Fifteen Questions: Benjamin L. de Bivort on Individuality, Forecasting, and the Politics of Science

Benjamin L. de Bivort is the co-chair of the Department of Organismic and Evolutionary Biology. This interview has been edited for length and clarity.

FM: Your research focuses on evolution through the lens of variations in animal behavior. What led you to make the connection between behavior and evolution?

BLD: We do consider evolution, but I would say we mostly study variation and behavior, and particularly variation we see among animals that are genetically identical and have been raised in the lab in standard environments. These are creatures where the nature is matched and their nurture is matched, and yet they’re still very different in their behaviors.

It’s a question of what is the source of that variation? And we think it arises in essentially random outcomes of the developmental process by which the body constructs itself and how the brain wires. You get these subtle differences, which, when the animal implements behavior, get magnified into individual differences that often exist for the lifetime of these animals, and are like personality. It does connect to evolution, but it’s not variation of the sort that evolution by natural selection normally acts on, because it can’t be passed on to progeny.

So in some ways, it’s a bit of an evolutionary dead end. But there’s also a metastrategy that evolution employs, which is, to what extent does variation as a trait confer adaptive advantages to animals? So we also look at that.

FM: Your work deals with this idea that behavioral variation and individuality can be linked to molecular variations. So if our individuality is linked to the chemical or molecular variations in our bodies, do you think we — or the organisms you work with — have a sense of agency, or can our personalities be described just by the molecular mechanisms behind the behavior?

BLD: I think so much work around biological mechanisms and neurobiology seeks to explain behavior. It can sort of feel like it takes away some of the magic of being an animal and exhibiting behaviors. To the extent we gain mechanistic understanding, we have taken away mystery. But, does that remove agency? I’m not sure. I think in my own experience, there’s something sort of humanistic about realizing that individual differences don’t have a compelling narrative behind them. These are quirks of circumstance, and yet they matter for how we interact with each other and how we move through the world and our political decisions and all that. But they don’t go back to deterministic stories, where maybe it opens up some freedom of action in the sense that we are not bound to our genetics nor the deterministic influence of the environment. But it’s a different kind of freedom. It’s a freedom just to be random, and not be bound by those other constraints.

FM: You use fruit flies as the model organism for your work. And I have done a little bit of fly work myself, in high school.

BLD: I imagine it was amazing.

FM: Well, it was more cancer, genetics, that kind of thing. The question that I would always get is: why are flies relevant to us as humans? And so my answer was always, because the genetics are really similar. But behaviorally, I feel like you would get that question a lot as well. Why are they a good model for behavior in particular?

BLD: Flies and humans solve a lot of the same problems in their ecology. They have to find food. They have to find shelter. They have to find mates. To choose whether to be curious and explore their environment or be conservative and shelter. These trade offs are the same for all animals. So the behavioral strategies that they evolve, such as maybe bet hedging, could potentially be strategies that we would evolve.

FM: To pivot a little more to your teaching work, I know you teach LS50, which is the integrated small group introduction to the life sciences. What drew you to that class?

BLD: Well, is 35 students a small group?

FM: Well, smaller than LS1a, which I took.

BLD: I was there for the very beginning of LS50.

I think the motivation for all the faculty and instructors starting the class was, what would we have wanted to learn as first year college students to put us on a trajectory to doing research in the natural sciences? We were drawing on our experience and our discovery of these themes and connections that run through the life sciences. Often they’re quantitative, and relate biological organization across many different contexts, like the way you approach dynamical modeling of an ecosystem might be the same way you approach dynamical modeling of a genetic switch. There are these approaches, these methodologies, that connect these disciplines across huge orders of magnitude, just radically different scales.

FM: And as a principal investigator, you have mentored, advised, guided so many people through various points in their scientific careers. How does working with doctoral or postdoctoral students differ from your work with undergraduates?

BLD: The longer somebody’s been in a research career, the more generally independent they can be. I wouldn’t say that there are necessarily category differences. It’s mostly quantitative differences.

I think one thing Harvard does well, because it has a high ratio of labs to undergraduates, is to be able to give undergraduate researchers a high degree of independence and ownership of projects.

Unsolicited research advice, if there are folks thinking about it: just dive in. Try it out. You can’t really understand what it’s like to have an independent research project until you try it, right? Because there’s rewards and frustrations, and you need to experience those firsthand to know if it’s for you.

FM: Was science something you were always interested in? Is there a story about how you first sparked an interest in science?

BLD: It kind of was. My parents are more humanitarian types. My dad studied the Arab-Israeli conflict for a PhD, and my mom was an artist and substitute teacher. But somehow, they turned out to have a kid who was into science.

