What Is Stonewalling

By Blubee Editorial, Editorial team · Updated August 10, 2026

On this page

Stonewalling is the withdrawal of a listener from a conversation: no response, no tracking, no visible reaction while the other person is still talking. John Gottman and Robert Levenson coded the behavior from video in a study of 73 married couples run in 1983 and 1987, where it separated couples heading for dissolution from couples who stayed together.

This page covers the operational definition and its origin, seven signs that show up in real arguments, the difference between stonewalling and physiological flooding, why the pattern starts, what it does to the person still talking, what stops it, and when silence is not stonewalling.

What stonewalling means#

Stonewalling is a listener behavior with a narrow technical definition: the person receiving the message stops emitting any signal that they are receiving it. Gottman and Levenson scored it from video and physiological recordings taken while couples discussed a real disagreement. Nothing in the measure comes from what either partner said about the argument afterwards.

The coded definition is stricter than everyday usage. In the 1992 paper, withdrawal from interaction was the sum of four negative listener behaviors: no response, not tracking, turning off, and incoherent talk. Intent is not part of the code. A person can be scored as stonewalling while privately trying to keep the peace, which is why the label often feels unjust.

In the 1992 coding, listener withdrawal is four measurable things: no response, no tracking, turning off, and incoherent talk. Intention never enters the definition.

The behavior sorts couples sharply. Across those 73 couples, listener withdrawal was scored 8.72 times per 15-minute conflict discussion in couples the researchers classified as non-regulated, against 4.82 times in regulated ones. By the 14-year follow-up published in 2000, the code had its own name in the scoring system, defined in the paper as listener withdrawal from marital interaction. Of the 79 couples in that longer study, 22 had divorced by year 14.

Stonewalling is the fourth of the behaviors Gottman named the Four Horsemen, after criticism, contempt, and defensiveness. It usually arrives last. The first three are things partners do to each other; the fourth is what one of them does after running out of capacity.

7 signs of stonewalling#

The signs of stonewalling are absences rather than actions, which is why they are easy to miss from the inside and impossible to miss from the outside. Each of the seven below describes something that stops happening while a conversation is still formally in progress.

1. The face stops moving#

The clearest marker is facial stillness. Ordinary listening produces a stream of micro-signals: eyebrows, blinks, small nods, a mouth that shifts with what it hears. In stonewalling these disappear and the face settles into a flat, unmoving mask.

Partners describe this as the moment the conversation died, and they are reading it accurately, because the listener has stopped transmitting. The stillness is not usually chosen. It is what a face does when the body has committed its resources elsewhere, which is why asking someone to look more interested produces only a worse imitation of interest.

2. Answers shrink to one syllable#

Speech does not always stop; it contracts. Replies collapse into fine, sure, okay, whatever, delivered at the same flat pitch regardless of what was asked. The content of the question stops affecting the shape of the answer, which is the actual tell.

A person who is still engaged gives different short answers to different questions. A person who has withdrawn gives the same one every time. Pressing for a fuller response tends to shorten the answers further, because each new demand adds load to a system that contracted because it was already overloaded.

3. The body turns away#

Withdrawal shows in posture before it shows in words. The shoulders rotate away from the speaker, the gaze fixes on a middle distance or a phone, the arms fold, the person stands up and starts doing something small and unnecessary.

Leaving the room is the visible extreme, but the same movement happens at a scale of a few degrees while both people are still seated. This is one reason arguments in cars escalate: neither party can rotate away, so the withdrawal has to happen entirely inside the voice and the face.

4. The conversation is redirected to logistics#

A subtler sign is the sudden appearance of admin. Mid-argument, the withdrawing partner asks whether the car needs fuel, mentions a work call tomorrow, or points out that it is late. This is rarely a manipulation. Practical talk is low-load and safe, and a nervous system looking for an exit reaches for the nearest one.

