Color Recognition: What we see vs animals with a larger color range

[Orodruin]

 

[collinsmark]

View attachment 340225

Many years ago I remember taking part in a rather deep discussion with @zoobyshoe here on PF regarding a study that was published, but not peer reviewed, on tetrachromacy. Most of that discussion involved criticizing, and breaking down that study.

The gist is that if a person was a true tetrachomat, that person could see clear differences between a true spectrum of white light and a reproduction created with only three colors: say one displayed on an RGB monitor.

That, and there's no way to display an image such that a "normal" tricrhomat could see what a tetrachromat "sees." It just doesn't make any sense. It would be akin to showing a color-blind person an image showing them what a non-color blind person sees.

Anyway, it was that discussion that inspired me to make my very first YouTube video:


Note the video is not a true minimum motion isoluminance test, as that cannot be done using a simple RGB display. The only thing the video does is simulate the experience of taking such a test.

 

[Tom.G]

The YouTube link in the immediately preceeding post

[MEDIA]=youtube]VVovqw0ywf4[/MEDIA]

doesn't seem to work. Clicking the Red arrow has no effect.

However, entering this OLD STYLE link in the browser address field does work. youtube.com/watch?v=VVovqw0ywf4.

The below shows different results depending on how the originating post was copied here... some things work and others don't.

Using the "Quote" facility, this has no link in it.

Anyway, it was that discussion that inspired me to make my very first YouTube video:


Note the video is not a true minimum motion isoluminance test, as that cannot be done using a simple RGB display.

(The situation in the following sentence is intermittent, the 1st save had a bad link, the 2nd save (after editing) had the last link working.
Using "Reply" facility has a working link ONLY IN "PREVIEW" when Creating or Editing this post. When Saved, the link does not work.)

Many years ago I remember taking part in a rather deep discussion with @zoobyshoe here on PF regarding a study that was published, but not peer reviewed, on tetrachromacy. Most of that discussion involved criticizing, and breaking down that study.

The gist is that if a person was a true tetrachomat, that person could see clear differences between a true spectrum of white light and a reproduction created with only three colors: say one displayed on an RGB monitor.

That, and there's no way to display an image such that a "normal" tricrhomat could see what a tetrachromat "sees." It just doesn't make any sense. It would be akin to showing a color-blind person an image showing them what a non-color blind person sees.

Anyway, it was that discussion that inspired me to make my very first YouTube video:


Note the video is not a true minimum motion isoluminance test, as that cannot be done using a simple RGB display. The only thing the video does is simulate the experience of taking such a test.

Cheers,
Tom

 

[collinsmark]

The YouTube link in the immediately preceeding post

[MEDIA]=youtube]VVovqw0ywf4[/MEDIA]

doesn't seem to work. Clicking the Red arrow has no effect.

However, entering this OLD STYLE link in the browser address field does work. youtube.com/watch?v=VVovqw0ywf4.

The below shows different results depending on how the originating post was copied here... some things work and others don't.

Using the "Quote" facility, this has no link in it.


(The situation in the following sentence is intermittent, the 1st save had a bad link, the 2nd save (after editing) had the last link working.
Using "Reply" facility has a working link ONLY IN "PREVIEW" when Creating or Editing this post. When Saved, the link does not work.)


Cheers,
Tom

Sorry about that. I posted the video to YouTube about 8 years ago, and maybe that has something to do with it.

I can see it from the post and click on the video just fine on two computers, one where I'm logged into PF and Google, and another computer where I'm not logged into either (but visiting PF as a guest). So I'm not able to reproduce the problem.

 

[DrClaude]

The gist is that if a person was a true tetrachomat, that person could see clear differences between a true spectrum of white light and a reproduction created with only three colors: say one displayed on an RGB monitor.

That, and there's no way to display an image such that a "normal" tricrhomat could see what a tetrachromat "sees." It just doesn't make any sense. It would be akin to showing a color-blind person an image showing them what a non-color blind person sees.

Knew someone at university who was partially color blind who asked a friend to stop wearing a particular pair of pants around her as they appeared to her as "fluorescent brown"

 

[BillTre]

they appeared to her as "florescent brown"

That's a pretty unique sensory description. Can't think of a fluorescent brown color I have seen.
It is imaginable though.
I find these things intriguing.

 

[Bandersnatch]

Can't think of a fluorescent brown color I have seen.

It reads florescent, not fluorescent. Don't know which was actually meant, and not any easier to imagine either way.
But, in general, probably a worthwhile distinction to keep in mind with Valentine's day around the corner.

