Collaborative Research: Social Signals and Reputation in Games with Facial Schematics

当代社会科学的成功之一是博弈论的发展和扩散。对于广泛的现象，博弈论对个体之间的战略互动产生了巨大的洞察力。它在预测大群体的行为方面显示出最大的力量--无论是在市场、政治选举、信息聚合的背景下，还是在面临大规模社会困境时。然而，博弈论也产生了对小群体行为的预测，这些预测没有在仔细控制(和重复)的实验室实验中得到证实。这项研究调查了社交信号如何有助于解释小群体行为。社交信号涉及代理人寻找关于他人意图的线索，并发送关于自己意图的线索；这些线索反过来又建立了对他人行为的预期。这项研究关注的是一组特定的线索：非语言面部表情。许多博弈论都假设参与者是匿名的，对他们伴侣的身份没有任何线索。另一方面，许多社交交流涉及面对面的互动，实验室实验表明，这种设置与匿名互动有很大不同。调查人员消除了语言交流，专注于一种相对简单的工具--风格化的面部表情--用它来解释一个不熟悉的交换伙伴最初的前科是如何形成的。这项研究使用标准博弈论模型，并借用进化心理学的新概念来解释导致一致和可预测结果的行为模式，这些结果与理论预测的结果背道而驰。实验室实验被用来检验关于社会信号、声誉前科的重要性和声誉发展的一些猜想。这是与弗吉尼亚理工学院和州立大学的凯瑟琳·C·埃克尔的合作研究。

One of the successes in contemporary social science has been the development and proliferation of game theory. For a wide range of phenomena, game theory produces enormous insight into the strategic interaction of individuals. It shows its greatest power in predicting the behavior of large groups of people -- whether this is in the context of markets, political elections, information aggregation or when confronting large-scale social dilemmas. However, game theory also generates predictions for small group behavior that are not corroborated in carefully controlled (and replicated) laboratory experiments. This study investigates how social signaling may help to explain small group behavior. Social signaling involves agents looking for cues about others' intentions and sending cues about their own intentions; these cues in turn build expectations for the behavior of others. This research focuses on a specific set of cues: nonverbal facial expressions. Much of game theory assumes actors are anonymous with no clues about the identity of their partner. On the other hand, much of social exchange involves face-to-face interaction, and laboratory experiments demonstrate that such settings are very different from anonymous interaction. The investigators eliminate verbal communication and focus on a relatively simple device - stylized facial expressions - using it to explain how initial priors are formed about an unfamiliar exchange partner. The research uses standard game theoretic models and borrows from new concepts in evolutionary psychology to explain behavioral patterns that lead to consistent and predictable outcomes that deviate from those predicted by theory. Laboratory experiments are used to test a number of conjectures about social signaling, the importance of reputational priors and the development of reputations. This is collaborative research with Catherine C. Eckel of Virginia Polytechnic Institute and State University.