property can contrast with another such property just by virtue of containing a conjunct that differs from a conjunct in that other property. Since the structural properties that characterize typical UMC objects incorporate many, many conjuncts – thousands, let us say – it will follow that one such structural property can contrast with other structural properties on thousands of different dimensions. What I will argue, however, is that this still will not be enough to show that there is contrariety among the structural properties that characterize typical UMC objects. Contrariety obtains only where a property differs from other properties to greater and lesser, but commensurable, degrees: there must be a well-defi ned phenomenon of “more and less different from.” But the diversity of dimensions on which these structural properties diverge from one another keeps this phenomenon from obtaining, I shall argue. My argument here parallels the arguments I gave in sections 5.4 and 6.3 concerning the structural properties that would characterize massive pluralities of microparticles – the pluralities to which causal exclusionists want to assign all the causal efficacy that common sense assigns to beliefs, desires, flying rocks, and barking dogs. But here I focus on differences among, specifically, properties that would characterize typical UMC objects. Let S1 be a structural property that characterizes a large and spatially discontinuous UMC object, all the parts of which are fundamental microparticles of physics. Each conjunct in S1 attributes, to some one microparticle, a particular charge or color or velocity or excitation state. Then compare S1 with two other structural properties. S2 . Each of its fi rst 1,000 conjuncts attributes to some one microparticle a feature that is slightly different, or in some cases moderately different, from the feature attributed by the corresponding conjunct in S1. S3. Each of its fi rst 5,000 conjuncts attributes to a microparticle a feature that is slightly different from the feature attributed by the corresponding conjunct in S1.

At fi rst blush, S3 appears to be more different from S1 than S2 is. But what if S2 further differs from S1 in the way that S4 does? 159

Familiar objects and their shadows S4. Each of its fi rst 1,000 conjuncts is exactly the same as the corresponding conjunct in S1, but S4 incorporates only 50 more conjuncts beyond these – S4 is a far less complex structural property than S1.

In that case it seems that S2 will be more different from S1 than S3 is. However, it all depends on whether S3 itself differs from S1 in the way that S5 does. S5. It is conjunct for conjunct the same as S1 up through the fi rst 3,000 conjuncts, but it incorporates 18,000 more conjuncts than S1 does – S5 is a vastly more complex structural property than S1 is.

It seems plausible that if S3 itself is similarly complex, it will after all be more different from S1 than S2 is, even if S2 is as (relatively) simple as S4 is. But what shall we say about S5 itself? If S3 is as complex as S5, S5 will be more like S1 than S3 is. But is S5 more like S1 than S2 is, or less? The same sort of question is posed (but from the other direction) by S6. S6. It is conjunct for conjunct the same as S1 up through its fi rst 50 conjuncts, and these are the only conjuncts that it has.

Is S6 more like S1 than S2 is, or less? I said that in this section we would consider also some structural properties that are accidental to UMC objects. Here is an example: S7. It is conjunct for conjunct the same as S1 up through the fi rst 4,000 conjuncts, but further conjuncts specify that each of the fi rst 4,000 conjuncts is satisfied at a point 1,000 miles removed from all points at which any other of the fi rst 4,000 is satisfied.

S7 is not a “mere list” structural property: it also specifies a relation that obtains between the objects described by many of its conjuncts. Does this peculiarity of S7 make it more like S1 than it would otherwise be, or less like S1, than it would otherwise be? S7 is a property that could accidentally characterize either a synchronic fusion of microparticles, or a diachronic fusion. S8, in contrast, can characterize only a diachronic fusion. S8. It is conjunct for conjunct the same as S1 up through the fi rst 4,000 conjuncts, but further conjuncts specify that each of the fi rst 4,000 conjuncts is satisfied at a time 500 years removed from all times at which any other of the fi rst 4,000 is satisfied.

160

A third false friend of familiar objects: UMC It would be wrong to say that S8 is an accidental property of some diachronic fusion. Any diachronic fusion is constituted by its associated “assignment,” and so it is not the case that the parts of any one diachronic fusion could have been located at times other than those at which they actually are located. But S8, like S7, does specify a relation between the objects described by many of its conjuncts. Is S8 more like S1 than S7, or less? How does its divergence from S1 compare with the divergences from S1 of S2 –S6? We must also consider divergences from S1 that rest on a structural property’s incorporating objects that seem to belong to different ontological categories from microparticles. S9. Each of its fi rst 1,025 conjuncts is exactly the same as the corresponding conjunct in S1. S9 incorporates 25 more conjuncts besides, and these are representative examples of the remaining 25: “and contains a part that has a boiling point of 202° F”; “and contains a part that requires on average 8.4 person–hours to produce”; “and contains a part that has a lifetime batting average of .284.” S10. Its fi rst 15 conjuncts match exactly the fi rst 15 conjuncts in S1. It contains 25 more conjuncts beyond these, and these are representative samples: “contains a part that has a height of over 10′ when mature”; “contains a part that is carnivorous”; “contains a part that is highly resistant to penicillin.”

Is S10 more like S1 than S9, or less? Intuitively, it seems that S10 features “the wrong sort of conjunct” at just 25 places – and precisely the same is true of S9. Is it decisive that there are proportionately more “wrong conjuncts” in S10 than in S9? If so, what does that say about the relation of S5 to S1? What this extended illustration shows is that there is systematic interference among the various dimensions along which variants on a structural property Sn, characteristic of a typical UMC object, can differ from Sn itself. The dimensions do not relate additively to one another. The consequence is that relative distances from Sn, along any dimension that appears to afford an ordering of “more and less different” for structural properties that are variants on Sn, will not in general amount, in and of themselves, to degrees of difference for those variant structural properties from Sn itself. Here is the general conclusion I draw concerning UMC objects. Initially, it appears that the typical UMC object is essentially 161

Familiar objects and their shadows characterized by a structural property of one sort, and accidentally characterized by various structural properties of a different sort. It appears that the typical UMC object is essentially characterized by a “mere list” structural property, and accidentally characterized by structural properties of a richer sort – properties the instantiation of which requires the obtaining of particular relations among the parts of the UMC object. But what I have argued, in this section, is that these putative accidental properties are not properties at all, since contrariety is not defi ned over them. In the previous section, I argued that contrariety is not defi ned over “mere list” structural properties either, and hence that they too are not genuine properties. I conclude that typical UMC objects have virtually no character – no qualitative make-up – at all. The typical UMC object is not essentially characterized by any property at all: there is no way that it is by nature, nothing that it is by nature like. It has virtually no accidental properties either; the sole exception would be aggregate mass. Entities that are this close to being literally nondescript have, I suggest, no place in serious ontology. 7.5

UMC a n d m e r eologica l e sse n t i a lism

This chapter has considered the thesis that, in addition to however many familiar objects there may be in the world – dogs, noses, and desks, for example – there are also UMC objects. It has construed UMC objects as objects such that necessarily, they exist exactly when and where their actual parts exist – as objects for which mereological essentialism holds true. It has taken the prime reason for believing in UMC objects to be an argument that Sider adapts from Lewis. But philosophers who take more leaves from Lewis’s book than just this single argument will be inclined to say that this chapter has given a confused account of UMC objects – an account that is in one way too weak, and in another too strong. This section is a postscript addressed to the alternative account that these philosophers would favor. Too weak: one might hold that the Sider/Lewis argument shows that the only composite objects are mereological fusions; in other words, that the only form of composition is mereological composition. Many philosophers may think that if many microparticles are in spatial proximity to one another, or are causally integrated with one another, or together form a separately movable mass of matter, 162

A third false friend of familiar objects: UMC then these factors – individually or in combination – entail that the microparticles together compose something. But there can be borderline cases of spatial proximity, of causal integration, and of separate movability. Hence if these factors did suffice for composition, there could be borderline cases of composition. P3, in the Sider/Lewis argument, would be violated. Perhaps then there is only one factor sufficient, and only one factor required, for a plurality of microparticles – or a plurality of anything else – to compose something. This factor is that the elements of that plurality exist. If many objects exist, then thereby, automatically, they jointly compose something – and the existence of these objects is all that can make for composition. Too strong: this chapter has held that mereological essentialism holds true for any UMC object. But philosophers who subscribe to Lewis’s counterpart theory will contest this claim. Might the very dog that stands before us – this very composite object – have existed, even if some of the microparticles that actually compose the dog had not existed at all? Counterpart theory seems to say: in the right conversational setting, the answer might well be yes. We consider a close possible world in which a few of the microparticles that actually compose this dog do not exist at all – or to speak more precisely, in which for a few of these microparticles, there are no counterpart microparticles. Even so, it could well be that a great many of the microparticles that actually compose this dog do have counterparts in that world, and that the counterpart microparticles are in close spatial proximity, are causally integrated with one another, and collectively move as a cohesive whole. As we envision this plurality of microparticles, we seem to see a dog that runs around and behaves much as the actual dog behaves. Then depending on the conversational setting, we might well be inclined to say that these microparticles together compose something. Indeed the something that these microparticles compose is so similar to the actual dog that we are inclined to say that this composite object “stands in” for the actual dog, and is our dog’s counterpart. Presto: our dog really exists in a possible world in which some of its counterpart microparticles do not exist at all. Mereological essentialism , in this conversational setting, fails. A similar scenario shows, these philosophers might argue, that it is not a necessary truth that our dog exists exactly where his actual component microparticles are found. Suppose that all the microparticles 163

Familiar objects and their shadows that compose the actual dog have counterparts, in some other close world, but that some of them are located on the moon. The vast majority of the counterpart microparticles are causally integrated with one another in just the way that the microparticles in the actual dog are causally integrated, and cohere as a separately movable (tailwagging, barking) mass, just as do the microparticles in the actual dog. Then we might say (depending on the conversational setting) that these connected microparticles compose a dog, and indeed compose the very counterpart to our actual dog. Presto: it is possible that even though the microparticles actually composing our dog were somewhat scattered, our dog himself would exist only on earth and not partly on the moon – would exist as a spatiotemporally continuous object. I concede to these philosophers that we might in fact make such judgements about counterparts to our actual dog. But would these be the right judgements? Suppose we were to consider these close possible worlds in the light of the lessons that these philosophers draw from the Sider/Lewis argument. This does seem appropriate: if those are true lessons, then it cannot hurt to bear them in mind, if we seek to make true judgements about this dog’s counterparts. Then we will judge that even if, in each of the possible worlds, certain microparticles are causally integrated and cohesively stuck together, forming a movable object that looks and sounds just like our actual dog, all that has nothing whatever to do with whether these microparticles between them compose anything. All that is relevant, to whether many microparticles between them compose something, is whether those many microparticles exist, we will remind ourselves. (In reminding ourselves of this, we will be shining a fl ashlight so as to dispel any shadows cast by nonexistent forms of composition.) All the same, we will indeed judge that there is a composite object located just where we have imagined ourselves envisioning a dog. It will be no more genuine than countless other composite objects located in those worlds, most of which are discontinuous, and many of which overlap the location of the envisioned dog. But yes, it will really exist. Yet should we judge that this composite object is the same composite object as our actual dog – more cautiously, that it is the counterpart to that actual composite object that is our actual dog? If all that renders our actual dog a genuine composite object is that all of 164

A third false friend of familiar objects: UMC his component microparticles exist, then it is hard to see how, in a world in which not all these component microparticles exist (i.e., have counterparts), we could have the same composite object as in the actual world (i.e., a true counterpart to the actual composite object). And what of the world in which some of the microparticle-counterparts are located on the moon? If all that renders our actual dog a genuine composite object is that all of his component microparticles exist, then it seems that the composite object in that world, that is the counterpart to our actual dog, is located partly on the moon. The point here goes back to the way the Lewis/Sider argument is set up. Its topic is: can composition be restricted ? The question is not whether composition can be extended – from familiar things to UMC objects. The starting presumption is that composition is found everywhere, in every plurality of objects, and the question is whether that presumption can be defeated. Why does that seem a reasonable starting presumption? Because UMC objects appear to be “an ontological free lunch.” The existence of any such object seems just to be the existence of its parts. But so long as we cling clearly to this starting idea, it will be very hard to argue that such an object can exist despite the nonexistence of some of its parts – or that it can exist in locations other than those occupied by its parts.

