A New Approach to the Analysis of Cooperation Under the Shadow of the Future : Theory and Experimental Evidence

The theory of infinitely repeated games lacks predictive power due to equilibrium multiplicity and its insensitivity to, for example, changes in certain parameters, the timing of players' moves or communication possibilities. We propose a new approach, which mitigates the shortcomings of the theory. In particular, we study a standard infinitely repeated prisoner's dilemma game and its variants with (i) heterogeneous tastes for cooperation, and (ii) strategic risk arising from incomplete information about the opponent's taste for cooperation. We theoretically show that the variants we study reduce strategic risk and boost cooperation and coordination on cooperation relative to the standard game, unlike what a theory based on pure material self-interest of players would predict. Our theoretical results are corroborated by the results of our experiments