This essay's title may seem paradoxical. Ontology is what there is; so how can there be a question about the relation between the “former” and the “latter”? The answer lies in distinguishing two types of ontological questions. Consider a philosophy professor who is wondering

This sentence can be taken in more than one way. The first way is to take it as

where (2) is indifferent to the ontological status of eyes but is to be answered affirmatively just in case not every student in question is either eyeless, or blue-eyed, green-eyed, hazel-eyed, etc. On the second construal (which is the type this essay will give to sentences of this sort), however, (1) asks a question that is different from (2) and receives an affirmative answer iff the empirical conditions necessary for an affirmative answer to (2) are satisfied and eyes have the ontological status of being entities. I will call questions of the sort expressed by (2) “ontological-basics questions” and questions of the sort expressed by (1), on the second construal, “ontological-status questions.”

This distinction must be distinguished from several other distinctions. It might be tempting to say ontological-basics questions are empirical, while ontological-status questions deal with matters that go beyond empirical facts. But this would be a mistake.

A characteristic principle of much contemporary antirealism is this: truth supervenes on evidence, in the sense that there can be no difference in truth value (between two statements, theories, worldviews, etc.) unless there is also a difference in epistemic value. In the first part of this essay, I will demonstrate this principle at work in the writings of two leading antirealists, Hilary Putnam and Nelson Goodman. In the remainder of the essay, I will argue that the principle is self-refuting in much the same way as the logical positivists' criterion of meaningfulness is: it fails to satisfy the requirements it seeks to impose on other views. There are ways of construing the principle to save it from selfrefutation, but they all have the effect of undermining its intended applications.

The Principle Exposed

Though never explicitly formulated, the supervenience principle plays a critical role in the arguments of Putnam and Goodman. Both writers make much of the fact that there are theories or descriptions of the world that are logically incompatible if taken at face value but that, at the same time, are in some sense equivalent. The equivalence in question is clearly supposed to be some sort of epistemic equivalence. It is not, however, to be understood in narrow empiricist fashion; it involves not merely covering equally well any relevant empirical data, but also fulfilling to the same degree any further constraints that reason may impose.

Introduction

Confusion Supervenes

Supervenience is a basically useful notion, but ‘supervene’ has become philosophers' jargon, waved like a wand to dazzle. ‘Supervene’ and ‘supervenience’, having been coined in their current philosophical sense around 1950, are now standard philosophical vocabulary. ‘Supervene’ in this use is purely a term of art, standing for several different concepts, as documented by Teller [p], Kim [cs], and me [s]. Yet the current mode is to take for granted that we all mean the same thing by ‘supervenience’. It is treated as though it were an everyday word with an ordinary, well-understood meaning we all intuitively grasp. The various technical senses would then be so many proposed explications of the assumed ordinary sense. But this common practice gets it upside-down. There is no ordinary philosophical sense of ‘supervenience’. There are only technical senses; their logical connections are complex. Here I will try once again to disentangle some of them.

A Relation between Families of Properties

A central confusion, or at any rate diversity, in the way ‘supervenience’ has come to be bandied about concerns the terms of the supervenience relation. As originally introduced, supervenience related two families of properties of the same type. For example, the mental properties of people might be held to supervene upon their physical properties. If the supervenient family is narrowed down to a single property, that's no problem.

A growing number of philosophers have turned to supervenience in hopes of finding a nonreductive form of determination. The search, however, is complicated by the fact that there is an entire family of supervenience relations, not all of which have the same reductive and determinative capacities. Among these various relations, two have come to receive a considerable amount of attention and interest, namely, strong supervenience and global supervenience.

Strong supervenience is a relation that holds between sets of properties and the individual objects that exemplify these properties. In particular, for two sets of properties A and B:

In contrast to strong supervenience, global supervenience is a holistic relation that applies to entire worlds rather than individual objects:

Some philosophers prefer global supervenience over strong, arguing that although strong supervenience might be strong enough to function as a relation of determination, it is not weak enough to be nonreductive. Others prefer strong supervenience over global, arguing that although global supervenience might be weak enough to be nonreductive, it is not strong enough to serve as a relation of determination.

