(p.142) Appendix: The Causal Theory of Properties
(p.142) Appendix: The Causal Theory of Properties
Throughout this work I have avoided tying the claims I have been defending to the causal theory of properties (CTP) I have defended elsewhere (in my 1980 and my 1998)—the view that all properties of concrete things are what in section I of Chapter 4 I called E‐properties, properties whose causal profiles are essential to them. I have relied only on the weaker thesis that each such property is individuated by a causal profile in the sense that it and it alone has that profile in the actual world and worlds nomologically like it. Those who reject CTP—and I take this to include most contemporary philosophers—could consistently accept all of the central claims in this work. Nevertheless, I do accept CTP, and I think that some of what I have said here gives it support. A full‐fledged defense of CTP would require me to address the various objections that have been raised against it, and that is something I will not undertake here. But I want to indicate, briefly, what the case in its favor is and how points made here strengthen it.
Basically, my case for CTP in my earlier work comes down to the claim that there is no plausible truthmaker for the identification of properties in different worlds having different causal profiles. Consider the set of causal profiles of all of the properties instantiable in the actual world. And consider an equally numerous set of causal profiles which belong to the properties instantiable in a world in which the causal laws are different. In both cases the causal profile of a member of the set will be specified in terms of its causal relations to other members of the same set. One can think of each of the sets as being characterizable by a vast Ramsey sentence which implies all of the laws that hold in the world in question. There are innumerable ways in which the members of the one set could be paired with those of the other. Is there anything that could make it true that the properties paired in one of these pairings are identical? Obviously we can't use sameness of causal profile as a basis for identifying properties in the two sets—for it will be true of each of the pairings that if the paired properties are identical then their causal profiles in the different worlds are different. Is there any relation between the two sets of causal profiles that would single out one of the possible pairings as giving us identity? It will no doubt be true that the different pairing will differ in how similar the causal profiles of the same property will be in the different worlds on the assumption that the pairing gives us identity. Perhaps there will be a pairing which if taken as giving us identity gives us a higher degree of similarity between the (p.143) causal profiles of the paired properties than what we get if regard any of the other pairings as giving us identity—although it seems more likely that some pairings will give us more similarity in some respects and others will give us more similarity in other respects. But even if there is a unique pairing that gives us maximum similarity, why should we say in such a case that the paired properties are identical, rather than saying simply that they are properties whose causal profiles are in certain ways similar?
A point I have emphasized in earlier work is that a natural way of thinking about the transworld identification of concrete objects is unavailable to us here. The way concrete objects can vary their properties across worlds mirrors the way they can vary their properties across time. I could have been a professional boxer if there is a possible history starting from some point in my actual career and terminating in my being a professional boxer. Applying this to properties, a property could have had a different causal profile in another world if there is a possible history starting from the property having the causal profile it actually has at some time and ending with its having a different causal profile. But if causal profiles are determined by the laws of nature, and these hold omnitemporally, there are no such possible histories. And even if the laws could change, there would be a question as to how we could be entitled to think that it is one and the same property that is first governed by one set of laws and then by another. Our normal criteria for asserting or denying that one and the same property is instantiated on different occasions assume that sameness of property goes with sameness of causal profile, and we have no criteria that could override these.
It may be objected that it is a mistake to suppose that there must be a constitutive account of what interworld identity of properties consists in—an account of why one of the pairings of actual world properties and properties in a nomologically different world is such that the paired properties are identical, despite the differences in their causal profiles. Identity facts, it may be said, need no truth makers other than themselves. Identity, or a least property identity, is a primitive, unanalysable relation.
But if this is true of inter‐world property identity, it ought to be true of intra‐world property identity. To hold that there is something that is constitutive of intra‐world property identity, but nothing that is constitutive of inter‐world property identity, would undermine the claim that there is a single relation, identity, common to the two cases. But if intra‐world property identity is a primitive, unanalysable relation, it cannot consist in sameness of causal profile. It would seem, then, that it ought to be possible for it to come apart from sameness of causal profile—i.e. for there to be cases in which the same property has different causal profiles at different times (just as, on the contingency view, the same property can have different causal profiles in different worlds), and cases in which different properties have, at different times, the same causal profile. And these are not possibilities we can rule out on empirical grounds. (p.144) All that we have to go on in making judgments of intra‐world property identity is sameness of causal profile. Going on this we would judge that we had the same property even if one property had been replaced by another having the same causal profile, and we would judge that we had different properties even if we had the same property with a different causal profile. The view that property identity is a primitive, unanalyzable relation only contingently related to sameness of causal profile even in the intra‐world case threatens to make property identity unknowable. And if we allow that there is a constitutive relation between sameness of property and sameness of causal profile in the intra‐world case, we should allow that there is one in the interworld case as well.
I argued in Chapter 4, section I that if there are both C‐properties (properties having their causal profiles contingently) and E‐properties, we can have no way of knowing of any property of which sort it is. There are two ways of avoiding the unknowability consequence. One is to deny that there are C‐properties—in other words, to embrace CTP. The other is to deny that there are E‐properties. But to do the latter requires holding either that there is no such property as being a braking system, or being an adding machine, or that such properties are C‐properties and have different properties in other possible worlds—i.e. that there are possible worlds in which the very property that in this world is the property of being a braking system has a causal profile that is irrelevant to the function of enabling braking. This seems counterintuitive, to say the least. So should we grit our teeth and accept the unknowability claim? But if the conclusion that properties like being a braking system are C‐properties rather than E‐properties is objectionable, surely the conclusion that for all we know such properties are C‐properties rather than E‐properties is equally objectionable. Such objectionable conclusions are avoided if we accept CTP.