Final Report Summary - SWARM (Empirical analysis and theoretical modelling of self-organized collective behaviour in three-dimensions: from insect swarms and bird flocks to new schemes of distributed coordination.)
Animal groups, like bird flocks and insect swarms, are paradigmatic examples of self-organized collective behavior. There is no group leader; rather, coordination occurs spontaneously as a consequence of the interactions between individuals. In this respect, there is a strong analogy with systems of interacting units, particles or spins, in condensed matter physics, where emergent collective phenomena, such us ordering and phase transitions, have been deeply investigated. The main objective of the project SWARM is to provide new knowledge about self-organization and collective behaviour in 3D animal aggregations, starting from experimental data on animal groups and using concepts and methods from physics.
To achieve its tasks, SWARM realized advanced stereo-experiments on flocks of birds and swarms of insects: natural groups were photographed in the field with multiple high-resolution machine vision cameras. We then processed the stereo-images with a novel three-dimensional (3D) tracking algorithm developed on purpose, and obtained the full 3D individual trajectories in groups of several hundreds individuals. These data allowed us to quantitatively characterize the collective properties of flocks and swarms in a way, which was impossible until a few years ago. First of all, we could measure those observables describing the degree of order in the group and the correlations between individuals. Then, we could use these measurements to retrieve information on the effective microscopic interactions between individuals.
We found that in flocks and swarms interactions are local, i.e. each individual directly coordinates its motion with the closest neighbors (either closest in metric space, as it happens in swarms, or in order of neighborhood, as it happens in flocks), and estimated their strength and range. We also showed that, despite the short-range character of interactions, individuals in a group are able to influence each other over the long distances. This occurs both in flocks, polarized groups with a common average flight direction, and in swarms, where the group has no global directional order. Indeed, we found that in flocks and swarms the correlations between individual directions and speeds are scale-free, i.e. the extension of mutual correlations (and mutual influence) is proportional to the group’s size. This behavior is not generic; rather it corresponds to what in physics is called critical behavior. For animal groups, this requires some kind of self-organization mechanism. Birds in flock balance imitation to neighbors and speed control in order to achieve large-scale speed correlations. Swarms tune their density and size so as to maintain a scaling behavior of the correlation function and exhibit a near-maximal degree of correlation at all sizes. A large degree of correlation enhances the collective response of the group to external perturbations and stimuli (predators or female arrival for mating) suggesting that these properties might be related to a general principle of biological efficiency.
Using the 3D trajectories we could also investigate how information propagates through the group, revealing new surprising features of collective decision-making. We looked at collective turns in flocks and found that the decision to turn begins with a few birds and then travels through the flock extremely fast, at a fixed speed and with negligible attenuation. We formulated a new theory that correctly reproduces these observations. Essential to the theory is the inclusion of the bird’s behavioural inertia, i.e. their resistance to change direction. Surprisingly, the resulting mathematical equations are analogous to the ones of superfluid transport, and predict that information transfer must be faster the stronger the group’s order, a prediction accurately verified by the data. Our results suggest that swift decision-making may be the adaptive drive for the strong behavioural polarization observed in many living groups.
To achieve its tasks, SWARM realized advanced stereo-experiments on flocks of birds and swarms of insects: natural groups were photographed in the field with multiple high-resolution machine vision cameras. We then processed the stereo-images with a novel three-dimensional (3D) tracking algorithm developed on purpose, and obtained the full 3D individual trajectories in groups of several hundreds individuals. These data allowed us to quantitatively characterize the collective properties of flocks and swarms in a way, which was impossible until a few years ago. First of all, we could measure those observables describing the degree of order in the group and the correlations between individuals. Then, we could use these measurements to retrieve information on the effective microscopic interactions between individuals.
We found that in flocks and swarms interactions are local, i.e. each individual directly coordinates its motion with the closest neighbors (either closest in metric space, as it happens in swarms, or in order of neighborhood, as it happens in flocks), and estimated their strength and range. We also showed that, despite the short-range character of interactions, individuals in a group are able to influence each other over the long distances. This occurs both in flocks, polarized groups with a common average flight direction, and in swarms, where the group has no global directional order. Indeed, we found that in flocks and swarms the correlations between individual directions and speeds are scale-free, i.e. the extension of mutual correlations (and mutual influence) is proportional to the group’s size. This behavior is not generic; rather it corresponds to what in physics is called critical behavior. For animal groups, this requires some kind of self-organization mechanism. Birds in flock balance imitation to neighbors and speed control in order to achieve large-scale speed correlations. Swarms tune their density and size so as to maintain a scaling behavior of the correlation function and exhibit a near-maximal degree of correlation at all sizes. A large degree of correlation enhances the collective response of the group to external perturbations and stimuli (predators or female arrival for mating) suggesting that these properties might be related to a general principle of biological efficiency.
Using the 3D trajectories we could also investigate how information propagates through the group, revealing new surprising features of collective decision-making. We looked at collective turns in flocks and found that the decision to turn begins with a few birds and then travels through the flock extremely fast, at a fixed speed and with negligible attenuation. We formulated a new theory that correctly reproduces these observations. Essential to the theory is the inclusion of the bird’s behavioural inertia, i.e. their resistance to change direction. Surprisingly, the resulting mathematical equations are analogous to the ones of superfluid transport, and predict that information transfer must be faster the stronger the group’s order, a prediction accurately verified by the data. Our results suggest that swift decision-making may be the adaptive drive for the strong behavioural polarization observed in many living groups.