A Study On Cognitive Characters And Affecting Factors In Bayesian Inference

Bayesian inference is a kind of inference about conditional probability reasoning. It has been noticed by psychologists because it involves complicative cognitive processing. Since 1970s, it has become an important project of probability inference research in cognitive psychology. The studies of the project have implications to explore cognitive psychological rules of probability inference and direct probability thinking, judgement, decision making and learning.The previous studies indicated that people's probability estimation was far from the Bayesian theorem when they made Bayesian inference. The aim of the present study is to explore and analyze the reasons of producing Bayesian inference deviation and the subjective and objective factors of affecting Bayesian inference.This dissertation contains two parts: literature review and exprimental study. The literature review introduces the psychological studies about probability inference and Bayesian inference in Western countries, it includes: the studies on phenomenon of base rate neglect; factors affecting Bayesian inference; main theories and methods, it also points out the limitations of the prevous studies: first, there exist controversial issues about the studies of phenomenon of base rate neglect. Second , the studies on relationships between problemrepresentation and Bayesian inference are seldom found. Third, Seldom studies have been done through combining subjective and objective factors. Fourth, the method of using correct rate as index to test relevant hypothesis seems to be singleness.In order to solve the above mentioned problems, the present study have designed expriments in three aspects: First, the phenomenon of base rate neglect is explored and analyzed in deepth and the function of false alarm rate in Bayesian inference is examined. Second, the cognitivepsychological mechanism of Bayesian inference is studied. Finally, the subjective and objective factors of affecting Bayesian inference are systematically discussed. The subjects in the expriments are university students. Multiple respondent variable, such as, Bayesian inference post probability estimation, probability estimation veracity ( deviation between subjects' estimation and normative correct percentage ), correct rate and reaction time, are adopted in accordance with different research purposes.The expriment part contains four groups of expriments. the first group of expriments discuss the functions of different probability information in Bayesian inference, the purpose is to clarify the dispute of phenomenon of base rate neglect and examine if false alarm rate affects the Bayesian inference deviation. This group of expriments include three expriments. expriment 1 discusses how different base rate affects Bayesian inference; expriment 2 discusses how the positions of base rate information affect Bayesian inference; expriment 3 discusses how the different proficiency of base rate and false alarm rate affect Bayesian inference.Generally speaking, the correct rate of probability estimation of the subjects is low in different exprimental conditions. As related to the standard value of Bayesian theorem, the subjectstend to estimate higher in the condition of lower base rate and estimate lower in the condition of higher base rate. This is a new finding of the present study, this phenomenon can be called as "teeterboard effect", the results of the expriments also show that the post probability estimation of higher base rate proficiency group is higher than the lower base probability proficiency group, but the correct rate of probability estimation in both groups are low, positions of base rate have no significant influence on post probability estimation.Different false alarm rate proficiency levels do not affect probability estimation value, but there exists interactoin between base probability proficiency and false alarm rate proficiency. In the conditions of low base probability, if the false alarm rate i...