Color vision, and seeing violet

[jostpuur]

Visible light spectrum: http://upload.wikimedia.org/wikipedia/en/7/79/Spectrum4websiteEval.svg

Wikipedia's article on color vision: http://en.wikipedia.org/wiki/Color_vision

Response intensities for cone cells S,M,L: http://upload.wikimedia.org/wikipedia/en/1/1e/Cones_SMJ2_E.svg

The way how a mixture of blue and green looks cyan, and how a mixture of green and red looks yellow, makes sense:

The cyan wave lengths stimulate both S and M cells, so the eye cannot tell the difference between cyan, and a mixture of blue and green.

The yellow wave lengths stimulate both M and L cells, so the eye cannot tell the difference between yellow, and a mixture of green and red.

However, the way how mixture of blue and red looks violet doesn't make sense equally, because the violet is not in between blue and red in the spectrum. The only way I could explain that blue and red appear violet, is that the L cell must have two response intensity peaks. One at centered at the red wave lengths, and other centered at the violet wave lengths. I don't see anything about this in the Wikipedia's article, but curiously, my guess is not in contradiction with the diagram of the response intesities: The red curve stops before it reaches the left corner! What would happen, if the curve was drawn all the way to the end of the visible spectrum? Would the red curve make a new peak there?

It looks like the person who has created the diagram must know that there is something strange happening with the red curve in the left corner, because otherwise he or she would not have been able to avoid making mistake, by avoiding to draw the curve all the way to the left, so carefully.

I don't have other sources at the hand right at the moment. But I remember seeing a same kind of figure in some book too: The red curve is not drawn completely.

The Wikipedia states:

violet light stimulates almost exclusively S-cones.

What does the "almost" mean there? If this claim is true, then why does the combination of red and blue show as violet to the human eye? And why isn't the meaning of the word "almost" made more explicit in the figure?

 

[Kalirren]

So I read up some on Wikipedia, and I came up with a partial explanation. Violet and purple are not the same thing.

Violet is a "spectral color," meaning that there is monochromatic light of a certain wavelength that looks violet.

"Purple," on the other hand, corresponds to no single wavelength. A stimulus that would register as purple can only be constructed via the combination of at least two spectral colors. The perception that purple lies between red and violet and connects them is an illusion created by the workings of the visual processing downstream of the photoreceptors in the retina.

If that explanation made no sense, I apologize: have no fear, just follow the link to where I pulled it from:

http://en.wikipedia.org/wiki/Purple#Purple_versus_violet

As for your point about the pigments, I don't believe the red pigment has another peak. The rhodopsin pigments are all based upon a cis-trans isomerization. That's a transition between two possible states, corresponding to one peak wavelength of energy. Wikipedia's "almost" just means that the absorption of violet light by the L pigment is not zero - indeed, that would be impossible, since the only way to get an absorbance of 0 at any frequency is to not have an electric field, which any molecule does.

 

[jostpuur]

This terminology convention with violet and purple does not seem to be relevant for the problem itself.

Emphasizing this

Violet cannot be reproduced by a Red-Green-Blue (RGB) color system, and must be simulated by a mixture of red and blue (purple).

(from Wikipedia)

is not relevant for the problem, because you cannot produce yellow light (light with the wave length corresponding to yellow) by combining red, green and blue lights either!

The real problem is that why does combination of blue and red appear to be violet/purple to human eye. The similar question for the yellow is not so puzzling.

 

[Kalirren]

So I looked up some papers which had better diagrams, and you're absolutely right, the L pigment -does- have another peak inside the visible. Both the L and the M secondary peaks are right underneath the primary peak of the S, and the S pigment has a secondary peak even further into the shorter wavelengths:

www.bg.ic.ac.uk/research/vision/Hiroshi/MomijiH06_Fig3.jpg[/URL]

So I was wrong in assuming that the cis-trans isomerization mechanism reduced the pigment to a single absorption peak.

That said, this still doesn't answer the question of why red and blue looks purple. (If the pigments' secondary peaks were the explanation, then blue and red alone wouldn't look purple because it would be missing the stimulation of the M pigment which also has a secondary peak under the S primary peak; blue and red -and green- would look purple!)

