Causality and the Notion of Necessity

Abstract

What is the main contention that Hume and contemporary Humeans share about inductive inference? It is essentially that no conclusions about the future or unobserved follow with certainty from propositions about what we have experienced in the past. The non-demonstrative nature of induction, it seems, follows inevitably from the truism that induction cannot qualify as deductive inference. The assertion of C and not E is never self-contradictory; hence one cannot say that E must occur. Necessity is not ascribable to causal propositions, except (and here the divergence begins) in a psychological sense (Hume) or in a ‘nomological’ sense (contemporary Humeans). There are, of course, other crucial differences among Humeans. For Hume, conclusions about the future do not even follow with any probability from propositions about past experience, a conclusion which Carnap, for example, tried to avoid by claiming that certain types of probability statements cannot be refuted by subsequent experience. But then the question arises of how such propositions can be construed as relevant in any sense to the future. In any case, we shall not pursue this question but will confine ourselves to the weaker and prima facie legitimate contention, shared by all Humeans, that there is no irreducible sense of natural necessity that is involved in causal judgments.

Revised version of a paper read at the Boston Colloquium for the Philosophy of Science on October 24, 1967. I am indebted to C. J. Ducasse for several points and examples and to Kenneth Barber, Robert Greenberg, and John Woods for helpful criticisms.