Friday, February 19, 2010

Is green a composite hue, modulated by blue-yellow opponent cells?

In the end of 2009, I had the opportunity to re-visit current models of human visual perception investigating color perception as a function of spatial frequency. Based on my engineering background, a simple sinusoidal sweep pattern modulated by different opponent colors appeared to be appropriate to examine some interesting observations.

Figure 1. 4 sinusoidal sweep patterns showing spatial bandwidth of cone bipolar cells; (leftmost) black-white demonstrating highest bandwidth without hue changes (achromatic), (center left) blue-yellow demonstrating lowest bandwidth with hue shift toward black-white, (rightmost) red-cyan demonstrating medium bandwidth with hue shift toward black-white, and (center right) magenta-green demonstrating bandwidth similar to blue-yellow with hue shift toward red-cyan.

The achromatic black-white sweep pattern (Fig. 1) does not change hue across spatial frequency. It is therefore the most stable stimulus with regard to color decoding carried out by the human visual system (HVS) using cone bipolar cells. On the other end, there is the chromatic blue-yellow cone bipolar cell system that demonstrates the lowest spatial bandwidth and changes its hue toward black-white with increasing spatial frequency. These are the two ‘pillars’ of opponent color theory widely accepted in color vision research.

But the red-green cone bipolar cell system often reported as the remaining opponent color component should be looked at more carefully. In tri-stimulus color space, a red-green opponent color combination does not lead to neutral gray, rather it adds up to yellow. Instead, highly reasonable candidates are magenta-green or red-cyan. Interestingly, red-cyan shows higher spatial bandwidth than magenta-green.

If one takes a closer look, one also notices that magenta-green changes its hue toward red-cyan (not black-white) with increasing spatial frequency. When one compares the progression of assimilation between blue-yellow and black-white (observing contrast along the border between the two patches) with the progression of assimilation between magenta-green and red-cyan, they appear very similar.

However, similar to the blue-yellow cone bipolar cell system, only the red-cyan opponent color combination changes its hue directly toward black-white with increasing spatial frequency and can therefore be considered mathematically compatible with black-white and blue-yellow.

Based on these observations, I suggest the following:
  1. The first cone bipolar system is black-white showing highest spatial bandwidth and no hue change across spatial frequency (achromatic).

  2. The second cone bipolar system is blue-yellow showing lowest spatial bandwidth and a hue change toward black-white.

  3. The third cone bipolar system is red-cyan showing intermediate spatial bandwidth, higher than blue-yellow, lower than black-white, and also a hue change toward black-white.

  4. Magenta-green can only be perceived when the blue-yellow cone bipolar system has not exceeded its spatial band limit and is therefore mediated by blue-yellow.

  5. Within the HVS, green is most likely a composite color (hue) where red-cyan and blue-yellow bipolar cells must be able to simultaneously generate a significant signal (of inhibition?) in the visual pathway.

I would be very interested in receiving your feedback on this. Do you know of any counter examples? By the way, if you have difficulties assessing the perceived effects of the sweep patterns, consider viewing them in monocular vision (covering left or right eye to avoid influence on binocular vision) and perhaps vary viewing distance.

Thank you very much in advance for your valuable comments.

Some links for further reading: