Your eyes have a clock sensor, and it is tuned to blue

What is melanopsin?

Melanopsin is a light-sensitive protein made by a small group of cells in the retina called intrinsically photosensitive retinal ganglion cells, usually shortened to ipRGCs. Most retinal ganglion cells only pass on signals that start in the rods and cones. These ones can also detect light by themselves, and the main place they send that information is the suprachiasmatic nucleus, the cluster of neurons in the hypothalamus that keeps the body’s 24-hour time (Berson et al., 2002; Hattar et al., 2002).

When the human gene was first described, the team surveyed 26 sites in the body and found melanopsin expressed only in the eye. Inside the retina it sat in the ganglion and amacrine cell layers, not in the photoreceptors that start vision, and the authors suggested from that location alone that it might handle jobs like setting circadian rhythms and suppressing melatonin rather than forming images (Provencio et al., 2000). The experiments that followed bore that guess out.

How scientists found a third light sensor

The clue came from blind people. In 1995 a team tested 11 patients who had no conscious perception of light at all, exposing them to bright light and measuring their melatonin. In three of them, melatonin fell about as much as it did in sighted volunteers, and when two of those three were tested again with their eyes covered, it did not fall. Those three slept without trouble and kept a normal daily rhythm. Most of the others had no melatonin response and a history of insomnia (Czeisler et al., 1995). Something in those three patients’ eyes was still telling the clock about light, even though the parts of the eye that build pictures no longer worked.

The gene arrived in 2000 (Provencio et al., 2000). Two years later, two papers in Science closed the loop. One showed that the retinal ganglion cells wired to the master clock respond to light directly, even when all input from the rods and cones is blocked, and that their sensitivity, colour tuning and slow response matched those of the clock’s light-setting mechanism (Berson et al., 2002). The other showed that every ganglion cell with that ability carried melanopsin, and traced their wiring in mice to the master clock and to the brain areas that control the pupil (Hattar et al., 2002).

Knockout mice then showed what the sensor does. Mice bred without melanopsin still kept a daily rhythm, but brief pulses of light reset their clocks far less than in normal mice (Panda et al., 2002). Their pupils also behaved oddly: normal in dim light, yet unable to close fully in bright light, which suggested the melanopsin cells and the rod and cone system share the work between them. The ganglion cells wired to the master clock in those mice were no longer sensitive to light by themselves, although their number, shape and wiring were unchanged (Lucas et al., 2003). In monkeys whose rods and cones had been blocked with drugs, the pupil still responded to light, and the slow, lasting squeeze of the pupil after a light switches off turned out to be driven mainly by the melanopsin cells, in monkeys and in people (Gamlin et al., 2007).

The most direct human evidence came in 2007, from two people who had lost their rods and cones entirely. In one of them, short-wavelength light suppressed melatonin, reset the clock and made him more alert than green light at 555 nm, the colour ordinary daytime vision is most sensitive to. In the other, the pupil was most sensitive at 480 nm, matching the melanopsin cells, and she could correctly report a faint flash at that wavelength but at no other (Zaidi et al., 2007). With no working rods or cones, the only sensor left that could explain those results was the one containing melanopsin.

The discovery, in order

Why is the clock sensor tuned to blue?

Every light sensor in the eye has a colour it responds to best. Daytime vision, run by the cones, is most sensitive at about 555 nm, in the green and yellow part of the spectrum (Lockley et al., 2003). The response that reaches the body clock is not. In 2001 two independent teams built action spectra for melatonin suppression, testing one narrow band of light at a time in the middle of the night.

In the first, 72 volunteers went through 627 tests, and the most potent band for suppressing melatonin was 446 to 477 nm, which is blue (Brainard et al., 2001). In the second, 22 volunteers went through 215 trials, and their curve showed the same strong short-wavelength sensitivity, quite unlike the curves of the rods or the cones (Thapan et al., 2001). The two groups reached the same answer: the photoreceptor that talks to the clock is tuned towards blue, and it is not one of the receptors we see with.

The next question was whether blue light also shifts the clock more. When volunteers sat under blue light at 460 nm for 6.5 hours, their clocks shifted two-fold more than under green light at 555 nm with the same number of photons, and their melatonin was suppressed about twice as much (Lockley et al., 2003). That is a ratio, and it is the honest way to quote this paper: the study reports the comparison between two colours, not a percentage that would apply to anyone’s bedroom.

