Abstract
[BACKGROUND] Visual performance differs both as a function of eccentricity and polar angle. At isoeccentric locations, performance is better along the horizontal than vertical meridian (horizontal-vertical anisotropy, “HVA”) and better along the lower than upper vertical meridian (vertical-meridian asymmetry, “VMA”). Computational modeling shows that polar angle variations in optical quality and cone density can account for only a small fraction of polar angle variations in behavior (Kupers, Carrasco, Winawer, 2019). Here, we compare asymmetries around the visual field in cones, retinal ganglion cells, and visual cortex. [METHODS] We computed HVA and VMA as percent difference in cone density along cardinal meridia using two datasets (post-mortem, Curcio et al. 1990; adaptive optics, Song et al. 2011) in two analysis toolboxes (ISETBIO; rgcDisplacementMap), and in midget retinal ganglion cell density using two quantitative models (Watson, 2014; Barnett & Aguirre, 2018). Cortical asymmetries were computed from the Human Connectome Project retinotopy dataset (Benson et al., 2018) as percent difference in V1/V2 cortical surface area in ±10º wedges centered on the cardinal meridia. [RESULTS] For cone density, results are consistent across datasets and toolboxes: a constant ~20% HVA from 2–40º eccentricity, a slightly inverted VMA from 0–5º (denser in upper visual field), and no systematic VMA beyond 5º. The two models of mRGC density show a common general pattern: larger HVA and VMA than cones (at 3.5º, HVA: ~23–31%, VMA: ~15%), both of which increase with eccentricity (at 40º, HVA: ~60%, VMA:~50%). Cortical surface area asymmetries are yet larger than mRGC density asymmetries (at 3.5º, HVA: ~46%, VMA: ~46%, each increasing with eccentricity). [CONCLUSION] Asymmetries around the visual field are amplified from cones to mRGCs and from mRGCs to cortex. It will be important to implement computational models to test whether mRGCs and V1 asymmetries can quantitatively explain visual performance differences.