Temporal associations supporting repetitions in free recall.

Abstract

The present studies use a novel approach to characterize how memory representations are updated with repetition. These studies use the free recall paradigm, which boasts greater memory advantages for spaced repetitions (Melton. Journal of Verbal Learning and Memory, 9, 596-606. 1970; Madigan. Journal of Verbal Learning and Verbal Behavior, 8, 828-835. 1969). However, a single recall of a twice-presented item precludes inferring whether the item's first or second presentation support its recall. The present studies leverage that, in free recall, transitions reflect stronger associations and are more likely between items studied nearby in time (Healey et al., Psychonomic Bulletin & Review, 26(3), 699-720. 2019). The critical analysis asks which transitions are more likely to a repeated item: temporal neighbors from its first presentation or from its second presentation? Transitions should be equally likely from neighbors of each presentation if the repeated item's presentations are stored independently. Transitions from second-presentation neighbors should be more likely if retrieval of item information from the first presentation strengthens the item representation during the second presentation, or if independent traces benefit from being studied more recently. Alternatively, retrieved context theory assumes that each studied item is associated with a slowly drifting temporal context, and repetition evokes study-phase retrieval of the context state from the first presentation (Howard & Kahana. Journal of Mathematical Psychology, 46, 269-299. 2002a; Siegel & Kahana. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40(3), 755-764. 2014). This context retrieval should strengthen the repeated item's associations to items with similar temporal contexts from its first presentation. As a result, retrieved context theory predicts more transitions to a repeated item from a first-presentation neighbor. Two studies provide support for the prediction of retrieved context theory, with implications for other theories.