Self-reorganization and Information Transfer in Massive Schools of Fish
Haotian Hang, Chenchen Huang, Alex Barnett, Eva Kanso · 2025
Read Self-reorganization and Information Transfer in Massive Schools of Fish by Haotian Hang, Chenchen Huang, Alex Barnett, Eva Kanso (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-05822.txt.
Introduction
A fish school turns as one, and signals seem to cross it instantly, but Hang and co-authors asked whether it survives at huge group sizes. They simulated schools of up to fifty thousand fish, with the water flow each swimmer creates feeding back on its neighbors. As the school grew, flow effects tore at its unity: the school split into clusters that kept breaking apart, scattering, and merging, much as real schools do.
Inside a coherent cluster, behavior stayed correlated across the group, the scale-free pattern seen in nature. Just before a breakup, the correlation length shrank and the group's responsiveness faded, leaving it exposed to predators. During a collective turn, news of the new direction spread at a steady linear pace. Merging sped the news several-fold and splitting slowed it. The authors suggest water flow helped set the sizes of living groups.
Catalog record
| CREATOR | Haotian Hang, Chenchen Huang, Alex Barnett, Eva Kanso |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 67 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-05822
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Self-reorganization and Information Transfer in Massive Schools of Fish Haotian Hang, Chenchen Huang, Alex Barnett, Eva Kanso Abstract The remarkable cohesion and coordination observed in moving animal groups and their collective responsiveness to threats are thought to be mediated by scale-free correlations, where changes in the behavior of one animal influence others in the group, regardless of the distance between them. But are these features independent of group size? Here, we investigate group cohesiveness and collective responsiveness in computational models of massive schools of fish of up to 50,000 individuals. We show that as the number of swimmers increases, flow interactions destabilize the school, creating clusters that constantly fragment, disperse, and regroup, similar to their biological counterparts. We calculate the spatial correlation and speed of information propagation in these dynamic clusters. Spatial correlations in cohesive and polarized clusters are indeed scale free, much like in natural animal groups, but fragmentation events are preceded by a decrease in correlation length, thus diminishing the group’s collective responsiveness, leaving it more vulnerable to predation events. Importantly, in groups undergoing collective turns, the information about the change in direction propagates linearly in time among group members, thanks to the non-reciprocal nature of the visual interactions between individuals. Merging speeds up the transfer of information within each cluster by several fold, while fragmentation slows it down. Our findings suggest that flow interactions may have played an important role in group size regulation, behavioral adaptations, and dispersion in living animal groups. Introduction Nature is in a perpetual state of reorganization. However, while these cohesive patterns are regularly documented in systems of small or moderate size [ 1 , 2 , 3 , 4 , 5 , 6 , 7 ] , it is unclear how they scale with increasing group size [ 8 ] : do large groups remain cohesive or do they undergo dynamic reorganization? We address this question in massive simulations of schooling fish, where individual swimmers interact through self-generated flows and follow behavioral rules inferred directly from experimental data in shallow water environments [ 2 , 9 , 4 , 5 ] . By optimizing our computational algorithms, we simulate over long times the motion of groups of up to 50,000 fish. We show that “more is different" [ 10 , 11 ] . Where smaller groups maintain cohesive and polarized formations, larger groups spontaneously reorganize, constantly fragmenting, scattering and reassembling, similarly to empirical observations of large flocks of birds [ 12 , 13 ] and schools of fish [ 14 , 15 , 8 , 16 ] . We analyze how this self-reorganization influences the collective responsiveness and speed of information propagation between members of the group [ 1 , 12 , 17 ] . Collective responsiveness in self-organized animal groups manifests in long-ranged spatial correlations [ 18 , 12 , 19 ] . Correlation measures how the change in the behavior of one individual influences the behavior of others in the group. An animal group exhibits maximal responsiveness to a perturbation, say, caused by an attacking predator [ 20 , 8 ] , when correlations are scale-free, that is, when the range of spatial correlations scale with the linear group size [ 12 , 20 ] . Analysis of empirical data of large bird flocks confirms that spatial correlations scale linearly with group size L [ 12 ] . But do these results translate to groups of swimmers? In physical models of flow-coupled swimmers, microscopic [ 21 , 22 , 23 ] and inertial [ 24 , 25 , 6 , 26 ] , perturbations get amplified as they propagate via the fluid medium, hindering group cohesion. These models do not enable individual swimmers to sense flows and respond accordingly. Biological swimmers, on the other hand, are flow sensitive [ 27 , 28 , 29 ] and seem to correlate their tailbeat frequencies and phase [ 30 , 31 , 32 ] , but the extent of flow-mediated correlations is limited in space [ 6 ] . Recent evidence suggests that vision is both necessary and sufficient for polarized schooling [ 33 ] . Even in robotic agents, visual interactions with immediate neighbors are sufficient to induce scale-free correlations in polarized groups [ 34 ] . But are these scale-free correlations universal to groups of individuals with long-ranged visual and hydrodynamic interactions? If so, how does dynamic reorganization within the group, including splitting and merging event, affect the extent of spatial correlations? Importantly, how fast does information travel within a polarized group? Inspired by the analysis of information propagation in bird flocks [ 17 ] , we consider the behavior of our interacting swimmers during spontaneous collective turns. We find that the information about the change in direction propagates linearly in time across the group, at speeds much faster than the individual swimming speed. This is in sharp contrast to the diffusive information propagation in symmetric, consensus-based models [ 35 ] , and in the absence of behavioral inertia [ 17 ] . We show that symmetry is broken due to the non-reciprocal nature of the interactions between individual swimmers [ 36 , 37 ] , much like in the game of telephone, where a player secretly shares a phrase with the next person, who then passes it along to the next player and so on. In this game, the interaction range is one, and the correlation length — representing how far the phrase spreads before becoming distorted — goes well beyond one, but does not scale with group size [ 38 , 39 ] . Importantly, the message is transmitted from one person to the next person who did not already have the information. This non-reciprocity inherently breaks symmetry and ensures that the message travels ballistically in time in one direction, as opposed to the diffusive propagation that occurs when each person randomly chooses to transmit the information in either direction [ 35 , 17 ] , left or right, irrespective of where it came from [ 40 ] . Surprisingly, when extending this analysis to quantify the speed of information propagation during self-reorganization, we find that merging of separate clusters speeds up the transfer of information within each cluster by several folds, while splitting and fragmentation slows it down. Results Dynamic reorganization, fragmentation, dispersal and reassembly in large fish schools. We numerically simulated the motion of a school of 50,000 fish coupled via visual feedback rules and flow interactions in an unbounded planar domain (Fig. 1 , Suppl. Movie 1). Each swimmer followed behavioral rules, modulated by an asymmetrical visual field representing frontal-biased perception [ 2 ] . These rules were derived empirically from shallow-water experiments, where each swimmer turned towards its Voronoi neighbors, aligned its heading with the same neighbors, and experienced rotational white noise [ 2 , 3 ] . Additionally, each swimmer generated a dipolar flow field and responded to the combined flow generated by all other swimmers [ 4 , 5 ] . We normalized the swimming speed U and intensity of rotational attraction by a proper choice of characteristic time and length scales [ 4 ] . Accordingly, with U=1 , three dimensionless parameters (I_{n},I_{a},I_{f}) distinguished the behavior of individual swimmers representing, respectively, the