How Do We Perceive Violet with Only Three Color Detectors in Our Eyes?

[resurgance2001]

If our eyes have only three types of color detector: one for red, blue and green, how do we perceive the color violet when we are looking at a spectrum?

If I am looking at the iphone or a TV or other device which effectively only emits combinations of pure red, blue and green light, I can see a violet color that has been produced by a combination of those primary colors, but when I see that violet at the end of spectrum, that is a pure violet that is made from simply one frequency, so how is it that the human eye is able to detect this pure frequency and the brain figure out it is violet and at the same time able to detect a mix of red blue and green and figure that out as violet?

I don't know if it is connected but I have noticed that when I connect my homemade spectroscope to various cameras, some cameras are able to 'see' and display the same violet that my eye sees, but some cameras are not able to and only display the color as blue.

Cheers

Peter

 

[Satwik]

Our eyes contain cells called cone cells , these are of different types .
Some are sensitive to red color , some to blue and some to green . When a we see a banana the yellow light reflecting from it reaches our eyes and that yellow light is perceived by the different cone cells and the information is sent to the brain to be processed which afterwards is interpreted as yellow .
In the same way voilet can be perceived though pure voilet color can not be seen very clearly through our eyes.
Hope this helps!

 

[FactChecker]

The eyes has 3 types of color sensitive cones and rods, which are not color sensitive. While the cones in our eyes respond primarily to high (blue), medium (green), and low (red) frequencies, they respond to other frequencies to a lesser amount. See http://hyperphysics.phy-astr.gsu.edu/hbase/vision/colcon.html#c1 for the curves of frequency response. For violet, there is low response from the blue and almost none from the green and red. So how does the brain know that it is just not a dim blue? The rods (see http://hyperphysics.phy-astr.gsu.edu/hbase/vision/rodcone.html#c4) in the eyes tell the brain that there is more light then the blue response indicates for blue. The brain can put it all together to know that the color must be lower than blue.

 

[my2cts]

If there is a signal on the blue receptor but none on the green and red, then the color must be violet.

 

[sophiecentaur]

The eyes has 3 types of color sensitive cones and rods, which are not color sensitive. While the cones in our eyes respond primarily to high (blue), medium (green), and low (red) frequencies, they respond to other frequencies to a lesser amount. See http://hyperphysics.phy-astr.gsu.edu/hbase/vision/colcon.html#c1 for the curves of frequency response. For violet, there is low response from the blue and almost none from the green and red. So how does the brain know that it is just not a dim blue? The rods (see http://hyperphysics.phy-astr.gsu.edu/hbase/vision/rodcone.html#c4) in the eyes tell the brain that there is more light then the blue response indicates for blue. The brain can put it all together to know that the color must be lower than blue.

All three sensors have sensitivity that stretches over the visible spectrum, albeit with low sensitivity 'out of band'. The brain uses all three signals and compares the relative amplitudes (the ratios of the signals), which allows it to fine tune its response over the range of each sensor. Without the other two sensors, all you would know would be the value of the luminance arriving. It's the overlap that is the clever bit. The response curves in this link show what I mean and, in particular, what happens at the blue end gives worse discrimination than anywhere else in the visible spectrum. The two longer wavelength sensors have pretty well run out by 'the blues' so working out their contributions is harder for the brain - producing uncertainty and even conflicting colour sensations e.g. 'spectral' violet and 'looks violet'.
@my2cts: you put it briefly but I wouldn't disagree with that concise statement.

 

[resurgance2001]

Thanks everyone for your replies but I am sorry I don't get it. Let's suppose that I have a spectrometer that gives the line spectrum for an energy saving light bulb.

I can shield over half the spectrum so that the only thing that is left are the two blue lines and one violet line. Now my eye is able to see the blue line as blue. And it is able to see the violet line as violet. There is no mixing of colours here because this is a line spectrum which has been produced by a diffraction grating. So my eye is able to see voilet light which and recognize it as voilet which would imply that the light sensitive cells in the eye are actually not being restricted to purely red, blue and green.

What is even more weird is that I can use a violet LED which is producing a light which is not a single pure voilet light but is composed of a mixture of different colors of light as demonstrated when I shone the light from the LED through the spectrometer, but my eye is still able to detect it as violet. In other words, mixing the correct combination of primary colours of light will produce a voilet color of light which the eye sees as voilet, or a single high frequency between blue and UV is also seen by the eye as pure voilet. So why is the eye able to do that?

