For the good and the bad, the world around us is full of distraction. In particular, onset stimuli that appear abruptly in the scene grab attention, thus disrupting the ongoing task. Different cognitive mechanisms for distractor filtering have been proposed, but prevalent accounts share the idea that filtering is accomplished to shield target processing from interference. Here we provide novel evidence that challenges this view, as passive exposure to a repeating visual onset is sufficient to trigger learning-dependent mechanisms to filter the unwanted stimulation. In other words, our study shows that during passive exposure the cognitive system is capable of learning about the characteristics of the salient yet irrelevant stimulation, and to reduce the responsiveness of the attention system to it, thus significantly decreasing the impact of the distractor upon start of an active task. However, despite passive viewing efficiently attenuates the spatial capture of attention, a short-lived performance cost is found when the distractor is initially encountered within the context of the active task. This cost, which dissipates in a few trials, likely reflects the need to familiarize with the distractor, already seen during passive viewing, in the new context of the active task. Although top-down inhibitory signals can be applied to distractors for the successful completion of goal-directed behavior, our results emphasize the role of more automatic habituation mechanisms for distraction exclusion based on a neural model of the history of the irrelevant stimulation. (PsycINFO Database Record (c) 2018 APA, all rights reserved).