Rapid disengagement through statistical learning of distractor-target spatial associations

Previous studies have demonstrated that attentional selection is biased toward locations likely to contain a target and away from those likely to contain a distractor. These biases are thought to arise from statistical learning, which adjusts weights within a spatial priority map to optimize attentional selection. While prior work has shown statistical learning based on across-trial probabilities, it remains unclear whether individuals can learn associations between distractor and target locations within a given trial. To investigate this, two experiments employed the additional singleton paradigm, in which, sometimes, the location of the distractor predicted the location of the target. Results from Experiment 1 revealed that participants responded faster when the distractor appeared at a location predictive of the target, suggesting successful learning of the within-trial association. Experiment 2, which incorporated eye-tracking, showed that this benefit was not because of proactive suppression, but instead driven by faster disengagement from the distractor, as reflected in reduced dwell times. These findings demonstrate a novel form of statistical learning based on within-trial spatial associations. They also indicate that the spatial priority map can be dynamically updated following attentional selection, a process potentially reinforced by the execution of a motor response. (PsycInfo Database Record (c) 2026 APA, all rights reserved).