Effect of the Desirability of Outcomes on Decision Making

Abstract

This experiment was designed as a direct test of the descriptiveness of an axiom first stated by Rubin. The axiom implies that adding a constant to any column of a payoff matrix will not alter S's choice of action. An auxiliary aim was to determine whether the effect of desirability of outcomes of events on frequency of prediction of the events would be observed when S was motivated by a small monetary incentive to make accurate predictions. S drew one card from each of a number of 10-card decks. On half the trials he won 50¢ if a marked card was drawn; on the other half he lost 50¢. In one condition S bet 10¢ on the type of card he would draw. In the other condition S predicted what he would draw after inspecting a payoff matrix in which the value of drawing a marked card and the value of the 10¢ bet were combined. The same pattern of results appeared in each condition: more predictions were made for drawing a marked card when S stood to win 50¢ than when he stood to lose 50¢ with objective probability constant. Thus, unless it can be shown that the utility function for these Ss was concave upward in an arithmetical plot and the Ss considered only the total outcome for each contingency, Rubin's axiom does not appear to be descriptive of decision making under these conditions.