CHAPTER 1
Omne ignotum pro magnifico
(âEverything unknown passes for something splendidâ)
âTacitus, 98 AD
Can we really build a science of love? Is love something that we can actually examine under a psychological microscope? The early great psychologists of the 19th century brought to psychological questions methods of systematic, unbiased observation and measurement. These methods were developed by animal psychologists and behaviorists who focused exclusively on observing behavior.
These early scientists in the emerging field of psychology claimed that the first task of science was description and good measurement. Their advice: Observe, describe, measure, and find patterns that replicate over studies. Then try to discover the principles that organize these patterns. So, with that history of psychology as a legacy, we all began bringing love into a laboratory.
However, my life-long colleague Robert Levenson and I were not limited to studying just behavior. We also include self-reports of experience, and human physiology. But our first task was simply to describe. Then our second task was to find replicable patterns. The third task was prediction over time. The fourth task was to understand that prediction, and build a theory. The fifth task was to use that understanding to help couples have successful love relationships. The accomplishment of these five tasks took many scientistsânot just our laboratoryâover four decades. Now we can share this good news with the world. What were the beginnings of this field of understanding love? It all began when two scientists recorded the conversations of one newlywed couple on their honeymoon.
RECORDING A COUPLEâS HONEYMOON
In the 1960s, two researchers, William Soskin and Vera John, inadvertently started a revolution in our ability to study love. They recorded one couple on their honeymoon. The couple had to wear huge expensive backpacks even when rowing on a lake to record their conversations. The backpack contained a radio transmitter that sent a signal of their conversation back to a control room so that all their conversation was recorded on audio tape.
Then Soskin and John transcribed every word the couple said to one another, and categorized every sentence, a process they called âcoding,â using a complex category system. Hereâs a clip of Roz and Jockâs conversation as they row on a lake resort:
Jock: | Come on! Yo ho, heave ho. You do the rowing. |
Roz: | Nuts to that idea. Youâre the big, strong man. Mmmmm. |
Jock: | Yeah, but I have a handicap. |
Roz: | Yeah, you have a handicap in your head. |
Jock: | (to attendant) Can we take out a boat? (They get a boat.) |
Roz: | Whoops! Donât get wet. You row for a while and then Iâll row. Okay? |
Jock: | All right. Itâs awkward rowing with the transmitter on. |
Roz: | Go on. Want me to take it while youâre rowing? |
Jock: | No, itâs okay. |
Roz: | Bet you donât know how to. |
Jock: | Oh, yes I do. I guess I just ⌠|
Roz: | Here, let me change. |
Jock: | Iâll just have to set this thing out here. |
Roz: | Let me take it. |
Jock: | Okay. Itâs really a clear lake, isnât it? |
Roz: | Itâs wonderful. Look, thereâs a big moth. I wish I had my book with me, then I could tell what kind it was. |
Jock: | (handing off the transmitter) Here, put it on. |
Roz: | Like this? I wouldnât want my speech distorted, since I usually have so much to say. |
Jock: | Arenât those cabins nice? |
Roz: | Yes, those are the ones we were supposed to be in. I keep telling you. |
Jock: | These there? Look how dark the water is down there. |
Roz: | You tip this boat over with me in it and Iâll be very upset. Uh, uh, huh, huh, huh, huh (chuckling). |
Jock: | I just felt the ⌠|
Roz: | (laughing) Jock, I just made a joke. Have you no sense of humor? |
Jock: | Look how ⌠|
Roz: | Why are we going way out in the middle? Iâll get sunburned. |
Jock: | What difference does it make whether youâre in the middle or not? |
Roz: | You get more reflection in the middle. |
Jock: | (scoffs) Oh! |
Roz: | Jock, I know! |
Jock: | How do you know? |
Roz: | I can see! You put on your sun specks before you get a headache, huh? |
Jock: | No. |
Roz: | No? Okay. Wanna take your shoes off? |
Jock: | No. |
Roz: | (taunting in a sing song way) ⌠Ah, Jockâs gonna be sore tomorrow because he insists on showing off. (Jock rocks the boat intentionally.) No! Now cut that out! Youâll ruin this $50,000 equipment. |
Jock: | Oh, look. Boy these are nice oars. |
Roz: | Youâre a good rower, honey. |
Soskin and John asked themselves, âHow does one scientifically âanalyzeâ these naturally-occurring conversations of these two people?â As a response, Soskin and John created a very complex set of categories (called a âcoding systemâ) for every one of Roz and Jockâs utterances. It took an enormous amount of time to categorize these utterances, and then to search for patterns in their data.
