Speaking is a lightning-fast relay race
A sentence is not one solid block: it is a line of small pieces of sound — the syllables — that have to set off one after another, at the right speed. For each one, the brain runs the same full lap: it gives the go, sends the command to the muscles of the mouth, the sound comes out into the air, returns through the ear, and along the way the brain checks that it sounds as expected.
When all goes well, that lap repeats dozens of times a minute without you giving it a single second of thought.
Stuttering is a go signal that doesn’t land right
What sets off each syllable is a signal produced by deep structures in the brain — the basal ganglia — in constant conversation with the cortex, the brain’s outer layer. This is the trail most of the research is following today: in a person who stutters, this signal struggles to arrive on time.
The syllable is ready. The person knows exactly what to say and exactly how to say it. But the “go” runs late. Sometimes it half arrives: the syllable takes off, the sound begins, then falls back, and it has to start over from the top — that is what you hear in “c-c-coffee”. Sometimes it never comes at all, and nothing comes out: that is a block.
Why it changes from one moment to the next
The same day can go from easy to impossible. On the phone, in front of a class, whenever the stakes go up, it gets harder: this starter is sensitive to pressure. Careful not to turn the cause around: stress is not what causes stuttering — it makes a difficulty that is already there worse.
And when singing, or speaking along with other people, it almost always disappears. Nobody is sure why. The usual explanation is that the rhythm then comes from outside and stands in for the inner go signal — though the researchers who proposed that say it does not account for singing, and a 2025 experiment found speaking in unison shifts timing in opposite directions in people who stutter and people who do not. What is not in doubt is that it happens, and that “just relax” does nothing at all.
Why some people and not others
It almost always begins between the ages of 2 and 5, exactly at the moment when speaking is becoming automatic. Genetics count for a great deal: it is common within a single family. Out of ten children who start stuttering, about seven stop on their own as they grow up — often slowly, over several years.
The inherited part is real, but there is no “stuttering gene”. The largest genetic study ever run, covering more than a million people, found dozens of places across our DNA involved, each one with a minuscule effect. Nothing the person did, and nothing their parents did, caused it.
So what is actually different?
We would love to be able to point at one part of the brain and say: there, that’s it. That is not what science found. When researchers compare hundreds of brain scans, they see very small average differences, gathered in two places: the speech network on the left side, where the words are put together, and the small deep areas that give the go signal. But the moment anyone tries to say exactly where, and exactly by how much, the teams stop finding the same thing. Most of these studies rest on a few dozen people at a time, which is part of why the picture keeps moving.
We have to be honest about what we do not know. We do not know whether these differences were there before the stuttering began: nobody has ever been able to scan a child’s brain before their first stutter, which arrives at around two or three years old. What we do know is that the differences seen in children and the ones seen in adults are not the same. And above all: several of them turn up in children who later stopped stuttering. Having a slightly different brain is therefore not a mark, not a prediction, and not a verdict.
So can you look at a brain and tell whether that person stutters? No. These differences are averages, and the two groups resemble each other so closely that they very nearly merge into one. It is a little like height: on average, adults are taller than twelve-year-olds, and yet, standing in front of someone five foot three, you cannot tell. The brain is the same, only far more subtle still. There is no scan anywhere that spots stuttering. Stuttering is something you hear when a person speaks; it is not something you see.
What this page does not claim
What you are looking at is a picture made for understanding, not a photograph of a brain: the areas and the pathways are simplified, and some distances are exaggerated to keep everything readable.
Above all, the exact cause of stuttering is not settled. The model shown here — a go signal that misses its moment — is one leading proposal among several, and the researchers who work on it call it speculative in their own 2025 review. No precise area of the brain has been pinned down reliably. A rival idea, that the trouble lies in how speech is checked by ear, is not settled either way — the same review calls it one of the most promising lines of inquiry. What does not hold up is the dopamine story, which rests on a single 1997 study of three people, never reproduced since; and the idea of something broken or cut through in the circuits — nothing is, every pathway is whole and in its place.
What is solid, on the other hand: this is a matter of speech circuits, not a matter of character.
Where this comes from
- Chang SE, Below JE, Chow HM, Guenther FH, Max L, Watkins KE, Bernstein Ratner N, et al. (2025). Stuttering: Our Current Knowledge, Research Opportunities, and Ways to Address Critical Gaps. Neurobiology of Language. free full text — the field’s own stock-take. Calls the timing model “speculative”, and auditory-motor integration “one of the most promising lines of inquiry”.
- Polikowsky HG, et al. (2025). Nature Genetics. free full text — 1,122,019 people (99,776 who stutter). 57 regions of the genome, each with a minuscule effect. This is the study behind “there is no stuttering gene”.
- Chang SE & Guenther FH (2020). Involvement of the Cortico-Basal Ganglia-Thalamocortical Loop in Developmental Stuttering. Frontiers in Psychology 10:3088. doi:10.3389/fpsyg.2019.03088 — the go-signal model this page draws.
- Alm PA (2004). Stuttering and the basal ganglia circuits: a critical review of possible relations. Journal of Communication Disorders 37(4):325–369. PMID 15159193 — where the timing account was first argued at length.
- Wu JC, et al. (1997). PMID 9106763 — the dopamine study. Three people who stutter, six controls, no imaging replication in the thirty years since. Included so you can see for yourself how thin it is.