This paper appeared in S.C. Gibb, E.J. Lowe, and R.D. Ingthorsson, eds., Mental Causation and Ontology. Oxford University Press, 2013, pp. 215-32.

The Identity Theory as a Solution to the Exclusion Problem
David Robb
Davidson College

This chapter is about a proposed solution to the exclusion problem, one I've defended elsewhere (Robb 1997, 2001; Heil and Robb 2003). Details aside, it's just the identity theory: mental properties face no threat of exclusion from, or preemption by, physical properties, because every mental property is a physical property. Here I elaborate on this solution and defend it from some objections. One of my goals is to place it in the context of a more general ontology of properties, in particular, a trope ontology.

The exclusion problem takes several forms. The version I confront here is generated by three principles:

Efficacy: Mental properties can produce physical effects.

Closure: Only physical properties can produce physical effects.

Dualism: Mental properties are not physical.

The problem is that each seems true, but the triad is apparently inconsistent. These principles depart in some ways from Robb (1997). For example, I here explicitly frame Efficacy (what I used to call Relevance) and Closure in terms of causal production (compare Kim 2007). Properties here are powers to produce characteristic physical manifestations in the appropriate circumstances (Molnar 2003; Heil 2003, ch. 8). It seems to me that it's here the exclusion problem is most clearly a problem of metaphysics, and even, as I hope to show, ontology. There are, granted, less metaphysically loaded versions of the problem, along with corresponding solutions. These appeal not to powers or production, but to, say, explanation (Burge 1993), counterfactual dependence (Yablo 1992; Loewer 2007), or causal intervention (Woodward 2008). I hope that much of what I have to say below sheds light on these alternative conceptions of the problem, but I will not explore these connections here.

I reject Dualism, at least on the most straightforward reading on which the principle is incompatible with the other two (more on this below). Every mental property is physical. In particular, every mental property is the physical property that would otherwise threaten to preempt it with respect to physical (and especially behavioral) effects. More simply, every mental property is what some would call its physical 'base' or 'realizer'. So this is an identity solution to the problem. This proposal, however, needs considerable refinement and defense, my aims in the rest of this chapter. I'll proceed by responding to a number of objections.

O1: Psychophysical property identity was undermined decades ago by the multiple realizability argument (Fodor 1974; Boyd 1980; Putnam 1980). Indeed, the contemporary debate is really just about how to save mental causation in the face of the non-reductive physicalism established by this argument. Proposing an identity theory at this point ignores these results and disengages from the contemporary debate.

R1: According to the multiple realizability argument, mental properties are not physical because they are multiply realizable in the physical. Put another way, when we ascribe mental properties, we abstract away from details of physical implementation. On the functionalist version of this argument, to instantiate a given mental property is to be in some state or other that plays the defining causal role of that mental property. Since this realizer or role-filler state can be any of a variety of physical properties, the mental property cannot be identified with any one of them. Mental properties are thus second-order properties, at best realized in, but distinct from, any physical property.

The objector is right that this argument is almost universally endorsed among those contributing to the mental causation literature. This may be what leads Yablo (1997, 255), for example, to say that accepting Dualism is included in 'the price of admission' to the mental causation debate. But while I grant that the argument is sound, its relevance to the exclusion problem is not so clear.

Start with a few roles properties are thought to play (compare Campbell 1990, 29; Oliver 1996). For example, properties are features; they are, that is, the truthmakers for (some) predications. The truthmaker for 'This apple is round', for example, is a property, roundness. Properties also sometimes appear as types. Here properties play the role of a one over many, something shared by objects of the same type or kind: various round things all have the same property, roundness. And properties also appear as powers, where this role can be indicated by the qua locution: an apple causes a certain kind of impression in soft clay qua round thing--in virtue of its roundness--but not qua red thing, in virtue of its colour.

Now it's at least not obvious that the same sort of entity answers to 'property' or 'roundness' in each case. The three roles--feature, type, power--may not be filled by the same thing. An ontology of properties would work this out. But for now, it's enough to point out that the multiple realizability argument is aimed (and, I grant, succeeds) at distinguishing mental and physical types. What the argument shows is that there is no one-to-one match between mental and physical types. So the psychophysical 'property dualism' warranted by this argument is type dualism. It remains open whether there is any such dualism of features or powers. But this question is central to the exclusion problem, for it's clearly properties as powers that appear in Closure and Efficacy. For all that's been said so far, type dualism is compatible with these principles.