165

8 Concluding Hegelian postscript

8.1

H ege l a n d con t e m por a ry sce p t icism a b ou t fa m i l i a r o bj e c t s

Contemporary metaphysicians are deeply sceptical of the familiar objects recognized by common sense and by many empirical sciences. What explains the scepticism? I shall begin this chapter by suggesting that contemporary metaphysics is dominated by the style of thought which Hegel – using the nineteenth-century vocabulary of faculties – called “the Understanding,” and that “the Understanding” is constitutionally antipathetic to familiar objects. But fi rst a few words about the style of thought that fi nds familiar objects congenial – the style of thought which Hegel identified under the title “Reason.” A prime characteristic of “Reason” is that it is willing to recognize what Hegel called “identity in difference.”1 “Identity in difference” is a form of sameness which articulates itself in difference. One example is the sort of persistence that seems to characterize familiar objects. Typically, a familiar object goes on being itself while passing through different phases or properties, that is, while differing from itself. Indeed in many cases – and especially if we count such properties as age among the relevant ones – a familiar object can go on being numerically the same object only by differing more and more from its earlier self. Another example is the relation of a familiar object to its parts. One and the same familiar object comprises different parts at different times; in the particular case of organisms, a familiar object can go on being itself only by expelling former parts and adding new ones. It is the same composite only because it is differently 1

All references to Hegel’s logic in this chapter (but one) are to Hegel, Logic of Hegel (1817), the version of the Logic that appeared in The Encyclopedia of the Philosophical Sciences, often called the Encyclopedia Logic. From here on I use the acronym EL. For the concept of “identity in difference,” see EL, §161.

166

Concluding Hegelian postscript composed. Then too there is a point about the properties by which familiar objects, and samples of familiar kinds of stuff, are characterized. Common sense is quick to agree that these properties fall into contrary ranges, each contrasting to graded degrees with its own proper rivals. The Hegelian claim – on which I shall focus at length in the later parts of this chapter – is that for any genuine property, its having the intrinsic positive character that it has just is (at least in part) its contrasting as it does with its own proper rivals. Its being just that property is its differing, in just those ways, from just those other properties. The rival style of thought, “the Understanding,” insists on recognizing only identities that are internally unfaceted , or (using Hegel’s term) “abstract.”2 Being-the-same, for the Understanding, excludes being different. This sets up, for the Understanding, a particular ontological agenda. “No entity without identity,” said Quine: and the Understanding recognizes only entities whose self-sameness is monochromatic and unvariegated. If there are persisting entities, they are entities that persist for so short a time that they cannot differ from themselves – temporal stages, which are shorter than any possible physical change. Composite entities must be regarded with scepticism, if their principle of composition allows for differences in the parts of which they are composed. So where common sense and empirical science see composite entities that undergo changes in their parts, the Understanding insists that there are only many physical simples, which at best are accidentally characterized by certain patterns of arrangement – dogwise arrangement or treewise arrangement or deskwise arrangement. Composition can indeed be allowed so long as asserting it says nothing more than that the many simple parts each exist. The mereological sums yielded by this principle of composition will indeed have the further merit, for the Understanding, that they are compositionally brittle: this mereological sum can be composed only of these parts, and can brook no difference in its composition. And which entities can the Understanding enlist to explain the causation apparently exercised by the messy difference-admitting entities in which common sense believes? Causal efficacy must get awarded 2

EL, §80. For the phrase “the abstract Identity of the understanding,” see EL, §115; see also the Zusatz to §115, and §116 and Zusatz. See also Hegel, Science of Logic , pp. 412–14.

167

Familiar objects and their shadows to mereological sums of physical simples. Each such sum will cause something, if it does cause anything, in virtue of a property that it has – a complex structural property. But since the Understanding draws no connection between the intrinsic positive character of such a structural property, and that property’s contrasting to graded degrees with its own proper contraries, the Understanding can overlook the fact that contrariety seems simply not to be defi ned over such properties. If this entails that such properties cannot instantiate invariances – and hence cannot, on the view of causation that I advocate, be properties in virtue of which a sum of simples causes anything – then the Understanding can always fall back on the device of tacitly adverting to shadows cast by the difference-admitting entities whose reality it disputes. Now Hegel did not, I (of course) admit, talk about exdurantism or mereological composition or causal invariances. That is precisely my point. He identified a style of thought, which he considered inexpugnable from human cognition, that believes only in unfaceted identities. There really is such a style of thought. The proof is that philosophers who have no inkling of proceeding from a Hegelian script take exactly those positions in metaphysics that the Understanding would favor. They take these positions in the face of obvious objections, technical difficulties, and incredulous stares. 8. 2 H e g e l o n p r o p e r t y- i d e n t i t y: t h e c ruc i a l p l ac e o f g r a d e d c o n t r a s t w i t h c o n t r a r i e s In chapters 5 –7 I trade heavily on a position on the nature of properties which Hegel articulates in the Logic. That position is that any property by nature contrasts with its own proper rivals to differing degrees – in the central case, to degrees that are quantitatively identifi able. In the present section I set forth the reasoning by which Hegel supports this position. In the next section I consider objections against the position – objections drawn from empirical science. The support for this key position on quality-identity occupies most of the fi rst part of the Logic, “The Doctrine of Being.” But before I enter into the details I should offer some very brief and rough remarks about the Logic as a whole. The Logic, almost all commentators agree, is a critical examination of a series of categories or concepts, beginning with very simple ones and ending with very 168

Concluding Hegelian postscript complex ones. Now it is a large and controversial question, in our day as it was in Hegel’s, just how much we can learn about the world by analyzing our concepts. But it seems hard to deny that it could be salutary, for philosophical practice, to learn that the contents we typically report our concepts to have – the ways we say they depict their targets – are incompatible with the use we make of those concepts. The main business of the Logic as a whole is to advance just such a contention. We seem to conceive of key elements of the world – its substances, its forces, its natural kinds, its laws – as independent, in both their nature and existence, from other elements both within and without their own ontological category. We seem to conceive each such element as essentially characterized by its own positive intrinsic character. But if one looks to the ways we use these concepts, Hegel argues – the ways we do use them and must use them – one sees that our initial reports about their contents are not entirely accurate. We do and must use these concepts of the world’s key elements as designating role-players in a complex, telos-inspired project. The roles are interdependent: what it is for substances to be substances, for kinds to be kinds, for laws to be laws, cannot be separated from there being in the world elements in the other ontological categories – including, ultimately, communal cognitive practices. And what is essential to the contents of each ontological category is something relational, namely the ways they function together with the contents of the other ontological categories in embodying and carrying out the complex, telos-inspired project. The world’s substances and forces, laws and kinds, could have had a different intrinsic character; they have some intrinsic character because it is only in virtue of this character that they serve their roles. The stage in this argument that is important for this book is the one that says: we imagine that our concepts of properties depict them as independent, both in their character and their existence, from other properties; but if one looks to the only stable use that can be made of the concept of a property, one sees we tacitly, but inevitably, treat properties as essentially “excluding” other properties – as each contrasting, to graded degrees, with its own proper contraries. This message is far less heady than the ultimate claim of the Logic, namely that we tacitly treat every element of the world as a role-player in articulating and achieving the life of Absolute Spirit. But I am happy to be confi ning my remarks to this more humble claim. I think that 169

Familiar objects and their shadows Hegel’s arguments for it are easier to follow, and more likely actually to succeed in establishing their conclusions – if not in the exact way Hegel puts them, then in mildly revised versions that are easy to suggest. To get a clear picture of the reasoning Hegel offers for his distinctive position on property identity, we must begin at the beginning of the Logic, with the much-discussed transition from Being to Nothing. First, a word on what “Being” is supposed to be. Hegel more than once appears to think that the logical progression traced by the Logic – in which the use made of ostensibly simple concepts in fact reveals that we are operating with concepts more complex than we give ourselves credit for – was reflected in the historical progression traced by Western philosophy. Certainly this is what he is thinking in the case of “Being”: Being is the same as the entity or element that Parmenides discussed, the entity or element which (in Waterfield’s translation)3 Parmenides called “what-is.” This is an entity, a “something,” whose core constitutive feature is just its existing. (In my own undergraduate lectures I call this entity “the Existent,” on the model of “the Just” or “the Equal” or “the Beautiful.”) The outcome of the argument that Parmenides makes about Being – an argument that can be pieced together from Plato and Aristotle, and from the fi rst part of Parmenides’ own poem – is that the only feature of Being is its existing. Hegel’s way of putting the point is that the only legitimate employment of the concept of Being is in the assertion “Being is.”4 But such an assertion can genuinely say something, Hegel points out, only if it manages to rule out some alternative.5 What might that alternative be? One suggestion is that it rules out the assertion “Being is not” (or, as I would prefer to say, “the Existent is not”). But since the whole constitutive character of Being consists in its existing, “Being is not” does not make sense, and we still have not specified a substantive assertion which “Being is” rules out. The prospects are better if we consider a variant on the subject, not on the predicate. The Existent (even Parmenides said this) is different from the Nonexistent: Being is different from Nothing, in Hegel’s terms. So perhaps “Being is” 3 4 5

Waterfield, First Philosophers, pp. 59–61. EL, §86. EL, §87 and Zusatz.

170

Concluding Hegelian postscript rules out “Nothing is.” But just what would it be for Being to be, rather than for Nothing to be? The concept of Being is empty except for the bare feature of being – which is precisely what we would have if “Nothing is” were true instead. As Hegel puts it, the claim that “Being is” converts to the claim that “Nothing is,” and vice versa.6 The situation can be remedied by seeing that tacitly, what we mean to be discussing in talking about Being is something the existence of which involves defi nite content. That is, we use “Being is” as excluding some genuine alternative, and thus tacitly treat Being as amounting to the category of Determinate Being 7 – or, as Hegel also says, the category of Quality.8 (In Hegel’s language, Determinate Being counts as an “overcoming” of Being and its attendant emptiness.9) To assert that some quality obtains is genuinely to assert something precisely because any given quality has by nature alternative qualities opposing it, the occurrence of which is ruled out by any assertion that the given quality obtains.10 Thus, to assert that red obtains is meaningful just because that assertion rules out an occurrence of brown or of green; an assertion that cold obtains rules out an occurrence of hot and of tepid. To put it differently, any instance of Determinate Being stands in contrary opposition to several rivals, and not just in contradictory opposition to one rival. If what redness excluded had nothing more to its nature than just being-excluded-by-redness – which would be the case if redness’s only rival were “nonredness,” “the nonred” – then we would be back at the same situation as with Being and Nothing. That is, if all that redness excluded were such 6 7 8 9

10

EL, §87. EL, §89. EL, §90. Hegel fi nds it convenient, in discussing particular qualities, to speak of items which have those qualities; yet he does not at this point wish to introduce the concept of a substance or “thing,” which could have more than one quality, or lose and gain different qualities. He wishes only to have some grammatical substantive for each property. Hence he employs the term “Etwas ,” which Wallace (faithfully to the German) translates as both “Something” and “Somewhat.” “A Something,” Hegel comments, “is what it is in virtue of its quality, and losing its quality it ceases to be what it is” (EL, Zusatz to §90) – a point to which Hegel returns in his subsequent discussion of the “Thing,” where he mentions “the ‘Somewhat’, which is still directly identical with its quality. Somewhat is what it is only by its quality …” (EL, Zusatz to §125). EL, §91 and Zusatz.

