This site is not maintained. Click here for the new website of Richard Dawkins.


← Sky-blue-pink. A colour never before seen?

ARL's Avatar Jump to comment 16 by ARL

I sort of doubt that unexperienced subjective colour labels reside in some nativist fashion in the brains of the colour-blind. After all, red-green colour blindness doesn’t literally mean not seeing red or green, or that you confuse the two, but that you have a two-dimensional colour-space rather than a three dimensional one. “Red” isn’t coded for by the firing of the long-wave cone, but by the distribution of activations across all cones (after the brain has taken account of relative brightness of surfaces at V4 to filter out constant illuminance). The long-wave cone is just maximally active when red is seen. So the colour-blind just have a flatter colour experience overall, and find it more difficult to discriminate colours because there are only two axes of comparison.

Remember that, from rather cruel experiments with cats, covering up a perfectly good eye from birth will cause neuroplastic changes such that the axons which would have referred to receptive fields at the covered-up retina are diverted for use at the uncovered retina. Once this has occurred, it is irreversible. Uncover the eye, and it will never see. Use it or lose it.

Also remember that newborns have a relatively underdeveloped colour experience, seemingly only able to discriminate colours close to the peak sensitivites of the three cones. More acute discrimination develops as the nervous systems develops. Natural selection can rely on certain inevitable experiences to drive certain aspects of neuro-development, only building in what it needs to.

Now, I would suggest that colour qualia are generated by the brain by relative comparison. It needs to discriminate and use information from opponent cells where retinal axons synapse, and there is the requirement that it make available to ‘consciousness’ the subjective sense that these opponent labels are maximally contrastive, and sufficiently different from one another (R-G, B-Y, Dark-Light). The particular sensitivity ranges of human photoreceptors would have been selected for by ancestral discriminatory demands, but I see no reason why selection would need to “build” in the labels for them. I would further suggest that colour develops in a (guided) neuroplastic process that expects certain inputs to be there. When there is partial poverty of input from the photoreceptors, the discrimination demands placed on it are lessened, and so the brain never needs to ‘make’ the colour.

Reflect on the fact that colour-blind people have a flatter colour experience, not because they can’t see a particular colour, but because they have reduced discrimination because they can’t ‘triangulate’ and thereby precisify. If all the colour labels were already in the brain, we might suppose that they didn’t have a flatter colour experience, for why should they? They still have M and L cones, but their sensitivity curves are too close for relative comparison. Why wouldn’t colours that were hard to discriminate switch between two distinct colour labels like Necker Cubes? If the labels were already there, then the only problem for the colour-blind brain would be ambiguity – that there are two equally possible reflectance profiles for this one surface. When this happens for 3D cubes, where there are no helpful contextual cues, the visual system happily flips between both models.

I suggest that this is what would happen in the colour blind if all the trichromatic spectrum labels were available to thei brain. Instead, certain colours are as indistinguishable to them as metamers are to trichromats (e.g. a sensation of “yellow” can be caused by equal amounts of ‘red’ and ‘green’, or by ‘yellow’ light, and we can’t tell the difference). If we had an fourth cone which peaked at ‘yellow’, we COULD differentiate these two ‘yellows,’ and our brains would have needed to develop a fourth axis to our subjective colour-space. But the R-G colour blind only have two points of comparison, and only need two axes. You would have to rewire complex systems of connections throughout the visual pathways to give them a third. Or perhaps, if you could fix the problem at the retina, the brain’s plasticity would happily do the job for you, as it compensates under so many other pressures. But I don’t think the colour label would be ready and waiting for the job.

Just a (long) thought.


Sat, 12 Jun 2010 16:32:17 UTC | #479646