Think-Before-Speak: From Internal Evaluation to Public Expression in Multi-Agent Social Simulation
The paper introduces a new framework called Think-Before-Speak (TBS) for multi-agent social simulation. TBS separates agents' internal reasoning from their public expressions, allowing for a more nuanced understanding of social interactions. The framework has been evaluated in simulated discussions on climate policy, revealing insights into how internal evaluations influence public dialogue.
- ▪TBS is an interval-based multi-agent simulation framework that distinguishes between private reasoning and public utterance generation.
- ▪The framework updates agents' internal states based on dialogue history and personal memory at each interval.
- ▪Results indicate that dissonance-related appraisal increases agents' willingness to speak, while silence-pressure appraisal decreases it.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2606.03137 (cs) [Submitted on 2 Jun 2026] Title:Think-Before-Speak: From Internal Evaluation to Public Expression in Multi-Agent Social Simulation Authors:Kaiqi Yang, Tai-Quan Peng, Sanguk Lee, Hui Liu View a PDF of the paper titled Think-Before-Speak: From Internal Evaluation to Public Expression in Multi-Agent Social Simulation, by Kaiqi Yang and 3 other authors View PDF HTML (experimental) Abstract:LLM-based multi-agent simulation offers a promising way to study social interaction, deliberation, and collective opinion dynamics.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