 

[BillTre]

not any easier to imagine either way

I think of fluorescent things as have more light coming from them (emitted) than their externally lite surroundings would lead one to expect.
Its a comparison of emitted brightness in a field of illuminated objects.
Brown is brown, but not found fluorescing to my knowledge.

If "florescent" is the proper spelling:

a particular pair of pants around her as they appeared to her as "florescent brown"

it would seem to refer to pants were the color of a brown flower. Doesn't make sense to me.

 

[DrClaude]

It reads florescent, not fluorescent. Don't know which was actually meant, and not any easier to imagine either way.
But, in general, probably a worthwhile distinction to keep in mind with Valentine's day around the corner.

Fixed

 

[DrClaude]

it would seem to refer to pants were the color of a brown flower. Doesn't make sense to me.

We all had problem trying to imagine what the color was like. She understood herself that "fluorescent brown" didn't make sense, but that was the best she could describe it.

 

[DaveC426913]

That, and there's no way to display an image such that a "normal" tricrhomat could see what a tetrachromat "sees." It just doesn't make any sense. It would be akin to showing a color-blind person an image showing them what a non-color blind person sees

It's not like they see new colours. They are merely differentially sensitive to the colours as we see them.

Apparently, tetrachromats have the ability to distinguish finer grades of blue-green. One woman said it was "obvious" to her that that blouse and those slacks were different shades of green, whereas no one else could.

So to simulate that for us mere trichromats, one would show a "before" pic, showing two swatches that appear identical, and next to it, an "after" pic showing two swatches where the erstwhile indistinguishable difference is exaggerated so that one is bluer/yellower than the other.

 

[collinsmark]

It's not like they see new colours. They are merely differentially sensitive to the colours as we see them.

Apparently, tetrachromats have the ability to distinguish finer grades of blue-green. One woman said it was "obvious" to her that that blouse and those slacks were different shades of green, whereas no one else could.

So to simulate that for us mere trichromats, one would show a "before" pic, showing two swatches that appear identical, and next to it, an "after" pic showing two swatches where the erstwhile indistinguishable difference is exaggerated so that one is bluer/yellower than the other.

Indeed, but we need to be careful on how we define "color."

For example, one could make two light sources:
(a) a combination of 612 nm and 549 nm of light, each wavelength having some relative amplitude, and
(b) a single wavelength of 580 nm of light.

If the relative amplitudes in (a) are adjusted just so, a typical trichromat could not tell (a) apart from (b). To the trichromat, they are "the same color." If we were to plot these two light sources on a chromaticity diagram, they would correspond to the same point, i.e., the "same color" (a linear combination of three colors, R, G, And B, map to a single point on the chromaticity diagram).

Figure 1. Example chromaticity diagram. This [two dimensional plot] is adequate for trichromats, but would need a new [third] dimension for tetrachromats.

The source (a) could be displayed using an RGB computer monitor. Source (b) cannot.

A tetrachromat, assuming they actually exist in humans in the first place (which is somewhat debatable), could tell the difference between (a) and (b), and more-so, the tetrachromat might claim that no computer monitor they have ever seen could reproduce (b).

If tetrachromacy actually exists in humans, and we wanted to reproduce "colors," as tetrachromats see "colors," we would not only need to revise our computer monitors and displays, but we would even need to modify the way our files store color data. Storing and displaying color data in RGB format would not be enough: we would need a new value to the mix, perhaps by adding yellow, Y, and displaying and storing color data in RYGB format.

 

[Baluncore]

Ignoring RGB constrictions. We could look at the real world through a broad filter, that would allow us to differentiate colours by brightness, that otherwise would look the same.

 

[collinsmark]

Ignoring RGB constrictions. We could look at the real world through a broad filter, that would allow us to differentiate colours by brightness, that otherwise would look the same.

Yes, of course. If we carried around prisms and/or diffraction gratings we could distinguish such difference by taking careful note of where the spectral bands/lines fell on the resulting spectrum. Or by switching carefully tailored filters in-and-out of our vision and taking note of differences in brightness. Both of those options might have problems in everyday practice though.

One the other hand, if we wanted to reproduce the "colors" as a tetrachromat sees "colors" in nature (again, assuming tetrachromacy actually exists in humans in the first place), we would need 4 primary colors rather than just 3. That means we would need to store image data with 4 colors per pixel (rather than just 3), and the same goes for displays (4 colors per pixel).

