Explaining causation: Towards a Humean theory of scientific explanation and causation.

Abstract

Any theory of explanation must capture the intimate connection between explanation and causation, or between explanation and a broader notion of <italic> dependence</italic>, which includes causal and, e.g., mereological dependence. This does not, however, require realism about causation. Following up on a suggestion of P. Kitcher's, I analyze causation in terms of explanation. In order to recover a relation of causation from an augmented and improved version of Kitcher's theory, I develop a notion of an ideal historical explanation: <italic> A</italic> is causally relevant to <italic>B</italic> just in case <italic> A</italic> and <italic>B</italic> are distinct events and there is an ideal historical explanation for <italic>B</italic> that makes essential reference to <italic>A</italic>. The unificatory structure of explanation is crucial to generating a relation with the right structure to be the causal relation. This structure also ensures that the unification approach has the resources to handle the problems of causal overdetermination and preemption. The resulting notion of causation is a regularity account, in which causal relations are underpinned by universal generalizations. This requires me to face the problem of distinguishing laws from accidental generalizations. I argue that the unification approach also has the resources to make this distinction, although the distinction turns out ultimately to be invidious: rather than there being a sharp and exhaustive division of generalizations into laws and non-laws, there is a continuum from the most fundamental laws of nature at one extreme to purely accidental matters of fact at the other. The picture of causation and law that emerges is broadly projectivist in spirit; that is, physical necessity is a projection by us of certain relations of logical necessity onto the world. Causation then has an ineluctable epistemic component. This does not make causation viciously relative or subjective. I distinguish various senses in which causation is and is not objective on the unification view, and argue that it is objective enough to satisfy our intuitions about causation.