Now it could turn out that features, types, and powers coincide, that the same sort entity fills all three roles. But this is a heavyweight ontological claim, one that goes far beyond O1's seemingly innocent appeal to the multiple realizability argument. And this is where a trope ontology becomes directly relevant, for according to this ontology, tropes are features and powers, but something else is a type. Typical among trope theorists is to say a type is a resemblance class of tropes (Campbell 1990; Bacon 1995; Williams 1966) or, what may amount to the same thing, a collection or plurality of resembling tropes. The mental type elation, for example, will be a class of tropes, all of which resemble one another closely enough to count as tropes of that mental type. But it's compatible with this that (1) all of the tropes making up elation are physical, (2) not all of these tropes resemble one another closely enough to count as a physical type, and (3) some of them do resemble one another closely enough to count as a physical type: each cluster of closely resembling tropes will make up a physical type that 'realizes' elation. (For more on this, see R6 below and Robb 1997.) What results is type dualism but trope monism: a psychophysical identity theory of features and powers.

Return then to the exclusion problem and the principles that drive it. Let the 'properties' in these principles appear consistently as powers, so that the principles are incompatible. The proposed solution is to reject Dualism in favour of an identity theory: mental powers are physical. If O1 protests that the multiple realizability argument undermines this identity theory, my reply is that O1 conflates properties as types with properties as powers. (Or if the objector insists on stipulating that the properties appearing in Dualism are types, then my reply is that the three principles are consistent after all, as the other two explicitly concern powers: the principles no longer generate a problem.) These remarks are not intended to establish a trope ontology, but just to show that O1's appeal to multiple realizability is not ontologically innocent: the nature of properties must be confronted directly if one is to claim, as so many do, that the multiple realizability argument blocks an identity solution to the exclusion problem.

O2: Even if the classic multiple realizability argument concerns types, a similar argument can be deployed concerning tropes, and thus powers. A mental trope is more compositionally plastic than any physical trope. Consider an elation trope E and the complex physical trope P with which E is allegedly identical. E could survive the change of a single neuron or particle, while P could not. The tropes are thus distinct, resulting again in Dualism, but this time it's explicitly a dualism of tropes, and so powers (compare Boyd 1980; Pereboom 2002).

R2: Unlike the original multiple realizability argument, this newer version threatens an identity solution to the exclusion problem, for it claims to directly establish a dualism of powers. But the intuitions driving this newer argument are far more controversial than those behind the original version. While it's no doubt true that an elation trope could exist in the absence of P, by what right does O2 claim that E itself, that very trope, could exist without P? If this is supposed to be just a brute intuition, it is one that I do not share.

Still, perhaps I can advance a bit beyond a clash of basic intuitions (though it's admittedly hard to avoid question-begging on this front). The judgments of type identity and distinctness in the classical multiple realizability argument are fundamentally driven by judgments of similarity, and especially imperfect similarity: physically diverse creatures, while not exactly resembling, are similar enough to fall under the same mental type. We can rely here on the fact that higher-level types by their nature abstract away from micro-differences, permitting less than perfect resemblance. But O2 cannot help itself to an analogous claim about tropes, for it's not nearly as clear that what's at the ontological ground level will, like types, permit higher, more abstract layers. Being ground-level and maximally determinate would seem to go hand in hand. Conflicting intuitions aside, then, there seem to be general reasons to be suspicious of the claimed plasticity of mental tropes.

O3: Multiple realizability is not the only reason to be a property dualist. There are several other arguments against the identity theory. For example: Mental properties are irreducibly subjective or private, while no physical property is (compare Jackson 1982). When ascribing mental properties, we are subject to normative or holistic constraints that do not bind us when we ascribe physical properties (Malcolm 1970; Davidson 1980). A being--my 'zombie twin'--could duplicate all of my physical properties yet lack my mental properties (Chalmers 1996). I might have existed disembodied, with all of my mental properties but with no physical properties (Yablo 1990). A psychophysical property identity would have to be necessary, but it appears contingent, and there is no plausible way to explain away this appearance (Kripke 1980). And there are others. These are directly aimed at distinguishing mental and physical features, which, at least for the trope theorist, are powers. One cannot, then, shrug off such arguments as irrelevant to Dualism and the exclusion problem.

R3: These do indeed threated an identity theory of mental and physical features, and thus of powers.[1] If any one of these arguments is sound, then it looks as if Dualism is unavoidable, so that either Efficacy or Closure must be abandoned. I have no Master Reply to these arguments: they must be confronted individually, something I won't attempt here. (I've taken on the zombie in Robb 2008.) However, I will make this more limited reply: much of the mental causation literature is conducted in the context of non-reductive physicalism, the pairing of Dualism with the thesis that the mental is always realized in the physical. The driving argument for Dualism in this context is the multiple realizability argument, which I hope to have shown is not nearly as favourable to a dualism of powers as some imagine. But the arguments in O3 are importantly different, for if they are sound, non-reductive physicalism is false, and a more robust form of property dualism--feature and power dualism--takes its place. In that case the dialectic changes considerably, and the aim of an identity theorist such as myself is not to accommodate the dualist argument--as I try to accommodate (classical) multiple realizability--but to confront it directly.

