Abstract

Collective motion is a familiar sight in nature; groups of distinct, self-propelled individuals appear to move as a coherent whole, exhibiting a rich behavioural repertoire that ranges from directed movement to milling to disordered swarming. Preeminent models of collective motion describe individuals in the group as self-propelled particles, subject to a combination of self-generated motion and social forces that depend on the state of neighbouring particles. In this work I introduce a new approach to modelling collective movement in animal groups based on active inference, a framework originating in theoretical biology that casts cognition and behaviour as consequences of a single imperative: to minimize surprise. Many empirically-observed collective phenomena such as cohesion, milling and directed motion, naturally emerge when considering individual behavior as the consequence of active Bayesian inference—this emerges without ever explicitly building behavioral rules or goals into individual agents. We show that active inference can recover and generalize the classic notion of social forces in agent-based models of collective motion, and numerically explore the parameter space of the belief-based model. In doing so we reveal non-trivial relationships between the beliefs of individuals beliefs and group properties like collective polarization and the probability of occupying different behavioural regimes. We also explore the relationship of individual beliefs about uncertainty, to the accuracy of collective decision-making. Finally, we show how agents that can actively alter their generative model over time, compared to those that can’t, form groups that are collectively more sensitive to external fluctuations and encode that information more robustly.