Abstract
Presentation Description :
Despite decades of progress in understanding retinal mechanisms and function, relatively little is known about how the retina processes natural images. Understanding vision in natural conditions presents important conceptual and technical opportunities. First, it provides an unusually strong opportunity to relate circuit mechanisms to computations of behavioral significance. Second, understanding how the retina encodes natural inputs is necessary for understanding the division of labor between retinal and cortical circuits. Third, engineering approaches to restore vision require better models of natural vision.
I will describe recent work probing how the receptive field (RF) surrounds of primate retinal ganglion cells (RGCs) integrate spatial structure in natural inputs. We find a strong asymmetry in the impact of spatial structure in the RF surround on responses of On and Off parasol RGCs: spatial structure decreases responses of Off parasol RGCs, as expected from classic center-surround antagonism, but increases responses of On parasol RGCs. As a result, On parasol RGC RFs differ substantially for spatially-homogeneous spots vs spatially-structured inputs such as gratings. The On/Off difference in integration of spatially-structured input in the classical RF surround has a significant impact on how natural images are encoded. Specifically, spatial structure in the classic surround suppresses responses of Off RGCs to natural inputs but can enhance responses of On RGCs.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.