Metaphysics

of Mind

metaphysics cartoon

  

Back to Home not avalaible

John Heil - Davidson College  


Introduction

Philosophy of mind is at a watershed. Recent debates are making it increasingly clear that answers to the deepest questions must be sought, not in the laboratories of neuroscientists, but in pursuit of old fashioned metaphysics. These questions are as baffling to the neuroscientists as they are to the philosophers.

Please do not misunderstand. I am not disparaging neuroscientific endeavors. On the contrary, as a materialist, I firmly believe that an understanding of the mind depends ultimately on our understanding the brain. Nor do I think a sharp line can be drawn between philosophy and the empirical sciences. I am convinced, however, that issues now at the forefront of the philosophy of mind are fundamentally metaphysical in character. (This is scarcely surprising. The philosophy of mind is, after all, a kind of applied metaphysics. It is crazy to think that philosophers of mind can ignore or remain neutral on questions of ontology.) These issues include

• the nature of mental properties and events and material properties and events, and the character of relations among these;

• the possibility of mental causation;

• epiphenomenalism and the problem of causal relevance;

• the nature of consciousness and qualities of conscious experiences, and relations these bear to the material world.

It is tempting to attack such issues individually. Picking and choosing among ontological theses by assessing their contribution to a solution of one or another problem in the philosophy of mind is ill-advised, however. Before a detailed account can be offered to the problem of mental causation, we should be clear on the nature of causation. Before we can make sense of the question whether mental properties are ‘causally relevant’ to the production of physical events, we need a sensible conception of properties and the bearing of these on causal transactions.

One worry, of course, is that a strategy of seeking answers to ground-level metaphysical issues before turning to questions in the philosophy of mind could easily result in our failing ever to get back to questions about the mind that excited us in the first place. My suggestion, however, is that questions that occupy philosophers of mind are at bottom special cases of more general metaphysical questions. Once this is appreciated, we are in a position to recognize assumptions that have colored ways we have tended to pose these questions and thereby influenced our conception of the space of acceptable answers. I can illustrate what I have in mind by reference to three largely unexamined theses that have done much to shape recent discussion in the philosophy of mind. I believe that each of these theses is false, and I suspect that many philosophers would find them unattractive. Their influence, however, stems in part from their remaining largely unexamined hence unavailable for criticism.

I have no illusions here. I do not imagine that we need only shift attention to metaphysics to tame all the recalcitrant puzzles in the philosophy of mind. We can be optimistic, however, that a move to consider metaphysical theses underlying familiar themes in the philosophy of mind might encourage ontological candor, a virtue noticeably absent in much recent work on the nature of the mind.

It is easy to be cynical about all this. Philosophers, it is said, are better at posing questions than answering them: there is no progress in philosophy. I am not so sure. The history of philosophy is replete with theories that provided satisfying answers to questions regarded by their authors as especially pressing. Notoriously, each of these leaves unanswered and unsolved a host of distinct problems. This is unsurprising. Theories of mind, for instance, were devised to enable us to cope with particular issues that were, at the time of their introduction, considered to be central. To the extent that a theory is successful, the problems on which it bears recede into the background, and those it leaves unresolved become salient. The result: a spate of new theories addressed to new issues together with the illusion that nothing has been accomplished.

Predicates and properties

The first, and in some respects the most pernicious, thesis can be formulated as a principle.

(F) When a predicate applies truly to an object, it does so in virtue of designating a property possessed by that object and by every object to which the predicate truly applies.

Although few philosophers are prepared to defend (F) outright, the influence of (F), or something very like (F), is patent. Some theorists seem willing to embrace a version of (F), provided the predicates in question are ‘projectable’ or figure in laws or lawlike generalizations. Because I doubt that qualifying (F) in this way affects what I have to say about it, I shall ignore this refinement for the moment (although I shall return to it later). In any case, officially endorsed or not, (F) has colored the discussion of mental properties in ways that outstrip its seeming banality.

