Decision, cause and counterfactuals

Abstract

Newcomb's problems have been advanced as counterexamples to any decision theory employing the principle of maximising conditional expected utility. David Lewis has proposed a revision of decision theory. He insists that only beliefs about the causal consequences of an act can count as a rational basis for choice. Lewis suggests that these beliefs are best expressed by counterfactual conditionals. I will argue that Lewis's claims about the way these expressions are standardly understood are unsupportable, and if these expressions are used in a decision context the paradoxical element in Newcomb's examples persist.