What to think about to fall asleep

The instruction that backfires

The phrase people actually type into a search box is how to turn my brain off, and the honest answer is that there is no switch. There is only what you point it at.

In a 1996 experiment, ordinary sleepers were asked either to fall asleep as quickly as they could or to fall asleep whenever they felt like it, while listening to one of two things: soft new-age music, or John Philip Sousa marches. Under the soft music, trying harder worked a little — the people told to hurry got there sooner. Under the marches, it inverted. The ones trying to hurry took longer than the ones not trying at all.

The authors read this through ironic process theory. To confirm you have stopped thinking, something in you has to keep checking your thoughts, and the checking is itself a thought. In a quiet room that costs little. In a loud one — a hard day, a live worry, a mind already running — the monitor costs more than the effort saves. "Just stop thinking about it" is not merely weak advice at that point. It is the mechanism keeping you up.

The experiment that handed people something instead

Six years later, Allison Harvey and Susanna Payne ran the complementary study. Forty-one people with insomnia were given one of three sets of instructions for a single night. One group was trained to choose an interesting, engaging scene and elaborate it in detail. One was told only to distract themselves, method unspecified. One was told nothing.

The trained-imagery group reported reaching sleep sooner than the no-instruction group, with pre-sleep thinking that was less frequent and less distressing. The researchers had also predicted that vague, unspecified distraction would come out worse than no instruction at all — and it didn't. That second prediction went unsupported, which is worth saying plainly, because the finding is narrower than the coverage it later attracted. Being handed a specific thing to think about did something. Being told to think about something else, unspecified, did not do much either way.

The explanation the authors offer is the part you can use. The imagery task worked, they argued, because it occupied enough "cognitive space" to keep the person from re-engaging with their worries — and because it was a specific alternative, something to search for rather than something to hold off. A mind can be pointed. It cannot be switched off.

What makes a subject work

Four properties, and a subject needs all four:

• True. Invented material has to be authored as it goes, and authoring is work. Facts arrive already finished.
• Concrete. Things you can picture — a lock gate, a rail, a glasshouse in rain — not concepts you have to hold in argument form.
• Emotionally flat. Nothing at stake, nobody you know involved. Your quarterly review is concrete and true and completely disqualified.
• Going nowhere. No arc, no build, no payoff withheld. If there is an ending worth staying for, you will stay for it.

Counting sheep fails on two of the four: it is not true and there is nothing to picture past the second sheep. A good podcast fails on the last one hardest — it is built by professionals whose job is to keep you listening. And this is the same requirement the cognitive shuffle meets from the other direction, by making the material so disconnected that no arc can form.

For 3am specifically

Waking at three and looking for something to think about is a harder case, because the mind is already mid-sentence about something real. Arguing with it loses. The move is the same one, applied faster: hand it a subject with no relation to the one it found, and let the new one be dull enough that losing your place in it costs nothing. The 3am wake-up has its own page here, because the reason it happens is not the reason bedtime thinking happens.

Where we come in, honestly

You can do all of this unaided tonight. Pick something you find mildly interesting and entirely non-urgent, and walk through it slowly: the standardisation of the railway gauge, the weather at the bottom of the sea, a Roman bathhouse at closing time.

The one cost of doing it yourself is that generating the material is mild work, and some people find the effort keeps them present. So we read it out instead — true, unhurried, flat, and with the ending given away in advance so there is nothing to stay awake for. Your attention follows out of politeness and loses its place around the fourth clause. That is the whole of low-arousal learning, and if you would rather be handed the subject than find one, that is what the stories are for.