Two decades ago, a team of researchers led by psychologist John Gottman set out to determine one thing:

Why do couples get divorced?

Gottman decided to answer this question by trying something very simple: Recording married couples talking for 15 minutes about a recent conflict that they were having in their relationship, and then carefully scrutinizing these recordings to see how happy and unhappy couples behaved differently. After all, every couple has problems; the simple act of fighting can't possibly be the only thing that drives a couple to divorce. There must be something in particular about the nature of the fights themselves that distinguishes happy from unhappy couples. After gathering these recordings from about 80 married couples throughout the Midwest, Gottman and his colleague Robert Levenson underwent the grueling task of coding these videos. This means that they made a note of every single time that certain things happened in the interaction. Was one partner angry? Was the other one getting defensive? How much did they use humor in their interaction? Did they show any affection? How about the nasty silent treatment - did that ever rear its cold, ugly head?

After keeping track of these couples and noting which ones ended up getting divorced over the course of the next 14 years, Gottman and Levenson eventually realized something incredibly important: They didn't actually need to note down all that much. In fact, there were just four behaviors that could be used to predict which couples would still be married 14 years later -- with 93% accuracy.

Yes; in case the enormity of what I just said didn't sink in quite yet, solely based on how often you notice four behaviors occurring in a single, 15-minute conversation, you can predict with 93% accuracy whether or not a couple will still be married 14 years from now.1

Now I'm guessing you probably want to know what these four behaviors -- or, as Gottman and Levenson call them, the Four Horsemen of the Apocalypse -- actually are. These four toxic behaviors are called contempt, criticism, stonewalling, and defensiveness.

And, funny enough, to understand what each of these behaviors looks like in action, one needs to look no further than America's favorite briefly-unhappily-married couple: Socialite Kim Kardashian and "basketball player" Kris Humphries.


Couples who eventually divorce express over twice as much contempt during disagreements as those who stay together for the long haul. In fact, Gottman himself believes that of the four "horsemen," contempt is the most significant one.

What does contempt look like? It's more than mere anger; all couples become upset or angry with each other at times, and this certainly does not mean that they will all divorce. Contempt in particular is a potent mix of anger and disgust. Expressing contempt involves speaking to your husband like he is "beneath" you, or mocking your wife in a cold, sarcastic way.

The clip below, from Keeping Up With The Kardashians, certainly elicited a lot of laughs when it aired. And many (including myself) thought it was kind of funny that Kris was clearly giving Kim a "reality check" about her likely-fleeting fame. Yet when considering their relationship quality, his response is completely toxic. It's clear in what Kris says to Kim that he didn't respect her or her priorities. It would be possible for these two to fight about where they should live without expressing contempt. Yet by telling her to her face that her career is essentially worthless - whether or not that is actually the case - he's expressing contempt towards her. No good for their ill-fated marriage.


The second horseman is criticism, which might immediately worry anyone who's ever complained about a partner forgetting to empty the dishwasher. However, the toxicity of criticism does not emerge in a disagreement where the partners are simply voicing any minor (or major) concerns that they might have. Criticism specifically involves turning your complaints into some sort of "defect" about your partner's personality. Rather than voicing constructive complaints about a behavior, situation, or incident, criticism specifically involves negative trait (not state) attributions.

In other words: A complaint focuses on the behavior. A criticism attacks the person.

We can see this in the following TV clip where Kim rants about her pet peeves. The very first one that she mentions is Kris's habit of brushing his teeth so vigorously that he gets toothpaste on the mirror (seriously, people -- you can't make up these scintillating conversations). But note how she says it. She doesn't say that it bothers her when he does this. She specifically notes that she hates the kind of people who brush their teeth so vigorously they get toothpaste all over the mirror. She has managed to take something fairly minor and, rather than phrasing it as a complaint ("It really bothers me when you do this. Could you try to brush over the sink, or at least wipe off the mirror when you're done?"), she has turned it into a weird, dental-centric criticism of his character ("You're the kind of person who messes up the mirrors when you brush your teeth!") Over time, these trait- (or personality-)based attributions can build up and lead to resentment or a lack of respect for one's partner, which will quickly breed that earlier sense of contempt.