 

[Baluncore]

Yes, of course. If we carried around prisms and/or diffraction gratings we could distinguish such difference by taking careful note of where the spectral bands/lines fell on the resulting spectrum.

We do not need a fine spectral filter, just something that splits the sensitivity of one of our color bands into two parts, like a rose colored glass. We only need to use it when we are unable to otherwise differentiate detail.

One the other hand, if we wanted to reproduce the "colors" as a tetrachromat sees "colors" in nature, we would need 4 primary colors rather than just 3.

The illumination of the scene, and the sensitivity of the pigments in our eyes, differ from those of an RGB color camera, and of the RGB display. What hope can there be reproducing color images in three colors, let alone four.

It takes a good white LED source, that includes several different phosphors, before I can be sure of the difference between red and orange resistor color codes. 2k2 = 33k.

 

[collinsmark]

We do not need a fine spectral filter, just something that splits the sensitivity of one of our color bands into two parts, like a rose colored glass. We only need to use it when we are unable to otherwise differentiate detail.

The illumination of the scene, and the sensitivity of the pigments in our eyes, differ from those of an RGB color camera, and of the RGB display. What hope can there be reproducing color images in three colors, let alone four.

It takes a good white LED source, that includes several different phosphors, before I can be sure of the difference between red and orange resistor color codes. 2k2 = 33k.

Yes, I agree that if you work at it, and swap your glasses on and off, you could, with effort, you could distinguish a difference. I have no argument with that. My point is that (as a trichromat) you couldn't tell an obvious difference between (a) and (b) just by glancing at the two sources without any additional tools.

Consider the hypothetical that you were the only person on Earth with green color cones in your retinas Everybody else just had red and blue cones. And consider that the red and blue sensitivity had sufficient overlap where green would have been. (I.e., you are the only trichromat; everybody else is bichromatic.)

Society has created red-blue (RB) monitors and displays instead of red-green-blue (RGB) monitors and displays. JPEG, GIF, MPEG, etc stored their data in only two channels (there would be no need for a third).

Art teachers would tell their students that all colors in the spectrum fall on the "color line" (there would be no such thing as a "color wheel").

When you look at the spectrum displayed on the monitor in post #3339 [that's post #1 now, as posts in this thread were moved from a different thread], it is of course red on one side, blue on the other side, but it's magenta in the middle. That's because it's displayed on an RB monitor. the monitor has no green component in its pixels.

But you have a prism and you use it to create a spectrum from sunlight. You point to the middle of the spectrum where you see green, "I want a monitor that can create that color."

"We already have one, see," they point to the monitor where the magenta is displayed in the middle of the spectrum.

"But that's not green," you retort. "That's magenta."

"What the heck is 'green'?!" they answer brusquely. They have no word "green" in their language. they don't understand what you mean.

"Surely you can see a difference between the middle of this spectrum, magenta, on the monitor, and this one here produced by the sunlight."

"They're both the same," they say, "magenta. They're both magenta. The middle of the spectrum is magenta."

You look out the window and point to the trees and grass (it's springtime, btw.) "What about the grass? Are you saying that grass is colored magenta?!"

"Of course it is," they answer. "Everybody knows that grass is magenta colored."

You try to explain that there's a difference between a combination of wavelengths at each end of the spectrum compared to a single one in the middle, but they have none of it.

"We all know from early art class," they insist, "that all colors of the spectrum can be constructed with different combinations of red and blue. And right in-between red and blue -- in the middle of the spectrum -- is magenta. 'Nuff said," and then they shut down the conversation.

-----


My point of that story is that if tetrachromacy actually exists in some humans, tetrachromats would have similar experiences with just the 3 primary colors society commonly uses in reproducing colors instead of the 4 they can see in nature. (Again, that's if human tetrachromats exist in the first place.)

[Edit: Interesting factoid: in this hypothetical society, "white" and magenta are the same thing. They don't have a word for "white" either. ]

 

[Baluncore]

I agree with your analysis.

Consider the hypothetical that you were the only person on Earth with green color cones in your retinas Everybody else just had red and blue cones.

Luckily, that is not the case, as we have evolved to survive to evolve. Green chlorophyll, red blood, and red berries, are important to us, but why do we need to see sky blue?
Animals that eat only greens, have red-blind eyes, that reflect red light.
Cats eyes reflect greens, as cats need to see their red/brown prey eating greens, against a green background.

If it was important, then tetrachromaticity would have been strongly selected. What advantage did, or will, a tetrachromat have?

