The Two-Person Harvard Game: An Experimental Analysis

Two subjects have to repeatedly choose between two alternatives, A and B, where payoffs of an A or B-choice depend on the choices made by both players in a number of previous choices. Locally, alternative A gives always more payoff than alternative B. However, in terms of overall payoffs exclusive choice of B is a better strategy. The equilibrium predicted by the theory of melioration is to exclusively play A, while the Nash equilibrium is to almost exclusively play B. The predictive values of such equilibria are analyzed under three different informational conditions. Special attention is paid to the learning processes exhibited by players.