Emergent Relational Order in LLM Agent Societies: From Collective Affect to Authority Stratification

2026-06-22 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Expert, quick

Summary

A new multi-agent framework, CAREB-MAS, utilizes large language models to simulate emergent relational order in agent societies, drawing on Affect Control Theory, Social Identity Theory, and Durkheimian collective affect. Published on 2026-06-22, this framework models agents reasoning through an emotion-ethics-belief chain and maintaining dynamic egocentric identities within a macro environment specifying individual production and preference-based allocation. Long-horizon simulations with CAREB-MAS spontaneously reproduce five core Differential Order phenomena: stable labor specialization, guanxi-based economic ethics, relational decay of cooperation, emergent relational authority, and clan-based center-periphery stratification. These simulated patterns adapt based on production structure, shifting from kin-centered integration towards greater functional interdependence. The research interprets Differential Order as a structure-sensitive emergent outcome of general social mechanisms, positioning LLM-based multi-agent simulation as a robust interdisciplinary framework for studying social structure and change.

Key takeaway

For research scientists exploring complex social systems, this work demonstrates how LLM-based multi-agent simulations can model emergent relational orders. You should consider CAREB-MAS or similar frameworks to investigate long-horizon social structures and their underlying mechanisms. This approach offers a powerful interdisciplinary tool for understanding how collective affect and individual identities shape societal stratification and cooperation dynamics.

Key insights

LLM-based multi-agent simulations can spontaneously reproduce complex social structures like Differential Order through emotion-ethics-belief chains.

Principles

• Social structures emerge from general mechanisms.
• Egocentric identities drive relational dynamics.
• Production structure influences social patterns.

Method

The CAREB-MAS framework simulates agents with emotion-ethics-belief chains and dynamic egocentric identities, interacting via minimal protocols to reproduce long-horizon social phenomena.

In practice

• Model social dynamics with LLM agents.
• Explore emergent behaviors in simulated societies.
• Analyze structure-sensitive social change.

Topics

• LLM Agents
• Multi-agent Systems
• Social Simulation
• Emergent Social Order
• Affect Control Theory
• Social Identity Theory

Best for: AI Scientist, Research Scientist

Related on AIssential

• Beyond Static Responses: Multi-Agent LLM Systems as a New Paradigm for Social Science Research — A six-tier framework maps LLM agent evolution from static tools to complex adaptive systems for social science research.
• Emergence of fragility in LLM-based social networks: an interview with Francesco Bertolotti — Interacting LLM agents spontaneously form social networks exhibiting emergent properties akin to human societies.
• Network Effects and Agreement Drift in LLM Debates — LLM agent opinion dynamics are shaped by network structure, group imbalance, and a systematic "agreement drift" bias.
• Social Structure Matters in 3D Human-Human Interaction Generation — LLMs can plan social structure for 3D human-human interaction, but require a separate motion executor to generate physically plausible movements.
• Agentopia: Long-Term Life Simulation and Learning in Agent Societies — LLMs can learn human-like social intelligence and improve role-playing through long-term, reward-driven social simulation.
• Scaling Behavior of Single LLM-Driven Multi-Agent Systems — LLM-driven multi-agent system performance shows diminishing returns due to coordination overhead, not monotonic scaling.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.