Finding continuity and discontinuity in fish schools via integrated information theory

Area of Science:

• Collective Behavior Analysis
• Information Theory Applications
• Animal Social Dynamics

Background:

• Collective behaviors are often studied for information processing, neglecting internal interactions.
• Existing methods rarely investigate qualitative shifts in small group sizes (e.g., 3 vs. 4 elements).
• Integrated Information Theory (IIT) quantifies information integration (Φ) and has potential for network analysis.

Purpose of the Study:

• To apply IIT 3.0 to analyze collective behavior in real fish schools (Plecoglossus altivelis).
• To identify novel emergent properties in collective dynamics using IIT.
• To explore the relationship between information integration and leadership in fish schools.

Main Methods:

• Applied Integrated Information Theory (IIT 3.0) to analyze real fish schools (Plecoglossus altivelis).
• Analyzed IIT on Boids simulations with varying coupling strengths for comparison.
• Measured information loss via minimum information partition to calculate Φ.

Main Results:

• Observed a discontinuity in 〈Φ(N)〉 distributions for schools of four or more fish, unlike mutual information or transfer entropy.
• Found that Boids simulations yielded different results compared to real fish schools.
• Identified a correlation between this discontinuity and the emergence of IIT-induced leadership, defined by group autonomy.

Conclusions:

• IIT reveals a critical transition in fish collective behavior at N=4, not detectable by other measures.
• IIT-induced leadership reflects group integrity and autonomy, offering a new perspective beyond traditional definitions.
• The findings suggest IIT is a valuable tool for uncovering hidden dynamics and emergent leadership in complex systems.