Introspection, Learning, and Equilibrium in Games: Theory and Experiment

博弈论越来越多地被用于社会科学，但对完美决策和完美预见的极端依赖引起了对其经验相关性的怀疑。实验室证据显示，行为模式系统地偏离了博弈论的预测，这进一步加强了这种怀疑。该项目涉及开发和测试更能描述实际人类行为的模型。一个目标是提供混合模型，能够再现有趣的模式的第一阶段发挥（内省），时间序列数据在重复的游戏（学习），和系统偏离静态平衡。另一个目标是应用成功的混合体来改进经济和社会机构的设计。

Game theory is increasingly being used in the social sciences, but the extreme reliance on perfect decision making and perfect foresight has raised doubts about its empirical relevance. This scepticism is reinforced by laboratory evidence showing behaviour patterns that are systematically biased away from game-theoretic predictions. This project concerns the development and testing of models more descriptive of actual human behaviour. One aim is to deliver hybrid models able to reproduce interesting patterns of first-period play (introspection), time-series data in repeated games (learning), and systematic departures from static equilibrium. Another aim is to apply a successful hybrid to improve the design of economic and social institutions.