EUVIMEDScopri EUVIMED
Open access · CC BY 4.0

Multi-Agent Reinforcement Learning for Language-Based Social Deduction

David R. Miller, Sophie L. Gagnon, James P. Wilson

Frontiers in Artificial Intelligence Research · 2026-03-10 · 3 (2) · 1-8

DOI: 10.71465/fair719 ↗

At a glance

Journal article in Frontiers in Artificial Intelligence Research, published online on 2026-03-10. Written by David R. Miller and 2 co-authors. The full text is freely available (open access, CC BY 4.0).

AbstractDati bibliograficiCiteRelated studies

Abstract (original language)

This study applies MARL to train LLM agents for language-based social deduction, where communication directly influences multi-agent outcomes. Speaking policies are optimized using rewards that capture the causal impact of messages on other agents’ beliefs, while listening models predict hidden state information from dialogue. Training on 12,000 simulated game episodes results in a 2.1× increase in win rate over standard RL baselines and demonstrates emergent strategic behaviors such as coordinated accusations and evidence sharing.

Cite

APA

Miller, D. R., Gagnon, S. L., & Wilson, J. P. (2026). Multi-Agent Reinforcement Learning for Language-Based Social Deduction. Frontiers in Artificial Intelligence Research, 3(2), 1-8. https://doi.org/10.71465/fair719

Vancouver

Miller DR, Gagnon SL, Wilson JP. Multi-Agent Reinforcement Learning for Language-Based Social Deduction. Frontiers in Artificial Intelligence Research. 2026;3(2):1-8. doi:10.71465/fair719

BibTeX
@article{Miller2026,
  author = {Miller, David R. and Gagnon, Sophie L. and Wilson, James P.},
  title = {Multi-Agent Reinforcement Learning for Language-Based Social Deduction},
  journal = {Frontiers in Artificial Intelligence Research},
  year = {2026},
  volume = {3},
  number = {2},
  pages = {1--8},
  doi = {10.71465/fair719}
}
Start for free →