Detection of Contextual Organization in Serial Learning

Abstract

Prior research based on anticipation data showed that performance differences among items in a sequence were strong and stable, with clear consensus among subjects. These performance effects also showed considerable predictability when an item and its context of neighboring items were considered, a finding that suggested the presence or use of organization. This form of prediction was attempted using data derived from orderlearning tasks that emphasize item recognition rather than response production. Although predictions remained strong, item effects became unstable and the effects of item context on performance appeared to become degraded systematically, in line with differences in task requirements.