What decides the colour of light?

[Don Jusko]

If it only reflected a very narrow band of colour then it would, unfortunately, look very dark so it can be a very difficult job.

That's not so unfortunate, it would be yellow's dark without the black (that's the way computers work, subtract light to make a darker color).
If you want a lighter color just add it's opposite color, blue light.
I'm a little curious as to just how dark of a yellow pure spectral yellow from a light source would be.

PY100 pigment from 1 light yellow to 9 dark brown.
Red to yellow darken to brown in the Real Color Wheel and element crystals.
Here is a link to show how dark spectral yellow is in pigment, plus the story.
http://www.realcolorwheel.com/colorwheel.htm

 

[DaveC49]

I always say that human colour perception is a really poor spectrometer. It is sooo easy to fool. And it doesn't matter at all.

Hi Sophie I don't think the operation of the human eye can be described as a spectrometer as such ( even a poor one) as the action of a spectrometer is to measure the intensity at a given wavelength whereas the brain ( not the eye) is inferring a color from the differential responses of three sets of chromo-receptors which each respond to different frequency/wavelength ranges which have some degree of overlap in the optical spectrum.

 

[sophiecentaur]

Hi Sophie I don't think the operation of the human eye can be described as a spectrometer as such ( even a poor one) as the action of a spectrometer is to measure the intensity at a given wavelength whereas the brain ( not the eye) is inferring a color from the differential responses of three sets of chromo-receptors which each respond to different frequency/wavelength ranges which have some degree of overlap in the optical spectrum.

As with all things 'evolutionary', the brain expends just enough effort on a problem to get by satisfactorily. A three filter analysis is good enough to distinguish between the spectra we see, reflected from most of the important objects in out lives. Skin can take on may different hues, depending on emotion, health and where we've come from. The tristimulus system does a great job there. Also, the range of colours from greenery / brownery is relevant to us and we do a great job there, too. No animal really cares about wavelength so we can't 'see' it. It was only when they started teaching kids about wavelength that this idea that 'wavelength = colour' and vice versa that any confusion arose.

 

[sophiecentaur]

If you want a lighter color just add it's opposite color, blue light.

You cannot "add blue light" in a subtractive colour mixing process (pigments or filters). If you want a more saturated colour with pigments, all you can do is subtract more light, making the surface darker. RGB colour synthesis is a lot easier to follow because the phosphors are fairly pure and bright. With CMY mixing, you can go more and more saturated but at the expense of brightness. The only way to produce a proper spectral yellow (i.e. a narrow band of wavelengths around the sodium yellow) is with a dielectric (interference) filter. A pigment can't do it - and it's particularly hard with a reflective surface. Insects and birds can have very saturated / bright colkours but not with pigments.
Those pictures of beakers of coloured water do not show spectral colours. How can they?
Printers and artists are pragmatic practitioners. They do the best they can to produce the colours they want. They do not claim to (or want to) produce spectral colours. I think there is an overlap in terminology which suggests some Physics that isn't really there in colour work.That doesn't matter at all - unless you try to equate the two fields of study.
That paper by Land, in an earlier post, is well worth reading in detail as it explains a lot about our perception of the colours of illuminated objects.

 

[Graeme M]

Sophiecentaur I will have to read that paper when I get a moment - I find the whole matter of colour intriguing, in particular that it is an entirely internal property or quality. Something I never realized before and which illustrates how little I actually know about this. So, a question from ignorance.

As I understand what's been said, the colour perceived is generated by the summing of responses from the cone cells which respond to the frequencies of incoming light. So I understand that colour is not a real physical property, rather light has frequency and wavelength and our brains work with frequency to generate a representational quality.

But I'm not clear about the matter of the shades or hues of colours. For example, the mention of say spectral yellow. I assume this term means 'pure' yellow? How have we derived (or more exactly, agreed upon) the values for those spectral colours? I would have thought that the frequencies of light at any particular point on the spectrum don't all necessarily sum exactly the same in all brains, so how do we know which specific frequencies are spectral yellow? Or does that not matter, it's just the statistically averaged perception of that colour?

Or am I just missing the point entirely?

 

[sophiecentaur]

I will have to read that paper when I get a moment

Yes - it is full of good stuff. Perhaps it's starting in on the subject a bit far along the road and you may find it 'challenging'.

