How birds see color

Hart (2001) reviewed bird photoreceptors across many orders: birds have five types of visual pigment and seven types of photoreceptor, namely rods, double cones and four types of single cone. Colored oil droplets inside the cones filter the light before it reaches the pigment, and both the pigments and the droplets vary with a species' habitat and ancestry.

In the budgerigar, the four single-cone pigments peak at about 370, 445, 508 and 565 nm (Goldsmith and Butler, 2005). In their tests, adding a small amount of ultraviolet (365 nm) to blue or yellow light changed its color dramatically for the birds.

In the budgerigar, the three longer-wave single cones each hold a colored oil droplet that works as a long-pass filter (it lets through only light of longer wavelengths), with cut-offs at about 445, 500–520 and 570 nm; the ultraviolet cone has a clear droplet (Bowmaker and colleagues, 1997).

A filter in front of a cone throws away light, so why have one? Vorobyev (2003) weighed the cost against the gain: the droplets reduce the overlap between neighboring cones, and his calculations show that this increases the number of object colors a bird can tell apart.

Our own calculation mostly points the same way: in three of the four pairs below, the droplets pull the two colors clearly further apart, while yellow and gray stay about where they were. We took a few everyday color pairs, guessed each color's spectrum, lit it with daylight and worked out what three sets of cones would catch: yours, the budgerigar's without droplets and the budgerigar's with them. All three are drawn by the same rule (longest cone as red, middle as green, shortest as blue, white stays white), so the rows can be compared with each other, but none of them is a picture of what a bird actually sees. The number is the color difference (ΔE 2000) between the two drawn swatches. On the scale we use across this site, below 3 is a close match, 6 or more is a visibly different color, and 10 or more is clearly different. Because every row follows this same rule, even «Your cones» looks off (the red ball comes out brown): compare the numbers, not the look.

The plumage of the budgerigar reflects near-ultraviolet light (300–400 nm). Pearn, Bennett and Cuthill (2001) removed that ultraviolet with filters: females consistently preferred males that reflected it, and the effect came from a change in the perceived hue, not in brightness.

Purple is a «nonspectral» color for us: no single wavelength of the rainbow looks purple; it needs blue and red light together. Birds, with a fourth cone sensitive to violet or ultraviolet, can have more such colors, for example ultraviolet plus green or ultraviolet plus red. Stoddard and colleagues (2020) showed in experiments with wild broad-tailed hummingbirds that they discriminate a variety of nonspectral colors. These birds most likely have the violet type of fourth cone: «Hummingbirds are believed to possess the VS-type visual system», the paper says. From 965 plumage and 2,350 plant spectra the authors estimated that hummingbirds likely see about 30% of the plumage colors and about 35% of the plant colors as nonspectral, against 7% and 19% for people.

The two swatches below are the same on your screen, because a screen emits no ultraviolet. If one of them also reflected ultraviolet, as many feathers and flowers do, a bird would see two different colors.