With the ideas42 Seminar Series, we invite leading scholars to share their insights and what inspires their exploration into human behavior.
Our New York office was pleased to host Ashton Anderson, a post-doc researcher in the Computational Social Science group at Microsoft Research in New York. Ashton has a Ph.D. in Computer Science from Stanford University. He is broadly interested in research that bridges the gap between computer science and the social sciences, and in particular, in combining machine learning/big data with the behavioral sciences. After giving a talk to the ideas42 team, Ashton was kind enough to share some of his thoughts on behavioral science:
What drew you to the intersection of machine learning and behavioral science?
I did my Master’s in game theory and my PhD in computer science, so combining machine learning and behavioral science was very natural for me. I work in computational social science now, where it has been enormously popular to bring computational methods and styles of thinking to bear on social questions. But there hasn’t really been an analogous surge of interest in bringing computational thinking to questions in behavioral science, where we aren’t only studying social phenomena, but also studying individual decision-making and behaviors. I think this is really important, so I started doing it.
What’s one of the most surprising discoveries about human behavior?
One of the most surprising discoveries to me is the sheer extent to which human behavior is affected by the environments people find themselves in. Studies like the Stanford Prison Experiment and Milgram’s obedience experiment (despite being ethically dubious by modern standards) illuminated how far outside their usual selves ordinary people will go in extraordinary circumstances. It’s natural to assume that the wide spectrum of behavior is accounted for by variation between people—that there are “good” people and “evil” people, independent people and obedient people, etc. But results like these suggest that a wide spectrum of behavior exists within each individual person. Rather than there being a world of different kinds of one-dimensional people, there is the world in every person — and the environment does a lot to influence who people are.
Tell us about your work in studying human error.
We are at a really exciting time in the study of human decision-making. Now, with the combination of algorithmic advances and the availability of large-scale corpora of human decisions, we can start trying to algorithmically characterize and predict instances on which humans are likely to make mistakes. Together with Jon Kleinberg and Sendhil Mullainathan, I assembled a massive dataset of human decisions in the context of chess, where we have perfect knowledge of the context of each decision, what decision a human took, and whether it was right or wrong. Using this dataset, we found a lot of interesting insights about decision-making in chess. For example, if you want to predict whether a chess player will make a mistake or not, knowing their skill level or how much time they have to make their decision has almost no predictive value at all. Instead, knowing how difficult the decision they face is makes all the difference — which further illustrates the power of the environment on human behavior. Furthermore, sometimes skill even hurts: we found chess positions in which strong players made mistakes more often than weak players.
How do you use behavioral science in your daily life (or recommend that people use behavioral science in their daily lives)?
The biggest difference behavioral science has made in my life has come from knowing about and constantly trying to mitigate present bias (the tendency for people to overvalue the present over the future, which can leads to behaviors like eating chocolate now and planning to diet in the future). I think even just realizing the variety of behaviors that can be explained by present bias helps in trying to avoid making these kinds of mistakes: once you start looking for it, you’ll recognize it in yourself all the time. But beyond that, developing strategies to help me make decisions for my future self (and sticking to them) has been very helpful in my daily life.