著者

苗村 弘太郎

出版者

京都大学文学部科学哲学科学史研究室

雑誌

科学哲学科学史研究 (ISSN:18839177)

巻号頁・発行日

vol.12, pp.21-42, 2018-03-31

---

Circumstantial (indirect) evidence is supposed to be unreliable in historical research. Historians, however, sometimes succeed in establishing a fact based on circumstantial evidence. The reason why circumstantial evidence can achieve it despite its infamous reputation is a question left to be answered. I will give an answer to this question by arguing that credibility of inference in historical research does not depend on whether it is based on circumstantial evidence but on whether its hypothesis is a good explanation in terms of IBE (inference to the best explanation). McCullagh(1984) argues that there are seven explanatory virtues that concern IBE in historical research: some explanation to evidence, explanatory scope, explanatory power, plausibility, ad-hocness, disconfirmation, relative superiority. This criteria can help us understand historical inference based on circumstantial evidence, but it has some problems. Therefore, I will try to modify his criteria in terms of Bayesiansim. I will argue three points. First, plausibility should be interpreted as prior probability in terms of Bayesianism. Second, ad-hocness should be turned into a virtue reflecting degree of unification. Third, how much weight is put on each of explanatory virtues depends on individual historian’s judgement. I will demonstrate these points by a case study.