Description:

Balletic displays of coordinated motion in bird flocks, fish schools, and human crowds are believed to emerge from local interactions between individuals. The crux of the problem thus lies in deciphering the rules of engagement that govern these local interactions and the neighborhood of interaction over which they operate, which may differ between species. I describe a bottom-up, behavioral dynamics approach to the rich problem of self-organized collective motion in human crowds.

What are the rules and the neighborhood that give rise to human ‘flocking’? What is the mechanism of self-organization? We approached these questions using VR experiments in which a participant walks in a virtual crowd, motion capture data on real crowds, and agent-based simulations. First we developed an omniscient model, in which each individual’s behavior is based on the position and velocity of neighbors, averaged within a neighborhood. I suggest that robust averaging in the neighborhood provides a positive feedback that leads to self-organized collective motion. 

If this is so, why aren’t people flocking all the time? We find that people ignore highly discrepant neighbors, which yields a bifurcation in crowd behavior. Such collective decisions may thus be explained by local interactions. Surprisingly, however, we find little influence of attention in a crowd – unless people specifically intend to follow familiar friends. 

Most omniscient models of collective motion (including ours) attribute local interactions to metaphorical social forces. We recently developed a visual model based on the optical information that actually governs local interactions. We find that social forces reduce to the visual coupling, and the neighborhood of interaction follows from the laws of optics. In sum, collective motion can be understood as emerging from the self-organizing dynamics of local visual interactions.

William Warren is the Chancellor's Professor of Cognitive, Linguistic and Psychological Sciences at Brown University. Warren’s research focuses on the visual control of action – in particular, human locomotion and navigation. He seeks to explain how this behavior is adaptively regulated by multi-sensory information, within a dynamical systems framework. Details on his talk will be posted soon.