The idea behind (F) is easy to grasp. Suppose you are a ‘realist’ about a particular predicate: ‘is in pain’, for instance. In that case, you will want to distinguish yourself from anti-realists. An anti-realist might hold that ‘is in pain’ serves a purely expressive function or that it is a mere ‘projection’ of an attitude toward some object or state of affairs. Realism about a predicate requires that the predicate ‘correspond’ in some fashion to something in the world. This something might be a property. Such reflections lead us to the idea that ‘is in pain’ designates a property, something common to every object to which the predicate truly applies.

So far so good. Prodded now by functionalists, we notice that objects to which we confidently apply ‘is in pain’ conspicuously lack a unique material property that might answer to the predicate. Perhaps human beings share complex neurological ‘pain properties’. But creatures belonging to other species, crustaceans, for instance, or mollusca, although apparently capable of feeling pain, possess vastly different kinds of nervous system. Perhaps, then, ‘is in pain’ is disjunctive in the way ‘is Jade’ is disjunctive as between jadite and nephrite.

A solution of this sort is unsatisfactory as functionalists and others of an anti-reductionist bent have made clear. In the first place, given the variety of creatures to which we seem warranted in applying the predicate ‘is in pain’, we should have to countenance an embarrassingly unwieldy disjunction. This tends to undermine the idea that ‘is in pain’ designates a property, something shared by its possessors. In the second place, we can apparently envision endless possible beings—creatures with silicon bodies, creatures with metallic constitutions, even disembodied creatures—to whom we might find it natural to ascribe pains. If the predicate ‘is in pain’ designates a disjunction of properties (or—whatever this could mean—a disjunctive property), we should like to know what the unifying principle—the principle warranting applications of the predicate to new kinds of case—could possibly be.

Functionalists propose a neat solution, one that seems to have satisfied nearly everyone, even opponents of functionalism. Predicates like ‘is in pain’ designate ‘higher-level’ ‘second-order’ properties of objects to which they apply. Functionalists take these ‘higher-level’ properties to be functional properties. The interesting move here is not the move to talk of functions, however, but the move to ‘higher-level’ properties. In this context, a ‘higher-level’ property is a property possessed by an object in virtue of its possession of some distinct ‘lower-level’ ‘realizing’ property. Functional properties are thought to be examples of ‘higher-level’ properties, although we need not suppose that every ‘higher-level’ property is a functional property.

We can say, then, that every object to which ‘is in pain’ truly applies shares a property, albeit a ‘higher-level’ property, one grounded in a variety—perhaps an endless variety—of ‘lower-level’ realizers. I am suggesting that functionalist arguments in support of this kind of ‘multiple realizability’ have persuaded many philosophers, including philosophers who disdain functionalism, that mental properties, although perhaps ‘supervenient’ on (that is, dependent on and determined by) ‘lower-level’ material properties are not themselves materially reducible. (Philosophers often accept the conception of property ‘levels’ promoted by functionalists but reject functionalism because they take mental properties to be irreducibly ‘broad’ or context-dependent, or because they suspect that functionalism excludes an elusive qualitative dimension of conscious experiences.)

Now, why should anyone think that ‘is in pain’ designates a ‘higher-level’ property? You might reason as follows. If we are realists about pain, we must suppose that ‘is in pain’ designates a property shared by every object of which it truly holds. It is unlikely, however, that (hokey disjunctive properties aside) every object to which we could apply this predicate shares a ‘lower-level’ material property in virtue of which the predicate applies to it. (The latter clause is important: objects to which the predicate applies could well share endless ‘lower-level’ properties. The pertinent question is whether they share a ‘lower-level’ property in virtue of which they satisfy the predicate ‘is in pain’.) We might suppose, then, that ‘is in pain’ designates a ‘higher-level’ property. This property unifies otherwise disparate instances of pain, instances that would appear utterly heterogeneous otherwise.

Alternatives to a view of this sort have not been warmly received. If we deny that ‘is in pain’ designates a property—a ‘higher-level’ property—we seem committed either to some other form of anti-realism about pain, eliminativism, for instance, or to an implausible reductionism. On the positive side, predicates like ‘is in pain’ are projectable; they figure in counterfactual-supporting generalizations that look for all the world like laws: psychological or ‘psycho-physical’ laws. To be sure, these laws are not exceptionless, but neither are the laws of biology, or geology, or meteorology. In any case, predicates that figure in such laws—and ‘is in pain’ is a winsome candidate—have a prima facie claim to be regarded as designating properties.

