Abstract
Classical memory theories propose visual working memory (WM) processing as a key pathway for long-term memory (LTM) generation. Yet exactly how WM processes, such as attentional prioritization of specific mnemonic contents, affect LTM remains elusive. To address this, we employed a two-stage visual WM-LTM task, while tracking observers' eye movements. In the initial WM stage, participants completed a WM-prioritization task with a retrocue indicating the relevant visual object for an upcoming search task. Following a break, participants performed a surprise LTM-recognition task, identifying the object seen in the earlier WM stage. Results at the WM stage revealed more accurate and faster searching for informatively cued objects versus neutrally cued ones, replicating cue-induced attentional prioritization in WM. Crucially, at the LTM stage, we found significantly higher recognition accuracy for objects that were cued during WM (but absent in the search array) compared to both uncued objects and trials with a neutral cue (which showed similar recognition accuracy). These findings show that attentional prioritization during WM improves subsequent LTM performance for the prioritized item without incurring an LTM cost for unprioritized items (while having equal sensory exposure). In addition, through observers' eye-movements we could track attentional allocation after the retrocue. This revealed how attentional allocation during WM affected LTM, with trials with faster attentional allocating in WM having a larger subsequent benefit in LTM. Overall, these results show that attentional prioritization during visual WM affects LTM, and how this is mediated by the effectivity of attentional allocation – offering a mechanistic interpretation that bridges the interaction between WM and LTM.