The notion that fairness can be determined by using impartial reasoning is examined and an argument is advanced that experimental methods can be used to determine what is fair in certain well-defined contexts. Implications for the prisoner's dilemma are examined. The standard game theoretic prescription of defection in the two-person prisoner's dilemma is contrasted with the prescription generated by rational self-interested players playing the game from an "impartial reasoning" point of view. The incentive structure in such a situation is shown to be similar to that in the familiar "You cut the cake and I choose" rule. The incentive is to be as fair as possible and it reveals the ethical content of rational self-interested behavior. When players choose in ignorance of whether the strategic choice is for themselves or their counterpart their choices are ipso facto fair. The indicated choice of mutual cooperation furnishes a partial explanation for the observed cooperation in experimental contexts. The moral prescriptions that individuals experience are shown to be a function of their risk propensities and to vary as a function of the cardinal properties of games. Implications for behavior in other games are briefly discussed.
1. The problem
Scholars have long argued that the structure of the "prisoner's dilemma ''l reflects a variety of important human interactions (Rapoport, 1968). Some have even argued that "... the logic underlying (the problem of collective action) is the same as that of the prisoner's dilemma" (Hardin, 1971). 2 Yet the analysis of the game has remained problematic.
The reasons for this are twofold. The first is theoretical. Game theory was formulated with the notion that optimal choices in strategic situations could be derived from the assumptions of rationality and self-interest. Yet when the behavioral assumptions of game theory are applied in the prisoner's dilemma, a paradoxical theoretical result emerges. The theoretical prescription is that * The generic arguments presented here have been developed jointly with Joe Oppenheimer as a result of our experimental work on distributive justice. Only the specific application to prisoner's and social dilemmas is the product of the current author although, as usual, discussions with Joe were helpful in clarifying the argument. I would also like to thank the Social Sciences and Humanities Research Council of Canada for funding and Nicholas Tideman for his helpful suggestions.