Abstract
To accommodate for our limited visual working memory (VWM) capacity, an attentional filter ensures only goal-consistent information is encoded into VWM. Despite its critical role in dynamic visual behavior, attentional filtering has traditionally been studied in contexts that restrict eye movements. Critically, saccades have been demonstrated to disrupt perception and object-location binding. Does control over the attentional filter persist across saccades, or do saccades disrupt this filter, resulting in the momentary loss of control post-saccade? In E1 (N=15), we established attentional filtering using a no-saccade design. On each trial, participants fixated on a point in one of four locations before seeing a black shape cue indicating that trial’s target shape (either a square or circle). Following a variable delay, they briefly viewed an array of four colored shapes—either two circles and two squares (2 targets/2 non-targets), or all four of the target shape—and were told to remember the colors of all the target-matching shapes. One of the target-matching shapes was then probed for report, and participants reported its color by clicking on a continuous color wheel. As predicted, we observed effective filtering: performance was better (i.e., less color report error) when only two items were relevant (SS2) relative to when all four were relevant (SS4). In E2 (N=20), we tested whether this filtering was disrupted by a saccade. On some trials, following the initial target shape cue, participants were prompted to make a saccade, and the stimulus array then appeared after either a short (50ms) or long (400ms) post-saccade delay. We observed significantly disrupted attentional filtering immediately following a saccade (short post-saccadic delay), whereas the filter was successfully reinstated following the long post-saccadic delay. We suggest that saccades disrupt attentional filtering and the filter must be reinstated following a saccade to efficiently filter non-targets from VWM encoding.