In my research, we filmed an interaction between a couple and had each partner turn a rating dial as they watched their tape afterward. On this graph (at left), you can see how one couple rated their interaction. The blue dots represent the wife’s ratings over 15 minutes of conversation; the red dots represent the husband’s ratings. When you add them together, these ratings are a constant, which means that in this interaction, her gain is his loss and his gain is her loss. This is what’s called in game theory a “zero-sum game.” You’ve probably all heard of the concept. It’s the idea that in an interaction, there’s a winner and a loser. And by looking at ratings like this, I came to define a “betrayal metric”: It’s the extent to which an interaction is a zero-sum game, where your partner’s gain is your loss. On the other hand, by trust we really mean, mathematically, that our partner’s behavior is acting to increase our rating dial. Even though we’re disagreeing, my wife is thinking about my welfare, my best interests. When we scientifically tested these so-called trust and betrayal metrics, we found that a high trust metric is correlated with very positive outcomes, such as greater stability in the relationship. In a 20-year longitudinal study of couples in the San Francisco Bay Area that I recently completed with UC Berkeley psychologist Bob Levenson, we found that about 11 percent of couples had a zero-sum game pattern, like in that graph. Every six years, we would re-contact all of the couples in the study, and they would come back to Bob’s lab at Berkeley. Yet we noticed that many of the zero-sum couples weren’t coming back. I thought maybe they dropped out because they found the whole thing so unpleasant. Well, it turns out that they didn’t drop out. They died.

Interestingly, the investors’ expectations about the back-transfer from the trustee did not differ between the oxytocin and placebo recipients. Oxytocin increased the participants’ willingness to trust others, but it did not make them more optimistic about another person’s trustworthiness.

Conventional economic theory maintains that people will always behave in a purely self-interested manner. According to this worldview, it makes no sense to trust, whether in a trust game or in real life, as any trust will be exploited. The trustee will always keep her entire windfall for herself, so the investor would be better off not transferring any money in the first place. And yet when researchers like Joyce Berg and others have had people play the trust game with real monetary stakes, they have repeatedly found that the average investor will transfer half of her initial endowment and receive similar amounts in return. Through the trust game, researchers have also discovered a number of factors that seem to drive levels of trust. Familiarity breeds trust—players tend to trust each other more with each new game. So does introducing punishments for untrustworthy behavior, or even just reminding players of their obligations to each other.