Abstract
Understanding collective behavior in animal groups requires quantitative frameworks that can capture how information is generated, shared, and transmitted among individuals. This talk introduces a suite of information-theoretic measures, beginning with Shannon entropy as the foundation for quantifying uncertainty. The discussion then develops through conditional entropy and mutual information, which characterize dependencies between agents. Extending these ideas to temporal dynamics, time-delayed mutual information measures coupling across time, while transfer entropy serves as a directional, time-asymmetric tool for identifying effective information flow. The main emphasis is on the theoretical underpinnings of these measures and their interconnections, with applications to key problems in collective animal behavior, such as detecting interaction structures, revealing leader–follower dynamics, and mapping the flow of information underlying coordinated movement.
About the speaker
I am investigating the collective movement of locusts using individual tracking of agents and incorporating information theory to infer different causal interactions that constitute the collective migration of locusts. In another approach, I also use the Eulerian framework, specifically Particle Image Velocimetry (PIV), to describe the macroscopic flow patterns in locust swarms. By combining causality analysis from information theory with PIV-based flow characterization, I aim to uncover the underlying principles governing active turbulence in collective migration.