Detection and resolution of drifting gratings by motion detectors in the fish retina

Fish have highly developed vision that plays an important role in detecting and recognizing objects in different forms of visually guided behavior. All of these behaviors require high spatial resolution. The theoretical limit of spatial resolution is determined by the optics of the eye and the density of photoreceptors. However, further in the fish retina, each bipolar cell may collect signals from tens of photoreceptors, and each ganglion cell may collect signals from tens to hundreds of bipolar cells. If we assume that the input signals in this physiological funnel are simply summed, then fine gratings that are still distinguishable at the level of cones should not differ from the homogeneous surface for the ganglion cells. It is therefore generally considered that the resolution of the eye is determined not by the density of cones, but by the density of ganglion cells. Given the size of the receptive field of ganglion cells, one can conclude that the resolving power at the output of the fish retina should be ten times worse than at its input. But this contradicts the results of behavioral studies, for, as it is known, fish are able to distinguish periodic gratings at the limit of resolution of the cones. Our electrophysiological studies with extracellular recording of responses of individual ganglion cells to the motion of contrast gratings of different periods showed that the acuity of ganglion cells themselves is much higher and is close to the limit determined by the density of cones. The contradiction is explained by the fact that ganglion cells are not linear integrators of the input signals, their receptive fields being composed of subunits with significantly smaller zones of signal summation where nonlinear retinal processing takes place.