Tomoji Shogenji Justification by Coherence from Scratch

Can coherence among independent pieces of evidence make them credible even if each piece has no individual credibility? It is generally agreed among recent epistemologists that coherence of independent pieces of evidence can make them more credible if each piece already has some credibility of its own. Take a group of somewhat questionable witnesses. Even if each witness’s report has only a modest amount of credibility, a close match among many independently produced reports makes people quite confident that they are true. Acknowledging this effect, even the opponents of coherentism in epistemology usually agree that coherence of independent pieces of evidence enhances their existing credibility. Our question, however, is whether coherence among independent pieces of evidence with no individual credibility at all can make them credible. We will call this type of evidential support “justification by coherence from scratch” (JCS for short). JCS has some initial appeal. To continue the example of testimonial evidence, even if we have no trust at all in what each witness says individually, a surprising match among their reports will prompt people to reconsider their assessment, provided it is clear that these reports are produced independently of each other. The reason is that it is highly unlikely—or so it seems intuitively—that unreliable witnesses testifying independently of each other happen to produce matching reports. However, intuition can be misleading especially in matters of probability (Kahneman, et al. 1982). This is not merely a general note of caution in the case of JCS. There have been formal arguments against JCS by the use of the probability calculus (Huemer 1997; Olsson 2002). If JCS is impossible by the rules of the probability calculus, we should regard its intuitive appeal as another instance of bias in our pre-theoretical intuitions. Against this background we aim at three goals in this paper. First, we will confirm formally that coherence does not make independent pieces of evidence credible if each piece has no individual credibility. There are some limitations in the recent formal