Multi-scale inference of interaction rules in animal groups using Bayesian model selection.

Area of Science:

• Collective animal behavior
• Animal interaction modeling
• Non-Markovian dynamics

Background:

• Understanding animal group movement often relies on large-scale system analysis.
• Recent studies utilize fine-scale animal motion for validating interaction rules.
• Collective motion in animal groups is a key area of ecological and behavioral research.

Purpose of the Study:

• To compare various interaction models for collective animal motion using Bayesian methodology.
• To investigate the collective motion of glass prawns (Paratya australiensis) and their response to density changes.
• To determine if memory of past interactions influences prawn collective movement.

Main Methods:

• Bayesian methodology for model comparison.
• Analysis of fine-scale and large-scale collective motion data.
• Fitting and evaluating mean-field, spatial Markovian, and non-Markovian models.

Main Results:

• Glass prawns display a density-dependent 'phase transition' into collective motion.
• Mean-field models capture large-scale behavior but not interaction locality.
• Traditional self-propelled particle models fail to capture fine-scale dynamics.
• Non-Markovian models accurately reproduce both fine-scale and global dynamics.

Conclusions:

• Glass prawn movement is influenced by both current and recent past interactions of conspecifics.
• Realistic self-propelled particle models for collective motion should incorporate memory and non-Markovian effects.
• The study suggests memory plays a crucial role in collective animal behavior across biological scales.