Learning in games by random sampling

We study repeated interactions among a fixed set of "low rationality" players who have status quo actions, randomly sample other actions, and change their status clue if the sampled action yields a higher payoff. This behavior generates a random process, the better-reply, dynamics. Long run behavior leads to Nash equilibrium in games with the weak-finite improvement property, including finite, super-modular games and generic, continuous, two-player, quasi-concave games. If players make mistakes and if several players can sample at the same time; the resulting better-reply dynamics with simultaneous sampling converges to the Pareto optimal Nash equilibrium in common interest games. (C) 2001 Academic Press.