Learning and Type Compatibility in Signaling Games
研究信号博弈中哪些均衡会出现，通过构建非均衡学习模型，分析发送者类型差异如何影响实验性信号发送，从而约束接收者的离径信念并实现均衡选择。
Which equilibria will arise in signaling games depends on how the receiver\ninterprets deviations from the path of play. We develop a micro-foundation for\nthese off-path beliefs, and an associated equilibrium refinement, in a model\nwhere equilibrium arises through non-equilibrium learning by populations of\npatient and long-lived senders and receivers. In our model, young senders are\nuncertain about the prevailing distribution of play, so they rationally send\nout-of-equilibrium signals as experiments to learn about the behavior of the\npopulation of receivers. Differences in the payoff functions of the types of\nsenders generate different incentives for these experiments. Using the Gittins\nindex (Gittins, 1979), we characterize which sender types use each signal more\noften, leading to a constraint on the receiver's off-path beliefs based on\n"type compatibility" and hence a learning-based equilibrium selection.\n