Byte by byte: Using data science to study human behavior

BY: ALEX JOHNSON

A popular thought experiment pushes people to consider the ethics of redirecting a trolley to hit a single person that is at the wrong place at the wrong time in order to avoid hitting five workers that the trolley is barreling toward. This puzzle has been debated for decades, but now researchers can enlist the help of computers to better understand the ethical conundrum. By testing social scientific theories in the framework of a rigorous, quantitative model, researchers can find actionable insights into social behaviors and interactions.

“We humans face some very complicated problems because we have a weak understanding of complex systems,” says Wesley Wildman, a Professor of Philosophy, Theology, and Ethics and Founding Member of the Faculty of Computing and Data Sciences at Boston University.  “I’m interested in finding the situations where the sciences of complex systems generate new insights.” Wildman aims to bridge the gap between science and the humanities by using data science to better understand the implications of social theories as part of the Hariri Institute’s Simulation Modeling for Population Health Focused Research Program (FRP) he helps lead with Brenda Heaton, Associate Professor of Health Policy & Health Services Research.

A trained philosopher with a background in mathematics, Wildman aims to put both disciplines to good use.  He works to find insights into real-world applications using theoretical models. “I’m trying to activate the whole of my skill set across the humanities and sciences to pull together stakeholders that can work on and solve human problems,” he says. Wildman overcomes many of the problems involved with understanding behaviors by using computer models to quantify abstract ideas like compliance, deviance, mental state, and violence. Social simulations involving numerous artificial agents in an artificial society allow Wildman and his colleagues to express the implications of theories about topics as diverse as immigration, religious radicalization, and suicidal ideation, and then to compare these simulated results against real world data.

In a recent project, Wildman investigated how likely people are to comply with a newly created COVID-19 management plan, and how an individual’s compliance impacts infection rates. “COVID models routinely make projections, but no one goes to check back later and see what aspects worked and what didn’t,” he says. Wildman built a flexible model of an “artificial university” that can be adapted to match the specific campus layout, student and employee populations, commuting patterns, etc. of real universities in order to determine which suite of COVID-19 prevention policies would be most effective. He then corroborated the model’s accuracy using real data provided by Northeastern University and two other schools about both policies implemented and infection rates. The model is applicable to any campus-based organization and will hopefully help organizations adapt quickly to the next pandemic.

Another benefit of computer modeling and simulation is that it allows researchers to safely study negative and harmful social behaviors that would never be approved for a clinical trial. For example, Wildman and his team used computer models to study religious radicalization and violent extremism. By shedding light on how the processes of radicalization take place, computer models can help inform public policy aimed at reducing radicalization and encouraging integration of cultural and religious minorities. This offers researchers a safe way to study harmful behaviors, and even predict negative outcomes before they occur. 

Applying data science to social scientific theories enables new insights into human behavior. As promising as these new technologies are, however, each brings with it a host of novel ethical challenges and potential pitfalls. Helping train computer scientists to think critically and proactively address these ethical pitfalls is, for Wildman, the most exciting part of his FRP. “As someone who studies both the humanities and computer science, I can appreciate the perspectives of both sides,” says Wildman.