Probabilistic Causal Interaction and Disjunctive Causal Factors

The basic idea in probabilistic theories of causality is that causes raise the probability of their effects. Of course, it is necessary to control for other causes of the effect in question in order to avoid being misled by “spurious correlations”, which arise in cases of “Simpson’s paradox”. But, even when we control for other causes, it is possible for a causal factor to raise the probability of a second factor in some situations and lower that probability in other situations, where these situations do not correspond to the presence or absence of other causes of the second factor. When this happens, the reason may be that the causal factor “interacts” with other factors. In this paper, I will explore the phenomenon of probabilistic causal interaction in detail and in general, and I will suggest a general way of accommodating the possibility of causal interaction in one common current understanding of probabilistic causation.