It has been shown that some dichromats can categorize colors using color names in quite similar manner to trichromats. The previous studies suggested some possible mechanisms to account for this trichromat-like categorical color naming ability, such as rod contribution, anomalous cone pigments, a nonlinear parallel chromatic channel or luminance cue. However none of them has not been able to fully explain the categorical color naming of dichromats. In this paper we tried to find out what observing conditions of color stimuli mostly affect the color categorization of dichromats to explore this naming mechanism. Dichromats and trichromats participated in the categorical color naming experiments, in which they named 424 OSA uniform color samples twice using only the Berlin and Kay’s eleven basic color terms under various chromatic illuminants. They also observed OSA color samples, simulated on a CRT monitor, in various observing conditions, i.e., a small visual angle, a short duration, equal luminance and blurred edge. A trichromat could use the same color names for the same stimuli in almost all observing conditions whereas a dichromat showed some confusion of color names along the green-red direction under some illuminants and for some CRT stimuli. This confusion was greater with a small visual angle and with equal luminance. Individual differences were much greater for dichromats than for trichromats. Our results suggest that dichromats might not have the same color category mechanism, but have different mechanisms that determine their degree of deficiency. These mechanisms must yield barely visible differences, not enough to be used for color discrimination, but enough for color categorization.

Many studies have suggested that color and luminance are separately mediated by different processing streams in the early visual system and that the two information cues play different roles in the perception of motion, depth and shape. One major suggestion in these studies is that luminance information is the main contributor to motion and depth detection. Although color information was not considered as an important cue, recent studies have shown that color signals contribute to the detection of motion and depth as well. To sort out the relative contributions of both information cues and to shed light on the visual mechanism behind motion and depth detection in general, we have investigated how color information is processed in the presence of luminance information.

Coding of color in the retina and lateral geniculate nucleus is dominated by cone opponency, and selectivities cluster around the corresponding color space axes. In the visual cortex, a more distributed representation for color stimuli is found. Comparison of this population code for color with the coding of other visual features such as orientation indicates similar coding mechanisms for both features. Further similarities can be observed with respect to contextual interactions. The visual context affects processing and perception of both color and orientation, and results of psychophysical measurements indicate that the properties of these interactions in both domains are similar. These findings suggest that the visual cortex makes use of the same neural mechanisms for the processing of color as for other visual features.

We can summarize the transformation of color signals in the early stages as follows. First, color is represented by the relative activities of three types of cone photoreceptors. Secondly, linear combination of cone signals occurs in regular manners and color is represented by two axes, namely L–M and S axes (two-axes representation). Finally, signals tuned to various directions in the chromaticity diagram starts to be formed in V1 resulting in hue selective neurons as well as neurons selective for saturation. We can call this third stage as having “multi-axes representation of color”. Color representation based on hue and saturation seems a principle way ubiquitous across different areas in the cortical visual areas.

Categorization and fine discrimination are two distinct aspects of our visual perception, both of which are clearly apparent in color vision. We can switch between these two modes of perception depending on the situation or task demands. To explore how visual cortical neurons behave in such situations, we recorded the activities of color-selective neurons in the inferior temporal (IT) cortex of two monkeys trained to perform a color categorization task, a color discrimination task and a simple fixation task. Many IT neurons changed their activity depending upon the task. A majority of neurons showed stronger responses during the categorization task. Moreover, for the population of IT neurons as a whole, signals contributing to performing the categorization task were enhanced. The task difference could be explained by the change in the response gain, indeed, the color selectivity was well conserved. These results imply that judgment of color category by color-selective IT neurons is facilitated during the categorization task and suppressed during the discrimination task as a consequence of task-dependent modulation of their activities.

Color spaces based on cone-response opponency (including CIE system) are useful for photometry. However, several psychophysical and physiological studies have suggested that they do not represent color appearance. Color signals for color appearance should exist somewhere in between the opponent and categorical levels:‘mid-level’of color-vision system. In this text, we would like to introduce some of our attempt on investigating color information processing systems at the mid-level by using psychophysical method and functional brain imaging technique (fMRI).