---
title: "Self-extinguishing maintenance: treatments that stop because they are working, and the relapse that follows"
kind: "Article"
topic: "Habits & maintenance"
date: "2026-10-04"
url: "https://canopticon.com/articles/self-extinguishing-maintenance"
author: "Chris Telford"
license: "CC BY 4.0"
---
Two teeth that do not quite touch trap food. Cleaning the gap after meals keeps the gum healthy, and within a week it looks fine — so the cleaning stops. A few weeks later the gum is sore again. Nothing about the teeth changed; the gap was there the whole time. The same happens with toes pressed too tightly together, with dry skin that only stays soft while it is moisturised, and with many medicines.
The research behind this article looked for a name for that whole cycle — treat until healed, stop, relapse, because the cause never went away — and found only names for pieces of it. Health psychology has one for the belief; adherence research has one for the stopping. What is missing from the literature is a general, cross-domain account of the full loop (symptom → maintenance → remission misread as cure → stopping → relapse caused by an unchanged structural cause). So it proposed two names, one for what happens and one for what you believe.
- The cycle | Self-extinguishing maintenance | a maintenance behavior that, by succeeding, removes the symptom that cues and reinforces it, so that the successful state is misread as cure and the behavior stops, while an unchanged underlying cause guarantees relapse. | In plain terms: the treatment works so well that you stop needing to think about it — and stop.
- The belief | Remission-as-cure error | "remission-as-cure error" (reading a maintained state as a resolved one) | In plain terms: taking "it is fine while I keep doing this" to mean "it is fine".
## Feeling better is the most common reason people stop treatment
The pattern is measured, across very different conditions. In a study of 198 adults hospitalised for asthma, 53% believed they only had asthma when they were having symptoms. Among 230 people treated for high blood pressure, newcomers were likely to drop out of treatment if: they had reported symptoms to the practitioner at the first treatment session, or they construed the disease and treatment to be acute. Across surveys in 13 countries of people who stopped antidepressants, the main patient-reported reason for discontinuation was feeling better (46.6%). And for blood-pressure drugs, about half of the patients who were prescribed an antihypertensive drug had stopped taking it within one year.
caption: Adults hospitalised for asthma who thought they only had it when they had symptoms
- Believed it | 53% | 53 | #2f7d52
- Did not | 47% | 47 | #cfc9bb
note: 198 adults
> **Objection: This is just non-adherence, which is already well studied**
>
> Each piece is studied; the loop is not. The research found that No source names the whole cycle across conditions, and none treats structural anatomy as the cause that never goes away. Adherence research records that people stop, and asks them why. It does not explain why the reason they give — feeling better — is so reliable.
“higher adherence went with stronger beliefs that the treatment is personally necessary (OR 1.742)”
## The pieces that already have names
Each piece of the loop is studied somewhere. The research's survey of what exists, and what each misses:
| Concept | Source / field | What it captures | What it misses |
|---|---|---|---|
| Common-Sense Model; "timeline" and "cure/control" beliefs | Leventhal and colleagues; Meyer et al. 1985 | People treat chronic conditions as acute and stop when symptoms go | Mostly medication; not structural causes |
| "No symptoms, no asthma" / acute episodic disease belief | Halm, Mora & Leventhal 2006 | Belief that the condition exists only when symptomatic; expecting a cure | Asthma-specific |
| Intentional non-adherence | Adherence literature; Morisky scale; NICE framing | Deliberate stopping "when I feel better" | Treats stopping as a decision, not a reinforcement loop |
| Necessity-Concerns Framework | Horne et al. 2013 | Adherence tracks perceived personal need | Doesn't model how need perception decays with success |
| Initiation vs maintenance | Rothman 2000, Health Psychology | Starting depends on expected outcomes; continuing depends on "perceived satisfaction with received outcomes" | Assumes satisfaction sustains behavior. In your case, satisfaction (healing) ends it |
| Proactive vs reactive therapy | Atopic dermatitis literature | Formal "get control, keep control" model | Framed clinically, not psychologically |
| Supportive periodontal therapy | Periodontology | "Life-long supportive maintenance phase" | Framed as attendance compliance |
| Prevention/preparedness paradox | Public health, risk management | Successful prevention makes itself look unnecessary | Applied to organizations and populations, not personal self-care |
*Copied from the research’s survey of existing concepts.[Details](doc:when-healed-is-the-treatment-working-the-stopping-mechanism-in-mainten)*
One commentary title captures the gap between the two expectations people carry: "Inside every chronic patient is an acute patient wondering what happened".
## Self-extinguishing maintenance: the problem was what kept you treating it
Why would anyone stop something that works? Because the problem itself was what kept them going. A sore gum is a reminder to clean the gap, and the relief when it stops hurting is the reward. Once the gum is healthy, there is no soreness to remind you and no relief to feel — so the cleaning quietly drops away. Behaviour analysts describe the same thing: once the discomfort is gone, the person would no longer be motivated to engage in the behavior. The treatment, by working, has removed both the reminder and the reward.
People also judge whether to keep going by how things look now. The leading account of why people maintain a behaviour holds that continuing depends on perceived satisfaction with received outcomes. That works for exercise, where feeling fitter keeps you going. For a maintained condition it does the opposite: the outcome looks finished, so the effort feels finished too.
