信号检测论与贝叶斯决策理论的关系  

The Relationship Between Signal Detection Theory and Bayesian Decision Theory

作　　者：胡啸 HU Xiao(Faculty of Psychology,Beijing Normal University,Beijing 100875,China)

机构地区：[1]北京师范大学心理学部,北京100875

出　　处：《心理技术与应用》2023年第9期542-558,共17页Psychology(Techniques and Applications)

基　　金：国家自然科学基金项目(32200841,32171045)。

摘　　要：信号检测论被广泛用于解释个体在不同类型认知任务中的决策过程。然而,经典信号检测论的重要不足之处在于难以进一步解释个体设置报告标准的过程对应怎样的内在心理机制。本文从贝叶斯决策理论的视角出发,深入探讨个体在信号检测任务中的决策规则。首先基于贝叶斯定理介绍贝叶斯决策理论的基本观点。随后探讨贝叶斯决策理论如何解释理想观察者的决策规则,以及在实际的信号检测任务中个体的决策结果与理想观察者之间的偏离。其次探讨经典信号检测论和贝叶斯决策理论在不等方差信号检测模型中的差异。最后简要介绍支持贝叶斯决策理论的实证研究证据。Signal detection theory(SDT)has been widely applied to explain the decision-making process in different types of cognitive tasks.However,one important limitation of classical SDT is that it is difficult to illustrate the mental mechanisms underlying the setting of response criterion.The current article discusses the decision rule in signal detection tasks from the perspective of Bayesian decision theory(BDT).I first introduce the basic idea of BDT based on the Bayes'theorem.Next,I discuss how BDT explains the decision rule of ideal observer,and characterizes the deviation between actual participants and ideal observer in empirical signal detection tasks.I then discuss the difference between classical SDT and BDT in unequal variance signal detection model.Finally,I briefly introduce the empirical research evidence supporting BDT.

关 键 词：信号检测论 贝叶斯决策理论 先验概率 似然函数 报告标准 

分 类 号：B841[哲学宗教—基础心理学]