Advancing the counterfactual analysis of causation
Abstract (Summary)What does it mean to say that one event is a cause of another? The simplest counterfactual analyses identify causation with one of two counterfactual-dependence relations: (1)Ã?Â if event c had not occurred, then (distinct) event e would not have occurred; (2)Ã?Â if c had not occurred, e 's probability would have been lower. These analyses enjoy some success. For the first: the dart-throw caused the balloon-pop, because if the throw had not occurred, the pop would not have occurred. For the second: suppose two radioactive samples, A and B, are introduced into a room containing a Geiger counter, and the counter clicks once due to an emission from an A-atom; then the introduction of A is a cause of the click; the click might have occurred without the A-introduction (a B-atom might have emitted), but we can say at least that without the A-introduction, the probability of the click would have been lower. Ultimately, however, these analyses fail, for two clear reasons. Preemption : add to the dart scene that Lucy would have thrown her dart if I had refrained--then although my throw caused the pop, the pop is not dependent on my throw. Failed potential causes : the probability of the click would have been lower without the B-introduction, but the B-introduction did not actually succeed in causing of the click. I defend the two simple analyses against various other objections; I then try to home in on the precise nature of their genuine problems; I examine almost all the attempts to date to improve upon the simple analyses; and finally I propose two new analyses. One of my analyses is deterministic, the other is confined to worlds that are purely indeterministic. Both analyses take causation to be primarily a matter of counterfactual dependence in the circumstances : holding certain features of the world fixed, effects and their probabilities do indeed depend (almost exclusively) on their causes (or their direct causes). My discussion of background issues--counterfactual semantics, objective chance, events--includes arguments for substantial simplifications of David Lewis's theory of events.
School Location:USA - Massachusetts
Source Type:Master's Thesis
Date of Publication:01/01/2003