Experiment 1 demonstrated distorted memory for exemplars of a social structure. Slides, each depicting one of six athletes finishing in an athletic event, were arranged into episodes consisting of from two ("TWOS") to five ("FIVES") slides, the first slide in an episode representing first place, the second slide, second place, and so on. During acquisition, TWOS, THREES, FOURS, and FIVES consistent with a linear ordering of skill were presented. During recognition, OLDS (episodes previously seen), NEWS (novel episodes consistent with the ordering), and NONCASES (episodes violating the ordering) were shown as TWOS, THREES, FOURS, and FIVES. OLDS and NEWS were equally confidently recognized; NON-CASES were rejected. Recognition confidence for OLDS and NEWS was a positive, linear function of episode length. Two forms of memory distortion occurred: NEW FIVES were falsely recognized, and they were recognized more confidently than OLD TWOS. Subjects apparently constructed an all-encompassing linearordering schema that determined recognition performance, since recognition confidence was invariantly a positive function of similarity to the schema. Experiment 2 disconfirmed two alternative explanations and confirmed that schema construction occurred.Understanding memory is essential for an adequate understanding of social cognition. Understanding the memory biases and distortions to which humans are subject is of particular importance, since these biases and distortions are important factors in accounting for the frequently erroneous character of social cognition.

No abstract

No abstract