Research map / Large language models / AI and society
Dancing in Chains: Strategic Persuasion in Academic Rebuttal via Theory of Mind
Zhitao He, Zongwei LYU, Yi R. Fung
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ToM analysis to training data (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 9.04 -> 9.42 | Improved |
| add strategy to training data (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 9.31 -> 9.42 | Improved |
| add thinking to training data (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 9.37 -> 9.42 | Improved |
| add RL after SFT instead of SFT-only (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 8.27 -> 9.42 | Improved |
| add SFT before RL instead of RL-only (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 8.79 -> 9.42 | Improved |
| use SFT+RL instead of DPO (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 8.68 -> 9.42 | Improved |
| add response reward (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 8.63 -> 9.42 | Improved |
| add analysis reward (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 9.23 -> 9.42 | Improved |
| add format reward (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 9.32 -> 9.42 | Improved |
| use Rebuttal-RM reward as the scorer instead of self-reward (ablation) | RebuttalAgent · R2-test (in-domain) | average holistic score: 9.42 -> 9.45 | No clear effect |
| apply framework to Llama-3.1-8B | RebuttalAgent · R2-test (in-domain) | average holistic score: 7.44 -> 9.20 | Improved |
| apply framework to Qwen3-4B | RebuttalAgent · R2-test (in-domain) | average holistic score: 7.69 -> 8.98 | Improved |
Closest papers by meaning
- Theory of Mind and Persuasion Beyond Conversation: Assessing the Capacity of LLMs to Induce Belief States via Planning and Action (NeurIPS 2026 · similarity 0.85)
- LLM Agents Can Be Choice-Supportive Biased Evaluators: An Empirical Study (AAAI 2025 · similarity 0.85)
- Measuring And Improving Persuasiveness Of Large Language Models (ICLR 2025 · similarity 0.84)
- ACED-Bench: Evaluating How LLM Agents Acquire and Act on Causal Evidence (NeurIPS 2026 · similarity 0.84)
- Scaling Inference-Time Computation via Opponent Simulation: Enabling Online Strategic Adaptation in Repeated Negotiation (ICML 2026 · similarity 0.84)
- Should we be going MAD? A Look at Multi-Agent Debate Strategies for LLMs (ICML 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
