The Science of Togetherness

Making Couples Therapy More Effective

Magazine Issue
September/October 2017
Two hands with an EKG diagram across them

Since the days of Freud, perhaps no movement has arrived on the therapy scene with the force and freshness of general systems theory. In the 1950s and 1960s, it announced an entirely new way of understanding how psyches are shaped, wounded, and healed, not merely within our skulls, but in the vast, humming spaces of our environment, in social interaction. The revolution picked up speed and became an international movement following the 1968 publication of a book titled General System Theory by German biologist Ludwig von Bertalanffy. 

The concepts of feedback, homeostasis, and a holistic view of a system captured the imagination of many psychotherapy pioneers, including Salvador Minuchin, Murray Bowen, Jay Haley, and Virginia Satir, and became firmly established in the training of every marriage and family therapist. Nevertheless, despite all the intellectual excitement, the hard truth is that, so far, the systems revolution hasn’t led to very effective ways of doing therapy.

Fortunately, a second revolution is quietly taking shape—a new wave of systems theory and therapy—that marries the wisdom of clinical intuition with the rigors of scientific inquiry. With more precision and accuracy, we can now begin to answer two key questions about relationships: what causes trouble between people and what helps them not merely survive together, but actually rekindle love and delight?

While most couples and family therapies over the past half-century have been short on scientific investigation and data, some research has now been conducted. What can we conclude from it? To explain the basic findings of these studies, we first need to get a bit technical and explain a unit called the standard deviation. Remember that bell-shaped curve you studied when you read about intelligence testing? You probably learned that the average IQ is 100, and that the standard deviation is 15. That means that a solid majority of people have IQs between 85 and 115—that is, within one standard deviation on either side of 100.

Now on to effect size, or how many standard deviations our intervention moves an untreated control group. In the 1968 science-fiction movie Charly, starring Cliff Robertson, Charly was a mentally disabled adult who underwent an operation that temporarily changed him into a very smart person. Charly’s IQ shot up from about 70 to 115, a change of 45 IQ points. That translates to an effect size of 45 divided by 15, which equals 3.0. That’s highly significant. An effect size of 1.0, which would’ve increased Charly’s intelligence just one standard deviation, would’ve boosted Charly’s IQ from 70 to 85. That’s some gain, and yes, it’s significant, but it’s not very meaningful—and it probably wouldn’t have led to a feature movie.

In general, of course, we want large effect sizes in our practice, and we want clinically meaningful effects. Here’s an example to make the concept of effect size clearer. In 2002, Donald Taylor and colleagues found probably one of the biggest effect sizes in medicine: the effects of exercising and quitting smoking. It turns out that doing those two things can increase a person’s lifespan by about 16 years, which is meaningful, without question! What’s the effect size here? About 2.0, because in lifespan, the standard deviation is about eight years.

Now let’s look at effect sizes for the systems-oriented couples and family therapies for which we have data. It’s not pretty. According to William Pinsof and Lyman Wynne’s exhaustive review of all the couple and family therapy studies, couple and family therapies have an average effect size of 0.5. What does this mean for real-life couples? The most effective therapies we have can take a couple that isn’t too unhappy to begin with and help make them a little bit happier. We can nudge them from mildly distressed to nondistressed. But in most cases, we can’t help them develop a great relationship, or even a very good one.

Furthermore, most of our actual clients are much lower in relationship satisfaction than the mildly dissatisfied couples who populate most research studies. In measuring marital satisfaction, sociologists developed reliable and valid measures that imitated the IQ tests, with a population average of 100 and a standard deviation of 15. Based on our clinical work, the typical client in couples therapy is about 3 or more standard deviations below that mean of 100, which is a marital satisfaction level of about 55.

Taking a couple from that point to a marital satisfaction level of 63 is an effect size of 0.5. That’s basically moving them from horribly unhappy to moderately miserable—and that’s what the 0.5 effect size means for couples therapy. If we start adding in the typical comorbidities of depression, anxiety disorders, domestic violence, extramarital affairs, trauma, addictions, and so on, the effect size is even smaller.

Let’s not mince words: the results of the empirical outcome studies in our couples literature are just plain embarrassing. Even though many of us claim that our research-based couples therapies are evidence-based, using the term is simply misleading, because it implies that the model results in meaningful positive change. Furthermore, so far, no one therapy truly stands out as more effective than any other, but the second wave of systems theory and therapy might right that.

