Retention, Truth-Content and Selective Realism

Abstract

Selectivism, the divide et impera approach, is arguably the most promising realist project today, despite lingering issues regarding the selection of truthful parts in successful theories. Second generation approaches address those issues by basing selection on detailed inferential analyses of impressive predictions drawn from the theory at hand. The leading reformed strategies advocate commitment to just theory-parts that are “absolutely indispensable” for advancing the derivations at hand. The minimalist emphasis involved seems reasonable at first, but it weakens the realist project, leaving many thinkers gloomy about the prospects of scientific realism. This paper has two parts. Part 1 examines the reformed strategies and traces their main difficulty jointly to the pledge they make to interpretive minimalism and a neglect of the epistemic (and realist) import of explanatory power. The result is selections of excessively modest content and vulnerable grounding that deliver less than is needed. Part 2 advances a critical revision and naturalist generalization of the basic strategy. The proposed alternative maintains the focus on inferential analyses of predictions and content-reduction, but without commitment to content minimalism. By moving in a naturalist direction, the suggested strategy brings the valuation of theory-parts in line with confirmational criteria that, in scientific practice, give salience to success and freedom from reasonable doubt. The ensuing proposal recognizes both prediction and explanation as necessary conditions for realist commitment to theory-parts.