What if humans could see ultraviolet and infrared?

Ultraviolet is one lens away, and some people already see it. Heat vision is a different story: the physics of a warm eye says a pigment for it would fire on its own, thousands of times a second.

Penstemon strictus flowers photographed in ultraviolet light, showing patterns invisible to the human eye
Photo: Michelle W. (鍾偉瑋) https://www.inaturalist.org/people/pufferchung, CC BY 4.0, via Wikimedia Commons

If humans could see ultraviolet and infrared, the ultraviolet half is nearly real already: your retina's cones respond down to about 300 nm, and it is the lens in front of them that blocks everything below roughly 380–400 nm, so people who have had the lens removed report seeing near-UV as a whitish violet. Near-infrared to about 1,000 nm is also within reach. Heat vision is the part physics forbids: a light-sensitive pigment cheap enough to be tripped by a 10 µm thermal photon would be tripped just as easily by the 37 °C body it lives in.

Key takeaways
  • 300 nm: how far into the ultraviolet human cones and rods can respond once the lens is out of the way.
  • About 380–400 nm: where the adult crystalline lens cuts ultraviolet off before it reaches the retina.
  • 1,000–1,060 nm: infrared laser pulses that 30 volunteers saw as faint green or red light in a 2014 PNAS study.
  • 9.3 µm: the peak of the glow your own body emits at 37 °C, by Wien's law. An eye tuned there would see itself first.
  • About 2,000 per second: the minimum thermal firing rate, per molecule, of a pigment able to catch 10 µm photons (my estimate from measured rhodopsin data).

The visible band, about 400 to 720 nm, is a thin slice of the spectrum, and it is not where the Sun's light ends. To answer the question properly I'll split it in two: what changes if we push the short end into ultraviolet, and what physics says about pushing the long end into infrared. The two halves give opposite verdicts.

Can a human eye see ultraviolet at all?

Yes, partly. The photoreceptors themselves can respond to wavelengths as short as about 300 nm, but the cornea absorbs below about 300 nm and the crystalline lens absorbs most light between 300 and 400 nm, so in a normal adult almost no UV reaches the retina. Remove the lens and the block is gone.

That condition has a name, aphakia. Before artificial intraocular lenses were invented after the Second World War, cataract surgery simply took the clouded lens out and left the patient aphakic. Vision researchers who are aphakic in one eye describe ultraviolet as a whitish blue to whitish violet, according to a 2023 review of the evidence by an optometrist writing for Healio. Not a new colour, then, because UV still lands on the same three cone types, mostly the short-wavelength S cones.

The story

In 1923 Claude Monet, then 82, finally agreed to cataract surgery. His right eye was down to light perception and his left measured 20/200; his late paintings had turned coarse and muddy. The surgeon removed the lens of the right eye. No implant existed, so Monet was left aphakic in that eye and never had the left one done. Afterwards his colour vision changed, and ophthalmologists have argued that the bluish cast in his last water lily canvases reflects ultraviolet that his unshielded retina could now register and that most viewers, with their lenses intact, cannot. Treat that last step as an interpretation, not a measurement: nobody tested Monet's spectral sensitivity. But the mechanism is sound and has since been measured in other aphakic observers.

Healio: The link between UV light, IOLs and Claude Monet's water lilies (2023)
Claude Monet's long painting Reflections of Clouds on the Water-Lily Pond, dominated by blues and violets
Monet's late water lily panels are heavy with blues and violets. One theory links this to his 1923 lens removal, which let near-ultraviolet reach his right retina.Photo: Claude Monet, Public domain, via Wikimedia Commons

Why does the lens block UV in the first place?

Douglas and Jeffery measured lens transmission in 38 mammal species and found that UV-blocking lenses occur in animals with high-acuity retinas, like ours. Short wavelengths scatter more inside the eye and focus at a different depth from long ones (chromatic aberration), so cutting them sharpens the image. Dogs, cats, ferrets and hedgehogs, with less sharp vision, have lenses that pass UVA.

The spectral transmission of ocular media suggests ultraviolet sensitivity is widespread among mammalsR. H. Douglas and G. Jeffery · Proceedings of the Royal Society B, 281: 20132995 · 2014

The authors measured how much light the cornea, lens and other ocular media of 38 mammal species, from 25 families and nine orders, let through across the UV and visible range. Many species, including dogs, cats, ferrets, hedgehogs and the okapi, transmit significant UVA and are therefore probably UV-sensitive even without a dedicated UV pigment. Lenses that block UV were found only in species with high-acuity retinas, supporting the idea that the block trades UV sensitivity for a sharper image. The limit: transmission tells you what reaches the retina, not what the brain does with it; behaviour has to be tested separately.

