A sound that helps a new ending stick

Imagery rehearsal asks you to do something simple: rewrite a recurring nightmare while you are awake, then rehearse the new version for a few minutes a day. It is the treatment the American Academy of Sleep Medicine recommends for nightmare disorder.^[5] A team in Geneva asked a further question: could a quiet sound help the new version reach the dreaming brain?^[1]

The idea: targeted memory reactivation

During sleep, the brain replays and strengthens some of what it took in during the day. Targeted memory reactivation, or TMR, tries to nudge that process. You learn something while a cue is present, such as a sound or a smell, and the same cue is presented again, softly, while you sleep. The cue seems to prompt the brain to revisit that particular memory.

A 2020 meta-analysis pooled 91 such experiments involving about 2,000 people. Overall, cues during sleep improved later memory by a small to moderate amount. The benefit was clear when cues were played in non-REM sleep, but in the studies available then, cues during REM sleep did not show a reliable effect.^[3] That is what makes the Geneva study interesting: REM is where most vivid dreams, and most nightmares, happen.

What the Geneva team did

Sophie Schwartz, Alice Clerget and Lampros Perogamvros recruited 36 adults with nightmare disorder at the sleep centre of the Geneva University Hospitals. They were 20 to 35 years old, had more than one nightmare a week, and said their nightmares were not linked to a traumatic event. People with severe depression, insomnia, psychosis, anxiety disorders, or medicines likely to cause nightmares were not included.^{[1, 2]}

1. One session of imagery rehearsal for everyone. Each person chose their most frequent nightmare, changed it so it was no longer distressing, and imagined the new version for five minutes.
2. A sound for half of them. Chosen at random, half heard a soft one-second piano chord every ten seconds while they imagined the new version. The other half did the same imagery in silence.
3. Two weeks at home. Everyone rehearsed the new dream every evening (the sound group with the chord playing) and slept with a wireless EEG headband that recognised REM sleep and played the same chord quietly during it.

The design is neat. Both groups heard the sound in REM sleep, so the only difference between them was whether the sound had been linked to the rewritten dream.^{[1, 2]}

What they found

Both groups improved, which fits what is already known about imagery rehearsal. But the group whose sound was linked to the new ending improved more, and the difference was still there three months later.^[1]

• After two weeks, nightmares in the linked-sound group fell from almost three a week to about one every five weeks on average. In the comparison group they fell from about 2.6 a week to about one a week.^[2]
• After three months, with no sound in between, the averages were about one nightmare every three weeks versus about one and a half a week.^{[1, 2]}
• Dream emotions. The linked-sound group reported more dreams containing joy. Fear in dreams fell similarly in both groups, and the groups did not differ on nightmare distress, sleep quality or mood.^[2]

The researchers suggest the sound may help consolidate the hopeful new version during REM, so that positive dreams gradually take the place of the old script. That is an interpretation of the results, not something the study can prove.

What it does not show yet

• It is small. There were 18 people in each group. Small trials often look more impressive than the larger ones that follow, so this needs repeating by other teams.
• The participants were a narrow group: young adults, 27 of them women, with nightmares not linked to trauma and no other major conditions. We do not know whether it would help trauma-related nightmares, older adults or people taking medication.^[2]
• Some questions stay open. There was no group without any sound, or without imagery rehearsal, so the study cannot separate every possible effect of sound on sleep. In the preprint, 28 of the 36 people returned for the three-month check.^[2]
• The equipment matters. The chord came from a research-adapted EEG headband with sensors on the forehead and the back of the head. It started only after about five minutes of stable REM, and paused whenever it detected movement or signs of waking.^[2]

Why a phone can’t do this the same way

REM sleep is defined by what the brain, eyes and muscles are doing. An EEG headband reads brain activity directly. A phone on the mattress or bedside table can only sense movement or sound. That says something about whether you are asleep, but much less about which stage you are in.

When researchers compared one popular phone sleep app with full laboratory sleep recordings in 20 adults, the app told sleep from wake reasonably well, but its sleep stages agreed poorly with the laboratory, and its stage-by-stage accuracy was under 50 per cent.^[4] A sound played at the wrong moment is unlikely to help, and it could wake you. Apps that simply play sounds through the night are not doing what this study did.

What this means for you

• Imagery rehearsal works without a sound. Both groups in this trial improved, and the method has years of evidence behind it.^{[1, 5]}
• The part you control is the rehearsal. A specific, vivid new version, rehearsed every day, is the core of the method. If you are new to it, start with what imagery rehearsal is.
• Watch for replications. If other teams find the same effect, sound cues during REM could become a useful add-on. Until then, it is a promising single study.
• If your nightmares are linked to trauma, do this kind of work with a professional beside you.

Baku does not play sounds into your sleep, and it will not pretend to know when you are in REM. Its Rewrite studio focuses on the part that is well established: writing the new ending, then rehearsing it daily.