摘要
关于贝叶斯主义的合理性,有两个最为重要的哲学问题:(1)形而上学问题:是否存在着一个关于先验概率P(h)的事实;(2)认识论问题:如何对一个先验概率归赋或者为一个概率陈述进行辩护。文章基于心智的预测加工模型,尝试勾画和发展一个回应先验概率的认识论问题的方案。文章将表明,从预测加工模型的层级化架构、从因果规律性在层级上的排布等特征来看,一方面可以从认知的内部机制出发,为先验概率奠定心理认知基础;一方面能够以外部世界的实际输入作为“客观”参数,对先验概率的内容进行修正与监督。基于这两方面的分析,可为先验概率的主观主义立场提供新的解释方案。最后,文章讨论这一方案如何可能回应客观贝叶斯主义的诘难。
There are two most important philosophical questions about the rationality of Bayesianism:(1)metaphysical question:whether there is a fact about prior probability P(h);(2)epistemological question:how do we assign a prior probability or defend a probability statement?Based on the predictive processing model of mind,this paper attempts to outline and develop an epistemological solution to the prior probability problem.We will show that,from the hierarchical framework of predictive processing model and the hierarchical arrangement of causality,on the one hand,we can lay the psychological cognitive foundation for prior probability from the internal mechanism of cognition.On the other hand,the actual input of the external world is taken as the objective parameter to correct and supervise the content of prior probability.Based on these two aspects,it can provide a novel interpretation for the subjective perspective of prior probability.Finally,we discuss how this scheme might respond to the challenge of objective Bayesianism.
作者
万舒婵
王静
WAN Shu-chan;WANG Jing(Department of Philosophy&Religious Studies,Peking University,Beijing 100871,China;School of Humanities,Tongii University,Shanghai 200092,China)
出处
《科学技术哲学研究》
CSSCI
北大核心
2023年第3期25-31,共7页
Studies in Philosophy of Science and Technology
基金
同济大学学科交叉联合攻关首批示范重点项目“以人为本的人工智能伦理治理”(ZD-110-202165)
关键词
先验概率
预测加工模型
知觉过程
贝叶斯主义
prior probability
predictive processing model
perceptual processing
Bayesianism
作者简介
万舒婵(1997-),女,山东青岛人,北京大学哲学系(宗教学系)博士研究生,研究方向为科学哲学、认知哲学等;王静(1968-),女,重庆人,同济大学人文学院哲学系教授,研究分析为科学哲学、分析哲学和现代知识论。
