Default Mode Network Connectivity Is Associated with Capture by Distractors Assigned to Learned Spatial Regularities

Abstract

Salient distractors interfere with visual search, but observers can learn to filter out distractors that occur frequently and in predictable locations. We examined whether within-default-mode-network (DMN) task-run connectivity was associated with distractor capture and its location-dependent modulation in task-fMRI data from two groups in which either orientation or color distractors carried a spatial-probability bias. Higher within-DMN connectivity was associated with greater distractor capture by the biased distractor dimension, which was also the most frequent distractor type, whereas capture by the unbiased dimension was essentially unrelated to connectivity. A hierarchical trial-level model also showed stronger DMN-distractor-capture coupling for the biased than for the unbiased dimension when both were evaluated under the same frequent-/rare-region weighting. DMN connectivity was also associated with spatial-probability learning, indexed by the rare-minus-frequent reaction-time difference on biased-dimension trials, but this association was closely tied to overall capture. When the physical distractor dimension was held constant, orientation showed stronger DMN-distractor-capture coupling when it carried the spatial bias; in contrast, the corresponding color comparison remained uncertain. The broader network analyses provided converging support for this DMN-distractor-capture coupling pattern. Higher DMN connectivity was thus not associated with poorer performance in general, but with less effective filtering of the distractor that the task set was tuned to suppress. This pattern is consistent with a DMN state that interferes with implementing the optimal task set.