While the conceptual reduction of the social sciences to psychology is unlikely, there is a weaker reading of Mill’s reductionist claims on which they might succeed. On almost anyone’s account of collective behavior, unless there are changes in individual psychological states or environmental conditions, there will be no change in the states of the collectivity that these individuals compose. However, if Mill is right to claim that the social sciences must be reduced to psychology, then every social fact must admit of some sort of reductive explanation; a more promising understanding of such reductions is to see them as requiring nothing more than each of social fact being explained entirely in terms of simpler entities. Perhaps the most promising way of spelling out such reductive explanations in the case of mental entities is suggested by David Chalmers (1996; Jackson 1998; Jackson and Chalmers 2001).57
Chalmers argues that if you want to reduce B-properties of one type to A-properties of another, then it will be a minimal condition on such reductions that B-properties supervene
57 Chalmers argues that if materialism is true every fact will admit of a reductive explanation in physical terms,
and although he’s not immediately concerned with the reduction of the social to the psychological, his story is helpful in spelling out an alternative account of reduction by way of supervenience.
on A-properties. Spelling out this notion of supervenience, Chalmers notes that B-properties supervene on A-properties just in case no two possible situations are identical with respect to their A-properties while differing in their B-properties. There are, however, a number of ways in which the relevant notion of possibility can be understood. Suppose someone wanted to explain why all worlds that are identical with regard to their physical properties would also be identical with regard to their biological properties. In offering an explanation of such similarities across worlds, one might begin with a notion of nomological or natural possibility that could constrain reductive explanations to all and only worlds that are identical to ours as regards the natural laws. Spelling out supervenience according to this notion of possibility, entails that any naturally possible situations with the same A-properties will have the same B-properties (Chalmers 1996). However, as Chalmers notes, this is a weak notion of possibility so it will not entail the sort of reductions that the Mill hoped for. A nomological relation between the social facts and the psychological facts cannot assure the deductive entailments that Mill requires between collective psychological explanations and individual psychological explanations.
The problem is this. If B-properties (e.g., the facts of the collective psychology) are only nomologically supervenient on A-properties (e.g., the facts of individual psychology), then it is possible to conceive of a world in which the A-facts hold but the B-facts don’t; and, provided conceivability is a good guide to possibility, this suggests that a world in which facts about individual psychology are the same and facts about collective psychology are different is metaphysically possible. The proponent of Millian deductions must recognize that such explanations require a stronger sort of necessity.58 Nomological supervenience is not
58 I am inclined to think that these sorts of deductions are unlikely to be forthcoming in any form. In fact, very
strong enough to guarantee that the introduction of B-facts into a world doesn’t offer something new that requires their own explanation in B-terms.
What Mill needs is a notion of supervenience that guarantees that if the collective psychological facts supervene on the individual psychological facts, any two situations that are identical in individual psychological facts will necessarily be identical in their collective psychological facts. To put this point another way, if Mill’s argument is to be successful, the supervenience relationship between collective psychological facts and individual psychological facts must be sufficient to guarantee that the individual psychological facts deductively entail the collective psychological facts. But, in order to guarantee this, we have to opt for a much stronger interpretation of the supervenience relationship, what Chalmers (1996) refers to as logical supervenience. Logical supervenience is the claim that B- properties supervene on A-properties just in case no two logically possible situations are identical with respect to their A-properties but different with respect to their B-properties. If this sort of relation obtains, then once God creates a world in which the A-properties are fixed, she doesn’t have any more work to do in fixing the B-properties. The B-properties are fixed as a matter of logical necessity once the A-properties are fixed and there is no conceivable world in which the B-facts differ while the A-facts remain the same. Suppose, for a moment, that this sort of relation obtains between the facts of collective psychology and the facts of individual psychology. How, then, would explanatory reductions work within this framework.
Jackson and Chalmers argue that there are a posteriori identities that obtain between the facts that articulated in macro-scientific explanations and their subvenience bases. This
required for explanation. However, this research project does seem to have some affinities with Mill’s reductive project, so I’ll entertain these possibilities.
thought seems, at least initially, quite plausible: the explanation of macroscopic phenomena takes the form of an analysis of the mechanisms that give rise to a particular sort of macroscopic phenomena. Thus, for example, if we want to explain why a car is accelerating slower than it typically does, the explanation should not be given at the level of the whole car; instead, the phenomena should be explained at a lower explanatory level by appeal to facts about clogged fuel injectors, bad spark plugs, or even old spark plug wires.
Similarly, as Dennett is fond of pointing out, when a person fails to behave rationally, for example when a person with Anton’s syndrome is obviously blind but denies being so, we explain her behavior not at the level of psychological phenomena but in terms of facts about her neurology; this, for example, might occur by way of an appeal to the sort of damage that has occurred to her occipital. Finally, if we want to explain why NC State fails to defeat UNC at basketball even though they’re playing to sort of offense that should cause problems for UNC, we’ll appeal to the athletic abilities of UNC’s players and to the breakdown of the Princeton Offense because of the particular mistakes made by the members of the NC State team.
Many explanations clearly do take the form of explanation in terms of simpler entities. In fact, this is precisely the sort of explanatory model on which the homuncular functionalism that I advanced in Chapter 1 is grounded. However, the truly astonishing claim advanced by Jackson and Chalmers, and the claim that is required for Mill’s reductive project, is that these sorts of explanations rest on conceptual truths that allow for an upward derivation of the macroscopic facts from their microscopic realizers. The reason for such a claim is that unless there is some good reason for thinking that the macroscopic facts
logically supervene on the physical facts, we will not have fully explained the macroscopic facts. Here, in brief, is the sort of argument Jackson and Chalmers typically offer:
1) If materialism is true, every fact will (eventually) admit of reductive explanation in physical terms;
2) Reductive explanation of B-facts in terms of A-facts requires logical supervenience of the B-facts on the A-facts;
3) Phenomenal facts don’t logically supervene on the physical facts;
4) So, phenomenal facts don’t admit of reductive explanation in physical terms; 5) So, materialism is false.59
There are, of course, a number of contentious premises in this argument—and I’m not going to argue against this position (Bill Lycan (2003) has done a nice job of pointing to a number of problems with this argument—so I’ll leave that to him). However, there is a version of this argument that captures Mill’s reductive intuition about collective psychology. Although Mill would not follow Jackson and Chalmers in their claim that physicalism is false, he would be concerned to avoid a similar untoward conclusion about empiricism. Mill’s empiricist version of this argument takes something of the following form:
1) If empiricism is true, then every psychological fact will (eventually) admit of reductive explanation in terms of individual psychologies;
2) Reductive explanation of B-facts in terms of A-facts requires logical supervenience of the B-facts on the A-facts;
3) But, empiricism is true;
4) So, collective psychological facts admit of reductive explanation in terms of individual psychological facts;60
59 A version of this reconstruction of this argument occurs in Lycan (2003)
60 This relies on a special case of (2): unless collective psychological facts logically supervene on individual
5) So, collective psychological facts logically supervene on individual psychological facts.
As with the Jackson and Chalmers argument, there are a variety of places at which to resist this argument. However, before turning to the ways in which one might resist this argument, let me turn briefly to the sorts of arguments that might be marshaled in favor of this reductive picture of explanation.