The Atlas of Reality
Page 19
Def D6.3 Sicceity. A property has a sicceity if and only if it is not individuated by its causal powers.
What sort of thing is a sicceity (in this sense)? Consider the following thought-experiment. Suppose that there is a parallel universe exactly like the actual world, but with a different color-sensation, red*, in place of the red sensations of the actual world. That is, imagine a parallel reality in which everything is exactly like the actual world, except that the human sensations that realize the red quale (that are phenomenologically red) in this world are not red at all but instead are red*, a different color-sensation (one that no one ever experiences in the actual world). Let's suppose that in this parallel world, red* sensations have all the active and passive powers that red sensations have in our world. Thus, looking red* and looking red have all of the same environmental, behavioral, intellectual, and emotional connections. The network of powers and of causal laws in the two worlds would be exactly the same, once we have substituted the name ‘red*’ for ‘red’ and vice versa. It seems plausible that the two worlds are really different, since there are different experiences in the two worlds. If so, there must be more to the property of phenomenal redness than its abstract causal profile. There must be a certain thusness to red sensations that is lost when red*-ness is substituted for phenomenal redness.
Here's another, similar example. Imagine a world exactly like ours except that in this alternative reality there are two kinds of electrons, instead of just one: alpha-electrons and beta-electrons. Suppose that the two kinds of electrons have exactly the same powers as electrons have in our world. Both types repel negatively charged things, attract positively charged things, have the same quantity of mass, charge and spin, interact with each other and with other particles in exactly the way electrons do, and so on. If that is a real possibility, then the properties of being an electron and of being an alpha-electron and a beta-electron are not individuated by their causal profiles. Each would have to have some sicceity, over and above its causal powers.
If there are no such sicceities, then it would be impossible for there to be two, indiscernible kinds of electron-like particles. The resulting world would just be the actual world itself, with some electrons arbitrarily labeled as ‘alphas’ and others as ‘betas.’ Without sicceities, each property would be individuated (distinguished from all other properties) only by the relative powers it confers on things. One would be committed to Strong Causal Individuation of Properties:
6.1T.1T Strong Causal Individuation of Properties. All fundamental properties are individuated by the set of powers they confer.
If even one property has a sicceity, one must deny this view.
Why does this issue matter? Who cares if particles are individuated by their powers or not? There is at least one way in which our position on this issue affects another important philosophical question. If properties are individuated by their causal powers, then there are no purely qualitative aspects to things. Every kind of thing would be definable functionally. This would make a certain materialist strategy in the philosophy of mind, namely ‘functionalism,’ much more attractive. Conversely, if properties have qualitative sicceities, this would be true of the secondary qualities (like colors and smells) that we experience consciously. Dualists have argued that these secondary qualities cannot be explained in purely physical terms. If these properties have sicceities, then dualism is more plausible.
The second major issue to consider is whether a property has its powers contingently or as a matter of necessity. We'll assume that if a property has its powers as a matter of necessity, this is not a brute necessity. Instead, we should suppose that the very essence of the property includes the power. Conversely, if a property has a power contingently, this is an “accident” of the property, in the sense of something super-added to its essence.
6.2T Essentiality of Powers (Weak Thesis). Some fundamental property has its causal profile of necessity.
6.2A Strong Accidentality of Powers. No fundamental property has any of its causal profile of necessity: that is, no property confers or fails to confer any power necessarily.
6.2T.1T Strong Essentiality of Powers. All fundamental properties have all of their causal profile of necessity: that is, each power is either necessarily conferred or necessarily not conferred by each property.
Is there a connection between Causal Individuation of Properties and Essentiality of Powers? We think so—if a property is individuated by certain features, it seems reasonable to assume that it has those features essentially. This may not be the case for concrete, material objects. For example, it is not immediately implausible to think that ordinary material objects are individuated by their spatial locations if you think it is impossible for two of them to occupy the same place at the same time. An ordinary material object's spatial location is, then, sufficient to individuate or distinguish it from all other things. But spatial locations aren't essential to those objects, since they can move around. However, it is hard to see how an abstract object like a property could be individuated by anything other than its essential features. The only plausibly fundamental contingent features of properties are what we might call properties of extensions, properties had by properties in virtue of the fact that they are instantiated by a certain group of things. For example, the property of being a dog has the property of being exemplified by exactly the things in the following set: {Elsie, Fido, Crackers, Spike,…}. These sorts of properties are insufficient to individuate properties because two distinct properties might be exemplified by exactly the same things without being identical. For example, the property of being a thing with a heart and the property of being a thing with a kidney have the same properties of extension because they are exemplified by exactly the same things. Since there are no contingent properties that individuate properties, then the following principle seems to be true of properties:
Individuation Entails Essentiality. If a property P is individuated by the features in set A, then P has all of the features in A as a matter of necessity.
