Aditya Prakash¹, Andrew Hollingworth¹; ¹University of Iowa

Attention can be guided by the content of visual working memory (VWM), often referred to as a search template. These templates can be positive (specifying feature values associated with the target) or negative (specifying feature values associated with distractors). In addition, the implementation of a search template can be either direct or indirect. For example, direct implementation of a cue to ignore red would be implemented by filtering stimuli on the dimension of color to reduce the relative priority of red items. In contrast, guidance could be implemented indirectly by marking, and subsequently de-prioritizing, the spatial locations of red items. Beck & Hollingworth (2015) proposed that negative templates maintained in VWM control attention indirectly through a mechanism of spatial recoding. This account predicts that the implementation of a negative template will depend on the ease of spatial recoding, with efficient avoidance when cued items are spatially segregated and less efficient avoidance when cued items are spatially intermixed. Here, we examined this prediction for oculomotor behavior in a search task employing negative color cues. Participants searched through arrays of objects drawn in multiple colors following a negative or neutral color cue. Consistent with previous work, we found that the probability of fixating negative-cue-matching objects was significantly below baseline. Critically, the efficiency of avoidance was influenced by a manipulation of spatial intermixing, with a lower probability of fixating negatively cued items when the level of spatial intermixing was lower. As argued by Beck & Hollingworth, direct implementation of a negative template may not be possible, as maintenance of the color in VWM leads to the obligatory enhancement of matching items. Instead, guidance by VWM-based negative cues appears to be implemented indirectly and reactively through the translation of the color cue into a spatial template based on the observed locations of cue-matching items.