If there is such a thing as a tetrachromat, then we seem to have lost that need, and what remains is a vestigial characteristic in the population gene pool, awaiting the return of, or some new advantage.

 

[collinsmark]

I agree with your analysis.

Luckily, that is not the case, as we have evolved to survive to evolve. Green chlorophyll, red blood, and red berries, are important to us, but why do we need to see sky blue?
Animals that eat only greens, have red-blind eyes, that reflect red light.
Cats eyes reflect greens, as cats need to see their red/brown prey eating greens, against a green background.

If it was important, then tetrachromaticity would have been strongly selected. What advantage did, or will, a tetrachromat have?

If there is such a thing as a tetrachromat, then we seem to have lost that need, and what remains is a vestigial characteristic in the population gene pool, awaiting the return of, or some new advantage.

For what it's worth, many animals have more than three color receptors. Apparently butterflies have 5, and, from what I'm told, mantis shrimp have 16*.

Speaking of which, back to funny science stuff. Here's something from The Oatmeal:
https://theoatmeal.com/comics/mantis_shrimp

It's better to go to the source link, but here are a couple of image grabs from the above link.

*[Edit: A more recent paper in the Proceedings of the National Academy of Sciences suggest the number of mantis shrimp color receptors/opsins might be as many as 33. https://www.pnas.org/doi/10.1073/pnas.1917303117 ]

 

[DaveC426913]

Indeed, but we need to be careful on how we define "color."

For example, one could make two light sources:
(a) a combination of 612 nm and 549 nm of light, each wavelength having some relative amplitude, and
(b) a single wavelength of 580 nm of light.

If the relative amplitudes in (a) are adjusted just so, a typical trichromat could not tell (a) apart from (b). To the trichromat, they are "the same color." If we were to plot these two light sources on a chromaticity diagram, they would correspond to the same point, i.e., the "same color" (a linear combination of three colors, R, G, And B, map to a single point on the chromaticity diagram).

View attachment 340644
Figure 1. Example chromaticity diagram. This [two dimensional plot] is adequate for trichromats, but would need a new [third] dimension for tetrachromats.

The source (a) could be displayed using an RGB computer monitor. Source (b) cannot.

A tetrachromat, assuming they actually exist in humans in the first place (which is somewhat debatable), could tell the difference between (a) and (b), and more-so, the tetrachromat might claim that no computer monitor they have ever seen could reproduce (b).

If tetrachromacy actually exists in humans, and we wanted to reproduce "colors," as tetrachromats see "colors," we would not only need to revise our computer monitors and displays, but we would even need to modify the way our files store color data. Storing and displaying color data in RGB format would not be enough: we would need a new value to the mix, perhaps by adding yellow, Y, and displaying and storing color data in RYGB format.

I'm just not convinced this is the whole story.

Here's a simulation of some real-world clothing swatches.
Two different materials, with different mixtures of colours; the threads are a mix of blue and green, producing a blue-green effect.

For a trichromat, the two materials look indistinguishable, but for the tetrachromat, who has an additional cones that peak more in the blue (dotted line at 498), the two colours are obviously different, because their fourth blue-green cones are stimulated more from swatch two than from swatch one.

I don't see why this effect would not extend to an RGB monitor.

The tetrachromat would be able to distinguish between RGB(000,133,059) and RGB(000,133,069) easier than we trichromats could.

It's kind of the colour equivalent of triangulation. They have an extra sensor at a different wavelength, than we do, to zero in on a colour contrast.

 

[BillTre]

I got no references on this, but:
I read somewhere that one or more of the mantis shrimp's many color receptors may specifically detect color signals involved in mating.
Some of the color detectors may only cover certain parts of the visual field (predator, prey, or mate detection.
Not all visual functions need be biologically equal and used for general visual purposes.

 

[collinsmark]

I'm just not convinced this is the whole story.

For what it's worth, let me state for the record that I'm not fully convinced that human tetrachromats even exist in the first place. The evidence for their existence seems a bit sketchy to me.

That said, there are people who claim to be tetrachromats. And there is some solid, biological evidence that some women actually have a 4th type of color receptor. And if it is accompanied by correspondingly separate mechanism of getting that 4th type of information to the brain, then maybe yes, there is tetrachromacy in some humans.

And there's plenty of evidence that even higher dimensions (e.g., pentachromacy) exist in the animal kingdom.

In other words, while I'm not convinced tetrachromacy exists in humans, I'm open to the possibility.

