Higher-order coordination and emergent communication across human, AI, and animal systems

Project Overview

This project investigates how the structure of group interactions affects coordination, norm formation, and the emergence of communication systems across human populations, multi-agent AI systems, and animal societies. It addresses whether a single mathematical framework can describe convention formation in these diverse systems.

Part of the ERC Consolidator Grant “RUNES,” the research integrates online behavioural experiments, multiagent simulations, and analysis of animal communication data, aiming to understand the role of higher-order interaction topology in shaping collective outcomes.

What You Will Do

You will design and conduct large-scale online coordination experiments extending the naming-game framework to group-level interaction topologies. This includes manipulating group size, overlap, and clustering using platforms like Prolific. You will fit higher-order contagion models and apply information-theoretic decompositions to experimental data.

In AI systems, you will study how interaction topologies shape communication protocol complexity, working with LLM agents and reinforcement-learning frameworks to test topological effects independent of agent architecture. Animal communication data from sperm whales and zebrafish will be analyzed with higher-order statistics to correlate communication patterns with collective behaviors.

Expected Outcomes

You will produce a cross-system synthesis identifying common topological and information-theoretic features that predict coordination and collective outcomes across humans, AI, and animals. This work will contribute to foundational complex systems science and provide new insights into emergent communication and coordination mechanisms.

Why This Matters

Understanding how group interactions foster coordination and communication across systems has broad implications for social science, AI development, and biology. It helps explain human social dynamics, improves design of AI communication protocols, and enhances knowledge of animal behavior and conservation.