Abstract

The problem of causation by absences is poorly understood; numerous accounts hold absence-causation to be different than "normal" causation, but none has fully explained the nature of the difference. Several recent theories have proposed persistence as the basis of the causal relation; these theories, which I call "process-persistence" theories, successfully resolve the preemption problem, but cannot account for cases of absence-causation because absences lack persisting properties. The differences in approach of process-persistence theory and traditional theories offer new insight into the nature of causation by absences, for which I use the term non-persistence causation. ;I demonstrate how process-persistence theory both solves the preemption problem and fails to account for absence-causation, and respond to objections that absences cannot be causes or effects. I observe, further, that process-persistence theory employs constraints on relata which require the "redescription" of many events in order to perform a causal analysis. The solution proposed for the redescription problem is that higher level events are realized by lower level events, based on a notion of realization understood as a "truthmaking" relation, with the realizers of events corresponding, roughly, to the various ways in which a higher level event could occur. On this view, the causal relation is grounded by persistence relations between particular properties of lower level events which I call base-realizers. Absences are higher level events which lack base-realizers because their causally relevant features do not persist and their persisting features are non-relevant. I apply the concept of realization to cases of non-persistence causation, proposing a methodology of establishing causation by identifying the relevant features of absences by examining the sequences in which we find them, and seeking persistence connections between the non-relevant properties of their realizers. I conclude that absence-causation is properly analyzed using the same criteria that we use for analyzing cases of "normal" causation, but that those criteria must be examined separately; consequently we cannot attain definitive causal determinations in absence-causation cases because of the fact that we have separated the notions of persistence and relevance