This paper develops a dynamic evolutionary model in which agents make choices on the basis of relative performance criteria. We distinguish two classes of learned behavior: imitative dynamics and a new class of dynamics, "introspective dynamics." Under imitative dynamics, agents compare payoffs of different agents in the population and tend to imitate more successful behavior in the population. Under introspective dynamics, agents compare their own current with past payoffs and tend to select actions that provide higher payoffs. With introspective dynamics, under weak regularity conditions, the stochastically stable set of states is contained in the set of Nash equilibria, providing a novel rationale for Nash equilibrium behavior. With imitative dynamics, under mild regularity conditions there is a unique stochastically stable state, bit it is not a Nash equilibrium. We consider both forms of dynamics in the contexts of games satisfying strategic substitutes and strategic complements.