Measurement theory and the foundations of utilitarianism

This article reconsiders the Harsanyi–Sen debate concerning whether Harsanyi is justified in interpreting his Aggregation and Impartial Observer Theorems as providing axiomatizations of utilitarianism. Sen's criticism and its formalization by Weymark are based on the claim that von Neumann–Morgenstern utility theory is ordinal, whereas Harsanyi's utilitarian conclusions require cardinal utility. Proposals for overcoming Sen's objection that appeal to formal measurement theory are considered. It is argued that one of these proposals due to Broome and Risse rightly points to a feature of expected utility theory that was ignored by Sen and Weymark, but that this proposal does not provide a normatively compelling justification for cardinal utility. The other proposal due to Broome is shown to make use of a strength of preference relation in addition to the axioms of expected utility theory.