A "pseudo-endowment" effect and its implications for some recent non-expected utility models