Carl R. Ingling

The relationship between spectral sensitivity and spatial sensitivity for the primate r-g X-channel

Which visual channel detects high spatial frequencies during careful fixation? Color vision models based on psychophysical data contradict electrophysiological results. According to electrophysiology, the channel which mediates foveal acuity originates in the small, tonic color-opponent r-g units of the X-cell pathway. However, psychophysical models assign acuity to the V lambda channel because when acuity is used as a criterion for equating luminosity it is additive. In all opponent-color models the r-g channel is subadditive and hence is excluded from mediating acuity. We show that the r-g channel adds cone signals for high spatial frequencies and subtracts them for low, and conclude that the major achromatic channel for human foveal vision originates within the r-g color-opponent channel. Quantitative analysis makes explicit the interaction between the spatial and spectral variables for the simple-opponent cells which predominate in primate foveal vision.

The spatiotemporal properties of the r-g X-cell channel.

Analysis of the simple-opponent r-g receptive field of the X-channel shows that it is tuned to both high and low temporal frequencies, high and low spatial frequencies, and that its spectral sensitivity is both chromatic and achromatic.

Perceptual correlates of magnocellular and parvocellular channels: seeing form and depth in afterimages.

The chromatic and achromatic channels of psychophysical models do not map simply onto the parvocellular and magnocellular channels of electrophysiology because the parvo channel carries both chromatic and achromatic signals. If vision of stabilized images be mediated by the parvo channel, then this channel mediates form and depth percepts.

Red - Green Opponent Spectral Sensitivity: Disparity Between Cancellation and Direct Matching Methods

The spectral sensitivity at the opponent stage of the visual system is traditionally measured by a hue-cancellation procedure. Comparison with a direct hue-matching method shows that cancellation overestimates short-wavelength sensitivity by as much as a factor of 30. The observation implies that different mechanisms control the perception of short-wavelength and long-wavelength redness.

The achromatic channel—I. The non-linearity of minimum-border and flicker matches

Abstract Flicker and minimum-border matches made at one intensity do not hold at other intensities. For flicker matches, more saturated lights must be increased relative to less saturated lights as intensity increases, in agreement with the hypothesis that a compression precedes the addition of the signals from the receptors. Minimum-border matches are more nearly linear. The two methods give the same radiances for an equal-luminance spectrum for a 100 troland standard; the methods depart for lower and higher intensities.

Simple-opponent receptive fields are asymmetrical: G-cone centers predominate

For quantitative models of color vision, the R-cone contribution to the r-g channel is less than half of the R-cone contribution to the V lambda channel. There is currently no explanation of how this different contribution of R cones to the two channels comes about. We propose an asymmetrical receptive-field arrangement to explain the difference in weighting. Because cones in receptive-field surrounds are weighted less than cones in centers, placing R cones predominantly in surrounds and G cones in centers provides a simple differential weighting mechanism. Electrophysiological and psychophysical evidence substantiates such an asymmetry of simple-opponent fields.

Spectral sensitivity for flicker and acuity criteria

Different channels in the visual system mediate the detection of flicker and the detection of high spatial frequencies. The magnocellular channel is optimized for flicker detection, whereas the parvocellular channel is optimized for color vision and spatial resolution. The spectral sensitivity of the magnocellular (flicker) channel is obtained by combining cone inputs in the ratio R/G = 5/3; the spectral sensitivity of the parvocellular channel is obtained with the ratio R/G = 2/3. However, when the parvocellular channel is used for resolution, the sensitivity changes from R/G = 2/3 to R/G = 5/3. By hypothesis, this occurs because only parvocellular centers resolve high spatial frequencies and because parvocellular centers are distributed in the same ratio as cones feeding magnocellular cells.

How neural adaptation changes chromaticity coordinates

Color-matching data obtained by Crawford and by Richards imply failures of additivity of color-matching functions. These failures raise serious doubts about the fundamental principles of trichromatic theory. If color matches are made by causing the lights on the two sides of a field to have equal quantum absorptions in the photopigments, and photopigment spectral sensitivities are invariant with adaptation, it is impossible for color-matching functions to be nonadditive. Since the data cannot be faulted, the solution of the dilemma must lie in finding how observers can make matches that violate the assumptions. Known neural interactions in the visual system suggest that the observers were not setting equal quantum absorptions on the two sides of the field. A model of the neural interaction accounts for the departures.

Stiles’s π 5 mechanism: failure to show univariance is caused by opponent-channel input

Stiles’s π5 field sensitivity approaches the sensitivity of the long-wavelength (L) cone as the duration of the test flash decreases. Further, for short test flashes mixtures of red and green backgrounds are additive. For long test flashes the backgrounds cancel, the mixture raising threshold less than expected. These results are explained if π5 is approximately the L-cone sensitivity combined with some opponent-channel sensitivity.

Tonic-phasic-channel dichotomy and Crozier’s law

The constancy of the dynamic range in a luminance-discrimination task is known as Croziers law; an old rule says that about half of a log unit spans the range from a low to a high frequency of seeing. For our conditions the slope of the psychometric function is steeper for short than for long test flashes; Croziers law requires a different constant when temporal parameters change. This result is substantiated by an analysis of Massofs data [Vision Res. 21,995 (1981)] on the variation of the slope of psychometric curves for different wavelengths. The change in Croziers constant between conditions may reflect the presence of more than one detection channel. If short test flashes are detected by phasic channels and long test flashes by both phasic and tonic channels, then our result implies a shorter dynamic range for phasic (Y-cell) than for tonic (X-cell) channels.