O4: The identity solution merely relocates the exclusion problem. While psychophysical trope identity may rescue the causal efficacy of mental powers (tropes), it still leaves open whether they are causally efficacious qua mental. Put another way: if mental powers are both mental and physical, why not think they are causally efficacious only in virtue of being physical? Closure would motivate this problem, and we're back with, if not the same exclusion problem, at least very a similar one (Noordhof 1998; Shoemaker 2003, 434; Macdonald and Macdonald 2006, 552-3).

R4: The general thought behind this objection seems to be that we will have solved the exclusion problem only if we finally arrive at something--of whatever ontological category--that's only mental, that is, mental but not physical. Anything that's both mental and physical, the thought goes, invites exclusion worries all over again. So, for example, Davidson's (1980, 1993) desired stopping point is at mental events, which, he says, are physical. But since mental events are both mental and physical, exclusion worries arise, so that we must show that mental events are causes in virtue of their mental features (powers). Similarly, the line goes, if our desired stopping place is with these mental powers, it had better turn out that they're not also physical, for then we get the same problem all over again, so that we must show that mental powers are causally efficacious in virtue of their mental features (higher-order powers). And so on.

But there's something suspect in this general line of objection. If psychophysical identity at the desired stopping place continues to invite exclusion worries, why shouldn't other psychophysical relations at the desired stopping place invite similar worries? Suppose, for example, that mental powers are in fact only mental--that is, mental but not physical--yet are immanent in (Yablo 1997, 275) or nothing over and above (Wilson 2005) the physical, where this may be spelled out in terms of, say, realization (Boyd 1980), the determinable-determinate relation (Yablo 1992), constitution (Pereboom 2002), metaphysical necessitation (Bennett 2008), de re, a priori determination (Jackson 2006), or something else. And suppose that immanence, in whatever form it takes, does in fact secure the efficacy of mental powers with respect to physical effects. Is there still a lingering worry that these powers are efficacious, not in virtue of being mental, but merely in virtue of being immanent in the physical (Lowe 1993, 632-3)? Maybe mental powers are efficacious, but only because they piggyback on their physical base powers. This appears to be, if not the original exclusion problem, one that's very similar to it.

Now the immanence theorists cited above may suspect that these worries are somehow ill-conceived. But then why should it be that when we move to the most intimate form of immanence--namely, identity--these worries are legitimate? It seems to me that whatever one's view of immanence, the way to stop these recurring qua questions is not to ban immanence in the physical at the desired stopping place. It's to show that qua questions--and the exclusion worries that threaten--at that place are somehow illegitimate. And this is where I think an ontology of properties will again be relevant.

In the work cited earlier, I've argued that at the level of powers, qua questions are illegitimate because powers do not themselves have higher-order features or powers. More simply: there are no tropes of tropes. Such higher-order tropes, I've argued, are explanatorily idle, threaten to start a vicious regress, and are just plain odd. Here I'll make just the oddity point. Start again with how the qua questions arise for an event identity theorist such as Davidson: even if mental events cause behavior, they should do so in virtue of their mental features. For example, the feel of a pain or the content of a belief must be efficacious. For this purpose, mental features (powers) are recognized. But now suppose these mental powers are physical, and one wants to raise qua questions again. Is the motive still that we want the mental features of a mental power to do some causal work? But this is what strikes me as odd. A phenomenal trope, for example, doesn't have a qualitative feel--at least not in the sense of having a qualitative feature--it is a qualitative feel. A mental power to cause a bit of behavior doesn't have a power to cause such behavior: it is such a power. To raise qua questions at this level, invoking features of features (powers of powers), looks strange, and appears to commit a category mistake.

A worry closely related to O4 is that while psychophysical property identity may secure the causal efficacy of mental powers, it doesn't secure 'distinctively' mental causation, or causal efficacy for mental powers 'in their own right' (e.g., Lowe 1993, 632; Wilson 2009, 150). But again, this seems to assume that mental powers on the identity theory have a dual nature, so that (in accordance with some suitably modified version of Closure) only the physical nature is engaged when mental properties are causally efficacious. But on the trope ontology, there is no such division within a trope: a mental power, its mental nature, and its physical nature are all one and the same. In this sense, a mental trope is both fully mental and fully physical.

O5: The identity solution advanced here combines psychophysical trope identity with type dualism: one and the same trope can be a trope of two types. But this is impossible, since tropes are individuated by their constituent types. (For more detailed discussion of this argument, see Ehring 1996; Whittle 2007.)

R5: The objector here is apparently thinking of tropes on analogy with Kimian events (Kim 1993), which are individuated, in part, by their constituent types. On such an analogy, if mental and physical types are distinct (and if a trope can have only one constituent type), no mental trope can be a physical trope.