Identifying the Set Point

In part, the problem of small effect sizes is bound up with the old general systems theory itself, which promulgates a set of concepts in couples and family therapy that have essentially eluded definition and remained imprecise metaphors. In other words, concepts like feedback, homeostasis, first-order change, second-order change, self-organization, and so on were useless in research because they couldn’t be precisely measured. They were also useless to the working therapist because they didn’t lead to specific and powerful therapeutic techniques.

Let’s start with homeostasis, a biological concept introduced in 1932 by physiologist Walter Cannon. The metaphor here is a thermostat in a room, which has a set-point temperature, and continually makes adjustments as the actual room temperature deviates from that set point. It’s a great metaphor, but in biology there’s always a specific, measurable set point that’s being regulated. An everyday example is the body’s regulation of blood pressure. In biology, that kind of precision about the set-point variable leads to the identification of mechanisms that explain how and why the regulation takes place, and what regulates health versus illness.

Let’s explore this a bit more. What’s a healthy blood pressure? It’s a set point of around 120/80. What’s an unhealthy blood pressure? It’s a set point that’s 140/110 or higher. That’s getting into a high-risk area for strokes, and nobody wants that. But if we want to provide effective treatment, we need to also ask, “What mechanisms control blood pressure?” There are many mechanisms, such as excessive contraction of the heart muscle, and each variable in the mechanism implies a specific and effective treatment, such as beta blockers for treating excessive cardiac contraction.

Herein lies a strategy for building a more effective couples and family therapy. First, find a valid and reliable homeostatic set-point variable, and describe the healthy and unhealthy set points. For example, we might identify a critical number of hostile messages a couple launches at one another in a conflict discussion; 2 in 15 minutes might be okay, but 15 might exceed a healthy set point. Then, we could try to identify the mechanisms that regulate that set point, showing us the kind of therapy that’s likely to be effective.

In our own research, the two of us started this way, by asking, “What variable gets regulated in a couple or family system? What’s a healthy relationship, and what’s an unhealthy one?” When Robert Levenson and John began doing longitudinal studies of couples in the 1970s at Indiana University, a simple answer emerged. They discovered that one homeostatic variable in a couple’s relationship was the ratio of positive-to-negative emotions expressed during a conflict interaction, as detected by trained observers. What’s a healthy couple’s set point? During conflict interactions, a positive-to-negative affect ratio of 5 to 1 or higher is healthy. That’s the average ratio in stable, happy couples. What’s an unhealthy couple set point? If the positive-to-negative emotions during conflict encounters is 1-to-1 or less, that’s unhealthy, and indicates a couple teetering on the edge of divorce.

Searching for the Mechanism

Okay, we’ve completed step one of our strategy by identifying a set point. Now we have to search for the mechanism that dysregulates this homeostatic set point. As part of the search, Robert Levenson and John discovered that not only was the conflict positive-to-negative emotion ratio 1 to 1 or less in unhappy couples, but in analyzing sequences of interactions, a universal pattern emerged.

John called this pattern The Negative Markov Absorbing State (forgive the academic jargon). In essence, in an unhappy couple relationship, neutral or positive affect (states in which the couple is calm, matter-of-fact, interested, affectionate, playful, humorous, or empathetic) gives way easily to negative affect (states in which the couple is critical, defensive, contemptuous, stonewalling, hostile, angry, hurt, sad, disappointed, belligerent, or domineering). Thus, for these couples, negative affect is easy to enter but hard to exit.

Perhaps a more descriptive name for this dysfunctional pattern is the Roach Hotel Model of Negativity. Paraphrasing the well-known advertisements for this trap, unhappily married couples check in to negative affect—but they don’t check out. Or this pattern could be called The Quicksand Effect. When things go that negative, it’s as if the couple has stumbled into a bog of quicksand. The more they wriggle, the deeper they sink into negativity. They can’t repair the negativity and exit that state.

Furthermore, when this negative state spills over into nonconflict interactions, the relationship will rapidly deteriorate. For example, with spillover, a couple is fighting a lot and they decide that a vacation in Hawaii might set things right. But they get to Hawaii and find themselves on a beautiful beach arguing about whose fault it was that nobody brought a beach towel. The negative affect quicksand bog has spilled over and ruined even their ideal vacation. At the point of spillover, any therapeutic intervention to try to pull the couple into a more constructive state is much less effective than it would be otherwise.

From the researcher’s viewpoint, the negative-affect ratio of a couple gives us a heightened ability to predict future relationship happiness and stability, or unhappiness and divorce. John and Robert Levenson aren’t alone in identifying this predictive ability; it’s probably the most highly replicated result in the literature on couples. In fact, this homeostatic set point even generalizes to families. The negative-affect ratio has been associated with the development of child psychopathology, and runaway negative affect also characterizes unhappy mother, father, and child interactions. Additionally, researchers have identified negative-affect regulation as a major indicator of healthy child development.

