PHI403 Lecture 20 — Plural Methods, One Causation?

Course's capstone, presenting its positive programme: ontological causal monism / primitivism + methodological pluralism.

Causal primitivism — causation is one single, unanalysable thing; it cannot be reduced to correlation, counterfactual dependence, manipulation, or probability-raising.
Symptoms, not definitions — each proposed theory of causation has isolated a symptom of causation, not causation itself. Symptoms: difference-making, regularity (though not constant conjunction), manipulability, probability-raising, energy transference.
A symptom accompanies the phenomenon "more or less reliably" (crying is a symptom of sorrow, not identical with it). None of causation's symptoms is universally present.

Evidential pluralism: use multiple methods, each attuned to different symptoms, to identify causation. This is the course's version of Feyerabend's "against method" and Longino's pluralism — no single approach is complete.

Default principle for method acceptance: a method is good if it correctly identifies cases of causation (and non-causation) via its symptoms more often than it fails.

When methods conflict, we need an evidence-ranking scheme — but one that is context-sensitive, not a fixed hierarchy like EBM's.

Connections

Back-link: sources/2023-anjum-rocca-phi403-causation-in-science. Concepts: concepts/causal-pluralism · concepts/methodological-pluralism · concepts/evidence-hierarchy · concepts/causation · concepts/dispositionalism.