Learning to Play Nash Equilibrium: An Experimental Research Program

[3111455]考克斯纳什均衡（NE）的概念已经成为博弈论和许多现代经济学的核心。它被用来理解和预测广泛的经济现象。NE实际上是一个相当有说服力的想法：每个人都没有动机采取不同于NE指定的行动，只要他相信其他参与者实际上会选择NE为他们指定的行动。但NE概念的说服力取决于游戏中的每个玩家对其他玩家的行为都有正确的信念。最近有几篇论文开发了一些模型，在这些模型中，玩家通过反复玩游戏来学习玩NE——他们学会了拥有相互正确的信念。这一建议描述了一个实验研究计划，旨在检验和测试一些最近的理论工作。实验研究的一个重点将是确定人们是否通常会学习一些NE而不是其他NE。因此，这项研究将潜在地支持和完善NE的使用。例如，它将为确定一种经济制度（无论是现有制度，还是针对经济问题新设计的解决方案）是否能产生理想的结果提供基础。当一个机构经常出现多个纳什均衡时，其中一些是社会期望的，而另一些则不是，这项研究将有助于确定人们是学会玩理想的结果还是不理想的结果。＊＊＊

9311455 Cox The concept of Nash equilibrium (NE) has become central to game theory and to much of modern economics. It is used to understand and predict a broad range of economic phenomena. NE is in fact a rather persuasive idea: Each individual has no incentive to act differently than the NE specifies he will, so long as he believes that the other participants will actually choose the actions the NE specifies for them. But the cogency of the NE concept depends upon each player in a game having correct beliefs about what the other players will do. Several recent papers have developed models in which players learn to play NE -- they learn to have mutually correct beliefs -- through repeated play of a game. This proposal describes a program of experimental research designed to examine and test some of this recent theoretical work. One focus of the experimental research will be to determine whether people generally learn to play some NE but not others. The research will therefore potentially both support and refine the use of NE. It will provide a foundation, for example, for determining whether an economic institution (either an existing institution, or a newly devised solution to an economic problem) produces desirable outcomes. When as often happens, an institution has multiple Nash equilibria, some of them socially desirable and some not, this research will help to determine whether people learn to play the desirable outcomes or the undesirable ones. ***