A neural marker of the representation used to guide visual search

Abstract

Guided search requires a top-down target representation to be maintained in visual working memory (VWM) so that it can be matched to information in peripheral vision. We used contralateral delay activity (CDA; an ERP indictor of VWM load), measured after target designation but before search display onset, to estimate the VWM load generated by the target representation. To investigate the relationship between target representation and later search guidance, we asked whether CDA was more pronounced on trials when the initial saccade during search was directed to the target (strong guidance) compared to when it was not (weak guidance). An arrow cue (200ms) directed attention to either the left or right side of the screen, followed by a 400ms delay, and then the presentation of two photorealistic objects (400ms) on each side of central fixation. This target preview was followed by a 1000ms delay during which CDA was measured, and finally by a four-object search display. One of the cued objects was always present, and observers indicated their localization decision by pressing a button while fixating the target. We found that the expression of CDA across parietal and occipital electrode sites interacted with search guidance (p [≤].05). For central-parietal and central-parietal-occipital sites we found our predicted relationship; CDA was larger (more negative) on strong guidance trials compared to weak guidance trials. This would be expected if search guidance improved with the number of target features coded in VWM. However, lateral-parietal sites showed the opposite relationship; significantly larger CDA on poor guidance trials. As well, these sites were negatively correlated with search guidance across observers. We conclude that CDA is a reliable neural marker of the VWM representation used to guide search to visually complex realistic targets. The different expressions of this relationship may indicate that CDA could be useful in revealing the feature weighting processes used to optimize guiding target representations. Meeting abstract presented at VSS 2012