大型语言模型中情绪推理的机制可解释性
Mechanistic Interpretability of Emotion Inference in Large Language Models
AI总结：
本研究探讨了自回归LLMs推断情绪的机制，发现情绪表征在模型特定区域功能定位且符合认知评价理论，通过因果干预可精确引导情绪文本生成，有助于敏感领域的安全与对齐。
摘要：
大型语言模型（LLMs）在从文本预测人类情绪方面展现出令人期待的能力。然而，这些模型处理情绪刺激的机制在很大程度上仍未被探索。我们的研究通过调查自回归LLMs如何推断情绪来填补这一空白，表明情绪表征在功能上定位于模型中的特定区域。我们的评估包括多样化的模型家族和规模，并得到了稳健性检验的支持。然后，我们借助认知评价理论（一种提出情绪源于对环境刺激的评价的成熟心理学框架），表明所识别的表征具有心理学上的合理性。通过对构建的评价概念进行因果干预，我们引导生成并表明输出与理论和直觉预期一致。这项工作突出了一种因果干预并精确塑造情绪文本生成的新方法，可能有益于敏感情感领域的安全性和对齐。
英文摘要：
Large language models (LLMs) show promising capabilities in predicting human emotions from text. However, the mechanisms through which these models process emotional stimuli remain largely unexplored. Our study addresses this gap by investigating how autoregressive LLMs infer emotions, showing that emotion representations are functionally localized to specific regions in the model. Our evaluation includes diverse model families and sizes and is supported by robustness checks. We then show that the identified representations are psychologically plausible by drawing on cognitive appraisal theory, a well-established psychological framework positing that emotions emerge from evaluations (appraisals) of environmental stimuli. By causally intervening on construed appraisal concepts, we steer the generation and show that the outputs align with theoretical and intuitive expectations. This work highlights a novel way to causally intervene and precisely shape emotional text generation, potentially benefiting safety and alignment in sensitive affective domains.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