But tropes here are not complex entities with types as constituents. Distinguish two conceptions of a trope (Daly 1997): as a complex entity (a substance's instantiating a universal), or as a fundamental entity. On the latter conception, which I endorse, tropes are of types, but they don't have types as constituents. Such a view is required by a trope ontology, on which tropes are the basic building blocks, the 'alphabet of being'. I take it no building block can have, as a constituent, the derived entities it grounds. And if tropes don't have types as constituents, there's no general barrier to a trope's being of more than one type.[2]

However, while there's nothing in general to prevent a trope's being of distinct types, there would be if the types in question were incompatible. For example, no trope can be a red trope and a green trope. Returning to the issue at hand, O5 might insist that mental and physical types are incompatible, again resulting in psychophysical trope dualism. The dualist arguments from O3, if sound, would deliver this result: if to be mental is, say, to be irreducibly subjective, and to be physical is to be irreducibly objective, then these types would exclude one another. But what about the anti-reductionist argument in play here, the multiple realizability argument? This, I claim, shows only that mental types are not physical; it doesn't show that they are anti-physical, that is, that they exclude the physical.

As a way of fleshing out this point, consider a version of O5 from Shoemaker (2003, 434, note omitted):

If we think of the instantiation of a property as the conferring on something of the conditional powers associated with that property, then when properties confer different sets of conditional powers, the instantiation of one of them is not identical with the instantiation of the other.

If 'property' here means type and 'instantiation' means trope (power) this looks like an argument that mental types are anti-physical in the relevant sense: no trope can be a trope of both a mental and physical type. In reply, I grant that no mental type has the same set of causal powers associated with it as any physical type. (Indeed, one might take this to be precisely the lesson of multiple realizability.) But this does not show that mental types are anti-physical. Suppose the causal powers of one type may be, as Shoemaker himself believes, a proper subset of the causal powers of another (see also Wilson 1999; Whittle 2007). Types standing in this intimate relation won't be incompatible, and this in fact entails that their corresponding tropes will be identical (assuming the background of a trope ontology). Returning to the case at hand, if the causal powers associated with mental type M are a subset of the causal powers associated with one of its physical realizers P, then trope identity follows: every P-trope will be an M-trope. And if M has only physical realizers, every M-trope will be some physical trope or other.

O6: Trope identity and type dualism are nevertheless incompatible. We can show this by assuming trope identity and deriving type identity. Take types to be resemblance classes of tropes, and assume trope identity, so that a given mental type, say elation, is a class of physical tropes. Now if this is to be a genuine type, its tropes must exactly resemble, since (Gibb 2004, 471):

It is only those classes of tropes with the greatest possible degree of unity, that is, sets of exactly resembling tropes, which can be substituted for universals. This can be seen by the formal properties of the relation of resemblance. Whilst all resemblance relations are reflexive and symmetrical, it is only in the case of exact resemblance that the relation of resemblance is transitive.

From here, the route to type identity is quick. The physical tropes making up elation must exactly resemble. But then elation is itself a physical type, for it's no doubt true--perhaps analytic--that exactly resembling physical tropes belong to the same physical type. So if elation collects all and only those tropes belonging to a given physical type, elation is that physical type. Moreover, if all of the tropes making up elation exactly resemble, we've lost any sense in which this mental type is multiply realizable in physically diverse creatures. The upshot is that the trope identity solution is forced into type identity: elation, and every mental type, is a physical type.

R6: It seems to me that this objection goes wrong at the beginning: physical tropes making up mental types needn't exactly resemble. In fact, for the trope theorist, this is the lesson of multiple realizability. Colour provides a standard analogy: while any determinate shade of red (shade type, that is) consists of exactly resembling tropes of that shade, red itself, the determinable, is a class of tropes with more relaxed resemblance standards: some red tropes exactly resemble one another--these exact-resemblance classes form the maximally determinate shades that realize red--but exact resemblance isn't required. For example, the type red contains, say, the scarlet trope of my blanket and the crimson trope of my chair, and these two inexactly resemble. Mental types work in much the same way: some of the tropes in elation exactly resemble. These classes of exactly resembling physical tropes form all of the physical types that realize elation. But exact resemblance isn't required for membership in elation. For example, a trope of human elation won't exactly resemble a trope of Martian or dolphin elation.

