Purchase this article with an account.
Brian McCann, Wilson Geisler, Mary Hayhoe; Decoding natural signals from the peripheral retina. Journal of Vision 2011;11(11):1193. doi: 10.1167/11.11.1193.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Peripheral ganglion cells have lower density and larger receptive fields than the fovea. Consequently, the quality of the visual signals that they relay is reduced. The information contained in peripheral ganglion cell responses can be quantified by how well they predict the foveal ganglion cell responses to the same stimulus. Here, we developed a model of human ganglion cell outputs combining existing measurements of the optical transfer function with the receptive field properties and sampling densities of P ganglion cells. Next, we simulated a small spatial population of P-cell responses to 1° patches from a large sample of luminance-calibrated natural images. For each image patch we simulated population responses for retinal eccentricities ranging from 0°–15°.
Spatial phase and orientation are largely preserved by circularly symmetric receptive fields. Therefore, we characterized the population of ganglion cell responses by their radially-averaged spatial power spectrum. A two parameter function adequately summarized these power spectra. One parameter describes the falloff of power with spatial frequency; the other describes the variance of the responses across cells in the population (power per ganglion cell). We found that the variance was constant with retinal eccentricity on average, but for a given patch the falloff parameter in the periphery was strongly predictive of the foveal variance. For example, at 15° eccentricity the percent error in the Bayes optimal prediction of foveal variance improved by a factor of 2 by taking into account both peripheral falloff and variance, as opposed to peripheral variance alone.
Humans could exploit this information when decoding peripheral P-cell responses. Decoding in this way might facilitate various known perceptual constancies (e.g., contrast and blur constancy) creating the common percept of a sharp peripheral image. Further, it could reduce the number of eye movements necessary to encode the image.
This PDF is available to Subscribers Only