Abstract

1.1 In 1955, John Harsanyi proved a remarkable theorem:1 Suppose n agents satisfy the assumptions of von Neumann/Morgenstern (1947) expected utility theory, and so does the group as a whole (or an observer). Suppose that, if each member of the group prefers option a to b, then so does the group, or the observer (Pareto condition). Then the group’s utility function is a weighted sum of the individual utility functions. Despite Harsanyi’s insistence that what he calls the Utilitarian Theorem embeds utilitarianism into a theory of rationality, the theorem has fallen short of having the kind of impact on the discussion of utilitarianism for which Harsanyi hoped. Yet how could the theorem influence this discussion? Utilitarianism is as attractive to some as it is appalling to others. The prospects for this dispute to be affected by a theorem seem dim. Yet a closer look shows how the theorem could make a contribution. To fix ideas, I understand by utilitarianism the following claims: (1) Consequentialism: Actions are evaluated in terms of their consequences only. (2) Bayesianism: An agent's beliefs about possible outcomes are captured probabilistically. (3) Welfarism: The judgement of the relative goodness of states of affairs is based..