171

Familiar objects and their shadows a self-effacing and insubstantial rival as that, redness would be as emptily positive as Being is. But Determinate Being does constitute a step forward from Being and Nothing. It does so because instances of it contrast variously with different positive properties – with contraries. Exclusion has advanced from contradiction to contrast. As Hegel puts it, “Negation is no longer an abstract nothing, but, as a determinate being and somewhat, is only a form of such being – it is as Otherness.”11 But the explanation why concepts of specific qualities can have meaningful use while the concept of Being cannot raises a new question. Meaningful assertion that one quality obtains is possible only if this assertion can be treated as an exclusion; and an exclusion can meaningfully be discerned only where anything that is excluded is a genuine quality in its own right.12 But that the latter quality is a genuine quality in its own right entails that its occurrence can, in turn, be treated as the exclusion of yet another quality. Hence, meaningful use of the concept of any one quality requires specifiability, not just one or a few “opponents” – “opponents” which provide opposition to the given quality – but of “opponents” forming an endless series, in which each “opponent” provides opposition to the previous quality. (This point is expressed by Hegel by the image of an “infi nite progression.” The concept of a given quality can meaningfully be used only where logical room is specified for an endless series of qualitative alterations; or, as Hegel metaphorically puts it, one quality becomes another, which becomes another, etc.13) The question which this observation raises is just how it is possible that successive qualitative “opponents” should be, in this way, specifi able ad infi nitum. The answer is that what particular quality comes next in 11 12 13

EL, §91. EL, §92 and Zusatz. Hegel’s actual words are not that one quality becomes another, but that any qualitative “something” or “somewhat” – a term which, as I indicated in note 9 of this chapter, Hegel intends as simply a noun form of a quality – “becomes an other: this other is itself somewhat: Therefore it likewise becomes another, and so on ad infi nitum” (EL, §93). The view as thus formulated might seem somewhat less awkward than a view that one quality becomes another – as I report Hegel to maintain – since a “something” might seem better able to retain its identity while undergoing such an alteration, than a “quality” would. But in fact a “something” in Hegel’s sense can as little alter, while retaining its identity, as can a “quality”: consider the remarks quoted in note 9.

172

Concluding Hegelian postscript such a series always is, at any point, predictable or computable. What comes next in such a series always is in a sense a further embodiment of qualitative opposition or contrast. It is, in other words, because we operate not only with the concept of this or that particular quality, but also with the concept of a universal common to all these particular qualities, that this series is for us specifi able ad infi nitum.14 And, hence, it is just because we operate with the concept of such a universal that concepts of particular qualities can have for us meaningful use. (Hegel could more persuasively have arrived at this conclusion by arguing that with regard to any quality-space, we must always be prepared to recognize new individual qualities that lie between, or at the extreme fl ank of, other qualities in the space. It is far less plausible to hold that in order to recognize any individual qualities we must recognize infi nitely many qualities, as Hegel appears to claim.) Indication of the role played by this concept of a universal common to particular qualities is the chief concern of Hegel’s discussion of Determinate Being, that is, Quality; the concept is also of importance for the idea that any quality stands in graded and quantifi able contrasts to its contraries, since in the further discussion of this concept 14

The concept thus introduced – as being what enables us to specify particular qualitative opponents ad libitum – is called by Hegel “Being-for-sel f ” (see following paragraph). Yet despite the fact that Being-for-self is introduced in this way, some interpreters regard Being-for-self as simply a diff erent concept from that of the particular qualitative feature (or “Something”), rather than as the universal of particular qualitative features. This interpretation assorts poorly with the text. Hegel generally uses “idea” in a Platonic sense, i.e., as meaning a universal rather than a minddependent abstraction – see EL, §213 and Zusatz, and cf. Zusatz to §143 – and hence he is saying that Being-for-self is the fi rst universal considered in the Logic when he writes, in EL §95, “[i]n Being-for-self enters the category of Ideality.” Further, that Being-for-self is a universal seems pretty clear from the passage quoted in the following paragraph in my text. But fi nally, that Being-for-self is a universal, of which qualitative Somethings are instances, is indicated by Hegel’s comments about “Reality” and “Ideality,” and about “the fi nite” and “the infi nite.” Hegel says that Reality – of which the only instance so far is the determinate Something (EL, §91) – must not be regarded as distinct from Ideality; that is, that the “real” (the determinate Something) can be understood only by seeing it as a case of the “ideal” (the universal, Beingfor-self – EL, Zusatz to §91, §95). It would be a mistake, Hegel says, to credit the fi nite – of which the only example so far is the determinate Something (EL, §92) – with a being of its own apart from the infi nite; the infi nite – of which Being-for-self is the only formulation thus far – is rather the inner essence of the fi nite itself (EL, §95).

173

Familiar objects and their shadows Hegel indicates parallels between the concepts of an individual quality and of an individual numerical unit. The universal subsuming particular qualities might be labeled “the qualitatively distinctive in general,” or “the distinctively specific,” but Hegel terms it “Being-for-self.” He narrates the transition to Being-for-self in this way: “Thus essentially relative to another, [the qualitative feature or] somewhat is virtually an other against it: and since what is passed into is quite the same as what passes over, since both have one and the same attribute, viz. to be an other, it follows that [the qualitative feature or] something in its passage into other only joins with itself. … Thus Being … is now Being-for-self.”15 The opposition, in other words, of one quality to its other is no entanglement with an extraneous or alien other; the one quality and its other are both at base “the qualitatively distinctive,” Being-for-self , and are as such mutually supporting. Hegel now makes explicit the root of the connection between qualities and quantities. When we see that particular qualities are just embodiments of Being-for-self, we see that qualities are in one way similar to individual numerical units.16 So much is this so, in fact, that Hegel speaks of numerical units as being themselves instances of Being-for-self. Being-for-self is said to be the principle of a “One” which distinguishes itself “from itself ” (“Repulsion”), and thereby “makes Many Ones”;17 this Many is then said to coalesce in turn into a One (“Attraction”).18 These statements may themselves seem more repellent than attractive: it may seem that particular qualities characterized by a qualitative distinctness from one another, and particular numerical units characterized purely by a numerical distinctness from one another, simply cannot both be instances of one and the same universal. What Hegel is claiming, however, is that insofar as qualities are instances of Being-for-self, they have formal features possessed more noticeably by numerical units. The consequence is that a developed system of qualitative concepts will inevitably provide room for a range of quantitative concepts. Talk of qualities cannot stand free from talk of quantities. 15 16 17 18

EL, §95. Cf. EL, §96 and Zusatz. EL, §97. EL, §98.

174

Concluding Hegelian postscript The fi rst feature about qualities made salient by talk of a “One” and a “Many” – the fi rst of two that make qualities conceptually inseparable from quantities – is “Repulsion.” Being-for-self, we have just seen Hegel say, is the principle of a One which “repels itself ” into a Many. Here is a reading of this claim. Consider Being-for-self as the universal common to particular qualities, as “the qualitatively distinctive in general”: we can speak of one instance of Being-forself only by also speaking of more than one instance of it, of “many” instances of it. For we can treat one item or feature as being qualitatively distinctive, or qualitatively contrasting, only if we speak also of some other item or feature from which it is distinguished, with which it contrasts; and this other item or feature will then itself count as a further case of “the qualitatively distinctive,” of Being-for-self. A parallel reflection holds true of particular numerical units. We can treat one numerical unit as being numerically separate and distinct only if we speak also of units from which it is numerically distinct; these items will then themselves count as having numerical separateness on their own, and will thus themselves count as further “ones.” The other feature Hegel attributes to instances of Being-for-self, by mention of “One” and “Many,” is “Attraction.” The distinctions by which the Many are held apart are, in principle, collapsible; “ones” deemed distinct can, with equal correctness, be regarded as identical.19 This claim of “Attraction” can readily be given a plausible reading, we shall see, for numerical units as such. It is more difficult to show how “Attraction” holds true of particular qualities – but even so, that is where we must begin. How might the distinctions between distinct qualities be, in principle, collapsible? The short answer is this: because Being-for-self is structurally unlike Animality and structurally like Number. Let me now say the same thing more slowly. Plato recognized universals that were instantiated in other universals. Animality is instantiated in Doghood and Birdhood; Number in Five (i.e., fivefoldness) and in Seven (sevenfoldness). Being-for-self is thus far like both Animality and Number: it is instantiated in diverse individual qualities. And each individual quality qualifies as a universal – or at least will qualify as a universal, once individual property-bearers are there on the conceptual map, as the Logic unfolds. In one other respect, however, 19

EL, §98.

175

Familiar objects and their shadows Being-for-self must be different from one of these two Platonic universals, since they are unlike one another. In the case of Animality, we can discuss or explain the differences between different particular forms. Thus Birdhood is a form of Animality involving the possession of wings, while Humanness involves the absence of wings and possession of rationality: we can mention differentiae which differentiate any one form of Animality from others. With Number, this is not possible. Fivefoldness is a different particular form of Number from eightfoldness. But if we attempt to discuss this difference, to state what grounds it, we fi nd there is nothing to say. It is of course true that fivefoldness is “smaller” than eightfoldness: but since two particular numbers can be distinct only in that one is bigger or smaller than the other, this comparison amounts only to an assertion that fivefoldness is a different form of Number from eightfoldness, and does not explain why. The question why – why, that is, fivefoldness is smaller or greater than eightfoldness – is, in fact, ill-formed. We do not ask, because we could not answer, why one particular form of Number is different from another. The distinctions we recognize between numbers are not qualitative, and are not explicable: we distinguish numbers not through differentiae, but through a conceptual species of pointing. Can we explain the differences between particular forms of distinctive specificity, i.e., of Being-for-self ? Redness is a different particular form of distinctive specificity from brownness. Can we say that redness is distinctively specific in a different way from that in which brown is distinctively specific – as we can say that Birdhood embodies Animality in a different way from how Humanness embodies Animality? The answer would seem to be No. Animality and Birdhood are distinct forms of Animality in that each has distinct qualities, i.e., differentiae. But distinct forms of Being-for-self are distinct in that they are different qualities. When it comes to distinct cases of Being-for-self, then, any attempt to explain or ground their mutual distinctness (via differentiae) runs up against the point that here, all qualitative distinctnesses qualify as explananda – as things-to-be-explained – and thus do not in their own right qualify as explanantia. But this is to say that the distinctness between distinct cases of Being-for-self does not permit explanation – does not permit it, and does not require it. The difference between redness and brownness, for example, is ungrounded or “brute.” What 176

Concluding Hegelian postscript follows is that distinctions we recognize between distinct cases of “distinctive specificity” – that is, between distinct particular qualities – are no more explicable than those between numerical units. (The same claim is made by Armstrong.20) We distinguish distinct qualities through a conceptual species of pointing: one quality is distinguished from another just in that the fi rst is “this” quality while the other is “that other” quality. The distinctions are not grounded, but are instead stipulative, defi nitional. We can now understand Hegel’s claim that “Attraction” applies to particular qualities. A particular quality or case of “distinctive specificity” is distinguished only through something like conceptual pointing, only as being “ just this form of distinctive specificity.” But the very same words apply equally to any other particular quality. Hence the very act of distinguishing parcels of Being-for-self ends up assimilating them one to another. As Hegel puts it, “[b]ut the Many are the same as one another: each is One, or even one of the Many; they are consequently one and the same. Or when we study all that Repulsion involves, we see that as a negative attitude of many Ones to one another, it is just as essentially a connective reference of them to each other. … The repulsion therefore has an equal right to be called Attraction; and the exclusive One, or Being-for-self, suppresses itself.”21 Not only, in fact, is there a “connective reference” among the different parcels of Being-for-self, the different particular qualities. The distinctions among these are stipulative; and it is a general truth, to which we shall return later, that distinctions drawn through stipulation can, with equal correctness, be collapsed. The distinctions between particular qualities are, as we put it above, “collapsible.” (The same point might seem more persuasive if put this way: there is no privileged segmentation of a particular quality-space into distinct individual qualities.) Having seen what Attraction amounts to in the case of particular qualities, we can now consider the Attraction which holds among numerical units. Each numerical unit must count as a unit discrete from others and unitary in itself. This requires, as noted above, that each numerical unit be spoken of along with other units from which it can count as distinct. Yet those other units, like the given unit, 20 21

Armstrong, “Causal Theory of Properties,” p. 36. EL, §98.

