Expectancies and the processing of social information: Implications for person *perception and person memory

Expectancies tend to produce confirmatory biases in social information processing, leading some researchers to hypothesise that expectancies act as schematic filters that facilitate the encoding of schema-consistent (versus inconsistent) information. Both theoretical and methodological arguments, however, provide challenges to the schematic filter perspective. The goal of the current research was to evaluate these claims. In distinguishing between two type of information that were either redundant or nonredundant with respect to activated schemata, I sought to disentangle remembering and reconstruction as indicators of social information processing, thereby providing insight into the process of encoding social information.;In three experiments, participants were (or were not) provided with information to set up an expectancy, and then formed an impression of an individual from a series of statements describing behaviours performed by that individual. After a brief delay, participants' memory for expectancy-consistent and expectancy-inconsistent behaviours was assessed. In Experiment 1, participants were asked to form an impression of a target individual with no cognitive constraints and no motivational distractions. Experiment 2 extended Experiment 1 by adding a processing goal manipulation to compare processing under optimal conditions with processing that occurs when social perceivers are either unmotivated or unable to concentrate primarily on forming an impression of a target person. Experiment 3 was designed to investigate how the extremity of the behaviours performed by the target individual influence encoding under different processing conditions.;The results provided converging evidence for the role of informational utility in guiding the processing of social information. All three experiments demonstrated that the activation of expectancies facilitated encoding that was efficient and driven toward maximising information value. In general, memory favoured non-schema-redundant information over schema-redundant information and schema-irrelevant information. The importance of the stored schema in driving these effects was highlighted by evidence that memory advantages for the ostensibly more useful information over the schema-redundant information were reliably greater when a strong rather than weak expectancy was activated within the experimental context, or when the task demands rendered the schema more (versus less) useful. These data suggest that social information processing is adaptively biased toward useful, non-schema-redundant information.