Comment by jrapdx3 | Hacker News Reader

Comment by jrapdx3 | Hacker News Reader

It's hard to compress what's known about the subject of color perception differences among people with nominally normal color vision.

Hypothetical tetrachromacy has been demonstrated, that is, M and L cones with minor differences in peak spectral response existing in a retinal mosaic. The question arises whether this difference is capable of being propagated through the bipolar connecting neuronal paths (which form the "blue-yellow" and "red-green" channel paths conveying color info to the brain visual cortex).

Several studies found no evidence for a "fourth channel", and some perceptual studies did not show confirmation of tetrachromacy. However a couple of studies do hint at rare female subjects who do seem to have good evidence for this trait.

It is postulated that because of the differences in the exact genetic encodings on the 2 X chromosomes in females, that up to 50% of women have a retinal mosaic of 4 populations of cones, S M M' L, S M L L', and so on. The theory is that this allowed greater distinction of yellow to red color distinction which may have helped survival by avoiding toxic plants based on subtle color distinctions.

In art, there are great colorists and those who are virtually incapable of subtle use of color. If anyone interested in finding out about their own abilities, check out: http://www.xrite.com/online-color-test-challenge

The on-line test is only a rough estimate, but it's fun and challenging. I suspect there's a spectrum, aka normal distribution of color discrimination ability. I should't say how well I did, though the results of trying it several times were remarkably consistent despite different computer systems that I used.

BTW I encountered one very intriguing article reporting on a large sample of retinas obtained post-mortem. As expected, ~50% of females showed retinal mosaic of 2 M or L cones where only one each was expected. Most fascinating was the fact that of male retinas 8% showed a similar tetrachromat pattern. How that could arise is a mystery considering males have only one X chromosome. (There are pigment gene variations with greater number of CNV and TR which could be activated? But it's still a mystery.)

> Hypothetical tetrachromacy has been demonstrated, that is, M and L cones with minor differences in peak spectral response existing in a retinal mosaic. The question arises whether this difference is capable of being propagated through the bipolar connecting neuronal paths (which form the "blue-yellow" and "red-green" channel paths conveying color info to the brain visual cortex).

Adding genes for extra cone types into normally dichromatic mammals (mice or something? I don’t remember off hand and I don’t have the citation at my fingertips) has seemed in studies to result in trichromatic vision, so it’s not too big a leap to suspect that tetrachromacy of the type described in the article might happen for some humans.

As you say there hasn’t been any study which showed this conclusively (that I’ve seen anyhow). I’d love to see more thorough research on these subjects who supposedly possess such vision.

> In art, there are great colorists and those who are virtually incapable of subtle use of color. If anyone interested in finding out about their own abilities, check out: http://www.xrite.com/online-color-test-challenge

Note, this online test is a bit easier than the paper version of the Farnsworth–Munsell hue test, since the colors in the screen version end up with some lightness differences that make it a bit easier to keep some sections of the chart in order. Either way, the test IMO mostly measures (a) whether someone has “normal” trichromatic vision, and (b) how patient and willing to fiddle with fine details they are. I don’t personally think Farnsworth–Munsell test scores are super meaningful, though a very poor score does indicate some color vision deficiency. [FWIW, if I take the time to do the test slowly, either on screen or on paper, I consistently score ~0, occasionally mixing up one pair or another.]

But anyway, I think the ability to be a good colorist in art has a whole lot to do with practice. Spending a lot of time mixing paint or color correcting photographs is likely to heighten awareness of tiny distinctions. For instance, I know that after a couple years of photography courses, several of my friends got much better at noticing color casts in photographs. Not that they couldn’t physically see them before, but after experience they more often spontaneously noticed the casts and started to have a feel for just how much adjustment in which direction would be necessary to counteract them.

> I think some of this has to do with practice. Spending a lot of time mixing paint or color correcting photographs is likely to heighten awareness of tiny distinctions.

I wouldn't doubt that practice could help, but the test (I believe) was designed to minimize practice effect. I've mixed a lot of color, but one can mix color to a shade only as far as one could determine it's right.

It used to be, color matching for printed material was done in large shops by workers visually comparing swatches to customer desired shades. This required training, but not every trainee could master it despite extra effort and time. Seemed it required native ability, talent as it were, to be able to be good at it.

Anyway, the online test is solely for one's own interest and information. I don't see there's anything substantial at stake.

> I think some of this has to do with practice. Spending a lot of time mixing paint or color correcting photographs is likely to heighten awareness of tiny distinctions.

I've never done either, score 12, age +- 50, male.

Interesting test, harder than I thought it would be.

I too have never done either, and also scored 12.

I scored 0 on the test first time (perfect color vision), no practice, and I have pretty much no experience with any kind of art. (27 male for ref)

EDIT: Had another 2 tries, got 4 and 12. Turns out practice does more harm than good - after 3 tries it's a bit disorienting trying to read off the screen. Gonna have to take a break from machine.

A test like this needs to isolate the subject to only one spectrum at a time. I gave up after the first two because of too much noise.

\tangent A computer-based color test might not be able to detect tetrachromacy, because monitors do not display a continuum of frequencies of light. They are intrinsically tied to trichromacy, by combining only three frequencies of light in different proportions. RGB.

