9 ms·

I disagree with the reasoning that purple looks closer to red than green because of the "left hump" in the red cone sensitivity curve. That IS why we see a purp

I disagree with the reasoning that purple looks closer to red than green because of the "left hump" in the red cone sensitivity curve. That IS why we see a purplish color in the rainbow (as opposed to it stopping at blue). But purple is close to red because of the circular nature of color because of the 3 types of cone cells. Even in the simplified model used by digital cameras, where it truly is red, green and blue, red is close to purple. This would be true even if (as is true on many camera sensors) the red sensitivity curve had no "left hump". See https://www.quora.com/Why-do-we-perceive-colors-red-and-violet-as-similar-for-example-they-are-next-to-each-other-on-a-color-wheel-when-they-are-at-opposite-ends-of-the-visible-part-of-the-electromagnetic-spectrum https://www.quora.com/Why-do-we-perceive-colors-red-and-viol...

Miiko 5y ago
Not sure I follow your arguments. Red sensitivity curves on the page you linked do have "left hump" (and 2 of answers there directly attribute closeness of red and purple to them). Could you please explain how 3 types of cones can cause "circular nature of color" without existence of that "left hump"?
- Jensson 5y ago
  A color is the combination of cells that gets activated. Without the left hump purple would still be purple even if that spectrum would no longer be perceived as purple. Red and purple are similar since they have many of the same cells getting activated.
  - Miiko 5y ago
    > Without the left hump purple would still be purple even if that spectrum would no longer be perceived as purple. Do not quite understand what you're trying to say (sorry, English is not my native language), here is my understanding - without the left hump purple would be still called "purple" but perceived as kind of deep blue (separate from regular blue, but not closer to red as it is perceived now). Is it what you were saying? > Red and purple are similar since they have many of the same cells getting activated. And this is the part I do not understand - without the left hump, purple (defined as spectral color with shortest wavelength) would NOT activate many red-sensitive (L) cone cells. Why it would have more similarities in cell activation to magenta (defined as red + blue) then? What specifically would cause L-cones input ignored in this case?
    - robbrown451 5y ago
      " without the left hump purple would be still called "purple" but perceived as kind of deep blue (separate from regular blue, but not closer to red as it is perceived now). " That's not correct. Without the left hump, almost everything would appear the same, with a few minor exceptions. You see purple and magenta when both the "red cones" and "blue cones" are activated, but not the "green cones". This can happen two different ways. One is very short wavelengths of light, which appear violet. The red cones are stimulated because of that left hump. This is where violet on the rainbow/spectrum comes from, and why a "black light" looks purplish rather than simply blue. But another way is to have some long wavelengths (which alone would appear red) and some short wavelengths (which alone would appear blue), mixed together. Note that to see a hot pink/magenta/fuchsia, this is the only way it will happen.
      - Miiko 5y ago
        You got me confused. In the original comment you wrote: > I disagree with the reasoning that purple looks closer to red than green because of the "left hump" but now you argue: > The red cones are stimulated because of that left hump. So do you agree with similarities of red and purple caused by "left hump" or not?
        robbrown451 5y ago
        I suggest reading the two links to my posts on Quora, where I go into a good bit of detail about it. But no, the red hump is not why red is close to purple. The red hump is simply why, in some circumstances, you can see purple (technically violet... which is a blueish purple) with only short wavelengths, rather than by mixing long wavelength and short wavelength light. In most cases, when you see something that appears purple, it is because you are seeing a mixture of red (long wavelength) and blue (short wavelength) light. In that case, the left hump is insignificant. Note that in all cases, when you see purple, your "red" cone cells (as well as your "blue" cone cells) are being stimulated. THAT is why red is close to purple.
        5y ago
- robbrown451 5y ago
  Red green and blue form a triangle, which is essentially cyclic. Here's another answer of mine on Quora that goes into more detail. https://www.quora.com/Why-do-we-use-RGB-to-represent-colors-Why-not-represent-colors-with-only-two-parameters-intensity-and-wavelength-of-the-light-that-color-represents/answer/Rob-Brown-13 https://www.quora.com/Why-do-we-use-RGB-to-represent-colors-...
