Spatial vision at the scale of the human foveal cone mosaic
Vision begins when light is imaged on the retina and sensed by the photoreceptors. Both optical blur and discrete spatial and spectral sampling of the retinal image by the receptors set fundamental limits on the information available for subsequent perceptual inferences. Despite these limits, the brain processes in the information in a manner that supports our exquisite spatial and color vision. In this talk, I will review psychophysical experiments and computational modeling that aim to study human vision at the fundamental cellular scale of the photoreceptor mosaic. The goal of the work is to clarify how the brain handles the information loss imposed by optics and photoreceptor sampling. An adaptive optics (AO) scanning laser ophthalmoscope was equipped with specialized hardware to allow precise delivery of stimuli that target specified retinal locations. Because the AO system corrects for optical blur, the stimuli can have a retinal size commensurate with the acceptance aperture of individual foveal cones. Using this system, we measured the area of complete summation for detection of increments (Ricco’s area). The data show that even with correction for optical blur and stabilization with respect to fixational eye movements, the neural summation area has a diameter of multiple cones. Computational modeling shows that the summation mechanism cannot be a simple linear filter, a conclusion supported by preliminary data from experiments that examine thresholds for combinations of increments and nearby decrements.