Liliana Bokros
Comenius University Bratislava
Since Van Fraassen’s argument in 1980, there has been a lively debate regarding the compatibility of Bayesianism and Inference to the Best Explanation with respect to the hypotheses assessment.
In response, various authors have challenged Van Fraassen’s position. Some, e. g. Weisberg (2009) and Psillos (2004), have stated that reconciling IBE and Bayesianism would require a radical, objectifying transformation of Bayesianism.
Others (e.g. Okasha 2000, Lipton 2004, Dellsén 2017) have maintained that IBE and Bayesianism can indeed be compatible in several aspects. Okasha (2000) suggested that the explanatory considerations central to IBE can inform the estimation of priors and likelihoods – key components of Bayes’ theorem. Similarly, Lipton (2004) proposed three possible strategies for integrating IBE and Bayesianism. Explanatory considerations might help the Bayesian formula to determine i) the likelihood; ii) the priors; and iii) the relevant evidence.
Dellsén (2017), on the other hand, views IBE as a heuristic within Bayesian reasoning—a tool for identifying the most probable explanatory hypothesis to serve as a working assumption for further investigation.
While I do not categorically deny the possibility of a compatible relationship between the two, I suppose that the circumstances and conditions of specific case of hypotheses assessment must be given in order to consider a model based on their cooperation. In this paper, I argue that the compatibilist position which treats explanatory considerations as determinative factor of prior probabilities in Bayes’s theorem is problematic so far as the epistemic status of the evidence is not clearly characterized.
This is mainly because what counts as the best explanation is determined by specific evidence that has already been obtained. Consequently, it is not possible to assign some degree of belief Pr(H) to a hypothesis without taking the evidence into account, since we already have it (we are not reliant on only background knowledge). Furthermore, the prior probability Pr(E), as a result of already having the evidence, would be 1. This would support the view, that the IBE and Bayesianism both operates on a different explications of term “evidence”, what (with its implications) should be considered by compatibilist before marrying the two.
Chair: Lucas Timmermann
Time: 03 September, 16:50 – 17:20
Location: SR 1.005
SOPhiA 2026 