177

Familiar objects and their shadows will have no feature beyond their numerical separateness. Hence the units from which the given unit is distinguished will be just like the given unit, except that the given unit is “this unit,” while the other units are “those units.” Hence the very assertion that there is a distinction between “this” unit and “those” will not be grounded or explicable, but will be a matter of stipulation or defi nition. As such, the distinction will be “collapsible.” The given “separate” unit can in principle be joined with others into a larger unit which will be just as genuinely separate, and just as genuinely unitary, as is the given unit itself.22 Hence there is an Attraction among numerical units. There is, equally, a reverse possibility: what we stipulate to be one “unitary” unit can, with equal correctness, be restipulated as a collection of smaller units, each of which will also be as genuinely discrete and unitary as the given unit. These points about the units in Quality are coupled, by Hegel, with a point about the particular numbers or “quanta” by which sets of such units are summed together.23 Since the units in any given set must, in principle, permit both division into more units and amalgamation into fewer, it follows that the number which expresses the sum of any given set must, in principle, be exchangeable with other numbers both greater and lesser. What we count as 12 units can equally well be counted as 36, or as 17; there is no one “fact of the matter.” Or, as Hegel puts it, “in quantity we have an alterable, which in spite of alterations remains the same.”24 We are now at the threshold of the key idea embodied in “the nodal line of Measure” – the idea which entails that qualities must contrast with other qualities in graded, typically quantifiable, degrees . 22 23 24

EL, §100 and Zusatz; see also §103 and Zusatz. EL, §101. This particular remark is in fact Hegel’s re formulation (in EL, Zusatz to §106) of the point expressed at the outset of the discussion of Quantity in such remarks as these: “In other words, when we say that the notion of magnitude lies in the possibility of being increased or diminished, we state that magnitude …, as distinguished from quality, is a characteristic of such kind that the characterized thing is not in the least affected by any change in it … [T]he distinction here given between quantity and quality is expressed by saying increase or diminution: the meaning being that, towards whatever side the determination of magnitude be altered, the thing still remains what it is” (Zusatz to §99); and “But in this way the quantum breaks up at the same time into an indefi nite multitude of Quanta or defi nite magnitudes” (Zusatz to §101).

178

Concluding Hegelian postscript We cross that threshold by asking the question which the previous discussion immediately raises. If an assertion that something amounts to 14 units is exchangeable with an assertion that it amounts not to 14 units but to 28, then what force could either assertion have? In general, what use can there be for concepts of specific quanta? Half of the answer to this question is obvious. The concept of this or that particular quantum is used only to indicate relations which hold between one collection or amount and others, and not some fact which holds true of that one collection or amount by itself. An assertion that some collection amounts to 14 signifies that that collection is or would be smaller than some other collection of 15, smaller still than some other collection of 16, considerably larger than some collection of 10, etc., but the assertion signifies nothing more: it indicates only that those relations of “smaller than” and “greater than” hold between the given collection and those others. Concepts of specific quanta have meaningful use, then, just because their use is to indicate such relations. As Hegel puts it, “the very notion of quantum is thus to push out and beyond itself.”25 Even so, this obvious point is only half of the answer that we need. A question still exists concerning the terms between which these relations of “smaller than” and “greater than” get indicated. A given amount gets counted, as we have seen, in units whose distinctness and unitariness is only stipulative, and hence a given amount of 14 can, with equal justice, be regarded as an amount of 28. What follows is that a given amount of 14 must permit both an assertion that it is smaller than some collection of 28 and that it is some collection of 28.26 Likewise for any other specific quantum. The remaining question, then, is how two such assertions can consistently be made. The answer is that they can consistently be made only if some means is available of distinguishing between the 28 with which a given 14 is identical, and the 28 – or rather those 28s – with which it contrasts. It must be possible to maintain that the difference between the given 14 and the fi rst of these 28s is “only conventional,” only a matter of the 25 26

EL, §104. In Hegel’s terms, when we recollect that a 14 is equally a 28 of units half as big, i.e., a 28/2, we see that a 14 is both an amount different from and “independent” of 14, and also something exchangeable, with “indifference,” with 28: EL, §106.

179

Familiar objects and their shadows way in which “the same thing” gets counted, whereas that between the given 14 and the other 28s is more than this. These distinctions can be drawn only if the difference between different quanta – e.g., between 14 and 28 – is such as sometimes to go along with difference of a further, more “substantive,” species, and such as sometimes not to.27 Hegel contends that this more “substantive” difference is qualitative difference. We can distinguish between some 28 which differs “only conventionally” from our given 14, and a 28 which genuinely is different from our given 14 – which genuinely, e.g., is greater – only because we can say of the former 28 that it does or would amount to something qualitatively indistinguishable from our 14, while saying of the latter 28 that it does or would amount to something much hotter, heavier, or otherwise qualitatively different from our given 14. Quanta which “genuinely” differ need not invariably be tied to different qualitative features, nor need different quanta be tied to just a single qualitative difference: a “200” can genuinely differ from 170 by being such as to amount to “scalding” rather than “hot,” or “heavy” rather than “normal,” or some other node in some other qualitative spectrum. But in Hegel’s view the quantitative divergences among quanta must be such as sometimes to have some qualitative import. Only so can the distinction be drawn between two quanta which genuinely are distinct and two quanta one of which is a recounted version of the other. What follows is that the concepts of specific quanta can have their assigned use only if these concepts are used in a system which makes use, also, of concepts of qualities. Talk of quantities requires talk of qualities. But the converse is also true. The assigned use of the concepts of specific quanta is to indicate amounts greater than some, and smaller than other, amounts, but it also is to indicate amounts which differ in varying degrees from a whole range of contrasting amounts. A 170 must count as an amount smaller than some 200, but smaller still than some 220, etc. And if the concept of “genuine” divergence among quanta requires use of the concept of qualitative difference, then the concept of degrees of “genuine divergence” among quanta requires use of the concept of qualitative difference more and more pronounced, and hence of qualities that contrast with other qualities to variable and quantifi able degrees. 27

EL, §§106 and 107 and Zusätze.

180

Concluding Hegelian postscript Property talk, I will now add in my own voice, is fully developed quality-talk. It is quality-talk accompanied by quantity-talk, quality-talk that can flesh out the distinctions required for quantitytalk to serve its assigned use. So the kinds of qualities that it recognizes – genuine properties – are capable of animating and giving content to multigrade contrasts among quanta. They are qualities that differ from other qualities in differing, quantitatively comparable, degrees. 8.3

O bj e c t i o n s t o t h e H e g e l i a n p o s i t i o n

But the Hegelian position on property-identity faces what seem, at fi rst blush, to be embarrassing counterexamples. Never mind the more contentious claim, incorporated within this position, that any genuine property contrasts with its own proper rivals to commensurable, and typically quantifiable, degrees. Even the less contentious claim that any genuine property contrasts with its own proper rivals – rivals in the plural – appears to face counterexamples. For it appears that there are some genuine properties which do not contrast with any rival properties unless – in what seems an ad hoc move – we count negative properties as their rivals. And it appears that there are many properties that contrast with, or exclude, only a single “opponent” property. In this section I examine a number of putative properties that appear to answer to each of these descriptions. I argue that in every case the appearance is delusive. The examples I will consider are drawn from everyday thought, from philosophical speculation, and most importantly from empirical theory. Everyday thought: it appears that being pregnant is a perfectly genuine property. But what other property does being pregnant exclude? The only readily apparent answer is not being pregnant – which seems to have questionable credentials as a property at all. Philosophical speculation: many philosophers endorse the view that everything real is physical, or that everything real is material. Within the philosophy of mind, this view gets articulated as the claim that every mental event is a physical event, or perhaps a material event. Now there is controversy, as debate within philosophy of mind has shown, as to just what such a claim says. What is it for an event to be “physical”: is it for it to be an event that can be specified using the vocabulary of current physics, or using the vocabulary of perfected physics – and if the latter, 181

Familiar objects and their shadows who is to say that some future physics may not recognize events that currently would seem, intuitively, to be “nonphysical”? But suppose that all these controversies can be resolved. Then if the metaphysical view that one might term “physicalism” (or “materialism”) is true, all real objects and events share a property (says this counterexample) to which no rival property is ever instantiated. Depending on how we answer the ancient and vexed question of what it is for there really to be such and such a property, this view may well entail that there is no property with which being physical (or being material) contrasts. Perhaps one might say, “well, there could have been nonphysical objects or events: nonphysical is a way that objects or events could have been, a property that they could have had; the physicalist is just saying that in fact, in the actual world, there are no such objects or events.” But this comment would seem to overlook the fact that “physicalism” (or “materialism”) is intended as a metaphysical view, as a claim as to what the world by nature is like. If the very nature of the world ensures that there are only physical objects and events, it seems hard to claim that there is any property which being physical rules out. Empirical theory: there are, we are told, six different kinds (or “fl avors”) of quarks. But quarks of whatever fl avor are characterized either by spin +1/2 (called “spin up”) or by spin −1/2 (called “spin down”). Here we appear to have two perfectly genuine properties, each of which excludes only a single “opponent.” Then too, quarks of every fl avor have either charge +2/3 or charge −1/3. Here too we seem to see single-opponent properties. A different example comes from the empirical results that are said to have signaled “the fall of parity.” Whether a glove points to the right or instead to the left seems to be just a perspective-relative matter: depending upon the vantage point chosen, one and the same glove can be said with equal justice to point to the right or to the left. Indeed whether a given glove is right-handed or left-handed can be said to be a perspective-relative matter. For if we rotate a given glove through “4-space” (a geometrical construct),28 a glove that originally qualifies as right-handed will come to qualify as left-handed instead; so right-handedness is a projection of one’s vantage point in 4-space.29 But empirical research has shown that some small particles – e.g. supercooled nuclei of 28 29

Gardner, Ambidextrous Universe, p. 155. Gardner, Ambidextrous Universe, pp. 150–52.

182

Concluding Hegelian postscript cobalt-60 – behave, in certain weak-force interactions, in an asymmetrical way that seems perspective-independent and “objective.” They emit particles which, by virtue of their participation in a magnetically southward pattern of emission, qualify as left-handed. And no possible alteration in perspective can bring about a right-handed “mirror image” of the asymmetry characteristic of the pattern of emission. Collectively, the emitted particles have a left-handedness which seems to be a genuine property, not just a way of appearing from a vantage point. But if it is a property, it is a property which excludes only one “opponent,” namely right-handedness. Let me discuss these apparent counterexamples in order. Many philosophers hold that properties are “sparse” – that there is not a real property corresponding to every predicate in our language. Certainly this much may be allowed to the proponent of the Hegelian position on property-identity. But then the proponent could simply deny, in response to the fi rst apparent counterexample, that being pregnant is a property at all. There is the predicate “is pregnant,” the proponent could say, and it is a determinable predicate. It is true of this or that woman (more broadly, female creature) in virtue of that woman’s being in one or another determinate gestational stage. The properties are being in the state characteristic of the first week of gestation, being in the state characteristic of the second week, etc. Each property contrasts with a plurality of rival properties. A roughly parallel response seems plausible in the case of the putative property is physical (or is material ) – a property to which, the counterexample alleged, there is no contrasting alternative. It is plausible to claim that the predicate “is physical” is a determinable predicate, true of an entity just in case that entity has some determinate microphysical character or other – or, perhaps more broadly, some determinate microphysical character or other or some determinate chemical character or other, some determinate biological character or other, etc. These characters would be properties, or clusters of coinstantiated properties. To each such character as a whole – or to each element in such a cluster – there would be a plurality of contrasting characters or properties. The more interesting apparent counterexamples to the Hegelian position are those suggested by quark theory. Let me begin with the properties apparently picked out by the predicates “has spin +1/2” and “has spin −1/2.” The predicates themselves appear to suggest 183

Familiar objects and their shadows that there are other values or forms of spin which, given the way the world works, quarks do not instantiate – do not, and perhaps cannot instantiate. The predicates appear to suggest, for example, that the property apparently picked out by “has spin +1/2” contrasts with has spin +1/4 or has spin +1 – which makes it seem as if the property apparently picked out by “has spin +1/2,” namely having spin +1/2, does after all exclude more than one opponent. Even so, one could argue that the threat here to the Hegelian position is serious. The opponent could say that at some very fundamental level of the world – a level populated by little more than quarks themselves – it is simply physically impossible that anything should instantiate spin properties other than just the ones apparently picked out by “has spin +1/2” and “has spin −1/2.” In that case one could argue that at this level, there are only these two spin properties. Our taxonomy may, in other words, suggest more alternatives than obtain in reality. Really, one might say, the properties picked out by “has spin +1/2” and “has spin −1/2” are as jointly exhaustive as the common synonyms – “spin up” and “spin down” – appear to suggest. My contention is that in this case, the distinction between “spin up” and and “spin down” would not signal a qualitative contrast between two distinct kinds of microparticles. Instead it would signal numerical distinctness between two discrete classes of microparticles. Between the two classes there would be many qualitative contrasts to be discovered; but there would be none as yet identified by the predicates we employ. To set forth the idea here, permit me to tell a story. A scientist enamored of theoretical simplicity surveys a domain that appears to be populated by two quite different types of microparticle, commonly called “Type 1” and “Type 2.” “Really,” the scientist announces, “there is just one type of microparticle here, which occurs in just two variant forms. I call microparticles of this one kind ‘kwarks’. The two forms which this one type of particle assumes are characterized by particular variants of The Property. What you call ‘Type 1’ microparticles are really just kwarks that have The Property, Variant Up; what you call ‘Type 2’ microparticles are kwarks that have The Property, Variant Down.” The first question that we should ask this scientist is: does having The Property amount to anything more than satisfying a bare 184

