Color Processing, Cortical
The transformation of color signals and chromatic properties of receptive fields within the visual cortex of primates.
Processing of Chromatic Signals in the Early Visual Pathways
The processing of chromatic signals in the retina and lateral geniculate nucleus (LGN) has been the focus of numerous studies and is well understood. Less is known about the fate of color signals in the cortex. This entry first reviews central aspects of color processing in the primary visual cortex (V1) and discusses how it differs from subcortical processes. It then discusses the processing of color signals in extrastriate visual areas.
Color in the Striate Cortex (V1)
Chromatic Properties of Individual Neurons
In the primate primary visual cortex, it had been estimated that about 50 % of the cell population is selective for color . Estimates of the proportion of color-selective cells, however, are complicated by the use of different criteria for the classification across studies. Interestingly, color signals in primary visual cortex have long been thought to be relatively weak relative to black and white stimuli. Studies based on imaging techniques of the active human brain, however, have now clearly demonstrated that primary visual cortex contains a large proportion of color-responsive and color-selective cells  The majority of color-selective cells in V1, like those of its subcortical input, the lateral geniculate nucleus (LGN), simply add or subtract their chromatic inputs. Indeed, most V1  cells’ responses to chromatic modulations are well accounted for by a model postulating a linear combination of the signals derived from the three cone classes. Although there are some V1 cells that are more selective for color than predicted , the model adequately fits the responses of the majority of V1 cells.
The preferred colors of V1 neurons , however, do not cluster around the three cardinal directions found in the LGN . Instead, cells often prefer colors that lie intermediate to these “cardinal” directions. For cortical cells, the classification of chromatic cells into red-green or blue-yellow opponent cells is therefore not valid anymore.
Moreover, most color-selective cells in V1  respond also vigorously to luminance variations, a property that seems ubiquitous in the visual cortex. It also illustrates the fact that color selectivity does not imply color opponency. A response to stimuli containing chromatic but no luminance information (isoluminant stimuli) does not imply that the neuron receives opponent inputs from two or three cone classes. True color opponency can be deduced in neurons that give stronger responses to isoluminant than to luminance stimuli, provided that the stimuli are equated for cone contrasts . It thus appears that in the cortex, there is a whole continuum of cells, ranging from strict color opponency to strict luminance .
Recent studies have also shown that, unlike earlier levels, primary visual cortex contains so-called double-opponent cells . Double-opponent cells are cells whose receptive field combines color and spatial opponency. The defining functional property of these cells is that they respond strongly to color patterns but weakly or not at all to uniform (full-field) color modulations. The existence of pure double-opponent cells, cells with non-oriented receptive fields that respond only to color patterns , has been hotly debated in the last years. While early reports have been contested, it now appears that double opponency does exist in primary visual cortex but linked with other functional properties such as orientation selectivity and a high sensitivity for luminance stimuli as well .
Clustering and Specialization of Color-Selective Cells in V1
In the last decades, the localization and functional specialization of color-selective cells within area V1 have been the matter of debate. A number of studies reported that color-selective cells in V1 are not orientation selective (a canonical property of virtually all primate V1 neurons) and that these cells are clustered within the cytochrome oxidase (CO)-rich blobs that have been described previously . Both claims, however, have been contested by a number of anatomical and physiological studies . Some of the controversy can be attributed to the different techniques and stimuli that have been used to localize and characterize color-selective cells. A number of extensive and careful studies using single-cell recordings combined with histological processing have clearly shown that color-selective cells are not restricted to the CO blobs.
The same studies have shown that color selectivity and orientation selectivity are not, as originally proposed, segregated within the V1 cell population. The most convincing, but not unique, evidence for conjoint selectivity to color and orientation comes also from single-cell recordings .
The Case of Blue-Yellow Signals
For a number of mostly technical reasons, many color-vision studies have focused primarily on red-green opponency and less on the evolutionary older blue-yellow system. The discovery of the koniocellular layers in the LGN and their predominant role in the processing of blue cone signals (necessary for blue-yellow opponency)  raised the possibility of a specialized treatment of blue cone signals within the primary visual cortex. To what extent blue cone signals are clustered within V1 and how much they contribute to cortical color processing is not clear yet. Recent studies seem to indicate that blue cone signals are distributed uniformly within V1 (no clustering) and that these signals might be combined with achromatic signals in the double-opponent cells described above.
Color in the Extrastriate Cortex
Although several studies showed that individual neurons in the dorsal visual pathway, in particular in area MT of the macaque monkey, can significantly respond to chromatic variations , these responses are typically smaller than those obtained with luminance stimuli and do not account for the animal’s behavioral performance. The following sections are thus restricted to areas of the ventral pathway that are known to play a critical role in color processing.