Here's a simulation of some real-world clothing swatches.
Two different materials, with different mixtures of colours; the threads are a mix of blue and green, producing a blue-green effect.

View attachment 340667

For a trichromat, the two materials look indistinguishable, but for the tetrachromat, who has an additional cones that peak more in the blue (dotted line at 498), the two colours are obviously different, because their fourth blue-green cones are stimulated more from swatch two than from swatch one.

View attachment 340666

Well, while some ranges of spectral properties of the source might look like that in your example, other ranges could have completely different qualitative colors that we, as trichromats, have not even imagined.

---

Suppose this same conversation was happening in the hypothetical world of the bichromats. One of the bichromat's says, "Well if @DaveC426913 is telling the truth, and he can make out a completely different color at wavelength of 534 nm, that just means he can better distinguish slightly different shades of magenta."

But you, being an actual trichromat, know quite well that the color "green" is qualitatively much more than just a slightly different shade of magenta.

I don't see why this effect would not extend to an RGB monitor.

The tetrachromat would be able to distinguish between RGB(000,133,059) and RGB(000,133,069) easier than we trichromats could.

It's kind of the colour equivalent of triangulation. They have an extra sensor at a different wavelength, than we do, to zero in on a colour contrast.

"Triangulation" is quite apt here. We, as trichromats can triangulate a three dimensional color (using three primary colors such as red, green and blue) onto a two dimensional color wheel, or a chromaticity diagram if you prefer.

Figure 1. Example chromaticity diagram. In such diagrams the intensity/luminosity/brightness is not indicated in the diagram; only the color quality is shown. In this way, 3-dimensional color vectors (that include brightness) can be projected onto a 2-dimensional color space diagram (which does not include brightness).

But if you had a 4th receptor, complete with all the biological mechanisms of passing that information back to the brain, the color information would not be 2-dimensional, it would be 3-dimensional and span a volume rather than an area.

Again, it would be like telling one of the hypothetical bichromats that "green" is qualitatively different than just a slightly different shade of magenta. We (trichromats) see the chromaticity diagram (Fig. 1) as two dimensional, while the bichromats are convinced that chromaticity is simply composed of a 1-D line.

So your example of "RGB(000,133,059) and RGB(000,133,069)" does not apply to a tetrachromat. You need four data points (a 4 dimensional vector) to fully represent the qualitative color (including brightness information) to a tetrachromat, not three.

 

[somegrue]

[Edit: Interesting factoid: in this hypothetical society, "white" and magenta are the same thing. They don't have a word for "white" either. ]

[Wikipedia says it's called "dichromacy", not "bichromacy".]

"White" seems the more basic concept, so it may be more apt to say that it's "magenta" that's surplus to requirements. In that sense, the entire idea of "colour" all but vanishes, really. Mixing red and blue no longer makes something worth a name like either "magenta" or "green" or anything else, it just makes them paler.

Which makes a lot of sense in hue-saturation-brightness terms, come to think of it. Throwing out one of RGB ought to mean that we get to throw out one of those as well. Brightness seems, again, too basic to touch, so in our terms, dichromats would have no notion of hue, or no notion of saturation, or perhaps a more palatable way to put it would be to say that the two notions collapse into one.

Aside: I watched a nature documentary a while ago that made an effort to get the audience to understand how come herbivores seem to have just as much trouble spotting a tiger's orange in front of lush green vegetation as they do spotting a lion's yellow in front of dry yellow vegetation, quite unlike we do. The(ir) reason was that herbivores are dichromatic, and to visualize that, they simply replayed the tiger footage and faded the colour to black-and-white - and voila, the tiger very much faded into the background as well. IIRC, my mental reaction at the time was a mixture of "nice!" and "surely that's cheating somehow". In the context of the previous paragraph, that was surprisingly spot-on, if I do say so myself: Taking it from trichromatic to monochromatic is clearly the right way to go, but overshoots the mark. They went from HSB to just B, when it should have been B+1.

Back to topic: Conversely, a true tetrachromat ought to be thinking about colour in terms of an HSBX tetrad, where X is a quality that's orthogonal to the trichromatic experience. Maybe it's fine to rephrase that as H²SB... like, even though I'm not actually able to conceptualize a two-dimensional hue, that seems only mildly weird. Which may have to do with typically thinking about one-dimensional hue as a circle instead of a straight line? Or maybe it's not fine.

Anyway, kindly let me know if any of this is unsound! :)