A Rationalization of Cooperation in the Iterated Prisoner's Dilemma

This paper consists of two parts. Its first part addresses an obvious and important lacuna in decision theory, our most basic and general theory of practical rationality. For all I can see, this lacuna is only insufficiently addressed in the literature presumably because it is not clear what to do about it and because the attempts to fill it lead onto very shaky grounds. Be this as it may, the first point of the paper will be to show a way how to close the gap. I believe there are a number of fruitful applications of this amendment to decision theory, and they are the source of my confidence in my proposal.1 However, in the second part of my paper I would rather like to consider whether the general ideas of the first part help to throw some new light on Prisoner's Dilemma and on Newcomb's Problem, the millstones around the necks of game and decision theory. I am not at all sure of this further application, but I am excited, I admit, by the outlook that it might work. The ideas of the first part are also formally developed in Spohn (1999, sect. 3). Here, however, I shall try to remain completely informal. This seems feasible for the following reason. Strategic thinking is essentially a recursive matter; one reasons back-