After all that âcodingâ work, they were mightily disappointed in the results of their analysis. They felt that these analyses totally failed to capture the complexity of this coupleâs relationship, that their coding just didnât capture the richness of the actual data. They concluded that the problem was that their categories for coding all this conversation by Roz and Jock had ignored both emotion and power. They felt that they really couldnât capture the playful teasing and the challenging contempt that Roz and Jock displayed in this clip, the attempts at control by Roz, and Jockâs resistance to being controlled, nor their repair in the last two lines at the end of this clip. Soskin and John wisely concluded that future coding of naturalistic conversations that a couple has should include emotion and power. They were right about that. That, in fact, was where we started, trying to describe the emotions.
OBSERVING LOVE: NEW METHODS MATERIALIZED
Although sociologists had been studying marriages since the 1930s, the use of observational methods was what we psychologists initially brought to this field. These sociologists had succeeded in accomplishing the huge task of defining, and reliably and validly measuring, relationship happiness in married couples. That difficult task took from 1938 to the mid-1950s. The accomplishment was an enormous advantage for us psychologists, once we started studying relationships in the mid-1960s.
By the early 1970s, psychologists who did therapy with families entered the fray, searching for what to observe in couples. The new technology of portable videotape made this process much richer. The intuitions of clinical psychologists led to many of the new breakthroughs. The therapists who began doing therapy with families taught us to focus on communication, on messages received and sent in just one interaction, and on the couple as an interacting system.
The decade of the 1960s had witnessed an outpouring of new writing by social workers, psychologists, and psychiatrists who were discovering new patterns in doing psychotherapy with couples and families. Instead of relying on narrative accounts of relationship events, they were observing these events in action.
Those insights were aided by breakthroughs in how to study nonverbal behavior, and emotion, particularly insights about the face that came from Paul Ekman and Wallace Friesen at the University of California at San Francisco. Inspired by Sylvan Tomkins, they and Carroll Izard taught us how to study emotions in the human face, in both adults and children.
Ekman and Friesen picked up on Darwinâs 1872 study of emotions, and created a new coding system that precisely described the motions of over 46 facial âactions.â They showed us how to study facial movement anatomically, and how to interpret facial movements to understand the universality of human emotions everywhere on our planet. They reversed the initial misleading declaration in the 1950s by the famous cognitive scientist Jerome Bruner that the face was a researcherâs nightmare, and not worth studying. Instead of a nightmare, Ekman and Friesen showed that it was a goldmine.
Being able to code and interpret facial expressions was just one gateway toward understanding emotion. Knowing what else to code in coupleâs interaction was a initially a huge challenge for the scientific study of couples, and it took scientists over 20 years to empirically learn what to observe in couplesâ interactions. Once we knew how to code emotions in the face, in the voice, in the body, and in language, we were off and running. A new era began once sophisticated observational methods began to be devised by psychologists like Gerry Patterson and Robert Weiss. We couples researchers also learned a lot from the scientists, such as Harry Harlow and Jim Sackett at the University of Wisconsin, who studied non-human primates.
