What the cognitive shuffling research actually shows

Does cognitive shuffling work?

Probably, for some people, in a modest way — and less certainly than most coverage suggests. The evidence points in a consistent direction: occupying a busy mind with simple, concrete imagery at bedtime seems to reduce the arousal and effort that keep people awake. What it does not yet show is that cognitive shuffling itself makes people fall asleep a particular number of minutes faster, because the one trial of the technique did not report that.

Two studies carry nearly every claim you will read. Here is each one, as it was published.

The cognitive shuffling trial (SLEEP 2016)

Luc Beaudoin, who developed the technique, tested its structured form — serial diverse imagining — with three colleagues from MacEwan University in Canada. The results were presented as a poster at SLEEP 2016, the joint meeting of the American Academy of Sleep Medicine and the Sleep Research Society.

• Who: 154 university students — 137 of them women — who complained of excessive mental activity before sleep.
• What: each was randomly assigned to serial diverse imagining (an app plays a concrete word every eight seconds and the listener pictures it), to structured problem-solving (a standard technique done earlier in the day, taking about fifteen minutes), or to both.
• Measured: questionnaires on pre-sleep arousal (mental and physical), sleep effort, sleep quality and sleep hygiene, at the start and again after one week and/or one month.
• Found: mental and physical pre-sleep arousal, sleep effort and sleep quality all improved compared with where people started, with large effects (partial η² from .43 to .71). Serial diverse imagining did as well as problem-solving, with the advantage that it is done at bedtime rather than before it — though the full poster adds that the trial had too little statistical power for a comparison between the methods to mean much either way.
• Rated: students who tried both methods were more likely to rate serial diverse imagining as the more helpful and the easier to use — a finding in the poster’s conclusion, not in its abstract.

What this trial cannot tell you

• Whether people fell asleep faster. Participants kept sleep diaries, which would record that, but the abstract and the poster both left the diary results out for space. Neither reports a figure in minutes.
• Whether the technique beat doing nothing. Every group received a treatment. Improvement from a starting point can come from time, expectation or simply paying attention to sleep, and without an untreated group those cannot be separated out.
• How it works for people with diagnosed insomnia, or for men, or outside a university. The sample was students, mostly women, selected for a racing mind rather than a diagnosis.
• How it holds up in full peer review. A conference abstract is a short summary, not a full paper with its methods open to scrutiny.
• How it holds up independently. The poster discloses that Beaudoin is a director and shareholder of CogSci Apps, the company behind the shuffling app mySleepButton. A replication by researchers with no stake in the technique would carry more weight.

One detail cuts the other way. Sleep hygiene — habits like regular hours and caffeine timing — got worse over the same period, which the authors put down to the academic term starting. Improvements in arousal and sleep quality against that background are, as they wrote, notable.

Where does “fall asleep 20 minutes sooner” come from?

From a different study, of a related technique. In 2002 Harvey and Payne, publishing in Behaviour Research and Therapy, gave 41 people with insomnia one of three sets of instructions to follow on one night:

1. Imagery distraction — after brief training, choose an interesting, engaging scene and imagine it in detail while falling asleep.
2. General distraction — distract yourself however you like.
3. No instructions.

The imagery group fell asleep sooner, and had fewer and less distressing pre-sleep thoughts, than the group given no instructions. The 2002 result is usually expressed as about 20 minutes sooner than those participants’ own baseline nights. The authors attributed it to the imagery occupying enough “cognitive space” to keep people from returning to their worries.

That is a real finding and a useful one, and it supports the same mechanism cognitive shuffling relies on. But it is a study of imagery distraction — elaborating an engaging scene rather than shuffling unrelated images — over a single night in 41 people. The sentence “cognitive shuffling makes you fall asleep 20 minutes faster” takes the minutes from this study and the technique from the other, and neither study supports it on its own.

Did the study show counting sheep makes it worse?

Not as published. A popular retelling says Oxford researchers found that insomniacs who counted sheep took longer to fall asleep. Harvey and Payne did predict that their general-distraction group would take longer to fall asleep than the group given no instructions — and the abstract says that prediction was not supported. The abstract does not mention sheep at all.

Whether the full paper’s general-distraction instructions named counting sheep as an example, we have not been able to check: the full text is behind a paywall. Until someone does, the defensible version is that counting sheep was not shown to help, and was not shown to hurt. The case against it rests on how it works rather than on this study — see does counting sheep work?.

Where does the figure in Sheeps’ own listing come from?

It is Harvey and Payne’s result expressed as a share of a typical time to fall asleep, and we would rather explain it here than have it found out.

Sheeps’ App Store listing and onboarding say people fall asleep about 42% faster, and that figure is the 20 minutes from Harvey and Payne’s 2002 study of 41 people with insomnia divided by 48 minutes — the average time to fall asleep at the start of the trials in which a 2016 meta-analysis compared online CBT-I with a waiting list (Seyffert and colleagues, 2016) — so it is arithmetic across two studies and not a test of Sheeps.

20 ÷ 48 is 41.7%, which rounds to 42; across that 2016 baseline’s confidence interval of 41 to 56 minutes, it runs from about 36% to 49%, which is why the app’s text says “around”. The weak joint is that the two numbers come from two different groups of people with insomnia. Below the figure and its explanation, the app’s evidence screen says that the study tested imagery distraction and not Sheeps.

The claims side by side

What would settle it?

A trial of cognitive shuffling with an untreated or placebo group, time to fall asleep measured objectively or by diary and reported, in people with insomnia rather than only students. None of that is exotic; it has simply not been published for this technique as far as we can find. Until it is, the honest description is a low-cost, low-risk technique with a sensible mechanism and encouraging early results.

For insomnia that has lasted months, the treatment with the strongest evidence is cognitive behavioural therapy for insomnia, which the American College of Physicians recommends as the first thing to try (Qaseem and colleagues, 2016).