Research

I am a mathematical ecologist by training and an interdisciplinary scientist at heart. Most of my work is connected by the common thread of information processing in biological systems. Below are a few areas I’m exploring, which will be updated as work progresses!

Quantifying Trust

By sharing information and making decisions collectively, groups of organisms may greatly increase their success in a complex environment. However, cooperation also makes individuals in a group vulnerable to direct or indirect exploitation. How can individuals form groups that minimize this vulnerability? In human terms, how do we decide if, and to what extent, we can rely on information communicated to us by others? Part of the equation must be how much we build trust in the information source. I build mathematical models of social learning to see how different models of trust and reputation evolve over time, as individuals form and dissolve groups at different scales.

Perceived Variance and Timescales

How do people make decisions about processes that are highly uncertain and take place over long timescales? Statistics offers answers on how one should handle these issues, but most people are not statisticians. I combine analysis of human behavioral data and mechanistic modeling to uncover how uncertainty affects the internal models of the world that people use to make decisions.

AI-assisted Model Discovery

Ecological systems are very complex, and studying quantities of interest has historically required carefully choosing simplifying assumptions in order to arrive at tractable math models. This practice is largely constrained by the intuition and experience of those constructing the model; critically important dynamics can easily be overlooked. For example, by adding just one small (and data-supported) nonlinearity to a classic fisheries model, (Gil et al.) identified counterintuitive results with major implications for the dynamics of the system. This suggests that the standard method for model construction poses a major constraint on our ability to capture important elements of a complex system.

Machine Learning and AI offer powerful tools for data analysis and pattern discovery, including for complex, nonlinear systems. These techniques are often “black boxes” which lack the interpretability and generality of classic mathematical models. However, these methods can be employed to augment the process of building symbolic math models, assisting in discovery of counterintuitive model features.

Higher-Order Evolution

How can a host evolve to protect itself from a repeated emerging threat, like respiratory viruses? One possibility is to create conditions in which the evolution of a pathogen may be constrained. I use nested within-host and population-level epidemic models to understand how challenges imposed by a host (different tissue types, immune pressure, and methods for pathogen propagation) can affect pathogen evolution.