'how everyday technology fools our brains into seeing more'

August 15, 2014 7:52 PM

Colour Mixing: The Mystery of Magenta

from the Royal Institution Of Science (Vimeo channel), via The Fox Is Black.

Ri also has things like: The Modern Alchemist on Air, Water and Earth

posted by the man of twists and turns (46 comments total)

Feh. The only "mystery" is why people have this weird notion that colors have to be spectral. It's like asking how you can recognize the sound of the letter S even though you can't produce it with a tuning fork.
posted by baf at 7:58 PM on August 15, 2014 [10 favorites]

Feh. Everyone's complaining about Magenta, but they're all good with Riff-Raff.
posted by eriko at 8:03 PM on August 15, 2014 [13 favorites]

I had that notion. I'd never considered this and I'm no stranger to the idiosyncrasies of human color perception. Consider me thoroughly charmed and educated by the video.
posted by figurant at 8:05 PM on August 15, 2014 [2 favorites]

That was really neat. Haters can can hold their breath until they turn magenta.
posted by exogenous at 8:24 PM on August 15, 2014 [1 favorite]

I didn't know magenta and purple are the same thing. It makes sense, I guess, but I always thought magenta was a pink-purple.
posted by bleep at 8:32 PM on August 15, 2014

Wikipedia would agree that magenta is not purple exactly. Maybe it's a British thing (like torch vs. flashlight).

But color classifications aside, the video is still neat.
posted by sbutler at 8:35 PM on August 15, 2014 [2 favorites]

"the formal name for purple is magenta"

Is this true somewhere? I thought in the UK it was violet.
posted by rahnefan at 8:46 PM on August 15, 2014 [1 favorite]

Well now, I just learned a new thing. Thanks!
posted by ThatCanadianGirl at 9:11 PM on August 15, 2014

Having gotten really into color theory recently I had already learned a lot of this, but this was still a great video. In particular his explanation of why red+green light is perceived as yellow was really helpful -- real yellow light also happens to stimulate the red and green cones about equally, so it makes sense that it would be perceived as metameric with red+green. I knew the facts but hadn't really connected them until now.
posted by vogon_poet at 9:18 PM on August 15, 2014 [2 favorites]

I didn't know magenta and purple are the same thing.

As a Formally Trained Artist, I was about to come in here to huff and puff after that "Purple = Magenta" line as well, and then I watched the rest of it it, and now I get that the purple I know and love is just a slightly more blue-leaning magenta. Science - it works!

(Offer not valid in pigment color mixing schemes.)
posted by Slap*Happy at 9:19 PM on August 15, 2014

But what about octarine?
posted by bashos_frog at 9:21 PM on August 15, 2014 [6 favorites]

Neat. Thanks. And now I see how CMY relate to one another, too. I'd always thought those seemed like arbitrary colours to base a colour system on, but they're not!
posted by If only I had a penguin... at 9:22 PM on August 15, 2014

Impossible colors are vaguely related.
posted by BungaDunga at 9:36 PM on August 15, 2014 [5 favorites]

"the formal name for purple is magenta"

Is this true somewhere? I thought in the UK it was violet.

Yeah, that's bonkers. As far as I'm concerned (Canadian FWIW, with a fine arts degree to boot) purple = violet. (Isaac Newton agrees.) Magenta is completely not purple. It's fuchsia.

Feh. The only "mystery" is why people have this weird notion that colors have to be spectral.

Unfortunately, that is how the world works. Colours correspond to wavelengths, which are actual, you know, lengths of waves, which are measurable. Bend light through a prism, and the frequencies come out in a smooth gradient. The visible spectrum is just the portion of the electromagnetic spectrum we can see. Below the visual spectrum (longer wavelengths) are infrared, radio frequencies, audible frequencies (i.e. sound) and on down; above the visible spectrum are ultraviolet, x-rays, and gamma rays.

And, yeah, magenta isn't anywhere on that spectrum. (Violet certainly is, however.)

Of course, we could all smoke a bunch of pot and get into that whole "are colours really colours, or do we just see them that way, man" thing. (They're not, and we just do. It is known.)
posted by Sys Rq at 9:38 PM on August 15, 2014

Impossible colors are vaguely related.

