Does Lewis’ Theory of Causation Permit Time Travel?

Phil Dowe

doi:10.3390/philosophies6040094

Abstract

David Lewis aimed to give an account of causation, and in particular, a semantics for the counterfactuals to which his account appeals, that is compatible with backwards causation and time travel. I will argue that he failed, but not for the reasons that have been offered to date, specifically by Collins, Hall and Paul and by Wasserman. This is significant not the least because Lewis’ theory of causation was the most influential theory over the last quarter of the 20th century; and moreover, Lewis’ spirited defence of time travel in the 1970s has shaped philosophers’ approach to time travel to this day.

Highlights

IntroductionLewis’ counterfactual theory of causation is to note some of its distinctive entailments about actuality: (1) A cause is a nomically necessary condition, given the actual circumstances, for its effect, where ‘nomically’ means according to the laws, which for Lewis are given by the best system analysis of the actual distribution of particulars; and (2) a cause exhibits asymmetry of overdetermination in the direction of the effect, that is, toward the past or toward the future depending on whether the effect is respectively in the past or in the future of the cause
In the case of a large complex effect like the appearance of a person, we have seen that we do not get causation because it is not possible to hold fixed the events in the background of the cause, and nothing in the time travel region can be held fixed by the similarity measure
The reason is that when a cause c has effects in its own causal past, the closest ~c worlds will be worlds which do not hold fixed the background of c, in general, an earlier effect of c will not be absent from all the closest ~c worlds; ~c worlds which do hold fixed the background of c contain large miracles, and are not the closest worlds

Summary

Introduction

Lewis’ counterfactual theory of causation is to note some of its distinctive entailments about actuality: (1) A cause is a nomically necessary condition, given the actual circumstances, for its effect, where ‘nomically’ means according to the laws, which for Lewis are given by the best system analysis of the actual distribution of particulars; and (2) a cause exhibits asymmetry of overdetermination in the direction of the effect, that is, toward the past or toward the future depending on whether the effect is respectively in the past or in the future of the cause. The condition for causation implicit in the similarity relation is not ‘a cause exhibits asymmetry of overdetermination in the direction (past or future) of the effect’, but rather, (2)* a cause exhibits asymmetry of overdetermination in the spatiotemporal region containing the effect compared to some other region very close to c To reject this point is to accept the specious argument, I would claim, and among other things that would leave one wondering how Lewis ever thought his account could get off the ground. The assumptions are very restrictive, perhaps too restrictive to be of interest, but it does give us an exception to the argument I am going to give

The Main Argument

Conclusions