Hempel's ravens, the natural classification of hypotheses and the growth of knowledge

Abstract

1. Most solutions offered to Hempel's paradoxes of confirmation have been formal. They have argued that since a true description of scientific justification cannot disobey basic logic, it follows that if it seemingly does so, then either the description is not entirely true or the logic employed not quite basic. Scientific justification is rendered formally consistent, intuitively sound and free of paradox by suggesting new explications of confirmation or modified logics of enquiry culminating in a notion of empirical support whose invariance under transformations of logical equivalence does not yield counterintuitive conclusions. In short they regard the paradoxes as not really being there at all, but as unfortunate and avoidable consequences of bad philosophy. I maintain that solutions of this kind have at most succeeded in shifting the problem to other levels of discourse, (leaving us with impossible methodology or with incomprehensible syntax) and that none have really solved the problem 1. Our interest in the present essay lies however in two of the less populated and far less popular group of writers who have argued that Hempel's paradoxes are a symptom of real incongruity between method and logic. They claim that confirmation by nature eludes basic logic and is therefore not expected to obey the Equivalence-Principle strictly. For them the paradoxes express not so much a puzzle to be solved, but rather facts of life to be studied, fenced in and understood. My reasons for choosing Hempel's Ravens of 1945 as my point of departure are by no means historical. I believe that they are symptomatic of what could be regarded as the central problem of the philosophy of science namely the problem of the rationality of science. For many, even today, the rationality of science means preserving the objective and logical character of the Realm of Justification, which in turn means for most of them simply its formalization. (By "Realm of Justification" I do not wish to imply justificationism of any kind, it is but a name for the activity involved in the treating of hypotheses following and opposed to their discovery.) The paradoxes are an embarrassment to the formalist for they lie at the