By appealing to inexact resemblance, I run up against the passage from Gibb quoted earlier. Following Armstrong (1989, 122-3), Gibb says that it is 'only those classes of tropes with the greatest possible degree of unity, that is, sets of exactly resembling tropes, which can be substituted for universals.' But granting this, I reply that mental types--that is, classes of mental tropes--don't substitute for universals (compare Whittle 2007, 71). Let them instead substitute for what Armstrong calls second-class properties, properties (types) that, while not universals, need to be recognized in any ontology as part of the manifest image. (Armstrong sometimes suggests colours as an example.) The idea is not foreign to trope theory. For example, it seems to have been Williams' (1966, 81) point when he says that classes of inexactly resembling tropes 'provide a less definite universal.' And Campbell (1981, 484) notes that 'The closeness of resemblance between the tropes in a set can vary. These variations correspond to the different degrees to which different properties [types] are specific.'[3] A second-class property, or a less definite universal, or a less specific property, is still a natural--not conventional, not gerrymandered--class of tropes, though it may be, to borrow a term from Lewis (1983), less than perfectly natural. It's just that it's a class of tropes whose requirements for membership are more relaxed than those more determinate classes that, I grant, are uniquely qualified to substitute for universals.

This may be the best place to address the following question: If all of the tropes in elation (or in any alleged mental type) are physical, what makes this a mental type? Put another way, if all of the tropes in this class are physical, what makes them mental as well, and in particular, elation tropes? Here I'm neutral and say: deploy your favoured theory of mentality. Functionalists, for example, will say that what makes all of these elation tropes is that they all have the causal profile definitive of elation. Those inclined toward qualia can say that what makes them elation tropes is their qualitative feel. Those who think intentionality is the mark of the mental can say that what makes them elation tropes is their representational content. I take no stand here on which is these is correct, only that each is compatible with inexact resemblance between tropes of the same mental type.

O7: But why tropes? If the exclusion problem calls for an identity solution, type identity is available, for there are versions of the type identity theory, such as Kim's (1998, 2005), that accommodate multiple realizability. Since an ontology of properties already requires types to fill the role of 'one over many', types might as well be pressed into service as powers as well.

R7: There are some close similarities between the trope identity solution here and the type identity solution favoured by Kim[4], which itself has some affinities with type identity theories from Lewis (1994) and Armstrong (1968). A full comparison between the views would take its own paper, but here I sketch what I take to be the main advantages of a trope identity solution over a Kim-style type identity solution.

First (Robb and Heil 2008, sec. 6.5), the trope-identity solution does not require, as Kim's type identity does, that mental types fragment into many structure-restricted types. On the trope identity solution, there is a single type elation, and many tropes of that type. What unifies them into a single mental type is their (inexact) resemblance. I consider this a slim advantage at best, however, for it seems that the plurality of structure-restricted 'elations' recognized by Kim might also be united into a single type by inexact similarity, even if Kim himself (1998, 111) is not inclined to recognize such unity.

Second, Kim concedes that his own reductivist account of mental causation is second-best (2005, 159):

The best, or the most satisfying, outcome would have been a vindication of mental causation along the lines of nonreductive physicalism; that would have allowed us to retain mentality as something that is causally efficacious and yet autonomous vis-a-vis the physical domain.

But the best outcome, as we saw, is not to be had. The next best outcome, in fact our only hope at this point if mental causation is to be saved, is physical reductionism. Physical reduction would save causal efficacy for mentality, at the cost of its autonomy. Reductionism allows only one domain, the physical domain, but the mental may find a home in that domain.

The suggestion here is that something valuable is lost in the move to a reductionist (or identity) solution to the exclusion problem. Kim says that what's lost is autonomy from the physical, but if autonomy here is just taken to be Dualism, then it's not clear why losing that should be mourned. Perhaps Kim has in mind here the distinctively mental contribution of mental powers, the efficacy of the mental as such mentioned above in R4. But if that's what is lost in Kim's account, then this is a reason to favour the trope identity solution, for there is no such loss on a trope ontology: for reasons given in R4, nothing distinctively mental is missing in either mental tropes or their casual efficacy.

A third reason to favour trope identity solution over its type identity sibling is a familiar ontological worry: it's not clear that types are the right sorts of the things to be powers, as they would have to be if a Kim-style solution is to be tenable. Types play the 'one over many' role for properties, and so whatever exactly types are, they seem to be 'spread out' over their various instances. (This is the case, I take it, even for Kim's structure-restricted types.) Powers, however, are local, in re, here-and-now. The point is clearest if types are taken to be resemblance classes, for a class, in addition to being abstract, is not local in a way a power must be (for a bit more on this, see Heil and Robb 2003, 175-6). Types may be useful in explanation, but this epistemic role should not be confused with the metaphysical role of powers, a role for which types seem to be ill-suited.

O8: Still, why tropes? What's essential to this identity solution is that mental powers are local, in re, and above all, physical. But for all that tells us, mental powers are not tropes but Armstrong-style universals (Armstrong 1989, 1997). Like tropes, Armstrong-style universals are both features and powers. The main difference is that Armstrong-style universals are, well, universals: they are wholly present in multiple instances, and moreover (in some cases) are types. But this difference doesn't seem to make a difference as far as an identity solution to the exclusion problem goes. Tropes or Armstrong-style universals would do just as well (Heil 2008; Maurin 2008; Whittle 2007, 70).