It's mathematically possible that a continuous spectrum of light also wouldn't pick it up, if its only effect was that different combinations of frequency that appear identical to an ordinary person could be distinguished. (Not that this would confer any evolutionary advantage.)

This test went around at work a while ago. At the time, I worked with a number of artists.

Many people got very different results depending on the monitor they were using.

This would probably be a very large factor in the score. I scored a 7 on a calibrated Dell UltraSharp (IPS) monitor at work. If I were to do this same test on my cheap uncalibrated Acer at home, I don't believe my score would be nearly as good. Even the difference between a calibrated and an uncalibrated cheaper LCD should be noticeable to most people.

I'm not sure how to describe it, I got 25... I could see the variances, in the tiles that looked "off" but wasn't always sure which direction to split to... I finally got frustrated and hit score. :-(

> But anyway, I think the ability to be a good colorist in art has a whole lot to do with practice.

Indeed. I scored 12 on that test, despite being colorblind with moderate-strong protanopia.

I got 39. I found the test interesting in terms of sorting algorithms. Typically when we think of sorting stuff, the comparison function is deterministic and cheap.

In this case, it gets increasingly hard and even erratic to tell adjacent colors apart; essentially the comparison function is increasingly expensive and non-deterministic.

I wonder if that has any consequences for the "optimal" strategy. In my case -- and I suspect most people did it the same way --, I used a kind of two-way insertion sort: for each field, I first decided if it should move left or right, then moved it as far as it "made sense". So I ended up with a list that got sorted from the outside towards the inside. In the end I had a few passes of "bubble sort", checking adjacent pairs for correctness. This was when it got expensive and erratic. :)

It would be a fun experiment to have an algorithm sort the list (based on whatever sorting algorithm) and simply use the human as a comparison robot: display two hues at the same time and let the human decide which of the two is "smaller". I'm sure you could easily measure and even predict the increase in "expense" (i.e. response time) and the increase in errors and ambiguity for close hues.

You could also do stuff like force the user to decide within a very short time, to determine if there's actually an increase in accuracy when staring at a given pair for 10 seconds vs a 500ms "hunch" decision (or heck, 10 500ms hunches, still a two-fold improvement).

I suspect there's a spectrum [of color perception among people with nominally normal color vision].

Indeed there is. The most common is a variant carried by around 2/3 of the human population (depending on ethnicity and gender), which results in a slightly different red response curve, as shown in color matching tests [1]; and might also be involved in some forms of actual color blindness. Other, less common variations involve mosaicism of both red and green pigments as well [see ibid, Fig. 8].

The tetrachromacy thing is a perennial media favorite these days; but IMO these subtler but actually widespread differences are much more interesting, at least from a philosophy of mind point of view.

[1] Deeb 2005 - The molecular basis of variation in human color vision. Link below:

http://www.mbfys.ru.nl/staff/j.vangisbergen/endnote/endnotep...

Thanks for the reference, I hadn't seen it previously. In males, the variation appeared to be more toward decreased color perception, except for one variation associated with increased red sensitivity.

It does raise the interesting but perhaps unanswerable question of the similarity or difference of the effect of the molecular variation vs. tetrachromacy on color perception.

Would greater red sensitivity of L cones in effect produce equivalent visual experience as having a fourth type of cone?

Since the visual system computes color on the basis of differential response in M and L cones at time t, having a greater separation of M and L spectra would be sufficient information to compute a larger range of intermediate colors vs. the normal L-M difference. The presence of another node L' between L and M could just be redundant input and not necessarily adding information to the color computation.

A question that arises here is if S M L L' females have better color perception is it because of having L and L' or would it reflect L' being more distant from M than the typical single L of female trichromats.

That may be significant in light of lack of clear findings of direct benefit to female color perception in those having L L' retinal mosaics.

EDIT: Actually, to my way of thinking, the issue is more precisely color discrimination, which is fairly directly measurable vs color perception, a less tractable concept.

I just did the test, and whoohoo I got perfect colour vision: http://i.imgur.com/WFoktvP.png (yay!)

But does that really mean I got all of them correct?! I was squinting and squinting (and cursing insertion sort), and I was pretty tired out near the end. I thought I might've been able to squeeze out some last tiny errors if I would have gone over all those bars again, flipping a few boxes that were not just quite right.

Also if I look at the rainbow bar in the screenshot, I'm assuming that's the full range of colours I sorted, I can clearly see there's a few errors and discontinuities in there still.

I wonder if it maybe just rounds down really small errors to zero because they're "close enough"? That's too bad, because I'm tempted to go back and try to do better :) But if it won't increase my score (decrease my error rate), I only have my own judgement of that final full rainbow hue bar to go by :)

(also, remember to disable Flux/Redshift. it probably doesn't help accuracy :) )

"considering males have only one X chromosome": Around 1 in 500 males have two X chromosomes (XXY), so it should be possible for males to end up as tetrachromats too (although this wouldn't explain 8%). See https://en.wikipedia.org/wiki/Klinefelter_syndrome

http://www.xrite.com/online-color-test-challenge

Doesn't work on mobile :-(

Reply on news.ycombinator.com