Nursena Ataseven¹ (), Sahcan Özdemir², Daniel Schneider², Wouter Kruijne¹, Elkan G. Akyürek¹; ¹University of Groningen, ²Leibniz Research Centre for Working Environment and Human Factors

Previous studies have suggested that working memory (WM) can leverage regularities in the environment to increase its capacity. This highlights how memory resources can be used more efficiently when associations in long-term memory support WM. However, the mechanisms underlying this interaction remain unclear. Here, we ask whether such associations might alter the representation of items maintained in WM. Are long-term memory (LTM) supported representations "compressed," thereby freeing up capacity for novel information? To address these questions, we recorded EEG data (N=20) as participants performed a delayed comparison task, where they compared a target orientation grating to a probe and judged whether it was rotated clockwise or counterclockwise. Before the target presentation, a central color cue (red, green, or blue) was shown. In half of the trials, this cue informed participants about the 90° range from which the upcoming target orientation was drawn, while in the other half the cue was uninformative. Participants learned the associations between colors and orientation ranges prior to the experiment. We analyzed the neural representations in working memory by decoding orientation from posterior channels during the target orientation display. This allowed us to examine how different cue conditions influenced the encoding of orientation representations and the potential role of LTM in optimizing memory processes. The results suggest that behaviorally, the informative cue improved memory performance. However, this difference was hardly reflected in the EEG decoding. During target encoding, no differences were observed. Throughout the maintenance period, we observed a very subtle difference that did not reach significance, but became larger when probe presentation was imminent. These findings suggest that the associated information hardly affected the representation in working memory, but might factor into behavior at a later stage, for example during the comparison with the probe.