Rapid inversion of singleton distractor representations underlies learned attentional suppression.

In visually complex and dynamically changing environments, humans must often filter out salient but task-irrelevant stimuli. Prior work shows that with repeated exposure to search arrays containing color singleton distractors, individuals can learn to divert attention away from these salient items. However, the neural mechanisms supporting such attentional suppression remain unclear. The present study examined the temporal trajectories of singleton distractor representations during visual search to address this gap. Using multivariate pattern analyses of EEG data (N = 40), we identified two clusters of decodable singleton distractor representations: an early cluster from 100-200 ms and a later cluster from 200-400 ms. Temporal generalization analyses showed that the later representations were transformed versions of the early ones. Importantly, stronger late, but not early representations, predicted faster search responses, suggesting that the later signals support distractor suppression. We hypothesized that this transformation facilitates suppressing singleton distractors in the spatial priority map. Comparing decoding evidence across locations revealed that target locations were consistently enhanced relative to non-singleton distractors, whereas singleton distractor locations were suppressed. Moreover, comparing the neural coding of locations revealed that the spatial organization in the singleton distractor neural space was inverted relative to that in the target neural space. Together, these findings reveal a rapid representational transformation underlying salient distractor suppression at the onset of visual search. This rapid inversion of singleton distractor signals was likely driven by top-down control mechanisms that transform bottom-up saliency signals, producing an inverted arrangement of target and distractor information within a shared neural space.