作者: Colin F. Camerer , Juin Kuan Chong , Teck-Hua Ho
DOI:
关键词:
摘要: Strategic thinking, best-response, and mutual consistency (equilibrium) are three key modelling principles in noncooperative game theory. This paper relaxes mutual consistency to predict how players likely behave one-shot games before they can learn equilibrate. We introduce a one-parameter cognitive hierarchy (CH) model to behavior games, initial conditions repeated games. The CH approach assumes that use k steps of reasoning with frequency f (k). Zero-step players randomize. Players using (≥ 1) best respond given partially rational expectations about what doing 0 through - 1 actually choose. A simple axiom which expresses the intuition thinking increasingly constrained by working memory, implies (k) has Poisson distribution (characterized by mean number τ ). CH model converges dominance-solvable equilibria when is large, predicts monotonic entry binary for < 1:25, predicts effects group size not predicted Nash equilibrium. Best-fitting values of τ have an interquartile range (.98,2.40) median 1.65 across 80 experimental samples matrix mixed-equilibrium dominance-solvable p-beauty contests. also economic value because subjects would have raised their earnings substantially if they had best-responded forecasts instead of making choices did.