Michael Mugno¹, Emma Sutcliffe¹, Timothy Vickery¹; ¹University of Delaware

Information retrieved from long-term memory (LTM) enters working memory (WM), and the amount of information that can be retrieved is constrained to the limits of WM (Fukuda & Woodman, 2017). Liu, Li, Theeuwes, and Wang (2022) presented evidence that when WM is full, retrieved information bypasses WM. However, it is possible that retrieved LTM items can still enter WM when it is already holding many items online, albeit at the cost of the fidelities of some or all items. We investigated this by introducing continuous reporting to their paradigm. If retrieved items bypass WM, then there should be no impairment of item fidelity for either the WM or retrieved items. If retrieved items enter WM, then item fidelities should suffer compared to when retrieval is not cued. We found that when WM is full (4 items), the fidelity of LTM items suffered compared to when WM was not full (2 items). Also, the contents of WM suffered when retrieval from LTM was cued compared to baseline, under both 2-item and 4-item WM load. An additional experiment showed evidence that these effects were not due to the task design, where LTM retrieval was nested within the WM task. However, the order in which LTM or WM items were cued did affect performance – the items that were cued first and tested last suffered from reduced precision. We conclude that LTM retrieval does not bypass WM, as it robustly interferes with WM representations (and vice-versa).

Acknowledgements: Supported by NIH NIGMS grant 5P30GM145765-03