A critique of expected utility theory: Descriptive and normative considerations