Oscillatory correlates of uploading long-term memory into visual working memory

Our visual working memory (VWM) allows us to represent a limited amount of information in a readily accessible state. To study how information is represented in VWM, studies thus far have almost exclusively examined situations in which visual stimuli are presented and maintained over brief retention intervals. These studies have successfully identified neural correlates of VWM representations. For instance, event-related potential (ERP) studies showed that the sustained negativity observed over parieto-occipital channels increases as the function of a set size up to individuals’ VWM capacity, plateauing at larger set sizes (e.g. Vogel & Machizawa 2004). Recently, we also found that alpha band power measured at the same time over the same channels shows the same set size function (Fukuda & Vogel 2011). However, it is still unclear how information retrieved from visual long-term memory (VLTM) is represented in VWM. In this study, we first had participants learn spatial layouts of colored objects that varied in their set size (i.e., 1 2 4 or 8 objects). When learning was complete, participants were instructed to retrieve learned arrays while we recorded their scalp EEG activity to examine previously established neural correlates of VWM. Here, we found that while the ERPs did not differentiate the set size of the learned arrays participants retrieved, the oscillatory activity did dissociate the set size of the arrays by mirroring the pattern observed when VWM is used to store information that was just presented in the visual field. That is, when participants were accessing the learned arrays, the alpha band power showed a monotonic decline up to the set size 4 with no further decreases for large arrays. These findings not only dissociate the ERP and oscillatory correlates of VWM, but also provide insights as to how VWM supports a diversity of operations during visual cognition. Meeting abstract presented at VSS 2013