See also chimerical colours by way of illusions like the Eclipse of Titan, which don't physically exist but are nonetheless perceivable by messing with the human eye.
posted by figurant at 9:45 PM on August 15, 2014 [4 favorites]

And, yeah, magenta isn't anywhere on that spectrum. (Violet certainly is, however.)

Are there any art pigments that reflect violet wavelengths exclusively, rather than reflecting some combination of red and blue light?
posted by evidenceofabsence at 10:10 PM on August 15, 2014 [1 favorite]

But, dude, what if, like, the colors that you see aren't the same colors that I see? Like, what if your blue and my blue are totally different?

Dude. Think about it.
posted by vogon_poet at 10:14 PM on August 15, 2014 [1 favorite]

Human color perception is fascinating. Here's an interesting article on Homer's much-debated "wine-dark sea".
posted by trip and a half at 10:30 PM on August 15, 2014 [2 favorites]

This rods and cones business always messes with my mind. I am stuck trying to mix pigmented oils on a palette; mixing paint is so far in my head that I can't get my mind around this business of red and green making yellow--it isn't true!
posted by Anitanola at 10:37 PM on August 15, 2014

Colours correspond to wavelengths, which are actual, you know, lengths of waves, which are measurable.

I think this is precisely backwards; there's nothing inherently green about 530 nm wavelength radiation (and indeed nothing inherently "visible light" about it) - it just happens to be a wavelength that triggers green receptors in human eyes. The notion of "color" is only meaningful in terms of something humans perceive; most people wouldn't call ultraviolet a color, but various flowers are vividly colored in the ultraviolet spectrum to attract certain pollinators.
posted by NMcCoy at 10:49 PM on August 15, 2014 [9 favorites]

See like a bee
posted by the man of twists and turns at 10:57 PM on August 15, 2014

Colors don't correspond to wavelengths, which is precisely what the video is pointing out... 1) you can perceive magenta, which isn't on the spectrum, meaning it has no wavelength; and 2) that yellow he produced isn't the wavelength of the light either. It was a mixture of red and green light. Colors occur in your head.

(Actually his lights were probably a mess of frequencies, but you'll get the exact same effect with monochromatic red and green laser light.)

(He didn't mention it, but black isn't created by light either. No, not because it's "the absence of light"-- the absence of light is a dark gray, what you see if you close your eyes. Black is a contrast effect created in your eyes/brain. Which is why you can see black on a TV or computer screen, a color darker than the surface of the screen when the power's off.)

It's slightly annoying that he talks about "red, green, and blue" cones, which is not true at all. They're better called L, M, S for the long/medium/short wavelengths they respond to. The L cones respond most strongly to a frequency of about 560 nm... which is greenish-yellow.
posted by zompist at 11:03 PM on August 15, 2014 [10 favorites]

I've been a big fan of this type of stuff for a while now. This is a good introduction to it!

I do disagree with his equating purple and magenta, though.

Anyway, there IS pure monochromatic violet light; it's the stuff around 400nm. And we perceive a mix of blue with a small amount of red the same way, because of the little uptick on the left of the red line on this graph. Note that none of our cones are all that good at that wavelength, which is why it's pretty hard to see that part of actual rainbows.

But magenta is red and blue photons, and there's no single wavelength you can use that will stimulate the red and blue cones equally, so his claim holds for magenta, but not for violet. (OED defines purple as being on the red side of violet, which is even more different from what this guy says, but I personally consider purple and violet interchangeable, so he's just regular wrong, not extra-double wrong. About that one little bit. The overall explanation is spot on, it's just purple = magenta part that's wrong.)
posted by aubilenon at 11:15 PM on August 15, 2014 [1 favorite]

"The only 'mystery' is why people have this weird notion that colors have to be spectral."

Yes and no.

In the sense that you're right, it's not that colors have to be spectral, it's that colors aren't spectral at all, they don't exist physically in that sense. They don't have to be, because they're not. So if you're already assuming the the rainbow is a pure physical manifestation of color (complete or not), then you're already confused.

In the sense that you're wrong, it's, um, that colors don't exist physically in any sense. Which I'm basically repeating what I already wrote, I know, but the essential problem is that people think that color exists out in the world, as opposed to only in our brains, and so pretty much all the other misconceptions follow from that.