What has any of this to do with principle (F)? We are being led by the idea that every respectable predicate designates a property. If we are convinced that a predicate like ‘is in pain’ is respectable, and if we cannot discover an obvious ‘lower-level’ (or ‘first-order’) property for it to designate, then this idea, codified in (F), will induce us to posit some ‘higher-level’ (or ‘second-order’) property to serve as the item named by the predicate.

Let me remind you of the fallout from this style of top-down ontologizing. First, there is the problem of ‘causal relevance’, the Gen-X version of the venerable mind–body problem. If mental properties are ‘higher-level’ properties dependent somehow on distinct lower-level realizers, then how could it be possible for mental properties to figure in causal transactions? If we assume that causal relata are events, and we have a relaxed conception of events, then we might allow that mental events (events possessing mental properties) can cause other mental events and physical events (events possessing physical properties). But now we are left with a problem of causal relevance: granted, a mental event can be causally efficacious, is its being mental—is its possessing some mental property—causally relevant to its effects?

Philosophers have offered ingenious solutions to the causal relevance problem, but none of the solutions advanced has attracted more than a handful of adherents. The mood is gloomy. We seem driven to choose between some wild reductionist thesis, on the one hand, and, on the other hand, epiphenomenalism concerning mental properties. In all of this I detect the hidden influence of principle (F).

What about (F), then? Taken on its own terms, (F) lacks plausibility. You need not be an ardent Wittgensteinian to doubt that meaningful predicates in general designate properties. ‘Is a game’ is only the most familiar example. But think of ‘is a stone’, ‘is red’, ‘is hot’, or ‘is a tree’. These predicates seem uncontroversially to hold of objects. They seem, moreover, to hold of objects in virtue of those objects’ properties. They do not seem to designate properties shared by or common to every object to which they apply, however. Must we then be anti-realists about games, stones, colors, temperatures, and trees? No, not unless we are convinced in advance of the truth of (F). Do objects to which such predicates apply then have nothing in common (or nothing in common in virt ue of which he predicates apply)? And, if not, what accounts for their apparent unity? What accounts for the ‘projectability’ of many of these predicates? What allows for them, or some of them, to figure in lawlike, counterfactual supporting generalizations?

Here is one story. Many predicates apply to objects in virtue of properties possessed by those objects. This idea is implied by, but weaker than, (F). It constitutes what is right about (F). Now envisage a predicate—‘is red’ will do—that applies to different objects, not because those objects share a property but because each possesses a similar property. The similarities captured by a predicate can be well-behaved (as I suspect those falling under ‘is red’ are) or they can be matters of family resemblance (as might be so with a predicate like ‘is a tree’). The function of these predicates is not to designate properties shared by objects, however, but to call attention to any of a range of similar, though not precisely similar, properties. The vast majority of the predicates we use to describe our world seem to be of this kind.

In saying this, in rejecting (F), would you be going anti-realist about all these predicates? Not at all. You could continue to think that there really are mind-independent trees, stones, hot bodies, and red things. Objects falling under each of these predicates are similar in some respect (this respect is the property in virtue of which the predicate applies), but they are not exactly similar. The similarities in question include causal or dispositional similarities. Red objects reflect light in similar, although not precisely similar, ways (and so yield similar, although not precisely similar, responses in observers). They do so because they possess similar, although not precisely similar, properties. It is in virtue of their possession of these properties that they reflect light as they do. The properties are similar, but not perfectly so. (If properties are universals, then, if A and B are precisely similar properties , A B; more on universals below.) No wonder then that objects possessing these properties behave similarly; no wonder predicates they satisfy are projectable; no wonder they can figure in formulations of counterfactual-supporting lawlike generalizations.