  - Miiko 5y ago
    > Red green and blue form a triangle, which is essentially cyclic. But the question then - why wavelength R=700, G=550 and B=450 form a triangle? For your answer on Quora, it says "Hue is cyclical rather than linear" without any explanation, and this is part - in my understanding - can be only explained by "left hump" (red cells being sensitive to blue light as well) and nothing else.
    - robbrown451 5y ago
      The triangle comes from their being 3, nothing more. There are three types of sensors. If you try to visualize them geometrically, you get a triangle. The gamut of all colors therefore is represented as a cube, with black, white, red, green, blue, magenta, yellow, cyan as corners. None of this is dependent on the left hump... the left hump is just a minor quirk that causes the shortest wavelengths to also stimulate the "red cones" a bit (the cone cells whose peak sensitivity is longer wavelengths). Its easiest to understand all this if you work with colors as RGB, as anyone who does things with colors on computers typically does (paint programs, html/css, etc). Note that when you see purple (or magenta, or violet) on a computer screen , it is by combination of red and blue, but by somehow making the wavelength of light coming off the screen actually be a shorter wavelength than blue.
      - Miiko 5y ago
        When I try to visualize 3 types of sensors (without left hump), I get a straight line not a triangle - that is technically still a (zero area) triangle, but inherently NOT "cyclic". Looks like you are mixing together magenta (red+blue) and violet (shorter wavelength than blue) - those are physically very different colors, but perceived by human eye as being the same / very similar. And the only explanation for that perception (= the reason why colors form a wheel) is that "left hump", nothing else
Jensson 5y ago
Yeah, red being closer to purple than red to green is about as strange as red being closer to orange than to green. The absolute difference between (1, 0, 1) and (1, 0, 0) is smaller than between (0, 1, 0) and (1, 0, 0).
- robbrown451 5y ago
  Right...if you think of color as RGB as you do (which is simplified, but mostly correct), it makes this concept a lot clearer.
OscarCunningham 5y ago
The red sensitivity curve for eyes (properly the cones are called Long, Medium and Short rather than Red, Green and Blue) also doesn't have a left hump. This is a misunderstanding originating from the fact that the X in CIE XYZ does have a left hump. But the XYZ functions are not intended to correspond to cone sensitivities. https://en.wikipedia.org/wiki/CIE_1931_color_space https://en.wikipedia.org/wiki/CIE_1931_color_space
- Miiko 5y ago
  According to what I could google, red sensitivity curve does have a left bump: https://www.researchgate.net/figure/Normalized-spectral-sensitivity-of-retinal-rod-and-cone-cells_fig7_265155524 https://www.researchgate.net/figure/Normalized-spectral-sens... So this is not a misunderstanding, but the real reason why magenta (red+blue) is perceived similar to violet (shorter wavelength than blue) by human eye.
  - dahart 5y ago
    What you’re finding is the CIE “color matching functions”, which is closely related but not the same thing as the raw physical response of the cones. Here is a paper describing the difference, and more current research than what you found: https://hal.archives-ouvertes.fr/hal-01565649/document https://hal.archives-ouvertes.fr/hal-01565649/document Look specifically at page 2, and the linear relationship between the color matching functions (where red has a left-bump), and the “fundamental sensitivity” plot, which is attempting to find raw physiological cone response to different wavelengths, rather than what the observer perceives.
dahart 5y ago
So what you mean is that any color that is a mixture of two wavelength that trigger two different cones is “close” in some sense to the perceived color of either one of those wavelengths? This doesn’t mean color is “circular” per se, right? I suppose color perception with 3 cones has a sort of circularity when we plot the color space in 2d, but I’m not sure that sufficiently explains our experience of color. Anyway, what you’re saying is that mixtures of two things can be perceived as close to one of the two inputs. The argument holds for any number of cones - if we had 10 cones, we could still trigger the two with the furthest apart wavelength response using a mixture of two separate wavelengths, and the result would be “close” to one of the two. But this also holds for any pair - the mixture of imagined cone #3 and imagined cone #6 would also yield a color that is “close” to #3, and also “close” to #6. Has anyone asked a tetrachromat this question? What would it mean if they disagree that red is close to purple? What would it mean if they agree?

