Under stochastic dominance Choquet-expected utility and anticipated utility are identical (Q756625)

According to L. J. Savage a decision maker chooses between acts, i.e. functions from a set of states into a set of consequences. In Choquet expected utility (CEU) a decision maker's preferences over acts f are represented by an integral \[ (1)\quad \int^{\infty}_{0}v(\{s\in S:\;U(f(s))\geq \alpha \}d\alpha, \] where S is a set of states, U a utility function, and v a Choquet capacity (i.e. a set function, increasing and normalized). In anticipated utility (AU) his preferences are represented by \[ (2)\quad \int_{S}U(f(s))d(\psi (P(s)) \] with some utility function U, underlying probability P, and some increasing transformation \(\psi\) of the unit interval. It is obvious that, with \(v=\psi \circ P\), (2) is a special case of (1), but not vice versa. Thus, CEU includes AU, but is more general. The author shows that if an underlying probability P is given and a CEU preference fulfils first degree stochastic dominance then there exists an increasing transformation \(\psi\) such that \(v=\psi \circ P\). Therefore, under first degree stochastic dominance CEU and AU coincide. Large parts of the paper are expository. They include a careful examination of some basic facts concerning the models of CEU and AU which may serve as a nice elementary introduction to the topic.