Abstract
Theories of actual causation formulated in the framework of causal models have made salient an interesting puzzle concerning the relationship between counterfactual dependence and causation. Even when theorists agree on the causal model appropriate for representing a particular situation, they sometimes disagree on which variable values represent the actual causes. This is puzzling, for the causal model represents all of the relevant counterfactual dependencies and thereby, apparently, all of the information that is relevant to prediction, control and explanation. What is the point of a concept that picks out one particular aspect of this structure? I give this question a more precise formulation, and suggest the outline of an answer.