Read the paper (free)

What would the world look like in ultraviolet?

An eye that saw down to 300 nm would find a hazier sky, flowers with targets painted on them, and paint that splits into old and new. Each of these follows from a measurable property of the materials.

  • Flowers. Many petals that look uniformly yellow to us carry UV-absorbing centres. Bees see them because their three receptor types peak at about 344, 436 and 544 nm, against roughly 420, 530 and 560 nm for human cones.
  • White paint. In reflected-UV photography, zinc white and titanium white appear dark because they absorb UV, which is how conservators spot modern retouching on old canvases. A UV-seeing gallery visitor would see those repairs as grey patches.
  • The sky and distance. Air scatters short wavelengths far more strongly, so the blue haze over a distant ridge would thicken into a bright veil (worked out below).
Horsetail milkweed flowers photographed in ultraviolet light, showing strong contrast between parts of the flower
Horsetail milkweed photographed through a UV-pass filter. Patterns like these are invisible to normal human eyes but obvious to bees.Photo: Michelle, CC BY 4.0, via Wikimedia Commons
Peak sensitivity of the three colour receptors: human vs honeybee
See the data as a table
Human cones ( nm)Honeybee photoreceptors ( nm)
Short (S)420344
Medium (M)530436
Long (L)560544

Source: Vasas et al. 2017 (bee peaks); Wikipedia: Cone cell (human typical values) · Human values are typical peaks and vary between people; the lens, not the S cone, sets the human UV limit.

Multispectral images of flowers reveal the adaptive significance of using long-wavelength-sensitive receptors for edge detection in beesVera Vasas, Daniel Hanley, Peter G. Kevan and Lars Chittka · Journal of Comparative Physiology A, 203: 301–311 · 2017

Using a multispectral image database of flowering plants, the team computed how bee S (about 344 nm), M (436 nm) and L (544 nm) receptors would respond to real flowers and leaves. The long-wavelength receptor gave the most consistent signal, with up to four times the signal-to-noise ratio of the other two, which explains why bees use it alone for edges and motion and save the UV-inclusive colour vision for telling flower species apart. The limit: these are modelled receptor responses from images, not recordings from bees in the field.

Read the paper (free)

Worked example: how much hazier is the sky in UV?

Molecular scattering in air follows Rayleigh's law: particles much smaller than the wavelength scatter light in proportion to one over the wavelength to the fourth power. It is why the sky is blue and sunsets are red.

\[ \frac{I_{350}}{I_{450}} = \left(\frac{450\ \text{nm}}{350\ \text{nm}}\right)^{4} = (1.286)^4 \approx 2.73 \]

So 350 nm light scatters about 2.7 times as strongly as 450 nm blue. A UV-seeing eye would find distant hills washed out much sooner, which is one reason landscape photographers fit UV filters. Distance would read less clearly; the sky would be brighter and the horizon softer.

Calculated in this post from Rayleigh's law (scattering proportional to 1/λ⁴) and two wavelengths chosen as example inputs: with these example inputs (350 nm and 450 nm) the ratio is (450/350)⁴, about 2.73.

Can people already see infrared?

Only at the very near edge, and only with lasers. In a 2014 PNAS study, all 30 volunteers perceived infrared laser beams scanned across the retina, and with short picosecond pulses near 1,000 nm they reported a colour close to half the infrared wavelength, so 1,040 nm looked greenish.

The explanation is two-photon absorption: two low-energy infrared photons hit the same pigment molecule almost at once and together deliver the energy of one visible photon. It only happens at the enormous peak intensities of pulsed lasers, which is why sunlight, rich in near-infrared, does not look any different to you.

Human infrared vision is triggered by two-photon chromophore isomerizationGrazyna Palczewska, Frans Vinberg, Vladimir Kefalov, Krzysztof Palczewski and colleagues · PNAS, 111(50) · 2014

Researchers who had noticed pale green flashes while working with a 1,060 nm laser tested 30 healthy volunteers, who matched the perceived colour of scanned infrared beams to visible light. Above about 950 nm the perceived colour matched roughly half the stimulating wavelength. With short pulses (about 1 ps) near 1,000 nm, the power needed to see anything was more than 100 times lower than with 0.5 ns pulses, and responses scaled with the square of laser power: the fingerprint of a two-photon process, confirmed in mouse retinas and in purified rhodopsin. A person with achromatopsia also saw the beams, showing rods respond too. The limit: this works only with pulsed lasers; it is not everyday infrared vision.

