A remark on the Harsanyi-Selten theory of equilibrium selection (Q800842)

We consider a class of 3-person games in normal form with two pure strategies for each player and two strict equilibrium points. To select one of these two strict equilibrium points as the solution, the equilibrium selection theory of Harsanyi and Selten is applied. The games are constructed in such a way that the a priori probabilities reflect somewhat poorly the risk situation of the players. It is argued and illustrated by examples that this might yield unreasonable results. The a priori probabilities would describe the risk situation of the players more completely if their definition were not based on the expectation of correlated decision behavior.