Deterrence, Maximization, and Rationality

Is deterrence a fully rational policy? In our world deterrence works sometimes. But in a more perfect world, in which actors rationally related their choices to their beliefs and preferences, and in which those beliefs and preferences were matters of common knowledge, could deterrence work? Some say no.' Others hold a conception of rationality that would commit them to saying no, were they to consider the issue.' I say yes. Deterrence can be part of a fully rational policy. I propose to demonstrate this. At the heart of a deterrent policy is the expression of a conditional intention. An actor A expresses the intention to perform an action x should another actor B perform an action y. If B would do y did A not express her intention, then we may say that A's expression of intention deters B from doingy. In expressing her intention as part of a deterrent policy, A seeks to decrease the probability of B's doing y by increasing his estimate of her conditional probability of doing x should he do y. We need better labels than x and y if our talk about deterrence is to be perspicuous. In at least some situations, A's deterrent intention is retaliatory; A expresses the intention to retaliate should B do y. So let us call x retal. And what A seeks to deter is an action that would advantage B in relation to A; let us then call y advant. We shall then say that an actor A expresses the intention to retal should another actor B advant. A seeks to affect B's estimate of her conditional probability of retal should he advant. Why does she expect her expression of conditional intention to have this effect? Let us suppose that A and B are rational; on the received view of rationality, an actor seeks to maximize expected