--

It seems to me that you are still confusing the color space (a 4-dimensional space consisting of all possible perceived colors defined by the responses of the S, M, L, and K (black rod pigment) pigments) with the color spectrum (a 1-dimensional space consisting of all spectral colors defined by wavelength.) The reason why I speculate this is that you seem to consider the perception of a yellow color (in the color space) created by the mixture of red and green spectral light as somehow more intuitive than the perception of a purple color caused by the mixture of red and blue spectral light, just because yellow lies between red and green but violet does not lie between red and blue (in the color spectrum).

This intuitive claim breaks down when you consider that the association of even the spectral colors with their perceived colors is fundamentally arbitrary.

I think this diagram would be helpful: [url]http://en.wikipedia.org/wiki/Image:CIExy1931.svg[/url]

Among the information contained in this diagram is the mapping of the visible light spectrum onto what essentially is a neurologically defined contour in the xyz-space of the diagram. As you can see, the contour of the spectral colors is not closed. Closing the contour with a straight line gives what is called the line of purples, and this contains all of the most saturated colors that are purple.

Now if you continue to ask about the neurological basis for the mapping diagrammed in that link, I have to admit ignorance. I'm not sure the field as a whole knows, either. It seems like a nontrivial problem.

 

[jostpuur]

www.bg.ic.ac.uk/research/vision/Hiroshi/MomijiH06_Fig3.jpg[/URL]
[/QUOTE]

This graph makes me happy :smile:

The second peak with the green graph is surprising though... And the over all shapes of the graphs are not actually agreeing with the previous graphs found in the Wikipedia. It would be interesting to see quantitatively how much the relative intensity differences can vary between different humans. And to learn how, where, and by whom these graphs are measured.

[QUOTE]
That said, this still doesn't answer the question of why red and blue looks purple. (If the pigments' secondary peaks were the explanation, then blue and red alone wouldn't look purple because it would be missing the stimulation of the M pigment which also has a secondary peak under the S primary peak; blue and red -and green- would look purple!)
[/QUOTE]

Clearly there is more details to be worked out, but already the existence of the secondary peaks makes the entire situation less paradoxical.

[QUOTE]
It seems to me that you are still confusing the color space (a 4-dimensional space consisting of all possible perceived colors defined by the responses of the S, M, L, and K (black rod pigment) pigments) with the color spectrum (a 1-dimensional space consisting of all spectral colors defined by wavelength.)
[/QUOTE]

This is surely not the case. Must be other kind of misunderstanding in communication.

[QUOTE]
The reason why I speculate this is that you seem to consider the perception of a yellow color (in the color space) created by the mixture of red and green spectral light as somehow more intuitive than the perception of a purple color caused by the mixture of red and blue spectral light, just because yellow lies between red and green but violet does not lie between red and blue (in the color spectrum).
[/QUOTE]

Actually not only intuitive, but explainable. This is how it goes: Because the yellow light is in between green and red, and because the intensity peaks of M and S cells are not too sharp, the yellow light stimulates both M and S cells, and this is what we have learned to see as yellow. It then follows that since the mixture of red and green lights also stimulate both M and S cells, the human eye cannot tell the difference between yellow light, and the mixture of red and green lights.

Do we agree that I understood this correctly?

If the secondary peaks did not exist at the violet end of the spectrum, then the same explanation could not be applied to explain why mixture of red and blue appears violet, so indeed the yellow light being in between the red and green was relevant.

 

[jostpuur]

That said, this still doesn't answer the question of why red and blue looks purple. (If the pigments' secondary peaks were the explanation, then blue and red alone wouldn't look purple because it would be missing the stimulation of the M pigment which also has a secondary peak under the S primary peak; blue and red -and green- would look purple!)

Qualitatively, I might immediately note that it is impossible to shine red light on the eye without, at the same time, stimulating the M cell (which is supposed to detect mainly green light). To me it seems that this is the key to solving this problem.

 

[Klockan3]

I thought a bit about this, since it is for example impossible to stimulate the green cells without doing the red one at the same time wouldn't we be able to produce another colour by eliminating the red ones response through other means?

Like me for example, I am colour blind in such a way that I lack the red cones, deep red looks black to me and computer red looks dark green, but that should also mean that my green is not the same as your green but instead be pure green.

 

[bballee]

How much variation from person to person is there in the ability to see the outlying wavelengths of the range of human vision? Can some people see into the ultraviolet or infrared ranges?