Person Memory: The Role of Processing Strategy, Expectancy, and Level of Incongruity in The Processing of Interindividual and Intraindividual Behavioral Variability

Abstract

Recent attempts to integrate levels of processing framework into a network model of person memory have led to several hypotheses that were examined in the reported research. In particular, the levels of processing approach suggests that information that is incongruent with a prior expectancy will be more difficult to integrate into an existing structure and, therefore, will tend to be better recalled than information that is congruent or irrelevant to a prior expectancy. Subjects were presented with a list of behavioral information about a target individual or group. Individual items varied in the level of incongruity they had to a prior expectancy and strong support was found for this hypothesis in several different experiments. Experiment 1 tested the further hypothesis that any variable that induces one to integrate discrepant pieces of information will lead to greater levels of recall. It was found that "impression formation" instructions led to higher levels of recall than "memory set" instructions and subjects recalled much more information about a single individual or a group of individuals that shared an important psychological bond than a group that did not share any obvious psychological bond. However, since none of the variables examined had a significant effect on recognition, it was concluded that these effects are isolated in the retrieval stage of information processing. Experiment 2 demonstrated that, holding the number of behaviors congruent with a prior expectancy constant, adding a number of incongruent behaviors to the list increases the likelihood that the congruent behaviors will be recalled. However, it was also demonstrated in Experiment 3 that adding congruent behaviors to the presentation list has no effect on recall of incongruent behaviors. It was concluded that behavioral items incongruent with a prior expectancy are more difficult to encode and therefore form a greater number of associations with other items than do congruent items. Analysis of the sequential order in which items were recalled supported this conclusion. Specifically, the probability that an incongruent item was recalled given that a congruent item was just previously recalled was much higher than the probability that a congruent item would be recalled given that an incongruent item had just previously been recalled. Implications of these results for a number of theoretical models are discussed.