A stochastic process that describes a payoff-based learning procedure and the associated adaptive behavior of players in a repeated game is considered. The process is shown to converge almost surely towards a stationary state which is characterized as an equilibrium for a related game. The...