On non-Nash equilibria

I consider generalisations of the Nash equilibrium concept based on the idea that in equilibrium the players' beliefs should not be contradicted, even if they could possibly be, incorrect. This possibility depends on the information about opponents' behaviour available to the players in equilibrium. Therefore the players' information is crucial for this notion of equilibrium, called Conjectural Equilibrium in general and Rationalizable conjectural. Equilibrium (Rubinstein-Wolinsky 1994) when the game and the players' Bayesian rationality are common knowledge. In this paper I argue for a refinement of Rationalizable Conjectural Equilibrium showing by propositions and by examples how this equilibrium notion works and how the suitable equilibrium concept depends on the players' information. (C) 1999 Academic Press.