A good comparison for what is happening with color is our experience of temperature. Physically, hot and cold are just measurements relative to each other along a single dimension. But, physiologically we have separate nerve receptors for "hot" and "cold" and our subjective experience of hot and cold are qualitatively distinct, not quantitatively. We do experience a quantitative difference with regard to the different levels of stimulation of each kind of those receptors -- so we do notice that one thing is hotter than another hot thing, and one thing is colder than another cold thing. But cold and hot are distinct. Even though physically, they're not, they're just quantitative differences.

Color is pretty much exactly the same thing, except that it's built around three separate receptors and, especially, how they interact. And, as the video explains, what's really interesting about building brain perceptual machinery around the interaction of those three receptors means that you can end up with sensations that are, in their relationship to the physics, really synthetically weird.

What I think is fascinating about this is that it's such a great evolutionary hack. You can see how an additional receptor would evolve to allow for a wider range of sensitivity to a sensation, but that wouldn't require that the two inputs be tagged, perceptually, as being two different inputs. They could be combined into a (subjectively) single qualitative sensation, just sensitive to a greater range of input.

Imagine that the range of sensitivity to the receptor that evolved to respond to light was only, say, right about 535 nanometers, only reaching down to 520 and up to 550. That is, green. And for everything outside that range, it's basically insensitive to completely. So the perception would be a monochrome image of everything that's green and all else would be dark grey/black.

But it would be very useful to see beyond that range, right? So it could only take a small mutation to the gene related to that receptor to move its sensitivity downward or upward. And then you could have a mutation where both receptor types are present in the phenotype, but the brain machinery hasn't changed and doesn't differentiate between the inputs it's getting from those two different kinds of receptors. Why would it?

The result would be a monochrome image that is sensitive to a wider range of light. Which would be great, evolutionarily speaking, assuming that this was environmentally important. And you can imagine how one or more chains of mutations would result with some number of different kinds of receptors that cover the entire range that is environmentally very important. For humans, you could end up with pretty much what we see with black and white movies (they don't match our amplitude sensitivity, but I'm only concerned here with frequency).

So then why color? Well, go outside into a natural environment with plants and animals and you can, um, see why. Assuming that we already have these different receptors that are sensitive to different ranges of light, and assuming that, say, we have a particular interest in the range of light that is blood red, or a leafy green, or a sky/water blue, then if some of our receptor types are variously more sensitive to light in one of those ranges then maybe it would be helpful to not merely combine the inputs into something undifferentiated but, instead, perceptually use these distinctions. Keeping this information means that there's an additional quality we experience, besides just brightness. Something red really stands apart from something green and something blue, which is very, very helpful to us. And so you get mutations coding for the parts of the brain that process these signals such that this extra information is retained and utilized.

Keep in mind that red or blue or green don't really exist, they are just ranges that different receptors are especially sensitive to and which our brains perceptually distinguish. Furthermore, given that these receptors are sensitive to ranges of frequencies, and the chemistry and biology of it means the sensitivity across these frequencies are curves, then they necessarily will overlap (because if they didn't, we'd have in-between ranges that we almost couldn't see at all). Given that, the brain can make use of the fact that light in some frequency ranges stimulate two receptors about equally, and, hey, it can tag that information as distinct, as well.

Outside of physical and, more realistically, biological/chemical limits, this is all pretty arbitrary. We could have fifty different receptors sensitive to fifty overlapping ranges of frequencies and six-hundred distinct colors which we experience as qualitatively distinct. Or fewer than we do. Or more receptor types but not experienced as more colors -- which is (almost) what happens with our night vision, which is a different receptor but pretty much not a different color. But it could be.

It's interesting to compare color vision to hearing. In fact, our hearing works physiologically in a similar way, we actually have a much greater variety of different receptors each sensitive to a smaller range of frequencies than we do with vision. The end result, though, is difficult to compare with vision. I'd argue that the precision of our sensitivity to frequency with hearing means that our brains end up pretty accurately translating that into a measurement of the quantity of frequency in a way that it doesn't do with light. Maybe if we did have hundreds of different photoreceptors we'd experience light the same way ... but I don't think so. I think that there's a more uniform environmental frequency distribution of the sound that we need to be able to hear such that it wouldn't really be useful for us to have some special frequencies experienced as qualitatively distinct, the way we do with different colors. Speculating about it, I'd say that we most need to be sensitive to the sound of movement, and the sounds that predators and prey make. But the sound of movement is spectrally widely distributed, so it wouldn't help there. And the sounds that predator and prey make are too easily altered. That's true of coat colors, as well, but not true of blood (red), or of the green of plants, or of the blue of water and sky. Those are pretty unchangeable and so we get a big return on the investment in perceptual machinery that emphasizes the differences between those different ranges of frequencies. With sound, we hear those differences, but they're all just as important as each other and the transitions are very regular and undifferentiated.

