<研究論文(原著論文)>信念を分節する: 推論主義による信念のパズルの解決

Abstract

Frege's puzzle of beliefs is one of the most traditional but still unsolved problems in analytic philosophy. While co-extensional terms are substitutable in referentially transparent contexts without change in truth value, they are sometimes not substitutable within propositional attitude reports. In addition, Saul Kripke suggests that this non-substitutability can lead to a certain irrationality or contradiction. These mysteries or “puzzles” threaten the standard representational conception of meaning. Robert Brandom proposes a non-representationalist alternative called “normative inferentialism” and argues that it can solve the puzzles of beliefs. Brandom claims that belief ascriptions must be interpreted from the believers' own perspectives, i.e., the inferential roles they assign to their beliefs. However, there are many substantive criticisms against Brandom's theory and his analysis of belief ascriptions. According to Carlo Penco and Ronald Loeffler, the analysis makes beliefs unintelligible to others and it is inadequate in the first place as a description of our ordinary practice of belief-ascription. In this paper, I propose a new solution to the puzzles from inferentialism by keeping the basic idea and modifying it to overcome these criticisms. I argue that the contents of belief ascriptions are determined by inferences that relevant persons, i.e., the attributors, hearers, and believers commonly endorse. By considering the contents of belief ascriptions to be partially constrained by the inferences the believers endorse, my analysis can appropriately predict when the substitutions of co-extensional terms fail without attributing any irrationality or contradiction to the believers. Since the contents also partially depend on the perspectives of the attributors and hearers, the beliefs are intelligible to them. Moreover, my analysis can explain our ordinary practice of belief-ascription better than Brandom's and Loeffler's.