Abstract
The active maintenance of visual items in working memory (WM) biases attention toward memory-matching objects. In dual-task paradigms that combine WM with visual search, the reappearance of the memory cue at a search distractor location results in slowed response times and misdirected first saccades (Soto et al., 2005), suggesting that spatial attention is captured by the memory-matching distractor. While WM-driven attention enhances early perceptual processing of subsequent targets at the memory-matching location (Pan et al., 2016), it remains unknown whether WM-captured attention enhances visual feature perception of distractor items, and if so, whether this might interfere with perception of the search target's features. For example, if spatial attention shifts to the memory-matching distractor before moving to the target, or is split between both locations, we might predict increased feature-binding errors (i.e., swap errors misreporting the distractor features or mixing errors blending the features; Golomb et al., 2014). Alternatively, attentional capture by the memory-matching distractor may be too transient or weak to disrupt perception of the target's features. Participants remembered a cue color across an intervening visual search, wherein they reported the orientation of a target grating, using a continuous-report procedure. Critically, search gratings were also colored. On half of the trials, the color of a distractor grating could match the memory cue. Probabilistic mixture models fit to orientation response distributions revealed decreased performance (higher proportion of uniform guessing) when a memory-matching distractor was present, consistent with WM-captured attention. However, we found no evidence of increased swap or mixing errors. Further experiments tested different combinations of visual features for memory and search stimuli and again found no increased swap or mixing errors. These data suggest that attentional capture by memory-matching distractors may slow or impair search performance, but does not interfere with perception of the target item's features.
Meeting abstract presented at VSS 2017