Proactive enhancement and suppression elicited by statistical regularities in visual search.

Abstract

The present study investigated how attentional selection is affected by simultaneous statistical learning of target and distractor regularities. Participants performed an additional singleton task in which the target singleton was presented more often in one location while the distractor singleton was presented more often in another location. On some trials, instead of the search task, participants performed a probe task, in which they had to detect the offset of a probe dot. This probe task made it possible to take a peek at the proactive selection priorities just at the moment the search display was presented. The results show that observers learn the regularities present in the search display such the location that is most likely to contain the target is enhanced while the location that is most likely to contain a distractor is suppressed. We show that these contingencies can be learned simultaneously resulting in optimal selection priorities. The probe task shows that both spatial enhancement and spatial suppression are present at the moment the actual search display is presented, indicating that the attentional priority settings are proactively modulated. We claim that through statistical learning the weights within the spatial priority map of selection are set in such a way that selection is optimally adapted to the implicitly learned regularities.

Statistical learning of across-trial regularities during serial search.

“…Clearly, the current findings cannot be explained in terms of some form of intertrial priming. Note that previous work investigating the distributional regularities of the target/distractor location by presenting the target or distractor much more often in one location than in all other locations do suffer from this potential shortcoming of repetition intertrial priming as the target or distractor has to be presented at the same location repeatedly (see also Goschy et al, 2014;Huang et al, 2021;Kabata & Matsumoto, 2012). The current findings can only be explained by assuming that observers have learned the embedded across-trial regularities without the need to assume any low-level repetition priming effects.…”

What to expect where and when: how statistical learning drives visual selection

“…Spatial priority map: a representation of a topographic space encoding the priority of individual locations combining signals from sensory input (bottom-up), current goal states (top-down or behavioral relevance), and statistical learning (history driven). Stimulus-driven (bottom-up) selection: when selection is automatic, is more likely to contain the target, which results in an attentional bias as evidenced by participants being faster to detect a target positioned in high-probability locations than in lowprobability locations [29][30][31].…”

“…In most experiments investigating statistical learning of the location of the distractor, at most, only a few participants are aware of the regularities present in the display [39][40][41]60,121,122]. In experiments in which regularities regarding the target are manipulated, awareness is much higher, with about two-thirds of the participants able to report the high-probability target location [31]. Note, however, that this all depends on the probabilities used, the number of trials, and the way awareness is assessed.…”

“…There is ample evidence that through statistical learning, locations that are likely to contain a distractor are suppressed proactively [31,67,127,128], indicating that the location within the priority map is suppressed before the display is presented. While this suppression is assumed to be implemented at a spatial priority map and thus generic, there is also evidence that under specific conditions spatial suppression can become tuned to specific distractor features and/or dimensions [37,60,129].…”

Pinging the brain to reveal the hidden attentional priority map using encephalography

“…In the current study we employed the additional singleton task (Theeuwes, 1992) with imbalanced target distributions, a paradigm that can be used to implicitly train participants to expect relevant information (i.e., targets) to appear in certain regions of space (Ferrante et al, 2018; Gao & Theeuwes, 2020; C. Huang et al, 2022).…”