Distinctiveness and interference in free recall: A test with the production effect

It is well established that when some words within a list are read aloud, or produced, and others are read silently, produced items are better recalled. According to the revised feature model (RFM), this benefit stems from additional features and enhanced distinctiveness. Because the model assigns forgetting to similarity-based retroactive interference, produced items should be better recalled when followed by a silent item than by another produced item. However, this mechanism had never been directly tested. In addition, Forrin et al. (2019) suggested a memory cost for silent items in mixed lists. Based on their production anticipation hypothesis, this cost derived from social anxiety and performance anticipation emerging within the experimental setting. Here, we tested these two competing accounts in four experiments in which 240 participants completed an immediate or delayed free recall task with 10-word lists. Produced and silent items were presented in two blocks of various lengths, and the occurrence of produced items within the list was predictable. We also manipulated the presence of the experimenter in the room during the task, so that the experimenter was present for Experiments 1A and 2A but absent for Experiments 1B and 2B. Overall, results offer full support for the RFM but very limited support for the production anticipation hypothesis. The main trends in the data were also modeled by the RFM. In sum, with the production effect, we provide the first comprehensive test of the retroactive interference mechanism described within the RFM and further inform our understanding of memory. (PsycInfo Database Record (c) 2026 APA, all rights reserved).