[rmbeard]

One could always use Choquet expected utility to get around any behavioral objections to the von Neumann or Savage axiomatizations of expected utility theory. Unfortunately this ergodic paper by Peters is simply not a constructive contribution to the debate but rather a red herring.