Concluding Hegelian postscript disjunction? That is, is there some common content to having The Property in either Variant Up or in Variant Down? Suppose the scientist answers “No.” Then, it seems, all that a kwark’s having The Property, Variant Up, comes to is its meeting the standard description that we formerly associated with “Type 1”; all that a kwark’s having The Property, Variant Down, comes to is its qualifying as what we formerly called “a microparticle of Type 2.” And now the scientist will have admitted that he has not really identified a commonality among what we formerly regarded as two quite different types of microparticle. His claim of theoretical simplification will have been exposed as a merely verbal maneuver. Then let us suppose, more charitably, that the scientist says there is after all some common content to having The Property, Variant Up, and The Property, Variant Down. What then distinguishes the two variants? They must embody different values or versions of that common content. The easiest way to make the suggestion concrete would be to say that the two variants embody quantitatively different values of that common content – that one embodies a greater version of the common content, and the other a lesser. But in any case it will be very hard to see why there might not be, at least as a matter of metaphysical possibility, yet other available variants of that common character. Perhaps the laws of nature forbid instantiation of these other variants. But they would still be genuine variants, genuinely alternative ways of embodying the common content. Or permit me an alternative story. Consider the case of a biological taxonomist, equally enamored of theoretical simplification. “Really,” she says, “what you call ‘trout’ and what you call ‘elephant s’ are not two different kinds of animal; they are just different forms of a single kind of animal, which I call ‘Kreeture s’. The two forms which Kreetures assume are characterized by particular variants of The Characteristic, of which there are only two. Kreetures that have The Characteristic, Variant Right, display gills and scales and fi sh-like morphology; Kreetures that have The Characteristic, Variant Left, have trunks, wrinkly skin, and a mature weight of 2 tons. All that having The Characteristic in general comes to is belonging either to what you call ‘trout’ or what you call ‘elephants’.” Then distinction between The Characteristic, Variant Right, and The Characteristic, Variant Left, would not, in and of itself, pick out a qualitative contrast between trout and 185

Familiar objects and their shadows elephants. It would serve only to signal a numerical distinctness between two classes of animals. Between the two classes numerous qualitative contrasts might obtain. But they would not obtain in virtue of the one class’s having The Characteristic, Variant Right, and the other’s having The Characteristic, Variant Left. These oneopponent “properties” would not underlie qualitative contrast: they would not really be properties at all. Now back to the imagined opponent of the Hegelian position – the philosopher who takes quark theory to show that, at a very fundamental level of the world, there are exactly two “spin” properties. There is spin up (= has spin +1/2 ), this philosopher says, and spin down (= has spin −1/2 ); each of these is a “single opponent” property. Well, if they indeed are properties, their disjoint instantiation must mark a qualitative contrast between different kinds (different fl avors) of quarks. But again we would have to ask: is there anything more to having some value of “spin” – one or another of the only two values that there are – beyond satisfying a bare disjunction? That is, is there some common content to possessing either value of “spin” – or does having spin up come to no more than belonging to numerically that class of quarks, while having spin down amounts to no more than belonging to this other, numerically distinct, class of quarks? If our philosopher endorses the latter answer, he admits that there is not after all a qualitative contrast between spin-up quarks and spindown quarks, but merely a numerical distinction between the two classes. If he endorses the former answer – if he says that there is some common content to having spin up and having spin down – then he admits that the spin-up quarks embody this common content in one way, the spin-down quarks in a different way. But these different ways of embodying that common content will be different by virtue of what each is like, and not just by virtue of the fact that this way is this way, while that way is that way. Room will therefore be created for the metaphysical possibility that there are yet other ways, unlike either of these and like something else, in which that common content can be embodied. We will be back to taking seriously the suggestion carried by our predicates: that there are contrasting alternatives such as spin +1/3, spin +1, etc. Our philosopher will have to retreat from the position that the properties spin +1/2 and spin −1/2 are “one opponent” properties. That is, he will have to admit that the

186

Concluding Hegelian postscript predicates “spin up” and “spin down,” taken as designating precisely those properties, are not jointly exhaustive of the properties available, in even this “fundamental” domain. Or, if he wished, he could cut the two predicates (“spin up” and “spin down”) loose from these precise semantic values, and could then insist that they are just as jointly exhaustive as they sound. He would have conceded that out there in nature, as metaphysically possible ways of embodying spin, there is a whole range of variant values of the common content. But each of the predicates “has spin up” and “has spin down,” he could say, is a determinable predicate, true of a microparticle just in case it possesses some one of the variant values in a part of the range. He could say that the predicates “has spin up” and “has spin down” bifurcate a range of properties. The properties which are the semantic values of these predicates, on various occasions, contrast with opponents, in the plural. The predicates by which we pick them out are only two. Our predicates furnish an impoverished taxonomy of properties – but not an impoverished reality. The same argument applies, mutatis mutandis, to the predicates “has charge +2/3” and “has charge −1/3.” But here there is perhaps less need for elaborate argument. There is something fictional about all versions of spin – these are supposedly rates at which a quark rotates on its axis, but of course we also know that quarks are not literally particles – and so it can seem excusable to just stipulate that spin up and spin down are mutually exclusive and jointly exhaustive. But there is nothing fictional about charge. We know a good deal about it, and one of the things we know is that charge admits of varied degrees. That is, not only do our predicates suggest that there are other, contrasting values of charge, beyond has charge +2/3 and has charge −1/3; the suggestion, it seems obvious, is correct. Now for the question whether the experiments that documented “the fall of parity” showed that there is a different pair of “one opponent” properties, namely right-handed and left-handed. What these experiments showed is that particles of certain kinds are all asymmetrical, in exactly the same ways as one another, in respect of various weak-force interactions. For example, when supercooled nuclei of cobalt-60 are aligned in a strong magnetic field, the nuclei all alike emit electrons and positrons in the direction of south in that magnetic field – at least, a significant majority of the electrons and

187

Familiar objects and their shadows positrons they emit get emitted in the direction south. 30 By virtue of their participation in this south-oriented pattern of emission, the emitted particles qualify as “left-handed.” But the left-handedness of the emitted particles is no mere function of the perspective from which they are viewed. The south-oriented pattern of emission, it appears, is a function of the laws of nature. Hence it is fair to infer that supercooled cobalt-60 nuclei simply cannot be brought to emit particles – at least, not a majority of the particles which they emit – in the direction north of a magnetic field in which they are aligned. No trick of perspective can bring about a “mirror image” version of the left-handedness by which the emitted particles collectively are characterized. The left-handedness is an objective, perspectiveindependent phenomenon, and distinct from what right-handedness would be. But is it really so obvious that the left-handedness by which the emitted particles are characterized, by virtue of their participation in the distinctive southward pattern, excludes only one “opponent” property? What qualifies the emitted particles as lefthanded is participation in a pattern of emission from which, it seems plain, there can be graded departures. There is a pattern of emission largely in the direction south; a pattern of emission in what is nearly a random scatter, but just barely favors south over north; there is emission in a perfectly random scatter. For that matter, there is also emission of particles strictly, entirely, in the direction south. These graded alternatives to what qualifi es the emitted particles as left-handed might seem to show that there are graded alternatives to their left-handedness itself. There is (or so it might seem) being fairly left-handed, weakly left-handed, markedly left-handed, etc. Our practices for using the predicate “left-handed” (or the predicate “right-handed”) may indeed forbid coupling the predicate with such adverbs as these. But it nevertheless seems plain that the phenomena out there in the world, which can serve as truth-makers for one or another attribution of that predicate, admit of graded departure from the phenomenon which makes it true to say that the emitted particles are “left-handed.” Each such phenomenon consists in the participation, by the particles emitted in a weak-force interaction, in a pattern of emission that is, in a perspective-independent 30

Gardner, Ambidextrous Universe, pp. 208–11.

188

Concluding Hegelian postscript way, biased to one degree or another in the same assymetrical way as is the pattern of emission characteristic of the particles emitted by the cobalt-60 nuclei. The predicate “left-handed,” as regimented by our practices, may be a determinable predicate, to which the only “opponent” determinable predicate is “right-handed.” But the properties which “left-handed” picks out will each contrast with many “opponents.” Again our predicates may furnish an impoverished taxonomy, but not an impoverished reality.

189

Appendix: “Mutually interfering” dimensions of difference Contrariety is not defi ned, I contend, for either the structural properties that typically characterize massive pluralities of microparticles, or for the structural properties that characterize typical universal mereological composition (UMC) objects. That is, there is no fi xed and stable phenomenon of “more and less different from” that relates any two such properties to a third. Suppose that c1, c2 , and c3 are structural properties of either sort, and at least appear to be contraries of one another. Then I contend that c3’s being farther removed than c2 from c1, along some dimension D1 of difference, does not in and of itself amount to c3’s being more different than c2 from c1. On the contrary, depending on how c3 and c2 stand towards c1 on other dimensions of difference, c3’s being farther away than c2 is from c1, along D1, may amount to c3’s being less different from c1 than c2 is. The same is true of any dimension of difference along which c3’s difference from c1 may be compared with c2’s difference. No dimension trumps all the rest. There is no fi nal, stable product of c2’s and c3’s relative differences from c1 along all the dimensions of difference. My experience shows that this contention is initially hard to believe, or perhaps just hard to understand. Initially, it can seem obvious that c2 and c3 each lie at some determinate distance from c1 along D1; that they are each also separated from c1 by some determinate interval along dimension D2; and so on through all the dimensions with respect to which both c2 and c3 both have locations. Once one has plotted c2 and c3 along all the relevant dimensions, it can seem, one then just needs to take the resultant vector leading from c1 to each. The longer vector will measure greater overall difference. There is a phenomenon of “more and less different from.” This initial response supposes that the space in which such structural properties stand at distances from one another is abstractly like the Munsell color solid. That is, suppose we fi rst determine that c3 190

Appendix lies at a greater remove from c1 than c2 does, along D1; and suppose next we determine that c3 is also farther removed than c2 is from c1 along D2. This response supposes that the difference between c3 and c1 that is measured along D1 “stays put,” when we add in the difference between c3 and c1 along D2 , and that if c3’s difference from c1 along D2 is again greater than c2’s, then c3 is even more different from c1, overall, than c2 is.1 That is, by the time we have taken account of positions along D2 , the vector running from c1 to c3 must be longer by an even greater margin than the vector running from c1 to c2. This is what I deny when I say that the dimensions that characterize this space are “mutually interfering.” Let me give some examples. First example. Suppose that the conjuncts in c1’s description are satisfied at fairly closely packed locations in space and time, that the conjuncts in c2’s description are satisfied at less closely packed spatiotemporal locations, and that the conjuncts in c3’s description are satisfied at locations more separated still from one another. Call this dimension of difference “D3.” An orthogonal dimension of difference measures how different the charges and energy states are that the conjuncts of c2 and c3 attribute to individual microparticles, from those that the conjuncts of c1 attribute. Call this dimension of difference “D4.” If c3 lies farther away from c1 than c2 does, along D4 as well, it could happen that the microparticles that figure in c3 attract one another, and influence one another, more strongly than do those that figure in c2. The microparticles in c3 could act in concert, to 1

From the standpoint of color-science, this comment about the Munsell color solid gets things a bit backwards. The Munsell solid is constructed to systematize the raw data about phenomenal similarity and difference between colors. (In this respect it is unlike the multidimensional spaces on which my argument focuses; in their case, there is no issue of phenomenality.) The raw data say that c 3 can, in one way, look just as different from (or similar to) c1 as c2 looks to be, while yet looking, overall (and hence in other ways), more different; and that c4 can look more different from c1 than c5 looks, in the fi rst of these ways, regardless of how c4 and c5 compare with c1 in the other ways. That is why the axes of the Munsell solid are orthogonal to each other; why, in other words, the very geometry of the solid ensures that the various differences from c1, as measured along one axis, “stay put” regardless of differences measured on other axes. There is a phenomenal phenomenon of “staying put,” and it comes before the geometry of the multidimensional space. In the multidimensional spaces that concern us, the dimensions are individuated conceptually, not by contribution to phenomenality.