A familiar strategy for nonreductive physicalists of mind to employ in their attempt to reconcile psychophysical event identity with the view that mental and physical properties of events are irreducibly distinct is to appeal to some version of the doctrine of psychophysical supervenience. Many who appeal to this doctrine do so in the belief that psychophysical supervenience not only establishes a covariance relation between mental and physical properties, in the sense that indiscernibility with regard to the latter entails indiscernibility with regard to the former, but establishes an asymmetric relation of dependency of mental upon physical properties that falls short of reducibility. Psychophysical supervenience is thus thought to be capable of showing that although mental properties are strictly speaking nonphysical, the ontology of the physical world in some sense both determines and exhausts what there is to the mental domain. I think it fair to say that many nonreductive physicalists of mind, myself included, believe that without supplementation by some version of a supervenience thesis, nonreductive monism is not worthy of the name ‘physicalism’.

Recent discussions of psychophysical supervenience have focused primarily on different conceptions of supervenience (strong, weak, global), on the question of which conceptions (if any) are appropriate to the psychophysical case, and on the question of whether those conceptions that may be appropriate to the psychophysical case lead to reducibility. The difficulty faced by nonreductive physicalists is to find a characterization of supervenience that is strong enough to meet the demands of asymmetric dependency without also leading to reducibility.

The Basic Concept

For most people who are not familiar with its many manifestations, analytical philosophy is the philosophy of reductionism par excellence. And the title is well earned when one recalls the string of reductionist programs that have left their mark on the first part of this century, ranging from the purported analytical reductions proposed by phenomenalism and behaviorism, to the weaker theoretical reductions of the later generations. Yet, starting with sporadic suggestions in the 1960s and 1970s, the philosophical literature is now rife with pronouncements of the wrongheadedness of all reductive programs. Perhaps surprisingly, the current literature associated with analytical philosophy is being swept by a wave of antireductionism.

Reductionism might be dead or dying, but the idea that certain entities we seem to talk and think about depend on others for their existence (and that they are somehow less real?) is still alive and kicking. This had led philosophers to search for a topic-neutral nonreductive dependence relationship that can be easily incorporated into the analytical toolbox of a variety of philosophical endeavors, performing at least part of the function reductive relationships were supposed to fulfill. Hence the recent philosophical interest in supervenience, which purports to be precisely this sort of relationship. Although the concept that the modern use of ‘supervenience’ aims to express has been around for some time, widespread interest in it is a relatively recent phenomenon.

Bertrand Russell (1924) placed at the heart of his logical atomism what he called “the supreme maxim in scientific philosophizing”: “Wherever possible, substitute constructions out of known entities for inferences to unknown entities” (p. 326). Rudolf Carnap took this maxim as his motto in explicating the logical structure of the world (1967, p. 5). Twentieth-century metaphysicians have often adopted Russell's maxim tacitly in pursuing reductive strategies.

In an increasing variety of areas, however, a consensus has grown that such strategies face dim prospects. We are still novices at neuroscience, but philosophers of mind no longer have much hope of reducing mental language to physical language. Physics has not yet found a unified theory of basic physical phenomena, much less a theory unifying them with the manifest image of the world, but philosophers of mind have lost faith that our everyday discourse about the world around us will eventually reduce to the language of an ideal physics.

A loss of faith in reduction has not led to a revival of faith in dualism, for its postulation of a realm of immaterial mental entities that have causal powers to affect each other and the material world seems unscientific and without explanatory power. To many, physicalism has remained the only respectable attitude toward the mind–body problem. The challenge has been to find a nonreductive physicalism that resists any commitment to immaterial entities while allowing that psychology and other higher-level theories are autonomous from physics.

Meanings

An attractive semantic view invokes abstract entities as the meanings of predicate expressions. According to this sort of view, people who use predicates enter into a three-term meaning relation in which a person associates a predicate with an entity that is the predicate's meaning. These entities may be Fregean senses, Platonistic properties, or some other sort of universal. The role they will have in the present inquiry is so schematic that differences among them do not matter. The idea will become clear enough from the sort of work that is supposed to be done by these ‘Meanings,’ as they will be called here.

