Abstract: | Kahneman and Tversky (1982) have proposed a simulation heuristic such that perceivers tend to substitute ‘normal’ antecedent events for exceptional ones in psychologically ‘undoing’ a given outcome. Recently Gavanski and Wells (1989) have demonstrated that exceptional outcomes tend to be perceived as caused by exceptional events and normal outcomes by normal events, a finding more in line with the representativeness heuristic than this ‘normalization’ principle. We argue that representativeness may be determined by the evaluative tone of events as well as by probability—namely that positive events are assumed to underlie positive outcomes and negative events, negative outcomes. Both normality and value were independently manipulated in order to test the relative effects of each of these factors. In contrast to Gavanski and Wells our data indicate that preference was given to the similarity of value between events and outcome for undoing both positive and negative and normal and exceptional outcomes. Some implications of these findings for counterfactual processing are discussed. |