Abstract

This article presents a new argument for incorporating a distinction between default and deviant values into the formalism of causal models. The argument is based on considerations about how causal reasoners should represent disagreement over causes, and it is defended against an objection that has been raised against earlier arguments for defaults.