Abstract
The contrast sensitivity function (CSF), how sensitivity varies with the frequency of the stimulus, is a fundamental assessment of visual function. Elucidation of its mechanisms is instrumental to understand how the visual system works in both health and disease. Under photopic conditions, the CSF measured with stationary gratings exhibits a well-known band-pass shape that typically peaks around 3–5 cycles and gradually transitions to a low-pass shape when gratings are temporally modulated. It is generally assumed that the CSF is largely shaped by the response characteristics of retinal neurons. However, the sensitivities of these neurons, as measured in experiments with immobilized eyes, considerably deviate from the CSF, especially at low spatial frequencies, where they exhibit much stronger responses than expected from the CSF. Under natural viewing conditions, humans incessantly move their eyes, even when looking at a fixed point. These fixational eye movements transform the visual scene into a spatiotemporal flow of luminance on the retina and are not present in neurophysiological characterizations of cell responses, when the eyes are normally immobilized. We used neuronal models to quantitatively examine the impact of eye drift on neural activity and compare the responses of retinal ganglion cells to the CSF of primates. We show that consideration of the retinal consequences of incessant eye drifts, coupled with the known spatiotemporal response characteristics of retinal ganglion cells, accounts for the band-pass shape of the CSF as well as for its transition to low-pass with temporally modulated gratings. Consideration of residual retinal motion with imperfect retinal stabilization also provides an explanation for the puzzling finding that visual sensitivity shifts to higher spatial frequencies under retinal stabilization. These findings make specific predictions both at the behavioral and neuronal levels and suggest a fundamental integration between perception and action beginning at the retina.
Acknowledgement: Michele Rucci: NIH EY018363 - NSF BCS-1457238, 1420212. Jonathan D. Victor: NIH EY07977