The coordinate system of suppression in dynamic contexts

It has been shown that observers can learn to suppress a spatial location that frequently contains a salient singleton distractor, which is referred to as learned spatial suppression. Our previous work showed that the default coordinate system of learned spatial suppression is retinotopic: Spatial suppression is naturally learned and transferred across gaze position in retinotopic (eye-centered) coordinates (Chang & Golomb, 2023). However, in the real world, spatiotopic (world-centered) coordinates are generally more beneficial to guide behavior because retinotopic representations are often unstable due to frequent eye movements. Can spatial suppression be learned in spatiotopic coordinates when the context is more conducive to spatiotopic learning? In the current study, participants performed a visual search task while the fixation point randomly changed between two locations (gaze positions 1 and 2) across trials. Participants searched four items for a shape oddball target (e.g., a diamond among circles) while ignoring a salient color singleton distractor that appeared on two-thirds of trials. Critically, the salient distractor appeared more frequently at a spatiotopic high-probability location, regardless of gaze position, creating a spatiotopically-weighted context. In one experiment, the target appeared equally often at each location, providing only distractor-related probability information (distractor regularities only; pure distractor suppression). In another experiment, in addition to the distractor probability manipulation, the target never appeared in the spatiotopic high-probability location, providing both target- and distractor-related probability information for suppression (both target and distractor regularities). Strikingly, the results showed that spatial suppression could be learned in spatiotopic coordinates when driven solely by distractor-related probability information (pure distractor suppression) but was learned in retinotopic coordinates when driven by both target- and distractor-related probability information. These results raise an intriguing possibility that pure distractor suppression is susceptible to spatiotopic-based learning, but target-related processing may be more hardwired for retinotopic representations.