后向归纳法的动态认知刻画
发布时间：2018-09-17 11:31
【摘要】：后向归纳法BI(Backward Induction)是求解动态博弈的经典算法,其认知机制的探讨多是基于静态的认知模型展开的。这样,为了给BI算法结果中具有反事实性的理性行动提供合理置信的解释,一些非平凡的条件被添加到这类认知模型中,形成多种较为复杂的条件知识(或信念)或层级式(Hierarchical)知识(或信念)系统。我们构建了一类博弈认知模型,基于公开宣告逻辑PAL(Public Announcement Logic),实现博弈认知模型的动态更新,论证了在完美信息动态博弈中,选手间的理性公共知识能够导致BI算法结果,为该算法的认知条件提供了一种新的逻辑刻画。这种刻画没有涉及选手策略等博弈概念,通过利用PAL中模型更新的动态性来描述动态博弈中的BI算法认知条件,不会受到通常BI算法认知刻画理论中所涉及的反事实(无论是主观还是客观)推理的影响,从而有效地避免了复杂的条件信念(或知识)系统或层级式知识(或信念)和信念修正的问题。
[Abstract]:Backward Induction (BI) is a classical algorithm for solving dynamic games, and the study of its cognitive mechanism is mostly based on static cognitive models. In order to provide a reasonable and credible explanation for the counterfactual rational actions in the results of BI algorithm, some non-trivial conditions are added to such cognitive models and formed. A variety of complex conditional knowledge (or beliefs) or hierarchical knowledge (or beliefs) systems are constructed. We construct a game cognitive model based on public announcement logic PAL (Public Announcement Logic) to dynamically update the game cognitive model and demonstrate the rational public knowledge among players in dynamic game with perfect information. It can lead to the result of BI algorithm and provide a new logic description for the cognitive condition of the algorithm. This description does not involve the game concepts such as player's strategy, and describes the cognitive condition of BI algorithm in dynamic game by using the dynamics of model updating in PAL. It is not subject to the counter facts involved in the cognitive Characterization Theory of BI algorithm. The influence of both subjective and objective reasoning effectively avoids the problems of complex conditional belief (or knowledge) systems or hierarchical knowledge (or belief) and belief revision.
【作者单位】： 中山大学哲学系、逻辑与认知研究所;
【基金】：国家社科基金资助项目(12CZX056) 教育部人文社会科学重点研究基地重大项目(15JJD720014) 广东省哲学社会科学“十二五”规划青年项目(GD11YZX03)的阶段性成果
【分类号】：B812.3
本文编号：2245748
[Abstract]:Backward Induction (BI) is a classical algorithm for solving dynamic games, and the study of its cognitive mechanism is mostly based on static cognitive models. In order to provide a reasonable and credible explanation for the counterfactual rational actions in the results of BI algorithm, some non-trivial conditions are added to such cognitive models and formed. A variety of complex conditional knowledge (or beliefs) or hierarchical knowledge (or beliefs) systems are constructed. We construct a game cognitive model based on public announcement logic PAL (Public Announcement Logic) to dynamically update the game cognitive model and demonstrate the rational public knowledge among players in dynamic game with perfect information. It can lead to the result of BI algorithm and provide a new logic description for the cognitive condition of the algorithm. This description does not involve the game concepts such as player's strategy, and describes the cognitive condition of BI algorithm in dynamic game by using the dynamics of model updating in PAL. It is not subject to the counter facts involved in the cognitive Characterization Theory of BI algorithm. The influence of both subjective and objective reasoning effectively avoids the problems of complex conditional belief (or knowledge) systems or hierarchical knowledge (or belief) and belief revision.
【作者单位】： 中山大学哲学系、逻辑与认知研究所;
【基金】：国家社科基金资助项目(12CZX056) 教育部人文社会科学重点研究基地重大项目(15JJD720014) 广东省哲学社会科学“十二五”规划青年项目(GD11YZX03)的阶段性成果
【分类号】：B812.3
【相似文献】
相关期刊论文 前1条
1 沈洁;潘天群;;动态博弈中的言语行为分析[J];天津商业大学学报;2008年01期
，本文编号：2245748
本文链接：https://www.wllwen.com/shekelunwen/ljx/2245748.html
