Visual responses of ganglion cells of a New-World primate, the capuchin monkey, Cebus apella

Abstract

1. The genetic basis of colour vision in New-World primates differs from that in humans and other Old-World primates. Most New-World primate species show a polymorphism; all males are dichromats and most females trichromats. 2. In the retina of Old-World primates such as the macaque, the physiological correlates of trichromacy are well established. Comparison of the retinae in New- and Old-World species may help constrain hypotheses as to the evolution of colour vision and the pathways associated with it. 3. Ganglion cell behaviour was recorded from trichromatic and dichromatic members of a New-World species (the capuchin monkey, Cebus apella) and compared with macaque data. Despite some differences in quantitative detail (such as a temporal response extended to higher frequencies), results from trichromatic animals strongly resembled those from the macaque. 4. In particular, cells of the parvocellular (PC) pathway showed characteristic frequency-dependent changes in responsivity to luminance and chromatic modulation, cells of the magnocellular (MC) pathway showed frequency-doubled responses to chromatic modulation, and the surround of MC cells received a chromatic input revealed on changing the phase of heterochromatically modulated lights. 5. Ganglion cells of dichromats were colour-blind versions of those of trichromats. 6. This strong physiological homology is consistent with a common origin of trichromacy in New- and Old-World monkeys; in the New-World primate the presence of two pigments in the middle-to-long wavelength range permits full expression of the retinal mechanisms of trichromatic vision.

Figure 1. Illustration of the LED waveforms in the various stimulus protocols

Continuous line, 636 nm LED; dashed line, 554 nm LED. A, to determine responses to step changes, 200 ms pulses (incremental and decremental luminance, and redward and greenward chromatic steps) were used. B, for the HFP protocol, 19 different ratios of 638/554 nm modulation depths were used, ranging from 0.1 (ten times the modulation in the 554 nm light than in the 638 nm light) to 10 (vice versa). As indicated, to achieve this range the modulation depths of both LEDs were modified through the series of ratios. C, for the phase protocol, modulation depth was held constant and the relative phase of the two LEDs was varied.

Figure 2. Templates and responses of dichromat cells to the HFP and phase protocols

A, in the HFP protocol, the relative modulation depths of counter-phase modulated 554 and 638 nm LEDs were manipulated. A relative amplitude of unity corresponds to the human luminosity spectral sensitivity function, where a standard observer sees minimum flicker. Based on cone pigment spectra, the expected cone signals’ amplitudes were predicted. The discontinuity in the 535 nm template is because modulation depths of both 554 and 638 nm diodes were manipulated to provide the different ratios. In the lower panels are shown data and fitted curves from the 563 and 535 nm phenotypes (one MC (□), one PC cell (▪) in each case, data of the MC cell being shifted up for clarity), shown to match the appropriate templates. B and C, response amplitude and phase templates for phase protocol. The relative phases of the 554 and 638 nm lights was manipulated at fixed modulation depth. Luminance modulation corresponds to a phase of 0 deg, chromatic modulation to a phase of ±180 deg. Again, the fit (one MC, one PC cell) of the templates to the data provided an unambiguous indication of the genotype. Modulation depth was 20 % for MC cells and 50 or 100 % for PC cells. Responses were averaged over 6 s for each point. First-harmonic amplitudes and phases were extracted by Fourier analysis. Measurements were at 19.8 Hz, 4 deg field, 2000 Td.

Figure 3. Spectral sensitivity of 563-535 nm trichromat achromatic (MC) cells

Spectral sensitivity of 563-535 nm trichromat achromatic (MC) cells was measured using the same protocols as in Fig. 2. A, amplitude and phase data for the phase protocol were fitted with a model assuming simple summation of cone signals. Stimulus parameters as in Fig. 2. B, flicker photometric protocol on the same cell. The null is close to unity, indicating that the 562:535 nm cone weighting for this cell was ca 2:1 as for a standard human observer. C, distributions of cone weights over the cell sample (32). Comparison macaque data were taken from Kaiser et al. (1990). Distributions of cone weightings are similar in the two species, arguing against significant patchiness in the cone array. Details of the physiological measurements as in Fig. 2.

Figure 4. Frequency-doubled responses in MC cells

A, averaged responses of MC cell to two cycles of counterphase modulation of the 554/638 nm lights at relative modulation depths as indicated at left. The first harmonic null is associated with a response at twice the modulation frequency. The phase of the frequency-doubled response is such that peaks lie midway between those in the upper and lower histograms. B, plots of 1st (▪) and 2nd harmonic amplitudes (•) as a function of relative 554/638 nm ratio (4 deg field). First-harmonic data fitted as in Fig. 3. The second- harmonic response rises to a peak at the first-harmonic minimum. C and D, data from an MC cell tested with two sizes of spot. The frequency-doubled response almost disappears with small stimuli, and can thus be localized to the receptive field surround. Details of the physiological measurements as in Fig. 2, except that 9.76 Hz modulation was used.

