Doctoral Dissertation Research: Learning in Three by Three Bimatrix Games; An Investigation With Experimental Data

项目摘要：本博士论文研究是由Nicole Bouchez在Daniel Friedman的指导下进行的。经济中充斥着买卖商品和服务的企业，但他们怎么知道竞争对手的激励措施呢？对于人们或企业群体如何在无法直接观察到的情况下了解彼此的激励措施，我们知之甚少。已经提出了几个相互竞争的理论模型，表明通过经济互动进行学习，但很少对这些模型进行实证比较。本项目进行了一系列计算机化的实验室实验，以考察三选双母线游戏中的小组学习。Bimatrix游戏允许玩家看到自己的选项和或有收益，但不允许他们看到对手的收益。这一结构具有广泛的适用性，是迈向一般性、多项选择、战略性互动的关键一步。收集的数据被用来比较不同的学习模式对个人行为的预测能力。

Project Abstract Project Abstract: This doctoral dissertation research is conducted by Nicole Bouchez under the supervision of Daniel Friedman. The economy is full of businesses buying and selling goods and services, but how do they know about their rivals' incentives? Little is known about how groups of people or businesses learn about each other's incentives when they are not directly observable. Several competing theoretical models have been suggested that show learning through economic interactions, however few empirical comparisons of these models have been made. This project conducts a series of computerized laboratory experiments to investigate group learning in three-choice bimatrix games. The bimatrix games allow players to see their own options and contingent payoffs, but do not allow them to see their opponents' payoffs. This structure is broadly applicable and is a key step towards general, multiple choice, strategic interactions. The data collected are used to compare how well the different models of learning predict individual actions.