Proportion of Color-Selective Cells
In extrastriate areas of the ventral pathway, the number of neurons whose responses are affected by color variations remains surprisingly constant despite the variability of the criteria used to classify neurons. This proportion reaches 50 % in area V2 and 54 % in V3 . Estimates in later areas of the ventral pathways (reviewed in 8) are more variable. In area V4, often but mistakenly considered a “color” area of the primate brain, original estimates ranged from less than 20–100 %. A more recent estimate of 66 % has been reported. In the IT cortex, it has been estimated around 48 %.
Chromatic Properties of Individual Neurons
As already described in V1 neurons, the chromatic properties of extrastriate neurons differ from those in the retina and LGN in two important ways. First, a significant proportion of neurons in each area possess a high degree of color selectivity. These neurons show a narrow tuning in color space. Narrowly tuned neurons have been reported, in different proportions, within areas V2, V3, and V4. The second major difference already described in V1 concerns individual neurons’ preferred colors. As in V1, the distribution of preferred colors of extrastriate neurons is not clustered in color space but is uniformly distributed .
Color Versus Other Visual Attributes
The question of conjoint selectivity for color and other visual attributes has been posed and answered for V1 neurons (see above). Similarly, the question whether color and other visual attributes such as orientation, motion, and size are segregated in the extrastriate cortex has been studied. Numerous studies (see ) in the last decade have shown that cells within V2, V3, and V4 can be concomitantly tuned for several dimensions of the visual stimulus. It thus appears that color is not processed independently but by the same neuronal populations that also code orientation or size. Note however that a recent study reported the existence of a significant subpopulation of V4 cells that respond to chromatic but not luminance variations leading the authors to suggest that color and luminance might be treated by different channels within area V4 .
Clustering of Color-Selective Cells
The debate concerning the organization of color-selective cells into clusters within a given area has extended to several visual areas beyond V1. In V2, the thin bands defined by CO staining have been reported to represent clusters of color-selective cells and to be the source of the color signals sent to area V4 and areas of the inferotemporal cortex. Moreover, optical imaging studies of area V2 concluded that color is represented in an orderly fashion within the thin stripes in the form of well-defined color maps that resemble those based on human color perception. As in V1, however, the clustering of color selectivity within the thin CO stripes has been challenged.
In areas more posterior within the temporal pathway, color-selective cells have been reported to cluster into subregions. This clustering is in fact often offered as an explanation for the large variance in estimates of the proportion of color-selective cells in the temporal cortex. Microelectrode recordings in dorsal V4 are thought to encounter many color clusters that seem to be much less prevalent in ventral parts of V4.
In the most anterior parts of the ventral pathways, the distinction between cortical areas is much less clear, and the relationship between areas reported in the macaque brain with those of the human is still controversial. It has nonetheless been suggested that color might be treated within a specialized pathway that extends across several of the ventral visual areas including V2, V4, and the dorsal portion of the posterior inferotemporal cortex (PITC) . Within PITC, color-selective cells would be clustered into islands themselves containing orderly, columnar color maps, reminiscent of the organization previously reported in V2.
Is There a Color Center in the Primate Brain?
Several of the issues discussed above are intimately related to the question whether the primate brain contains one color center, a cortical area whose main function would be to support most or all aspects of color perception. The notion of a color center is a natural consequence for the proponents of a strictly modular view of cortical organization. Most researchers agree that there are areas in the ventral cortex that are highly activated by color stimuli, but none of these areas have been established as predominant over the others.
Moreover, experimental lesions in macaque V4, the most popular candidate for the role of color center, do not result in a complete loss of color vision. These results cast doubts on the notion of a unique color center and support the idea that color, like many other visual attributes, is treated within a network of neuronal populations distributed within the ventral occipitotemporal pathway.
Important perceptual phenomena such as color constancy, unique hues, or color categorization remain largely unexplained. Further studies reconciling imaging or lesion results in the human brain with those obtained by methods revealing single activity in the macaque brain are thus necessary to fill these gaps. Moreover, the relationship between color signals and those associated with other visual attributes such as object shape or motion needs further scrutiny, particularly in extrastriate areas. Today, it is safe to say that a full understanding at the neuronal level of perceptual phenomena associated with color is still eluding the vision science community.
- 3.Lennie, P., Krauskopf, J., Sclar, G.: Chromatic mechanisms in striate cortex of macaque. J. Neurosci. 10(2), 649–669 (1990)Google Scholar
- 11.Leventhal, A.G., Thompson, K.G., Liu, D., Zhou, Y., Ault, S.J.: Concomitant sensitivity to orientation, direction, and color of cells in layers 2, 3, and 4 of monkey striate cortex. J. Neurosci. 15(3), 1808–1818 (1995)Google Scholar