How they see
How does a zebrafish see the world?
Four cone channels and a near-panoramic view
A zebrafish carries four cone classes—ultraviolet, blue, green and red—in a retina built for bright shallow water. Its side-set eyes survey 339 degrees around the body, while adult behavioural tests place spatial acuity at only 0.56 to 0.58 cycles per degree. Fine detail blurs, but colour and near-panoramic coverage remain central.
See through their eyes
Human ≈ 60 cpd (20/10)
How sharp is a zebrafish's vision?
Its eyes resolve about 0.57 cycles per degree (No direct Snellen equivalent; 0.56–0.58 cpd behavioural OKR), while a human reaches around 60 cpd (20/10). Its field of view spans about 339°.
How does a zebrafish see a human?
Its eyes resolve about 0.57 cycles per degree against your 60, so a face has to be roughly 105 times closer for it to pick out the detail you do. Your outline and your movement read clearly; your expression does not.
Colour comparison
How each colour shifts when seen through this animal's eyes, using the same colour model as the app. Tones a screen can't emit (such as ultraviolet) are shown as an approximation.
What colours can a zebrafish see?
Four cone classes feed colour-opponent retinal circuits. The ultraviolet channel is especially specialized in the upper frontal field for detecting UV-bright prey in larvae.
Photoreceptors
Tetrachromat (UV, blue, green and red cones)
Cone peak sensitivities
- UV362 nm
- S (blue)415 nm
- M (green)480 nm
- L (red)570 nm
How does a zebrafish see in the dark?
A duplex retina contains rods and cones; rod responses are roughly 40–220 times more photosensitive than cone responses, while colour is a bright-light function.
How does a zebrafish's vision compare to a human's?
Humans use three cone classes and do not see ultraviolet; zebrafish use four and cover 339 degrees rather than roughly 200. An RGB screen can only use false colour and measured blur to communicate that difference, never reproduce it literally.
What is special about a zebrafish's vision?
- Four cone classes including ultraviolet
- 339° adult visual coverage with only a 21° rear blind sector
- Fronto-dorsal acute zone; larval UV cones specialize for prey capture
What the research shows
Zebrafish cone peaks depend on how they are measured. The canonical in-situ values of 362, 415, 480 and 570 nm, which the page displays, come from Robinson et al. (1993), whereas Sato et al. (2025), recording from dark-adapted, dissociated adult cones, obtained 374, 401, 484 and 553 nm. Neither set is wrong; opsin expression and preparation shift the result. Connaughton & Nelson (2010) addressed a different question, whether all four signals are actually used, by recording tetraphasic and UV-dominated responses in horizontal cells, the first stage of retinal processing after the photoreceptors.
Acuity shows the gap between behaviour and anatomy. Tappeiner et al. (2012) tested the optokinetic reflex at stimulus speeds of 10 to 30 degrees per second and measured 0.56–0.58 cycles per degree, while Pita et al. (2015) calculated a ceiling of 1.89 cpd from peak ganglion-cell density and lens size; the simulation uses the behavioural 0.57 cpd. Pita et al. (2015) also mapped a 33° binocular field and a 21° blind sector at the rear. No flicker-fusion number is given, because optokinetic results are task-dependent and not interchangeable with one, and the larval ultraviolet gain reported by Yoshimatsu et al. (2020) is not assumed to hold in adults. A single phone camera cannot capture a 339° field.
Guides: Animals that see ultraviolet · Which animal has the best eyesight? · All guides
Scientific sources
- Robinson et al. (1993), PNAS 90:6009–6012
- Connaughton & Nelson (2010), J Neurophysiol 104:2906–2921
- Tappeiner et al. (2012), Front Zool 9:10
- Pita et al. (2015), PeerJ 3:e1113
- Yoshimatsu et al. (2020), Neuron 107:320–337.e6
- Baier & Scott (2024), Annu Rev Neurosci 47:255–276
- Sato et al. (2025), Sci Rep 15:11651
Download Animal Vision
Point your camera at anything and watch the world transform into this animal’s vision in real time.
