Abstract
There is some evidence that visual working memory can bias selective attention by providing a competitive advantage to sensory inputs that match the contents of visual working memory. However, it is unclear whether this process is automatic or can be flexibly controlled. The present study investigated this issue by using ERP recordings, focusing on N2pc (a measure of attentional deployment) and Pd (a measure of attentional suppression). Participants were instructed to memorize the orientation of a colored rectangle in a memory array, and memory was tested at the end of the trial by a test array. In Experiment 1, task-irrelevant circles (probe stimuli) were briefly presented between the memory and test arrays. We found that the task-irrelevant circles matching the color of the rectangle in memory did not elicit N2pc but instead elicited Pd. This suggests that the item matching memory was detected but was suppressed before attention could be captured by it. In Experiment 2, a search-target square was presented on 50% of the trials between the memory and test arrays, requiring a button-press response. In the other 50% of the trials, task-irrelevant circles were presented, as in Experiment 1. We found that the target square elicited an N2pc, and N2pc onset latency was earlier if the target matched the color being held in visual working memory. In contrast, the task-irrelevant circles again elicited Pd if they matched the color being held in visual working memory. These results suggest that sensory inputs matching the contents of visual working memory are detected and have an advantage in competing for attention, but these inputs can be either suppressed or facilitated depending on task demands. Thus, items that match memory automatically receive priority for attention, but the actual deployment of attention is flexibly controlled by top-down factors.