ASMR · Sheffield · finger clip

The towel fold dropped 3.41 beats.

JM Thomas · 18 August 2026

A 3-minute towel-fold clip dropped a finger sensor 3.41 beats a minute. Quiet scalps watched the same clip.

They folded a towel.

The pasta chef kept more of the beat.

A library whisper splits two rooms

Poerio's PLOS paper opens in a quiet library, two people whispering behind you, a page turning, a pencil scratching a note you did not ask to hear.

That whole room is one sound bed, and the paper's first job is to split the people who want to leave from the people whose crown starts to crawl.

For a lot of us that's irritation. For the other room it's a warm crawl from the crown down the neck.

They named that crawl Autonomous Sensory Meridian Response. Viewers already had the YouTube word. They already had the sleep caption. Sheffield asked whether the clip moved a body they could wire.

Giulia Lara Poerio's line on the live university news page from 21 June 2018 still sits there: the relaxing effect showed up in people who already get the feeling. Then the watch.

Only-then-watch.

That's the door. The reprint printed 3.41, and the news page rounded it to 3.14, which is the kind of compact caption that travels farther than the table that actually held the watch.

Table 3 keeps 3.41.

A thousand screens ran three minutes

Study 1 lived online. Twitter, Facebook, a dedicated site, a University of Sheffield mailing list. Three months. 2073 people started. 898 bounced before the end. 173 sat longer than an hour, so they got cut as a possible two-session crawl. 1002 remained. 48 percent female. Mean age 29.40, spread 10.79, range 18 to 77.

813, 81 percent, classified themselves as people who experience ASMR after the clips or from everyday life. 189 said no.

Each person watched three minutes of a spoken ASMR clip, three minutes of a sound-only ASMR clip, and three minutes of a control clip that copied the camera and the talking without the slow hands or the soft voice. Order shuffled. After each clip they marked tingle frequency from 1, none of the time, to 7, all of the time, and they marked how they felt now against before on a 1-to-7 change sheet.

Spoken ASMR made the tingle group tingle more than the quiet group, mean difference 1.63, d = 1.04. Sound-only 1.38, d = 0.89. Control still split a little, 0.28, d = 0.31, so even a pasta-style clip can leak a bit of tingle in people who already have the switch.

Calmness on the spoken clip: tingle group minus quiet group 0.91, d = 0.90. Stress dropped 0.69, d = 0.73. Sadness dropped 0.38. Spoken clips also raised connectedness 0.21, d = 0.31. Sound-only connectedness stayed flat, p = .115.

Sexual arousal did not split. F(1.90, 1903.66) = 0.48, p = .611.

Control clips printed no affective split. Excitement, calmness, stress, sadness, connectedness. All of those sat on the same side of zero. The tingle answer needed the ASMR clip and the tingle scalp. One without the other kept the sheet quiet.

Inside the 813, 74 percent named soft speaking. 70 percent named whispering. They reported 6.76 of 13 listed triggers, spread 3.30. Mean years of knowing they had it: 12.19. Calculated onset 15.37. 83 percent already watched clips to fire it. 51 percent watched about once a day or several times a week.

Sheffield wired the middle finger

Study 2 sat in the psychology department with 56 tingle people and 56 quiet people matched on age and gender until two records failed the box and left 110, 58 percent female, mean age 26.14, spread 8.63, range 18 to 59.

They wrapped a photoplethysmograph around the middle finger and two electrodes around the index and ring of the same hand on a ProComp5 Infiniti box sampling 256 times a second, heart rate in beats, skin conductance in microsiemens, the whole chain running while a clip played.

Three minutes of sitting. Then three clips, order balanced: a standard ASMR clip, a control clip, and a self-picked ASMR clip. Tingle people had to stay off YouTube for three days first. Quiet matches watched the same self-picked clip their pair brought in.

The standard clip was a woman folding a towel, speaking softly in Russian, hands slow, camera on the fold. That clip had pulled the strongest tingle and pleasant-affect mark in Study 1.

