Why animals swirl and how they group

Area of Science:

• Animal behavior
• Computational neuroscience
• Collective motion

Background:

• The biological function of animal swirling motion, where groups orbit a common center, remains poorly understood.
• Existing models often rely on complex physical interactions, lacking broad applicability.

Purpose of the Study:

• To propose a novel explanation for animal swirling behavior using reinforcement learning (RL).
• To determine the biological advantages of swirling motion in animal groups.
• To analyze coordinated group locomotion in viscous fluids using RL.

Main Methods:

• Modeling animal learning processes with reinforcement learning (RL) algorithms.
• Developing a simple set of learned rules for agents to achieve swirling behavior.
• Analyzing the effects of RL-trained swirling on swarm resistance to perturbations.
• Applying RL to model energy dissipation in group locomotion.

Main Results:

• Reinforcement learning successfully models swirling as 'escort behavior,' where individuals maintain a specific distance from the swarm center.
• Swirling behavior significantly enhances a swarm's resistance to external perturbations by orders of magnitude.
• RL optimizes animal disposition for energy-efficient locomotion in viscous fluids.

Conclusions:

• Reinforcement learning offers a parsimonious and effective framework for understanding animal collective motion, including swirling.
• Swirling motion serves a crucial biological function by increasing swarm stability and resilience.
• RL-based models can elucidate energy efficiency in coordinated animal movement.