Incentive constraints in games with bounded memory

Abstract

In this paper I analyze a repeated coordination game between a bounded memory player and a fully rational player. A bounded memory player is a player who is subject to imperfect recall, but, conditional on this constraint, acts optimally every point in time given her beliefs over the actual history of the game. I show that the results contrast with the ones in the finite automata literature. In particular, a result denoted the 'tyranny of the weak' in Gilboa and Samet (Games Econ Behav 1: 213-221, 1989) is no longer present here.