Exemplar-Based Explanation

The idea that natural phenomena can be explained by means of exemplars or explanations of previous phenomena has often been proposed in the literature but never formally worked out. To the best of my knowledge, no exact mechanism exists that models new phenomena on previous ones, or that describes how problem solutions can be reused to explain new cases in science. In this paper, I attempt to develop such a model. It is not difficult to find inspiration in the literature for constructing an exemplarbased model of explanation. Thomas Kuhn urged that "Scientists solve puzzles by modeling them on previous puzzle-solutions, often with minimal recourse to symbolic generalizations" (Kuhn 1970: 189-190), In the same vein, Ronald Giere claims that scientists possess a large collection of exemplars which makes it possible for them to recognize a new situation as similar to previous situations (Giere 1999). But neither Kuhn nor Giere show how exemplars or problem solutions can be put to use to explain new phenomena. Perhaps Philip Kitcher is on the right track when he urges that "Science advances our understanding of nature by showing us how to derive descriptions of many phenomena, using the same patterns of derivation again and again" (Kitcher 1989: 432; my italics). This brings me the following proposal: new phenomena can be explained by combining partial derivations of previously explained phenomena. But how can we formally do this? One way to go is to formalize derivations in physics by means of derivation trees that describe each step in getting from laws to a description of a phenomenon, and to next use subtrees from derivation trees to derive new phenomena (cf. Bod 1998). Let's start with a simple, idealized example. In the textbook Physics, Alonso and Finn give the following derivation of the Earth's mass from the Earth-Moon system (Alonso and Finn 1996: 247):