Abstract

You drop the glass. It shatters. Here there are two distinct events, related by causation. What is this relation? ;I argue that the causal relation is best understood as the relation of being a probability-raiser of a process. I take the causal relata to be property instances at spatiotemporal regions, analyze the notion of a process in terms of sequences of events related by nomic subsumption , and understand probability-raising as counterfactual chance dependence in the style of David Lewis. Thus causes are understood as property instances on which the sequence of property instances constituting the lawful production of the effect depends. ;The understanding of causation offered here synthesizes the leading extant analyses of causation, which divide into probability-raising and process-linkage . I argue that the synthesis offered improves on its predecessors by developing a series of four problem cases: preemption, which shows causation without probability-raising, overlapping, which shows probability-raising without causation, double prevention, which shows causation without process-linkage, and mere traces, which shows process-linkage without causation. These objections interlock: the preempting cause and trace noncause are both process-linked to, but not probability-raising of, their alleged effects, while the double preventing cause and overlapping noncause are both probability-raising of, but not process-linked to, their alleged effects. Understanding causes as probability-raisers of processes resolves this problematic. A continuity requirement is then added which preserves these solutions, renders the relation continuous and transitive, and resolves a further problem case combining preemption and double-prevention involving causation without probability-raising or process-linkage