Adaptive Dynamics in Coordination Games

这项研究旨在更多地了解协调问题是如何解决的。分析的部分动机是希望了解在最近的一些协调实验中观察到的纳什均衡之间引人注目的、持久的歧视模式；这些模式随着参与者的数量和收益函数而系统地变化，其方式不能用传统的均衡改进来解释。这项研究的基本前提是，由于协调博弈中的参与者通常不确定其他参与者将如何对他们的多种均衡做出反应，因此在重复博弈中从经验中学习可能在决定结果方面发挥决定性作用。第一步是开发一个现实但易处理的学习过程模型，显示玩家信念的分散如何与学习相互作用，利用其对极限结果(通常是潜在协调博弈的均衡之一)的概率分布的影响来解释观察到的歧视模式。最终目标是更全面地了解参与者的数量、他们互动的方式、他们的收益是如何确定的，以及他们的沟通机会如何影响协调结果。

This research is directed at learning more about how coordination problems are solved. The analysis is motivated in part by a desire to understand the striking, persistent patterns of discrimination among Nash equilibria observed in some recent coordination experiments; these patterns varied systematically with the number of players and the payoff function, in ways that cannot be explained by traditional equilibrium refinements. The basic premise of the research is that, because players in coordination games are normally uncertain about how other players will respond to their multiplicity of equilibria, learning from experience in repeated play is likely to play a decisive role in determining outcomes. The first step is to develop a realistic but tractable model of the learning process that shows how the dispersion of players' beliefs interacts with learning, using its implications for the probability distribution of the limiting outcome (normally one of the equilibria of the underlying coordination game) to explain the observed patterns of discrimination. The ultimate goal is to understand more generally how the number of players, the ways in which they interact, how their payoffs are determined, and their opportunities to communicate influence coordination outcomes.