"I think this is precisely backwards..."

and

"Colors don't correspond to wavelengths, which is precisely what the video is pointing out."

I had the same reaction, but he does understand this (you can tell by his "they're not, and we just do"). I agree that he otherwise seems to be asserting the naive, mistaken understanding of color, though.
posted by Ivan Fyodorovich at 11:33 PM on August 15, 2014 [14 favorites]

"2) that yellow he produced isn't the wavelength of the light either. It was a mixture of red and green light. Colors occur in your head."

That's another neat thing -- that color is in our heads and so color imagery is engineered, intentionally or not, around what's happening in our heads. Which means that it only works because of what's happening in our heads and absolutely is not (necessarily or usually) reproducing the physical conditions that gave rise to that perception.

A long time ago I heard that the Voyager disc contains color photographs. Okay, so when I first heard this, I didn't think anything of it. Much later, though, I thought, huh? Seriously? Surely those scientists couldn't possible have thought that would make any sense to an alien?

Well, it turns out, though I couldn't find a lot of information about it, they weren't quite so naive. I believe that they provided actual spectral information about the source and reflected light, at least in some cases and to some degrees.

Even if they hadn't done so, though, what's really interesting is that regular old three color process imagery would be something that sufficiently intelligent hypothetical aliens would probably be able to figure out (eventually) and it would tell them a lot more about some important things than would images with more and more accurate spectral information.

Well, okay, unless we were being crazily and exhaustively comprehensive with the more accurate spectral images, we'd probably only cover our visible range, which would tell them a lot about both our star and our atmosphere. But the three color process, though, tells them that and implies quite a bit about how our vision works (receptors and processing) and what our environment of adaptation most likely was.

What the three color process images would tell them less about are the properties of the things depicted in the imagery and the light which illuminated them.

It's kind of a trade-off. If these were hostile aliens, then the spectral images would be helpful for them to evaluate the environment and levels of technology while the three color process images would imply quite a bit to them about what humans are like.

Lucky for any hostile aliens, we went into great detail about what we're like, including our science and technology, as well as including both spectral imaging and three color process imaging, so they'll be all set for a successful invasion.
posted by Ivan Fyodorovich at 12:12 AM on August 16, 2014 [3 favorites]

Ivan Fyodorovich: colors don't exist physically in any sense

I agree, but the philosophers at BBS, including Daniel Dennett, may disagree.
posted by Gyan at 12:21 AM on August 16, 2014

That's a neat point about the Voyager stuff. I believe our ability to print full-color-spectrum reproductions are not very good. Because the overwhelming majority the technology for color reproduction we've developed in the last 5000 years is designed to trick the eye, not emulate the light. Anything else is only useful for relatively exotic experimental purposes. So mostly what we have is a whole bunch of dyes and pigments which have a bunch of different absorptive and reflective profiles that we can mix together.

For instance a while ago I converted a DSLR I had to capture only near-infrared. The red, green, and blue sensor sites still had different spectral response curves, though, so I could get some false color information out of it. In this picture, you can see a rendering of the completed building - but the ink which is sky-color in visible light is nothing like the sepia sky in the IR spectrum. Similarly to someone who can tell the difference between 590nm light and a mix of equal parts 530nm and 700nm, like a space-mantis-shrimp, none of what we do to color would make any sense at all. Color film photography, electronic displays, dyes, pigments, all of it is fundamentally engineered around the limits of human physiology.
posted by aubilenon at 12:36 AM on August 16, 2014 [6 favorites]

"I believe our ability to print full-color-spectrum reproductions are not very good."

