Stages for extracting colour information: How the brain processes colour

Abstract

Colour information is processed by many stages in the visual system. Primate colour vision relies on three photoreceptors, cones, which sample visible light and send signals to the second stage, cone–opponent units. Surprisingly this stage determines not only the threshold detection for chromatic patches, but also matching surface colours under various illuminations. Hue discrimination at detection thresholds reveals contribution of the third stage, colour-opponency, which determines colour categories, or unique hues. These hues remain constant for para-central, peripheral field locations, providing the reference for veridical vision. However, more challenging tasks of colour identification require contributions from higher colour centres (V4–complex, TEO, IT).The aim of this study is to present key experiments illustrating the contributions of the cone opponent stage and unique hues in low-level colour vision, whereas the effect of damage to various stages will be discussed in terms of the distributed representations in the visual system.The experiments showed that the colour apparatus uses signals derived from the entire visual field (i.e. panoramic viewing), perhaps using some form of spatial and temporal integration. This observation is supported by the fact that when a matching paradigm is used to compare peripheral and central colour samples, peripherally invariant hues correspond to unique hues. And, finally, there is a trade-off between sensitivity and selectivity, and it seems likely that neurons form flexible assemblies which can act together to extract either hue, giving fine hue discrimination but poor sensitivity, or extract contrast giving high chromatic sensitivity but having poor hue discrimination.