The tuning shifts a little with age. The lens filters out more blue light as the eye ages, yet in a small study the older group’s melatonin was suppressed just as much as the younger group’s, and their peak sensitivity moved only slightly, from 484 nm in the young to 494 nm in the older participants (Najjar et al., 2014). The authors suspect the system compensates somehow, and say they do not yet know how.

Do rods and cones matter too?

They do. The melanopsin cells receive signals from the rods and cones as well as making their own, so what reaches the clock is a blend (Gooley et al., 2010).

A 2010 study tested this directly by giving people blue light at 460 nm or green light at 555 nm near the time their melatonin normally rises, at a range of intensities. At the start of the exposure, green light suppressed melatonin as well as blue did, which pointed to the cones helping. As the exposure went on, the green light’s effect faded compared with the blue. The authors concluded that cones contribute a lot at the beginning of a light exposure and at low light levels, while melanopsin is the main driver during long, bright exposures (Gooley et al., 2010). The mouse work had already hinted at the same division of labour (Lucas et al., 2003).

So “only blue light matters” is wrong. Brief or dim light of other colours can reach the clock through the cones. What the 2010 study found is that melanopsin is the main driver during long, bright exposures (Gooley et al., 2010), and melanopsin leans towards blue.

What does the clock sensor control?

The melanopsin cells do more than set the clock. The studies on this page tie them to five separate things your body does with light:

• Clock timing. Light at the wrong time of day shifts the clock later or earlier, and short-wavelength light shifts it more than green light (Lockley et al., 2003; Panda et al., 2002).
• Melatonin. Light at night suppresses it, most strongly in the blue band (Brainard et al., 2001; Thapan et al., 2001).
• The pupil. The lasting constriction after bright light depends mainly on the melanopsin cells (Lucas et al., 2003; Gamlin et al., 2007).
• Alertness. In a man with no working rods or cones, blue light raised alertness more than green light did (Zaidi et al., 2007).
• Rudimentary visual awareness. A woman with no working rods or cones could correctly report a faint flash near 480 nm but not at other wavelengths (Zaidi et al., 2007).

That last point means the same light that suppresses melatonin can also raise alertness (Zaidi et al., 2007). How much ordinary evening light it takes, and how alert it makes people, is the subject of another article in this library, ordinary evening light moves your body clock.

Measuring light the way the clock sees it

Lux, the unit on a light meter, weights light by how bright it looks to daytime vision, so it is dominated by the green and yellow light the cones like best. Once it was clear that a different sensor runs the clock, researchers from across the field proposed a new way to measure light that accounts for every input the clock receives, melanopsin included, and offered simple suggestions for lighting designers and regulators (Lucas et al., 2014).

One measure of this kind is melanopic illuminance. It counts how much light reaches the eye and weights it by how strongly each wavelength drives melanopsin. When one researcher pooled 19 laboratory studies, melanopic illuminance was the best available predictor of how the human circadian system responded across all of them (Brown, 2020). Two lights that look equally bright can differ a lot on this scale, which is why a lamp’s colour and its brightness both matter at night.

What this means for screens at night

A computer screen is a source of light at eye level, often for hours in the evening. When a tablet, an e-reader and a phone were measured, all three gave off light with a strong blue peak (Gringras et al., 2015). Two things lower how much a screen drives the clock sensor: making it warmer, so less of its light falls in the blue band, and making it dimmer, so less light arrives at all.

The trials that tested only the first lever are a useful warning. On an iPad, both of Night Shift’s settings still suppressed melatonin, the low and high colour-temperature settings did not differ significantly, and the authors concluded that changing a screen’s colour without changing its brightness may not be enough (Nagare et al., 2019). The full set of screen studies, including the ones that found no effect, is on the page on screens at night and sleep, and the Night Shift trials have a page of their own.

Sundown is built on exactly this. It warms a Mac’s display down to a deep red, where Night Shift stops at a warm white, and dims it as well, so it works on both the colour and the amount of light. Dr. Alexander Wunsch measured it across the blue band, 380 to 500 nm, the range that takes in melanopsin’s peak, and at deep warmth Sundown eliminates blue light. The method is on how we tested Sundown.