Abstract
“The receptive field centers of midget ganglion cells of the parvocellular pathway are usually assumed to derive from a single cone. The input to the surround is controversial; cone-specific or mixed-cone surround models have been suggested. We have measured spatial frequency (SF) tuning curves to luminance, red-green chromatic, or L- and M-cone isolating gratings. A satisfactory model should be able to predict responses to all grating types.
With L- or M-cone isolating gratings, for most cells the slope of the SF tuning curves are often too shallow to be predicted by a single Gaussian mechanism. An inverse cosine transform of the SF curves gives spatial structure for M- and L-cone receptive fields. There is much inter-cell variability; some cells seem to have strong, local opponency with broad opponent flanks, some have more classical center-surround opponent structure. We could describe the data satisfactorily with a dual opponent-mechanism model, a local, opponent receptive field added to a much wider opponent mechanism. Most cells were approximately spatially isotropic and well-behaved in the phase domain.
These results indicate receptive field structure of midget ganglion cells is more complex and variable than expected. Standard difference-of-Gaussian models are unlikely to be adequate. It is feasible that mechanisms exist to adjust cone weighting (which is usually well balanced in M,L opponent cells) to local cone distributions. Lastly, the extensive spatial summation observed with psychophysical detection of red-green perturbations may have in part a retinal substrate.”