A satisfactory theory of strategic behavior in games not only must identify the outcomes that are equilibrium points, but also must explain the origin of mutually consistent strategic behavior. Recently, there has been a renewed interest in the dynamic systems approach to the origin of mutually consistent strategic behavior. This literature can be divided into four broad categories: learning by trial-and-error, best responding to naive beliefs, best responding to sophisticated beliefs, and evolutionary processes. Unlike theories of strategic rationality, it seems very difficult to select amongst the competing theories using introspection. The research will use the scientific method to discriminate between those abstraction assumptions underlying strategic rationality that lead to accurate theories of strategic behavior in observable games and those that do not. It will focus on the abstraction assumptions needed to map an observable game into an abstract game and, hence, needed for the extant theory of abstract games to be relevant for an analysis of strategic behavior in observable games. The ultimate goal is to develop an empirical foundation for a convincing and scientific explanation of the origin of mutually consistent behavior in observable games in which the scientist can observe the rules of the game, the actors selecting actions, the outcome of those actions, and the resulting scores.