Okay, just as with healthy blood pressure, we now have our set point for healthy couple and family relationships. So what? How does that help us be better therapists? Well, we need to ask the next important question, “What might be a mechanism that controls this set point?”

In our observational study of repair in our Love Lab—an apartment laboratory equipped with computers, video cameras, physiological sensors, and other equipment—in which we studied interactions between 30 couples for three years, we found that every conflict discussion is characterized by many repair attempts. Some attempts fail, while others succeed. We discovered that couples who successfully repair can exit the negative-affect state early in the conversation, before it becomes too negative and hurtful. Effective repairs are emotional, vulnerable, and foster understanding and validation. We found that the two most powerful repair approaches were beginning the conversation gently and taking responsibility for even a part of the problem. Couples were then likelier to avoid the attack-defend mode and move instead into a collaborative mode.

In our research with newlyweds in our apartment laboratory, we discovered a subtle process as partners moved about the lab and we followed them with three cameras. Frequently, one person would request something from a partner, which we called a “bid.” For example, the first camera would record a wife going to the window and saying, “Oh, it’s so pretty out there. There’s a beautiful boat.” The second camera would focus on the husband. When he responded, even minimally, by saying, “Oh, yeah, it is,” we coded that as “turning toward” the bid. When the partner made no response, we coded it as “turning away.” We discovered that the 17 couples who divorced six years after their wedding had turned toward their partner’s bids an average of 33 percent of the time, while the 113 still-married couples had done so a whopping 86 percent of the time. When they turned toward their partner’s bids at a high rate, repair attempts during conflict were more successful.

In the observational study of the conflict interactions of 130 newlywed couples, we discovered that the reason why unhappy couples get stuck in this negative absorbing state is the failure of repair attempts. We’ve all worked with couples who enter our office hopelessly mired in this negative emotion bog. Research reveals that what lies at the heart of unhappy couple relationships can best be thought of not as some quality inherent in the partners, but as a failure to repair the inevitable conflicts and disjunctions that occur in any couple.

Restoring Trust

The first book to have a real impact on couples therapy was published in 1968. In The Mirages of Marriage, authors Don Jackson and William Lederer argued that a key mechanism for marital success was the “quid pro quo”—the reciprocal exchange of positive acts and believing that this reciprocity is the essential marital contract. The recommended therapy inspired by the book was called reciprocal contingency contracting. The basic idea was that couples therapy should be highly rational, with the goal of helping couples negotiate sensible reciprocal agreements, even if both partners were operating from a position of self-interest. It was assumed that this reciprocity was all that was necessary for positive emotions to blossom again.

However, in 1997, Murstein, Cerreto, and MacDonald studied both relationship satisfaction and this reciprocal quid-pro-quo kind of thinking in both friendships and marriages. They show-ed that the proposed quid-pro-quo mechanism was actually 180 degrees wrong. In fact, quid pro quos turned out to be characteristic of ailing marriages and friendships, not happy ones. People didn’t start thinking about reciprocity until the relationship was already going wrong. So the quid-pro-quo therapy was actually teaching couples to emulate the patterns of failing relationships, not healthy ones. It was militating against trust, because people don’t tend to become emotional accountants and worry about reciprocity until the relationship already seems unfair. Instead of a give-to-get orientation, people in happy relationships simply give without expecting a return on their investment, a give-to-give orientation.

In our research on trust, we discovered that a satisfying agreement in which positive affect prevailed—our healthy set point—could be created only if each partner was working from a position of mutual interest, not self-interest. Trust is essential to get to that 5-to-1 healthy set-point balance of positive and negative affect.

We also discovered that what builds trust is a process we called attunement, which is created when one partner listens compassionately and nondefensively to another partner’s negative emotion. The motto of high-trust couples seems to be, “Baby, when you’re upset, the world stops and I listen.” That includes listening when the partner is the target of the upset. In successful—happy and stable—relationships, a partner might say, “I’m angry because you’re on your cell phone at dinner, and I miss you.” The response would tend to be something like, “Okay, that makes sense. I’m listening. What do you need?” With a newlywed couple that eventually divorces, the response would tend to be defensive, like, “Well, you aren’t so perfect either. You bounced that check last week.”