The converse implication is not so plausible. One might think that properties confer powers essentially even though they are individuated, not by those powers, but by their sicceities. In fact, we will argue that such a view is a metaphysically attractive position.1
6.2.1 The causal theory of properties: causal structuralism
Sydney Shoemaker is the most prominent proponent of Causal Structuralism or the Causal Theory of Properties (1980, 1998). Causal Structuralism combines Strong Causal Individuation of Properties with Strong Essentiality of Powers. The combination of the two means that all properties are pure powers. There are no sicceities, since it is impossible for two distinct powers to have the same causal profile.
Since all properties are pure powers, this means that all powers are powers to affect other powers. We can never leave the “circle of powers.” In any given world, we can abstract a directed network or graph of powers. How ought we think of this network? If P1 is an active power, for example, then P1 is the power to affect things with some passive power P2 in such a way that those things come to have some new (active, passive or immanent) power P3. From these internal facts about the powers, we could abstract the law of nature L1:
(L1) It is a law of nature that: if a P1-thing comes into proximity to a P2-thing, the latter becomes a P3- thing.
Or, in a more abbreviated form:
(L1) P1(x) & P2(y) & C(x,y) ⇒ P3(y)
Each power, whether active, passive or immanent, would correspond to one or more such law formulas. If we put all the laws together, we constructed the world's “law-book.” From the law-book, we can abstract the causal profile of each power, by simply conjoining the laws into one long statement, and replacing the name of the power with a variable. For example, the following is part of the causal profile of P1, given law L1:
(P1 Profile 1) V1(x) & P2(y) & C(x,y) ⇒ P3(y)
All we have done is replace the name of P1 with a variable, ‘V1’, in law L1. Su
ppose that we simultaneously replace all of the names of properties with variables in the conjoined statement of the law-book. The resulting formula will be the simultaneous causal profile of all of the world's properties. Here is an example of part of such an abstracted law-book:
(L1*) V1(x) & V2(y) & V3(x,y) ⇒ V4(y)
Since a property can, for Shoemaker, be identified with its causal profile, to be property P1 is nothing more than to be something that can play the role of V1 in the fully abstract version of the world's law-book. Ditto for all of the other properties. This type of abstract network of causal relations is all there is to the being of any of the world's properties.
AN ARGUMENT FOR ESSENTIALITY OF POWERS: THE BRANCHING-WORLDS ARGUMENT Sydney Shoemaker has offered an argument for Essentiality of Powers from the nature of possibility. It seems reasonable to suppose that the causal profile of a property is something the property has permanently. It would be very odd if the property of being fire were to lose its power of heating or if the property of being water were to lose its passive power of freezing when cooled to 0 °C. If that is right, then the permanence of power provides us with some reason for thinking that powers are essential to their properties.
Shoemaker attempts to substantiate this argument by assuming that every possible world has ‘branched off’ from the actual world at some time in the past. In other words, every possible world shares an initial segment of its history with the actual world. This is what we will call an ‘Aristotelian’ conception of possibility, an idea that we will explore further in Chapter 15. Let's call Shoemaker's assumption the ‘Branch Principle’.
Branch Principle. For every possible world w, there is a time t such that w and the actual world are exactly alike up until time t.
If we accept the Branch Principle (Branch, for short), and we accept that causal profiles are permanent features of properties, then it immediately follows that every property has its causal profile as a matter of necessity. As we shall see, there is a great deal to be said on behalf of Branch. One obvious drawback is that if determinism were true, then Branch would entail that there is only one possible world. This would make every truth necessarily true. So, the plausibility of Branch requires the world to be necessarily indeterministic.
ARGUMENTS FOR CAUSAL STRUCTURALISM Shoemaker offers two arguments in favor of Causal Structuralism: one appeals to semantics and a second to epistemology. A third argument, based on simplicity, is also relevant.
1. The semantic argument. Shoemaker argues that we could not refer to properties unless they were individuated by their causal powers. Suppose, to the contrary, that there were two distinct properties P1 and P2 that had exactly the same causal profiles. That is, each property causes and is caused by exactly the same properties. In such a case, we could never distinguish one from the other, since we can only distinguish properties by their effects, including their effects on our senses and our minds. If we cannot tell one from the other, how could we refer to one rather than the other?
It isn't clear how this is supposed to establish that it is impossible for two properties to share their causal profiles. It's not even clear why this should give us reason to suppose that in fact there aren't two properties with the same profile. All that it establishes, if it establishes anything at all, is that we could not refer to one rather than to the other in such a case. So what? Why should such a linguistic limitation on our part have any metaphysical implications?
