Rapid inversion of singleton distractor representations underlies learned attentional suppression

In visually complex and dynamically changing environments, humans often face the challenge of filtering out salient stimuli that are presently irrelevant to their tasks. Recent evidence suggests that through repeated exposure to search arrays containing salient color singleton distractors, individuals can learn to divert their attention away from such salient but irrelevant stimuli, even before they capture attention. However, the mechanisms underlying such attentional suppression remain unclear. The current study examined trajectories of singleton distractor representations during visual searches to address this gap. Using multivariate pattern analyses on EEG data (N = 40), we found that singleton distractor representations underwent a rapid inversion approximately 200 ms into the search. These inverted representations were coded in a shared subspace with target representations, but in a reversed orientation, presumably to downweight their salience in the spatial priority map. Target locations were consistently enhanced compared to non-singleton distractors, while singleton distractors were suppressed. Our findings reveal a novel mechanism of rapid representational transformation underlying salient distractor suppression at the onset of visual search. The rapid inversion of pop-out singleton distractor signals resulted in an inverted arrangement of target and distractor representations in a shared neural subspace, which facilitates subsequent read-out of both target enhancement and distractor suppression signals in the computation of spatial priorities to successfully guide search.