Yifan Fang¹ (), Kevin Ortego¹, Viola Störmer¹; ¹Dartmouth College

People learn from statistical regularities in their environments in support of more efficient behavior. For example, during visual search, people can rapidly learn which color or object frequently associates with a task-relevant target, improving search performance. Previous studies using simple features like color suggested that early and later stages of neural processing are involved in this learned prioritization (Ortego et al., 2024; Wang et al., 2023). Here, we investigate the neural underpinnings of learned attention during real-world object search. Participants (n = 24) performed a visual search task to indicate the direction of a tilted object (target) among three other upright objects (distractors) while EEG was recorded. On each trial the same four objects were present, but unbeknownst to the participants, one object appeared more frequently as the tilted target object during an initial “training phase”, which was followed by a “testing phase” where target frequencies were equal across all four objects. We found that participants rapidly learned these statistical regularities, reflected in faster response times during search for the frequent relative to infrequent objects during the training phase. These behavioral benefits were accompanied by changes in the N2pc component, an index of early attentional selection, and the later LPC component, demonstrating that selecting one specific object more frequently modulates neural processing and behavior. However, these learning effects did not extend reliably into the testing phase, suggesting that attentional tuning to specific objects may quickly adjust to the current regularities in the environment. Together, our results suggest that similar neural processes may give rise to how selection history influences search behavior for both simple features and complex, real-world objects.