. For example, the mention of say spectral yellow. I assume this term means 'pure'

When you meet up with a fresh term like 'spectral colours' in a post, Google can be your friend. I put in that term and the first hit was a wiki article with just what you need. Look at the CIE chromaticity diagram on that link. You can't expect to get all you need to know from conversations on PF (Q and A can be a very inefficient form of learning - when you don't actually have a personal tutor, sitting next to you).

 

[Graeme M]

Well yes, but that doesn't answer my musings. The idea of spectral colours described there is pretty much as I assumed it to mean. That is, it's evoked by a single or narrow set of wavelengths. Though here I am not sure why it talks of wavelength rather than frequency but that just means I know little about em radiation. Regardless I can see how that works.

What I am more getting at is that if we consider yellow which is evoked by way of light stimulating the L and M cones in some proportion, and we have some 6 million cone cells in total, there seems to me to be a fair amount of potential latitude in the exact numbers of cone cells stimulated. That is, how likely is it that for arguments sake exactly 1.8 million Ls and 1.65 million Ms are stimulated in each person's retina.

As colour is not a real thing how is it that we can agree on what spectral yellow looks like? Yes it might be light at 570nm, but does it follow that every human retina responds to that in exactly the same proportion? Or does it not matter in that near enough is good enough? Or do we derive the agreement on what spectral yellow looks like by statistical sampling? That is, if we want to create spectral yellow we can't simply produce something that reflects at 570nm, we have to judge by eye surely?

Yes I realize we can measure the frequency and wavelength via an instrument, but the instrument has no idea what yellow looks like, it can only measure the physical property. We still have to agree on what spectral yellow is before we can assign a wavelength to THAT colour.

Don't we?

 

[sophiecentaur]

Though here I am not sure why it talks of wavelength rather than frequency

I went into that. It's just historical and it is now the convention. There would be no point in changing, at this stage.

there seems to me to be a fair amount of potential latitude in the exact numbers of cone cells stimulated.

Yes, I'm sure there is. Our actual memory for colours is pretty poor (which is why we have to take the curtains into the shop when choosing the wallpaper and one's wife would not allow you to go out and choose a top for her birthday, without taking the skirt with you). But we can distinguish between the 'millions of colours' that your TV monitor can give you, in some really critical material (large areas of nearly the same colour). The 'experiment' you describe would not be a good one because it would be illuminating the whole retina so there would be no reference with which the eye could calibrate itself. I also previously mentioned the poor colour memory in a darkened cinema, compared with viewing TV in the home.

but does it follow that every human retina responds to that in exactly the same proportion?

This has also been mentioned before. Is it not well known that the colour sense varies a lot from person to person? (And animal to animal) What is far more important is the discrimination between adjacent areas - revealing patterns and shapes with predators and prey are against a similar coloured background and the slight blush of embarrassment or pleasure on another person's face. We are very very good at that.

We still have to agree on what spectral yellow is before we can assign a wavelength to THAT colour.

I can't imagine a 'Scientist' going to a lot of trouble to name the colour of a spectral line in an experiment - except in very broad terms. The whole point of assigning a wavelength to a spectral line is to make it possible to refer to it with precision. Otoh, an artist, who would never be dealing with spectral lines (there may be exceptions to that statement but it wouldn't involve pigments) will be using an entirely different way of referring to the colours (see the 'Colour Wheel" system) which doesn't refer to wavelength at all. If you look at the CIE diagram, it is surely pretty obvious that the majority of colours in that colour space do not lie on the spectral arch, over the top.
There is no disagreement between the Colour and Wavlength descriptions. They are just appropriate in different contexts. The only thing is to avoid using them for the same thing.
P.S.

if we want to create spectral yellow

. . .we wouldn't use a reflective surface or a filter. We would use a light emitter - probably a sodium discharge lamp. Nothing else would give a totally pure match. (Google colour synthesis in TV and read about the principles behind it). PF can only do so much.

 

[Graeme M]

Thanks sophiecentaur. That still doesn't quite answer my question so I'll assume my question indicates a basic misunderstanding on my part. Oh well, I run into that a LOT! :)

 

[vela]

What I am more getting at is that if we consider yellow which is evoked by way of light stimulating the L and M cones in some proportion, and we have some 6 million cone cells in total, there seems to me to be a fair amount of potential latitude in the exact numbers of cone cells stimulated. That is, how likely is it that for arguments sake exactly 1.8 million Ls and 1.65 million Ms are stimulated in each person's retina.