Abandoning (F) and the philosophical baggage that accompanies (F) is a liberating experience, one anybody engaged in the philosophy of mind who is not already heavily invested in the fruits of (F) might try it. The thought that giving up the idea that mental predicates designate mental properties in the sense mandated by (F) will seem reductionist or eliminativist only to someone beguiled by (F). Again: relinquishing (F) does not require an anti-realism about mental predicates, nor does it oblige you to define or analyze mental predicates in purely non-mental terms. The position encouraged here is reductionist in another sense perhaps. It is ontologically reductive. It foster s a single-level (or, better, a no-level) ontology of properties. It affects, not the claims advanced by psychologists or neuroscientists, but our conception of what grounds those claims.

Universals

According to (F), when a predicate applies truly to an object it does so by virtue of designating a property possessed by that object and shared by every object to which the predicate truly applies. Talk of ‘sharing’ properties or of having ‘the same’ property leads philosophers to visions of universals. Consider two spatially noncontiguous objects at a particular time. Suppose both objects possess the same property (both are spherical, for instance, or positively charged). If we read ‘the same’ as registering strict identity, then there is something, some one thing, both objects share. You and I can share a pizza or an umbrella, but the sharing of properties is not like this. If distinct, noncontiguous objects share a property at a time, then the property is wholly present in each object at once. When you and I share a pizza, you and I eat distinct slices of the pizza; if you and I share an umb rella in a rains torm, you are under one part of the umbrella, I am under another part. In contrast, when two objects share sphericity (construed as a universal), sphericity is not partly in one and partly in the other; sphericity is wholly present in each.

This is a mystery I am not in a position to illuminate. In any case, I intend to sidestep the problem by suggesting that we need not regard properties as universals. True, we speak of objects ‘sharing’ properties and of their having ‘the same’ property. We speak of people as sharing a taste for Wagner, however, or wearing the same dress to a party. In such cases, we seem not to have in mind strict identity. Two women who appear at a party wearing the same dress appear wearing exactly similar dresses. If you and an acquaintance share a taste for Wagner, then you are both similarly disposed toward Wagner.

Perhaps this is all we mean when we speak of objects sharing properties or possessing the same property. Two spherical objects would be exactly similar with respect to their shapes. But the sphericity of one object need not be strictly identical with the sphericity of the other. Devotees of universals will speak here of distinct ‘instances’ of a single property. But maybe properties are nothing more than particular instances. In that case, we collect similar instances together under a single predicate as ‘tokens’ of a single ‘type’, but it would be a mistake to imagine that this requires us to regard a property either as something in addition to its instances or as something wholly present in each of its instances.

If you take properties in this way, you are not thereby denying that properties exist, only that properties are universals. You are thinking of properties as ways particular objects are. The baseball Mark McGwire hit for his 69th home run in 1998 is spherical; its sphericity is similar to, though numerically distinct from, the sphericity of the baseball McGwire hit for his 70th home run. This is to treat properties as ‘tropes’, although I shall avoid this label here. Many self-described ‘trope theorists’ extend the view in unsavory directions. Trope theorists tend to be ‘bundle theorists’, for instance, holding that objects are nothing more than ‘bundles’ of ‘compresent’ tropes. We can steer clear of these murky waters, however, in granting just that talk of properties, talk of objects sharing properties, does not imply that properties are universals.

Ah, but taking properties to be universals might be thought to provide important theoretical benefits we would miss otherwise. If objects share the same property (in the sense that a property that one possesses is strictly identical with a property all the others possess) we can understand why the objects would be similar and behave similarly qua bearers of that property. Universals, in this way, afford us an understanding why objects are as they are, and why similar objects are similar. If you reject universals, however, you do not thereby lose these explanatory advantages, you merely transform them. Objects possessing exactly similar properties can be expected to be similar and behave similarly (at least qua bearers of those properties). This will be so whether the properties are taken to be strictly identical or taken to be exactly similar.

What you lose when you abandon the idea that properties are universals is a reduction of similarity to strict identity. Objects are similar with respect to their properties. Crimson objects are similar with respect to color; spherical objects are similar with respect to shape. If you take properties to be universals, then this similarity stems from identity: the objects are identical—strictly identical—in some respect. If, in contrast, you take properties to be particularized ways (ways particular objects are), you will regard this similarity as primitive and intrinsic. Objects are similar when they possess similar properties; properties are similar tout court. This ball and that ball are similar with respect to their sphericity. This ball’s sphericity and the sphericity of that ball are not similar with respect anything: they are flatly similar. This kind of similarity, note, is an objective, mind-independent, fundamental feature of our world.