In therapy, this means that the clinician helps a couple become nondefensively responsive, to attune to the partner’s negative emotions. How does the therapist do this? By teaching the couple that behind every negative affect there’s a longing, and within that longing there’s a positive need. For example, within the emotion of anger is usually a blocked goal, such as wanting loving attention. So we teach our couples how to turn toward a partner’s anger and discover the need within it. Within the emotion of fear, there’s information about what a partner needs to feel safe. We therefore teach our couples how to turn toward a partner’s fear. When someone is sad, there’s an unexpressed loss. So we teach our couples how to turn toward a partner’s sadness.

For example, in working with a couple who’d recently had their first baby, the wife said to the husband, “I’m really mad at you. I feel neglected. When you come home at night you don’t usually want to hear about my day. I’m afraid you think it’s totally boring.” Before therapy, the husband would’ve been likely to respond defensively. But after being coached to attune to her anger, he responded, “I actually think your day is a lot more interesting than my day, and I love hearing about the baby. I was really sad to have been on that business trip when she took her first steps. I just get obsessed with being successful at work, and when I bring that stress home, I forget to ask about your day. I’m sorry. Tell me more of what you need.” This is the kind of interaction that leads couples to establish a healthier set point in their relationship.

Creating a Safe Haven

Part of building trust—what couples researcher Susan Johnson calls a safe haven—is the ability to stay physiologically calm, and to help one’s partner stay calm, even in the face of conflict. In their very first Love Lab study, Robert Levenson and John discovered that couples whose relationships deteriorated over three years were those who became “diffusely physiologically activated” during conflict. That means that their hearts beat faster, their blood flowed faster, they sweat more from the eccrine glands in their palms, and they jiggled around in their seats more than couples whose relationships were happy, or became happier, over time.

Later, psychologist Janice Kiecolt-Glaser and her husband, immunologist Ronald Glaser, teamed up with endocrinologist William Malarkey to study physiological changes in 90 newlywed couples as they discussed a high-conflict topic. Having taken a small blood sample of each individual, the researchers were able to measure the amount of adrenaline and cortisol the couples secreted—and from this data, predict the fate of their relationships 10 years later. Compared to couples who remained married, divorced couples were 34 percent higher on adrenaline during their conflict discussion 10 years earlier.

Why does all of this matter? Because when a person’s heart rate is above 100 beats a minute, or their oxygen is below 95 percent, they can’t listen very well. They can’t empathize. They lose access to their sense of humor. They’re secreting two major stress hormones: adrenaline and cortisol. At home, this can mean escalating arguments between partners, with one agitated person shouting over the other, hearing nothing of their partner’s needs. In the consulting room, it can mean stalled progress. For instance, if the therapist asks one partner to summarize and validate what the other partner has just said, they can’t do it well if they’re physiologically aroused, or physiologically flooded.

But how is a therapist supposed to know if a client is flooded? Our experience shows that clinical intuition isn’t a reliable way of determining that. Instead, we’ve developed and tested direct ways of measuring physiological arousal with inexpensive and reliable pulse oximeters, an instrument that measures heart rate and the blood oxygen level. Pulse oximeters fit comfortably on a person’s index finger and can ring an alarm when the heart rate exceeds 100 beats a minute, or goes less than 95 percent oxygen concentration in the blood. We’ve concluded that’s the only way a therapist can know if a client is physiologically flooded.

Then, we use biofeedback to build self-soothing ability using Heart Math’s emwave device. The emwave is a simple biofeedback device that builds physiological resilience; our clients use an ear clip and learn to turn a light from red to green. With regular use, a client who gets physiologically flooded can learn to increase a neurological dimension of self-soothing called vagal tone. These very good pulse oximeters currently cost around US $15. The Heart Math’s emwave device currently sells for under $100.

Since you can’t tell what a person’s heart rate or blood oxygen concentration is just by looking at them, therapists need to use these gadgets to measure physiological arousal in clients. Otherwise, since flooded clients can’t empathize, we might falsely conclude that they’re narcissists. But if we know when they’re flooded during therapy sessions, we can teach them to soothe themselves and one another. In one session with a policeman and his wife, for example, as she was talking about wanting to sell their house and move, his face remained calm while his heart rate soared to 140 beats a minute. Instead of responding to his silence with either her own silence or anger, the wife said, “Wow, you’re really flooded. What’s going on with you right now?”

The husband was too flooded to answer, so the therapist guided him in paced breathing and mindful meditation until his heart rate went back to his baseline. He then talked about how stressed he felt by the idea of selling their house, and why he felt that way. In the end, the pulse oximeter opened up an important conversation that ordinarily (at home) would’ve escalated into an argument or remained unaddressed.

