Thursday 18 February 2010

Rational Transitions and Hurley’s Monkey

I believe that an agent may instantiate a rational transition even though that agent lacks the concept of a rational transition. In support of this claim, I will be exploiting the example due to Susan Hurley (see below).

In general, the distinction between possessing the concept of a certain state of affairs and instantiating that state of affairs is relatively straightforward. Consider, for example, the concept of a mammal. One could not plausibly be in possession of the concept without knowing that mammals are furry, warm-blooded animals that nurse their young. However, one may instantiate mammal-hood—i.e., by being a furry, warm-blooded animal that nurses its young—even if one does not know that mammals are furry, warm-blooded animals that nurse their young. Analogously, I maintain that it is possible to instantiate a rational transition even if one does not possess the concept of a rational transition. In the account that follows, I will attempt to sketch and motivate a conception of rational transitions that can accommodate the above intuitions.

To say that a psychological transition is rational is, by my lights, to say that it is reason-conferring; it is to say that the agent has a reason for beliefs based on that transition. I hold that a plausible account of a rational transition must be able to accommodate the distinction between possessing the concept of a rational transition, on the one hand, and instantiating a rational transition, on the other. We may motivate this claim by considering what I take to be an example of a rational transition—namely, a transitive inference. In order to have the concept of a transitive inference one must believe that such inferences may be employed in different, but logically similar, contexts. In other words, one must believe that it is generalisable. However, an agent may instantiate a transitive inference even though that agent lacks such a belief. This is illustrated by an example due to Susan Hurley, who describes the following empirical possibility:
An intentional agent’s reasons for action can be bound to specific contexts and not generalise; there can be islands of practical rationality. For example, a primate could have reasons in social contexts that she cannot generalise to nonsocial but logically similar contexts. Suppose a monkey observes that conspecifics A is dominant over B and that B is dominant over C and, never having observed A and C together, registers that A is dominant over C, and is able to use this information in instrumentally appropriate ways in relation to various goals. Nevertheless, she might be unable to generalise the ability to make transitive inferences to foraging contexts, such as: tree A has more fruit that tree B, which has more than tree C, so tree A has more fruit that tree C. Evolution might have conferred the ability to make transitive inferences in social contexts, if it was most valuable there, without conferring the conceptual abilities needed to transfer it readily to other contexts (Hurley [2003], "Animal action in the Space of Reasons", pp. 238-239).
The lesson of Hurley’s monkey is that an agent’s belief that a particular inferential pattern is generalisable does not necessarily account for that agent’s reliance on that inferential pattern. There are a number of ways that this claim may be unpacked. Since I am sympathetic to the dispositional account of belief defended by Ruth Barcan Marcus (See her paper: "Some Revisionary Proposals about Belief"), I will rely on an argument that exploits that conception. Let us say that an agent believes that a particular inferential pattern is generalisable only if that agent is disposed to employ that inferential pattern in a number of different contexts. Ex hypothesi, Hurley’s monkey is not disposed to employ transitive inferences in a number of different contexts. It follows that Hurley’s monkey lacks the belief that transitive inferences are generalisable. Moreover, if we assume that an agent possesses the concept of a transitive inference only if that agent believes that transitive inferences are generalisable, then Hurley’s monkey fails to fulfil a necessary condition for possessing the concept of a transitive inference. Consequently, insofar as Hurley’s monkey represents an empirical possibility, it follows that an agent may instantiate a transitive inference even though that agent lacks the concept of a transitive inference.