Seminars

Revisiting behavioural-inference in fish: schooling as a noise-induced effect

I will talk about some recent work concerning collective motion in schooling fish, and how careful analysis of the data reveals features that are absent in canonical descriptions 'flocking', or 'swarming'. In particular, I will show that schooling--- highly ordered and coherent group motion--- bears all the hallmarks of a noise-induced effect, arising from finite-size stochasticity. Moreover, I will describe how such rare empirical evidence puts an extremely tight constraint on the possible interactions between individual fish, validating previous detailed studies of turning angle correlations. More generally, I will argue for a re-appraisal of the treatment of noise, not only in descriptions of collective motion, such as the celebrated Toner-Tu class of hydrodynamic models, but for behavioural inference in general. In the context of the latter, our work demonstrates that, rather than simply obfuscating otherwise deterministic dynamics, noise is actually fundamental to the characterization of emergent collective behaviour(s) and should therefore be treated as such.

Details to be filled in

no seminar (colloquium)