Behavioral Game Theory Experiments and Modeling

This chapter reviews recent experimental data testing game theory and behavioral models that have been inspired to explain those data. The models fall into four groups: in cognitive hierarchy or level-k models, the assumption of equilibrium is relaxed by assuming agents have beliefs about other agents who do less reasoning (i.e., some are non-strategic, and others are more strategic and understand they are playing non-strategic players). A different approach, quantal response equilibrium, retains equilibrium expectations but adds stochastic response (of which players are aware). Learning theories explain choices in games played repeatedly as a result of past actions and payoffs that were received (as in classical reinforcement) or foregone payoffs (model-directed learning). Finally, many studies reject the joint hypothesis of equilibrium expectations and optimization, along with self-interest in valuing outcomes. Social preference models have emerged to explain these data, capturing concepts like inequity-aversion, reciprocity, and social image.