It's not the number of cones that are stimulated that matters in perceiving color.

As colour is not a real thing how is it that we can agree on what spectral yellow looks like? Yes it might be light at 570nm, but does it follow that every human retina responds to that in exactly the same proportion? Or does it not matter in that near enough is good enough? Or do we derive the agreement on what spectral yellow looks like by statistical sampling? That is, if we want to create spectral yellow we can't simply produce something that reflects at 570nm, we have to judge by eye surely?

You were likely taught as a child that bananas are yellow. In your brain, you perceived some color and assigned it the name yellow. For all you know, the color you perceive as yellow in your brain is what my brain correlates with the color red. There's no way to know what each person actually perceives. But we all agree that whatever color we see that a banana has is called yellow.

 

[Graeme M]

Vela, I think that's exactly what I was getting at. As I understand it, colour is not a property of the physical world. It's an internal representation. Light has the physical properties of wavelength, frequency and so on which we can measure instrumentally, but colour is not a measurable property, or so I thought. A spectrometer will show us the relevant physical properties but it doesn't tell us which wavelengths are which colour, so that must be a subjective judgement?

How then do we settle on 570nm as being the wavelength that represents spectral yellow rather than 580 or 560? If it's being perceived according to an organic perceptual system there must be biases and variability between individuals, so it just seems unlikely to me that every person agrees that a particular wavelength represents a pure colour. So on what basis do we conclude that spectral yellow is 570nm.

I did do a little research/googling but none of the references I found talked about that, they all just operated from the basis that a particular colour has a particular wavelength. The implication seems to be that colour is a physical property and variability between people's perception is just a subjective interpretation of an objective property, but I had thought that colour is not an objective property.

So perhaps I just misunderstand what is meant by colour being represented internally.

Note: I know very little about light, EM radiation, spectroscopy etc so my use of terminology might be a bit (or a lot!) suspect. The question itself is simply a conceptual one about the perception of colour.

 

[sophiecentaur]

For all you know, the color you perceive as yellow in your brain is what my brain correlates with the color red. There's no way to know what each person actually perceives.

IF what you say is true then colour printing and TV displays couldn't work at all. Whilst it is true that the fine detail of peoples' perceptions of colour have a spread, it has been found that people agree, largely with which synthesised colour matches a given original colour. If what you suggest were true then there could be no, (well established) CIE colour space diagram. You would need to turn bits of it inside out, according to who was using it.
I know that people quote colour names associated with sea, sky, blood etc etc differ a lot between cultures but that could well be because the average actual colours actually are different in different climates and lattidudes. Also, skin colours are very different in different places, so the appearance of blood will also be different. It is wrong to confuse the 'names' of colours with how they can be matched to certain mixes of primaries. The latter is a pretty well established bit of psycho-engineering. (Nikon, Cannon and Pentax sell the same cameras throughout the world and they don't need to be tweaked to fit the users in each country.

 

[sophiecentaur]

How then do we settle on 570nm as being the wavelength that represents spectral yellow rather than 580 or 560?

There are many different "spectral yellows". All that is necessary is that sit on the spectral curve and viewers assess them as 'yellow'. You can be more precise and call it 'Sodium Yellow", which nails it to the narrow pair of sodium emission lines. But what you have written implies to me that you are only considering the colours of monochromatic light. I don't know how many times I have to make it clear that most colours are not formed of monochrimatic light. All wavelengths can be assigned a colour but that doesn't imply that all colours can be assigned to a wavelength. If you haven't read statements to that effect then you have not been reading publications about colourimetry. Many (otherwise well informed) people are incredibly sloppy about this issue.

 

[sophiecentaur]

That still doesn't quite answer my question

Which question is the one that's not been answered? What have you done your homework on, so far? Have you seen a CIE colour chart? Have you seen how colours (points) on that chart can be matched with combinations of other points (primaries)? I suspect that you are trying to get your understanding from this PF thread alone. It can't work that way.

 

[DaveC426913]

For all you know, the color you perceive as yellow in your brain is what my brain correlates with the color red. There's no way to know what each person actually perceives. But we all agree that whatever color we see that a banana has is called yellow.

