Smile and the world smiles back. Can looking at faces lower aggression?

I collected the data for an experiment trying to reduce anger by manipulating biases in emotional processing. I was sceptical, but the results surprised me
Before I started my PhD, I worked as a “research assistant”. That’s a fancy title for an academic dogsbody; well, it can be. I was lucky and had some great bosses in the five years I had that job, but sometimes it can involve menial tasks like data entry, or running experiments you think are a complete waste of time.
One such experiment, that I was asked to run by my boss while we waited for ethics approval on another study, was published last week in the journal Psychological Science.
Shows what I know! It showed that a simple task involving looking at faces and judging their emotion could reduce anger and aggression in a population of aggressive young people, and a group of controls at Bristol University.
When I was asked to run the experiment, I was sceptical because it was a novel way to try and reduce aggression. There is a suggestion that antidepressants work by changing the way a person processes emotions, reducing a negative emotional bias that is a symptom of depression. But would the same be true for aggression?
The premise is: perhaps aggression is at least in part caused by biases in emotion perception. So if you see a neutrally emotional face, due to biases in your emotional processing you view that face as angry or hostile. You respond with hostility towards the owner of that face, and, like a vicious circle, they see your hostility and respond accordingly. The experiment was simple. We showed people a series of emotionally ambiguous faces: happy and angry expressions blended together at different ratios to create a spectrum of faces. Those at either end were clearly happy or angry, but others in the middle were ambiguous.
Each person will have a slightly different boundary point, where on one side they see happy, and the other side angry. If you work out where a person’s boundary is, you can then give them biased feedback to try and “shift” it so they see “happy” further along the spectrum than previously.
The feedback, “yes, that was angry” or “no, that was happy” (for example), is given after every face is responded to, and is tailored to each individual’s boundary, attempting to shift their perception of happy to include a larger range of ambiguous faces.
But would this really make people less aggressive? I was tasked with running the pilot experiment. Half the participants got biased feedback, and half got unbiased feedback consistent with where their boundary actually was, as a control condition. They were asked about their mood with a couple of standard mood questionnaires before and after the experiment.
As I watched people doing the task, I confidently predicted to myself that those who were getting the biased feedback would be angrier afterwards. After all, they were being told they were wrong a lot of the time! When the experiment was over and we looked at the data, I was astounded to see that the opposite was true. Those who had received the biased feedback had not only shifted their boundary by the end of the experiment, but they rated themselves as less angry than the control group afterwards (there was no difference in anger before the experiment). It didn’t affect how happy or sad they were.
This is one experiment, and the plan is to replicate it in different groups of people, to see how robust the effect it. Also, while two weeks post-training is a long time in terms of an experiment, it’s not that long in terms of the behaviour of a youth with a high probability of offending. It would be great to run the experiment over a longer timescale, and with more realistic outcomes, such as re-offending rates.
But this unexpected finding astounded me, and may have really useful benefits. And it just goes to show that ideas are worth testing, even if your data monkey (in this case, me) has doubts! Courtesy The Guardian