The causal grounds of really statistical explanations

Abstract

A number of philosophers have argued that purely statistical theorems can non-causally explain empirical matters of fact. Marc Lange has coined the term Really Statistical Explanations (short: RS explanations) for such explanations. A toy example would be an explanation of the average value of a fair die after a large number of throws in terms of the purely statistical law of large numbers. More substantial examples include certain explanations of trait-fitness distributions, and explanations of the average number of particles emitted from a radioactive source. This paper takes issue with the view that RS explanations are non-causal. More specifically, it argues that Lange’s attempts at elucidating what makes such explanations non-causal are unsuccessful. Lange agrees that RS explanations supply information about causes. Yet, he argues, they are not causal, since they do not derive their explanatory power from supplying information about causes. I discuss several readings of this claim, and argue that none of them establishes Lange’s intended conclusion. Finally, I discuss a recent argument by Lange that tries to show that a causal construal of RS explanations cannot do justice to their unifying power, and argue that it is inconclusive.