The Mind Science of Practice

Five months of running a hypnotic ritual practice, written down honestly. The part that worked was ordinary and measurable. The part that did not work was the part everyone writes about, and the two were never entangled in the first place.

Five months, one claim, and a sensor on the bedside table

A fixed sequence, a fixed time window, a word repeated with a count, and a closing action. Done daily for five months, tracked with a sleep sensor and a notebook, aiming at one thing: nights where the practice shortened the time it took to get back to sleep after waking.

This is a deliberately small claim, and the smallness is the design. The aim names an observable event with a start and an end, which means it can be counted, and a countable aim is the only kind that can tell you anything when the month is bad.

The part that worked, and it was not the hypnosis

A fixed sequence reliably changes breathing and heart rhythm within minutes, and this was the most reliably measurable effect in the whole notebook. Heart rate down, variability of one kind up, breathing rate down. It appeared within ten minutes and it appeared on the boring days too.

The word hypnotic describes the framing. The measurable thing is attention plus repetition plus a fixed duration, which is a physiological event rather than a psychological one. Nobody in the tradition that gave me the word would have been happy about that sentence, and it is the reason the practice is worth keeping.

Expectancy, and where the ceiling is

Trials in which participants were told directly that what they were given was inert still show reported changes. That puts a ceiling on the simple story that the whole effect is belief, and it means the placebo component here is real and modest rather than either absent or total.

Modest is the operative word and it took me most of the second month to accept. The plausible size of an expectancy effect is small enough that a single bad night is indistinguishable from noise, which is exactly why the aim needed to be countable and why the notebook needed to have a column for what I would have done in any case.

A hundred calls, and the number they gave back

Five symbols, one in five, follows from the design of the deck regardless of who is guessing. I ran a hundred calls as a calibration exercise and got a number that sat exactly where chance would put it, which is the least satisfying and most useful result in the notebook.

Its purpose was not to settle anything about the mind. It was to find out how I counted. Almost everyone who runs a hundred calls discovers that they had been counting selectively, and I was no exception. That discovery transferred directly to the sleep record, where the same kind of selective attention was doing a great deal of work.

Where the weaving went wrong

Predictive processing treats perception as hypothesis management, and it explains how expectation can change experience without anybody consciously lying to themselves. It is a good account and it is a model. I had been treating it as a finding for about three weeks before I noticed what I was doing.

An explanation that accounts for a phenomenon is not evidence for the phenomenon. Once I sorted those two apart, the notebook got more honest, because I had been writing entries in the voice of someone reporting a mechanism rather than someone reporting a week.

The three months where nothing happened

Month three, a full month with no change in the sleep figures beyond what the previous baseline predicted. No effect, no signal, nothing. The honest response was to keep the record and change nothing, because changing three things at once after a flat month is how people end up with a practice that works and no idea which part.

Month four was the same. Month five produced two short nights out of thirty-one, which is well inside what the baseline accounted for, and I want to be precise about that: two out of thirty-one is not a result, it is a coincidence I nearly wrote up as one.

The costs, which nobody in this material mentions

A daily practice with a fixed word and a count is monotonous, and monotony is a cost. I was also more inclined to treat a bad night as evidence about the practice rather than about the night, which is a subtle drift and it took two months to notice.

There is a real one in the anxiety direction. Self-monitoring reliably increases what you notice, and for some people that means a practice aimed at sleep becomes a practice about noticing sleep, which is the exact opposite of the aim. The version that helped was the one where I stopped checking the sensor on the nights I was anxious.

Four claims ranked by how well the notes support them

A fixed sequence changes the autonomic picture within ten minutes. That is the finding, and it is a modest one, and it is a real one. Expectancy has a small and documented effect on reported outcomes. The card deck is arithmetic. The model is a model.

What the record does not establish is that the practice shortened sleep onset latency, because over five months the sleep figures are indistinguishable from baseline. The most defensible sentence I can write is that the practice reliably changed my state and I cannot show that it changed the thing I aimed at.

Why I kept it

Because the ten-minute autonomic change is worth having on its own, and because a practice that produces a measurable change in one thing is a practice that has run, which is a separate property from whether it produced the outcome you wanted.

That distinction is the useful legacy of five months. Many practices are abandoned because they are measured against an aim they were never going to move. This one survived because the aim was countable, the count came back flat, and the thing that did change was worth keeping anyway.

The versions I would run instead

Fewer components. The fixed word and the count did the work and the imagery did not, and I spent four months on imagery. One aim per cycle, because two aims in a month and no result in either is the same as one aim and no result, except you have learned less.

And a shorter commitment, three months rather than five, with the option to stop written down at the start. A practice you are locked into for half a year is a practice you will defend, and defending a practice is a form of self-monitoring that will make your record worse.

The question I would ask a beginner

What would you count, and how many would it take to convince you that nothing is happening? Ask it before you start. If the answer is a feeling, the record will support you indefinitely and you will never learn anything.

The second question is what you would keep if the count came back flat. If the answer is nothing, the practice has a single point of failure and will collapse the first bad month. If the answer is a real behaviour, you have a practice that can survive the data.

What five months of notes actually returned

The sequence changed my body inside ten minutes, the symbol did not do anything measurable, the sleep did not move, and the practice is still on the schedule because the first part is worth having.

That is the whole finding and it took five months. It is smaller than the popular version by a long way, and it is the version you can build a practice on, because every clause in it is something a stranger could check.

Frequently Asked Questions

Did the hypnotic ritual actually work?

The measurable part worked and the aimed-at part did not. A fixed sequence reliably changed breathing and heart rate within ten minutes, across five months and on the unremarkable days. The sleep figures I was aiming at stayed indistinguishable from my own baseline. Those are two separate results and reporting only the first would be dishonest.

Is the placebo component of ritual practice real?

Real and modest. Trials where participants are told directly that what they were given is inert still show reported changes, which limits the simple belief story. The size is small enough that one bad night is indistinguishable from noise, so any practice in this area needs a countable aim or the effect size cannot be distinguished from a good week.

Why run a hundred card calls before anything else?

For calibration. Five symbols means one chance in five whatever the guessing, so the number is a fixed point rather than a finding. What it actually teaches is how you count. I found I had been counting selectively, and the same selective attention was distorting the sleep record, which made the exercise worth more than the experiment.

Can self-monitoring make a practice worse?

Yes, and it is the failure nobody warns you about. Noticing increases what you notice, so a practice aimed at sleep can turn into a practice about inspecting sleep, particularly on the anxious nights. I got more out of the version where I stopped consulting the sensor when I was worried than from the version where I checked every morning.

How do I know whether a practice is running at all?

Separate the two questions. Whether the practice changed anything you can measure is one question, and whether it produced the outcome you aimed at is a different one. A practice can do the first and fail the second, and it is still a practice that runs, which is a reason to keep it that the popular framing does not allow.