The skin and the gums show how fast the reversal is. When long-term moisturiser use stops, dryness rises within 1 day after discontinuation. Gums inflamed by a few weeks without cleaning recover quickly once cleaning resumes, and That quick reversibility is exactly what teaches the wrong lesson: the tissue *looks* cured within days, but only while cleaning continues.
> **Objection: The reward-and-cue account is a theory, not a finding**
>
> Correct, and the research says so: The reinforcement framing, the cue-loss framing, the proposed name, and the McLuhan reading are this report's synthesis. They are consistent with established principles but not tested as such. The stopping is measured; the explanation for it is built from well-established principles and has not been tested as a whole.
The symptom was also the reminder. a routine cued by a symptom loses its trigger when the symptom goes. And while the routine works, the evidence that it matters never appears: While the routine runs, you never see what would happen without it, so the healthy state gets credited to "my body healed" rather than "my routine is holding it." The research flags these as inferences from established principles, not tested findings.
The skin case is precise. After long-term moisturiser use stops, dryness "reverted to baseline levels after 6 days" by objective assessment. Dermatology already runs a maintenance-forever model for eczema, called proactive therapy, and it works: a meta-analysis found proactive fluticasone (RR 0.46) and tacrolimus (RR 0.78) reduced flare risk compared with vehicle. Adherence is still poor: in one survey at least 56.6% of AD patients did not use emollients as often as recommended.
The teeth case has its own numbers. In 104 adults, gaps where teeth did not touch showed more probing depth (+0.27 mm) and attachment loss (+0.48 mm) than the same people's closed contacts. Gums that went uncleaned developed gingivitis within about 9–21 days, and people differ: Later work identified "fast," "slow," and "low" responder phenotypes. Stopping long-term gum care has a measurable cost; one meta-analysis reported a pooled tooth-loss risk ratio of 0.56 for regular versus erratic compliers. The research is candid that the evidence for interdental cleaning itself is low-certainty.
## The remission-as-cure error: reading "under control" as "fixed"
The belief half has a partial name already. The belief error has a name in health psychology: the acute (or "acute episodic") model of a chronic condition. People treat a lasting condition as if it were a passing one. The remission-as-cure error is that model applied to a routine: the condition looks resolved because the routine is holding it, and the routine gets the credit for nothing.
Most of what people experience as illness does heal: cuts, colds, sprains. It is reasonable to expect the same of everything. The research puts it fairly: The misreading is not stupidity. It is a reasonable generalization from most of one's experience, applied to the minority of conditions where it fails. The minority is the conditions whose cause does not go away — chronic physiology, or plain anatomy, like two teeth that do not touch and so trap food. Geometry does not heal.
Being told to keep going rarely survives this. Experiments on behaviour learned from instructions found it relatively insensitive to consequences — and a rule followed because a parent or clinician said so has nothing left to hold it once the person feels fine and no one is checking. The belief that does predict continuing is the sense that a treatment is personally needed: across 94 studies, higher adherence went with stronger beliefs that the treatment is personally necessary (OR 1.742). And that sense of need is exactly what success erodes.
> **Objection: Sometimes stopping when you feel better is right**
>
> It is, when the cause has gone. The research flags its own antibiotic figure for this reason: stopping antibiotics early is not always wrong. An infection clears; the gap between two teeth does not. The distinction the loop hides is not between feeling well and unwell but between a cause that has ended and one that is still there.
The research draws the told-versus-understood distinction more finely. Behaviour analysts separate following a rule for approval from following it because it describes the world: A child told "floss every day" without the mechanism learns pliance. Once parental enforcement is gone, the behavior has neither a social nor an experienced basis, so it falls apart. Motivation research points the same way; a meta-analysis of 184 data sets found internalised motivation related to better health outcomes: "Because I understand why" beats "because I was told."
## Success makes the routine invisible
Public health has a version of this for whole populations. A commentary on why prevention is resisted lists as an obstacle that the success of prevention is invisible, lacks drama, often requires persistent behavior change, and may be long delayed. Self-extinguishing maintenance is the same thing inside one person: while the routine works, there is nothing to see, and the only way to see what it was doing is to stop and watch the problem come back.
## How to tell a routine that ends from one that does not
The research turns the distinction into a few working tests. The first is about the cause: Transient causes (injury, infection, one-off exposure) → heal and done. Persistent causes (anatomy, spacing, genetics, chronic physiology) → maintenance until the structure changes. The second treats repetition as information: Two recurrences at the same site with the same trigger is strong evidence of a standing cause.
The third moves the reminder off the symptom: Tie the routine to a fixed context: after the shower for the toes, after each meal for impaction sites. And the last asks what would end the routine for good: Where the cause can be removed, removal beats lifelong maintenance.
## What the two names change
Naming the belief corrects it: a symptom-free period is not a cure, it is the routine working. Naming the cycle explains why the correction is so hard to keep. It changes what "better" means. A symptom-free period on a routine is not evidence the problem has gone; it is evidence the routine is doing its job. "Controlled" describes the condition with the routine, not the condition underneath. It moves the useful question from "do I still feel it?" to "is the cause still there?" — and a second relapse at the same spot, with the same trigger, answers it. And it explains why the lesson usually arrives late and one condition at a time: each relapse teaches it again for that one thing, because nothing names the pattern across them. The inhaler, the moisturiser and the floss are one problem, and the person who stopped each of them was reading the evidence correctly — it was the evidence that was misleading.