Miiko 5y ago

Not sure I follow your arguments. Red sensitivity curves on the page you linked do have "left hump" (and 2 of answers there directly attribute closeness of red and purple to them). Could you please explain how 3 types of cones can cause "circular nature of color" without existence of that "left hump"?

Jensson 5y ago
A color is the combination of cells that gets activated. Without the left hump purple would still be purple even if that spectrum would no longer be perceived as purple. Red and purple are similar since they have many of the same cells getting activated.
- Miiko 5y ago
  > Without the left hump purple would still be purple even if that spectrum would no longer be perceived as purple. Do not quite understand what you're trying to say (sorry, English is not my native language), here is my understanding - without the left hump purple would be still called "purple" but perceived as kind of deep blue (separate from regular blue, but not closer to red as it is perceived now). Is it what you were saying? > Red and purple are similar since they have many of the same cells getting activated. And this is the part I do not understand - without the left hump, purple (defined as spectral color with shortest wavelength) would NOT activate many red-sensitive (L) cone cells. Why it would have more similarities in cell activation to magenta (defined as red + blue) then? What specifically would cause L-cones input ignored in this case?
  - robbrown451 5y ago
    " without the left hump purple would be still called "purple" but perceived as kind of deep blue (separate from regular blue, but not closer to red as it is perceived now). " That's not correct. Without the left hump, almost everything would appear the same, with a few minor exceptions. You see purple and magenta when both the "red cones" and "blue cones" are activated, but not the "green cones". This can happen two different ways. One is very short wavelengths of light, which appear violet. The red cones are stimulated because of that left hump. This is where violet on the rainbow/spectrum comes from, and why a "black light" looks purplish rather than simply blue. But another way is to have some long wavelengths (which alone would appear red) and some short wavelengths (which alone would appear blue), mixed together. Note that to see a hot pink/magenta/fuchsia, this is the only way it will happen.
    - Miiko 5y ago
      You got me confused. In the original comment you wrote: > I disagree with the reasoning that purple looks closer to red than green because of the "left hump" but now you argue: > The red cones are stimulated because of that left hump. So do you agree with similarities of red and purple caused by "left hump" or not?
      - robbrown451 5y ago
        I suggest reading the two links to my posts on Quora, where I go into a good bit of detail about it. But no, the red hump is not why red is close to purple. The red hump is simply why, in some circumstances, you can see purple (technically violet... which is a blueish purple) with only short wavelengths, rather than by mixing long wavelength and short wavelength light. In most cases, when you see something that appears purple, it is because you are seeing a mixture of red (long wavelength) and blue (short wavelength) light. In that case, the left hump is insignificant. Note that in all cases, when you see purple, your "red" cone cells (as well as your "blue" cone cells) are being stimulated. THAT is why red is close to purple.
        5y ago
robbrown451 5y ago
Red green and blue form a triangle, which is essentially cyclic. Here's another answer of mine on Quora that goes into more detail. https://www.quora.com/Why-do-we-use-RGB-to-represent-colors-Why-not-represent-colors-with-only-two-parameters-intensity-and-wavelength-of-the-light-that-color-represents/answer/Rob-Brown-13 https://www.quora.com/Why-do-we-use-RGB-to-represent-colors-...
- Miiko 5y ago
  > Red green and blue form a triangle, which is essentially cyclic. But the question then - why wavelength R=700, G=550 and B=450 form a triangle? For your answer on Quora, it says "Hue is cyclical rather than linear" without any explanation, and this is part - in my understanding - can be only explained by "left hump" (red cells being sensitive to blue light as well) and nothing else.
  - robbrown451 5y ago
    The triangle comes from their being 3, nothing more. There are three types of sensors. If you try to visualize them geometrically, you get a triangle. The gamut of all colors therefore is represented as a cube, with black, white, red, green, blue, magenta, yellow, cyan as corners. None of this is dependent on the left hump... the left hump is just a minor quirk that causes the shortest wavelengths to also stimulate the "red cones" a bit (the cone cells whose peak sensitivity is longer wavelengths). Its easiest to understand all this if you work with colors as RGB, as anyone who does things with colors on computers typically does (paint programs, html/css, etc). Note that when you see purple (or magenta, or violet) on a computer screen , it is by combination of red and blue, but by somehow making the wavelength of light coming off the screen actually be a shorter wavelength than blue.
    - Miiko 5y ago
      When I try to visualize 3 types of sensors (without left hump), I get a straight line not a triangle - that is technically still a (zero area) triangle, but inherently NOT "cyclic". Looks like you are mixing together magenta (red+blue) and violet (shorter wavelength than blue) - those are physically very different colors, but perceived by human eye as being the same / very similar. And the only explanation for that perception (= the reason why colors form a wheel) is that "left hump", nothing else

