The argument gets louder and something in the body switches off. The words are somewhere; they cannot be reached. What comes out is I don't know, or nothing at all, and from the other side of the table the nothing reads as a verdict: you do not care, you are not listening, you have already left.
None of that is what is happening inside. What is happening inside is one of three things, and they look identical from outside, which is why the silence is so reliably misread.
85%
of stonewallers in Gottman's lab were men
74
studies of demand and withdraw, 14,255 people, pooled in one analysis
13%
of a non-clinical sample reported freezing during a lab stressor
20 min
the minimum break Gottman's research recommends for a flooded body to come down
Silence one: the freeze
Freezing is the third of the four stress responses, after fight and flight and before fawn, and it is the one with the oldest biology. Karin Roelofs's 2017 review in the Philosophical Transactions of the Royal Society described what the body does: the heart slows rather than races, attention narrows, and the whole system goes still in a state that is not passive at all. Freezing is preparation. The animal that goes still is gathering information before it moves, and so is the person who goes blank in a kitchen.
It is measurable in ordinary people. Norman Schmidt and colleagues gave a non-clinical sample a laboratory stressor in 2008, a single breath of carbon dioxide enriched air, and thirteen percent reported the sensation of being unable to move. The people who froze were the ones who reported more panic and more anxiety during the challenge. The freeze is not rare, and it is not confined to people with a history.
What it feels like from inside is specific. The words are present and unreachable. Time slows. The urge to say something and the inability to say it coexist. And it ends, usually within minutes, at which point the sentence that was needed arrives, in full, too late. The entry on fight, flight, freeze and fawn sets this response next to its siblings.
Silence two: flooding
The second silence is longer and has a different engine. John Gottman's lab at the University of Washington spent decades wiring couples to heart monitors during real arguments, and found that above a certain level of arousal the conversation effectively ends whether or not anyone is still talking. Gottman called the state flooding: the pulse high, the stress hormones up, the part of the mind that handles nuance losing bandwidth to the part that handles threat. A flooded person cannot take in new information, cannot read the other face, and cannot produce anything more constructive than repetition or retreat.
Stonewalling is what flooding looks like from outside. The listener withdraws, shuts down, stops responding, and the Gottman Institute's own summary of the lab findings is blunt about who does it: eighty-five percent of the stonewallers they studied were men. The explanation offered is physiological rather than moral. Male cardiovascular systems, on average, react harder to conflict and take longer to return to baseline, so the point at which a conversation becomes unbearable arrives sooner and stays longer.
The Gottman research also produced the one intervention with real evidence behind it. A break of at least twenty minutes, because that is roughly how long the body takes to come down. Not a break that is taken by leaving the room in silence, which is withdrawal and lands as abandonment. A break that is announced, with a return time.
Silence three: the punishment
The third silence is the one the other two get mistaken for. The silent treatment is chosen. It is withdrawal used as a message, and the research on it is unambiguous about the cost.
A 2026 systematic review in Frontiers in Psychology covered fifteen studies and 2,417 participants. The silent treatment was associated with low agreeableness and high neuroticism in the person using it, was triggered by unmet needs and perceived neglect, and was strongly tied to rumination during the silence itself, which in turn predicted emotional exhaustion for the person doing it. On the receiving end: loneliness, lowered self-esteem, and the feeling of being excluded from one's own relationship. Greater commitment predicted less use of it, which is the review's most hopeful finding. People who are invested tend to reach for repair instead.
| The silence | What it feels like inside | What it looks like from outside | What it needs |
|---|---|---|---|
| Freeze | The words are there and cannot be reached; it ends within minutes | Blankness, I don't know, a stare | One prepared sentence, said badly, to end the moment |
| Flooding | Too much, too fast; the only available move is out | Shutting down, leaving, monosyllables | An announced break with a return time, at least twenty minutes |
| Silent treatment | A decision, with a message attached | The same shutting down, held for hours or days | A different conversation, about the need underneath the punishment |
The middle column is the whole problem. From across the table, the three are indistinguishable. From inside, they are three different things with three different fixes, and the person who freezes is very often being treated as though they were choosing silence three.
