Do the requirements of morality and those of rational self-interest dictate performance of the same acts in every particular situation? In this thesis I examine and evaluate various proposed answers to this age-old philosophical question. I focus on a particular kind of situation in which the two sorts of requirement seem to be at odds with one another. These are situations of contract-keeping that are prisoner's dilemma-like. In such situations, if you are moral, then it appears that you should comply (...) with an agreement to do the "cooperative thing." If you are rational, then it seems that you will do best for yourself if you refrain from cooperating. Reformulating the central question of this essay, is it rational in some sense of 'rational' to do the cooperative thing in, and so to "escape" a prisoner's dilemma? ;I begin with Hobbes and submit he would answer in the negative. In analysing Hobbes' position, I critically discuss Jean Hampton's and Gregory Kavka's views on Hobbes on state-of-nature cooperation. I then consider more recent replies paying particular attention to David Gauthier's. I argue that his defense of an affirmative reply--the desirable reply--is flawed. I arrive at a similar verdict about Edward McClennen's opinion. Finally, I advance my own conclusion--there may be situations in which people must act in a way that is either immoral or irrational. (shrink)

InMorals by Agreement, David Gauthier (1986) argues that it is rational to intend to cooperate, even in single-play Prisoner's Dilemma games, provided (1) your co-player has a similar intention; (2) both intentions can be revealed to the other player. To this thesis four objections are made. (a) In a strategic decision the parameters on which the argument relies cannot be supposed to be given. (b) Of each pair ofa-symmetric intentions at least one is not rational. But it is impossible to (...) form symmetric intentions to cooperate conditionally. For the condition on which the decision depends cannot be fulfilled without deciding. (c) If one's intention has to be ascertained on the basis of information about one's past performance, it is straightforwardly rational to intend to cooperate, but there is no reason to do so in a single-play PD. (d) The argument cannot be extended ton-person games which are Gauthier's principal concern. (shrink)