40 Years of Gottman Research on Contempt

A research roundup on contempt: the single strongest predictor of divorce in Gottman's longitudinal data. What it is, what it is not, and what to do tonight if it has slipped into your relationship.

# What 40 Years of Gottman Research Actually Says About Contempt

If you read one piece of relationship research in your life, make it the work on contempt. Across 40 years and thousands of couples, John Gottman's lab found that one specific behaviour predicts divorce more powerfully than anything else they measured. Not arguing. Not money fights. Not differences in libido. Contempt.

This post is a research roundup, not opinion. We will define contempt the way the research defines it, summarise the studies, separate it from the things people often confuse it with, and translate it into something you can actually do something about tonight.

## What the research found

Gottman's "Love Lab" began studying couples in the 1980s. By following hundreds of couples for over a decade in some studies, the team identified four behaviours so corrosive they called them the Four Horsemen: criticism, defensiveness, stonewalling, and contempt.

All four predict relationship distress. Contempt predicts divorce specifically. In the longitudinal data, the presence of contempt in the way one partner spoke about or to the other was the single strongest behavioural predictor of separation, even when controlling for other factors.

Contempt also showed up in physical health data. Couples in whom contempt was frequent had measurably more infectious illness over the following years than couples without it. The hypothesis: contempt elevates chronic stress in both the contempt-giver and the contempt-receiver, which suppresses immune function. The research is correlational, but consistent.

## What contempt actually is (and what it is not)

This is where most popular summaries get sloppy. Contempt is not anger. Anger is hot, present-tense, and often productive. Anger says "I am upset." Contempt says "I am better than you." It is the position of looking down on your partner from above.

Specific markers of contempt in the research:

- **Mockery and sarcasm**, especially in the form of repeating what the partner just said in a mocking voice. - **Eye-rolling**, especially in response to the partner speaking. - **Name-calling**, even in apparent jest. "Idiot." "Lazy." "Crazy." Especially when said with a smile. - **Hostile humour at the partner's expense**, particularly in front of others. - **Moral disgust**, the position of "I cannot believe I am even with someone like this."

Things often mistaken for contempt that are not:

- Sharp criticism. (That is criticism, the first horseman, also corrosive but different.) - Going quiet during a fight. (That is stonewalling, the third horseman, often a flooding response.) - Strong disagreement. (Couples can disagree intensely without contempt.)

The distinguishing question to ask: am I expressing that something my partner did is wrong, or am I expressing that my partner, as a person, is beneath me. The first is repairable. The second is the predictor.

## Where contempt comes from

The research suggests contempt almost always grows out of long-running, unaddressed admiration deficit. When one partner stops actively noticing what they admire about the other, the unnoticed gap fills with the things they do not like. Repeated for long enough, those become the default lens. The partner is now seen primarily through the negative.

This is one of the few places where the research and the practical advice align cleanly. The most reliable behavioural antidote to contempt is not "stop being contemptuous." It is "rebuild active admiration." Contempt cannot easily coexist with frequently expressed, specific appreciation.

## Specific moves the research supports

The Gottman team and others have studied which interventions actually move couples out of contempt. The best-supported are not therapy techniques in themselves; they are behavioural practices the couple does daily. The three with the most evidence:

**1. The 5:1 ratio.** In stable relationships, partners exchange roughly five positive interactions for every one negative one, even during conflict. Couples in distress are often closer to 1:1 or worse. Bringing the ratio back up does not require grand gestures. It requires small, frequent, specific positives. "I noticed you got the kids to bed without me asking." "Thank you for the coffee." Daily.

**2. Daily appreciation, said out loud.** Not thought. Not implied. Said. Once a day, name one specific thing your partner did that you appreciated. The specificity matters. "Thanks for being you" does not move the needle. "Thank you for handling the call to your mother last night" does.

**3. Repair attempts, accepted.** The research found that what distinguished couples who recovered from rupture was not the absence of negative behaviour. It was that when one partner made a repair attempt (even a clumsy one, like a corny joke in the middle of a fight), the other partner accepted it. The willingness to accept the repair, more than the elegance of the repair itself, is the variable.

## What this means for live two-person AI sessions

Contempt rarely shows up in a single sentence. It shows up in tone, frequency, and pattern across sessions. One of the things a live AI session is genuinely useful for is naming, in real time, when contempt markers appear:

> "I noticed that just now, when you repeated what your partner said, the tone shifted. That is the kind of move that, repeated, becomes contempt. Can you say what you were actually feeling underneath that?"

The intervention is not punitive. The intervention names the marker, asks for the underlying feeling, and gives the partner a chance to express it without the contempt frame. Done in the moment, this is one of the most repair-positive things either a human therapist or an AI session can do.

## The bottom line

Of all the things to be vigilant about in a long relationship, contempt is the one. Not because it is the worst feeling, but because it is the one with the strongest predictive validity in the longest-running couples research we have. If you can keep contempt out of your relationship, you have already moved your odds significantly.

Sources for further reading: - Gottman, J. M., & Levenson, R. W. (2000). The timing of divorce. *Journal of Marriage and Family*, 62(3). - Gottman, J. M. (1994). *What Predicts Divorce?* Lawrence Erlbaum. - Gottman Institute research summaries (gottman.com).