This strategy can best be summed up by one, simple line: "It's not me. It's you."

Do you often "play the victim" in your fights, maybe by making it seem like everything that happens is your partner's fault? Do you regularly deny responsibility for any role that you might play in a conflict? Do you find yourself trying to "prove" during a fight that your partner is "more wrong" than you are?

If so, you might be guilty of defensiveness.

It's only natural to believe in a disagreement that we are right and our partners are wrong. After all - if we didn't think that we were right, we probably wouldn't be fighting! But the important thing about defensiveness is that it involves a tendency to consistently blame things on one's partner, paint oneself as the "martyr" or "victim" who does nothing wrong, and make it seem like the partner is responsible for anything that goes wrong in the relationship.

Below, we can see Kim flip her switch to Defensiveness Mode the very second that something goes wrong in an interaction with Kris. She is perfectly happy to be playing around with Kris by the water, but as soon as something goes wrong (she loses an incredibly expensive earring in the ocean), the entire incident immediately becomes his fault. She accepts no blame for any part she might have played in the disagreement, like also horsing around...or even simply wearing earrings that expensive around the ocean. Instead, she becomes the "victim," and Kris becomes the villain. Unfortunately, no one likes being the villain all of the time -- which is why defensiveness can suck the life out of your relationship.


Stonewalling is when, during a conflict, one (or both) partners will completely tune out from the discussion -- maybe by texting, turning on the TV, or simply not responding to a partner's attempts at conversation. Stonewalling is usually accompanied by increased physiological responses like an accelerated heart rate, higher blood pressure, and sweating. This indicates that it might be a response to physiological overarousal; after dealing with too much physiological stress, someone's response to a conflict may simply be to "shut down" and block out the argument.

Stonewalling is toxic because it shuts down productive conversation. Rather than hashing out any differences in a mature way, someone who stonewalls just succeeds in raking over the underlying issues, not really resolving anything, and also making one's partner feel like he/she is not being taken seriously or heard.

Below, we can see Kris engaging in some very clear stonewalling. Kim has told him that she does not plan on changing her last name, and this clearly bothers him. However, rather than talking this out and coming to some sort of compromise or reasoned conclusion, Kris completely shuts her out. He turns to his phone while she's trying to speak to him, doesn't really listen to (or process) what she is saying, and disengages from the conversation. As mentioned above, this creates a twofold problem -- not only does it not solve the true, underlying problem, but it makes Kim feel devalued.

What To Do?

Now that Kim and Kris have conveniently demonstrated the four worst things that you can possibly do in a relationship, how are you supposed to combat them?

Gottman and Levenson suggest that each of the four horsemen actually has a healthy counterpart.

  • Rather than phrasing a complaints as a criticism of your partner's personality or traits, try to emphasize state attributions. Instead of complaining about your partner's personality, raise complaints or problems about the situation or the behavior.
  • Instead of "playing the victim" and getting defensive, accept responsibility for your role in a conflict. This doesn't mean shouldering all of the blame; rather, it means recognizing and acknowledging anything that you might have done, or any way in which you are not blameless.
  • Build a "culture of appreciation," so both partners see each other with respect and appreciation instead of contempt, resentment, and disgust. Emphasize the importance of respecting each other and seeing each others' interests, hobbies, and passions as worthwhile.
  • If you are someone who frequently ends up stonewalling, or disengaging from potential conflict, figure out if this is because the fight is too overstimulating. If so, you can engage in something called "physiological self-soothing," which basically just means taking deep breaths and trying to mindfully relax. By focusing on the breath and calming down, it can become easier to have a difficult conversation without becoming so emotionally overaroused that disengagement seems like the only way to cope.

It's also important to note that if you've noticed any of the Four Horsemen in how you or your romantic partner tend to behave, this should not be interpreted as a "death knell" for your relationship. Although these studies do not establish the direction of any presumed causality, researchers in the field believe that the truly important contribution of this work is not the ability to identify doomed relationships and "predict divorce" as a cute party trick. The important note is that these behaviors themselves are presumed to be the cause of marital dissatisfaction, not the other way around. Just as we learned that it's important to attribute your partner's mistakes to situational (rather than dispositional) factors to avoid engaging in too much criticism, we can think about applying this state vs. trait idea to our attributions about the overall relationship as well. If you tend to be defensive and your partner tends to stonewall, it is much more productive (and optimistic!) to focus instead on addressing the individual behaviors themselves that are causing problems, rather than assuming that this indicates that your relationship is destined to fail. Based on what Gottman and his colleagues have argued, being aware of these four behaviors and trying to actively combat them is not a futile task -- it should actually greatly increase your odds of staying happily committed.

