Repository URL:
Sebastian Lutz
preprint description
Elliott Sober has suggested his contrastive criterion of testability as an improvement over previous criteria of empirical significance like falsifiability or a suggestion within Bayesianism. I argue that Sober’s criterion entails that if one group of people is justified in believing a claim, every group is, and that it tacitly relies on an inconsistent interpretation of probabilistic inequalities. Furthermore, the criterion’s restrictions on the use of auxiliary assumptions are in part redundant and in part unjustified. Most importantly, they are so weak that almost all theories can be contrastively tested. On the basis of these results, I suggest a modification of Sober’s criterion that avoids these problems without abandoning Sober’s core idea.

This preprint has 0 Wikipedia mention.