Saturday, March 24, 2007

The inverted spectrum problem (part 3)

In the previous post, I had asserted that the colors perceived by humans when the actual paths corresponding to different cone cells could be exact. But I also think there might be cases of mild and severe color inversions where the above may not hold true.

Let me explain why. In humans, there are 3 kinds of cones which correspond to blue, green and red (roughly speaking). Actually, it is not that for each monochromatic light, only one type of cone gets activated. All three types can get activated, but each type of cone is more sensitive to some wavelengths than others. The portion of the brain responsible for color perception decodes the information from the three types of cones to determine the actual hue perceived by the consciousness.

Human vision is trichromatic, meaning that the color perceived is a point in 3 dimensional color space. Typically we take red, green, and blue to be orthogonal components and any hue can be represented by a point in this space. If we normalize the intensity, all hues are on the positive octant of the unit sphere with origin at 0.

Note that the wavelength to color mapping is not one to one. Purple is a combination of red and blue (at opposite ends of the spectrum), but a monochromatic light having an average of red and blue wavelengths appears green and not purple.

There is nothing in theory that requires vision to be trichromatic. In fact, many birds have four types of cones, making their vision possibly tetrachromatic. It is quite likely that they perceive hues which cannot even be imagined by us (that is, outside of the qualia space of humans). In fact, the "white" perceived by birds might be different from our white and something that we cannot even imagine.

Now, it is true indeed that for a variety of reasons we consider red, green, and blue to be the primary colors. This has got to do printing, and CRT and LCD screens. In a monitor, there are R, G and B pixels and other colors and intensities are got by illuminating these at various levels.

But in principle, if we consider the color space to be 3-D, ANY set of three mutually orthogonal directions would do. But the main question is, is there a preferred set of reference hues which lead to invariance between different humans to exact precision? In other words, are there eigenhues that have a simple mathematical relationship so that all humans perceive the same color along these axes exactly? (It may turn out that these eigenhues may not even be orthogonal on the color sphere, but I suspect they would. Also, I think there would be an overdetermination with respect to the white point, which I believe would also exact for two individuals.)

Now, I wish to stress that these eigenhue axes are intrinsic to the "projector" of the Cartesian theater. Actual stimulation of the retinas of two individuals corresponding to these eigenhue axes will not lead to the same results because of the variability in the cone responses from individual to individual. I am talking about something a lot more fundamental.

Now lets take the case of color blindness. There are many types of color blindnesses and most have to do with the front end (the cones themselves). In some cases, the brain itself might be involved (as in the case of achromatopsia), where the subject sees only in black and white.

There is also evidence of some women having a fourth type of cone which allows them to differentiate hues better than other humans. This condition has been mistakenly quoted as tetrachromaticity, but unless the brain itself is wired for 4-D color space as opposed to 3-D, one cannot call it that (and there is no reason to believe that these women have such a radically different brain organization). They still would perceive the same range of hues as other normal humans, although the mapping to the actual visual spectral composition might be slightly different. (Actually, to tell the truth, I myself have not come across any content on the web establishing that birds indeed do have a 4-D color space, but I suspect this to be the case.)

So, we have seen that mix-ups can happen at the front end (the rods and cones), and the processing unit (the brain). What about the back end (the Cartesian projector)?

Now this question sure leads us to Cartesian dualism and the mind-body problem! But that doesn't mean we should get scared away and not attempt at conjecturing such conditions. We do know that our subjective experiences do depend on physical phenomena in the brain. After all, drugs like LSD can alter visual qualia and so no one can deny that there is a physical basis for our subjective experiences.

I think the color pathways offer an unique opportunity for a lot of such experiments compared to other senses like sound, touch, smell, etc. The reason is, assuming a 3-D color space for humans, there should be a great deal of identical brain processing for the three pathways except at the end of the chain, namely the colors of the lenses of the Cartesian projector. Note that these colors would be the eigenhues which I discussed earlier. An analogy can be found in the case of real life component video amplifiers. The component signals (R, G and B) undergo the same kind of processing in any cable chain and the amps can be exchanged as long as the inputs and outputs still correspond to one another.

Now we know why the front end (the cones) have different color sensitivity profiles. It has nothing to do with the mind-body problem and is in the purely physical realm. It has to do with the pigments photopsins and rhodopsins in the cone cells.

But what about the color lenses of the Cartesian projector? Now this is a totally different ball game! But I believe that the color lenses are due to simpler molecules (than photopsins and rhodopsins) and the quantum-mechanical wavefunctions of the excited states of these molecules correspond to the color perceived in qualia space. Well, I have not attempted to describe what QM has to do with perceived qualia, but I am certainly not Pinker to dismiss the consciousness-QM connection here. But at the same time, I am not sure if Penrose's argument is right here either.

So what does this imply? For one, I feel that there is a very restricted family of these molecules able to "project" on to the Cartesian screen. There should be three such in humans with possibly a fourth in the case of birds and other lower forms of life. Since I feel that these molecules aren't even proteins (which might have different allele-variants), but something much simpler, I feel that the hues of these would be the same for any two humans, because their eigenstates would be the same. These molecules would fail in their function with even the slightest mutation, and it is possible that some cases of brain achromatopsia might have to do with mutant forms of these molecules, rather than the wiring in the brain itself.

If my theory is right, this provides a new twist to the color inversion problem! For if, for the moment, we assume that the brain pathways themselves are the same for the three eigenhues, what would happen if these molecules get switched? Then we could indeed expect some people to be born with the classic "inverted spectrum" condition! I call this the "severe color inversion syndrome", since the subject indeed sees red different from a normal person. (Actually there would be many forms of this depending on which eigenhues get switched, but I lump them together for now. On top of that, there might also be color anomalies resulting from cone defects, but I do not want to discuss such complications which are unnecessary to the basic understanding.)

Now what happens if both the inputs and the outputs of the amps (read brain pathways) get switched together? If the amps are identical, it shouldn't matter. But I feel there might be minor differences in the amps themselves. I would term this condition "mild color inversion syndrome". This condition, if it exists, should not be discernible as easily as the severe one, even with subjective tests.

I think that most inversion syndrome conditions (if they indeed exist) would be the mild type. This is because, I think, any mutation that causes the eigenhue molecules to get switched also strongly causes the amp inputs to get switched. Perhaps because there might be a shared protein in the synthesis of both the eigenhue molecule and the "input selector". Evolution would have favored such a scenario.

I also think it is possible for a much rarer condition (which I term "alien color inversion syndrome") in which one of the normal eigenhue molecules in the human is replaced by the fourth one present in birds. I don't know of the possibility of this. But in that case, the colors perceived by someone from this condition would have no correspondence with that of normal people!