Abstract
Statistical learning can shape visual attention by enabling spatial suppression of locations that frequently contain distractors. Although learned suppression may be context-dependent, the underlying factors driving this dependency remain unclear. While previous research has largely focused on the nature of the context itself, less attention has been paid to how context–regularity pairs are presented, particularly whether they are blocked or intermixed across trials. Multiple uninterrupted pairings may be necessary for less salient contextual cues, such as background colors or images, to support context-dependent learning. The present preregistered, high- powered study (N = 136) therefore used a blocked presentation, with two blocks each pairing a unique background color with a high-frequency distractor location, followed by an intermixed testing phase with equiprobable distractor locations. Results show that although distractors reliably captured attention, the background context did not affect response times in the testing block, with Bayesian analyses providing decisive evidence for context-independent suppression. These results show that implicitly learned distractor suppression is highly robust, persistent, and generalizes across non-task- relevant contexts, even when context–regularity associations are repeated and uninterrupted during learning.