A Naturalist Generalization

If the previous considerations are correct, then recognizing specific theoretical content as crucial (or not) to deriving successful predictions from a given theory should rest on compelling specific reasons.

One appealing platform, invoked by naturalists of a previous generation, rests on scientific appraisals of scientific success and freedom from compelling specific doubts—in Dudley Shapere's vocabulary (1982, 1984). Can this “first approxi­mation” platform help the inferential selectivist approach? No single prediction outcome taken in isolation can stamp much of a seal of reliability or unreliability on any functioning posit. To do better, one needs to analyze clusters of derivations of different predictions (and phenomena) in which the theory-part in question plays a role. Acting on this suggestion, the remainder of this paper tries to develop an alternative to minimalism, picking out for realist commitment only claims that are taken to be both predictively successful and free of compelling specific doubts by scientific standards. How might a selectivist convincingly identify theory-parts likely to have high truth-content, and do this with projective force? Two selection strategies common in scientific assessments seem especially relevant to the task at hand (Cordero 2015, 2016). One focuses on inference analyses applied to repre­sentative sets of the predictions a theory T makes as it gains applications. The other strategy looks for independent support (from sources external to the theory) that may be available for parts left by T as assumptions or postulates that then play a role in the derivations at hand. As these two lines do their jobs during the life of a theory, they may concentrate confirmational weight on some theory-parts; when that occurs there is a promising case for selective realism.

5