R8: There are indeed multiple and substantive similarities between a trope ontology and Armstrong's ontology of universals. For example, in their respective ontologies, both tropes and Armstrong-style universals are: (1) features; (2) powers[5]; (3) spatiotemporal, at least when we restrict our attention to physical properties; (4) instantiated in re; (5) dependent beings, depending, in particular, on the objects that instantiate them; (6) maximally determinate: neither ontology permits the sort of 'layering' one finds among types; (7) responsible for sameness of type; that is, they are the grounds or truthmakers when two objects are of the same type or kind.

The similarities between the two ontologies are so striking that it's tempting to deny there is a deep difference here. Indeed, at one point Armstrong (1989, 139) wonders, following a suggestion from H. H. Price, whether the difference between these apparent rivals is merely one of alternative languages: what the trope theorist describes as exactly resembling features, Armstrong describes as one and the same. Both describe, in their own ways, the same underlying facts. There are a number of ways this suggestion could play out. Let S1 and S2 be two exactly resembling features. The suggestion could be the relatively innocuous claim that S1 and S2, while strictly distinct, are nevertheless identical in a looser sense. This seems to be the lesson of Williams' (1986) later view on universals, what Campbell (1990) calls 'painless realism'. Alternatively, one might appeal to the doctrine of relative identity (Geach 1980), so that S1 and S2 are distinct tropes, but the same universal. Or the suggestion could be a conventionalism of sorts, so that there's no objective fact of the matter of whether S1 is S2: the apparently distinct ontologies result from the neutral facts plus either of two conceptual or linguistic overlays chosen for, say, pragmatic reasons (Carnap 1956; Sidelle 2002). I can't follow through on these lines here, or maybe anywhere. But supposing it turns out that one of them is correct, what becomes of O8? If the worry is that Armstrong's ontology would do just as well as the trope ontology in response to the exclusion problem, my reply would be that it would do just as well, but only because it's the same ontology, differently described.[6]

This is as far as I want to pursue this option. Suppose there is a genuine ontological difference, as there appears to be, between Armstrong-universalist and trope-theoretic versions of the identity solution. Is there any reason to favour one over the other? I doubt an advantage will be found within the confines of the exclusion problem. The ontologies appear isomorphic in any respect relevant to this problem. If either has advantage, it would have to come from more general considerations.

One minor edge the trope ontology has is that it can give a uniform account of types. As before, call the ground-level 'properties', whether tropes or universals, features. For the universalist, all types are resemblance classes of features except for maximally determinate types. Every maximally determinate type is itself a feature, a single universal. But for the trope theorist, there's no such abrupt change at the most basic level: all types, even the most determinate, are classes of resembling features. What's distinctive of the basic level of types is that the features of those types exactly resemble. But there's no categorial difference at this level for the trope theorist. Here ontological continuity strikes me as a virtue, though this is admittedly a small advantage.[7]

Another potential advantage for the trope ontology may be found in the metaphysics of causality. It's typical for a trope theorist to insist that tropes are better suited than universals--even Armstrong-style universals--to play the role of powers: Campbell (1981, 480), for example, makes the point, as does Honderich (1992, 246-7). Representative is this passage from Campbell (1990, 23):

It is not the stove, the whole stove, that burns you; not even the whole stove here now. For its solidity, iron structure, enamel surface and smoothness have nothing to do with it. It is the temperature that does the damage. Moreover, it is not any temperature, or temperature in general, but this particular case of temperature, among the myriads in the world, and even among the many the stove has during its life. Yesterday's stove temperature is quite innocuous. It is today's that burnt you.

Again, accommodation of the ontology of causes into the trope scheme is so smooth because what is required is an element that combines particularity with a very restricted qualitative character, since causes are always features (almost always a small selection from the host of features present) and every particular cause is a particular feature or constellation of features.

Campbell's point here is complicated by his taking tropes to be the causal relata, while I'm neutral on this matter. But the argument applies equally as well if a causally efficacious property of a cause--a power--must, like causes themselves, combine elements of particularity and qualitativity.[8] Moving beyond Campbell's argument, one might also motivate the trope ontology in the context of a more general metaphysics of causation. Here I have in mind, for example, Ehring's (1997) recent defense of the transference theory, one which makes essential use of tropes. In any case, however one evaluates Ehring's trope-based metaphysics of causation, its relevance to O8 helps to confirm a primary theme of this chapter: the ontology of properties is not optional when confronting the exclusion problem.[9]


[1] I'll assume in what follows that features and powers coincide: the same sort of 'property' (namely, a trope) fills both rolls.

