Integrating Retrieved Context Theory and Item-Order Theory: A Hybrid Model of the Distinctiveness Effect

Abstract

Remembering one item in free recall often triggers recall of other items encoded nearby in time to the initial item. Retrieved Context Models can account for this temporal contiguity effect (TCE) but cannot account for how various experimental manipulations modulate the size of the TCE. We attempt to model the finding that orthographic distinctiveness dramatically reduces the TCE. Across six experiments, we found that the TCE is sometimes attenuated (though never eliminated) by distinctiveness. We developed a model of the distinctiveness effect by integrating the item-order framework with Retrieved Context Models. This model assumes orthographic distinctiveness increases the amount of processing devoted to items themselves, thereby reducing the functionality of the model mechanisms that generate temporal contiguity: context drift rate during study and the formation of new associations between items and experimental context. This model was fit to recall and TCE data from pure-list conditions and accounted for the data quite well. We then used the parameter values resulting from pure-list fits to simulate mixed-list data. Despite the substantial differences between pure- and mixed-lists, the model fit the mixed-list data quite well. The results support an item-order account of the distinctiveness effect.