Modes of information flow in collective cohesion

Area of Science:

• Complex Systems
• Statistical Physics
• Behavioral Ecology

Background:

• Pairwise interactions are crucial for collective behavior and group cohesion.
• Quantifying individual influence on collective dynamics is a persistent challenge.
• Existing methods like mutual information and transfer entropy have limitations in measuring information flow.

Purpose of the Study:

• To identify pitfalls in current information measures for analyzing agent trajectories.
• To develop a novel method for accurately quantifying information flow between individuals in collectives.
• To expose the roles of individual and group memory in collective behavior.

Main Methods:

• Decomposition of information measures into three distinct modes.
• Analysis of information flow between single pairs of agents.
• Investigation of many-body nonadditive interactions without conditioning on additional agents.

Main Results:

• Time-delayed mutual information and transfer entropy can yield misleading interpretations of information flow.
• Decomposed information modes accurately capture pairwise mutual influence.
• The method reveals many-body interactions without explicit conditioning on other agents.

Conclusions:

• The pairwise decomposed modes of information flow offer an improved diagnostic tool for mutual influence in collectives.
• This approach overcomes limitations of traditional information-theoretic measures.
• Understanding decomposed information flow enhances the study of collective behavior.