Yeah, it's a problem at recording and reproduction, obviously. What I meant to say was that when I looked into what they did for the Voyager disc, is that I found some indications that they tried to provide some information about the spectral characteristics of what was depicted, and didn't just naively include three color process images with no explanation. I don't know how much spectral information they included, or how they did it.

I sometimes wistfully imagine an image capturing system that records a spectrogram for every unit of resolution as well as, somehow, a reference for the illuminating source(s).
posted by Ivan Fyodorovich at 1:02 AM on August 16, 2014

So the point seems to be that for most colours that humans perceive there are essentially two physic ways to produce the same response in humans.

1) light of a single particular wavelength = X. (ie so 570 nm will appear yellow)

2) a mix of light of wavelengths Y + Z (which could be say 400nm + 800nm and will appear exactly the same yellow!)


Humans are unable to see a difference between X and (Y + Z)
- they look exactly same to us but they are actually physically different. - if you had a more accurate light wavelength measuring device then you could determine the difference between the two.

His point seems to be that for some mixtures of (Y + Z) there is no corresponding X that would produce the same response in humans. - which actually seems quite sensible when you think about it.
posted by mary8nne at 1:07 AM on August 16, 2014 [2 favorites]

That's correct, except that there aren't just two ways to produce color X... there's an infinite number of ways.

And yes, you can measure the light of each frequency, and distinguish X, Y + Z, and all the other possible combinations, using a spectrograph.
posted by zompist at 2:10 AM on August 16, 2014

I admire the elegance and efficiency of the brain's hack.
posted by maxwelton at 2:19 AM on August 16, 2014

What's fascinating to me is the triple set of cones in humans. Fish and birds are mostly tetrachromats (four colour receptors) inherited from early vertebrates, sometimes much more into the UV range. Placental mammals mostly lost two, and only primates got a new third one - other mammals are mostly dichromats, and cannot distinguish between green and red (long wavelengths). Some insects are trichromatic, with UV, blue and green.

Some people only have two of the three working properly, and are thus dichromatic - commonly called colour blind. It mostly affects men, as red-green blindness (the most common) is a mutation on the X chromosome. A woman can be a carrier on one of her X chromosomes, but not colour blind, but her sons have a 50% chance of getting the mutated X chromosome and thus being colour blind - a woman has to have the mutation on both X chromosomes to be colour blind.

That means though that men who are red-green blind have one mutant cone and two normal ones. That also means that their mother and daughters have four cones; one mutant, and three normal ones - which means some women are tetrachromatic - they might be able to see more colours in the red-green range.

Only one woman so far has been found to actually have the ability to do so. Of course, she can't describe what it's like, any more than we can describe what it's like to see red to a dichromat. One supposition why so many tetrachromats are inactive is because all our materials and printing work is done for trichromats, and natural colours they might see are rare, so they never develop the ability to distinguish those extra colours that us trichromats would see as equivalent.
posted by ArkhanJG at 4:47 AM on August 16, 2014 [8 favorites]

>>Colours correspond to wavelengths, which are actual, you know, lengths of waves, which are measurable.

>I think this is precisely backwards; there's nothing inherently green about 530 nm wavelength radiation (and indeed nothing inherently "visible light" about it) - it just happens to be a wavelength that triggers green receptors in human eyes. The notion of "color" is only meaningful in terms of something humans perceive; most people wouldn't call ultraviolet a color, but various flowers are vividly colored in the ultraviolet spectrum to attract certain pollinators.

1. That is what "correspond" means.
2. Read the last bit of the comment you quoted.
posted by Sys Rq at 6:35 AM on August 16, 2014

I love reading about these things and that was a nice demonstration [Though dammit magenta is a thing and I retreat into the "my own damn eyes" cliche]. I've been fascinated to read about this for a while.

From the wiki, Mantis shrimp, "The midband region of its eye is made up of six rows of specialised ommatidia. Four rows carry up to 16 different photoreceptor pigments, 12 for colour sensitivity, others for colour filtering. The vision of the mantis shrimp is so precise that it can perceive both polarised light and multispectral images." I get that bees can see what I can't outside my spectrum. I have a bit of trouble trying to intuit that there are "colors" invisible to me within my spectrum. The video explains it nicely.

Language, somefuckinghow also influences eyesight it seems. Here is an article, and here is a bit more detail.

