mlsanityData to 7 Oct 2026

Interaction Protocol Shapes Moral Judgment in Multi-Agent Debate

arXiv 2510.10002 ↗First posted 11 Oct 2025cs.AIcited by 2

On the map: LLM agents and tool use → Social simulation and multi-agent debate

Abstract (arXiv)

As agentic AI systems are deployed in advisory and evaluative roles, understanding how multi-agent interactions shape behavior becomes essential. Multi-agent debate has been studied as a mechanism to improve accuracy, but less is known about how debate structure -- the interaction protocol -- affects the values, dynamics, and consensus patterns that emerge when models navigate contested, real-world decisions. We address this gap by facilitating multi-agent debates among three models (GPT-4.1, Claude 3.7 Sonnet, and Gemini 2.0 Flash) to collectively assign blame in 1,000 everyday dilemmas from Reddit's ``Am I the Asshole'' community. We compare synchronous (parallel) and round-robin (sequential) interaction protocols, mirroring two fundamental ways multi-agent systems are orchestrated in practice. Across more than 30,000 total debates, our findings show striking behavioral differences, which we characterize through two dynamics: inertia and conformity. In the synchronous setting, GPT-4.1 showed stronger inertia (0.6-3.1% revision rates) than either Claude 3.7 Sonnet or Gemini 2.0 Flash (28-41% revision rates). Meanwhile, in round-robin debates, GPT-4.1 and Gemini 2.0 Flash stood out as highly conforming relative to Claude 3.7 Sonnet, with their verdict behavior strongly shaped by order effects. We further characterized the values invoked during debate, finding that GPT-4.1 emphasized personal autonomy and honest communication relative to its debate partners, while Claude 3.7 Sonnet and Gemini 2.0 Flash prioritized empathetic dialogue. Together, these results show how interaction protocol shapes moral reasoning in multi-turn debates, establishing it as a substantive sociotechnical design consideration in multi-agent systems.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Placed in Social simulation and multi-agent debate, one of 16 directions in LLM agents and tool use. Builds on 42 of its 61 references that are in the map. First posted on arXiv 11 Oct 2025.

Builds on

42 of 61 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025Debate or Vote: Which Yields Better Decisions in Multi-Agent Large Language Models?24 Aug
  2. 2025Stop Overvaluing Multi-Agent Debate -- We Must Rethink Evaluation and Embrace Model Heterogeneity12 Feb
  3. 2023Exploring Collaboration Mechanisms for LLM Agents: A Social Psychology View3 Oct

Other references · 39

  1. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,655
  2. 2023A Survey of Large Language Modelscited by 2,537
  3. 2023AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversationcited by 1,906
  4. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,739
  5. 2023Towards Understanding Sycophancy in Language Modelscited by 988
  6. 2018AI safety via debatecited by 384
  7. 2025Accelerating scientific discovery with Co-Scientistcited by 353
  8. 2024Humans or LLMs as the Judge? A Study on Judgement Biasescited by 314
  9. 2024Debating with More Persuasive LLMs Leads to More Truthful Answerscited by 255
  10. 2024TradingAgents: Multi-Agents LLM Financial Trading Frameworkcited by 182
  11. 2024Emergent social conventions and collective bias in LLM populationscited by 115
  12. 2025ELEPHANT: Measuring and understanding social sycophancy in LLMscited by 115
27 more
  1. 2025Large Language Models Often Know When They Are Being Evaluatedcited by 105
  2. 2025Measuring Sycophancy of Language Models in Multi-turn Dialoguescited by 91
  3. 2023Moral Foundations of Large Language Modelscited by 84
  4. 2024DailyDilemmas: Revealing Value Preferences of LLMs with Quandaries of Daily Lifecited by 57
  5. 2024Large Language Models Reflect the Ideology of their Creatorscited by 56
  6. 2023Towards Reasoning in Large Language Models via Multi-Agent Peer Review Collaborationcited by 48
  7. 2024Calibrating the Confidence of Large Language Models by Eliciting Fidelitycited by 47
  8. 2025Values in the Wild: Discovering and Analyzing Values in Real-World Language Model Interactionscited by 47
  9. 2024Moral Alignment for LLM Agentscited by 39
  10. 2024Large Language Models in Mental Health Care: a Scoping Reviewcited by 34
  11. 2025Probing and Steering Evaluation Awareness of Language Modelscited by 31
  12. 2025The Hawthorne Effect in Reasoning Models: Evaluating and Steering Test Awarenesscited by 29
  13. 2025Normative Evaluation of Large Language Models with Everyday Moral Dilemmascited by 23
  14. 2025LLM-powered Multi-agent Framework for Goal-oriented Learning in Intelligent Tutoring Systemcited by 21
  15. 2024ValueBench: Towards Comprehensively Evaluating Value Orientations and Understanding of Large Language Modelscited by 19
  16. 2025Synthetic Socratic Debates: Examining Persona Effects on Moral Decision and Persuasion Dynamicscited by 19
  17. 2025Will AI Tell Lies to Save Sick Children? Litmus-Testing AI Values Prioritization with AIRiskDilemmascited by 17
  18. 2024Measuring Social Norms of Large Language Modelscited by 16
  19. 2025Literature Review Of Multi-Agent Debate For Problem-Solvingcited by 15
  20. 2025MAGI: Multi-Agent Guided Interview for Psychiatric Assessmentcited by 14
  21. 2024The Moral Turing Test: Evaluating Human-LLM Alignment in Moral Decision-Makingcited by 12
  22. 2025We Should Evaluate Real-World Impactcited by 12
  23. 2025When Two LLMs Debate, Both Think They'll Wincited by 10
  24. 2024MentalAgora: A Gateway to Advanced Personalized Care in Mental Health through Multi-Agent Debating and Attribute Controlcited by 8
  25. 2025Arbiters of Ambivalence: Challenges of Using LLMs in No-Consensus Taskscited by 8
  26. 2023LLMediator: GPT-4 Assisted Online Dispute Resolutioncited by 6
  27. 2024SocialGaze: Improving the Integration of Human Social Norms in Large Language Modelscited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Preserving Disagreement: Architectural Heterogeneity and Coherence Validation in Multi-Agent Policy Simulationcited by 0
  2. 2026Measuring AI Accountability Through Argumentation Analysis: Can Model Reasoning Withstand Scrutiny?cited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Talk Isn't Always Cheap: Understanding Failure Modes in Multi-Agent Debate
  2. 2026The impact of multi-agent debate protocols on debate quality: a controlled case study
  3. 2026Multi-Agent Debate Strategies: Survey, Taxonomy, and Challenges
  4. 2025When and Why Does Multi-Agent Debate Fail and Does It Really Underperform?
  5. 2026The Cost of Consensus: Isolated Self-Correction Prevails Over Unguided Homogeneous Multi-Agent Debate
  6. 2026Multi-Agent Consensus as a Cognitive Bias Trigger in Human-AI Interaction
  7. 2025Voting or Consensus? Decision-Making in Multi-Agent Debate
  8. 2026Not All Flips Are Conformity: Decomposing Stance Convergence in Multi-Agent LLM Debate