Published by Unseen Progress, an independent publisher of caregiver research. Last reviewed 2026-05-10. Part of the selective mutism research overview.
Short answer. Plateaus and regressions are predictable phases in selective mutism treatment, not signs of failure (Bergman et al., 2013; Kotrba, 2015; McHolm, Cunningham & Vanier, 2005). Most stalls fall into one of four patterns — generalisation gap, setting transition, premature step-up, or unaddressed comorbidity — and each has a different research-backed response. The clinical literature treats plateau as a diagnostic event: it tells you which variable to adjust, not that the protocol has stopped working.
Behavioural change in selective mutism is not linear. The early gains — first words to a clinician, first slide-in success with a grandparent — accumulate rapidly because the child's existing speaking circle is being incrementally extended. Each new person is structurally similar to the last. The first time the protocol meets a categorically different setting — the school classroom, the doctor's office, a peer's birthday party — the speaking behaviour generalises more slowly. This phase shows up as a plateau even when the underlying treatment mechanisms are still working.
Kotrba (2015) and McHolm, Cunningham and Vanier (2005) both describe this as the generalisation tax: the cost paid every time the child encounters a setting where the speaking behaviour has not yet been demonstrated. The tax is highest at the first instance of each setting category — first school, first peer, first authority figure outside the family — and decreases as the child accumulates evidence that new categories are survivable.
The child is speaking comfortably with the trusted adult and the introduced new person, but the speech is not transferring to similar people or settings on their own. Sliding-in is working with each individual but not generalising across the class.
What the research suggests: explicit generalisation programming. Vary the setting, the time of day, the activity, and the new person in deliberate combinations rather than expecting transfer to happen passively. Bergman's protocol and the SMart Center materials both describe this as planned generalisation — the protocol does not generalise unless you make it generalise.
The child is speaking in one setting (e.g. small therapy room) and the protocol moves to a meaningfully different setting (e.g. classroom). Speech volume drops or disappears entirely. This is not regression in the technical sense — it is the same skill being asked of a categorically harder environment.
What the research suggests: treat the new setting as a fresh slide-in problem. Re-establish the speaking baseline with the trusted adult in the new setting before introducing new people there. The fact that the child speaks freely in the old setting is preserved progress, not erased.
A successful slide-in or ladder rung was followed by a larger jump than the child was ready for. The next session showed reduced speech or full freeze. The protocol overshot.
What the research suggests: step back one or two rungs and rebuild. McHolm, Cunningham and Vanier (2005) and Kotrba (2015) both emphasise that "stopping while still in the speaking zone" is more important than session length. A plateau caused by overshoot resolves within 2–4 sessions of stepping back to a known-comfortable rung.
The plateau coincides with — or is caused by — something the protocol is not targeting: a new sibling, a school transition, an emerging comorbid anxiety, an undiagnosed learning issue, parental stress, a move, illness. The treatment plan was correct, but the surrounding variables changed.
What the research suggests: audit the broader context before adjusting the protocol. A plateau after a move, a new teacher, or a major family change is rarely a treatment failure — it is a load increase. The protocol may need to pause at the current rung until the surrounding load decreases, rather than push forward.
Regression — actual loss of previously demonstrated speaking behaviour — is real and meaningfully different from plateau. The research-backed distinction:
Most caregiver reports of "regression" are mild regression, not significant regression. The literature describes mild regression as expected after summer breaks, school transitions, holidays, and family disruptions (Kotrba, 2015). The protocol does not need to restart from zero — it needs to re-establish the previously demonstrated rung quickly, then continue.
A particular case worth naming: most parents observe a regression in selective mutism progress around the start of each school year. The child finished June speaking quietly to one teacher, returns in August/September unable to speak to anyone in the new classroom. This is the most common single regression pattern in the clinical record (Kotrba, 2015; McHolm, Cunningham & Vanier, 2005) and is predictable enough to plan for. The Selective Mutism Association recommends explicit school-side sliding-in protocols that begin before the first day of school, not after the regression has happened.
1. Wait 4–8 weeks before concluding plateau. Selective mutism progress is measured in months, not days. A two-week slowdown is not yet a plateau. 2. Identify which of the four patterns applies. Run through generalisation gap, setting transition, premature step-up, and unaddressed comorbidity. The pattern dictates the response. 3. Re-establish the previously demonstrated rung before pushing forward. Especially after any regression — even mild — the next session should be set at a rung the child has previously succeeded at. 4. Audit the surrounding context. New school year, new teacher, family change, illness, the parent's own stress. The protocol does not work in isolation from these variables. 5. Re-evaluate comorbid anxiety. If a plateau persists beyond 8–12 weeks with no contextual explanation and no protocol adjustment having worked, the literature suggests reassessing whether comorbid social anxiety has intensified (see the comorbid social anxiety question) and whether an SSRI consultation is appropriate (see the SSRI question). 6. Resist restarting the entire ladder. A common error is to scrap accumulated progress and start over. The previously demonstrated rungs are preserved; the plateau is a signal to adjust one variable, not to abandon the plan.
Plateau is not evidence the child cannot recover. Plateau is not a reason to stop treatment. Plateau is not a moral failing of the parent or child. The longitudinal record (Bergman et al., 2013; Kotrba, 2015) shows that nearly every successfully treated case included visible plateaus inside the timeline — they are part of the path, not a deviation from it.
---
Unseen Progress publishes long-form caregiver research and builds research-backed daily trackers for the families covered. See the full selective mutism research overview for the complete framework.