Abstract
Theories of attention propose that working memory representations are used to guide visual attention. Empirical tests of this prediction using behavioral measures have led to mixed results. Some conclude that working memory representations automatically guide attention, while others suggest that an individual's strategy determines whether working memory representations guide attention. Here, we used the N2pc component of observers' event-related potentials to directly measure the effect that visual working memory representations had on how covert attention was deployed. Participants maintained an object representation in working memory in service of a change-detection task while performing visual search. On some trials, the search array contained a distractor that exactly matched the object held in working memory and across trials we manipulated its location relative to the search target. If working memory representations automatically bias attention to select matching inputs, then the memory-matching distractor should elicit an N2pc regardless of the location of the target. However, the results did not support this hypothesis. Instead, we found the visual search target elicited a larger amplitude N2pc when the memory-matching item was in the opposite hemifield, suggesting that covert attention was actively directed away from the memory-matching item. The findings of these experiments are inconsistent with the proposal that working memory representations automatically guide attention to memory-matching items, and lend support to the proposal that the influence of working memory representations on attention during visual search is under strategic control.