PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
May 7, 2026International Journal of Advanced Computer Science and Applications0 citationsOpen Access

Interruptible Multi-Agent Debate: Sentence-Level Disclosure and Urgency-Based Turn-Taking for Early Error Correction

View Full Paper
AKAkikazu KimuraKFKen FukudaYTYasuyuki Tahara

Key Points

  • To develop an interruptible multi-agent debate framework that enables early error correction and improves reasoning ability.
  • Introduced sentence-level disclosure and urgency-based turn-taking.
  • Implemented a shared public-token budget for agents.
  • Conducted evaluations on 1,000 MMLU questions with three agents.
  • Achieved highest final accuracy in two-incorrect-one-correct setting (49.5%).
  • Improved accuracy in one-incorrect-two-correct setting (79.2%).
  • Demonstrated that interruptions generally enhance listeners' responses.

Abstract

Multi-agent debate (MAD) has emerged as a promising approach for improving the reasoning ability of large language models (LLMs). However, existing turn-taking schemes typically disclose a speaker’s entire utterance before other agents can respond, allowing erroneous premises to spread through the shared context and making early correction difficult. This study proposes an interruptible MAD framework that enables early error correction through sentence-level disclosure and urgency-based turn-taking under a shared public-token budget. Each non-speaking agent continuously generates an action plan including its current assessment, action choice, urgency, and supported answer. The next speaker is then selected dynamically from agents requesting to speak or interrupt, while silent turns are allowed when no intervention is necessary. By revealing only one sentence at a time and discarding undisclosed sentences after interruption, the proposed framework is designed to prevent misleading claims from expanding into long incorrect explanations. Under a controlled evaluation on 1,000 MMLU questions using three agents with conditioned initial states containing both correct and incorrect answers, the proposed framework achieves the highest final accuracy in both the two-incorrect-one-correct setting (49.5% vs. 37.2% and 43.7%) and the one-incorrect-two-correct setting (79.2% vs. 68.7% and 73.8%). Analysis of intermediate answers further show that interruptions improve listeners’ answers more often than they worsen them. These results suggest that fine-grained, interruptible turn-taking can suppress misinformation propagation and stabilize consensus formation under the evaluated setting.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Kimura et al. (2026) studied this question.

synapsesocial.com/papers/69fbe3ca164b5133a91a32b1https://doi.org/10.14569/ijacsa.2026.0170403
Ask AI
Helpful
Bookmark
Share
View Full Paper