Physics of Collective Animal Behavior

Animals often need to make non-trivial decisions, individually or collectively, with short- and long-term fitness consequences. From the perspective of most animals, who are ignorant about the timescales and forces of evolution, decisions are made as direct responses to physical inputs gathered by their sensory systems and in order to obtain an immediately tangible outcome. In our lab, we develop agent-based and physics-inspired models that assume the perspective of the animals and describe the observable dynamics of the system based on explicit rules of interaction, behavior, and decision-making, which can in principle be deduced from spatio-temporal measurements of animal movement and behavior. These models interface with modern tools that allow the acquisition of high-resolution quantitative data and are therefore becoming increasingly important in the study of animal behavior. 

Our recent work on animal contests (fights) over resources showcases this approach. In orb-weaving spiders (Trichonephila clavipes), we studied the spatio-temporal dynamics of male-male competition taking place within the confined two-dimensional arena of a female’s web. Inspired by the interactions between physical particles, we modeled these spider contestants as active particles which interact through attractive and repulsive effective forces. This approach allowed us to explain the increased competitive advantage of larger contestants as the density of contestants increase in this system. Building on this, we developed a general model to describe the dynamics of animal contests more broadly. By using effective interaction potentials to represent contest behaviors and rules of motion, we could simulate various contest scenarios. Our model accommodates different rival assessment strategies and the effects of fighting costs, providing a comprehensive tool for understanding contest dynamics and outcome. These studies illustrate how our physics-inspired models can provide new insights into the mechanisms driving competitive interactions in animals, enhancing our understanding of animal behavior in adversarial settings.

Our recent works on the collective behavior of ants are another great illustration of our modeling approach. In Camponotus sanctus ants, we studied collective decision-making in emigrating colonies during nest selection. Treating the ants as independent decision makers, we developed an experimentally motivated model that captures the dynamics of the colony. The model predicted that cohesive emigration, without fragmentation, is achieved only by intermediate sized colonies, and this prediction was subsequently verified in experiments. Under an imposed conflict between colony subgroups regarding the desired emigration target, we found that individuals concede their potential benefit to promote social consensus. 

In Paratrechina longicornis ants, we studied cooperative transport in a confined geometry (a ’piano movers puzzle’ solved by ants) using an Ising-like model of coupled ’ant-spins’. Based on the model, we developed physical simulations for the dynamics of these tiny puzzle solvers. These simulations allowed us to better understand and mechanistically explain the superior puzzle-solving performance of large ant groups compared to small ant groups. These studies have provided new insights into the collective strategies ants use to solve complex problems, highlighting the power of agent-based and physics-inspired modeling in studying collective animal behavior.

Components of our physics-inspired model for animal contests. Top – landscapes of effective ‘resource potentials’, which attract contestants and bring them into contest range. Bottom – landscapes of effective ‘contestant interaction potentials’, which encode the behavior of contestants during contests. Adapted from: 
A. Haluts, S. F. G. Reyes, D. Gorbonos, R. I. Etheredge, A. Jordan, and N. S. Gov, Spatiotemporal Dynamics of Animal Contests Arise from Effective Forces between Contestants, Proc. Natl. Acad. Sci. U.S.A. 118, (2021)
A. Haluts, A. Jordan, and N. S. Gov, Modelling Animal Contests Based on Spatio-Temporal Dynamics, J. R. Soc. Interface. 20, (2023).
After the destruction of their original home, the ants had to choose between two nests of different quality – a “good” nest (low light intensity) and a “poor” nest (high light intensity). An automatic gate at the entrance of each nest provides precise control over the access of specific ants. Illustration from: 
H. Rajendran, A. Haluts, N. S. Gov, and O. Feinerman, Ants Resort to Majority Concession to Reach Democratic Consensus in the Presence of a Persistent Minority, Current Biology 32, 645 (2022).
Snapshot from a simulation of ants cooperatively carrying a T-shaped load in the direction of their nest.

 

Related papers:

  • A. Haluts, A. Jordan, and N. S. Gov, Modelling Animal Contests Based on Spatio-Temporal Dynamics, J. R. Soc. Interface. 20, (2023). DOI: 10.1098/rsif.2022.0866
  • H. Rajendran, A. Haluts, N. S. Gov, and O. Feinerman, Ants Resort to Majority Concession to Reach Democratic Consensus in the Presence of a Persistent Minority, Current Biology 32, 645 (2022). DOI: 10.1016/j.cub.2021.12.013
  • A. Haluts, S. F. G. Reyes, D. Gorbonos, R. I. Etheredge, A. Jordan, and N. S. Gov, Spatiotemporal Dynamics of Animal Contests Arise from Effective Forces between Contestants, Proc. Natl. Acad. Sci. U.S.A. 118, (2021). DOI: 10.1073/pnas.2106269118
  • "What Ants Can Teach Us on Democracy and Social Cohesion". Article in Haaretz