Position: Simulating Society Requires Simulating Thought

2025-06-08

AI Summary

Existing LLM-based social behavior simulations predominantly rely on surface-level generation without grounding in cognitive principles, resulting in severe deficiencies in internal consistency, causal reasoning, and belief traceability. Method: We propose Generative Minds (GenMinds), a cognitive modeling paradigm that implements realistic reasoning and decision-making through structured, revisable belief representations. We further introduce RECAP—a novel evaluation framework integrating belief revisability, causal traceability, and demographic anchoring into social simulation assessment. GenMinds integrates cognitive science theory, structured knowledge representation, and causal inference modeling to support dynamic belief evolution and intervention-response analysis. Results: Experiments demonstrate significant improvements in agents’ causal inference accuracy, belief consistency, and robustness to interventions. GenMinds establishes an interpretable, verifiable generative infrastructure for high-stakes applications such as policy simulation.

Technology Category

• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior
• Knowledge Representation and Reasoning: Reasoning with Beliefs
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

Simulating society with large language models (LLMs), we argue, requires more than generating plausible behavior -- it demands cognitively grounded reasoning that is structured, revisable, and traceable. LLM-based agents are increasingly used to emulate individual and group behavior -- primarily through prompting and supervised fine-tuning. Yet they often lack internal coherence, causal reasoning, and belief traceability -- making them unreliable for analyzing how people reason, deliberate, or respond to interventions. To address this, we present a conceptual modeling paradigm, Generative Minds (GenMinds), which draws from cognitive science to support structured belief representations in generative agents. To evaluate such agents, we introduce the RECAP (REconstructing CAusal Paths) framework, a benchmark designed to assess reasoning fidelity via causal traceability, demographic grounding, and intervention consistency. These contributions advance a broader shift: from surface-level mimicry to generative agents that simulate thought -- not just language -- for social simulations.

Problem

• Simulating society requires cognitively grounded reasoning
• LLM-based agents lack internal coherence and causal reasoning
• Generative Minds paradigm supports structured belief representations

Innovation

• Generative Minds paradigm for structured beliefs
• RECAP framework for reasoning fidelity
• Cognitive science-based generative agents