Poster Presentation

Poster C in Poster Session C: Wednesday, August 5, 9:30 – 11:15 am, Kimmel Center, Shorin & Rosenthal Rooms

Persistent polar-angle asymmetries in performance and system-level computations despite equating cortical size

David Tu¹, Shutian Xue¹, Jonathan Winawer¹, Marisa Carrasco¹; ¹New York University

Presenter: David Tu

Visual performance is location-dependent: Typically, it decreases with eccentricity and is higher along the horizontal than the vertical meridian–termed the horizontal-vertical anisotropy (HVA)–and along the lower- than the upper-vertical meridian–vertical meridian asymmetry (VMA). We have previously studied how (i) system-level computations vary across the visual field: Internal noise increases with eccentricity and gain varies around polar angle, yielding both an HVA and VMA, and (ii) cortical surface area modulates performance: equating cortical size (M-scaling) eliminates contrast threshold differences across eccentricity, but only reduces polar angle asymmetries. Here, we investigated whether and how scaling stimulus size to equate cortical surface area across locations (M-scaling)—on a per individual basis—modulates performance and system-level computations. To do so, we estimated contrast thresholds using an equivalent-noise protocol (signal embedded in varying noise levels), and fit a perceptual template model to estimate gain and internal noise at each location, using both constant and M-scaled stimulus sizes. Across eccentricity, internal noise increased when stimulus size was constant. M-scaling eliminated this eccentricity effect by equating internal noise, while gain increased in the periphery. Around polar angle, gain was higher at the horizontal than the vertical meridian and at the lower- than the upper-vertical meridian, but internal noise did not vary. M-scaling diminished the HVA in gain and resulted in a VMA in internal noise. These results indicate that visual performance differences across eccentricity and around polar angle reflect different computations: the former in internal noise and the latter in gain. Critically, equating cortical surface area across locations on an individual basis eliminates eccentricity differences, but polar angle asymmetries in performance and system-level computations persist, albeit in an attenuated manner.

Topic Area: Computational Models of Vision & Visual Cortex