Abstract
Given the severe capacity limit of our visual working memory (VWM), it is crucial that we frequently update VWM representations depending upon the current task demands. Yet, it remains unclear how we "drop" no-longer necessary items from being held in VWM. Here, we measured the contralateral delay activity (CDA), a neural index of the number of items in VWM and asked participants to purge recently encoded items from VWM. In Experiment 1, we briefly (100ms) presented 4 colored squares and on some trials cued the participants during the retention period to either discard these items from memory. When the cue instructed participants to discard all four squares from VWM (i.e. forget-all condition), the CDA diminished completely following the cue. However, when a physically equivalent cue instructed participants to drop three items leaving one square remaining, the drop in the CDA was negligible. These results suggest that participants can delete all active VWM representations far more effectively than they can selectively drop a subset of items. In Experiment 2, we examined these same mechanisms under circumstances in which the stimuli remain continuously visible. Previously, we provided evidence that 1) we are subject to the same capacity limit even when the stimuli to represent are continuously visible, and 2) the capacity limit is manifested by the same neural correlate, the CDA (Tsubomi, Fukuda, and Vogel, VSS 2011). Using the same cue procedure as in Experiment 1, we replicated the finding that the CDA was almost entirely diminished in the forget-all condition. By contrast, however, we found a sizable CDA reduction even in the selective forgetting condition. These results suggest that we can selectively delete active VWM representations so long as the items are still visually available for resampling into memory.
Meeting abstract presented at VSS 2013