I think we can know a little more than that. All colours are not made equal. For example, when we produce colours by means that are controllable (in terms of brightness, saturation, etc.) we still all agree that a 'standard' yellow is a brighter colour than the other colours. And we agree that the 'standard' blue is darker. Likewise, combining them, colours that combine with yellow still produce lighter than average intermediate colours, etc.

If you were seeing red when I was seeing yellow, there should be a discrepancy between how we rate what we are seeing.

Yet we all agree that yellow is the preferred colour to paint signs that need to catch the eye at night, that red is not as visible as yellow, and that blue would be a poor choice because it's so dark. We will also spot a banana out of a field of neutral grey noise fast than we will spot a purple/blue eggplant.

 

[vela]

Yet we all agree that yellow is the preferred colour to paint signs that need to catch the eye at night, that red is not as visible as yellow, and that blue would be a poor choice because it's so dark. We will also spot a banana out of a field of neutral grey noise fast than we will spot a purple/blue eggplant.

Sure, but that's because the eye responds to certain wavelengths of light more strongly than other. That's independent of how the brain interprets those signals.

 

[Ravyan Asro]

The beam of light should disperse the colours with different wavelengths and the colours will deviate according to their wavelength.

 

[sophiecentaur]

This thread keeps drifting away from the path of righteosness, I'm afraid. People are not sticking to the principles of PF and they are quoting personal views rather than finding out the actual facts and figures of colourimetry. There are numerous links that give the standard models of colour vision and they are based on a lot of measurements and statistics. If it hadn't been sorted out pretty well, then TV and colour printing would never be as good as it is for nearly everyone (proof of the pudding again). This link is full of good stuff and this wiki article is worth getting into and doing more than just skimming.
It is so easy to get the wrong idea about this topic and it ought to be treated in the same way that 'regular Physics' is treated, with a certain amount of reverence for the established theories. It should not be assumed to be an easy chatty topic.

 

[madness]

It seems to me that the basic question of this thread should boil down to whether retinal cone cells are sensitive to the wavelength or the frequency of light. This wiki page goes into the details of how phototransduction in the retina works - https://en.wikipedia.org/wiki/Visual_phototransduction. I haven't found a concrete answer to whether it is the wavelength or frequency that matters most, however.

 

[vela]

This thread keeps drifting away from the path of righteosness, I'm afraid. People are not sticking to the principles of PF and they are quoting personal views rather than finding out the actual facts and figures of colourimetry.

I think you're just missing the point of Graeme's question, which has nothing to do with perception in the context of colorimetry. Any time you start talking about perception, there's a subjective element, and this subjectiveness is what's at the root of Graeme's question and confusion. Suppose a child comes up to you and asks you what the color red is. You'd be hard-pressed to explain what that color is other than showing him a red object and saying "this color is red." The child perceives something and associates it with the color red. If the child is red-green colorblind, he's likely not seeing the same thing you see, yet he still has some notion of "red."

So Graeme's question arises because he has it backwards. We can't define "spectral yellow" in terms of perception and then figure out what frequency of light it corresponds to, because we don't know what another person sees in their mind's eye. For all I know, what you see in your mind's eye would look like a picture from a clown college in mine, and vice versa. Instead, we define that spectral yellow as light of a certain frequency, and we associate that our individual perception with the name "spectral yellow."

 

[DaveC426913]

It seems to me that the basic question of this thread should boil down to whether retinal cone cells are sensitive to the wavelength or the frequency of light. This wiki page goes into the details of how phototransduction in the retina works - https://en.wikipedia.org/wiki/Visual_phototransduction. I haven't found a concrete answer to whether it is the wavelength or frequency that matters most, however.

It is pretty straightforward. The retinal molecules are not stimulated by wavelength. They can only be stimulated by frequency.

How would a molecular bond be able to detect the wavelength of light? One cannot determine a wavelength unless one knows the speed of light and the length of the wave. Molecules do not know this.
What they know is energy levels.

 

[sophiecentaur]

Suppose a child comes up to you and asks you what the color red is.

But that would be Colourimetry. A child would ask you for an example of something that's red. Imagine that child spoke no English and wanted to know the meaning of the word Red. All you could do would be to give an example of two or three disparate objects whose only common feature was their colour (red). A rational child would (or could) then appreciate that all objects of the same sort of colour were red. Totally subjective and it would not ever involve a source of monochromatic light.
Assuming that they had not been misled by you, there would be no way that they would think that a blue balloon was red. They would already have mapped the various colours of objects around them into groups with their own private names for the colours. Unless they had non-standard colour vision, they would not place bannanas amongst a set of objects that we would all recognise as blue. People seem to suggest that 'one person's red' can be 'another person's yellow' but I have yet to read of anyone for whom that is actually true.

