The Two Sides of Interventionist Causation

Publication Year:
Usage 162
Downloads 162
Social Media 6
Tweets 6
Repository URL:
Evans, Peter W.; Shrapnel, Sally
Most Recent Tweet View All Tweets
preprint description
Pearl and Woodward are both well-known advocates of interventionist causation. What is less well-known is the interesting relationship between their respective accounts. In this paper we discuss the different perspectives of causation these two accounts present and show that they are two sides of the same coin. Pearl’s focus is on leveraging global network constraints to correctly identify local causal relations. The rules by which global causal structures are composed from distinct causal relations are precisely defined by the global constraints. Woodward’s focus, however, is on the use of local manipulation to identify single causal relations that then compose into global causal structures. The rules by which this composition takes place emerge as a result of local interventionist constraints (or so the claim goes). We contend that the complete picture of causality to be found between these two perspectives from the interventionist tradition must recognise both the global constraints of the sort identified by Pearl and the local constraints of the sort identified by Woodward, and the interplay between them: Pearl requires the possibility of local interventions and Woodward requires a global statistical framework within which to build composite causal structures.