以部分后验概率表征后验概率是可行的吗?

斯特雷文斯用部分后验概率P(H/┐(H&A))(文中的Q(H))来表征和预计后验概率P(H/E)(文中的P+(H)),尝试为奎因-迪昂问题(Q-D)的解决以及特设性假设评价标准问题提供一种客观贝叶斯主义方案。这一方案是否可行的关键是P(E/H&┐A)/P(E/┐(H&A))=1+ε是否是可满足的。斯特雷文斯揭示出在重要的Q-D情况(文中的情况3)下,该条件是可满足的。笔者通过对斯特雷文斯方案的分析,进一步回应费特尔森等对此方案的质疑,认为此方案是基本可行的。有待进一步研究的问题则是独立性假设的合理性问题。

Is It Feasible to Represent the Posterior Probability by the Partial-posterior Probability?

Strevens tries to provide an objective Bayesian approach to solve the Quine-Duhem(Q-D)problem and the problem of ad hoc hypotheses by using Partial-posterior Probability to represent the Posterior Probability.Whether it is feasible depends on whether the condition "P(E/H&┐A)/P(E/┐H&A)=1+ε"could be satisfied.Strevens shows that in the important Q-D circumstance,the condition could be satisfied.According to the arguments of Stevens and my replying to the objection generated by Fitelson and Waterman,I claim that Strevens’ appoach is basically feasible.However,it leaves over a problem about the rationality of the hypothesis of Independence for further inquiry.