ON THE ANALYSIS OF CAUSATION
According to the regularity theory of causation, one set of conditions (or factors or events or states of affairs) causes another to occur only if a set of conditions similar to the latter occurs whenever a set of conditions similar to the former occurs. A well-known difficulty for the regularity theory is its failure to distinguish between cases of accidental correlation and causation. Thomas Reid's example of the coming of day regularly following the coming of night and Russell's example of the workmen leaving the Manchester factory regularly following the sounding of the hooter at the London factory illustrate this difficulty. The regularity theory, further, cannot be salvaged by the following sort of reply. 'Similar', in the regularity formula, means 'relevantly similar': the hooting at the London factory does not cause the workmen's leaving in Manchester because these sets of conditions lack a relevant similarity, for example, that they occur in the same general area. This attempted defense of the regularity theory fails because the resulting analysis is viciously circular. In this context 'relevantly similar' can only mean 'similar in the causally relevant respects'.
Problems of this kind have led some philosophers to suggest analyses of causation in terms of natural laws. Accordingly, one set of conditions causes another to occur only if there are natural laws from which it can be deduced that the latter set of conditions occurs given that the former set occurs. Since 'whenever the hooter sounds at the London factory, the workmen leave the Manchester factory' is an accidental universal and not a law of nature, Russell's case does not constitute a counterexample to the laws-of-nature theory. However, there are difficulties for this theory that closely resemble those of the regularity theory. There are several kinds of natural laws. Consider the developmental law which states that whenever the formation of the circulatory system of an human embryo occurs, the formation of the lungs in the embryo occurs. On the laws-of-nature Synthese 21 (1970) 222-227.