Abstract

This book focuses upon the crucial metaphysical question “What is it for one event to cause another?,” and the view defended is, first, that all reductionist accounts of the nature of causation are open to decisive objections, so that a realist account of causation must be correct, and second, that among realist accounts of causation, the only type of account that can possibly be correct is one on which causation is a theoretical relation between events since causal relations cannot be directly and immediately observed. After a brief introduction and overview, then, the idea that causation is a relation that, like spatial relations, can be directly observed, is considered and shown to be impossible, since for any possible experience where it might be maintained that one is directly aware of a causal relation between two events, there could be an experience that was perceptually identical in which there was no causal relation between events. With the view that causal relations are directly observable ruled out, the idea of reductionist analyses is considered, and here there are several contemporary approaches that need to be examined, including accounts in terms of laws of nature, interpreted in a Humean fashion; counterfactual accounts of causation; probabilistic accounts involving empirically based, statistical relations; and metaphysically robust accounts in terms of powers, propensities, and dispositions. With all such reductionist accounts shown to be untenable, an extremely simple realist account in which the idea of logical probability is central is set out and defended against objections.