Harsanyi’s theorem without the sure-thing principle

We study a version of Harsanyi’s theorem (Harsanyi, 1955) in a framework involving uncertainty. Without assuming the sure-thing principle but sticking to consequentialism, we obtain that a Paretian social aggregation should be affine and that all decision makers must use additively separable preferences that resembles expected utilities. Whenever preferences are state independent, we find that decision makers must be expected utility maximizers and share the same beliefs. The sure-thing principles is hence necessary for the social aggregation to be possible. And consequentialist non-expected utility models cannot be obtained as an aggregation of preferences.