Read the paper (free)
3004007001,000 nm3 µm10 µmWavelength (log scale)Normal human eyeAphakic eye (lens removed)HoneybeePulsed laser, two-photonThermal cameras, pit vipersBody glow peaks at 9.3 µm
Who sees what. The normal eye's band is set by the lens at the short end and by pigment chemistry at the long end. Heat imaging lives far to the right, out of reach of any photopigment.

Why can't eyes evolve thermal infrared vision?

Because a warm body is itself a bright infrared source, and a pigment sensitive enough to catch heat photons would be set off by heat. Two laws make this precise: Wien's displacement law says where a warm object's glow peaks, and the Boltzmann factor says how often random thermal jostling supplies a given energy.

Step one: where does a human glow?

Wien's law says the peak wavelength of a blackbody's emission is inversely proportional to its temperature, with a constant of 2,898 µm·K.

  1. Sun. Surface about 5,772 K. \( \lambda_{max} = 2898 / 5772 \approx 0.50\ \mu m \), right in the green. Our eyes are tuned to the light that is there.

    Calculated in this post from Wien's displacement constant (b = 2,898 µm·K) and the Sun's nominal effective temperature of 5,772 K, as defined by the IAU (Wikipedia: Effective temperature).

  2. You. Body core at 310 K (37 °C, the widely cited normal body temperature). \( \lambda_{max} = 2898 / 310 \approx 9.3\ \mu m \).

    Calculated in this post from Wien's constant (2,898 µm·K) and a core temperature of 310 K.

  3. The room. Walls at 300 K peak at 9.7 µm. Almost the same wavelength as you.

    Calculated in this post from Wien's constant (2,898 µm·K) and an assumed wall temperature of 300 K, a premise for this example.

\[ \lambda_{max} = \frac{b}{T}, \qquad b = 2898\ \mu\text{m·K} \;\Rightarrow\; \lambda_{max}(310\ \text{K}) \approx 9.3\ \mu\text{m} \]

By the Stefan–Boltzmann law, a 310 K blackbody radiates \( \sigma T^4 = 5.67\times10^{-8} \times 310^4 \approx 524\ \text{W/m}^2 \), against about 459 W/m² for a 300 K wall. The signal you want, the difference between a warm animal and its background, is small next to what the inside of your own eyeball is radiating at the retina.

Step two: what a heat photon can and cannot trigger

A photon's energy is \( E = hc/\lambda \), conveniently about 1,240 eV·nm divided by the wavelength. A 500 nm photon carries 2.48 eV; a 10 µm photon only 0.124 eV. Thermal energy at body temperature is \( kT = 8.617\times10^{-5} \times 310 \approx 0.0267 \) eV, so a 10 µm photon is worth less than five \( kT \).

Calculated in this post from hc (about 1,240 eV·nm, Photon energy), the Boltzmann constant (8.617×10⁻⁵ eV/K) and a body temperature of 310 K.

Rhodopsin works because its thermal barrier is high. Measurements on toad rods (Firsov et al., 2002) put the activation energy of spontaneous "dark events" at no more than 20–24 kcal/mol (the paper's own corrections use 22, roughly 0.95 eV per molecule) and the rate at about 5 × 10⁻¹² per molecule per second at 16.5 °C. A pigment can never need more energy to fire than the photon it is meant to detect delivers, so a 10 µm pigment must have a barrier of 0.124 eV or less.

\[ \frac{r_{10\,\mu m}}{r_{rhod}} \ge \exp\!\left(\frac{E_{rhod} - E_{photon}}{kT}\right) = \exp\!\left(\frac{0.954 - 0.124}{0.0267}\right) = e^{31.1} \approx 3\times10^{13} \]
  1. Scale rhodopsin to 37 °C. With a 0.954 eV barrier, going from 289.7 K to 310.2 K multiplies the rate by about \( e^{2.53} \approx 12.5 \): \( 5\times10^{-12} \to 6.3\times10^{-11}\ \text{s}^{-1} \).
  2. Apply the Boltzmann ratio. \( 6.3\times10^{-11} \times 3\times10^{13} \approx 2{,}000\ \text{s}^{-1} \) per molecule, as a lower bound.
  3. Count molecules. A toad rod outer segment holds about 3 × 10⁹ pigment molecules. That cell would register on the order of 10¹² false photons a second, in total darkness.

