博弈论中重复可允许算法的认知分析  

An Epistemic Analysis of Iterated Admissibility Algorithm in Game Theory

作　　者：崔建英[1,2] 

机构地区：[1]西南大学逻辑与智能研究中心 [2]中山大学逻辑与认知研究所

出　　处：《逻辑学研究》2010年第3期44-66,共23页Studies in Logic

基　　金：教育部招标课题重大攻关项目(批准号:05JZD0008);教育部人文社科青年基金项目(批准号:08JC72040002)

摘　　要：博弈论中重复可允许(Iterated Admissibilty)算法对于快速约简博弈模型、寻找合理置信的纳什均衡具有重要意义,但该算法的认知基础存在悖论。本文构建一个完备的博弈认知逻辑系统EL_G,利用该系统语言描述博弈相关概念和性质,使得我们可以基于EL_G逻辑刻画重复可允许算法,从而达到为该算法提供合理的认知基础,解决算法背后的认知悖论的目的。The algorithm of Iterated Admissibility is significant for the rapid reduction of game models and seeking reasonable Nash equilibria in Game Theory. But there is a paradox in the Epistemic foundation of Iterated Admissibility algorithm. In this paper, we build a complete game epistemic logic systems ELG, and describe some concepts and properities related to a game in the logic language, so that we can characterize the Iterated Admissibility algorithm based on ELG. Thus, we can provide a reasonable epistemic foundation for the algorithm, and in a sense resolve the epistemic paradox for the algorithm.

关 键 词：重复可允许算法 博弈认知逻辑 理性 公共知识 

分 类 号：B812[哲学宗教—逻辑学]