Russell, Causation, Determinism

Two arguments due to Russell are examined, and found to show that the notion of causation as full determination doesn’t mesh easily with deterministic global physics and the distinction between effective and ineffective strategies. But a local notion of causation as involving a certain kind of counterfactual dependence is, I argue, compatible with Russell’s conclusions. I defend it from a resurgent form of Russell’s microphysical determinism argument by making some mildly contentious claims about the autonomy of the events posited by the special sciences. 1 Russell’s Arguments Russell (1913) takes the relation of causation to be a relation of determination: c causes e just when c determines e to occur. This relation is supposed to be asymmetric and plausibly transitive as well. The fundamental law of causality is supposed to be that every event has a sufficient cause, one that is guaranteed to bring that event about and in fact did so. This intuition about the deterministic nature of causation is not a Russellian idiosyncrasy: it originates in Hume’s ‘constant conjunction’ regularity analysis (if c and e are constantly conjoined, the appearance of c should be sufficient for the appearance of e), even later accounts like Suppes (1970) keep the idea that individual causes partially determine their effects, and deny that every event has a sufficient cause to avoid the supposed ‘universal law’. Russell thinks that this notion of causation as a determination relation between events doesn’t appear in physics, and hence should be jettisoned from a properly scientific world view. (Perhaps there might be some pragmatic sense in which causation is useful, but there are no deep metaphysical truths about causation.) Field (2003) identifies two arguments in Russell to this conclusion. The first rests on the claim that the equations of microphysics are bi-deterministic. If we fix the microphysical state s of some system R at t, that fixes the whole trajectory of R through the space of states both before and after t.1 If all macroscopic events are constituted by ∗This is a draft version—please do not cite without permission. Originally presented at Causal Republicanism, Centre for Time, University of Sydney, 16/7/03. Thanks to an audience there, in particular Chris Hitchcock, Helen Beebee, Huw Price, Daniel Nolan, Graham McDonald, Charles Twardy, Mathias Frisch, Jason Grossman, Adam Elga and Toby Handfield. †Philosophy Department, Princeton [email protected] 1Earman (1986).