Causal Modeling and the Efficacy of Action

Abstract

This chapter brings together naïve action explanation with interventionist causal modeling to demonstrate a method to produce causal models capable of incorporating actions alongside other causal relationships. Standard causal modelling is supplemented with a new assumption that, if it holds, enables stronger causal inferences: the Rationalization condition. The internal structure of action, where stages of a temporally extended action are unified by rationalizations, cannot be reduced to mere causal connection without genuine loss. Despite the fact that action explanation necessarily exceeds causal explanation in strength, existing causal modeling techniques can be expanded to model action in some cases by relying on the Rationalization condition, as an addition to conditions such as Causal Markov or Causal Faithfulness. Converting rationalization as found in naïve action explanation into such a condition for structural causal modeling strengthens existing causal modeling techniques’ inferential power in cases where the causal system being modelled involves actions.