Adaptive Learning in Imperfect Monitoring Games

This paper deals with the problem of specifying a general learning model, the rationality of which is not situation dependent. I propose a very general model of adaptive learning suitable to study learning problems in games with imperfect monitoring, such as extensive form games. In this context I relate adaptive learning with a general notion of equilibrium. In particular I provide a dynamic characterisation of conjectural equilibria: a "stable" strategy profile is consistent with adaptive learning if and only if it is a conjectural equilibrium. Journal of Economic Literature Classification Numbers: C72, D83, D82