A semantic approach that appeals to Meanings as the meanings of predicates is familiar. Let us briefly review a few important aspects of it. If it is said, for instance, that some object ‘has a spherical shape’, then this predication attributes to the object the Meaning being spherically shaped. This particular abstract entity is what is meant by the predicate, because it is the entity with which English speakers enter into the mental relation of meaning, or intending, by the predicate. Quite similarly, when someone asserts in English, ‘The Loch Ness monster is a dragon’, this is an attempt to predicate being a dragon of something that lives in Loch Ness, because that Meaning is what is intended by English speakers who predicate the expression ‘is a dragon’.

Contemporary physicalists typically reject two opposing positions concerning the relation between physical and psychological properties: Cartesian dualism and reductive physicalism. In doing so, they embrace some kind of nonreductive physicalism entailing that psychological properties depend on, but are irreducible to, physical properties. This essay assesses the viability of such nonreductive physicalism. In particular, it examines whether nonreductive physicalism can, with help from certain nonreductive supervenience relations, steer a clear course between Cartesian dualism and reductive physicalism. We shall see that the course is far from clear, owing to complications facing a physicalist account of the pertinent notion of dependence. Physicalist psychofunctionalism in particular, we shall contend, raises serious problems for current conceptions of nonreductive mind-body supervenience relations, owing to its requiring relatively local causal mechanisms for the occurrence of psychological properties. Let us begin with sketches of the two positions nonreductive physicalists aim to avoid: dualism and reductionism.

Dualism and Independence

Cartesian dualism affirms the ontological independence of the psychological from the physical, in a sense to be specified. Cartesian substance dualism implies that psychological substances (e.g., thinking individuals) do not depend ontologically on physical substances or properties. Cartesian property dualism implies that psychological properties – psychological features that can be exemplified by individuals – do not depend ontologically on physical substances or properties. Cartesians, following Aristotle, might distinguish substances and properties roughly as follows: properties are predicable of things and are multiply realizable, but substances are individuals that are not thus predicable or realizable.

In recent years, philosophers bent on defending physicalism have been attracted to the notion that apparently nonphysical features of our world supervene on its physical features. Mental and moral properties, for instance, though not perhaps identifiable with or reducible to physical properties, are thought nevertheless to depend on, and to be determined by, physical properties. More generally, supervenience claims are taken to range over collections or “families” of supervenient properties (A-properties) regarded as depending on distinct families of subvenient B-properties. I shall designate claims of this sort “A/B-supervenience” claims. Sometimes the supervenience relation is distilled into a pair of slogans: “No A-difference without a B-difference”; “Two objects identical with respect to their B-properties must be identical with respect to their A-properties.”

According to Richard Miller, appeals to supervenience on behalf of physicalism are empty; it is “trivially” true that “the nonphysical supervenes on the physical.” Worse,

Miller advances a pair of interesting and important claims. First, the supervenience of the nonphysical on the physical is “trivial.” Second, the supervenience relation is invariably symmetrical: when As supervene on Bs, the supervenience of Bs on As is “all but guaranteed.”

We do not need an individualistic or Cartesian psychology in order to develop satisfactory psychological explanations for the workings of the physical mind. More pointedly, there are various ways in which a Cartesian psychology would be explanatorily impoverished relative to its wide rivals, and this makes individualism not only unnecessary but implausible. Part of my general argument has involved bringing out non-individualistic strands to existing psychological explanations, using these as a check on the normative claims that individualists have made about psychological explanation and mental causation; part has involved rethinking the more general claims on which individualists have based their views about psychology in particular. Since one of the central motivations for individualism is a commitment to some form of physicalism or materialism about the mind, many of these more general claims have been made about the nature of science and scientific explanation, these being paradigms of triumphs in our attempts to understand the physical world.

