Screens before sleep: what is solid, what is not

Few sleep topics have travelled further from their evidence than screens at night. The core finding is genuine and well established. Several of the claims built on top of it are much weaker than their confidence suggests, and separating the two is more useful than repeating either the alarm or the backlash.

What is solid: evening light suppresses melatonin and shifts the circadian clock later. This is thoroughly established chronobiology, described in this section's articles on melatonin and the circadian rhythm. Light in the hours before bed tells the internal clock that the day is not over, delaying the melatonin rise and pushing the whole sleep window later. A controlled study comparing reading from a light-emitting e-reader before bed with reading a printed book found exactly this: with the e-reader, participants took longer to fall asleep, showed suppressed melatonin, had their circadian timing delayed, and were sleepier the following morning. The mechanism is real and it has been demonstrated directly.

What is less solid than the headlines: that the blue wavelength specifically is the main culprit for most people. The underlying science is correct — the retinal cells that report light to the circadian clock are most sensitive to short-wavelength light — but the leap from that to the everyday claim about phones is where things stretch. What reaches those cells depends on total intensity as much as on spectrum, and a phone at arm's length in a lit room delivers far less light than the outdoor daylight a person is exposed to routinely. The effect of a typical device on a typical evening is real but considerably smaller than the framing implies.

And the fix that follows from the blue-light story has underwhelmed in testing. Night-mode filters and amber-tinted glasses should work if wavelength is the whole mechanism, and trial results on them have been mixed and generally modest. Studies comparing devices with and without night mode have often found little difference in sleep outcomes. That is an awkward result for the popular account, and it should update it: if removing the blue does not fix the problem, the blue was probably not the whole problem.

What likely matters at least as much is what the device does to you rather than what it emits. Content is arousing. Work email, an argument, an unresolved news cycle, a game, an algorithmic feed engineered to defer the moment you stop — these raise alertness and rumination, and pre-sleep arousal is one of the better-supported obstacles to falling asleep. There is also the simplest mechanism of all, and possibly the largest: displacement. A device does not need to affect your physiology at all to cost you an hour of sleep if you were going to be asleep and instead you were not. Time-use research consistently finds screen time near bed predicts later bedtimes, and a lost hour is a bigger effect than anything spectral.

The practical conclusion is therefore different from the usual advice, and less about hardware. Total evening light matters, so dimming the room is a bigger lever than adjusting one small screen in it. Morning daylight matters more than most people expect, and it pulls the clock in the useful direction. What you do on the device deserves more attention than which filter it has. And the hour you were going to be asleep is worth defending on its own terms.

For dreams specifically, the connection is indirect but real. A clock shifted later and a shorter night both cut disproportionately into the REM-rich hours before waking, which is where most remembered dreaming comes from. People who habitually lose an hour to a screen at night are not dreaming differently so much as getting less of the part of the night dreams are recalled from — an effect covered in this section's articles on sleep cycles and sleep debt.

Frequently asked

What this is based on

• The controlled study comparing light-emitting e-readers with printed books before bed
• Chronobiology research on light suppression of melatonin and circadian phase shifting
• Trials of blue-light filtering interventions and their mixed results