Purchase this article with an account.
Caroline Robertson, Katherine Hermann, Anna Mynick, Dwight Kravitz, Nancy Kanwisher; Panoramic Memory Shapes Visual Representations of Scenes. Journal of Vision 2016;16(12):323. doi: https://doi.org/10.1167/16.12.323.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
As we navigate around our visual environment, our awareness of the place we are in seems to extend beyond the specific part of the environment that is currently in view to include a broader representation of the scene all around us. Here, we tested whether memory of a broad, panoramic space influences the ongoing representations of discrete views from within that panorama. Specifically, we introduced participants (N=21 behavioral study; N=12 fMRI study) to dynamic fragments of novel 360° spatial expanses, some of which contained overlapping visual content (Overlap Condition), and others of which did not (No-Overlap Condition) (Study Phase). Then, we tested whether discrete, non-overlapping snapshots from opposite edges of these spatial expanses become associated in memory, and acquire representational similarity in the brain, as a function of whether the visual information connecting them is known or unknown (Overlap vs. No-Overlap). On each trial of the fMRI study, participants were presented with single snapshots from the studied spatial expanses. Classification analyses showed that all independently-localized regions of the scene network (PPA, RSC, OPA) were able to discriminate individual scenes (all p< 0.001) and spatial layout (open/closed) (all p< 0.05). Importantly, representations in one region of the scene network were also sensitivity to broader spatial knowledge: the RSC showed significantly greater representational similarity between pairs of snapshots that had appeared in the Overlap vs. the No-Overlap condition (p< 0.008). Behaviorally, memory for the association between two snapshots was higher if they were were drawn from the Overlap vs. No-Overlap conditions (all p< 0.02), as was spatiotopic position memory (p< 0.02). Our results demonstrate that images of visual scenes evoke greater representational similarity in the brain when the visual information that unites them has been previously observed, suggesting that moment-to-moment visual representations are shaped by broader visuospatial representations in memory.
Meeting abstract presented at VSS 2016
This PDF is available to Subscribers Only