In arguing that a science of the mind does not require a Cartesian psychology, I have neither rejected materialism nor claimed that individualists suffer from so-called scientism about psychology, that is, a faith in science to provide all of the answers that it is reasonable to ask about the mind. This book expresses no scepticism about the cognitive sciences, only a reluctance to accept all that has been claimed on their behalf and a rejection of certain construals of what an interdisciplinary, scientific understanding of the mind must be like. It is not an external critique of individualism but a challenge to individualism on its own terms.

Although the debate over individualism has attended to empirical research in psychology, it has not focussed explicitly on the criteria for evaluating causal explanations. In this chapter I introduce two related aspects to explanatory power, causal depth and theoretical appropriateness, and offer the following argument against individualism. We do and should, ceteris paribus, prefer causal explanations that exhibit these explanatory virtues. When other things are equal, wide psychological explanations of behavior are sometimes causally deeper and more theoretically appropriate than their narrow rivals. Therefore, in at least some cases, we should reject narrow explanations in favor of wide explanations of behavior because of the latter's greater explanatory power.

Some of the details of this argument against individualism also provide reasons for rejecting neurosdentific explanations in favor of intentional explanations of behavior; the argument is thus secondarily directed at eliminative materialism, particularly versions that concentrate on our folk psychology (P. M. Churchland 1981; P. S. Churchland 1986; Ramsey, Stich, and Garon 1991). Individualists and eliminativists both ascribe primacy to intrinsic physical properties in scientific explanation, and so focus on the issue of how a properly scientific psychology is to be integrated with the developing neurosciences and computer science. This chapter aims to show why this view and the focus it engenders should be rejected.

The complete argument is developed in four stages. First, I provide an account of the explanatory virtues of causal depth and theoretical appropriateness (§1).

A central part of my critique of the argument from causal powers examined in Chapter 2 was that it equivocates on ‘causal powers’, using that notion in both its extended and restricted senses. I claimed that this equivocation reflects a deep problem in this a priori argument for individualism, one that indicates a fundamental incompatibility in the claims that need to be true for any version of this argument to be sound. Less explicit in that chapter was my more general scepticism about the prospects for a priori arguments for individualism. In this chapter, I argue more directly for both the depth of the identified equivocation and this more general scepticism.

A PRIORI ARGUMENTS FOR INDIVIDUALISM

The intuition that causal powers occupy a special place in taxonomy and explanation was expressed in the argument from causal powers by the claim that sciences taxonomize by causal powers (global individualism). One reaction to my critique of the argument from causal powers that I have found common in discussion is to grant the basic points of the critique (e.g., concede that global individualism is false) but to suppose that there is some other, closely related argument from a premise about causal powers to individualism that is immune to the critique. The frequency of this reaction reflects the strength of the intuition that causal powers occupy a special place in scientific taxonomy and explanation.

Implicit in a number of views we have already discussed is the idea that a proper understanding of the nature of causation leads one to individualism. For example, Fodor claims that global individualism follows from relatively uncontroversial claims about causal explanation (Chapter 2); McGinn considers his analysis of causal factors into powers and parameters to be a ‘rather elementary’ point about causation (Chapter 5). But thus far we have not focussed on the intuition behind such views: that making sense of mental causation requires accepting individualism.

There are two complementary parts to this intuition. One part is the idea that denying individualism commits one to unacceptable views about mental causation. For example, it has been said that denying individualism is tantamount to positing ‘action at a distance’ in psychology, commits one to the existence of ‘crazy causal mechanisms,’ and ‘violates supervenience’. These claims have found their way into the philosophical subconscious and contribute to the intuitive pull that individualism has for many philosophers. If denying individualism committed one to any of these views, individualism would be compelling, if only by default, in much the way that it is often held that some version of materialism about the mind must be true because the various forms of dualism are metaphysically unacceptable. In §1 I state the arguments for and in §2 offer responses to each of these three related objections to the denial of individualism.

The flip side to these objections is the idea that individualism itself follows from a basic and unobjectionable claim about causation: that causation is loca.