Similarity comes in degrees. Properties can be exactly similar or less-than-exactly similar. Two less-than-exactly similar complex properties might have exactly similar constituents. But even utterly simple, non-complex properties, if there are any, can be more or less similar to one another. Pretend for a moment that being red, being yellow, and being blue are simple properties. Red is more similar to yellow than either is to blue, though the similarity falls well short of perfection. Devotees of universals who admit that non-complex universals could be similar, must admit this kind of irreducible, intrinsic similarity relation. The mere fact that such a relation is required by a conception of properties as ‘particularized ways’, in no way places such a conception at a disadvantage.

These are heady issues. My point here, however, is just that it is far from obvious that you lose any kind of theoretical advantage when you treat properties as particulars. And of course you gain the advantage of not having to countenance entities that can literally be wholly present in many non-contiguous places at once.

You may think this unfair. Perhaps you are attracted to the idea that universals are ‘Platonic’ entities: ‘forms’ with distinct spatio-temporal instances. The instances are numerically distinct particulars, but the universal is a single entity. If you hold a view like this, you owe the rest of us an account of the instantiation relation, a relation obtaining between a non-concrete, non-spatio-temporal entity and instances of that entity. It is difficult to feel optimistic about the success of such a project.

Let us take stock. You can choose between properties as in rebus universals, properties as transcendent (ante res) universals, and properties as particulars. (I leave aside nominalist views according to which there are no properties or properties are identified with classes of objects.) Universals of either sort allow a simple answer to one kind of similarity question, reducing similarity to the strict identity of universals. If you take properties to be particulars, you must accept property similarities as irreducible and primitive. This might seem a liability until you recognize that, if you grant that non-complex universals could exhibit degrees of similarity, you will have to posit irreducible similarity relations among universals as well. All things considered, then, it is hard to see that any envisaged advantage of appeals to universals could outweigh their liabilities. Such considerations favor a conception of properties as particulars: ways particular objects are.

Perhaps you are worried, as many philosophers profess to be, about ‘individuating’ properties construed as particulars. Indeed the mere formulation of the individuation question expressed in a certain tone of voice is sometimes regarded as a devastating criticism of the view. The issue goes beyond the scope of this discussion, but there are a few things to note here. First, problems about individuation arise, if at all, for simple, non-complex properties. Complex properties, like complex objects, can be individuated by reference to their constituents. Second, simple properties can be partly individuated by reference to the objects to which they belong: if A and B are properties of distinct objects, they are distinct properties. Because it may be necessary to appeal to properties in the individuation of objects, there is a risk of circularity here. That, however, is to be expected when you are dealing with the most fundamental ontological elements, those in terms o f which conditions of individuation of less fundamental items are to be given. Third, following a suggestion of David Robb, you might say that A = B just in case A and B spatially and temporally coincide and are exactly similar.

More needs to be said on this topic, but perhaps I have said enough to dispel the impression that a view according to which properties are particulars—ways particular objects are—is a hopeless nonstarter. This is enough for my purposes.

Before moving on, let me point out a kind of synergy between Principle (F) and the doctrine of universals. I have suggested that you could accept (F) without accepting a commitment to universals, so long as you construe talk of ‘the same’ property or of a property’s being ‘shared’ as expressing similarity rather than strict identity. It is nevertheless quite natural to read (F) as ranging over universals. Likewise, anyone predisposed toward universals could well find (F), or the thought (F) expresses, attractive. Universals incorporate an element of weirdness, so why should we regard with suspicion the idea that predicates, generally, express universals?

I propose now to move on to consider what I regard as a third dogma pervading the philosophy of mind. As I hope to show, a discussion of this dogma will also reveal less obvious features of properties as they might be conceived by someone put off by universals.