Thanks for the post. tmotat.
posted by vapidave at 6:58 AM on August 16, 2014 [1 favorite]

So the point seems to be that for most colours that humans perceive there are essentially two physic ways to produce the same response in humans.

1) light of a single particular wavelength = X. (ie so 570 nm will appear yellow)

2) a mix of light of wavelengths Y + Z (which could be say 400nm + 800nm and will appear exactly the same yellow!)


I'd disagree there. The vast majority of colors that humans perceive cannot be produced by a single wavelength at all. The only colors that can be produced by a single wavelength are fully saturated colors.
posted by baf at 7:05 AM on August 16, 2014 [2 favorites]

But, dude, what if, like, the colors that you see aren't the same colors that I see? Like, what if your blue and my blue are totally different?

So like you're reading this on Metafilter and I'm reading it on AskMetafilter? That's deep, man.
posted by yoink at 7:28 AM on August 16, 2014 [1 favorite]

radio frequencies, audible frequencies

Electromagnetic waves (radio light etc.) Are not the same as sound.

Completely different phenomona. For one thing, light can propogate in vacuum.
posted by Pogo_Fuzzybutt at 8:02 AM on August 16, 2014 [2 favorites]

> the essential problem is that people think that color exists out in the world, as opposed to only in our brains,

For that matter, people tend to think there's a world out there to exist in.
posted by jfuller at 9:12 AM on August 16, 2014

Cool. I have a discussion on this in my design class at school.

One small thing I was aware as part of the process that hasn't been mentioned is that the red receptors have a bit of sensitivity in the blue range, which the greens don't. Don't have a reference for that right now, and I'm not sure how much that has to do with the process, but there you go.

Also, a very big issue that hasn't been discussed that affects color perception and blending is the difference between additive and subtractive colors. That's the difference between the RGB image we see on a computer screen (additive) and the CMYK image we see on the printed page (subtractive) and why the two never match. That's why when you mix red and green light you get yellow, and when you mix red and green paint you get brown.
posted by Mcable at 9:22 AM on August 16, 2014 [2 favorites]

baf: "I'd disagree there. The vast majority of colors that humans perceive cannot be produced by a single wavelength at all. The only colors that can be produced by a single wavelength are fully saturated colors."

It's a pretty narrow gamut, but it's it's still a spectrum of light. And yea, I suppose that most light in the real world is a mixture of colors. Otherwise LEDs and lasers would be pretty boring. I'm not surprised people think of color as spectral; children are taught about the light spectrum in grade school, but I don't recall any art teachers telling people about retina response functions.

Personally that just makes it even more interesting. Like chemistry for photons in a way. You have elemental light, that when combined produce compound colors, some of which appear as some other color. But you also have compounds that don't have a corresponding single wavelength. The S sounds, if you will.
posted by pwnguin at 9:45 AM on August 16, 2014

I think of magenta as #FF00FF, or the last color in the ANSI color palette that you use because if you accidentally put magenta letters against a green background the color burns into your soul.
posted by RobotVoodooPower at 10:23 AM on August 16, 2014

Electromagnetic waves (radio light etc.) Are not the same as sound.

Completely different phenomona.

Gosh, someone better tell Alexander Graham Bell.

Electromagnetic frequencies in the audible range, when converted to mechanical waves of equal frequency, result in vibrations of the tympanic membrane that we perceive as sound.

You know full well that's what I meant, pogo_fuzzybutt.
posted by Sys Rq at 10:23 AM on August 16, 2014

Color Vision by Peter Gouras looks to be a good summary of color vision neuroscience.

the geometry of color perception from color vision. Looks like an awesome blog!
posted by Golden Eternity at 10:59 AM on August 16, 2014 [1 favorite]

Those of us sitting in the Colorblind section are smiling politely as if you're all speaking in tongues.
posted by tommasz at 11:43 AM on August 16, 2014

I sometimes wistfully imagine an image capturing system that records a spectrogram for every unit of resolution as well as, somehow, a reference for the illuminating source(s).

Me too! That and lightfield displays.
posted by aubilenon at 12:44 PM on August 16, 2014

"Blue and Yellow Don't Make Green: How to Mix the Color You Really Want"- Every Time Paperback – March 10, 2002

Best color theory ever.
posted by RavinDave at 4:45 AM on August 17, 2014 [2 favorites]