 

[FactChecker]

would imply that the light sensitive cells in the eye are actually not being restricted to purely red, blue and green.

Yes. Several posts have said that.

In other words, mixing the correct combination of primary colours of light will produce a voilet color of light which the eye sees as voilet,

I think that is very possible. A proper mixture of colors could easily induce the cones to respond as though the light was a pure voilet

 

[my2cts]

I am not at all an expert in the biophysics of the eye, but my guess is the following.
The violet line leads to three readings on the red, green and blue cones.
For the violet line I expect a red signal of zero, a weak green signal and a much stronger blue signal.
From the ratio green/blue the color can be found. This assertion is unambiguous.
No two different sets of input can lead to the same conclusion.
That is how I would set it up it with three calibrated sensors with overlapping spectral response.

 

[sophiecentaur]

Thanks everyone for your replies but I am sorry I don't get it. Let's suppose that I have a spectrometer that gives the line spectrum for an energy saving light bulb.

I can shield over half the spectrum so that the only thing that is left are the two blue lines and one violet line. Now my eye is able to see the blue line as blue. And it is able to see the violet line as violet. There is no mixing of colours here because this is a line spectrum which has been produced by a diffraction grating. So my eye is able to see voilet light which and recognize it as voilet which would imply that the light sensitive cells in the eye are actually not being restricted to purely red, blue and green.

What is even more weird is that I can use a violet LED which is producing a light which is not a single pure voilet light but is composed of a mixture of different colors of light as demonstrated when I shone the light from the LED through the spectrometer, but my eye is still able to detect it as violet. In other words, mixing the correct combination of primary colours of light will produce a voilet color of light which the eye sees as voilet, or a single high frequency between blue and UV is also seen by the eye as pure voilet. So why is the eye able to do that?

That is the whole point of the system, as has been mentioned several times already, already.. All three sensors have some level of response at all wavelengths. Your brain looks at the combination of the three output levels and assesses the ratios between them. Have you looked at the response curves that are in the links, posted? Without looking at, or being familiar with them, none of this will mean anything. The blue -end response has a peak of sensitivity and drops off either side. A spectral line with wavelength a bit longer than that peak can have exactly the same effect as for a line that's a bit shorter wavelength. How do you distinguish, because just the one signal will only tell you how bright the object is (monochrome vision)? You also look at the other two sensor outputs. If the line is shorter wavelength than the peak, the other two responses will be much lower - so your brain can tell that the wavelength is shorter than its peak response. For a line of longer wavelength, the two other responses will be higher - so the brain will know that the wavelength is longer than the blue peak because it gets two bigger signals. The problem (if you want to call it that) with the Violet thing is that the response of the other two is so low in the violet direction that the brain has difficulty in working things out. It will confuse a spectral line with what you can produce with a combination of R G and B phosphors. (Look at the principles of colour TV) But, as far as the evolving hominid was concerned, does this matter? The answer must be NO, because, if it did matter, you can bet your life that we would have evolved to make this discrimination better. Not much worth knowing about in the blue region of the spectrum, obviously and Nature is more lazy than a second year student. We do not experience the same aliasing at the red end because out system needs to work better there.
There is a similar thing with hearing and judging musical 'notes' we have progressively worse and worse intonation discrimination as the frequencies get higher and higher. 'Why" would that be? Probably because it's outside the range of our voices and we just don't need to be so much aware of the pitch of high (and low) notes.
The question has been raised about how we get 'calibrated'. Our eyes are open all our waking hours and we see some colours very frequently (faces, blood and leaves). Our brain is constantly exercising itself to get the maximum information it can from scenes. Every time young humans opens their eyes, they are practising this - plus movement awareness and spatial awareness etc. etc..

 

[DrStupid]

For the violet line I expect a red signal of zero, a weak green signal and a much stronger blue signal.

Photopsin I has a minor absorption maximum around 400 nm and the resulting red signal should be slightly more intense then green (see http://upload.wikimedia.org/wikipedia/commons/6/65/Cone-response.svg ).

 

[sophiecentaur]

Does "Photospin 1" refer to a particluar graph? I can't find it. The graph in this link (quoted earlier) implies that the low level responses on the skirts of the curves are rather uncertain. That means it's anyone's guess as to what they really are. It's a matter of 'the proof of the pudding is in the eating', I think and the very fact that there is confusion / redundancy in the Violet region means that something funny happens in the violet region of the two other passbands. There doesn't seem to be much left to say about this unless someone can come up with a very reliable reference for the sensitivity curves. The curves all seem to have been derived in a roundabout way, using various colour matching techniques and, if our eyes are actually not reliable in the violet region, then neither are the inferred sensitivity curve values. (I think this is referred to as 'ill conditioned'??)There may not actually be a proper answer to this at all, until they can strap a probe on the optic nerve and look at the actual signals.