The partner still in the conversation hears it as proof that the relationship does not matter, which is close to the opposite of what is happening inside the other person. Both readings are sincere, both are held confidently, and they are incompatible.

5. Afterwards, the content is missing#

People who stonewall frequently cannot reconstruct what was said. They remember that the argument happened and they remember its temperature, but the specific sentences are gone. This is not necessarily convenient forgetting: the withdrawal is itself a marker that arousal was high while those sentences were being said.

The practical consequence is that agreements reached in the last twenty minutes of a bad argument are unreliable, because one person may never have processed them. Anything that matters should be restated later, in a calm conversation, rather than treated as settled because nobody objected at the time.

6. The shutdown arrives earlier each argument#

Stonewalling has a learning curve, and it runs the wrong way. In a new relationship the shutdown may take twenty minutes of hard conflict to appear. Years later it can arrive within the first two sentences, triggered by a tone of voice rather than by anything actually said. The nervous system has learned to predict where this class of conversation ends, and starts protecting early.

This is the point at which couples describe not being able to talk about anything at all. It is a description of timing rather than of subject matter, and it explains why the list of forbidden topics grows without anyone agreeing to it.

7. Everything returns to normal once the subject changes#

The most confusing sign is how fast the person comes back. Ten minutes after a conversation ended in a wall, they are cheerful about dinner, warm, apparently fine. To the partner still carrying the unfinished argument this reads as indifference, or as punishment being lifted. Read physiologically it is simpler: the load was removed, so the system recovered.

The distinction matters, because a person who has shut down and a person withholding warmth to make a point need opposite responses.

Stonewalling vs flooding vs the silent treatment#

Stonewalling, flooding, and the silent treatment are three different things that produce the same silence, and confusing them is the most common error in writing about this topic. Flooding is what the body is doing, stonewalling is what an observer can see, and the silent treatment is a tactic with a goal. Only the third one is aimed at the other person.

Flooding, which Gottman also calls diffuse physiological arousal, is the internal state: heart rate climbing, blood vessels constricting, attention narrowing. The Gottman Institute uses a heart rate above 100 beats per minute as the practical threshold past which a person can no longer take in what a partner is saying, however hard they try. Stonewalling is what that state looks like from outside when the response is freeze rather than fight.

The silent treatment differs on the one criterion that is hard to fake: it survives the removal of load. Someone who was flooded warms up once the pressure is off, sometimes embarrassingly fast. Someone using silence as leverage stays cold, because staying cold is the instrument.

Why stonewalling happens#

Stonewalling begins as an attempt to stop a conversation from getting worse, and it is strongly patterned by physiology. Gottman and Levenson measured husbands stonewalling more than wives across their 79 couples, at 1.07 codes per turn at speech against 0.84, and the Gottman Institute puts the male share of stonewallers in that body of research at 85 percent. The standing explanation is that the male cardiovascular system takes longer to return to baseline after conflict arousal, so withdrawal becomes the cheaper option sooner.

The 1998 study of 130 newlywed couples pushed this further. The absence of physiological soothing of the husband was among the processes that predicted divorce, in a study that predicted divorce and stability with 83 percent accuracy. The shutdown itself is not the fatal element. What matters is whether anything in the couple’s repertoire brings the aroused partner back down.

History matters too. People raised where raised voices preceded something frightening shut down faster. People who learned that expressing a need produced withdrawal escalate faster.

Those two histories pair up with striking regularity, which is where the four attachment patterns become useful as a map. The partner who avoids closeness under stress and the partner who reads silence as abandonment are each doing something that made sense earlier in life.

What stonewalling does to the partner who keeps talking#

Stonewalling is not a neutral pause for the person on the other side of it. Human beings treat an unresponsive face as a signal, not as an absence. The standard response is to raise volume, precision, and frequency in pursuit of any reaction at all. That raises the load on the withdrawn partner and makes the next shutdown faster.

The loop has a name in the research literature: demand and withdraw.