This is my estimate, and it assumes the same attempt frequency as rhodopsin; real pigments differ by a factor of ten or so either way. That does not rescue it, because the answer is off by about twelve orders of magnitude. A warm-blooded photochemical eye cannot see heat, because its own heat would fire the pigment.

Sensitivity check: where does the wall really fall?

Run the same bound across wavelengths. Below about 1,300 nm the photon energy exceeds rhodopsin's barrier, so nothing forces extra noise. At 1,400 nm the bound gives about 2 false events per rod per second, already comparable with dim starlight signals. At 2 µm it is roughly 50,000 per second, and at 4 µm about 6 × 10⁹ per second. The practical edge of photochemical vision in a warm eye sits near 1.3–1.5 µm, just past the edge the 2014 laser experiments probed.

Calculated in this post from hc (about 1,240 eV·nm, Photon energy) and the 0.954 eV barrier and 3 × 10⁹ molecules per rod from Firsov et al., 2002, at 310 K; the wavelengths 1,300, 1,400, 2,000 and 4,000 nm are example inputs.

Use this to check what you just read: pick a wavelength and see the photon energy and the minimum thermal noise a pigment for it would carry, then try Wien's law on any temperature.

Could a warm eye see this wavelength?

Verdict: possible with changes, for near-UV and near-IR only

Ultraviolet vision down to about 300 nm needs only the removal of the lens, which has already happened to millions of cataract patients; the price is a less sharp image and a retina exposed to UV damage. Near-infrared to about 1.3 µm is chemically possible with a new pigment. Thermal infrared at 8–14 µm is impossible for a warm photochemical eye: by the Boltzmann factor, a pigment for 0.124 eV photons would fire at least about 2,000 times a second per molecule from body heat alone. Heat sensing needs a different organ, as pit vipers and thermal cameras show.

Calculated in this post from the photon energy at 10 µm (0.124 eV, hc about 1,240 eV·nm, Photon energy) and rhodopsin's measured thermal rate and barrier (Firsov et al., 2002); the working is in the steps above.

What would photos, paint and skin look like to a UV-and-IR eye?

Every image made for human eyes would look wrong to it, because screens, inks and photographs are engineered only for the 400–700 nm band. The table sets out the main differences, band by band.

BandPhoton energyWhat you would noticeWho sees it now
Near-UV, 300–400 nm3.1–4.1 eVFlower bullseyes, dark modern white paint, stronger hazeBees, many birds, aphakic humans
Visible, 400–720 nm1.7–3.1 eVOrdinary colourHumans
Near-IR, 0.7–1.1 µm1.1–1.8 eVLeaves glowing bright, as in IR photosConverted cameras, satellites
Short-wave IR, 1–3 µm0.4–1.2 eVHot metal glowing, water darkCooled InGaAs sensors
Thermal IR, 8–14 µm0.09–0.16 eVBody heat, warm enginesPit vipers, thermal cameras
Near-infrared photograph of rocks and trees in which the foliage appears bright white
A near-infrared photograph from a converted camera. Healthy leaves reflect a large share of near-infrared light, so foliage glows white; this is the same physics satellites use to map crops.Photo: David from Colorado Springs, United States, CC BY 2.0, via Wikimedia Commons

Leaves, screens and photographs

Near-infrared vision would make every tree look lit from within. NASA's vegetation index works because healthy leaves absorb most visible light (0.4–0.7 µm) for photosynthesis but reflect a large share of near-infrared (0.7–1.1 µm), scattered by the leaf's cell structure. The index is simply \( \text{NDVI} = (\text{NIR} - \text{VIS}) / (\text{NIR} + \text{VIS}) \), and dense green canopy scores 0.8–0.9.

Meanwhile, your phone screen would look oddly dim. Displays emit only red, green and blue primaries, and an sRGB image has no channel at all for UV or IR. A photograph of a sunflower would lose its UV bullseye; a printed poster would have no infrared data. This is the same problem I covered in RGB vs CMYK, one band wider: every reproduction is a reduction.

Below 300 nm300 to 400 nm400 to 720 nm and near-IRCornea: stops UVB and UVCLens: stops most UVARetina: cones responddown to about 300 nm
The eye's filters in order. The cornea stops the harshest ultraviolet, the lens stops most of the rest, and the retina behind them could respond further into the UV than we ever experience.

What do cameras already capture that eyes can't?

Almost every camera you own sees further than you, and manufacturers spend money to stop it. Silicon sensors respond well into the near-infrared, so cameras carry a "hot mirror" filter to block infrared that would otherwise wash out colour. Take that filter out and you have a full-spectrum camera, the tool conservators use for UV and IR examination of paintings.

