![[2601.15715] RebuttalAgent: Strategic Persuasion in Academic Rebuttal via Theory of Mind](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2601.15715] RebuttalAgent: Strategic Persuasion in Academic Rebuttal via Theory of Mind
Summary
The paper presents RebuttalAgent, a framework using Theory of Mind for strategic persuasion in academic rebuttals, addressing the complexities of effective communication in research disputes.
Why It Matters
This research is significant as it tackles the underexplored challenge of academic rebuttal, which often lacks effective strategic communication. By integrating Theory of Mind, the RebuttalAgent aims to enhance the quality of rebuttals, potentially improving the academic discourse and peer review processes.
Key Takeaways
- RebuttalAgent introduces a novel approach to academic rebuttal using Theory of Mind.
- The framework includes a ToM-Strategy-Response model for effective persuasion.
- RebuttalBench, a large dataset, was created to train the agent through critique-and-refine methods.
- The agent shows an 18.3% improvement over baseline models in automated evaluations.
- Rebuttal-RM, a specialized evaluator, achieves high scoring consistency with human preferences.
Computer Science > Computation and Language arXiv:2601.15715 (cs) [Submitted on 22 Jan 2026 (v1), last revised 25 Feb 2026 (this version, v3)] Title:RebuttalAgent: Strategic Persuasion in Academic Rebuttal via Theory of Mind Authors:Zhitao He, Zongwei Lyu, Yi R Fung View a PDF of the paper titled RebuttalAgent: Strategic Persuasion in Academic Rebuttal via Theory of Mind, by Zhitao He and 2 other authors View PDF HTML (experimental) Abstract:Although artificial intelligence (AI) has become deeply integrated into various stages of the research workflow and achieved remarkable advancements, academic rebuttal remains a significant and underexplored challenge. This is because rebuttal is a complex process of strategic communication under severe information asymmetry rather than a simple technical debate. Consequently, current approaches struggle as they largely imitate surface-level linguistics, missing the essential element of perspective-taking required for effective persuasion. In this paper, we introduce RebuttalAgent, the first framework to ground academic rebuttal in Theory of Mind (ToM), operationalized through a ToM-Strategy-Response (TSR) framework that models reviewer mental state, formulates persuasion strategy, and generates evidence-based response. To train our agent, we construct RebuttalBench, a large-scale dataset synthesized via a novel critique-and-refine approach. Our training process consists of two stages, beginning with a supervised fine-tuning phase to e...