Abstract
Humans perceive the world veridically and what we physically “see” remains largely invariant across individuals. The complex visual process of “seeing” involves a cascade of image projections from our retina to the retinotopically organized visual cortical areas. What we know from fMRI scans is that visual inputs projected onto the brain are distorted. So how do those distorted projections relate to one another in the visual processing cascade and ultimately transform our perception of the physical world into a non-distorted one? Existing retinotopic maps show that when visual inputs are passed from the retina to V1, V2, … to Vn, geometric distortions take place in each step. But the precise way in which visual inputs are distorted in each step has remained an unexplored mystery. We precisely quantified the geometric distortions from retina to V1 and V2 using recently developed quasiconformal quantification framework of retinotopic maps. The framework applied the principles of computational conformal geometry and quasiconformal Teichmüller theory to quantify the precise amount and direction of visual input distortion in each local region of the cortical surface using the Beltrami coefficient (BC), a differential geometry concept that measures angle distortions in quasiconformal maps. The geometric distortions from retina to V1 and V2 were compared using their respective Beltrami coefficient maps (BCM) with retinotopic data from the Human Connectome Project (n=181). The average measurement of mean BC distortion from V1 to V2, defined as the ratio of V2 to V1, was found to be 0.969±0.503 (ratio of 1 means no distortion). The average variance of mean BC distortion was 0.184±0.715. The results suggest that retinal image projections to V1 and V2 are largely the same. With advances in retinotopy technology and better fMRI data, we will apply the framework to analyze retinotopic maps in higher order visual cortical areas.