Cumulative prospect theory and the St. Petersburg paradox