Blog

In 1984, a research team did something to forty-three headache sufferers that would be hard to get approved today.

Everyone in the study had recurrent tension headaches. They were hooked up to a biofeedback machine that displayed muscle activity — and told they were being trained to relax those muscles. If tension headaches come from tense muscles, teach people to untense the muscles. 

Except the researchers rigged it. For half the participants, the machine was secretly set up backwards. Their feedback rewarded increased muscle tension, not decreased. The subjects were being trained, without knowing it, to tighten up.

Then there was a second layer. Independently of what the machine was actually doing, participants were shown fake progress displays. Some were convinced they were succeeding dramatically. Others were convinced they were making only modest progress.  So you end up with people who genuinely relaxed but thought they were mediocre at it, and people who genuinely tensed up but thought they were crushing it.

The people who believed they were succeeding, improved by 53 percent.  The people who believed they were doing just okay improved by 26 percent.  Which direction their muscles actually moved made no meaningful difference.  That’s the opposite of what the whole treatment rationale predicts. The group whose forehead muscles got tighter — but who thought they were winning — did better than the group whose muscles genuinely relaxed.

Why this isn’t a fluke.  The obvious response is that something went wrong in one odd study. It didn’t.  The muscle explanation for tension headache had been shaky for years before this. When researchers measured muscle activity and headache severity in the same people, the relationship was weak. When they tracked how much muscle activity dropped over a course of biofeedback, that drop didn’t predict who got better. The treatments worked. The stated reason they worked kept failing to show up in the data.  This is worth sitting with, because “tension headache means tense muscles, so relax the muscles” is still the standard advice. It’s intuitive. It’s anatomically sensible. And when it’s been tested directly, it hasn’t held up as the explanation.

What tracked with improvement was something psychologists call self-efficacy — not optimism, not positive thinking, but a person’s specific belief that they have effective actions available to them in this particular situation.  That’s what the fake success displays were manipulating. Not muscle tone. Confidence in one’s own capability.

And it isn’t a small or short-lived effect. In a later trial of 176 people with frequent migraines — people who had already failed to respond to optimized drug therapy — behavioral treatment produced large increases in headache self-efficacy and in the sense that headaches were something they could influence rather than something that simply happened to them. Those changes were still there twelve months later.

To be clear about what this does and doesn’t mean: it does not mean the pain is imaginary. It means the nervous system is constantly making judgment calls about ambiguous signals, and what is believed about the capacity to handle those signals feeds into those calls. The belief isn’t sitting on top of the physiology. It’s participating in it.

Where hypnosis comes in

Look again at what the researchers were actually manipulating: a vivid, believable experience of succeeding. They had to fake it, because in 1984 the only way to hand someone that experience in a lab was a rigged display. But deception isn’t the only route to it, and it was never the interesting part. Hypnosis produces the same thing directly and honestly: a focused state in which you rehearse handling the situation and register the result as something you did. That’s the mechanism the study accidentally isolated. It’s also why I’m skeptical of the framing that hypnosis works by relaxing you. Relaxation is pleasant and often useful, but the headache literature already tried explaining these results with relaxation and it didn’t hold. What changes is the expectation of your own capability — and that turns out to be the thing worth changing.

Here’s the practical part, and it follows directly from the research.  Self-efficacy is built primarily from evidence of our own success. Not from encouragement, and not from being told you’re capable but from watching yourself handle something.  Most people who track their headaches track the wrong variable. They log intensity: a 6 today, an 8 yesterday. That record is a catalog of things happening to you, and reading it back builds the opposite of what you want.

Try logging actions and outcomes instead. Not “7 out of 10 at 3pm,” but: what did I do, and what happened after?  Stepped outside for four minutes — dropped a little. Unclenched my jaw twice during the meeting — didn’t build the way it usually does. Went to bed on time — woke up clear.

You are collecting evidence that your actions have effects. Over a few weeks, that record becomes something you can actually consult on a bad afternoon, and it’s built from your own experience rather than someone’s reassurance.  That’s not a trick to feel better about the pain. Based on the evidence, it’s aimed at the part of the system that turns out to be doing the work.

A note on the obvious exception: this is about ordinary recurring stress headaches. A headache that arrives suddenly and severely, follows a head injury, comes with fever, vision changes, weakness, or confusion, or breaks sharply from your usual pattern is a reason to call a doctor, not to start a log.