Why You Freeze When Someone Yells at You

Someone starts shouting and you are gone. Not gone as in you left — gone as in the part of you that speaks has stopped being available. Your face goes still. You can hear the sound but you have stopped extracting sentences from it. Later, walking away, you will have a fully formed response, and by then it is useless.

If that is familiar, this page is about the specific case: not a row you were part of, but being yelled at. A parent. A manager. A partner whose volume climbs while yours does not exist.

Those are different situations, and lumping them together is why generic conflict advice so often fails to land here. If your experience is closer to shutting down partway through a two-sided argument with someone you are equal to, our guide on why you shut down during arguments fits that better. If it is going blank when someone challenges or accuses you directly, why you freeze when confronted covers that. This page stays with the one-directional case, where someone with more power in the moment is raising their voice at you.

Volume is its own signal

Notice something about the experience: you often cannot report afterwards what was actually said.

That is not carelessness. A raised voice functions as a threat cue in its own right, independent of content, and it is registered fast — before the deliberate, interpretive part of your thinking has produced anything. This is why being shouted at about something trivial can produce the same shutdown as being shouted at about something serious, and why people are sometimes bewildered by their own reaction to a raised voice in a supermarket that had nothing to do with them.

For a nervous system that learned, early and repeatedly, that shouting was the thing that came immediately before something worse, the sound is not the prelude to the danger. It is the danger. The learning is doing exactly what learning is supposed to do. It is simply calibrated to a situation you are no longer in.

Why freeze, and not fight or flight

The four survival responses are not selected at random. Which one your system reaches for reflects, in large part, what actually worked.

Consider the position of a child in a household where an adult shouts. Fighting back escalates it and is likely to make things worse. Leaving is not available — you cannot walk out, and often you cannot even leave the room. What remains is becoming small, still and unprovocative until it passes.

That is not weakness. In that setting it is the correct answer, and it was probably arrived at without any deliberation at all. The difficulty is that the strategy is durable. It outlasts the household. Twenty years later a manager's voice climbs in a meeting and the same programme runs, in a situation where you have options the child did not have — the ability to leave, to object, to end the relationship entirely.

Your body does not know that yet. It has not been updated.

What the research actually supports here

This is a topic where confident claims outrun the evidence, so it is worth being precise about what is established and what is inference.

What is well supported. Being subjected to parental verbal aggression is not a minor exposure. Teicher and colleagues assessed 554 people aged 18–22 and found parental verbal aggression associated with moderate-to-large effects on psychiatric symptoms — including dissociation, "limbic irritability", depression, anxiety and anger-hostility. In that sample the effects were comparable to those associated with witnessing domestic violence or non-familial sexual abuse, and larger than those associated with familial physical abuse. Exposure to several forms of maltreatment together frequently had an effect greater than the sum of the parts (*American Journal of Psychiatry*, 163(6), 993–1000, doi:10.1176/ajp.2006.163.6.993).

The dissociation finding is the one that speaks most directly to the going-blank experience this page is about.

The limits of that study. Its participants responded to advertisements, so the sample is self-selected rather than representative, and the design is cross-sectional — it establishes association, not causation. It is strong evidence that verbal aggression is not harmless. It is not evidence that it caused your particular freeze response.

What is inference, and should be labelled as such. The most rigorous research on involuntary freezing in humans comes from studies of assault, where immobility can be measured against a discrete event. In one study of 425 recent sexual assault survivors, the factors the researchers labelled *Dissociation & freeze* and *Reactive surrender* were far more pronounced than active resistance (doi:10.1080/20008066.2026.2673777). Related work finds that the severity of peritraumatic tonic immobility predicts later dissociative symptoms and negative post-traumatic beliefs — though notably, in that study, not overall PTSD symptom severity (doi:10.1002/jts.23182).

That literature demonstrates that involuntary freezing under threat is real, measurable and common. It does not study being shouted at. Applying it to a raised voice in a kitchen is a reasonable extrapolation, and it is still an extrapolation. Anyone telling you that a specific study proves what happens in your body when your partner shouts is going further than the research does.

What tends to help

None of the following is treatment, and none of it will make the reaction stop. What it can do is shorten it, which in practice matters more.

Get your eyes moving. Freeze tends to come with narrowed, fixed attention. Deliberately looking around — naming three objects to yourself, tracking the edges of the room — is one of the more reliable ways to interrupt the fixity, and it is invisible to the other person.

Change something physical. Pressing your feet down, shifting your weight, uncrossing your arms, standing if you were sitting. Small motor actions tend to be recoverable when speech is not.

Have one sentence ready. During shutdown, generating language is the hard part; retrieving a rehearsed phrase is much easier. "I need a minute." "I'm not able to talk about this right now." "I'll come back to this." Pick one, keep it short, and practise it when you are calm so it is available when you are not.

Expect the delay, and use it. The response that arrives forty minutes later is not wasted. Writing it down, or returning to the conversation once you can access it, is a legitimate way to participate — it is simply asynchronous. Many people find this reframe removes a large amount of shame.

Attend to the actual situation. This deserves saying directly: some of the time, freezing is information. If someone in your life yells at you regularly, the useful question is not only how to regulate yourself through it. Repeated verbal aggression is a pattern worth examining on its own terms, and that is a conversation for a therapist or a domestic abuse service rather than a self-help article.

A note on what this page is

Educational material, not a clinical assessment. Freezing when someone shouts does not by itself indicate a trauma history, a diagnosis, or anything about your character.

If you want a structured way to see which of the four responses is most familiar to you, our free trauma response quiz takes about two minutes and runs entirely in your browser — nothing is stored or sent anywhere. You may also want to read the fuller picture of the freeze response, or how survival responses affect decision-making, which is where a lot of people notice the after-effects.

Sources

• Teicher, M. H., Samson, J. A., Polcari, A., & McGreenery, C. E. (2006). Sticks, stones, and hurtful words: relative effects of various forms of childhood maltreatment. *American Journal of Psychiatry*, 163(6), 993–1000. doi:10.1176/ajp.2006.163.6.993
• Kreis, I., et al. (2026). Peritraumatic reactions during sexual assault: associations with assault characteristics and mental health. *European Journal of Psychotraumatology*. doi:10.1080/20008066.2026.2673777
• Barbano, A. C., et al. (2025). Posttraumatic stress disorder symptoms, posttraumatic cognitions, and dissociative experiences following rape. *Journal of Traumatic Stress*, 38, 963–975. doi:10.1002/jts.23182