The evidence on it is substantial. A 2014 meta-analysis pooled 74 studies covering 14,255 participants and found a moderate association between demand/withdraw and worse individual, relational, and communicative outcomes, at r = .360 overall and r = .413 in distressed and clinical samples. The pattern is not a stylistic quirk. It tracks with how the relationship goes.

The loop is symmetrical in cost even when it is asymmetrical in appearance. Whoever is pursuing is also making the shutdown arrive sooner.

Being the person who keeps talking is its own experience: the sense of arguing with furniture, then the escalation, then guilt about the escalation. If that fits, the partner who pursues harder is worth reading alongside shutting down under pressure, because the two describe one loop from opposite seats. A short questionnaire will show where you land on avoidance and on anxiety, with results shown immediately and no email required.

How to stop stonewalling#

Stopping stonewalling works by changing the physiology first and the conversation second. The order is not negotiable: a flooded person cannot use communication advice. Each step below carries a condition.

Say what is happening before you go quiet#

The single highest-value change is a sentence delivered before the shutdown completes: something like I am at my limit and I am going to stop making sense soon. This converts an invisible internal event into shared information. It works only if it is said early, while speech is still available. Said after the wall is up, it sounds like an exit line, because by then it is one. Deciding the sentence in advance rather than in the moment is the same mechanic as changing one behavior deliberately: one action, one moment it fires.

Take at least twenty minutes#

The Gottman Institute puts the minimum useful break at 20 minutes, on the grounds that the stress chemistry driving the shutdown clears through the bloodstream rather than switching off on command. Shorter pauses return both people to the argument still aroused, which is why the second round is usually worse than the first. Spend the time on something genuinely absorbing, not on rehearsing the argument, which keeps arousal exactly where it was.

Name a return time, and return#

A break without a stated return time is indistinguishable from abandonment, and the pursuing partner will treat it that way. Name a specific time before leaving the room, then keep it, including when the subject has lost its urgency. Keeping the appointment is what makes future breaks acceptable to the other person. Skipping it twice teaches them that asking for a break means the conversation is over.

Change the first minute, not the whole argument#

Withdrawal is often a response to how a conversation opens rather than to its content. A specific complaint about a situation, rather than a summary judgment about a person, keeps the load low enough that the listener stays in the room. This is the cheapest intervention and the least satisfying, because it asks the partner who is not withdrawing to move first.

Use apps for tracking these patterns to check whether the openings are actually changing, and other self-assessments here for a wider read on the dynamic.

When stonewalling is not the explanation#

Stonewalling is a specific coded behavior, not a synonym for being quiet, and the label is applied far more often than the behavior occurs. Silence during a difficult conversation has several ordinary causes, and treating all of them as one of the Four Horsemen produces a false accusation with a scientific accent.

The clearest boundary is tracking. Someone who pauses before answering, asks for a minute to think, or says they need to sit with something is still receiving the conversation, and the original coding scheme would not score them as withdrawn. Thinking is not stonewalling, and neither is processing more slowly than a partner would prefer.

Sensory or cognitive overload is a second exception. Some people go non-verbal under noise, fatigue, or accumulated demand in ways that have nothing to do with the relationship, and the useful response there is environmental rather than relational. Grief, acute illness, and the weeks around a work crisis produce the same picture temporarily.

The gender framing deserves its own caution. Men do stonewall more frequently in the measured samples, but frequency is not harm. The 2014 meta-analysis found essentially the same effect size for wife-demand and husband-withdraw, r = .380, as for husband-demand and wife-withdraw, at r = .392.

The pattern costs a relationship at a comparable rate whichever partner takes which half of it. A version of this conversation in which one gender owns the problem will not fix it.

Finally, silence used deliberately to control someone else’s behavior is a different pattern with a different remedy, and no self-soothing technique addresses it. If quiet is applied and lifted according to whether you gave in, the useful conversation is with a licensed professional about the relationship as a whole, not about conflict tactics.