 

[Graeme M]

Yes Vela, that's largely what I was getting at. It was just a simple idle question, very very basic. While I understand that we call something yellow because we all agree on what yellow is (whether we really 'see' yellow or blue) I was curious about how we actually settle on some set wavelengths to define various colours if the property colour is not physical. Somewhere in this thread I think it was mentioned that people can detect a wide range of colours (and here I might mean hues or shades, I don't really know anything about colour), but if all they are perceiving is the responses of rods and cones which is a biological process, it seems unlikely they are all perceiving exactly the same thing. So I am just asking how we can assign a physical correlate to a neural correlate with any precision.

 

[rootone]

. People seem to suggest that 'one person's red' can be 'another person's yellow' but I have yet to read of anyone for whom that is actually true.

Hmm, well I can remember an ex partner described a dress she was wearing as green, but I would have called it blue.
There are some in-between shades which people don't agree on.

 

[sophiecentaur]

nothing to do with perception in the context of colorimetry.

Could we be using the term "colourimetry" at cross purposes?
I have noticed that Chemists and Biologists tend to refer to the measurement of the absorption spectra of solutions etc. as colourimetry. That is a different application of the term and is not aimed at colour reproduction. I guess that, originally, many tests were based on change of colour and would have used colour charts. Now, you can buy colourimeters that do a similar trick but automatically; they are effectively, spectrometers of various qualities.
My experience of colourimetry is to map, as accurately as possible, the perception of colours and to reproduce those colours with a 'Metameric Match'. The chemist's approach is expressly to cut out the subjective bit as much as possible. (to eliminate the 'colour' totally from the exercise)

 

[madness]

It is pretty straightforward. The retinal molecules are not stimulated by wavelength. They can only be stimulated by frequency.

How would a molecular bond be able to detect the wavelength of light? One cannot determine a wavelength unless one knows the speed of light and the length of the wave. Molecules do not know this.
What they know is energy levels.

But it's got nothing to do with energy levels. It's a structural change in a protein (a macromolecule) in a process called photoisomerisation. Given that the wavelength of visible light is around the same length scale as conformal changes in macromolecules, I see no reason why wavelength couldn't be the causally relevant factor.

 

[sophiecentaur]

But it's got nothing to do with energy levels. It's a structural change in a protein (a macromolecule) in a process called photoisomerisation. Given that the wavelength of visible light is around the right length scale for conformal changes in macromolecules, I see no reason why wavelength couldn't be the causally relevant factor.

I googled photoiomerisation and, dang me, if the first diagram on the page didn't have hν on it. ν (greek letter nu) stands for frequency and hν is the photon energy.
How about that? Ol' Dave is not raving mad at all.

 

[vela]

I haven't found a concrete answer to whether it is the wavelength or frequency that matters most, however.

This just doesn't seem like a distinction worth making as wavelength and frequency are not independent of each other. Their product has to equal to the speed of light in the medium, which is established by properties of the medium. If you know the wavelength of light in the eye, you know the frequency and vice versa.

 

[madness]

I googled photoiomerisation and, dang me, if the first diagram on the page didn't have hν on it. ν (greek letter nu) stands for frequency and hν is the photon energy.
How about that? Ol' Dave is not raving mad at all.

If you google retinal cone, every website says that cones are sensitive to the wavelength of light. If you take that as the standard then we're not going to get very far.

Edit: see also this http://pubs.rsc.org/en/content/articlelanding/2012/cs/c1cs15179g#!divAbstract, which says "Azobenzene undergoes trans → cisisomerization when irradiated with light tuned to an appropriate wavelength.". I'm not claiming this as evidence that wavelength rather than frequency is the causal factor, I just wanted to show that this is not a good way to settle the issue, since previous studies likely haven't distinguished wavelength and frequency.

 

[madness]

This just doesn't seem like a distinction worth making as wavelength and frequency are not independent of each other. Their product has to equal to the speed of light in the medium, which is established by properties of the medium. If you know the wavelength of light in the eye, you know the frequency and vice versa.

It would be possible to put photoreceptive cells in other media in which light travels at at a different speed in order to decouple the frequency from the wavelength.