Bayesian inference for identifying interaction rules in moving animal groups.

Area of Science:

• Collective behavior
• Animal movement ecology
• Statistical modeling

Background:

• Collective patterns in animal groups often exhibit universality, masking underlying interaction details.
• Inferring specific animal interactions from macroscopic dynamics is challenging due to model ambiguity.

Purpose of the Study:

• Develop a Bayesian framework to learn animal interaction rules from fine-scale movement data.
• Address the inverse problem of inferring interaction rules from simulation models.
• Quantify confidence in parameter fitting for animal interaction models.

Main Methods:

• Utilized a Bayesian framework for analyzing fine-scale recordings of animal movements.
• Applied techniques to infer interaction rules from simulation models.
• Assessed the impact of data collection rates and tested topological vs. metric neighborhood models.

Main Results:

• Demonstrated parameter inference from a limited number of observations.
• Showed reliable estimation of attraction and alignment terms during milling behavior.
• Identified limitations in measuring interaction radius in specific configurations (e.g., torus shape).

Conclusions:

• The developed methodology aids in designing experiments for studying animal interactions.
• Provides guidance on optimal data analysis techniques for collective animal behavior.
• Highlights the importance of fine-scale data for understanding biological interactions.