Cultivating Commitment

Why is commitment also vital to creating a safe haven? To put it simply, it creates a vital loyalty, and it may also prevent cheating. In Caryl Rulbult’s three decades of research on couple commitment, she found that the critical variable in predicting infidelity is the process of negatively comparing one’s partner to real or imagined alternative lovers. We call these negative comps, and they’re the key turning point in predicting infidelity. Cheating, she discovered, becomes likelier when people seethe with resentment for what’s missing in their partner. By contrast, commitment develops by building gratitude, both by cherishing what’s there in your partner and by comparing your partner favorably with real or imagined alternative relationships. That comp—positive or negative—is the key ingredient in building a safe marital haven or creating a breeding ground for betrayal.

In our lab, we also discovered that commitment is built by turning toward everyday bids for emotional connection. Turning toward bids is a second major way couples build commitment on an everyday basis. In fact, we found that most of the arguments couples have tend to be about feelings hurt by failed bids for connection. We also discovered that turning toward bids at a high level creates shared humor and hilarity during conflict, which reliably reduces heart rate. Humor is a great physiological soother.

The Magic Trio

So now we’ve identified a mechanism for regulating a couple’s affective set point, which we call the magic trio. This trio defines precisely how to build a safe haven in a relationship. Based on carefully observing couples and coding their responses, we’ve found that the key requirements for a safe haven are: (1) physiological calm, which is built by physiologically self-soothing and soothing one’s partner; (2) trust, which is built by attunement; and (3) commitment, which is built by cherishing (positive comps) and by turning toward bids for emotional connection.

Let’s review what the research has uncovered so far that can open up a path to effective therapy. The first element is a set point in a couple or a family during conflict that maintains at least a 5 to 1 positive-to-negative emotion balance. The relationship is in danger when the positive-to-negative emotion balance during conflict is 1-to-1, or lower. Second, we’ve identified a clear mechanism that regulates this set point: the magic trio for creating a safe haven.

So if therapists can help clients build calm, trust, and commitment with their partners, will their intimate relationship measurably improve? If our proposed new general systems theory and therapy are effective, they ought to be able to create effect sizes bigger than the typical 0.5. So far, we’ve completed five outcome studies, with two more underway. Our effect sizes vary between 1.0 and 4.0, so we’re on the right track.

We realize that five studies is only a beginning. Moreover, some of the studies focus on couples workshop participants, rather than longer-term therapy clients. So we need many more trials, with larger samples, to replicate our results. Nonetheless, the work done so far gives us a window on the impact of the magic trio.

When couples dedicate themselves to building the qualities of the trio, does it significantly improve their relationships? Much more research will be needed to provide firmer scientific evidence, but we believe we’ve identified the quantifiable variables that will lead to more truly empirically based treatment methods in the future. And we believe that our field should no longer be satisfied with therapies that produce weak effects. Looking ahead, we need to develop a therapy that can take an ailing relationship and help a couple create a happy and lasting relationship, not just one that’s slightly less miserable.



Pinsof, W. M., & Wynne, L. C. (1995a). The efficacy of marital and family therapy: An empirical overview, conclusions, and recommendations. Journal of Marital and Family Therapy, 21, 585–613.

Pinsof, W. M., & Wynne, L. C. (1995b). Family therapy effectiveness. Washington, DC: American Association for Marriage and Family Therapy Press.

Pinsof, W. M., & Wynne, L. C. (2000). Toward progress research: Closing the gap between family therapy practice and research. Journal of Marital and Family Therapy, 26, 1–8.

Taylor, Jr., D. H., Hasselblad, V., Henley, J. S., Thun, M. J., & Sloan, F. A. (2002). Benefits of smoking cessation for longevity. American Journal of Public Health, 92, 990–996.


Photo © Jonathan Knowles/Getty Images

John Gottman

John Gottman, Ph.D., was voted one of the Top 10 Most Influential Therapists of the past quarter century and was recently honored with the 2021 Lifetime Achievement Award by the Psychotherapy Networker. Professor Emeritus in Psychology at the University of Washington, Dr. Gottman is known for his work on marital stability and relationship analysis through scientific direct observations, self-report and physiology. He is the author or co-author of over 200 published academic articles and more than 45 books, including the bestselling The Seven Principles for Making Marriage Work; What Makes Love Last; The Relationship Cure; and Why Marriages Succeed or Fail. He is the co-founder of The Gottman Institute and Affective Software, Inc., which has created a tele-therapy technology that will live on cell phones, computers, and tablets to assist therapists in couples therapy and offer direct relationship building services directly to couples. Learn more at