Date created: | Last Updated:

Category: Project

Description: The production effect is a well-established finding: If some words within a list are read aloud, that is, produced, they are better remembered than their silently read neighbors. The effect has been extensively studied with long-term memory (LTM) tasks. Recently, using immediate serial recall and short-term order reconstruction, Saint-Aubin et al. (2021) reported informative interactions between the production effect and serial positions. Here, we asked whether these interactions would also be observed with the LTM tasks used in the field. In Experiment 1, pure and mixed lists of 8 words were presented in both order reconstruction and free recall tasks, with a 30-second filled retention interval. In Experiment 2, the list length was extended to 24 words; in Experiment 3, 10-word lists were used with a 2-minute retention interval. Results from all experiments aligned well with those observed in short-term memory. With mixed lists, where produced and silently read words alternated, produced items were better recalled, leading to sawtooth serial position curves. With pure lists, compared to silently read items, produced items were better recalled when studied in the last serial positions, but were less well recalled for the primacy positions. Results were readily accounted for by the Revised Feature Model, originally developed to explain short-term memory performance. The findings and model suggest that produced items are encoded with more item-specific, modality-related features, that this generates a relative distinctiveness advantage in short- and long-term memory. However, the richer encoding comes at a cost: it appears to disrupt rehearsal.