Abstract
Multiple models suggest that working memory (WM) provides the online workspace for apprehending items during visual search. In line with this hypothesis, we found that WM capacity strongly predicts search efficiency. Previous work, however, has found that search remains efficient when observers are given a concurrent memory load during search (Woodman et al. 2001). Although this finding suggests that working memory is not needed for apprehending search items, our hypothesis is that subjects may have dropped the stored items from WM in anticipation of the search phase of the dual task. In this case, observers could have relied on retrieval from recently activated long term memory to respond to the memory probe at the end of the trial. To test this hypothesis, we measured storage-related neural activity (contralateral delay activity; CDA) while subjects performed a WM/search dual task in which the memoranda occupied one visual field and the intervening search array was presented in the other. If search and WM are independent processes then the storage-related CDA should be equivalent in the single and dual task conditions. Conversely, if subjects drop the memory items in anticipation of the search array, then CDA amplitude contralateral to the memoranda should drop in the dual task condition. The behavioral results replicated the strong correlation between WM capacity and search efficiency. Second, CDA amplitude in the dual task condition dropped during the interval between memory encoding and search onset. Finally, the size of this drop in storage-related neural activity predicted behavioral costs in both memory and search during dual task relative to single task conditions. These data suggest that visual search and WM storage rely on a common resource, but that observers minimized dual task costs by offloading items from WM when search was required. Thus, visual working memory is integral to visual search.
Meeting abstract presented at VSS 2012