Figure 5. Phase protocol for MC cells at different frequencies

A, amplitude and phase of 1st harmonic response for an on-centre cell at three temporal frequencies as indicated. B, similar data for an off-centre cell. At low temporal frequencies the amplitude of minimum response is displaced away from ±180 deg. As frequency increases, this effect diminishes. The continuous lines show the fit of a model described by Smith et al. (1992). Modulation depths were 50 % for the on-centre, 20 % for the off-centre cell. Six-second activity averaged for each point; other details as in Figs 2–3.

Figure 6. Phase protocol for MC cells: effects of stimulus size and mean chromaticity

Shown are response amplitude and phase, and fits, for an on-centre MC cell. A, effect of stimulus size: 4 deg and 0.5 fields (2.44 Hz, mean chromaticity 595 nm). For the 4 deg field, the response minimum is shifted away from ±180 deg to ca 80 deg. This shift is absent with small stimuli. B, the phase effect is also absent on adding a 460 nm, 1000 Td background. These effects are also found in macaque MC cells.

Figure 7. Summary of phase of minimum response for MC cells

The phase of minimum response was obtained for each cell and condition by fitting with a cosine function, and the data are plotted here against temporal frequency. Mean and 95 % confidence limits are shown. Comparison macaque data were obtained from Smith et al. (1992). A similar shift is present in both species, although variability is greater in capuchin.

Figure 8. Phase protocol for PC cells at different frequencies

A, response amplitudes and phases for a +M − L cell. B, similar data from a +L − M cell. At low temporal frequencies, there is a maximum response to chromatic and a minimum response to luminance modulation. As frequency increases, the phase of minimum response shifts, in opposite directions for the two cell types. The continuous lines show the fit of a model described by Smith et al. (1992). Modulation depths were 50 % for both cells.

Figure 9. Summary of phase of minimum response for PC cells

A, phases of minimum response were obtained by fitting a cosine function for each cell and conditions. L- (5) and M-cone (7) centre cells show shifts in different directions. A difference in latency of the opponent cone mechanisms by a few milliseconds is responsible for the shift. Data resemble those of macaque. B, cone weightings in opponent cells derived from model fits; comparison of capuchin and macaque. A value of 0.5 gives an equal cone balance, with no response to luminance modulation (and a vigorous response to chromatic modulation). Data from both species cluster around this value.

Figure 10. Responses to step changes in luminance and chromaticity of MC and PC cells of trichromat

A, responses of MC cell. B, responses of +M − L PC cell. Percentages are modulation depths; 100 % chromatic modulation corresponded to a mean cone modulation of near 45 %. MC cells give a phasic response to luminance increments and decrements even at low contrast, and little or no response to red or green chromatic perturbations. The cone-opponent cell gives a vigorous, sustained response to chromatic perturbations and little response to luminance. Average of 20 stimulus repetitions, 16 ms binwidth.

Figure 11. Responses to step changes in luminance and chromaticity of MC and PC cells of dichromat

The same format is followed as in Fig. 10. There is no chromatic response from the PC cell, but its luminance response is also weak.

Figure 12. Contrast response relations of different cell types in dichromats and trichromats

A, response amplitude and phase from MC cell of trichromat. Two frequencies, 0.61 and 19.5 Hz, are shown in these and other panels. MC cell responses at 19.5 Hz rapidly saturate accompanied by an advance in response phase. B and C, PC cell responses to luminance and chromatic modulation. Less saturation is evident, and no phase advance. The saturation of response that does occur appeared to be linked to response rectification. For the chromatic modulation, mean cone contrast has been plotted on the ordinate. Responsivity is high even at 0.61 Hz. For luminance modulation, PC cell responsivity is low at low frequency but increases at high frequency due to a centre-surround latency difference no longer causing full cancellation of opponent signals. D and E, response amplitude and phase of MC cells and PC cells of a dichromat. Data resemble trichromat luminance data. Data have been fitted with Naka-Rushton functions. Physiological data collection details as in Fig. 2.

Figure 13. Comparison of temporal response at different contrasts for capuchin and macaque

Data such as those of Fig. 12 plotted in an alternative form and compared with macaque results. Response amplitude is plotted as a function of frequency at different contrasts. A, for the MC cells (luminance modulation), the frequency of maximum response increases with increasing contrast, and there is an indication of a resonance peak at ca 50 Hz. B, for the PC cells (chromatic modulation), the curves at different contrasts show no such change in curve shape as contrast increases. Macaque data were obtained under identical conditions to the capuchin data (Smith et al. 1990).

Figure 14. Comparison of temporal modulation transfer functions for trichromatic and dichromatic capuchin and the macaque

Contrast gain is the initial slope of the Naka-Rushton functions fitted in Fig. 12. Mean data from five well-studied cells was averaged in each case. The form of the curves are similar in different phenotypes and species, but capuchin cells continued to respond to higher temporal frequencies.