 

[DrStupid]

Does "Photospin 1" refer to a particluar graph?

Photospin I is the protein-pigment complex responsible for the color red and the red graph represents the corresponding spectra.

There doesn't seem to be much left to say about this unless someone can come up with a very reliable reference for the sensitivity curves.

The data from http://rspb.royalsocietypublishing.org/content/royprsb/220/1218/115.full.pdf (page 121) result in the same curves:

Is this reliable enough?

 

[sophiecentaur]

Is this reliable enough?

It looks good and better than my searches found. There is a difference between the picture p121 and the coloured one you posted, which shows a rising response at short wavelengths. I don't understand that. Yet another source, perhaps? There is some discussion in the paper about the reliability of results and it's obviously a bit speculative in the regions of low response of the curves. Still no actual signal level measurement though - a lot of inference needed, to arrive at answers.

It does seem to give a clue about the violet confusion thing. Spectral violet will give a low blue response, which is not much higher than that of the other two curves. It isn't too difficult to imagine getting the same colour sensation with some spectral blue ( for which the blue sensor gives a much higher reading) and a sniff of some R and G phosphors, to emulate what the three sensors would get (relatively) with spectral violet. But that goes for all colour synthesis systems. Perhaps it would be more fruitful to look at why we have such good colour discrimination of colours around the red end of the spectral curve on the CIE chart. It must be to do with the fact that the long and medium wavelength sensor characteristics are close enough together to improve resolution. On that graph of yours, the blue response drops right down at the green peak - which would give a good way of deciding which side of the peak of green that a colour falls. But, of course the perceived colour with an RGB display depends upon the levels of R,G and B inputs. It's pretty amazing that it's possible to synthesise colours so accurately with a simple triad of phosphors. Some very clever thinking in the early work, I reckon.

 

[DrStupid]

There is a difference between the picture p121 and the coloured one you posted, which shows a rising response at short wavelengths. I don't understand that.

My plots are based on the data in table 2. The graphs in figure 2 (based on the data in table 4) only show the main peaks. According to the description on page 122 this figure has been optimized in order to make the shapes of the spectra similar. That explains why the beginning second peak has been omitted. It seems the authors were interested in the main peaks only.

 

[sophiecentaur]

Thanks for the explanation. Have you any comments on my ideas about the violet confusion? I'm sure that it just shows up an inadequacy that is but in on a 'need to know' basis (as with most of our brain systems). Is there anything like an equivalent in the far red, I wonder?

 

[RobS232]

The eyes has 3 types of color sensitive cones and rods, which are not color sensitive. While the cones in our eyes respond primarily to high (blue), medium (green), and low (red) frequencies, they respond to other frequencies to a lesser amount. See http://hyperphysics.phy-astr.gsu.edu/hbase/vision/colcon.html#c1 for the curves of frequency response. For violet, there is low response from the blue and almost none from the green and red. So how does the brain know that it is just not a dim blue? The rods (see http://hyperphysics.phy-astr.gsu.edu/hbase/vision/rodcone.html#c4) in the eyes tell the brain that there is more light then the blue response indicates for blue. The brain can put it all together to know that the color must be lower than blue.

No.
The red sensing rods also detect light in the deep blue. A response that is both red and blue is seen as a combination of red and blue. Our eyes cannot differentiate between a mixture of two colors and deep blue. Both look violet.
http://midimagic.sgc-hosting.com/huvision.htm

 

[THE HARLEQUIN]

If I am looking at the iphone or a TV or other device which effectively only emits combinations of pure red, blue and green light, I can see a violet color that has been produced by a combination of those primary colors, but when I see that violet at the end of spectrum, that is a pure violet that is made from simply one frequency, so how is it that the human eye is able to detect this pure frequency and the brain figure out it is violet and at the same time able to detect a mix of red blue and green and figure that out as violet?

there's nothing called pure violet sir .. ... something that is colored violet is also a combination of primary colors

 

[RobS232]

there's nothing called pure violet sir .. ... something that is colored violet is also a combination of primary colors

The frequency range near 400 nm is defined as being "violet" . Therefore the color "violet" does exist, by definition. You are confusing this with the problem that the human eye has differentiating between some combinations of color. The eye responds in an identical manner to a two color source blue and red as to a single color source, pure 400 nm violet. But that is a problem with our eye, not the wavelength of light.

