Cooperation and bounded recall (Q804483)

Let G be a two-player game with a unique Pareto-optimal outcome. Consider the supergame which infinitely repeats G, and perturbations of this supergame, in which a player plays with small probability a fixed strategy with bounded recall. Then the perturbed supergame has a pure strategy Nash equilibrium, and payoffs at this equilibrium are close to optimal. This remarkable result shows how optimality is a rational outcome in a game with boundedly rational players.