This research will extend dynamic theories of games in several directions. Part 1 contributes to the growing literature on evolution and learning. It will examine how a stable pattern of behavior might emerge when players adjust their behavior by trial and error. In particular, the study will focus on the long run implications of experimentation or mutations in such a process. This approach has the potential to make a unique prediction, even when the underlying game possesses multiple locally stable equilibria. Part 2 examines the nature of implicit collusion in a realistic situation, where information about the past history is not common knowledge. The investigator will formulate a model of a discounted repeated game with privately observable outcomes, and will explore the role of communication. This study is important because it will improve our understanding in situations, prevalent in industrial organization literature and organizational settings, in which players receive private signals about past play and implicit agreements are developed.
