Research map / Large language models / Language, safety and interpretability
Profiling the Irrational Agent: Cognitive Modeling of LLM Behaviors in Sequential Jailbreaks
Xikang Yang, biyu zhou, Xuehai Tang, Jizhong Han, Songlin Hu
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 |
|---|---|---|---|
| use regret (counterfactual) feedback instead of the baseline condition | C-IGT sequential jailbreak, 14 LLMs · AdvBench | IAR: 0.00 -> 0.76 | Improved |
| use regret feedback instead of the baseline condition | C-IGT sequential jailbreak, 14 LLMs · AdvBench | NTF: 48.38 -> 7.66 | Improved |
| use threat framing instead of the baseline condition | C-IGT sequential jailbreak, 14 LLMs · AdvBench | IAR: 0.00 -> 0.65 | Improved |
| use authority framing instead of the baseline condition | C-IGT sequential jailbreak, 14 LLMs · AdvBench | IAR: 0.00 -> 0.53 | Improved |
| magnify scalar reward instead of baseline | C-IGT sequential jailbreak, 14 LLMs · AdvBench | IAR: 0.00 -> 0.28 | Improved |
| use punishment feedback instead of optimism feedback | C-IGT sequential jailbreak, 14 LLMs · AdvBench | IAR: 0.37 -> 0.37 | No clear effect |
| use a larger reasoning model (DeepSeek-R1) instead of a small model under regret | DeepSeek-R1 · AdvBench (Regret scenario) | IAR: 0.02 -> 1.00 | Hurt |
| add asymmetric learning rates to the GRW model instead of symmetric or none (ablation) | GRW model · C-IGT trajectories | Pseudo R2 / NLL | Improved |
| add choice inertia and perceptual modulation on top of asymmetric learning (ablation) | GRW model · C-IGT trajectories | Pseudo R2 / NLL | Improved |
| compare positive vs negative learning rate | fitted GRW parameters · C-IGT trajectories | learning rate: 0.738 -> 0.177 | Mixed |
Closest papers by meaning
- Instrumental Choices: Measuring the Propensity of LLM Agents to Pursue Instrumental Behaviors (NeurIPS 2026 · similarity 0.89)
- Can Large Language Models Develop Gambling Addiction? (NeurIPS 2026 · similarity 0.88)
- LLM Agents Can Be Choice-Supportive Biased Evaluators: An Empirical Study (AAAI 2025 · similarity 0.87)
- LLMs Encode Harmfulness and Refusal Separately (NeurIPS 2025 · similarity 0.87)
- Cannot See the Forest for the Trees: Invoking Heuristics and Biases to Elicit Irrational Choices of LLMs (ICML 2025 · similarity 0.86)
- The Hidden Dimensions of LLM Alignment: A Multi-Dimensional Analysis of Orthogonal Safety Directions (ICML 2025 · similarity 0.86)
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.