What 74 studies say about the dance
The silence almost never happens alone. It is half of a pattern that has its own name in the research, demand and withdraw: one partner pursues, raises the issue, presses for a response, and the other retreats, and each response makes the other's more likely. Christensen and Heavey described it in 1990. In 2014 Paul Schrodt and colleagues pooled every study they could find.
- 255 people: 0.36
- Wife demands and husband withdraws: 0.38
- Husband demands and wife withdraws: 0.39
- Couples already in distress: 0.41
- Couples not in distress: 0.35
Three things in that chart matter. The association is moderate and consistent, which in this field is a large effect. It is very nearly the same whichever partner does the withdrawing, so the pattern is the problem rather than the gender. And it is present in couples who are not in trouble, at almost the same strength as in couples who are. The dance does damage before anyone would call the relationship distressed.
The mechanism is easy to see once named. The pursuer reads the silence as a verdict and pursues harder. The withdrawer reads the pursuit as attack and withdraws further. Both are responding to a stress response in the other as though it were a position, and neither has been told which of the three silences they are looking at.
Silence as regulation, silence as withdrawal
Not every retreat is a retreat from the relationship. The evidence is clear that a flooded person who keeps talking does more damage than one who stops. The question is not whether to go quiet. It is whether the quiet is declared.
The last row is the connection to sleep. An argument that ends in an undeclared silence has nowhere to go, so it waits for the pillow, and the piece on sleep after an argument describes what it does there. Haydon and Salvatore's 2023 study of 208 couples wearing sleep monitors found broken sleep highest on days of relationship stress, and the link strongest in people with avoidant attachment, who are, not coincidentally, the ones most likely to withdraw.
Reading your own silence
None of this is treatment. A freeze that arrives in ordinary conversations, not only in heated ones, or a shutdown that never fully ends, belongs with a professional. What a questionnaire can do is smaller and often enough: say which of the three silences is yours, and which one your partner is seeing.
The Stress Response test reads the first move. The Conflict Style test reads the second. Read next to a partner's results, the two turn an argument about who does not care into a picture of two nervous systems doing what they were built to do, which is a much easier thing to negotiate with than a person.
The first change is not feeling more confident in arguments. It is a frozen moment that ends in a sentence rather than in leaving.
Sources
- Schrodt, P., Witt, P. L., Shimkowski, J. R. (2014). A meta-analytical review of the demand/withdraw pattern of interaction and its associations with individual, relational, and communicative outcomes. Communication Monographs, 81(1), 28-58.
- Gottman Institute. The Four Horsemen: stonewalling.
- Roelofs, K. (2017). Freeze for action: neurobiological mechanisms in animal and human freezing. Philosophical Transactions of the Royal Society B, 372(1718).
- Schmidt, N. B., Richey, J. A., Zvolensky, M. J., Maner, J. K. (2008). Exploring human freeze responses to a threat stressor. Journal of Behavior Therapy and Experimental Psychiatry, 39(3).
- Dubey, A., et al. (2026). Antecedents and consequences of silent treatment in close adult relationships: a systematic review. Frontiers in Psychology, 17.
- Christensen, A., Heavey, C. L. (1990). Gender and social structure in the demand/withdraw pattern of marital conflict. Journal of Personality and Social Psychology, 59(1).
- Haydon, K. C., Salvatore, J. E. (2023). Relationship stress, arguments, and sleep quality: a causal process analysis. Journal of Social and Personal Relationships.
- Bracha, H. S. (2004). Freeze, flight, fight, fright, faint: adaptationist perspectives on the acute stress response spectrum. CNS Spectrums, 9(9).
- YouGov (2024). People-pleasing poll: 71 percent of US adults say they often go to great lengths to avoid conflict.