We demonstrate that a combination of responses of various types of spectrally opponent sustained cells of the macaque lateral geniculate nucleus (LGN) may be related to equidistant color space. Response curves of such cells to stimuli of different luminance ratios and wavelengths are similar to the first stage opponent coordinate functions of the new SVF color-difference formula [T. Seim and A. Valberg, Color Res. Appl. 11, 11 (1986)]. Mathematical simulation of the responses of these cells to a variety of color stimuli is possible through computation of cone excitations and subsequent sums and differences of cone signals. When the response functions thus obtained for cells are used to replace the corresponding coordinates of the SVF diagram, the distributions of equiluminous stimuli with the same sensory differences from an achromatic stimulus approximate ellipses about the white point, and loci of constant hue approximate straight lines. Improved uniformity may be obtained by linear combinations of these cells' outputs or by including more cell types with best responsiveness to other directions of color space. This indicates possible roles of LGN cell types for color scaling in primates, in that color scaling observed psychophysically is an implicit property of these cells' responses.

Building upon electrophysiological recordings from the lateral geniculate nucleus (LGN) of the macaque monkey, we describe a model for neural processing of color and brightness/lightness information that starts in the cone receptors and continues in the opponent cells of the retina, LGN, and visual cortex. The excitation of the three cone types to direct stimulation by light is modified in accordance with a hyperbolic response function before providing inputs to retinal ganglion cells. Using weighted differences of such cone outputs, we simulate the responses of common types of opponent ganglion and geniculate cells to light modulation along the chromatic and luminance dimensions. Extrapolating the results of the simulation, we suggest a way in which the brain might combine inputs from the geniculate to obtain correlates of chromatic and achromatic color vision and of brightness/ lightness perception. In particular, we demonstrate for the first time how combinations of ''L-M'' and ''M-L'' parvocellular ON-and OFF-opponent-cells may lead to a quantitative account of brightness and blackness scaling.

Comparisons of S- or prepotential activity, thought to derive from a retinal ganglion cell afferent, with the activity of relay cells of the lateral geniculate nucleus (LGN) have sometimes implied a loss, or leak, of visual information. The idea of the "leaky" relay cell is reconsidered in the present analysis of prepotential firing and LGN responses of color-opponent cells of the macaque LGN to stimuli varying in size, relative luminance, and spectral distribution. Above a threshold prepotential spike frequency, called the signal transfer threshold (STT), there is a range of more than 2 log units of test field luminance that has a 1:1 relationship between prepotential- and LGN-cell firing rates. Consequently, above this threshold, the LGN cell response can be viewed as an extension of prepotential firing (a "nonleaky relay cell"). The STT level decreased when the size of the stimulus increased beyond the classical receptive field center, indicating that the LGN cell is influenced by factors other than the prepotential input. For opponent ON cells, both the excitatory and the inhibitory response decreased similarly when the test field size increased beyond the center of the receptive field. These findings have consequences for the modeling of LGN cell responses and transmission of visual information, particularly for small fields. For instance, for LGN ON cells, information in the prepotential intensity-response curve for firing rates below the STT is left to be discriminated by OFF cells. Consequently, for a given light adaptation, the STT improves the separation of the response range of retinal ganglion cells into "complementary" ON and OFF pathways.

An improved quantitative description of the equidistant color and lightness spacing of the Munsell and OSA‐UCS Systems is presented. The basic idea is to apply physiologically acceptable hyperbolic response functions to relate light absorption in and the excitation of the three cone types with visual response. A similar nonlinear equation describes the relationship between the luminance factor Y and a lightness magnitude. Subsequent linear combinations of the nonlinear functions describe opponent‐color transformations. The resulting formulation, called the SVF formula, is superior to the best available empirical formulae, such as CIELAB and CIELUV, in describing the Munsell color space, and it performs even better than the OSA L g j formula on the OSA‐UCS data.

A particular version of a spectral integrator has been designed. It consists of a xenon lamp whose light is dispersed into a color spectrum by dispersing prisms. Using a transmissive LCD panel controlled by a computer, certain fractions of the light in different parts of the spectrum are masked out. The remaining transmitted light is integrated and projected onto a translucent diffusing plate. A spectroradiometer that measures the generated light is also attached to the computer, thus making the spectral integrator a closed-loop system. An algorithm for generating the light of a specified spectral power distribution has been developed. The resulting measured spectra differ from the specified ones with relative rms errors in the range of 1%-20% depending on the shape of the spectral power distribution.