AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Do LLMs “Feel”? Emotion Circuits Discovery and Control

Chenxi Wang, Yixuan Zhang, Ruiji Yu, Yufei Zheng, Lang Gao, Zirui Song, Zixiang Xu, Gus Xia, Huishuai Zhang, Dongyan Zhao, Xiuying Chen

ICML 2026 · Large language models › Language, safety and interpretability · approach: Mechanistic analysis of attention and representations

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 changedWhereMeasuredEffect
modulate the assembled global emotion circuit instead of prompting or a single steering vectorLLaMA-3.2-3B-Instruct · held-out test setemotion expression accuracy (%): 91.22 -> 99.65 (steering vs circuit)Improved
use circuit modulation instead of steering for surpriseLLaMA-3.2-3B-Instruct · held-out test set, surpriseemotion expression accuracy (%): 67.71 -> 100Improved
ablate top-ranked emotion neurons (k=2, 4) per layer (ablation)LLaMA-3.2-3B-Instruct · emotion score on test setemotion score changeHurt
ablate randomly chosen neurons or heads instead of top-ranked ones (ablation)LLaMA-3.2-3B-Instruct · emotion score on test setemotion score changeNo clear effect
enhance top-k neurons with emotion difference vectors (ablation)LLaMA-3.2-3B-Instruct · emotion score without emotion instructionemotion score changeImproved
enhance a few top attention heads (ablation)LLaMA-3.2-3B-Instruct · emotion score without emotion instructionemotion score changeImproved
ablate top-2 neurons on Qwen2.5-7B (ablation)Qwen2.5-7B-Instruct · anger emotion scoredelta s: -100.30 (top-2) vs -0.27 (random)Hurt
ablate random attention heads on Qwen2.5-7B (ablation)Qwen2.5-7B-Instruct · emotion scoredelta s: -322.77 (random k=8) vs -544.44 (anger k=8)Hurt
apply emotion steering vectors to Qwen2.5-7B negative emotionsQwen2.5-7B-Instruct · held-out test setemotion expression accuracy (%): anger 0.60, fear 0.00 vs happiness 93.30Mixed
circuit-based steering at scale 2.0 on Qwen2.5-7BQwen2.5-7B-Instruct · held-out test setemotion expression accuracy (%): disgust 100.00, anger 18.33, average 66.18Mixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy