It’s Not That You Fight. It’s What Happens Inside the Fight.

By the time a couple starts wondering whether the marriage is in real trouble, both partners are usually keeping a private tally. Two fights this week. Four last month. One of you counts them as evidence the marriage is failing. The other counts them as evidence that bringing things up only makes it worse. The tally feels like a diagnosis.

It measures almost nothing. For more than three decades, John Gottman and Robert Levenson recorded couples during actual disagreements, coded the interactions moment by moment, and then followed those couples for years to see whose marriages survived. How often couples fought did not separate the lasting marriages from the failing ones. Specific, observable patterns did. Eight of them, and most run quieter than a raised voice.

Anger Was Never the Predictor

Therapists in the 1970s assumed the poison was anger answered with anger. The observational data said otherwise: Stable, happy couples trade anger in kind too. What predicted divorce was escalation, the fight that climbs. A complaint gets answered with a counterattack, the counterattack with contempt, and within minutes the original topic has left the room entirely.

Four behaviors drive that climb, the ones Gottman named the Four Horsemen: Criticism, defensiveness, contempt, and stonewalling. Of the four, contempt, the eye-roll, the mockery, the talking down, is the strongest single predictor of divorce in his research.

Underneath the escalation question sits a balance question. During conflict, stable couples in Gottman’s lab produced about five positive interactions for every negative one. Couples headed for divorce ran closer to 0.8 to 1. Read that carefully, because the target is not zero negativity. Complaint, irritation, and anger do necessary work in a marriage: They flag what is not working, keep two people honest with each other, and force the relationship to keep adapting instead of calcifying. A marriage scrubbed of every negative feeling is not at peace. It is sedated. The goal is regulation, not eradication.

The Quiet Version Is More Dangerous Than It Looks

Some couples present the opposite picture: No blowups, and also no laughter, no questions, no interest, no affection. Two capable adults managing logistics side by side, neither one reaching for the other anymore.

The research finding here is one of the most sobering in the field. In their 14-year longitudinal study, Gottman and Levenson found that the two failure modes run on different clocks: Escalated negative affect during conflict predicted the divorces that came early, in roughly the first seven years, and did not predict the later ones. The divorces that came later were predicted by something subtler: The absence of positive affect during conflict. No warmth, no humor, no interest while disagreeing.

Loud marriages tend to break early. Quiet ones break late. If the two of you have stopped fighting because you have mostly stopped reaching for each other, the calm is not recovery. It is the roommates stage, and it has a longer fuse, not a smaller charge.

When the Friendship Erodes, Perception Flips

Robert Weiss named a mechanism he called sentiment override: The running verdict you hold about your partner starts overriding the actual content of what they say and do. In negative sentiment override, a neutral comment lands as a jab. A genuine olive branch reads as manipulation.

This is not a metaphor. Researchers Elaine Robinson and Gail Price placed trained observers in couples’ homes to count positive behaviors, and trained the partners to count the same things. In happy couples, the partners’ counts matched the observers’. In unhappy couples, partners registered only about half of the positive things their partner actually did. Half of the kindness was landing unrecorded, not because it was absent but because the filter no longer let it through.

The filter feeds on a familiar accounting trick: Your own lapses get situational explanations (exhausted, stressed, bad week) while your partner’s lapses get character explanations (selfish, cold, never really cared). Left running, it eventually recasts the past too, and the story of how you met starts getting told differently.

Here is the part that changes treatment: You cannot argue a perception filter back into accuracy, and litigating each misread comment one at a time goes nowhere. The filter is downstream of the friendship. When fondness and admiration are rebuilt, perception recalibrates on its own. When they are not, no amount of communication technique survives the subtext.

Your Body Ends the Conversation Before You Do

Every couple knows the moment a discussion stops being a discussion. Heart rate climbs, stress hormones dump into the bloodstream, and the body shifts into threat management. Hearing narrows. Working memory shrinks. Creative problem-solving goes offline. Each of you starts repeating your own position, louder and simpler, taking in nothing new. Gottman calls the state flooding, and once it starts, continuing the conversation produces damage, not progress.

On average, men flood faster and stay flooded longer, often replaying the fight internally in ways that keep the alarm switched on, but either partner can be the one who shuts down. This is also the hidden engine inside the pursue-and-withdraw loop: The partner who goes silent usually is not indifferent. They are physiologically overwhelmed, and the silence is a smoke alarm, not a verdict.

The intervention is unglamorous and effective: A real break of at least twenty minutes, spent actually calming down rather than rehearsing the argument, then returning to the topic. That is not avoidance. It is physiology management, and without it the rest of the skill set never gets a chance to run.

Influence Runs Both Ways or the Marriage Runs Down

In a study of 130 newlywed couples, Gottman and his colleagues John Coan, Sybil Carrère, and Catherine Swanson found that a husband’s unwillingness to accept his wife’s influence predicted divorce. In their heterosexual sample, wives were, on average, already accepting influence, which is why the predictive variance sat on the husband’s side. The refusal shows up two ways: Escalating hard against a partner’s ordinary complaint, or checking out of the conversation entirely. Either way, both of you end up alone with your own positions.

Accepting influence is not capitulation. It is the practice of letting your partner’s perspective actually move you, and in the data it behaves less like a courtesy and more like a load-bearing wall.

The Couples Who Last Are Not the Ones Who Never Blow It

Every marriage generates regrettable incidents: The sharp word, the missed cue, the fight that went somewhere ugly. The research is blunt that these are standard equipment, not defects. What separates lasting couples is not a cleaner record. It is whether repair attempts land: The half-joke mid-argument, the touch on the arm, the “let me try that again.”

And repair attempts land in proportion to everything above. The same olive branch reads as peace or as sarcasm depending on the sentiment override running underneath it, which is why repair skill and friendship repair are one project, not two.

What This Changes About Getting Help

None of these eight patterns is a personality verdict, and none of them is fixed. They are observable, measurable interaction habits, which means they can be assessed precisely and retrained deliberately. That is why couples work here starts with assessment and runs as a structured program with a defined endpoint rather than open-ended weekly conversation: The research names the specific systems that fail, and specific systems can be rebuilt.

If the two of you recognized yourselves more in the quiet sections above than the loud ones, take that seriously. It is the most common profile in struggling marriages, it is the one couples wait longest to address, and it is the one this work was built for. Schedule an initial consultation and we will map which of these patterns is actually running in your marriage.

FAQ