[2] An analogy to Davidsonian events is suggestive, though it's just an analogy: I would not want to say that tropes are events, Davidsonian or otherwise.

[3] Both this and the Williams passage are quoted in Bacon (1995, 17), though Bacon himself insists on exact resemblance, as does Macdonald (1998, 334).

[4] Kim presents this brand of type identity as the best way to save mental causation, but he stops short of endorsing it: see Kim 2005, 161.

[5] There is a difference here: tropes in themselves bestow causal powers, while Armstrong-style universals do so only given the laws of nature. But both are powers in the sense that they are those entities that are causally efficacious, whether or not such efficacy depends on laws of nature.

[6] I take causal efficacy to be, like causation itself, extensional: a power to produce a certain effect is so no matter how described. Matters would be different if, say, explanation rather than causal efficacy were in play. Explanation is intensional, so that even if the two ontologies are equivalent, there may be explanatory reasons to favour the conceptual apparatus of one over that of the other.

[7] The advantage for the trope theorist would be greater if the universalist were forced to say that the higher-level types are not genuine (because not universals). In that case, the trope theorist but not the universalist could admit elation, e.g., as a legitimate type: compare Ehring's (2003, 384) response to O8. But it seems to me that a universalist can be a realist about higher-level types (whether or not Armstrong himself is). They just will not count as genuine universals, but something else, such as the 'second-class properties' mentioned in R6.

[8] Combination, however, is only a metaphor here: for reasons given earlier, a trope is its particularity and qualitativity.

[9] A version of this chapter was read in September 2009 at Davidson College and at The New Ontology of the Mental Causation Debate, a conference held at Durham University. I thank the audiences on those occasions for helpful comments, questions, and objections.


