Reflections of a bashful Bayesian: A reply to Peter Lipton.

Citation: Salmon, W. C. (2001) Reflections of a bashful Bayesian: A reply to Peter Lipton.. Explanation (pp. 121-136). Springer Netherlands. (RSS)
Internet Archive Scholar (search for fulltext): Reflections of a bashful Bayesian: A reply to Peter Lipton.
Tagged:

Summary

In this paper, Salmon responded to the inference to the best explanation proposed by Lipton, and then made some comments, especially for the relationship between Bayesian and Explanation. Salmon distinguished the difference between explanationist and Bayesian by discussing the probability and informative value. The explanationist suggests that a good hypothesis is based on higher probability. On the contrary, Bayesian thinks that explanatory power is based on higher informative value instead of higher probability.

Based on Lipton’s argument which loveliest explanation is a guild to likeliest explanation, Salmon thinks explanationism and Bayesianism may be complementary to Lipton. Lipton suggests that explanatory consideration will improve the Bayesian mechanism in four ways: first, the explanatory consideration can determine the likelihood in order to move from prior probability to posterior probability. Second, we can infer the conditions based on the prior probability. Third is to determine which data are relevant to hypothesis. Fourth, the explanatory consideration might become a theory.

However, Salmon disagreed that explanatory consideration can help the Bayesianism. For the likelihoods, he thinks that without explanatory consideration, we can also evaluate the likelihood for something based on previous observation. In addition, prior probability was determined by existence of evidence or probability of the hypothesis, which means that simple hypotheses will be more successful due to higher probability. This leads to the problem of oversimplification duo to less explanatory power, and it is hard to apply to social sciences.

Another difference between Salmon and Lipton is the methodology. Lipton thinks that the explanation is descriptive, but Salmon suggests it is more normative. Salmon criticized that Lipton’s inference to the best explanation model does not provide a normative basis for judging inference. In addition, which is the best explanation is hard to determine duo to many different pieces of evidence of real case. Salmon then concluded that Lipton’s model is not scientific but good commonsense explanation.