191

Appendix much the same degree as do the microparticles in c1; in contrast, the microparticles in c2 might have less influence over one another than do the microparticles in either c1 or c3. In this case, the intuitive thing to say – in at least some contexts – would be that the satisfaction of c3 gives us a complex microparticle event, more like c1 than c2 is. Greater distances along D3 and D4 will have given us less overall difference. Vector length does not measure difference. At the same time, note that we cannot accommodate this fact by just inserting a “wrinkle” or a “warp” in the space defi ned by D3 and D4. For there might be a structural property c4, very much like c3, which attributes charges that diverge to just an equal degree (as in c3) from those attributed by c1, but which differs from c3 in which charges it attributes to particular microparticles – it often attributes “positive” where c3 attributes negative, or the color-charge “red” where c3 attributes “green.” Then c4 would be found at just the same place as c3 is, in the multidimensional space, but the microparticles that satisfy it would not act in concert to a greater degree than do those in c2. The “warp” would have shortened the vector from c4 to c1, but wrongly so – if the vector is supposed to measure difference. Another way in which we cannot accommodate the problem posed by the fi rst example is by saying that at the point on D4 occupied by c3 and further out, greater distances along D4 yield greater similarity to c1. That will not in general be true. Suppose that the conjuncts in c1 depict microparticles as bearing fairly strong charges and energy states, the conjuncts in c2 depict charges and energy states yet stronger, yet the conjuncts in c3 depict microparticles as only very weakly charged. Then the microparticle event depicted by c3 may after all be more different from c1 than is the microparticle event depicted by c2. In any case, here is the crucial point, so far as this “fi rst example” is concerned. Any of these intuitive verdicts about “greater and lesser difference” from c1 is vulnerable to being overturned, as soon as we consider relative distances from c1 along yet other dimensions of difference. Suppose that 300 conjuncts in c1, c2 , and c3 all depict exactly the same sorts of microparticles moving in exactly the same sorts of ways, and bearing exactly the same charges and energy states. Then any story that makes us judge, intuitively, that c3 is more like c1 than c2 is can be overturned by adding that c3 incorporates 4,000 conjuncts, while c1 and c2 incorporate only those 300 conjuncts. If a story 192

Appendix prompts the reverse verdict – that c2 is more like c1 than c3 is – then that verdict can be reversed by just making c2 be the vastly more complex microparticle event, while c3 is “ just the same size” as c1. Second example. Consider microparticle events that differ from c1 by being farther and farther removed from c1 along D3. As we move very far along D3, it seems intuitively that we get microparticle events that do not act in concert at all – that simply do not do anything. Intuitively, this utter lack of efficacy may seem to swamp any differences from c1 along D5, “number of conjuncts that attribute different properties from those attributed by corresponding conjuncts in c1.” All such events may seem to differ from c1 to an equal degree, intuitively. But again, this intuitive verdict can always be overturned. Vast differences in complexity, for example, might seem to overturn it. Third example. Suppose that thirty-six conjuncts in the description of c2 depict the microparticles they describe as lying within a volume which common sense would suppose to be occupied by a dog, and suppose that fi fty-three conjuncts in the description of c3 do the same. Call this dimension of difference “D6.” Intuitively, it may seem that difference in distance from c1 along D6 is a “difference that makes no difference.” It is as if one makes marks along D6 at different removes from c1, and as soon as one picks up one’s pen, the marks jump right back to the origin. (The marks jump, that is, if one supposes that vector length measures degree of difference.) Fourth example. Suppose that thirty-six conjuncts in the description of c2 again depict the microparticles they describe as lying within a volume which common sense would suppose to be occupied by a dog, and forty conjuncts in the description of c3 depict the microparticles they describe as lying within a volume which common sense would suppose to be occupied by a pond. Suppose that number-ofconjuncts-depicting-pondish-location is dimension D7. Is the vector from c1 to c2 longer than the vector from c1 to c3, or shorter? Wait – don’t answer! Common sense supposes that the dog mentioned in the description of c2 is swimming in a pond; but as to the pond mentioned in the description of c3, common sense makes no suppositions as to what may be swimming in it. So now it turns out that if c3 has any determinate position along D6, its position is at a zero remove from 193

Appendix that of c1, though it is noticeably removed from c1 along D7. As to c2 , it is removed from c1 along D6 by a distance nearly as noticeable as the one just mentioned – is this what one should say? – but c2 is removed from c1 along D7 by an equally great distance as well. So the vector from c1 to c2 is longer than the vector from c1 to c3. But intuitively, c2 does not seem to be more different from c1 than c3 is. The lesson generalizes: there will be many dimensions of difference, removal along which from c1 will make for longer vectors, but will not make for genuinely greater difference from c1. Despite these points about the dimensions of difference on which structural properties such as those discussed here depart from one another, there still could be a phenomenon of “more and less different” that obtained among such properties. There could be such a phenomenon, that is, if there were differences between such properties that were brute – or, as one might better say, “disrespectful.” These would be differences between one structural property and others that were not, in any way, functions of how the fi rst differed from the others in this respect, in that respect, in that other respect, etc. That is, they would be differences that did not obtain in virtue of aspects of the structural properties which they separated from one another. They would be differences that were brute, that were not grounded in what the properties separated by them were like. But the very idea of such differences appears to be nonsensical. One property’s difference from another does have an intimate connection with what each such property is intrinsically like – and if the properties in some family differ from one another in different ways, they differ in different respects, and these respectful differences determine their overall differences.

194

References

Armstrong, D. M., A World of States of Affairs (Cambridge University Press, 1997 ). “The Causal Theory of Properties: Properties according to Shoemaker, Ellis, and Others,” Philosophical Topics, 1, no. 2 (1999): 25 –38. Baker, Lynne Rudder, The Metaphysics of Everyday Life (Cambridge University Press, 2007 ). Persons and Bodies: A Constitution View (Cambridge University Press, 2000). Balashov, Yuri, “About Stage Universalism,” Philosophical Quarterly, 57 (2007 ): 21–39. “On Vagueness, 4D and Diachronic Universalism,” Australasian Journal of Philosophy, 83 (2006): 523 –31. Persistence and Spacetime (Oxford University Press, forthcoming). Bennett, Jonathan, Events and Their Names ( Indianapolis: Hackett , 1988). Bigelow, John and Pargetter, Robert, Science and Necessity (Cambridge University Press, 1990). Carter, W. R., “Will I Be a Dead Person?,” Philosophy and Phenomenological Research, 59 (1999): 167–71. Davidson, Donald , “Emeroses by Other Names,” Journal of Philosophy, 63 (1966): 778 –79. “Mental Events,” in Lawrence Foster and J. W. Swanson, eds., Experience and Theory (Amherst: University of Massachusetts Press, 1970), pp. 79–101. Dennett, Daniel, “Quining Qualia,” in A. Marcel and E. Bisiach, eds., Consciousness in Contemporary Science (Oxford University Press, 1988), pp. 42–77. Elder, Crawford , “Conventionalism and Realism-Imitating Counterfactuals,” Philosophical Quarterly, 56 (2006): 1–15. “Conventionalism and the World as Bare Sense-data,” Australasian Journal of Philosophy, 85 (2007 ): 261–75. “A Different Kind of Natural Kind,” Australasian Journal of Philosophy, 73 (1995): 516 –31. “On the Place of Artifacts in Ontology,” in Eric Margolis and Stephen Laurence, eds., Creations of the Mind: Theories of Artifacts and Their Representation (Oxford University Press, 2007 ), pp. 33–51. “Physicalism and the Fallacy of Composition,” Philosophical Quarterly, 50 (2000): 332 –43.

195

References Real Natures and Familiar Objects (Cambridge, Mass.: MIT Press, 2004). Findlay, J. N., Meinong’s Theory of Objects and Values (Oxford University Press , 1963). Gardner, Martin, The Ambidextrous Universe: Left, Right, and the Fall of Parity ( New York and Toronto: New American Library, 1964). Grandy, Richard , “Artifacts: Parts and Principles,” in Eric Margolis and Stephen Laurence, eds., Creations of the Mind: Theories of Artifacts and Their Representation (Oxford University Press, 2007 ), pp. 18–32. Hall, Ned , “Two Concepts of Causation,” in John Collins, Ned Hall, and L. A. Paul, eds., Causation and Counterfactuals (Cambridge, Mass.: MIT Press, 2004). Haslanger, Sally, “Persistence through Time,” in M. J. Loux and D. Zimmerman , eds., The Oxford Handbook of Metaphysics (Oxford University Press, 2003), pp. 315–54. Hawley, Katherine, How Things Persist (Oxford University Press, 2001). Hawthorne, John, Metaphysical Essays (Oxford University Press, 2006). Hegel, G. W. F., Hegel’s Science of Logic, trans. A. V. Miller ( London: George Allen & Unwin, 1969); originally 1831. The Logic of Hegel, Pt. i of The Encyclopaedia of the Philosophical Sciences, trans. William Wallace, 2nd edn. (Oxford University Press, 1892); originally 1817. Philosophy of Nature, trans. A. V. Miller ( Oxford University Press, 1970); originally 1830. Heil, John, and Mele, Alfred , eds., Mental Causation (Oxford University Press, 1993). Hirsch, Eli, The Concept of Identity (Oxford University Press, 1982). Dividing Reality (Oxford University Press, 1993). “Physical-Object Ontology, Verbal Disputes, and Common Sense,” Philosophy and Phenomenological Research, 70 (2005): 67–97. “Rules for a Good Language,” Journal of Philosophy, 85 (1988): 694 –717. “Strange Thoughts of the Third Kind,” in Peter A. French, Theodore E. Uehling, Jr., and Howard K. Wettstein, eds., Midwest Studies in Philosophy, XII: Realism and Antirealism ( Minneapolis: University of Minnesota Press , 1988), pp. 3–24. Hitchcock , Christopher, “The Intransitivity of Causation Revealed in Equations and Graphs,” Journal of Philosophy, 98 (2001): 273 –99. “A Tale of Two Effects,” Philosophical Review, 110 (2001): 361–96. Horgan, Terence, “Mental Quausation,” Philosophical Perspectives, 3 (Philosophy of Mind and Action Theory) (1989): 47–76. Horgan, Terry, and Potr č, Matjaž, “Blobjectivism and Indirect Correspondence,” Facta Philosophica, 2 (2000): 249 –70. Husserl, Edmund , Cartesian Meditations, trans. Dorion Cairns ( The Hague: Martinus Nijhoff, 1970); originally 1929. Johnston, Mark , “Hylomorphism,” Journal of Philosophy, 103, no. 12 (2006): 652 –98.