Once the initial huge problems of measurement were solved, new methods also had to be developed for analyzing this rich stream of observational data from two interacting people. This mathematics of the new field was christened âcybernetics.â It had been developed by M.I.T. mathematician Norbert Wiener during World War II. Wiener was part of a project to develop the new math as part of the task of designing an anti-aircraft gun that could anticipate and follow its target (instead of throwing up an array of flak, and hoping an airplane bumped into it). The new math all had to do with observing events unfold over time, a field christened âtime-series analysis.â A âtime-seriesâ is a graphical plot of a variable over time. Just as the Dow Jones industrial average tracked the stock market, we could track a conversation between two people using two Dow Jones graphs of a conversation. Every morning daily newspaper had examples of these stock market time-series graphs.
With time-series analysis, we could search for rhythmic patterns during conflict. French mathematician Jean Baptiste Fourier had proved an amazing theorem in 1822 that helped us see the wave nature of patterns that repeated over time. Using the new methods of time-series analysis, we could now quantitatively assess how interconnected two people were. We could even assess the connection between two peopleâs hearts. Statistician James Ringland and I worked out a statistical test for making causal inferences in two time-series, one for each partner. Using time-series methods that Gene Glass and Victor Wilson and I developed in 1973, we could also assess whether specific events preceded significant changes in a time series. Therefore, by the early 1980s a whole new technology for analyzing time-series patterns could be brought to bear on this task of bringing love into a laboratory and watching lovers interact.
But, what if the sequence of data was just a series of categories, like âRoz-angryâ followed by âJock-angry,â rather than continuous numbers as in the Dow Jones average? How would these data get analyzed?
The answer lay in Claude Shannonâs information theory, developed in a small monograph published in 1949. Shannon was actually a student of Norbert Wiener. The vague concept of âinformationâ was now defined precisely in terms of the statistical reduction of uncertainty. It took 25 years for Shannonâs information theory to be applied to the study of couplesâ interaction. A brilliant clinician and scientist named Harold Raush applied the new mathematical âinformation theoryâ techniques Shannon had devised during World War II. Raush conducted a groundbreaking longitudinal study at the National Institute of Mental Health of couples having their first baby.
Finally, because of Raushâs pioneering study, our analytic techniques could match the subtlety that Soskin and John could not capture in Jock and Rozâs honeymoon talk. Now, instead of merely tallying how often some observational category each partner occurred in the interaction, we could (for example) describe how Roz tended to react when Jock challenged her knowledge. We could describe these probabilistic sequences in transitional probabilities. That means that we are not saying that that Roz would respond to Jock that way every time, but just more likely than chance alone would predict. We talk about reducing uncertainty in what Roz might do next by knowing what Jock had just done. If our prediction of Rozâs doing behavior Y, once Jock had just done behavior X, was significantly better than her overall probability of doing behavior Y, then thatâs all we needed to determine to get the probabilistic sequence Jock X â Roz Y. This is what we need to do to detect statistical patterns of sequences in our data.
We could also build longer, and much more interesting, sequences. In information theory we now had our basic tools for describing what patterns of behavior were in Roz and Jockâs and other couplesâ data. Statistical tests could then be devised to ferret out sequences in Jock and Rozâs codes. This idea of probabilistic sequential pattern was an enormous conceptual advance in understanding interactions between two people. And our field owed it all to that World War II project developing a more sophisticated gun. Now, with information theory and time-series analysis, we could detect complex sequences that actually captured the complexity of a coupleâs interaction. We were off and running, with observational techniques and analytic tools to match the richness of what we were observing in our labs. If we found patterns that characterized unhappy marriages (a big âifâ), and if we then found different sequences that characterized happy or stable marriages (another big âifâ), then we could see if these results replicated, and if they did (another big âifâ), then we could try to explain these patterns. We had the tools now. We just needed the data.
Yet, big problems emerged. The amount of data we generated in one study was simply overwhelming. For example, if we had only 40 codes for each partnerâs behavior, in just two-step sequences we had 40 Ă 40 = 1,600 possible two-step patterns. In just three-step sequences we had 64,000 possible patterns! How could we ferret out the dance in which Roz and Jock were engage...