Representation of color stimuli in awake macaque primary visual cortex

Abstract

We investigated the responses of single neurons in primary visual cortex (area V1) of awake monkeys to chromatic stimuli. Chromatic tuning properties, determined for homogeneous color patches presented on a neutral gray background, varied strongly between cells. The continuum of preferred chromaticities and tuning widths indicated a distributed representation of color signals in V1. When stimuli were presented on colored backgrounds, chromatic tuning was different in most neurons, and the changes in tuning were consistent with some degree of sensitivity of the neurons to the chromatic contrast between stimulus and background. Quantitatively, the average response changes matched the magnitudes of color induction effects measured in human subjects under corresponding stimulus conditions.

Figure 1. Chromatic Tuning of Two V1 Neurons

(A and B) Tuning to isoluminant stimuli. Horizontal axis, L-M cone contrast variation; vertical axis, S cone contrast variation. Stimulus hue changes as a function of polar angle. Data points in the polar plots show mean spike rate in a time window 50–150 ms after stimulus onset, error bars denote standard error. Tuning curves are linear interpolations of data points. Thin circles around the centers of polar plots indicate spontaneous firing rates. Insets show peri-stimulus time histograms for responses to stimuli in the corresponding directions. (C and D) Tuning contours in three-dimensional color space, derived from responses to stimuli in 26 directions. (C) Broadly tuned, nonopponent neuron. (D) Color-opponent neuron tuned narrowly to different hues (azimuth) and more broadly tuned to stimuli with luminance differences (elevation).

Figure 2. Chromatic Tuning for Isoluminant Colors

Tuning parameters were derived from interpolated responses in the (L-M,S) plane (see text). (A) Distribution of tuning peak directions. The inset shows a polar plot of the same data, smoothed by replacing each data point with a Gaussian of 5° width. The distribution is continuous, nonuniform, and asymmetric. (B) Distribution of tuning curve widths (full width at half height). The dashed line denotes the width of a cosine function, corresponding to linear combination of cone inputs. The distribution indicates that most neurons combine cone inputs nonlinearly.

Figure 3. Effect of Background Chromaticity on Tuning Curves

Tuning curve pairs of eleven neurons for stimuli on gray background (black curves) and on colored background (gray curves), respectively. Coordinate axes as in Figure 1 (A and B). Inset at top right indicates the chromaticities of the eight stimuli (black dots) and the eight possible background chromaticities (gray circles) in the isoluminant plane. In each plot, a small gray circle indicates the direction of the chromaticity of the colored background. For each neuron, both tuning curves were obtained during the same experiment, in which the combination of stimulus and background color was chosen randomly from trial to trial. For most neurons, a response reduction was observed with chromatic background for stimuli in or near the direction of the background chromaticity. Several neurons showed a response enhancement for stimuli near the opposite direction.

Figure 4. Time Course of Responses

Average responses of 23 cells that had been tested with stimuli on gray and on (+L-M) backgrounds. Solid lines show mean spike rate to (+L-M) stimuli presented for 500 ms. Dashed lines show activity for trials without stimulus, i.e., the stimulus region remained gray during the trial. Black line, gray background; gray line, (+L-M) background. Responses to stimuli on colored background are lower than to stimuli on gray background.

Figure 5. Effect of Background Chromaticity on Response Strength

(A) Distribution of background modulation index, measuring the reduction of responses to stimuli on a chromatic background relative to the responses for the same stimuli on the neutral gray background. Stimuli had a cone contrast of 0.15 in the same direction as the chromatic background with contrast 0.1. The median of the distribution is 0.69. Black data points with crosses indicate corresponding values for psychophysical data of eight subjects (median: 0.79). (B) Mean response as a function of stimulus contrast. Stimuli were from an axis in the (L-M,S) plane in the direction of the chromatic background color. Zero stimulus contrast corresponds to a gray stimulus. Negative contrast corresponds to a stimulus chromaticity opponent to the chromatic background. Black, responses to stimuli on gray background; gray, responses to stimuli on chromatic background. Background contrast was 0.10. Data are means of 61 cells that had been tested with stimuli of different contrasts. For chromatic backgrounds, responses were reduced. The response for stimuli with a contrast of 0.15 (vertical dashed line) on a chromatic background corresponded to the response to a stimulus with lower contrast on a neutral background (dotted lines). Conversely, stimulus contrast needs to be increased on chromatic backgrounds to elicit the same response as on a neutral background.

Figure 6. Responses Are Influenced by Remote Color Fields

Response histograms for a neuron tested with stimuli and three remote color fields of the same chromaticity. Remote fields were 2° squares at a lateral distance of 5° from the receptive field, presented during the 500 ms stimulus period. (A and B) Control conditions without stimulus. (C and D) Responses to stimulus. (D) When remote fields are presented in addition to the stimulus, the response is slightly reduced. Insets show stimulus configurations (circle, receptive field; cross, fixation spot; display elements are not drawn to scale).

Figure 7. Influence of Remote Color Fields

Distributions of response ratios for stimuli with and without remote fields. (A) Remote fields had same chromaticity as stimuli. Median: 0.88. Black data points with crosses indicate corresponding values for psychophysical data of eight subjects (median: 0.92). (B) Remote fields and stimulus had opponent chromaticities. Median: 1.02. Note that in only a subset of experiments two remote field colors were used. The different results for the two conditions indicates that the influence of remote fields is color specific.

Figure 8. Estimation of Stimulus Color from Population Response

(A) Perceptual effect of background color. The rows of color patches marked by asterisks are physically identical, but they are displayed on different backgrounds and therefore look different. For example, patch (b) looks more similar to the physically different patch (a) than to the physically identical patch (c). (B) Four of the tuning curves of Figure 3, rotated in color space such that the respective directions of the background chromaticities (gray circles) are aligned, to mimic the background conditions in (A). Responses to patches (a)–(c) were estimated from the data points on the appropriate tuning curves in the 0° and 45° directions. (C) Estimated responses to patches (a)–(c) for the four neurons in (B). Responses are scaled relative to maximum firing rate for each neuron individually. The response pattern elicited by patch (b) is more similar to that for patch (a) than to that for (c), corresponding to the perceptual situation. (D) Induced color shifts estimated from 94 tuning curve pairs. For responses to six stimulus chromaticities (open dots) on bluish background (bluish circle), those chromaticities (black dots) were determined that yielded the most similar responses when presented on a gray background. Responses were considered as 94-dimensional vectors and similarity was measured by Euclidean distance. The dark purple curve shows, for the 90° (+S) stimulus on bluish background, the response vector distance as function of chromatic direction. The minimum occurs at a direction that is shifted from the 90° direction, away from the background direction. The dashed dark purple circle denotes the minimum distance, for easier visual inspection of the distance curve. For the other directions tested, similar shifts were obtained. Thus, qualitatively, the background has the same inducing effect as in perception.