Reasonable utility functions and playing the cooperative way

In this essay I dispute the widely held view that utility theory and decision theory are formalizations of instrumental rationality. I show that the decision theoretic framework has no deep problems accommodating the reasonable qua a preference to engage in fair cooperation as such. All evaluative criteria relevant to choice can be built into a von NeumannMorgenstern utility function. I focus on the claim that, while rational choicedriven agents are caught in the Paretoinferior outcome, reasonable agents could solve the PD and cooperate. Not so, I argue. If reasonable people find themselves in PD situations they too would follow the dominant defect strategy. The difference between instrumentally rational agents and those who are also reasonable is not that they would behave differently in Prisoners Dilemmas, but that reasonable people are more successful at avoiding the Prisoners Dilemma and tend to play more cooperative games.