I think the hinge points for me were a really effective sixth grade science teacher who made it all very enjoyable, maybe also participating in little physics and engineering competitions in high school and building mouse trap racers that could go 150 feet or something like that. I was in a rock and fossil club as a kid, so I credit my parents for exposing me to the natural world. And when I went to college, I was pre-med for maybe three weeks before I said, “nope.” So now I do the basic side of biology.

FM: What do you do when you’re not in the lab?

BLD: Well, I still enjoy collecting rocks and fossils — that came back a bit during the pandemic. It was a nice outdoor, distanced activity. I do puzzle competitions. There’s a thing called the MIT mystery hunt, and I'm on a team for that.

FM: I had seen that you and your team had won the MIT mystery hunt back in 2019. And then you wrote the competition the next year.

BLD: Tradition is, if you win that competition, you have to make the next year’s.

FM: What was that like, to write such an intense contest?

BLD: As a bit of background, when we ran the Mystery Hunt, there were about 4,000 competitors. Our team, though the MIT puzzle club, had a budget of, I think it was $80,000 to run it.

Our team had about 70 people, and I had a minor leadership role, but really we had three people who were able to take a whole year off of work to be the triumvirate that led our production of this puzzle hunt. We did more than 200 puzzles, which was the largest at the time.

FM: On your lab website, you had listed forecasting as one of your interests — what is forecasting?

BLD: So this is the idea that people can pick up the skill of predicting geopolitical outcomes in the future, and was the subject of study by a professor named Phil Tetlock at the University of Pennsylvania, and Barbara Millers. And some of their provocative findings were that people who are professionally pundits actually do a bit worse at predicting future outcomes than non-experts. That was the first result, and then the later results include that you can generate more accurate forecasts by aggregating predictions across a lot of people — wisdom of the crowd effects. And then these studies were conducted as contests, and so I was a participant in some of these forecasting competitions, and did well enough to be put in a category called a Super Forecaster, where we were reliably in the sort of top percentiles of predicting some of these outcomes, and got to go head to head against the intelligence agency and various AI entities.

FM: And what’s the best thing you’ve predicted? Or, I don’t know if best is the right word for geopolitics.

BLD: A classic example was will Mario Monti resign by the end of the calendar year? And this was 10 years ago.

You have to assess, what is the sort of base rate of turnover of Italian prime ministers? And then are there particulars about Mario Monti and the stability of his government? This is the way you approach those questions. I think this was the year I was doing it with teams, so you could have some discussion with your teammates and generate a consensus prediction.

Phil Tetlock wrote a book on this, and was shopping around for anecdotes, and he asked me for one — didn’t make the cut into the book but it was on, ‘would China violate the air defense zone over Taiwan within some time frame?’ And my contributions to the team discussion on this were to see if we could approach it from a Bayesian perspective. Can we put forward a prior and then update it based on news events as they came along to get a posterior estimate for that probability?

FM: I’ve often heard it said that one reason science is in crisis is because it’s hard for complex findings to be made understandable to the public. Is that something you agree with, and what can be done to rectify that?

BLD: It’s true that communicating sophisticated scientific concepts to non-specialists can be challenging, but there’s a saying, at least on social media, which is: “We’re not going to sci-comm our way out of this crisis.” I agree with that. I think that the problems are more fundamental and have a different root cause than a lack of communication.

This desire that people have for science to be apolitical — I think we’re seeing just how untenable that is. The popular support for science and the support among political parties and our governmental institutions, it waxes and it wanes, and we’re in a period of reduced support, both at a popular level and the institutional level. And I think it’s fairly straightforward to connect this to larger problems around trust in institutions and trust in the governance of elites. Science is an expensive kind of elite activity, and if people don’t have trust in the work of the government, we’re going to see support for science fall.

I heard it stated that the cultural prestige of science tends to grow or decline in proportion to the alignment of science with the interests of capital. I think there’s some legs to that argument as well. When you have science speaking out about the dangers of AI to the environment, to our information ecosystem, to our cultural resilience to misinformation, that puts science in conflict with powerful political interests. I think that’s part of the story also.

FM: You’ve been vocal about what many are calling the Trump administration’s “war on science.” What impact has the administration’s budget cuts had on your work?

BLD: I had a grant cut. I had NIH support, which was terminated, along with hundreds of grants here at Harvard. I’m a department chair now, so I acutely see the effects of the loss of financial support on our ability to recruit graduate students, to appoint new postdoctoral fellows, the potential cuts to staff. These are all related, right?

There’s an ecosystem of funding that’s needed to keep scientific enterprises going, and when the money dries up, the scientific activities have to dry up at some point.

—Magazine writer Sophie Gao can be reached at sophie.gao@thecrimson.com. Follow her on X @sophiegao22.