196

References Jubien , Michael, Ontology, Modality, and the Fallacy of Reference (Cambridge University Press, 1993). Kim , Jaegwon, Supervenience and Mind (Cambridge University Press, 1993). Koslicki, Kathrin, “The Crooked Path from Vagueness to FourDimensionalism,” Philosophical Studies, 114 (2003): 107–34. Laycock , Henry, Words without Objects (Oxford University Press, 2006). Lewis, David , “Causation,” in Philosophical Papers, vol. II (Oxford University Press , 1986), pp. 159–213. Parts of Classes (Oxford: Blackwell, 1991). “Survival and Identity,” in Philosophical Papers, vol. I ( New York: Oxford University Press, 1983), pp. 55–77. Lowe, E. J., “In Defense of Moderate-Sized Specimens of Dry Goods,” Philosophy and Phenomenological Research, 57 (2003): 705 –10. Kinds of Being: A Study of Individuation, Identity and the Logic of Sortal Terms (Oxford: Blackwell, 1989). Mackie, J. L., “Causes and Conditions,” American Philosophical Quarterly, 2 , no. 4 (1965): 245 –64. Markosian, Ned , “Brutal Composition,” Philosophical Studies, 92 (1996): 211–49. McDermott, Michael, “Redundant Causation,” British Journal for the Philosophy of Science, 46 (1995): 523 –44. McLaughlin, Brian, “On Davidson’s Response to the Charge of Epiphenomenalism,” in Heil and Mele, eds., Mental Causation. Merricks, Trenton, Objects and Persons (Oxford University Press, 2001). Millikan, Ruth, Language: A Biological Model (Oxford University Press, 2005). Language, Thought, and Other Biological Categories (Cambridge, Mass.: MIT Press , 1984). On Clear and Confused Ideas (Cambridge University Press, 2000). Varieties of Meaning (Cambridge, Mass.: MIT Press, 2004). White Queen Psychology and Other Essays for Alice (Cambridge, Mass.: MIT Press , 1993). Mills, Eugene, “Interactionism and Overdetermination,” American Philosophical Quarterly, 33 (1996): 105 –17. O’Leary-Hawthorne, John [see John Hawthorne], and Cortens, Andrew, “Towards Ontological Nihilism,” Philosophical Studies, 79 (1995): 143 –65. Parfit, Derek , Reasons and Persons (Oxford University Press, 1984). Putnam , Hilary, “Realism and Reason,” Proceedings and Addresses of the American Philosophical Association, 50 (1977 ): 483 –98. Reason, Truth and History (Cambridge University Press, 1981). “Why There Isn’t a Ready-made World,” Synthese, 51 (1982): 141–67. Rea , Michael, “In Defense of Mereological Universalism,” Philosophy and Phenomenological Research, 57 (1998): 347–60. World without Design (Oxford University Press, 2002). Rea , Michael, ed., Material Constitution: A Reader ( Lanham, Md.: Rowman & Littlefield , 1997 ).

197

References Rosen, Gideon , and Dorr, Cian, “Composition as a Fiction,” in Richard M. Gale, ed., The Blackwell Guide to Metaphysics (Oxford: Blackwell, 2003), pp. 151–74. Sanford , David , “Fusion Confusion,” Analysis, 63 (2003): 1–4. Saucedo, Raul, “Parthood and Location,” in Dean Zimmerman , ed., Oxford Studies in Metaphysics (Oxford University Press, forthcoming ). Schwartz , Stephen, “The Essence of Essence,” Australasian Journal of Philosophy, 87 (2009): 609 –23. Sidelle, Alan, “Modality and Objects,” Philosophical Quarterly, 60 (2010): 109 –25. Necessity, Essence, and Individuation ( Ithaca, N.Y.: Cornell University Press, 1989). “A Sweater Unraveled: Following one Thread of Thought for Avoiding Coincident Entities,” Noûs, 32 (1998): 423 –48. Sider, Theodore, “All the World’s a Stage,” Australasian Journal of Philosophy, 74 (1996): 433 –53. Four-Dimensionalism: An Ontology of Persistence and Time (Oxford University Press , 2001). “What’s So Bad About Overdetermination?,” Philosophy and Phenomenological Research, 67 (2003): 719 –26. Sosa, Ernest, “Existential Relativity,” in Peter A. French and Howard K. Wettstein, eds., Midwest Studies in Philosophy, vol. XXIII: New Directions in Philosophy ( Boston: Blackwell, 1999), pp. 132–43. Spelke, Elizabeth, “Initial Knowledge: Six Suggestions,” Cognition, 50 (1994): 431–45. “Principles of Object Perception,” Cognitive Science, 14 (1990): 29 –56. Sturgeon, Scott, “Physicalism and Overdetermination,” Mind, 107 (1998): 411–32. Thomasson, Amie, Ordinary Objects ( New York: Oxford University Press, 2007 ). Thomson, Judith Jarvis, “Parthood and Identity across Time,” Journal of Philosophy, 80 (1983): 201–20. Unger, Peter, “I Do Not Exist,” in G. P. Macdonald , ed., Perception and Identity ( Ithaca, N.Y: Cornell University Press, 1979), pp. 235–51. van Inwagen, Peter, “Can Mereological Sums Change their Parts?,” Journal of Philosophy, 103 (2006): 614 –30. Material Beings ( Ithaca, N.Y.: Cornell University Press, 1990). Waterfield , Robin, The First Philosophers (Oxford University Press, 2000). Wells, Alexander Frank , Structural Inorganic Chemistry (Oxford University Press, 1962). Wheeler, Samuel ,“On That Which Is Not,” Synthese, 41 (1979): 155 –73. Wiggins, David , Sameness and Substance (Cambridge, Mass.: Harvard University Press , 1980). Sameness and Substance Renewed (Cambridge University Press, 2001). Woodward , James, Making Things Happen ( New York: Oxford University Press, 2003).

198

References Woodward , James, and Hitchcock , Christopher, “Explanatory Generalizations, Part I: A Counterfactual Account,” Noûs, 37 (2003): 1–24. “Explanatory Generalizations, Part II: Plumbing Explanatory Depth,” Noûs, 37 (2003): 181–99. Zangwill, Nick , “Aesthetic Functionalism,” in Emily Brady and Jerrold Levinson, eds., Aesthetic Concepts: Sibley and After (Oxford University Press , 2001).

199

Index a priori knowledge, 10 abeternal, 55 adolescent, 63–65 as a natural kind, 61–62 aggregate, 146 Alexander’s dictum, 117 alleles, 91, 93 alteration, 3, 44 qualitative, 13 analytic, 10 kind terms, 25 truths, 8 Animality, 175–76 anomalous monism, 117 antirealism, 2 about persistence, 76 conventionalist, 31 apcean, 21, 34, 39–41 apples, 34, 42–44 application conditions, 32 , 36 reapplication conditions, 32 appluark, 43–44 Aristotelian substances, 141–42 Aristotle, 58, 154, 170 aspen trees, 144 assignment, 151, 161 Attraction, 174–75, 177–78 attunement of behavior, 97–98, 100, 108–12 successful, 99

of belief, 97 to circumstances, 97 to pursuit of desire, 97 axon, 107, 110 Baker, Lynne Rudder, 64–66 Balashov, Yuri, 74 baldness, 143 bananas, 83 banyan trees, 144–45 baseball, 106, 115 shattering a window, 129 beaver tail-slap, 17 behavior, 14 goal-pursuing, 97 outcome, 4, 88–91, 106, 111–12 shaping, 90, 93, 97, 111 causal, 90 variant, 99 Being, 170–71 Determinate, 171 Being-for-self, 173–74, 176–77 belief, 14, 98, 108–10, 112–13 accurate, 97 and relevance to desire, 99 content, 112 false, 97 historical, 97 irrelevant, 97 -producing device, 97–100, 112

200

Index production of, 98 proper function, see proper function: of beliefs stored, 97 -tokening, 107–8 true, 96 variant, 97–99 belief–desire psychology, 126 beliefs and desires, 88–90, 93, 95, 98, 106–8 Bennett, Jonathan, 80 berries, 98 Bigelow, John, 155, 195 biological description, 92 biological device, see biological trait biological products, 90 beliefs and desires as, 90 biological trait, 90–92 biological type, 92 birth control pills, 95, 137 blank check, 37 bootstrapping, 13 border contrast, 50 borders, 50–51 boundaries, 5 brain, 88–89, 93, 95, 98, 100, 107–8, 110 devices, 90 event, 89, 107 state, 107 brute differences, see differences, brute butter, 146 car, 42–43, 45 car horn, 106 carple, 39–41, 44 cat, dying, 86–87

causal exclusion, 3, 88–89, 100–6, 107–12 , 159 arguments, 3 causality, 1, 3–4, 28, 89–94, 100, 108–10, 118, 167 and shaping of behavior, 90 as counterfactual dependence, 94 as exceptionless law, 126–27 as invariance, 80, 128 causal chain, 132–37 collective, 108 competition, 111, 125, 127 control, 27–28, 46, 67, 69 and persistence conditions, 77–79 lack of, 46 counterfactual account of, 80–81 detection of, 42 , 44 efficacy, 3, 88, 159 bearers of, 116–18 in shaping behavior, 113 of beliefs and desires, 89 of dogwise arrangement, 116, 120 of mental events, 91 of microparticles vs. dogs, 117 explanation, 94, 120–21, 134 familiar objects, 88 governance, 11 history, 92 immanent, 82 in science, 93 interaction, 47 mediated, 137 microphysical, 92 possible-cause generalization, 135 regulated property cluster, 27 remote, 47

201

Index causality (cont.) significance of evolutionary biology, 92 sufficiency, 106 token, 118, 132 , 134–37 type, 135, 137 chairwise arranged, 119 closest possible worlds, 105 closure of physics, 3 clustered characteristics, see properties: clusters cobalt-60, 183, 187, 189 coffee, 146–47 cohesive mass, 50, 122–23, 164 Coincidence, 73, 150 spatial, 100 colocation, 8 color, 102 , 157 complex event, 127, 129–31 microparticle, 100–3, 105, 130, 192–93 microphysical, 111 neurochemical, 107 complex property microphysical, 93 structural, 100, 109 composite objects, 5, 162–63, 166–167 composition, 4, 57, 63, 115–16, 118, 123, 142 , 146–48, 162–164, 167 borderline, 143–44, 146, 163 brittle, 5, 167 flexible, 5 mereological, 162 , 168 restricted, 142 , 165 universal mereological, 123 conceptual scheme, 38–39, 48 conjunctive predicate, 100–2

conjunctive property, 157 conservation principles, 12 prediction of accidental properties, 47 constitution, 63 consumer device, 16, see representation: consumer continuous series, 142 contrariety, 159, 162 see also contrary properties contrary properties, 5, 11, 42–44, 99, 154–55, 159, 167–169, 171–74, 181, 190 conventionalism about kind terms, 25 about persistence, 28 see also modal conventionalism convergent implementation, 52 conversational setting, 163 count nouns, 146 counterfactual dependence, 133 causation as, 94 non-foretracking, 95 counterpart microparticles, 163 modal, 71 relations of samples of matter, 73 of statues, 73 temporal, 3, 72 temporal, 71 theory, 163–65 covariation, 11, 27 cpersons, 21–22 Danielson, Connecticut, 157 Davidson, Donald, 39, 91, 117, 126–27, 131

202

Index “Mental Events” (paper), 90, 117 dendrites, 107 density, 51 dependence, 94 descriptive content, 2 desire, 15, 98, 112–13 -producing device, 96, 113 production of, 98 proper function, see proper function: of desires stored, 96 -tokening, 107–8 unactionable, 96 unrealistic, 96 variant, 97, 99 destruction, 44 determinable predicate, 183, 189 D-fusion, 19–20, 46, 150–51 minimal, 151 diabetes sufferers as natural kind, 61 diachronic counting, 83 difference degrees, 105, 109, 130–31, 161, 180–81, 191–94 commensurable, 159 mutually interfering, appendix quantifiable, 178 dimensions, 99, 102, 111, 190–94 independent, 99 mutually interfering, 99 systematic interference among, 161 differences, brute, 194 dimensions additive relations, 161 disjunctive generalizations, 41 disjunctive predicates, 41–42 dog, 71, 84, 115–16, 118, 163–65

bark, 120–121, 128–29, 131, 136–37 reflecting off shed, 123 effects of, 116, 121 hole at base of fence, 121 surface of, 120 dogwise arrangement, 4, 138–39, 167 Dorr, Cian, 115 Eiffel Tower, 157 Eleatic Stranger, 3 electron, 60–61 elephants, 185 eliminativism, 150 empirical knowledge, 10–11 empirical learning, 11, 13 empirically revisable, 11 endurantism, 70 motivations, 72 error, see possibility of error Etwas, 171 event c, 100–106, 111–12 event e, 100–106, 111–12 evolution, see natural selection evolutionary biology, 90, 92 exceptionless generalization, see exceptionless law exceptionless law, 90–93, 117, 126–27, 131–32 causation as, 88 exdurantism, 70–87, 168 motivations, 72 existence without persistence, 56 exogenous variable, 95–97, 99, 128, 132 explanatory invariance, 109 see also invariance: explanatory

