Emotional stimuli engage corticolimbic circuits and capture attention even when they are task-irrelevant distractors. Whether top–down or contextual factors can modulate the filtering of emotional distractors is a matter of debate. Recent studies have indicated that behavioral interference by emotional distractors habituates rapidly when the same stimuli are repeated across trials. However, little is known as to whether we can attenuate the impact of novel (never repeated) emotional distractors when they occur frequently. In two experiments, we investigated the effects of distractor frequency on the processing of task-irrelevant novel pictures, as reflected in both behavioral interference and neural activity, while participants were engaged in an orientation discrimination task. Experiment 1 showed that, compared with a rare distractor condition (20%), frequent distractors (80%) reduced the interference of emotional stimuli. Moreover, Experiment 2 provided evidence that emotional interference was reduced by distractor frequency even when rare, and unexpected, emotional distractors appeared among frequent neutral distractors. On the other hand, in both experiments, the late positive potential amplitude was enhanced for emotional, compared with neutral, pictures, and this emotional modulation was not reduced when distractors were frequently presented. Altogether, these findings suggest that the high occurrence of task-irrelevant stimuli does not proactively prevent the processing of emotional distractors. Even when attention allocation to novel emotional stimuli is reduced, evaluative processes and the engagement of motivational systems are needed to support the monitoring of the environment for significant events.

Understanding natural scenes involves the contribution of bottom–up analysis and top–down modulatory processes. However, the interaction of these processes during the categorization of natural scenes is not well understood. In the current study, we approached this issue using ERPs and behavioral and computational data. We presented pictures of natural scenes and asked participants to categorize them in response to different questions (Is it an animal/vehicle? Is it indoors/outdoors? Are there one/two foreground elements?). ERPs for target scenes requiring a “yes” response began to differ from those of nontarget scenes, beginning at 250 msec from picture onset, and this ERP difference was unmodulated by the categorization questions. Earlier ERPs showed category-specific differences (e.g., between animals and vehicles), which were associated with the processing of scene statistics. From 180 msec after scene onset, these category-specific ERP differences were modulated by the categorization question that was asked. Categorization goals do not modulate only later stages associated with target/nontarget decision but also earlier perceptual stages, which are involved in the processing of scene statistics.