Abstract
Pelli & Farell (1999) showed that visual sensitivity can be factored into efficiency and equivalent input noise Neq. Efficiency is only weakly dependent on size and nearly independent of eccentricity 𝜑 (Rosen & Pelli, submitted to JOV), so the strong dependence of visual sensitivity for simple targets on size and eccentricity is due to changes in the equivalent input noise. In the fovea, Pelli & Raghavan (Cosyne 2016) showed that the equivalent input noise is a sum of photon noise and a cortical noise. The photon noise Nphoton depends only on luminance, and is independent of letter size; the cortical noise is proportional to letter area A. Thus, Neq = Nphoton + aA, where a is a constant. We extended their measurements to the periphery, measuring threshold energy E for identification of a letter of size 0.5 to 16 deg at eccentricities 0 to 32 deg with and without a white noise background (RMS contrast 0.26). We estimated the equivalent input noise Neq = N E0/(E-E0), from the threshold energy E in noise, threshold energy E0 without noise, and noise power spectral density N. Supposing that the periphery is like the fovea, but scaled by the cortical magnification factor, we had expected to find that the cortical noise term, like cortical magnification, would depend on eccentricity. However, the data reveal a simpler result, and are well fit by Pelli & Raghavan's foveal equation with an additional term that is proportional to eccentricity 𝜑, Neq = Nphoton + aA + b𝜑, where b is a constant. We tentatively attribute this new noise term to the density of retinal ganglion cell spikes, which is inversely proportional to eccentricity.
Meeting abstract presented at VSS 2017