Milton Erickson had a problem that every therapist has and most never solve.
His patients might seem to be recovering rapidly, but their apparent susceptibility to the therapy masked a deep resistance: They would soon relapse into old habits, blame the doctor, and stop coming to see him.1
The rapid recovery was the symptom. A patient improving too easily is a patient who has not engaged, and the crash is coming.
His response was not to work harder on the resistance.
Erickson began ordering some patients to have a relapse, to make themselves feel as bad as when they first came in—to go back to square one. Faced with this option, the patients would usually "choose" to avoid the relapse—which, of course, was what Erickson really wanted.1
Greene's generalisation: *"This is a good technique to use on children and other willful people who enjoy doing the opposite of what you ask them to: Push them to 'choose' what you want them to do by appearing to advocate the opposite."*1
The instruction changes something structural, and the change is easy to miss because the technique gets summarised as reverse psychology.
Before the instruction, relapsing is defiance. It is the patient's own move, made against the therapy, and it carries a payoff: it proves the doctor wrong, restores control, and confirms that nobody can tell them what to do.
After the instruction, relapsing is compliance. It is the doctor's homework. Doing it means obeying.
Nothing about the relapse itself has changed. What has changed is what it would mean, and for a patient whose real project is not being managed, the meaning was the entire appeal.
So the technique does not persuade anyone of anything. It removes the payoff from the option you do not want by reassigning its authorship — and it does that without argument, without pressure, and without the patient having to concede a point.
This is one of the three genuinely risky methods in Greene's list, and the chapter does not mark it.
The patient can take the relapse.
Greene's own hedge is doing quiet work: "the patients would usually 'choose' to avoid the relapse." Usually. Which means sometimes they went back to square one, on their doctor's written instruction, and Erickson owned that outcome completely.
Compare Colour the Choices, where every branch is one Kissinger proposed and no outcome is a loss. Same list, same page, and one of them can produce a patient who has deteriorated on your advice while the other cannot produce anything worse than a policy you also drafted.
The three risky entries — this, the Weak Man on the Precipice, and Brothers in Crime — are printed flat alongside four riskless ones, and the criterion that sorts them is stated two pages earlier. See The Six Methods of Controlling Options.
Erickson could run this because of three things Greene does not mention, and removing them changes the technique entirely.
He could observe the outcome. The patient comes back next week. If the relapse was taken, he sees it and can respond. A technique with a live losing branch is only safe where the loss is visible and recoverable.
Both parties wanted the same thing. The patient came to get better. The resistance was not opposition to the goal but to being handled, which is why reassigning authorship resolves it — there is no genuine conflict of interest to survive.
He carried the liability. A doctor who instructs a patient to relapse has taken on professional responsibility for the consequence, and that responsibility is what disciplines the use.
Strip all three out — which is what happens when the manoeuvre is extracted into a general power technique — and you have: tell someone to do the harmful thing, so they will do the other thing, in a situation where you cannot see the outcome, you want something different from what they want, and you bear no cost if it goes wrong.
That is not the same instrument. Five of the seven methods on this list were performed inside a professional role with its own constraints, and the extraction is the book's central move.
Greene names the target type — "children and other willful people who enjoy doing the opposite" — and the scope condition is tighter than it looks.
The technique requires a person whose opposition is reliable and general, not selective. If someone resists you only on things they actually disagree with, advocating the opposite does nothing: they evaluate the proposal and take it or leave it on the merits.
It works on people whose contrarianism is a stance rather than a judgement — where the reflex fires before the content is assessed. And that is a real and recognisable type.
But it means the technique is self-limiting in a way the chapter does not say. A person operating on reflex will eventually notice a pattern of getting what they wanted by opposing you, and the reflex is not stupid, it is fast. Erickson could run it once per patient on one question. Run it three times on the same person and you have taught them to check.
Someone is doing the thing you keep asking them not to do, and the asking is what is keeping it alive.
First, be honest about the diagnosis. Does this person oppose you, reliably, across topics? Or do they simply disagree with you about this? If it is disagreement, none of this applies and you are about to be manipulative and ineffective at the same time.
If it is genuine reflexive opposition, the move is not to advocate the opposite. It is to remove the defiance payoff, and there is a low-risk version and a high-risk version.
The low-risk version: stop supplying the resistance. Say nothing about it. Take your position out of the room entirely, so that continuing costs them the pleasure of continuing against something. This is most of the value with none of the exposure.
The high-risk version is Erickson's — actually instruct the thing you do not want. Only do this when you can see the outcome, when you can respond if they take it, and when the harm of them taking it is one you are willing to own. Erickson had all three. Most people reaching for this have none.
