| Collective animal behaviour has attracted enormous interest among physicists in recent years. Self-organization of individuals into coordinated groups indeed strongly reminds ordering phenomena in condensed matter systems. How much can we push the analogy with physical systems ? Can we describe animal aggregations in the same way we would do with a system of particles or spins ? Despite the intense work in theoretical studies and numerical modelling, the scarce feedback with experimental data has restrained to give a clear answer to these questions. In this talk I will show that, in some cases, this can actually be done. Starting from field data of large flocks of starlings we indeed construct a maximum entropy model, which describes the statistics of individual flight directions in the group. This model is of the same kind as models used to describe ferromagnetic ordering and we can study and solve the statistical mechanics associated to it. In this way, we prove that interactions between individuals in a flock are local (a bird interacting with a finite number of neighbours) and topological (the number of interacting neighbours being independent of group density). The model quantitatively predicts the propagation of order throughout the flock, using no free parameters, even in very large aggregations. |
|