Properties, dispositions, qualities

Are properties dispositional or categorical? Some philosophers contend that properties are dispositional: every property makes a particular kind of contribution to the ‘causal powers’ or dispositionalities of objects possessing it. Other philosophers argue that properties, or at any rate first-order properties, are categorical (where ‘categorical’ means ‘non-dispositional’). Either there are no dispositional properties; or dispositional properties ‘supervene’ on categorical properties as ‘second-order’ properties; or perhaps laws of nature empower objects in virtue of those objects’ categorical properties. Still other philosophers, hoping to claim the middle ground, argue that some properties are dispositional and some are categorical: neither dispositionality nor categoricality requires independent grounding.

What is rarely noticed is that there is another option: every property is both dispositional and categorical. If we take ‘categorical’ to mean ‘non-dispositional’, then, following C. B. Martin (who is following Locke), it is less misleading to express this view as holding that every property is both dispositional and qualitative. (Henceforth, I shall use ‘qualitative’ in favor of ‘categorical’.) This is not the thesis that every property is a complex entity consisting of a dispositional component and a qualitative component. A property’s dispositionality is not a part or a property of that property; it is that property. Similarly, a property’s qualitativity (if I may) is not a part or a property of that property; it is that property. This yields what Martin calls ‘the surprising identity’, what I shall call the identity thesis. According to the identity thesis, dispositionality does not supervene on qualitat ivity, or vice vers a, at least not in the sense in which one is the ground for the other. (Of course, given that a property’s dispositionality and qualitativity are identical with the property each trivially supervenes on the other.)

The traditional debate over dispositionality exhibits a rigid dialectic. One side, noting that non-dispositional properties could make no difference to their bearers, argues that every property (or every first-order, or real, or natural property, I shall ignore such qualifications here) is dispositional, and concludes from this that properties are purely dispositional. The other side points out that a world in which every property is purely dispositional is a world in which nothing is ever accomplished: the manifestation of every disposition would itself be a pure disposition for the manifestation of a pure disposition for…. From this it is concluded that properties, or at any rate the basic ones, must be qualitative, hence non-dispositional. A friend of the identity thesis, however, would contend that both sides are right, both wrong. Pure dispositionality and pure qualitativity are equally mythical. Every property distinctively affects both the dispositionality and the quality of objects to which it belongs. We can consider a property’s dispositionality or its qualitativity but in so doing we are just differently considering the selfsame property.

A view of this kind can help itself to the negative arguments of both sides of the traditional debate. A positive defense might begin with simple examples. Consider sphericity. Sphericity would seem to be a paradigmatic quality. Spherical objects differ qualitatively from non-spherical objects. But spherical objects also differ from non-spherical objects dispositionally, and their so differing stems from the presence or absence of sphericity. It seems natural to conclude that sphericity—the property of being spherical—is both qualitative and dispositional. We can separate these in thought, in the sense that we can consider sphericity as a quality (or disposition) without thinking of it dispositionally (or qualitatively). Indeed philosophers often do exactly this.

Such reflections might tempt you to regard sphericity’s qualitativity and dispositionality as themselves properties, perhaps distinct second-order properties of a single underlying grounding property, or perhaps as distinct but co-occurring properties (where the co-occurrence might be nomologically or metaphysically guaranteed). You might be led down one of these unpromising paths if you believed something like this

(S) If an entity, e, can be considered, now as f (and not y), now as y (and not f), e must possess distinct properties corresponding to f and y.

A proponent of the identity thesis will reject (S). Such a philosopher will think of the case of sphericity as a potential counter-example to (S). Of course, a proponent of (S) will regard the identity theorist’s take on sphericity as question-begging. So it goes when you get down to the basic issues in ontology.

Sometimes, in considering an object now one way, now another, we concern ourselves with distinct aspects or properties of the object. You can consider a tomato as something red or as something round. You can, as well, consider the tomato’s redness and its roundness, two properties of the tomato. But it does not seem obvious that in considering sphericity, now as a quality, now as a power, you are considering distinct aspects or properties of sphericity. Rather you are considering one and the same property—sphericity—in two ways. This might be what you do when you see, for instance, a Necker cube now one way and now another.

Necker Cube

In this case you consider the very same ‘aspects’ or properties of the drawing, now one way, now another.