Armstrong, D. M. (1968). A Materialist Theory of the Mind. London: Routledge and Kegan Paul.
---- (1989). Universals: An Opinionated Introduction. Boulder: Westview Press.
---- (1997). A World of States of Affairs. Cambridge: Cambridge University Press.
Bacon, J. (1995). Universals and Property Instances: The Alphabet of Being. Oxford: Blackwell.
Bennett, K. (2008). 'Exclusion Again'. In J. Hohwy & J. Kallestrup (eds), Being Reduced: New Essays on Reduction, Explanation, and Causation. Oxford: Oxford University Press.
Boyd, R. (1980). 'Materialism with Reductionism: What Physicalism Does Not Entail'. In N. Block (ed), Readings in Philosophy of Psychology, Vol. 1. Cambridge, MA: Harvard University Press.
Burge, T. (1993). 'Mind-Body Causation and Explanatory Practice'. In J. Heil and A. Mele (eds), Mental Causation. Oxford: Clarendon Press.
Campbell, K. (1981). 'The Metaphysic of Abstract Particulars'. Midwest Studies in Philosophy, 6: 477-88.
---- (1990). Abstract Particulars. Oxford: Basil-Blackwell.
Carnap. R. (1956). 'Empiricism, Semantics, and Ontology'. In Meaning and Necessity. Second Edition. Chicago: University of Chicago Press.
Chalmers, D. A. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford: Oxford University Press.
Daly, C. (1997). 'Tropes'. In D. H. Mellor and A. Oliver (eds), Properties. Oxford: Oxford University Press.
Davidson, D. (1980). 'Mental Events'. In Essays on Actions and Events. Oxford: Clarendon Press.
---- (1993). 'Thinking Causes'. In J. Heil and A. Mele (eds), Mental Causation. Oxford: Clarendon Press.
Ehring, D. (1996). 'Mental Causation, Determinables and Property Instances'. Nous, 30: 461-80.
---- (1997). Causation and Persistence: A Theory of Causation. Oxford: Oxford University Press.
---- (2003). 'Part-Whole Physicalism and Mental Causation'. Synthese, 136: 359-88.
Fodor, J. A. (1974). 'Special Sciences (Or: the Disunity of Science as a Working Hypothesis)'. Synthese, 28: 97-115.
Geach, P. T. (1980). Reference and Generality, Third Edition. Ithaca: Cornell University Press.
Gibb, S. C. (2004). 'The Problem of Mental Causation and the Nature of Properties'. Australasian Journal of Philosophy, 82: 464-76.
Heil, J. (2003). From an Ontological Point of View. Oxford: Clarendon Press.
---- (2008). 'Modes and Mind'. In S. Gozzano and F. Orilia (eds), Tropes, Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag.
Heil, J. and D. Robb (2003). 'Mental Properties'. American Philosophical Quarterly, 40: 175-96.
Honderich, T. (1992). 'Causation: One Thing Just Happens After Another'. In L. E. Hahn (ed), The Philosophy of A.J. Ayer. La Salle: Open Court.
Jackson, F. (1982). 'Epiphenomenal Qualia'. The Philosophical Quarterly, 32: 127-36.
---- (2006). 'On Ensuring that Physicalism is Not a Dual Attribute Theory in Sheep's Clothing'. Philosophical Studies, 131: 227-49.
Kim, J. (1993) 'Events as Property Exemplifications'. In Supervenience and Mind. Cambridge: Cambridge University Press.
---- (1998). Mind in a Physical World. Cambridge, MA: MIT Press.
---- (2005). Physicalism, or Something Near Enough. Princeton: Princeton University Press.
---- (2007). 'Causation and Mental Causation'. In B. P. McLaughlin and J. Cohen (eds), Contemporary Debates in Philosophy of Mind. Oxford: Blackwell.
Kripke, S. A. (1980). Naming and Necessity. Cambridge, MA: Harvard University Press.
Lewis, D. (1983). 'New Work for a Theory of Universals'. Australasian Journal of Philosophy, 61: 343-77.
---- (1994) 'Lewis, David: Reduction of Mind'. In S. Guttenplan (ed), A Companion to the Philosophy of Mind. Oxford: Blackwell.
Loewer, B. (2007) 'Mental Causation, or Something Near Enough'. In B. P. McLaughlin and J. Cohen (eds), Contemporary Debates in Philosophy of Mind. Oxford: Blackwell.
Lowe, E. J. (1993). 'The Causal Autonomy of the Mental'. Mind, 102: 629-44.
Macdonald, C. (1998). 'Tropes and Other Things'. In S. Laurence and C. Macdonald (eds), Contemporary Readings in the Foundations of Metaphysics. Oxford: Blackwell.
Macdonald, C. and G. Macdonald (2006). 'The Metaphysics of Mental Causation'. The Journal of Philosophy, 103: 539-76.
Malcolm, N. (1970). 'Scientific materialism and the Identity Theory'. In C. V. Borst (ed), The Mind/Brain Identity Theory. London: Macmillan.
Maurin, A. (2008). 'Does Ontology Matter?' In S. Gozzano and F. Orilia (eds), Tropes, Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag.
Molnar, G. (2003). Powers: A Study in Metaphysics. Oxford: Oxford University Press.
Noordhof, P. (1998). 'Do Tropes Resolve the Problem of Mental Causation?' The Philosophical Quarterly, 48: 221-6.
Oliver, A. (1996). 'The Metaphysics of Properties'. Mind, 417: 1-80.
Pereboom, D. (2002). 'Robust Nonreductive Materialism' The Journal of Philosophy, 99: 499-531.
Putnam, H. (1980). 'The Nature of Mental States'. In N. Block (ed), Readings in Philosophy of Psychology, Vol. 1. Cambridge, MA: Harvard University Press.
Robb, D. (1997). 'The Properties of Mental Causation'. The Philosophical Quarterly, 47: 178-94.
---- (2001). 'Reply to Noordhof on Mental Causation'. The Philosophical Quarterly, 51: 90-4.
---- (2008). 'Zombies from Below'. In S. Gozzano and F. Orilia (eds), Tropes, Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag.
Robb, D. and J. Heil (2008). 'Mental Causation'. In E. N. Zalta (ed), The Stanford Encyclopedia of Philosophy [Fall 2008 Edition, available online]:
Shoemaker, S. (2003). 'Realization and Mental Causation'. In Identity, Cause, and Mind, Expanded Edition. Oxford: Clarendon Press.
Sidelle, A. (2002). 'Is There a True Metaphysics of Material Objects?' Philosophical Issues, 12: 118-145.
Whittle, A. (2007). 'The Co-instantiation Thesis'. Australasian Journal of Philosophy, 85: 61-79.
Williams, D. C. (1966). 'The Elements of Being'. In Principles of Empirical Realism. Springfield, IL: Charles Thomas.
---- (1986). 'Universals and Existents'. Australasian Journal of Philosophy, 64: 1-14.
Wilson, J. (1999). 'How Superduper Does a Physicalist Supervenience Need to Be?' The Philosophical Quarterly, 49: 33-52.
---- (2005). 'Supervenience-based Formulations of Physicalism'. Nous, 39: 426-59.
---- (2009). 'Determination, Realization and Mental Causation'. Philosophical Studies, 145: 149-69.
Woodward, J. (2008). 'Mental Causation and Neural Mechanisms'. In J. Hohwy & J. Kallestrup (eds), Being Reduced: New Essays on Reduction, Explanation, and Causation. Oxford: Oxford University Press.
Yablo, S. (1990) 'The Real Distinction Between Mind and Body'. Canadian Journal of Philosophy, Supplement 16: 149-201.
---- (1992). 'Mental Causation'. The Philosophical Review, 101: 245-80.
---- (1997). 'Wide Causation'. Philosophical Perspectives, 11: 251-81.