Zveriki Zveriki

The World Through a Chick's Eyes. Part One: Color and Perception

Short version: only the sections with diagrams, the diagrams themselves and explanations of how to read them.

1. On the Color Vision of Humans and Chickens #

In the upper left corner are the sensitivity graphs of the human S, M, L receptors to a pure spectral wave. In the upper right corner is an animation of the full color gamut: all the colors a human can possibly see (well, at least as far as a display can reproduce them). At the bottom are 5 strips illustrating the human color gamut at different color saturation.

I will dwell separately on the last two diagrams, because the terminology presented there will be used further on. For color coding it is often convenient to use the HSV model — Hue, Saturation, Value.

Hue is encoded on a color wheel, where red, green and blue each have their own polar coordinate — the rotation angle that points to a color or a mix of colors. You can see such a color wheel in the animation in the upper right corner.

Value (brightness) is encoded as a linear quantity from 0 to 100%, where 0 is black and 100 is the lightest.

Saturation is encoded, like brightness, from 0 to 100%, where 0 saturation means the absence of any pronounced hue (shades of gray), and 100% is the most vivid color.

Later in this text I will often mention saturation, and sometimes brightness as well.

2. The Spectral Color Space of the Chicken #

Let us look at the sensitivity diagram of the chicken's visual receptors. Besides S, M, L, as in humans, the chicken has a fourth type of receptor, sensitive to ultraviolet, called V cones. As on the human cone sensitivity graph, all curves are normalized to the sensitivity of the most sensitive receptor; in the chicken, unlike in humans, this is V.

Let us immediately settle how we will display the UV component of color. Purple (magenta) is already “taken” in human vision as the mixture of red and blue, but we still need a way to show ultraviolet. From here on, color noise will be used to show ultraviolet. When mixed with other colors, it slightly raises or lowers both the brightness and the saturation of the color it is mixed with. The density of the noise indicates the intensity of the effect on the V cone.

The diagram below shows the basic colors — red, yellow, green, cyan, blue and magenta — mixed with ultraviolet. For each color, the vertical axis shows brightness from 0 at the bottom to 100 % at the top, and the horizontal axis shows the increase of ultraviolet from 0 to 100 %.

Now let us compare the perception curves of the human and the chicken on one diagram:

On the perception diagram, the human curves are at the top, the chicken curves are at the bottom, and the colors of the spectrum are drawn as a strip in between.

This is the same “unrolled HSV cylinder” as for the human, only for the chicken. The horizontal axis is hue (the part of the spectrum seen by the S, M and L cones), the five strips are saturation 100, 75, 50, 25 and 0 %, and inside each strip the vertical axis is brightness from 100 % (top) to 0. The three panels show how the picture changes with the amount of ultraviolet U = 0, 50 and 100 %: ultraviolet is shown as animated color noise, and the denser the noise, the stronger the V cone is excited.

In the diagram one can see a pronounced dark band in the violet region. The point is that spectral violet looks reddish-blue to humans because short wavelengths weakly excite the L cones as well as the S cones; the same S + L combination from mixing red and blue light gives the non-spectral purple. For the chicken this does not apply, since it has a V type of receptor, but more on this below.

The bottom row of the diagram shows two curves on which the distortions are built (the horizontal axis shows the hues R, Y, G, C, B, M): orange is the degree of hue blur (the higher the curve, the worse the chicken distinguishes shades neighboring that color), light blue is color saturation relative to the human (the lower it is, the more the chicken falls behind the human in discrimination). They show that the chicken's discrimination is uneven. It distinguishes neighboring shades noticeably better than a human in the regions of blue (about 450 nm), green (510–520 nm) and yellow-orange (570–580 nm): for example, the chicken will tell two neighboring yellows apart about twice as well as a human. But in the region of light blue and cyan (470–500 nm), and of orange and red (600–640 nm), it distinguishes shades much worse than a human. For the violet–blue edge of the spectrum (below 430 nm) discrimination is weak in both.

Let us draw a diagram that illustrates the resolving power of human and chicken color vision across pure spectral shades:

In the center, the strips show a comparison of how strongly neighboring colors are discriminated: the worse a human distinguishes shades, the paler the spectrum is shown in those places (the upper strip in the center), and likewise for the chicken (the lower strip in the center).