It looks like things are turning around for Kim K. She's now in a relationship with rapper Kanye West, and their first child together is due over the summer. The two are not officially engaged, but have hinted that they plan to wed in the future.

Regardless of whether or not Kim ties the knot for the third time, taking a good look at her past relationship behavior might help her figure out what went so terribly wrong in the past. Hopefully, she will bring less criticism and defensiveness into her relationship with Kanye, and maybe Kanye won't be as likely to stonewall during arguments or treat Kim with contempt.

And if nothing else, at least Kim's 72-day marriage to Kris Humphries did provide the world with one valuable thing:

A play-by-play demonstration of exactly what not to do in your own romantic relationships.


Gottman, J., & Levenson, R. (2002). A Two-Factor Model for Predicting When a Couple Will Divorce: Exploratory Analyses Using 14-Year Longitudinal Data. Family Process, 41, 83-96.

Carrere, S., & Gottman, J.M. (1999). Predicting Divorce among Newlyweds from the First Three Minutes of a Marital Conflict Discussion. Family Process, 38, 293-301.

Gottman, J.M., Coan, J., Carrere, S., & Swanson, C., (1998). Predicting Marital Happiness and Stability from Newlywed Interactions. Journal of Marriage and the Family, 60, 5-22.


For more information on Gottman's research:

Psych Your Mind on the Four Horsemen

Further Citations for Gottman's work

The Gottman Institute

Image Credits:

Kim Kardashian: Photograph by David Shankbone via Wikimedia Commons. Available under Creative Commons Attribution 3.0 Unported License.

Kris Humphries: Photograph by Keith Allison via Flickr. Available under Creative Commons Attribution-Share Alike 2.0 Generic License.

Kim & Kanye: Photograph by Noel Vasquez via Flickr. Available under Creative Commons Attribution 2.0 Generic License.


1. This figure of "93% accuracy" in predicting divorce can be a little bit confusing, so I want to clarify what it means. In the first study that Gottman and his colleagues ran, they conducted analyses after collecting data on many couples over the course of about a decade, and they identified the most important factors that distinguished the couples who happily stayed together, the couples who unhappily stayed together, and the couples who got divorced. In six future studies, the researchers then took those factors (identified from the first study) and, using entirely new sets of couples, tested different "models" of relationship satisfaction and divorce. What this means is that they designed a bunch of different ways that the important factors might combine together in order to influence divorce rates, and then used that information to create what is basically like an "algorithm" (or calculator) for analyzing a large number of couples. The models can use these factors and the information entered abotu the couples to perform a "prediction analysis," where the models automatically sort the couples into categories based on its predictions about their future marital status. In these studies, the models that sorted these couples into "Divorced" or "Still Married" groups were able to correctly identify the "divorced" couples 85-95% of the time. This is a rough description of the process, of course, but should generally convey what actually goes on in this research. Gottman and his colleagues never actually take individual couples and say "Will Divorce" or "Will Not Divorce." However, as Gottman and his colleagues argue (and rightfully so), you can predict divorce in the sense that you can look at an individual couple and assess the degree to which they resemble the couples that were deemed "Will Probably Divorce" by the model, knowing that the model exhibited ~90% accuracy for the couples that were assessed.

As an example, in the 1992 paper, there were 47 couples, 7 of whom had gotten divorced. The model's prediction analysis predicted that 37 of the couples would still be married (all of whom were), and that 10 of the couples would be divorced (7 of whom were). Therefore, the only mistake that the calculator made was classifying 3 still-married couples as "probably divorced." This means that the calculator was correct about 44 of the 47 couples, making it 93.6% accurate.

For more information, you can read the FAQs on Dr. Gottman's site.