mlsanityData to 14 Oct 2026

From Passive to Persuasive: Localized Activation Injection for Empathy and Negotiation

arXiv 2511.12832 ↗First posted 16 Nov 2025cs.CL cs.AIcited by 2

On the map: Mechanistic interpretability → Activation steering methods

Abstract (arXiv)

Complex social behaviors, such as empathy and strategic politeness, are widely assumed to resist the directional decomposition that makes activation steering effective for coarse attributes like sentiment or toxicity. We present STAR: Steering via Attribution and Representation, which tests this assumption by using attribution patching to identify the layer--token positions where each behavioral trait causally originates, then injecting contrastive activation vectors at precisely those locations. Evaluated on emotional dialogue and negotiation in both single- and multi-turn settings, localized injection consistently outperforms global steering and instruction priming; human evaluation confirms that gains reflect genuine improvements in perceived quality rather than lexical surface change. Our results suggest that complex interpersonal behaviors are encoded as localized, approximately linear directions in LLM activation space, and that behavioral alignment is fundamentally a localization problem.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability. Builds on 15 of its 31 references that are in the map. First posted on arXiv 16 Nov 2025.

Builds on

15 of 31 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024AtP*: An efficient and scalable method for localizing LLM behaviour to components1 Mar
  2. 2023Steering Llama 2 via Contrastive Activation Addition9 Dec
  3. 2023Steering Language Models With Activation Engineering20 Aug

Other references · 12

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,307
  3. 2022Locating and Editing Factual Associations in GPTcited by 2,691
  4. 2020Aligning AI With Shared Human Valuescited by 723
  5. 2023Emergent Linear Representations in World Models of Self-Supervised Sequence Modelscited by 319
  6. 2023Evaluating Large Language Models in Theory of Mind Taskscited by 222
  7. 2020Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claimscited by 165
  8. 2018Decoupling Strategy and Generation in Negotiation Dialoguescited by 162
  9. 2019What makes a good conversation? How controllable attributes affect human judgmentscited by 140
  10. 2020ConvoKit: A Toolkit for the Analysis of Conversationscited by 72
  11. 2023Reinforcement Learning with Human Feedback: Learning Dynamic Choices via Pessimismcited by 69
  12. 2025Mind the Value-Action Gap: Do LLMs Act in Alignment with Their Values?cited by 51

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Studycited by 1
  2. 2026Detection != Reliable Control: Decodable Empathy Directions Yield at Most Partial Shifts in Automated Empathy Scorescited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Conversations: Love Them, Hate Them, Steer Them
  2. 2026Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs
  3. 2025What Can We Actually Steer? A Multi-Behavior Study of Activation Control
  4. 2026Activation Steering via Generative Causal Mediation
  5. 2026Steering Follows Geometry, Not Labels: Emotion Directions in a Full-Duplex Speech Model
  6. 2026Emotional intelligence in large language models is fragmented across perception, cognition, and interaction
  7. 2026Psychological Steering of Large Language Models
  8. 2026Activation Steering Induces Emergent Misalignment: A More Comprehensive Evaluation