My aim is not the impossible one of providing a knock-down argument to the conclusion that every property is, at once, dispositional and qualitative. I want only to get the view on your radar. Now that it is there, imagine for a moment that it is correct. What follows? One immediate consequence is that we are no longer faced with a mystery as to why certain qualities bestow particular ‘causal powers’ on objects possessing them. The connection between qualities and powers has long been regarded as contingent. If we accept the identity thesis, we turn our backs on this vintage prejudice. This might worry you. I suspect we are inclined to see qualities and powers as contingently related so long as we focus attention on complex secondary qualities, colors, for instance, or odors. A proponent of the identity thesis will want to start with the most basic properties. The idea is that here, at the basic level, a property’s dispositionality and qualitativity are clearl y exhibited as the property itse lf differently addressed. From these basic cases, we move to progressively more complex cases and arrive eventually at colors and odors.

I put it to you: is it so obvious that it is possible to have two objects that are qualitatively indiscernible, yet discernible with respect to their dispositionalities; or two objects dispositionally indiscernible but qualitatively discernible? Humeans have theoretical reasons for thinking that such things are possible, but the rest of us need not be moved by those reasons.

Of course if you remain at a relatively high level of abstraction and rely on pure conceivability judgments as a standard of possibility, then many things become possible, among them ‘Zombies’. A Zombie is a being precisely similar to a conscious being with respect to its dispositionalities but lacking consciousness—where consciousness is taken to be a qualitative accompaniment of certain properties, properties in virtue of which a being has a compliment of causal powers or dispositionalities. But if we accept the identity thesis, then sameness of causal powers—exact dispositional similarity—yields qualitative sameness. Zombies, on this view are impossible.

I do not offer these thoughts as a disproof of Zombies. The idea, rather, is to make it clear that the Zombie possibility rests on a substantive thesis concerning the nature of properties. At the very least, this thesis requires defense. The usual strategy of attacking (or defending) Zombies by sticking to higher-level themes in the philosophy of mind disguises this point. If nothing else, the discussion makes clear that we cannot hope to settle fundamental issues in the philosophy of mind without confronting ontology in a serious way, that is, on its own terms. When we do this we can see that the identity thesis deserves a run for its money. On the face of it, the thesis fits a plausible ontology of properties (whether, note, we take properties to be particulars or universals). It makes sense of phenomena that appear mysterious otherwise. And it has a kind of internal plausibility that can seem attractive to anyone not already committed to a going view.

Concluding remarks

I contend that the philosophy of mind could benefit from our taking metaphysics more seriously. This requires a willingness to accept ontology on its own terms and an openness to possibilities marginalized by orthodox philosophy of mind. Top-down metaphysics—metaphysics tailored to yield outcomes favored by theory-wielding philosophers of mind—is the tail wagging the dog. It inhibits progress in our attempts to understand the mind and it warps our overall view of the world.


Bibliographic Note

Ideas discussed here have their origins in many sources. They are spelled out in more detail in a forthcoming paper written with David Robb, and they are sketched in Philosophy of Mind: A Contemporary Introduction (London: Routledge: 1998). The latter includes a substantial bibliography as well as compilations of suggested readings on selected topics. D. M. Armstrong’s Universals: An Opinionated Introduction (Boulder: Westview Press, 1989) provides an excellent introduction to the metaphysical background. Despite its title, Armstrong discussion is inclusive and fair-minded. I discuss principle (F) and the ‘no levels’ view in ‘Multiple Realizability," American Philosophical Quarterly 36 (1999): 189–208 . A defense of the identity thesis (every property is qualitative and dispositional) can be found in C. B. Martin’s ‘On the Need for Properties: The Road to Pythagoreanism and Back’, Synthese 112 (1997): 193–231; and in C. B. Martin and John Heil, ‘The Ontological Turn’, Midwest Studies in Philosophy, forthcoming. Zombies were introduced to the philosophical world by Robert Kirk in ‘Zombies vs. Materialists’, Proceedings of the Aristotelian Society, Supplementary vol. 48 (1974): 135–52. (Kirk’s more recent reflections on zombies can be found in his Raw Feeling, [Oxford: Clarendon Press, 1996].) David Chalmers has brought zombies to special prominence in his The Conscious Mind: In Search of a Fundamental Theory (New York: Oxford University Press, 1996).

not avalaible                                                                                              Top of page