Understanding Causal Inference (David Papineau)

The Logic and Metaphysics Workshop will meet on March 7th from 4:15-6:15 (NY time) in-person at the Graduate Center (Room 5382) for a talk by David Papineau (King’s).

Title: Understanding Causal Inference

Abstract: The current pandemic has focused attention on the techniques used by epidemiologists and other non-experimental scientists to infer causal hypotheses from correlational data. These techniques, which hinge on assumptions about the way causal connections manifest themselves in conditional and unconditional correlations, pose an obvious philosophical challenge. What is it about causation that allows them to work? None of the mainstream accounts of causation—counterfactual, process, dispositional, regularity—casts any light on this question. Probabilistic and interventionist theories of causation do offer a direct response to the challenge, by positing a constitutive connection between causes and correlations, but I shall argue that these theories do not dig deep enough. Instead I shall develop an older idea—which goes back to H.A. Simon in the 1950s—that relates causal relationships to systems of structural equations with probabilistically independent exogenous variables. The attraction of this structural equations approach is that it allows us to view the correlational patterns as fallible evidence for causal relationships, rather than constitutive of them. I shall consider whether this approach can lead to a full reduction of causation and how it might accommodate quantum mechanical unpredictability.

Leave a Reply

Your email address will not be published.

Need help with the Commons? Visit our
help page
Send us a message