ScioMind: A Cognitive Multi-Agent Simulation Framework
Researchers have introduced ScioMind, a cognitively grounded multi-agent social simulation framework that combines structured opinion dynamics with LLM-based reasoning. It integrates memory-anchored belief updates, hierarchical memory architecture, and dynamic agent profiles derived from corpus-grounded retrieval. The framework aims to bridge fixed update rules and unconstrained LLM interactions for studying social opinion dynamics.
Key facts
- ScioMind is a cognitively grounded simulation framework.
- It combines structured opinion dynamics with LLM-based agent reasoning.
- It includes a memory-anchored belief update rule with personality-conditioned anchoring strength.
- It features a hierarchical memory architecture for persistent belief formation.
- Dynamic agent profiles are derived from a corpus-grounded retrieval pipeline.
- The framework enables heterogeneous personalities and evolving internal states.
- It was introduced in a paper on arXiv (2605.13725).
- The approach bridges fixed update rules and unconstrained LLM interactions.
Entities
Institutions
- arXiv