Joseph C. Pitt

SummaryHempel's Deductive‐Nomological model of explanation is compared to Sellars' brand of essentialism. The source of their differences is shown to lie in their views on the explanatory role of inductively based generalizations. An adequate explanation requires a reasoned account of why an empirical generalization fails. On Sellars' view this entails concentrating on the nature of the things whose behavior is in question. We thereby remove ourselves from the misleading positivist methodology in which one counterinstance renders a generalization uninteresting. It is suggested instead that “disconfirmed” generalizations are of crucial positive methodological importance for purposes of theory development.RésuméLe modèle déductif‐nomologique de Hempel est comparéà l'essentialisme dans la version qu'en propose Sellars. L'auteur montre que leurs différences proviennent de ce qu'ils jugent différemment le rôle explicatif des généralisations inductives et les raisons de l'échec d'une généralisation empirique. D'après Sellars, il faut par conséquent se concentrer sur la nature des choses dont on observe le comportement; on évite ainsi les problèmes méthodologiques soulevés par les théories positivistes standard de la confirmation. L'auteur conclut en montrant que des généralisations infirmées ont une importance positive capitale pour le développement des théories

The essays in this volume grew out of a seminar examining the possibility of the emergence of a new consensus in the philosophy of science. While that issue is not resolved, we are presented with the most thorough examination of problems associated with the deductive-nomological model of explanation and its variants since the publication of Hempel's Aspects of Scientific Explanation and other Essays in the Philosophy of Science. The discussion begins with Wesley Salmon's monograph-length review of the past forty years of work in the tradition initiated by Hempel and Oppenheim in their groundbreaking article "Studies in the Logic of Explanation". As one of the major players in the debates, Salmon's personal account is informative; it provides a useful introduction to the topic and covers the recent history of work on explanation in a manner that allows the uninitiated to follow the arguments and intricacies of the essays that follow. A main theme in Salmon's essay, as in much of his work, is the relation between the theory of explanation and the concept of causation. Given that the notion of causal laws plays a dominant role in DN explanations, this discussion is most welcome. Causation is also a clear concern of many of the other papers. The resolution of some of the tensions between Salmon's approach to causation, and the various roles of explanation in the philosophy of science, is, in fact, one of the main objectives of the extended concluding piece by Philip Kitcher. For Kitcher, explanations serve as unifying mechanisms for theories within different disciplines. Unification in turn also serves as a criterion for choosing between competing theories. The interplay between, on the one hand, Salmon's attention to the problems deriving from Hempel's initial formulation of DN and, on the other hand, Kitcher's concerns with Salmon's conceptualization of causation, produces one of the more fascinating dialectics of the volume. By attending to their dialogue, we see just how far the theory of explanation has come in the last forty years and yet how slow progress can be when fundamental problems are deeply entrenched. Causation continues to bedevil us. With one exception, the remaining contributors to this volume seem to confirm that judgment. Matti Sintonen, Paul Humphreys, David Papineau, Nancy Cartwright, James Woodward, and Merilee Salmon explore the many facets of causation as it applies to clarifying our explanatory objectives. Peter Railton is an exception. In his paper Railton reminds us that behind the epistemological and pragmatic concerns of most of the other contributors, Wesley Salmon included, there lies a set of metaphysical concerns. A thorough analysis of causation and, hence, a resolution to many of the DN model's problems requires a logically prior assessment of the relation between the epistemology of causation and the metaphysics of necessity. How one decides that issue or cluster of issues, Railton reminds us, will play heavily in the controversies between realists and nonrealists over the proper use of the principle of inference to the best explanation.