Abstract
Recent research has demonstrated that visual statistical learning guides attentional priority by proactively suppressing locations where distractors are more frequent. While there is evidence to suggest that both attention and working memory (WM) consume a common pool of cognitive resources, there is limited evidence on the nature of the interaction between the learned distractor suppression and WM. The present study investigated whether the learned spatial suppression of an external distractor has a carry-over effect that impairs the WM representation of objects encoded at that location. Across two experiments, participants first completed a visual search task designed to induce statistical learning. A salient singleton distractor appeared more frequently in a specific quadrant, while appearing less frequently in other locations. This was followed by a delayed discrimination task, in which participants had to encode and maintain target shapes presented at either the high-probability or low-probability distractor location. Experiment 1 imposed a low WM load (one target), while Experiment 2 imposed a high WM load (two targets). The results showed that the carry-over effect of suppression from a visual search task to a WM task was contingent on WM load. When the load was high, participants exhibited significantly reduced sensitivity (d') and a higher false alarm rate for targets presented at the high-probability distractor location. These findings suggest that statistical learning operates by down-regulating sensory gain at suppressed locations. This pattern supports a non-modular, shared resource account of attention and WM, suggesting that history-driven attentional biases can modulate the WM representations.