Collaborative Research: Sophisticated Learning and Strategic Teaching in Repeated Games

协作研究：重复游戏中的复杂学习和策略教学

• 批准号: 0078911
• 负责人: Colin Camerer
• 金额: $ 24.46万
• 依托单位: California Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2000
• 资助国家: 美国
• 起止时间: 2000-08-01 至 2003-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0078911&HistoricalAwards=false
• 关键词: Collaborative; Research; Sophisticated; Learning; Strategic

Camerer #0078911Our research addresses how people and organizations learn from experiencein strategic situations like bargaining, coordinating joint actions(teamwork), choosing prices and features for new products, bidding inauctions, etc. In previous research we discovered a mathematical formulawhich explains how people appear to learn from experience, but thenumerical details of the formula (its "parameters) seem to vary fromsituation to situation, as if people are learning in different ways. Wetherefore propose to explore why these parameters seem to vary. Inaddition, most mathematical theories of strategic learning assume thatpeople only look back at past experiences. We also propose to extend thesetheories to allow for people who realize that other people are learningfrom experience, and are able to therefore outguess what others will dobased on what happened in the past. If players are "sophisticated", inthis sense, it pays for them to take actions that are not perfect in thenear-term, to "teach" other players who are learning to take actions whichwill benefit the "teachers" in the long-term. This teaching can bebeneficial for the teacher but bad for society (e.g., when firms scareaway innovative competitors by threatening illegal retaliation), orbeneficial for everyone (e.g., when firms reassure others that they can betrusted). Our research develops a precise mathematical theory of howsophisticated players behave and when it pays for them to teach. We usethe theory to explain observations from experiments and predicts whetherteaching will occur in new situations.

Camerer #0078911我们的研究探讨了人们和组织如何在谈判、协调联合行动等战略情境中从经验中学习（团队合作），为新产品选择价格和功能，在拍卖中出价等。在以前的研究中，我们发现了一个数学公式，它解释了人们如何从经验中学习，但公式的文字细节（它的“参数”）似乎因情况而异，好像人们以不同的方式学习。因此，我们建议探讨为什么这些参数似乎有所不同。此外，大多数关于策略学习的数学理论都假设人们只会回顾过去的经验。我们还建议扩展这些理论，使人们能够意识到其他人正在从经验中学习，因此能够根据过去发生的事情猜测其他人会做什么。从这个意义上说，如果玩家是“老练的”，那么他们采取短期内不完美的行动，“教导”其他正在学习采取行动的玩家，从长远来看，这些行动将使“老师”受益，这是值得的。这种教学对教师有益，但对社会有害（例如，当公司通过威胁非法报复来吓跑创新竞争对手时），或者对每个人都有利（例如，当公司向别人保证他们可以被信任时）。我们的研究开发了一个精确的数学理论，来解释老练的球员的行为，以及他们什么时候可以支付教学费用。 我们用这一理论来解释实验观察到的现象，并预测在新的情况下教学是否会发生。