Abstract
The evidence is mixed as to whether attentional biases caused by a visual working memory (WM) representation depend on whether or not that representation is task relevant. Moreover, recent work shows long-term memory templates quickly begin guiding attention when searching for the same target over trials. In many laboratory paradigms, the target representations that guide attention may therefore not be maintained in WM at all. Our goal was to develop a paradigm for measuring the effect of task-relevant and task-irrelevant visual WM representations on attention. Participants searched 6-item arrays for targets while maintaining in WM the search color and another color for a subsequent memory task. This ensured task-relevance was not confounded with representation in WM versus long-term memory. In addition, we removed bottom-up differences between task-relevant and task-irrelevant search items by randomizing the target and memory colors and focusing on the first saccade to ensure all items were an equivalent distance from the current fixation. We tracked observers’ eye movements to the task-relevant search target, the task-irrelevant memory match, and distractors during search. First saccades were strongly biased to the task-relevant WM representation (i.e., the target) over the task-irrelevant memory match (respectively 80.9% vs. 5.9%, p <.0001). A control condition ensured that this was not due to differing strengths of the two WM representations. We also found a weak, but significant prioritization of the task-irrelevant WM matches over the other distractors whether or not the search target was present (ps <.05). This demonstrates that the strong bias to task-relevant WM matches is not due to a slight competitive advantage over task-irrelevant WM matches, but that the top-down attentional control from task-relevant working memory templates is qualitatively different than an automatic influence of WM representations on attention.
Meeting abstract presented at VSS 2012