Horse color vision: carrots, colored doors, and a different palette
Horse color vision has been tested with something more persuasive than a stable-yard opinion: two doors and a carrot reward. In a study published in 2008, Lina Roth, Anna Balkenius and Almut Kelber trained horses to choose between blue and green panels. The rewarded door opened onto carrot pieces; the other remained locked. The important feature was the changing brightness of the panels. A horse that kept selecting the rewarded color could not succeed simply by following the lighter patch. Color discrimination was a measurable behavior, with a food reward waiting behind it.
The team used five brightness versions of blue and seven of green. Across the test combinations, the rewarded panel could be brighter or darker than its partner. Three horses began testing, but motivation limited the results from two. Chap, the remaining gelding, selected the rewarded blue panel in 31 of 40 presentations at a light intensity comparable to moonlight. When he failed at a dimmer level, a return to brighter conditions restored his performance. That control helped distinguish a visual limit from a horse that had simply stopped participating.
The biological machinery differs from the familiar human arrangement. Horses have two cone types involved in color perception, while most humans have three. Roth and colleagues used cone sensitivity peaks of 428 and 539 nanometers in their calculations. These are peaks in the receptors' responses to light, rather than a declaration that horses see exactly two isolated paint colors. Rod cells provide greater sensitivity in dim conditions, while cones contribute color discrimination. The reflecting layer behind the retina also helps return light toward the receptors.
A 2025 Frontiers review by Lina Roth and Paul McGreevy makes that distinction especially clear. Horse color vision extends along a continuous range associated with blue and green-yellow sensations. Two cone types do not divide every object into two rigid categories. The review describes a neutral region near 480 nanometers, where light can stimulate the cone types in proportions resembling the response to gray. Certain blue-green samples can consequently be difficult to separate from gray, even though other colors are distinguishable. Difficulty with a particular comparison therefore cannot establish a wholly colorless visual world.
That review also explains why a brightness control matters. A single photoreceptor registers absorbed light; color information requires comparison between receptors with different spectral sensitivities. Two equally bright objects can differ in their reflected spectrum. Conversely, an apparent boundary in brightness can come from a shadow rather than from two different objects. Comparing spectral information supplies another cue to their separation. This is the mechanism that makes a carefully controlled color task stronger evidence than a casual observation that a horse noticed one bright object and ignored another.
The sporting consequences appear in Sarah Catherine Paul and Martin Stevens's original study of racing obstacles, published in Applied Animal Behaviour Science in 2020. They assessed orange fence markers at 11 British racecourses using models of equine color and brightness vision. They also examined alternative colors under different light and weather conditions. Orange often provided relatively poor contrast against the surrounding scene. White, fluorescent yellow and bright blue offered stronger visibility in the comparisons, although their relative advantages depended on the background and illumination. A color conspicuous to a person was not automatically the strongest marker for a horse.
Their jumping trials tested behavior as well as modeled visibility. Horses encountered fences with orange, fluorescent yellow, bright blue or white takeoff boards and midrails. Bright blue increased the takeoff angle, white increased takeoff distance, and fluorescent yellow shortened landing distance relative to orange. These were changes in measured jumping responses. The experiments did not establish that any single paint color would prevent every fall, or that visibility alone controls a jump. Experience, the surrounding scene and other features can also influence performance.
Rørvang, Birte Nielsen and Andrew McLean's 2020 review places those findings alongside earlier obstacle studies. It describes blue obstacles as easier to detect than green ones against grass, and emphasizes the role of the surroundings. A rail and its background form a visual comparison; their color names alone leave out that relationship. Horse color vision therefore matters to the whole approach to an obstacle, including which components contrast with the turf and which stand against the fence itself.
The same review distinguishes seeing objects in low light from distinguishing their colors. Horses can retain useful sensitivity to objects after color comparisons become difficult. The advantage of a large eye, abundant rods and a reflecting retinal layer does not mean every color remains available in darkness. The practical picture is a visual system with real color information, specific confusions and changing performance as illumination falls. A course painted for human eyes can overlook all three.
- Roth, Balkenius and Kelber, The Absolute Threshold of Colour Vision in the Horse, PLOS ONE (2008) — pmc.ncbi.nlm.nih.gov
- Paul and Stevens, Horse vision and obstacle visibility in horseracing, Applied Animal Behaviour Science (2020) — pmc.ncbi.nlm.nih.gov
- Roth and McGreevy, Horse vision through two lenses: Tinbergen’s Four Questions and the Five Domains, Frontiers (2025) — www.frontiersin.org
- Rørvang, Nielsen and McLean, Sensory Abilities of Horses and Their Importance for Equitation Science, Frontiers (2020) — www.frontiersin.org