Linfeng Tony Han¹ (), Russell A. Epstein¹; ¹University of Pennsylvania

People rely on representations of places to navigate through the world. To learn a new place, the visual system must integrate discrete views into a unified representation. There are at least two ways that this might be done: first, by integrating across the panorama of views obtained when standing at a single location; second, by integrating across different views of the same scene or landmark obtained when standing in different locations. Guided by previous work (Marchette et al., 2015; Robertson et al., 2016; Berens et al., 2021), we hypothesized that these two viewpoint-integration processes would have different neuroanatomical substrates in scene-selective cortex. To test this idea, we familiarized participants with a route through a virtual city containing 24 storefronts that were controlled for their low-level visual similarities. The storefronts were locally associated in pairs, which could either be storefronts on different buildings that were directly across from each other on the same street (same-panorama condition), or storefronts on different sides of the same building visible from different streets (same-landmark condition). These associations were learned through multiple viewings of a movie sequence that strictly controlled the time that each item was viewed and the interval between them. After learning, participants were scanned with fMRI while viewing the storefronts and performing a judgment of relative direction task. Preliminary analyses of multivoxel activation patterns in scene-selective retrosplenial complex (RSC) and parahippocampal place area (PPA) showed an interaction between type of association (same-panorama, same landmark) and brain region, whereby RSC showed relatively stronger association for same-panorama storefront pairs and PPA showed relatively stronger association same-landmark storefront pairs. These results support the existence of dissociable mechanisms for integrating across views to represent places as either the location of the observer (same panorama) or the location observed (same landmark).

Acknowledgements: This work was supported by NIH grant R01EY022350 to R.A.E.