Luke Rosedahl^1,2, Isabella Cabaleiro¹, Yuka Sasaki², Takeo Watanabe²; ¹Florida Atlantic University, ²Brown University

Category learning is fundamental to how we interact with the world and supports performance in many important professional domains. Unlike visual perceptual learning (VPL), category learning was historically thought to generalize across locations, reflecting the belief that it uses location-invariant visual information. However, recent work found that learning for information-integration (II) categories (categories that combine information across different feature dimensions) is location-specific (Rosedahl, Eckstein, and Ashby, Nature Human Behaviour 2018; Rosedahl and Watanabe, JCE 2024). However, the mechanism of specificity remains unclear, partly because no paradigm has yet induced location transfer. Here, we tested whether performing a subsequent unrelated task with the categorization stimuli, combined with feedback, can induce location invariance. We utilized a paradigm shown to induce VPL location transfer (subsequent double-training) either with or without response feedback for the double-training task. Participants performed category learning in the trained visual field location followed by a 2IFC contrast discrimination task with the categorization stimuli on the opposite side of the visual field and finished with category learning post-test for both locations. Half the participants (N=15) received response feedback for the 2IFC task while the other (N=15) did not and eye-tracking was used throughout to enforce fixation. While the group without 2IFC feedback showed location specificity, the group with 2IFC feedback showed no location specificity. This work provides the first paradigm that induces location transfer of category learning, addressing a key gap in understanding location-specific learning. One possible explanation is that feedback during double-training triggered reinforcement-related neural processes (such as spatially-diffuse dopamine-mediated plasticity in the striatum) that enabled mapping from previous category learning to the new visual location. More broadly, this work adds to the growing list of similarities between category learning and perceptual learning and raises the question of whether feedback also plays a critical role in VPL location transfer.

Acknowledgements: This work was supported by the National Eye Institute of the National Institutes of Health under award numbers [K99EY034891, R00EY034891, R01EY019466, R01EY027841, and R01EY031705] and NSF-BSF under award number [BCS2241417].