  • Reflected-UV photography needs a full-spectrum camera, a UV-pass filter and UV lamps around 365 nm. It maps zinc and titanium white retouching.
  • Infrared photography through a visible-blocking filter shows underdrawings beneath paint layers and the glowing foliage above.
  • Thermal imaging uses a different sensor entirely, usually a microbolometer, because silicon cannot detect 10 µm light.

If you shoot RAW, you already work with sensor data that your eyes never saw directly. My guides to RAW and dynamic range and white balance and Kelvin cover how that data is squeezed into the human band, and how phone cameras compute a photo shows how much processing sits between sensor and screen.

Thermal camera image of a clothed person, with warm skin shown in bright colours and cooler clothing in dark colours
A thermal image renders 8–14 µm radiation in false colour. No eye, only a cooled or bolometric sensor, can make this picture.Photo: Emilian Robert Vicol (byref), CC0, via Wikimedia Commons
The lens is a UV filter we are born with. The heat wall is one we cannot remove, because we are the heat.From this post

Would UV be a new colour?

For an aphakic eye, no: UV excites the existing S cones, so it maps onto violet-white. A genuinely new hue would need a fourth cone type and the wiring to compare it with the others, as bees and birds have. The 2025 "olo" experiments, where lasers stimulated only M cones in five people, show the brain can report a colour outside normal experience when the cone signals are unusual. Design systems built on perceptual colour spaces such as OKLCH assume three cones; a fourth would break every colour model we have.

What if next?

Questions people ask

What colour would ultraviolet be if we could see it?

People without a lens describe near-UV as a whitish blue or whitish violet, because it stimulates the same S cones as violet light. A truly new colour would need a fourth receptor type, which humans do not have.

Can humans see infrared light?

Not in daily life. In lab tests, pulsed infrared lasers around 1,000–1,060 nm were seen as faint green or red flashes because two photons combined in one pigment molecule. Ordinary infrared from the Sun or a remote control is invisible.

Do people see more colours after cataract surgery?

Many notice colours look bluer and brighter, because the yellowed old lens filtered blue. Most modern implants block UV, so true UV vision after surgery is uncommon. Any change should be checked by an eye doctor.

Why can snakes see heat but we can't?

Pit vipers do not use a pigment. Their pit organ is a thin membrane warmed by infrared, read by very heat-sensitive TRPA1 channels. It is a thermometer, not an eye, which sidesteps the thermal-noise problem.

More from the studio: the Kala world page collects every design and photography post on this site.

Sources

  1. Douglas and Jeffery, Proc. R. Soc. B: The spectral transmission of ocular media suggests ultraviolet sensitivity is widespread among mammals (2014) (source for lens UV blocking and high-acuity link)
  2. Palczewska et al., PNAS: Human infrared vision is triggered by two-photon chromophore isomerization (2014) (30 volunteers, half-wavelength colour, pulse comparison)
  3. Firsov, Donner and Govardovskii, Journal of Physiology: pH and rate of dark events in toad retinal rods (2002) (rhodopsin thermal rate 5 × 10⁻¹² /s, 20–24 kcal/mol barrier, 3 × 10⁹ molecules per rod)
  4. Vasas et al., J Comp Physiol A: Multispectral images of flowers (2017) (bee receptor peaks 344, 436, 544 nm)
  5. Healio: The link between UV light, IOLs and Claude Monet's water lilies (2023) (Monet's 1923 surgery, aphakic UV perception, 300 nm limit)
  6. Wikipedia: Cone cell (typical human cone peaks)
  7. CHSOS: Technical photography of white pigments (zinc and titanium white dark in UVR)
  8. CHSOS: Ultraviolet reflected photography (UV-pass filter, 365 nm lamps)
  9. Cosentino, Conservar Património 21: Practical notes on ultraviolet technical photography for art examination (2015) (full-spectrum camera modification)
  10. NASA Earth Observatory: Measuring vegetation (NDVI and EVI) (leaf NIR reflectance, NDVI formula)
  11. Gracheva et al., Nature: Molecular basis of infrared detection by snakes (2010) (TRPA1 in pit organs)
  12. Cell Press: Near-infrared colour vision in humans enabled by upconversion contact lenses (2025)
  13. Straight Dope Message Board: What if we could see ultra violet or infrared? (reader question)
ultraviolet visioninfrared visionaphakiaWien's lawBoltzmann factorrhodopsinbee visionUV photography
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of Artblog.

Keep reading in Artblog

From the other blogs