Cone receptive field in cat retina computed from neural circuit model

We have derived the cone receptive field in cat retina by two independent methods. First, we used the pattern of anatomic convergence from cones to beta ganglion cells to compute what cone receptive field would produce the published physiology of beta cells. This derivation was based on the assumption that visual signals from cones are summed linearly through known anatomical pathways to produce the beta receptive field. We used a deconvolution method to successively approximate the best fit for a cone receptive field. The resulting cone receptive field at an eccentricity of 1 degree had a broad center of 50 um diameter at 1/e amplitude, and a wider shallow surround of about 3% peak center amplitude.; We corroborated this conclusion using a compartmental model to simulate the physiological responses of the neural network responsible for a single cone's receptive field. The model consisted of an array of 900 cones (30 x 30), 9 type A and 36 type B horizontal cells, and included a visual stimulus, the effects of optical blur and scatter, and synaptic parameters such as presynaptic threshold and postsynaptic reversal potential. To facilitate building this model, we developed a neuron description language based on the C language which allowed the neural network to be concisely described and modified easily at a high level. A computer simulation program translated the neural description into a low level compartmental model which was simulated with difference equations based on the Crank-Nicholson method of numerical integration.; We obtained a combination of parameters which after convolution would approximately match the beta ganglion cell receptive field. Our results suggest that: optical blur and cone-cone gap junctions together create a broad dome-shaped center cone response with an exponential skirt, and negative feedback from type A and B horizontal cells creates a weak distal surround and stronger proximal surround to narrow the center skirt. These results imply that the receptive fields of all retinal neurons have broad centers and weak antagonistic surrounds originating in the outer plexiform layer.