mlsanityData to 7 Oct 2026

The Contagion Tensor: A Framework for Measuring Output-Distribution Coupling in Multi-Agent LLM Systems -- and Auditing the Claims It Enables

arXiv 2606.28839 ↗First posted 27 Jun 2026cs.LGcited by 0

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

We introduce the Contagion Tensor, a measurement framework for quantifying how large language model (LLM) output distributions couple across modalities, agents, and time steps. From the tensor we derive the Coupling Amplification Factor (CAF), a family of ratio-based metrics sharing the form CAF = E[T_condition] / E[T_baseline], providing unitless, baseline-referenced measurement with bootstrap confidence intervals. We instantiate CAF in four variants and evaluate the strongest in a complete 2x2x2 block-orthogonal simulation design with modality-specific ablation. The ablation reveals that an apparent image-condition super-linear effect (CAF = 1.40) collapses to sub-linear (CAF = 0.87) when the image perturbation module is disabled, a shift of -0.53 with zero effect on text conditions. We supplement with real-API experiments across two model families: DeepSeek-Chat (R=30) and GPT-4o-mini (R=15, real vision). Under uniform personas, text-only communication produces CAF approx 1.0 in both models. Diverse personas drive convergence (CAF = 0.88). A within-model comparison on GPT-4o-mini reveals: C3 (text) CAF = 1.02 vs. C5 (real vision, R=30) CAF = 1.72 [1.700, 1.733], delta = +0.70, validating the simulation's super-linear image-condition prediction. Of 11 conditions, 5 have been tested on real APIs and 6 remain unverified. Our contribution is two-layered: (1) a measurement instrument that makes output-distribution coupling quantitatively falsifiable; and (2) a transferable ablation protocol that any modular multi-agent simulator can adopt to distinguish genuine coupling from design artifacts.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 17 of its 30 references that are in the map. First posted on arXiv 27 Jun 2026.

Builds on

17 of 30 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025MultiAgentBench: Evaluating the Collaboration and Competition of LLM agents3 Mar
  2. 2023MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework1 Aug
  3. 2023CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society31 Mar

Other references · 14

  1. 2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,500
  2. 2023AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversationcited by 1,906
  3. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,739
  4. 2020CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Modelscited by 703
  5. 2021BBQ: A Hand-Built Bias Benchmark for Question Answeringcited by 683
  6. 2018Gender Bias in Coreference Resolutioncited by 438
  7. 2023Improving Language Model Negotiation with Self-Play and In-Context Learning from AI Feedbackcited by 202
  8. 2020Evaluating the Evaluation of Diversity in Natural Language Generationcited by 127
  9. 2025G-Memory: Tracing Hierarchical Memory for Multi-Agent Systemscited by 112
  10. 2025Do Large Language Models Solve the Problems of Agent-Based Modeling? A Critical Review of Generative Social Simulationscited by 28
  11. 2025Emergent Coordination in Multi-Agent Language Modelscited by 22
  12. 2025EcoLANG: Efficient and Effective Agent Communication Language Induction for Social Simulationcited by 5
2 more
  1. 2025Detecting Perspective Shifts in Multi-agent Systemscited by 4
  2. 2026CASPIAN: Online Detection and Attribution of Cascade Attacks in LLM Multi-Agent Systems via Cross-Channel Causal Monitoringcited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026BOUNDARY_SYNC: Measuring Communication-Induced Representational Coupling in Multi-Agent LLM Systems
  2. 2025Can Compressed LLMs Truly Act? An Empirical Evaluation of Agentic Capabilities in LLM Compression
  3. 2026Characterizing Performance-Energy Trade-offs of Large Language Models in Multi-Request Workflows
  4. 2026A Calibrated Instrument for Measuring How Inference Optimizations Affect Output Quality
  5. 2026UniACE: A Unified Framework for Evaluating LLM Agentic Capabilities
  6. 2026Peer-Voted LLM-Agent Stress Tests Find Feed-Induced Lexical Convergence but No Reliable Matched-Exposure Advantage for Distributed Sources
  7. 2025Leveraging the true depth of LLMs
  8. 2025Emergent Convergence in Multi-Agent LLM Annotation