Kevin Ortego¹, Douglas Addleman², Viola Stoermer¹; ¹Dartmouth College, ²Gonzaga University

Statistical regularities in the environment influence how our brains process and prioritize information, even when we are not explicitly aware of these regularities. For example, a target item occurring more frequently in a particular location or color will lead to faster response times in visual search tasks. Similar behavioral benefits are observed from learning that certain features are commonly irrelevant (e.g. that items in a particular color are frequently distractors). It is currently debated whether these learning benefits are realized via enhancements at early processing stages or through later modulations of response and decision thresholds. We tested this by recording EEG in two experiments in which participants (N=32, N=25, ongoing) viewed a circular visual search array and identified the gap direction of a Landolt C with a left or right gap among seven distractors with top/bottom gaps. Critically, and unbeknownst to participants, we manipulated the color probabilities such that the target occurred more frequently in one particular color (target feature learning), or such that distractors occurred more frequently in a particular color (with equiprobable target colors; distractor learning). This resulted in “valid” trials where the targets/distractors matched the most likely color and “invalid” trials where this association was reversed. As predicted, responses were faster for valid relative to invalid trials in both experiments, and correspondingly the N2pc, an ERP component associated with early attentional selection, occurred approximately ~30ms earlier on valid trials across both experiments, demonstrating that both relevant and irrelevant feature learning can modulate attentional selection within ~200ms of stimulus onset. Later response and memory processes, indexed by the LPC component, were also modulated by the probability structure, but the pattern differed for relevant and irrelevant feature learning. Together our results suggest similar attentional processing benefits, but differential memory-related effects across these two modes of learning.