Evolutionary Learning in Signalling Games

We study equilibrium selection by evolutionary learning in monotone signalling games. The learning process is a development of that introduced by Young for static games extended to deal with incomplete information and sequential moves; it thus involves stochastic trembles. For vanishing trembles the process gives rise to strong selection among sequential equilibria. If the game has separating equilibria, then in the long run only play according to a specific separating equilibrium, the so-called Riley equilibrium, will be observed frequently. Also if the game has no separating equilibrium a particular behavior will emerge as the only one observed frequently in the long run. It may or may not correspond to a pooling equilibrium, but if it does, it is to one where both types of sender choose the signal that is best for the ''high'' type when all signals are responded to as if they came from the ''low'' type. This selection is stronger than, and only partly in accordance with, traditional selection based on restrictions on ''out-of-equilibrium'' beliefs.