Abstract
Visual working memory (WM) is known to exhibit attractor dynamics, wherein mnemonic representations drift toward discrete, stable attractor states [1,2]. Maintenance in WM is also accompanied by specific patterns of synchronization and desynchronization in parieto-occipital alpha-band (8-12 Hz) oscillations [3]. Yet, the link between alpha desynchronization and attractor dynamics in WM remains unexplored. We tested n=24 human participants on a visual WM task involving delayed (~2500 ms) reporting of a retro-cued grating’s orientation. Although grating orientations were uniformly distributed across trials, participants’ orientation reports systematically favored the nearest diagonal orientations and were biased away from cardinal orientations, indicating stable and unstable fixed points (attractors) at these orientations, respectively. We investigated the behavioral and neural correlates of these attractor dynamics. We quantified the magnitude of bias in orientation reports (“attractor strength”) for each participant, and divided (median-split) participants into “weak” (n=12) and “strong” (n=12) attractor groups. Weak attractor participants were significantly more precise with reporting stimulus orientations, both for the WM cued (p=0.017) and uncued stimuli (p=0.030). Interestingly, weak attractor participants exhibited stronger alpha desynchronization as compared to strong attractor participants. In fact, the level of alpha desynchronization correlated robustly (and negatively) with attractor strength across individuals (p<0.05). Moreover, cue-induced alpha desynchronization -- the difference between alpha power in the parieto-occipital cortex contralateral versus ipsilateral to the WM cue -- also strongly predicted attractor strengths, across individuals (p<0.05). Contrary to recent reports [2], these results show that attractor dynamics do not produce an obligatory increase in working memory performance. Rather, alpha-band desynchronization may constitute a key neural mechanism governing the strength of attractor dynamics in visual WM. References: 1. Bae et-al, 2014. PMID:24715329 2. Panichello et-al, 2019. PMID:31358740 3. Wianda et-al, 2019. PMID:30887701