 

[THE HARLEQUIN]

The frequency range near 400 nm is defined as being "violet" . Therefore the color "violet" does exist, by definition. You are confusing this with the problem that the human eye has differentiating between some combinations of color. The eye responds in an identical manner to a two color source blue and red as to a single color source, pure 400 nm violet. But that is a problem with our eye, not the wavelength of light.

a violet beam makes the eye respond just the way a violet colored object makes it respond ... as far as i know

 

[FactChecker]

No.
The red sensing rods also detect light in the deep blue. A response that is both red and blue is seen as a combination of red and blue. Our eyes cannot differentiate between a mixture of two colors and deep blue. Both look violet.
http://midimagic.sgc-hosting.com/huvision.htm

Thanks. It's interesting that there is a "bump" of sensitivity that red cones have down in the far blue side, That reinforces the fact that the mixture of responses from the types of cones and rods is interpreted by the brain as a single color. Nothing is simple.

 

[RobS232]

a violet beam makes the eye respond just the way a violet colored object makes it respond ... as far as i know

That is true. There are other color combinations of light that lead to ambiguity in interpretation. There is also variation on the color response in the eye among human populations (color blindness is an extreme case) that lead to differences in interpretation. A great deal of research has been carried out on this topic by diverse industries - color printing, color film manufacturers, digital camera manufacturers, color display producers.

 

[abecedarian]

Thanks. It's interesting that there is a "bump" of sensitivity that red cones have down in the far blue side, That reinforces the fact that the mixture of responses from the types of cones and rods is interpreted by the brain as a single color. Nothing is simple.

Also, notice that the wavelength of violet light is close to being a 2nd harmonic of red light so the 'waves' will become increasingly coincident, but since the violet's frequency is also nearing twice that of red, sensitivity will be decreased.

 

[sophiecentaur]

The frequency range near 400 nm is defined as being "violet" . Therefore the color "violet" does exist, by definition. You are confusing this with the problem that the human eye has differentiating between some combinations of color. The eye responds in an identical manner to a two color source blue and red as to a single color source, pure 400 nm violet. But that is a problem with our eye, not the wavelength of light.

BE careful not to confuse 'wavelength' and 'colour'. Colour is the response / sensation that the eye and brain experience from a single or a combination of wavelengths. There are a million possible combinations of primaries that can produce a given colour sensation, depending on the actual primaries used.
I would also make the point that the Primary Phosphors, used (which are still the basis of the colorimetry used -afaik) are by no means monochromatic. (spectral) The phosphors were based on what could be done best with a CRT and the choices were made on the basis of producing bright enough primaries for a useful display - hence, fairly broad band sources. Since the introduction of LEDs, things may be different in practice but I have a feeling that history will still affect the present colorimetry in displays.

 

[FactChecker]

BE careful not to confuse 'wavelength' and 'colour'. Colour is the response / sensation that the eye and brain experience from a single or a combination of wavelengths.

As far as pure, single frequency colors go, you may have to take this up with Newton and all the other people who refer to Newton's "color theory". Trying to define colors by the response of the eye has required defining a "standard" eye, which changes from time to time -- a bad foundation for science. It's like defining temperature by how it feels to the "standard" finger. Extending the defined colors to non-pure mixtures should be done by referring to ratios of the single-frequency colors or to some weighting function.

 

[sophiecentaur]

You may have to take this up with Newton and all the other people who refer to Newton's "color theory". Trying to define colors by the response of the eye has required defining a "standard" eye, which changes from time to time -- a bad foundation for science. It's like defining temperature by how it feels to the "standard" finger. I hope this trend to take color back to the middle ages can be stopped.

Colour is the response of the eye. It's subjective. The eye is not a spectrometer - not a scientific instrument. It is as accurate as it needs to be for gaining an appreciation of the environment by analysing every combination of wavelengths it sees with just three colour receptor responses. Many animals do not need such colour discrimination and their vision is correspondingly simpler. If you want to be 'scientific' about the spectrum of visible light then you use a spectrometer and measure levels for small divisions of wavelength.
I wasn't aware that Newton was very well informed about how modern colorimetry works. Did he have knowledge of the tristumulus basis of colour vision?
If you don't want a "standard eye" then you would have to close down the colour TV display industry - which is built entirely on the concept. And it does pretty well for most critical viewers.