And once. You get to do this once with a given person on a given question. It is not a practice, it is a card.
Strongest evidence. A named clinician, a specific and unusual instruction, and a stated mechanism — plus Greene's own "usually," which is more honest than the surrounding entries and is the only acknowledgement in the whole list that a method can fail.
Tension — a live losing branch, unmarked. Documented above.
Tension — the clinical constraints are removed in the generalisation. Observability, aligned goals, and owned liability all disappear when the move is offered to a general reader.
Tension — self-limiting, and presented as a technique. A reflex that gets rewarded gets examined.
🚩 [POPULAR SOURCE] · 🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no case study, no paper, no date beyond "the 1950s". 🚩 REPLICATION RISK — a pre-2015 clinical claim about a therapeutic technique's effectiveness, given with no outcome data, no controls, and no indication of how often it was used or how often it failed. Erickson's clinical corpus is substantially anecdotal and self-reported; treat "the patients would usually choose to avoid the relapse" as a practitioner's account of his own practice, not as evidence about efficacy.
Open questions. What happened to the patients who took the relapse — the hedge implies they exist and nothing in the account follows them. And is the paradoxical instruction doing the work, or is it the attention — a doctor who cares enough to give an unusual instruction may be producing engagement by a much more ordinary route.
Erickson is on his fourth page in this vault, and this one is a fragment of a doctrine the others carry properly.
Erickson and the Utilization Technique is the parent: patients arrive with "countless ways of controlling the relationship" — withholding information, resisting trance, questioning his competence, insisting on the hopelessness of their problems — and the method is to use the resistance as the vehicle rather than fight it. Ordering a relapse is one application of that principle.
The difference between parent and fragment is the whole ethical content. Utilization is a therapeutic stance: it takes the patient's actual behaviour as material and works with it toward what the patient came for. Force the Resister is the manoeuvre with the stance removed and a different beneficiary installed.
Greene does this consistently and it is worth naming as a corpus-level pattern rather than a one-off complaint: the technique survives extraction and the constraint does not. The same happens to Rikyū's tea ceremony at Law 30 and to Ninon's courtship advice at Law 28.
Against Colour the Choices in the same list: both aim to make the target arrive at your conclusion unassisted. Kissinger achieves it by advocating nothing; Erickson by advocating the opposite. The one with no advocacy in it is the one with no losing branch, which is a useful general observation about this whole family.
Clinical psychology — the doctrine this is a fragment of. Erickson and the Utilization Technique holds that resistance is not an obstacle to the work but the material of it, and that the therapist's job is to find the version of the goal that the patient's actual behaviour is already moving toward.
The insight the pairing produces: utilization requires that a version of the goal exists which the resistance can serve. That is a precondition, and it is the thing that makes the technique honest — you are not redirecting someone, you are locating the overlap between what they are doing and what they want. Greene's extraction assumes the overlap always exists, or does not care whether it does. So the diagnostic that makes this safe is not about the target at all: it is whether you can name the goal that their resistance is already serving. If you can, you are doing utilization and it will hold. If you cannot, you are running a trick on a reflex, and it works once.
Behavioural economics — the option whose value is its meaning. Costly Signaling holds that an action's value to the actor often lies in what it signals rather than in what it materially delivers.
The relapse is exactly that. Its material content — feeling as bad as before — is purely negative; nobody wants it for itself. Its entire value to the patient is as a signal of autonomy, and signals are cheap to destroy because their meaning is assigned by context rather than carried by the act.
What that produces is the general form of the technique, wider than reverse psychology: when someone is doing something whose payoff is symbolic, you do not have to change their incentives or their mind. You change what the act would mean, and the payoff evaporates while the act stays available. Erickson did it by prescribing. You can also do it by making the behaviour ordinary, or expected, or somebody else's idea. What you cannot do is argue, because argument confirms the meaning — it establishes that there is something to defy, which is the thing being purchased.
Sharpest implication. The relapse's whole value to the patient was as defiance, and prescribing it converted the identical act into compliance — nothing material changed, only the authorship. That generalises well past reverse psychology: any behaviour whose payoff is symbolic can be defused by reassigning what it would mean, and arguing is the one move that reliably fails, because argument confirms there is something worth defying. But it is one of three genuinely risky entries in Greene's list, Erickson's own "usually" is the only failure admission in the chapter, and every constraint that made it safe — observability, aligned goals, owned liability — is stripped out in the generalisation.
Generative questions.