203

Index explosivism, 2–3, 5, 8–9, 14, 45, 70, 72 , 75, 150 about numerical persistence, 19–20, 22–23 fallacy of composition, 109 familiar objects, i, 1–6, 22 , 31, 42 , 55, 70, 76, 78, 81–82 , 86, 88, 106, 114, 122 , 127, 139–140, 149, 162 , 166 fictionalism, 119–20 fitness-enhancing effect, 91 fluid, 10 fly consumption trigger, see neural signal S folk ontology, 119 folk physics, 127 folk psychology, 95–96 fourcylortart, 43 fox, 121, 123 frog, 15 fungus, 144 fused objects, 140 fusion, 86, 142–143, 145 diachronic mereological, 150, 152 , 160–61 mereological, 162 synchronic mereological, 152 genome, 11 gold, 58 Grandy, Richard, 122 gricular, 39 grincular, 39 group subject, 48 H 2O, 63, 65 as a natural kind, 69 haecceity, 57–59

hamburger, 157 hardwired, 11, 13 Hawley, Katherine, 82–83, 86–87 Hegel, G. W. F., 145, 154, 189 Logic, 168–69, 170, 175 Hirsch, Eli, 21–22 , 25, 34, 40, 45–46, 50, 122 historically normal, 90, 92 Hitchcock, Christopher, 42 , 80, 89–90, 93–95, 101, 118, 128, 132–33, 196, 199 homogeneity, 50 horse, 37 Hôtel Biron, 157 human being as a natural kind, 66 diabetic, 65 life stages, 63 organism, 5 human diabetic, 66 Husserl, Edmund, 52–53 ice, 65 as a natural kind, 61–62 , 69 identity, 7 fi xed by conceptual scheme, 32 kind, 2 , 7–11, 26–27 numerical, 7–8, 140 see also persistence: numerical property, 58 quality, 168 unfaceted, 167–68 identity in difference, 166–67 incar, 21–22 , 34–35, 45–46 individuality attenuated, 145 induction, 10, 26, 28–29 inductive inference, 9 and natural kinds, 67

204

Index intervention, 134, 136 on variables, 94, 101 intuition, 133–34 INUS, 80 invariance, 42 , 44, 89, 94, 97, 100–1, 109, 111, 118, 128, 132 , 134, 137, 168 as account of causality, 80 between beliefs/desires and behavior, 95 between dog barking and observer hearing, 129 explanatory, 89–90 range, 94 single-valued, 131 two or more, 137 Johnston, Mark, 22 judgement about kinds, 26 of folk observers, 120 kinds animal, 9–10 artifact, 10 edible, 99 families of, 10 fluid, 9 food, 9–11 membership, 26 metal, 11 natural, 2 , 8, 36 as language-dependent, 37 higher-level, 61 lower-level, 61, 66 non-primary, 65 plant, 9 primary, 65 property, 99

strange, 39–44 kind-sameness, see identity: kind klable, 21, 45–46 scratched, 47–48 Koslicki, Kathrin, 141, 144–45, 147 Kreetures, 185 kwark, 184–85 language organ, 107–8 Laycock, Henry, 141, 146 learning, 11–12 , 14, 17, 98 Leibniz’s law, 62 , 73 lemonade, 148 Lewis, David, 85, 94–95, 132–33, 141–42 , 162–163 on modality, 71 Mackie, J. L., 80 manipulation, 94, 128, 136–37, see also intervention Markosian, Ned, 57 material objects, 2 materialism, see physicalism matter-names, see noncount nouns maximal, cohesive, separately movable mass, see cohesive mass Meinong, Alexius, 56 mental causation, 3, 88–90 mental event, 91 mereological simples, see simples mereological sum, 4–5, 123, 140, 167 mereology essentialism, 140, 162–63 Merricks, Trenton, 111, 115, 117 methane, 153, 155–57 amplified, 157

205

Index microparticles, i, 1, 3–5, 57, 88–89, 92–93, 100–1, 103–6, 111, 114, 116–26, 128–29, 132 , 137, 139–40, 145–49, 159–65, 184, 190–93 divergent aggregations, 92 microphysical event complex, 91–92 microphysics, 91, 114, 139 laws of, 92 microstructure, 7, 11 Millikan, Ruth, 13–19, 90, 95, 112–113 mind-dependent, 6 mind-independent, 1–2 , 6, 28, 36, 40, 49, 51 see also realism people, 53 modal conceptualism, see modal conventionalism modal conventionalism, 1–2 , 8–10, 14 about numerical persistence, 28 molecular structure, 69 molecules, 107 momentary slices, see temporal part motion, 12 , 50 Munsell color solid, 102 , 156–57, 191 natural kind, 10, see kinds: natural natural selection, 6, 11, 14–16, 90–91, 93, 95–99, 107, 112 shared history, 90 naturally selected device, 95 naturally selected programs, 90 necessity de dicto, 91

de re, 91 metaphysical, 58 neural event, 107–108 neural realization, 107 neural signal S, 15 Neurath’s boat, 5 neuroanatomy, 107–8 neurochemistry, 107–12 events, 108 states, 88–89, 108 neurons, 93, 107–10 neurotransmitter, 107–9 noncount nouns, 2 , 32–33, 65, 146 Nothing, 170–71 Number, 175–76 numerical sentence, 143–44, 146 truth conditions, 147 numerical units, 177–78, 179–80 objects course of existence, 34 incomplete, 56 individuated, 145 macroscopic, 106 material, 2 mereological, 140, 142 diachronic, 141 spatially discontinuous, 152 , 157, 159 spatially unitary, 50 strange, 21–23, 45–46 unitary, 139, 146 object-stage, 73 occasion sentence, 15 ontological free lunch, 139–40, 165 ontological relativism, 2 , 31, ch. 2 and empirical discovery, 37–38 master argument for, 38, 40, 47

206

Index OR, see ontological relativism ordering of variations, 101 outcar, 21–22 , 45–46 overdetermination, 3, 89, 93, 100–6, 107–12 , 126 symmetrical, 134 systematic, 126 owl, barred, 121, 123 Panthéon, 157 parcel, 146 Pargetter, Robert, 155, 195 Parmenides, 170 parts, 139, 166 pattern detection, 13 patterns of continuity, 29, 46 perceptual tracking, 13 perdurantism, 70–71, 85 motivations, 72 persistence, 2–3, 7–9, 13, 70, 150–151, 167 across time, 2 brute, 56–57 conditions, 2 , 26, 28, 35–36, 55 and conceptual schemes, 35 mind-dependent, 49, 51 numerical, 12–13, 28, see persistence9 propertyless, 58 semantic assessment of persistence claims, 73 strange, 23, 45–46, 72 vicarious, 71–72 physicalism, 182 Plato, 3, 170, 175 poker, 71 possibility of error, 7–8, 11 possible-cause generalization, 136 prediction, 13

pregnancy, 95, 137, 181, 183 primary kind, see kind: primary prime matter, 58–59 producer device, see representation: producer projection downward, 3–4, 72 proper function, 96 of beliefs, 96–97 of desires, 96 properties accidental, 42 , 47, 152–53, 158, 161–162 accidental structural, 162 aggregate mass, 162 bearer, 153–54 biochemical, 107 border-insensitive, 51 standard, 51 border-sensitive, 51 clusters, 11, 27, 29, 41, 67 disjunctive, 43 dispositional, 51, 77–79, 99 essential, 28, 158 structural, 161 transmission of, 79–80 genuine, 5, 142 , 154, 162 , 167, 181 genuine but noncontrasting, 181–82 historical, 74 immediate phenomenal, 51 individuating, 11 intrinsic, 71 as relational, 71 lingering, 74–75 nature of, 168 of aggregate mass, 153 physical, 181, 183

207

Index properties (cont.) ranges, 10, 26–27 social, 99 sparse, 183 structural, 5, 109–111, 142 , 148, 153–54, 158–62 , 190 “mere list”, 157–58, 160, 162 accidental, 160 and contrary properties, 158 and UMC objects, 151 complex, 168 rich, 154, 162 unitary, 157 variant, 110, 161 surface-level observable, 99 variant, 11, 27 pushmipullyu representations, 13–19, 23, 25–26, 28–29 Putnam, Hilary, 37–38, 52–53 pyramidal cell, 110 qualitative principle, 62 and natural kinds, 60 quanta, 178, 180–81 quark, 43–44 spin of, 182–85, 186–187 quausation, 89, 112–13 Quine, W. V. O., 167 Rana pipiens, 15 Rea, Michael, 115, 140 realism, 2 , 9, 118 about cell biology, 55 about dogs, 124 about existence of material objects, 2 about neuroanatomy, 55 about plate tectonics, 55 about sameness, 2

and courses of existence, 55 and natural kinds, 55 and persistence conditions, 55 qualitative principle, see qualitative principle sparse, 13 Reason, 166 recommendation letter, 133 as cause of death, 133, 136 referential intentions, 2 reidentification, 11 relativism, 32–33 representation, 14 consumer, 14–15 of obtaining states of affairs, 96 producer, 14 reproductive fitness, 91 Repulsion, 174–75, 177 retrodiction, 13 Rosen, Gideon, 114 Rube Goldberg machine, 135–36 sameness “soft”, 44 sameness judgements, 17 scepticism about composition, 4 about familiar objects, 166 Schwartz, Stephen, 57–58 semantic relevance of sortals, 76 semantic rules for sortals, 37 semantic value, 112 sense-data, 51–52 sensory experiences, 116, 118, 120–121 shadow, 100, 168 shape, 50 shared history, 91 ship of Theseus, 3

208

Index Shoemaker, Sydney, 45 Sidelle, Alan, 37–38, 45, 150 Sider, Theodore, 19–20, 83–85, 141–45, 162 simples, 1, 114–15, 119, 122 , 139, 167–68 simultaneity, 100 size, 50 snow, 24 snowballs, 24 snowdiscall, 21–24 snow-perfect-sphere, 22 solipsism, 52 sorites, 142 sortals, 2 , 32–33, 36, 44, 48 and matter-names, 36 shared, 49 Sosa, Ernest, 21–22 , 39 space–time “worms,” 71–72 , 83 spatiotemporal relations, 7 species, animal, 11 squirrel, 61 stage theory, 3, see exdurantism standing sentence, 17 statue, 115 and clay, 62–63 and gold, 73 of Atlas, 62 , 73 strange languages, 22 , 34, 39–44 classificatory strangeness, 34 individuative sameness, 34 strange sortals, 39–44 Strangers, 41–43, 46 stuff, 49 primordial, 150 subvenience, 88–89 supervenience, 88, 100 sweetness, degrees of, 99 synapse, 110

system of belief and desire production, 98 table, 45 scratched, 47–48 t-counterpart relations, 73 mind-independent, 76 nonstandard, 75 nonsupervenient, 83 Ted, 74 Ted, Fred, and Ed, 84–85 temperature range, 42 temporal action at a distance, 81 temporal counterpart relations, see t-counterpart relations temporal part, 3, 72 , 85, 167 thrombosis, 95, 137 Tolstoy, 74–75 transcendental ego, 6 transference principles, 46–47 trees, 45 treesorchips, 45 triggering event, 77 trout, 185 trout–turkey, 19–20, 23, 72 , 151 truth conditions, 8–9, 13, 28 UMC, see universal mereological composition UMC objects, typical, 190 Understanding, the, 166–68 unitary thing, 157 universal mereological composition, 139, ch. 7 universals, 175 unrestricted mereological composition, 4–5, see universal mereological composition

209

Index vagueness, 35, 86–87, 144 van Inwagen, Peter, 114–15, 119, 122 , 140 variable, 94 cause and effect, 94–95 independent, 132 multivalued, 94, 101, 128–29 correlation among, 118 single-valued, 101 two-valued, 128 velocity, 50 viscosity, 51

War and Peace, 74–75 water as H 2O, 31 Weehawken, 157 what-is, 170 Wiggins, David, 65 wine, 146 Woodward, James, xi, 18, 42 , 80, 89–90, 93–95, 101,117–18, 128–29, 132–36, 198–99 writer–cucumber, 75, 151

210