In any case, it seems (as Hawthorne 2001 argues) that we could in fact refer to one rather than the other, simply by encountering one rather than the other on a single occasion and making subsequent reference back to that specific occasion. Suppose that P1 and P2 share the same profile, and that at noon on 10 October, I encounter an instance of P1 but not of P2. I could stipulate that by ‘P1’ I shall mean the property an instance of which I encountered at noon on 10 October. This stipulation will enable me to refer to P1 rather than to P2, even though it is true that I can never be sure, on any future occasion, whether I am then encountering a case of P1 or P2. In effect, Hawthorne was applying to this case Saul Kripke's arguments against the descriptive theory of names in Naming and Necessity (1980).
2. The epistemological argument. Shoemaker's epistemological argument is also directed toward supporting Strong Individuation of Properties by Powers.
Suppose that properties are not individuated by their causal profiles.
Then we would be unable to discriminate one property P1 from its causal “twin” P2.
This entails that we could never tell whether two things have the same property (both P1 or both P2) or two different properties (one P1 and the other P2).
Hence, we could never know that any two things shared the same properties.
But, obviously we can know in many cases that two things have the same property.
Therefore, Strong Causal Individuation of Properties is true.
As Hawthorne (2001) points out, the weakness in this argument occurs at step 4. Suppose, for example, that property P1 occurs with great regularity in our world, while P2 is extremely rare. In that case, we might well know that we are encountering another instance of P1, even though we cannot tell the difference between P1 and P2. The trouble is that Shoemaker's argument depends upon the following principle:
Discrimination Principle. If we cannot discriminate between cases in which A is true and cases in which B is true, then we can never know that A is true.
However, Discrimination leads very quickly to radical skepticism. One cannot tell cases in which one is a brain in a vat from cases in which one is not. Yet it seems reasonable to think that one can know that one is not a brain in a vat, or at least that one can know things that imply that one is not a brain in a vat, such as the proposition that one has arms and legs. It is plausible to think that one can know that p, even though one cannot tell whether it is the case that p or the case that q, so long as the alternative that q is sufficiently weird, remote, outlandish, or improbable. If so, then Shoemaker's argument establishes at best that we rarely encounter properties with identical causal profiles, not that such a situation is metaphysically impossible.2
3. Appeal to simplicity. Finally, the defender of Causal Structuralism can point out that it is an attractively simple theory. It posits only a single kind of property: pure powers. There are no sicceities, and no pure non-powers or “dual-aspect” properties, properties with both a causal profile and a sicceity. If we don't need to posit additional kinds of properties, why should we?
ARGUMENTS AGAINST CAUSAL STRUCTURALISM If there are good arguments against Causal Structuralism, then simplicity alone can't carry the day. We will now consider three such arguments: the problem of circularity or vacuity, the problem of arbitrariness, and the problem of symmetrical systems of causal laws.
1. The problem of circularity or vacuity. Bertrand Russell objected to an early version of Causal Structuralism on the grounds that it was vacuous and viciously circular:
There are many possible ways of turning some things hitherto regarded as “real” into mere laws concerning the other things. Obviously there must be a limit to this process, or else all things in the world will merely be each other's washing. (Russell 1927: 325)
Causal Structuralists imagine a world in which every property is a pure power, a power to bring about other pure powers or to be brought about by them, with the whole network of powers never “bottoming out” in anything other than pure powers. The whole picture seems metaphysically anemic or colorless—a world of furious activity but lacking any bottom line of qualities or forms. In particular, we might object that Causal Structuralists cannot give an adequate account of the qualitative aspects of our conscious experience, which seem to have a thusness or sicceity that consists in more than their mere powers or dispositions to produce further states.
2. The problem of arbitrariness. There is one relation that Causal Structuralists treat as having a sicceity, independent of its place in the cau
sal/nomological network: the causal relation itself. As Hawthorne points out, it would make little if any sense to move all the way to “Hyperstructuralism” in which even causation itself is individuated by its causal profile. What sort of causal profile does causation have? If we replace even the name for the causal relation in the law-book with yet another variable, the law-book would have no content at all. But if the causal relation has a sicceity, why not concede that other properties might have sicceities as well? Causal Structuralism seems inherently unstable or unprincipled.
3. The problem of symmetrical law-books. The final objection Hawthorne (2001) raises concerns the possibility of symmetrical law-books. Suppose that the world's law-book was made up of exactly the following immanent causal laws:
A = {the power, when combined with property B, of producing a change from D to C; thepower, when not combined with B, of producing a change from C to D}