Cartwright on Probabilistic Causality: Types, Tokens, and Capacities

Abstract

I think Nancy Cartwright has written an interesting, informative, and penetrating book that defines a promising combination of realism and empiricism in the philosophy of science. It combines philosophical depth with a commitment to applications to and lessons from actual science. As the reader no doubt already knows, the main theses are that (1) science needs to appeal to unobservable causal connections (or to capacities), and hence that (2) the most promising-or only scientifically tenable-empiricist program in the philosophy of science should be about how to measure the strengths of these connections (how to measure capacities)-taking it that these exist. An important part of Cartwright's argument involves the relations between correct analyses of singular (or token-level) and generic (or type-level, or property-level, or population-level) causation. Roughly speaking, traditional empiricist, or Humean, understandings of causality attempted to reduce whatever there is to the idea of causality to the idea of constant or at least regular associations between or among event types, while Cartwright commits herself to the idea that there is something real about actual individual (singular) causes that bring about effects (that something real being nature's capacities). In this paper, I will not be at odds either with the Humean idea (when appropriately modalized), or with Cartwright's idea that there exist physical capacities. However, Cartwright argues that a correct scientific understanding of generic causal claims (e.g., Taking birth control pills causes thrombosis) requires an understanding of singular causal claims (e.g., Jane's taking birth control pills caused her to get thrombosis). This is intended to support the case for capacities in that (roughly), if it is correct, then, if science needs generic causal laws, then it needs capacities. In this paper, I will dispute Cartwright's argument for a dependence of the type level theory on the idea of singular causes.