贝叶斯条件化原则及其辩护

The Principle of Bayesian Conditionalisation and Its Justification

一、贝叶斯定理与动态假设 贝叶斯方法论（Bayesian methodology）或贝叶斯主义（Bayesianism）早在上世纪20-30年代就被提出，直到上世纪80—90年代才逐渐成为科学哲学领域的一个势头强劲的学派。贝叶斯主义又叫做“主观主义”（subjectivism）或“私人主义”（personalism）。其理论特征主要有二：其一是把概率解释为一个人的“置信度”（degreeofbelief），

Bayesians regard Bayesian inference as the core of inductive inference and hence regard the logicality of Bayesian inference as a solution to the problem of rationality of induction.However,Hacking indicated that the so-called principle of Bayesian rationality depended on an unproven assumption,namely P'(h)=P(h/e),which was termed'principle of Bayesian conditionalisation'.Therefore,it turns the problem of rationality of induction into the problem of rationality of principle of Bayesian conditionalisation.The principle of Bayesian conditionalisation is a special case of the rule of Jeffrey conditionalisation in which P'(e)=1,and both of them imply the requirement of invariance of conditional probability.Howson has made an attempt to justify this requirement,but this attempt is not successful.I put forward here the principle of minimum initial probability to justify the principle of Bayesian conditionalisation and explain its local rationality.