Don't like angry faces? We have oxytocin for that.

I'm sure anyone who reads this blog is by now well aware of the many roles of oxytocin: In men, women, in sex, and especially in emotionally related things like pair bonding, trust, and facial recognition related to emotional states. People who just call oxytocin the "love molecule" don't get the HALF of it. Oxytocin has so many roles and so much influence that it really seems like a bit much to try and pin it down to a one-word definition.

Sci was particularly interested in the role of oxytocin in facial recognition, because it's NOT facial recognition like you would normally think of, just recognizing a face as a face or a face as THE face of a specific person. No, oxytocin appears to be related more to the recognition in a face of a certain social stimuli, in this case, the social stimuli of angry or happy faces. Evans et al. "Oxytocin Decreases Aversion to Angry Faces in an Associative Learning Task" Neuropsychopharmacology, 2010.


What we know so far: people use social cues to weigh the value of OTHER things, like financial incentive. For example, give a person a happy face, and an angry or sad face to choose from. Each face (happy or angry, say) is linked to the prospect of either winning or losing money (one face will win more often over the other). When you pick a face, it will tell you whether you won or lost, and thus give you feedback for the next trial. The cool thing about this is that people prefer the happy faces, EVEN when the happy faces don't predict monetary reward, they pick the happy faces more often than they should.

The question here was how oxytocin (which conveniently comes as a nasal spray and thus is very convenient to use in humans), which has a great deal to do with social recognition, would affect this game.

So they took 18 guys (why guys?! I would have liked to see the effect in women at different stages of the menstrual cycle), and gave them oxytocin or nothing. The they had them do this task:


In this case, one of the faces was rewarded 60% of the time, while the other was rewarded 40% of the time. They compared the number of times they picked the happy face vs the numbers of times they picked the angry face, as compared to the amount of times the faces were rewarded.

The graphs above are really pretty confusing, but the ones I want you to focus on are C and G. Both of these show the percent feedback for happy face evidence (when happy faces get rewarded) on the bottom, and the choices that people made along the side. C is the probability of the choices, while G is what people actually did (from what I can tell). As the probability of being rewarded for a happy face gets lower (lower than 0.5, which is 50% of the time), the probability of being rewarded for a sad or angry face is thus higher. You can see that the probability estimates showed that the people on placebo (no oxytocin) would pick the unhappy faces less. Even when the unhappy faces were the ones rewarded most often, they would still choose the happy face more than they should. On the other end, when they happy face was rewarded the MOST option, the people on placebo OVERPICKED it, and in fact over preferred it to the amount that it was rewarded. The prediction was that, on oxytocin, people would prefer the happy face LESS, and pick it only the amount it was rewarded on the positive side, while showing less preference for it on the negative side (when happy faces were not rewarded).

As you can see from G, that's sort of what they saw, but only sort of. Rather than oxytocin decreasing their preference for happy faces when happy faces were rewarded to normal levels, people on ocytocin still over-preferred those happy faces, above what the reward should predict. However, at the other end, when negative faces were the most rewarded, the people on oxytocin followed the predictive model and picked the angry faces more often, suggesting that oxytocin prevents the aversion to angry or sad faces that people on placebo show. This was all with no effect on mood, and it did not matter whether the face was angry or sad, as long as the emotion displayed was negative.

So it appears that oxytocin decreases our AVERSION to negative faces, while not necessarily making more attracted to positive faces. This means that people with the oxytocin on board ended up making better financial choices in the task (being rewarded more often) while people without oxytocin on board lost out by continuously preferring the positive faces, though none of the conditions actually made people LOSE money, they only didn't win. The whole thing might be different if there was actual financial loss involved.

What does this whole thing mean? It means that there may be a role for oxytocin in how we respond to negative faces as well as positive ones. It may not be so much of an increase in trust so much as a decrease in suspicion. There could also be a role for oxytocin in treating social issues that go along with problems like schizophrenia, when people show an over-reaction to negative faces which could be helped with oxytocin. Of course, who knows what else oxytocin can do, that little molecule is EVERYWHERE.

Evans S, Shergill SS, & Averbeck BB (2010). Oxytocin decreases aversion to angry faces in an associative learning task. Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology, 35 (13), 2502-9 PMID: 20844475

3 responses so far

  • Jon says:

    Interesting study. You say

    "This means that people with the oxytocin on board ended up making better financial choices in the task (being rewarded more often) while people without oxytocin on board lost out by continuously preferring the positive faces"

    But it looks to me (from Figs a & e) that the Drug group actually had slightly lower accuracy (although non-significant if the error bars anything to go by). So they made better decisions (relative to placebo group) in one condition but worse decisions in the other.

    Am I missing something?

    Also, the optimal strategy would be to work out (consciously or otherwise) which was the most rewarded face and then stick to that one until the end of the trial. Do they mention anything about learning effects? Is it that the Drug group start off with a lower aversion to negative faces? Or is it that they're quicker to overcome the initial bias towards happy faces in the face of negative feedback? If that makes sense?!?

    • scicurious says:

      I see your point in the figures, but the error bars and the results say nonsignificant, so in effect they were not less accurate.

      Your second question is a very good one, I was expecting them to have trained them on the task previously to rule that out, but they DIDN'T. An interesting question, and I wonder how the learning curves looked. It's actually not that they have a bias toward happy faces so much as there's a bias AGAINST unhappy ones, but if they have better or worse learning curves, it might also address your first question of accuracy.

  • david ropeik says:

    Hi. My expertise is in the psychology of risk perception (as a journalist who has investigated what various sciences tell us about that). Trust is a huge factor that influences how afraid we are, or aren't, and oxytocin plays a major role in that. I've summarized what some of the literature on oxytocin and trust says - in association with angry faces - in my current book, "How Risky Is It, Really, Why Our Fears Don't Always Match the Facts". Which may be of interest to someone who writes a blog like yours!

Leave a Reply