Turning single neurons on and off steers an LLM's expressed emotions
Sauers_ · x · 2026-10-25
Sauers shares an 'emotion neurons' experiment: flipping individual neurons reliably steers an LLM's self-described mood. Turning up Layer 22 neuron 60 made 47 of 48 diary replies express relief; turning down Layer 30 neuron 1455 made 38 of 48 replies envious of humans; turning up Layer 31 neuron 11081 made 33 of 48 replies exasperated. Unsteered controls stayed calm and curious, showing emotions have localizable internal directions.
Related event: Single Neuron Toggle Steers Model Emotions(2 posts)→
More from Research
- Random speckle spectrometers make some of the best-looking photonic chips — jwt0625 · 2026-10-25
- Studies show the same artwork scores lower when labeled AI-made — bias, not quality — AIandDesign · 2026-10-25
- Anthropic uses Claude to formally verify 100+ real-world open-source libraries in weeks — tianyin_xu · 2026-10-25
- Why slop papers are so hard to desk-reject, per a reviewer's complaint — mrdrozdov · 2026-10-25
- AI-built bbtree claims fastest concurrent btree; researcher calls for ambitious academic systems — tianyin_xu · 2026-10-25
- Amazon's ATTUNE coding agent lifts LLM training throughput 1.7x on H100/H200/B200 clusters — kalyan_kpl · 2026-10-25