Long / Position Papers
May consist of up to 9 pages of content, plus unlimited pages for references and appendix.
NLP/CORE Unranked
SocialAgent@NeurIPS 2026
Second Workshop on Large Language Models for Social Reasoning and Simulation
The SocialAgent@NeurIPS 2026 workshop brings together researchers to advance the principled, responsible, and empirically grounded use of large language models as socially situated agents for reasoning, simulation, and evaluation of social behavior. It focuses on LLM-based agent systems that model decision-making, norms, cooperation, and collective dynamics, with an emphasis on validity, ethical implications, and pluralistic evaluation beyond task accuracy.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Short / Abstract Papers
May consist of up to 4 pages of content, plus unlimited pages for references and appendix.
Research areas in scope
01
Workshop Themes
Social reasoning & cognition: Theory of mind, moral judgment, and genuine reasoning vs. prompting artifactsSocial simulation & its validity: LLMs as proxies for individuals and populations, and the validity standards that make simulations trustworthyPluralistic alignment & evaluation: Contested values, emergent norms, and evaluation beyond task accuracy
02
Topics
Agent-Based Social Simulation with LLMs: Coordination, conflict, cooperation, collective action, and information diffusionEmergence, Norm Formation, and Cultural Evolution: The development of norms, beliefs, narratives, and collective behaviorPersona Fidelity and Population Heterogeneity: Psychologically plausible agents, synthetic populations, and demographic variationPlatform and Network Contexts: Exposure structures, feedback loops, incentives, moderation, and governanceInterventions and Counterfactual Experiments: Simulations for policy, system design, and stress testingGrounding, Realism, and Evaluation: Calibration, reproducibility, privacy, and comparisons with human behaviorEthical and Societal Implications: Validity limits, misuse risks, and responsible applications in sensitive domains
Policies worth checking twice
- The reviewing process will be double-blind.
- At least one author of each accepted paper should register and present their work in person at SocialAgent.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026