100 J.J. Haldane
Now, however, I want to go further and argue that there are grounds
for not regarding action and psychological explanation more generally as a
species of causal explanation in the sense required by the physicalist argu-
ment. Assuming the law-like nature of efficient causation, and the claim that
giving reasons is giving efficient causes, it ought to be the case that there are
psychological laws connecting psychological states to one another and (as
reasons) to actions. As Paul Churchland has been concerned to emphasize,
there are indeed well established psychological generalizations of an appar-
ently law-like form – what he calls the ‘explanatory laws of folk psychology’.
Consider the following examples:
(1) For any subject x and any propositional content p: if x fears that p then
x desires that it not be the case that p,and
(2) For any subject x and any propositional contents p and q: if x believes
that p, and believes that if p then q, then barring confusion, distrac-
tion, etc., x believes that q.Notice first that while (1) is an unrestricted generalization it is patently
false, and when one tries to accommodate counter-examples, cases where
someone fears that p but does not desire that not p, by introducing a ceteris
paribusclause, or, as in (2) by various exclusions, it quickly becomes apparent
that the character of other things being equal and that of relevant exclusion
conditionscannotbe fully specified. No genuine, universal psychological
generalizations – that is to say ‘laws’ – can be specified. Furthermore, such
reason /action generalizations as seem to approximate to law-like status are,
if true, a priori. Consider:
(3) For any subject x and any action type A: if x believes that A is logically
impossible then x cannot sincerely try to A.Unlike an empirical causal law, hypothesized on the basis of observed
sequences, this principle identifies a relation between elements in a rational
order – ‘the sphere of reasons’. This comes out in the fact that such principles
constrain the application of psychological concepts. If we had good reason to
maintain that someone believed that a course of action was logically impos-
sible, then we rationally could not describe him or her as sincerely trying to
effect it. Anything that supported attributing the belief would ipso facto be
reason for not attributing the attempt, and vice versa.
How then do action explanations work, if not by citing antecedent (efficient)
causal factors? Part of my general approach has been to resist reductions,