The control clip was a man showing pasta on the same teaching camera with hands in frame, ordinary voice, ordinary speed, the kind of kitchen demo that should have stolen the pulse if "any slow video calms you" were the mechanism.

It didn't.

They rated how the lab tingle compared with daily life, 1 much less intense to 5 much more. Mean 2.72, spread 1.05. The room under-fired the bedroom.

The towel fold beat the pasta chef

Table 3 is the watch.

Tingle heart rate sat at 76.30 at baseline, spread 9.40. Control pasta 74.10, spread 8.84. Towel fold 70.43, spread 8.32. Self-picked 70.95, spread 8.58.

Quiet-arm hearts sat at 75.91 at baseline, 73.32 on pasta, 71.74 on the towel, and 71.04 on the self-picked clip, so a still chair already stole a couple of beats before the tingle split even showed up.

They scored each ASMR clip as a change from the pasta clip so the number would be the whisper, not "watching a screen." Tingle people dropped more than quiet people, F(1, 106) = 4.95, p = .028, partial eta squared .05, mean difference 1.48 beats, d = 0.45.

Average tingle drop against pasta: 3.41 beats. Towel 3.67. Self-picked 3.15. Heart-rate n after outliers: 53 tingle, 55 quiet.

Quiet scalps dropped some too. Pasta to towel is 73.32 to 71.74. That's a clip doing a little work on anybody who sits still. The tingle group still lost more of the beat.

I've sat through whisper clips that left my scalp blank. The file already named that body. It's the quiet arm, and the quiet arm is a result, not a broken pair of headphones.

The skin still woke 0.30

Skin conductance on the tingle arm sat at 3.04 microsiemens at baseline, 3.49 on pasta, 3.76 on the towel, 3.82 on the self-picked clip.

Quiet-arm skin sat at 2.96 at baseline, 3.60 on pasta, 3.64 on the towel, and 3.63 on the self-picked clip, almost a flat line once the screen was already on.

Tingle people rose more from pasta to the ASMR clips, F(1, 105) = 5.92, p = .017, d = 0.46. Mean rise 0.30 microsiemens. Towel 0.27. Self-picked 0.33. Conductance n: 52 tingle, 55 quiet.

A nap story wants both numbers to fall. They didn't. Heart slowed. Skin lit. The authors called that a mixed state, calm plus a bit of charge, and they pointed at work that already split cardiac and electrodermal channels instead of treating them as one arousal knob.

Excitement rose with calmness in the tingle arm, d = 0.81 and 0.55. Sadness, stress, connectedness, and sexual arousal did not split from the quiet arm in the lab. The big online calm-and-stress split shrank once they put a clip on a finger in a room that already under-fired daily life.

A song chill raises that same watch

Aesthetic chills from music, the frisson file on this site, sit with a faster pulse in the papers Poerio cites. Grewe, Benedek, Sumpf. Piloerection with a climb.

Sheffield's tingle clip went the other way on the heart, a drop where the music papers print a climb, which is why a crown-crawl caption is a terrible reason to mash this URL into the frisson file.

Barratt and Davis 2015 already had the survey: people use the clips for relaxation, sleep, and stress, first tingle often in childhood. Smith 2016 had a tiny resting-state set, 11 tingle brains, reduced default-mode coupling. This page does not rerun those files. It keeps the finger clip.

The other sensory maps sit in the Library.

A quiet crown is the result

People hunt a better whisper when the first one does nothing. The 1002-person sheet already sorted that. Quiet scalps watched the same spoken clip and stayed near 1.42 on the tingle mark. Tingle scalps sat at 3.05.

Buying new headphones to "get" ASMR is a different errand than sitting on the side of the split the file actually printed.

Self-ID is a hole. Nobody ran a week-apart consistency test. Demand can push a tingle rating. Physiology is harder to fake, and even there the lab felt weaker than the bedroom, mean 2.72 on that 1-to-5 intensity mark.

Expectancy is named in the discussion, familiarity is named, and they did not run a blind they could hide, because a tingle person already knows the genre and a quiet person already knows they came in as the match.

Pulse file, two rooms