Victoria L. Jacoby^1,2 (vjacoby@bwh.harvard.edu), Jeremy M. Wolfe^1,2; ¹Brigham and Women's Hospital, ²Harvard Medical School

Visual search often relies on category-based target detection, particularly in foraging tasks that require locating multiple targets among distractors, such as picking berries or detecting tumors. If categories are unfamiliar or perceptually similar, observers must learn how to distinguish targets from distractors. While incidentally learned information about targets and distractors can aid search performance (e.g., Addleman & Störmer, 2023), prior work has primarily examined statistical learning of single features and locations or memory for individual items. Here, we asked whether observers could move beyond individual item learning to discover and extract perceptual patterns among items when targets and distractors are drawn from similar, perceptually based categories. In Experiment 1, participants were asked to collect rocks from one geological rock category while avoiding those from another. Participants were shown three examples from each category. Next, a pretest evaluated their ability to categorize rocks based on this minimal training. This was followed by visual foraging with feedback and post-foraging recognition and classification assessments. Assessments included previously seen and novel items from the two primary rock categories, as well as items from additional, unseen categories. Results revealed above-chance classification performance, with significant improvements after foraging. Exemplar recognition was strongest for target items, and classification accuracy was highest for previously seen targets, suggesting greater exemplar-level learning for targets. Categorization of novel exemplars was equivalent for targets and distractors, indicating similar category-level learning. In Experiment 2, we removed the pretest and all explicit cues indicating that distractors belonged to the same category prior to the assessments and found similar results: generalizable category learning was observed for both categories. These results suggest that visual foraging supports the learning of perceptual patterns in both targets and distractors, enabling generalization to novel items and influencing how observers search for these categories.

Acknowledgements: This work was supported by the Harvard Medical School Alber’s Fund