3. Let Us Try to Draw Color Wheels for Humans and Chickens #

Let us try to model the color wheel “head-on”, placing around the circle in sequence red, green, blue, ultraviolet, green + UV, red + UV, then blue again, so that one wheel gives all possible pairwise combinations. For better expressiveness we will change the strength of the light for the transitional tones: where the distinguishability of shades is higher than in humans, saturation is boosted, and where it is lower, it is, conversely, reduced. And ultraviolet, as before, we will denote with color noise.

Since a human mostly does not see UV, only the color component from the mixing of the two colors remains in the right part.

4. What Happens in the Chicken When Two Pure Spectral Colors Are Mixed (for Example, in Reflected Light) #

To draw the diagram we will take two pure spectral colors pairwise (i.e. light of strictly one wavelength), mix them in different proportions and calculate the stimulus from the responses of the V/S/M/L cones. First we will get the result for the human, then pick the corresponding one for the chicken.

UV, as before, we denote with color noise (the density is set by the response of the UV cones). If in a given region the chicken perceives the color more strongly than the human, we make the color more saturated; otherwise, paler.

Axes: the horizontal and vertical axes are the wavelengths λ1 and λ2 of the pure colors (300–650 nm, including ultraviolet); each cell shows the result of mixing them 50 : 50 across the RGB channels.

For the human this is a simple average of the two colors. A human does not see ultraviolet: to the left of ~400 nm the blue channel simply decreases (down to zero by ~370 nm), and the other channels do not change.

On the right is the shift of the color balance in mixing: for red hues the color on the vertical (λ2) dominates in the mixture, for blue hues the color on the horizontal (λ1); brightness is the magnitude of the shift.

5. Color Wheels of Mixing Two Spectral Colors #

The same picture can be presented differently — as color wheels — and at the same time we can add different shares of the mixed-in pure spectral color. In each wheel the angle sets the wavelength of the first color λ1 (400–650 nm, clockwise from the left edge), and the radius sets the wavelength of the second color λ2 (400 nm at the center, 650 nm at the rim). The first color is taken with weight 100 %, the second with weight p = 25, 50, 75 and 100 % of the first. The human is shown at the top (weighted average of RGB), the chicken at the bottom: the saturation of colors is changed according to the perception weights, and UV is shown by noise (the density follows the sensitivity of the UV cone to the mixture).

6. The Human Color Space for Mixtures of Two Spectral Colors #

We move from illustrations to the full color space: we remove the illustrative boosting and weakening of saturation and add the colors as they are. We start with the human. The first square is the reference (the second color is mixed in at 0 %, i.e. the pure first color remains); in the others the share of the second color is 25, 50, 75 and 100 %. The horizontal axis is λ1 (the first color, weight 100 %), the vertical axis is λ2 (the second color, weight p of the first), both axes 400–650 nm. The stimulus is the sum of two spectral lines (power 1 and p), the responses of the S/M/L cones to it are added up, and the color is obtained from the total response (constant brightness).

7. The Chicken Color Space for Mixtures of Two Spectral Colors #

Now the same for the chicken. The squares are built the same way as for the human: a reference with 0 % of the second color and four squares with 25, 50, 75 and 100 %, both axes 400–650 nm. The responses of the four cones (V/S/M/L) are added up; the color from the S/M/L part is matched to the chicken colors from the other diagrams, without additional boosting or weakening of saturation. UV is shown by noise whose density is determined by the total response of the V cone.

8. Comparison of the Human and Chicken Color Spaces for Mixtures of Two Spectral Colors #

Now everything is ready to model the full color space of the chicken and compare how it differs from the human one. The human is shown on the left, the chicken on the right: the horizontal axis is λ1 (the first color), the vertical axis is λ2 (the second color), both axes 400–650 nm. In the animation, the admixture of the second spectral color smoothly grows from 0 to 100 % and just as smoothly decreases back. The stimulus is the sum of the responses of the S/M/L cones in the human and of the V/S/M/L cones in the chicken; the color from the chicken's responses is matched to the chicken colors from the other diagrams, and UV is shown by noise whose density is determined by the total response of the V cone.

Full article and references →

Author: ymik · November 19, 2026