Skip to contentSkip to filters

We use cookies for analytics.

Unravelling the Structure of Behavioral Variation from Data

Antonio Costa

Paris Brain Institute

Abstract

Animal behaviour varies widely, both within an individual over time and between individuals. Often overlooked, this variation reflects hidden control variables and mechanisms shaped by evolution: variation in behavioural traits can help populations withstand environmental change, and atypical motor patterns in neurological disorders may offer clues for personalised therapies. Comparing such complex behaviours is difficult, because when dynamics are nonlinear and unfold over multiple timescales, standard metrics based on summary statistics often miss meaningful differences.

The talk introduces a framework that encodes multiscale dynamics to compare behaviour from data. By modelling nonlinear dynamics probabilistically, using transfer operators inferred from time-series data, it defines a distance metric that captures behavioural differences across timescales. Tailored to finite, noisy datasets, the approach identifies principal axes of variation and enables rigorous clustering of individual trajectories. Applications to bacterial chemotaxis and larval zebrafish locomotion show that the inferred axes of behavioural variation reflect underlying physiological variables and developmental histories.

Topics