To pursue your idea of a 'standard finger' for assessing temperature then I would suggest the equivalent would be 'wind chill' which is a fuzzy idea based on the effect on a human body under some standard conditions, including temperature.

 

[FactChecker]

I wasn't aware that Newton was very well informed about how modern colorimetry works. Did he have knowledge of the tristumulus basis of colour vision?.

Well, Newton called them colors and he was there before you.

 

[bahamagreen]

The first reply mentioned yellow, which is different; yellow is neither a pure nor composite signal from the three types of cones' responses...

The signal for yellow is synthetically comprised at a higher later stage in visual processing. Yellow is the only color that is done differently like this in the human visual system, and somehow (...perhaps because its generation is "closer" to the later stages of processing? ...or perhaps it incurs a slight additional processing delay with respect to the signals from the cones?) this makes yellow a subtly more "interesting" color subjectively - generally a top choice for danger and warning signs, barricade and police line tape, road lane lines, fast food restaurant and consumer food packaging, classic Post-it notes, etc.

 

[sophiecentaur]

Well, Newton called them colors and he was there before you.

How is that relevant? Newton got many things wrong - we later found. He can be excused for confusing colour with wavelength because he had only a limited experience to work on. He probably never considered applying his theory to the colour Magenta.

 

[sophiecentaur]

The first reply mentioned yellow, which is different; yellow is neither a pure nor composite signal from the three types of cones' responses...

The signal for yellow is synthetically comprised at a higher later stage in visual processing. Yellow is the only color that is done differently like this in the human visual system, and somehow (...perhaps because its generation is "closer" to the later stages of processing? ...or perhaps it incurs a slight additional processing delay with respect to the signals from the cones?) this makes yellow a subtly more "interesting" color subjectively - generally a top choice for danger and warning signs, barricade and police line tape, road lane lines, fast food restaurant and consumer food packaging, classic Post-it notes, etc.

That's an interesting response. Are you using the term "yellow" as a general description for what are often referred to as 'skin tones'? Humans are very much fine tuned to discriminating between different skin tones because they are good signals of emotion. Do you have an evolutionary 'reason' (or advantage) for this special treatment of Yellow? There almost certainly has to be one, which warning signs are deliberately exploiting - snakes and wasps, perhaps?
I really think this thread should be discussing non-spectral colours more. After all, we never (not too strong a word, I would say) actually see spectral colours - certainly not in everyday life. You have to try really hard to find examples of totally saturated Yellow or any other 'rainbow' colours.

 

[bahamagreen]

Yellow might relate to current skin tone perception but I imagine lighter skin tones appeared long after the peculiar human processing for the color yellow.

Visual processing is one of those things that is just increasingly amazing the closer one looks; here are some other things to consider is this general discussion...

Some creatures have more than three types of cone response (fish, amphibians, reptiles, and birds have four) and some less (dogs only have two). The retinal density of some bird's cones is 25x that of the human eye, providing profound resolution.

In humans the red-green trichromatic channel is involved in the perception of both color and form. The blue-yellow channel is not involved in form, only color.

There are various types of "horizontal" layer processing in the retina (which has ten layers); one of these has two forms responding to color opponents red-green or blue-yellow, another codes only for brightness (no opponent color response between cones).

The processing streams of visual information split into four different functional / physical pathways in which form, color, "dynamic form" (location), and motion are each processed separately as attributes. There are five subsequent locations of processing in the streams for motion and location, six for color, and seven for form. That is to say, when you play with a tennis ball, the shape, color, location, and motion of the ball that you see as one thing are being individually processed in different regions of the brain, yet the appearance of these attributes is integrated so seamlessly well you can toss and catch the ball easily.

The "wiring" from the optic nerve to the rods and cones in the retina is not behind the retina but to the front of the retina, inside the eye, and these connections lie all over the retinal surface. The "blind spot" is where these connections pass through the retina to become the optic nerve behind the eye (in a sense, your retina is installed "backwards", but so is the image upside down and left right reversed... and the focal "plane" projected into the inside of a sphere...).

There is a complicated system that conspires to prevent eye movement (both using the muscles around the eyeball to rotate the eyes' orientation with respect to the head, and gross rotations of the head) from presenting a disorienting "whooshing by" of the visual field... this system integrates complexly with the suppression of motion detection processing to stabilize the apparent visual field.

The brain's surface (cortex) has ten physical layers of functional processing... as does the retina. Evolutionarily, it is as if the brain has managed to migrate a bit of its surface out to interface with the external world...!