Abstract
Causal models, such as structural equation models and causal Bayes nets, naturally provide a conception of causal paths and path-specific effects. These ideas have also been used in accounts of actual causation. In this talk, I explore the questions of why we are interested in causal paths, and why we tend to be interested in some kinds of causal path rather than others. I will argue that knowledge of causal paths is not necessary for one-shot decisions, but is crucial for plans that involve multiple steps.