Jensson 5y ago

Yeah, red being closer to purple than red to green is about as strange as red being closer to orange than to green. The absolute difference between (1, 0, 1) and (1, 0, 0) is smaller than between (0, 1, 0) and (1, 0, 0).

robbrown451 5y ago
Right...if you think of color as RGB as you do (which is simplified, but mostly correct), it makes this concept a lot clearer.

OscarCunningham 5y ago

The red sensitivity curve for eyes (properly the cones are called Long, Medium and Short rather than Red, Green and Blue) also doesn't have a left hump. This is a misunderstanding originating from the fact that the X in CIE XYZ does have a left hump. But the XYZ functions are not intended to correspond to cone sensitivities. https://en.wikipedia.org/wiki/CIE_1931_color_space https://en.wikipedia.org/wiki/CIE_1931_color_space

Miiko 5y ago
According to what I could google, red sensitivity curve does have a left bump: https://www.researchgate.net/figure/Normalized-spectral-sensitivity-of-retinal-rod-and-cone-cells_fig7_265155524 https://www.researchgate.net/figure/Normalized-spectral-sens... So this is not a misunderstanding, but the real reason why magenta (red+blue) is perceived similar to violet (shorter wavelength than blue) by human eye.
- dahart 5y ago
  What you’re finding is the CIE “color matching functions”, which is closely related but not the same thing as the raw physical response of the cones. Here is a paper describing the difference, and more current research than what you found: https://hal.archives-ouvertes.fr/hal-01565649/document https://hal.archives-ouvertes.fr/hal-01565649/document Look specifically at page 2, and the linear relationship between the color matching functions (where red has a left-bump), and the “fundamental sensitivity” plot, which is attempting to find raw physiological cone response to different wavelengths, rather than what the observer perceives.

dahart 5y ago

So what you mean is that any color that is a mixture of two wavelength that trigger two different cones is “close” in some sense to the perceived color of either one of those wavelengths? This doesn’t mean color is “circular” per se, right? I suppose color perception with 3 cones has a sort of circularity when we plot the color space in 2d, but I’m not sure that sufficiently explains our experience of color. Anyway, what you’re saying is that mixtures of two things can be perceived as close to one of the two inputs. The argument holds for any number of cones - if we had 10 cones, we could still trigger the two with the furthest apart wavelength response using a mixture of two separate wavelengths, and the result would be “close” to one of the two. But this also holds for any pair - the mixture of imagined cone #3 and imagined cone #6 would also yield a color that is “close” to #3, and also “close” to #6. Has anyone asked a tetrachromat this question? What would it mean if they disagree that red is close to purple? What would it mean if they agree?