Abstract
Competing theories of visual search make diverging claims about whether we can suppress salient but irrelevant information and, if so, where in cortex attentional priority maps reside. Here, we used fMRI to assess whether the representation of a salient distractor can be suppressed below baseline in early visual cortex. Subjects (n = 12) performed an additional singleton search task in which they searched for a target (e.g., green diamond) amongst non-singleton distractors (e.g., green circles); an irrelevant color singleton distractor (e.g. red circle) appeared on 75% of trials. We manipulated trial history to induce attentional capture (“color variable”, colors swapped unpredictably) or to eliminate capture (“color constant”, colors were fixed). Using a spatial encoding model and fMRI, we estimated the representation of each item’s location in the 4-item search array. When participants were behaviorally captured (“color variable” condition), we observed that model-based estimates of the target and distractor locations were stronger than the non-singleton distractor location in early visual cortex (V1-V3), p < .01. and in IPS0-3, p < .005. When participants were not behaviorally captured (“color constant” condition), we observed that model estimates of nearby singleton distractor locations were suppressed below the level of non-singleton distractors in early visual cortex (V1-V3, p < .01) and some other visual areas (e.g., hV4-VO2, p < .05; IPS0-3, p < .01), consistent with the signal suppression hypothesis. Furthermore, model-based estimates of distractor suppression related to behavior. On fast trials (i.e., successful suppression), there was robust suppression of the irrelevant singleton distractor below the non-singleton distractor baseline (p < .005). This suppression was not observed for slow trials in this condition (p = .28). Our findings support accounts in which successful suppression leads to the down-weighting of distractors in spatial priority maps throughout visual cortex, bolstering evidence from EEG and eye-tracking studies.