Language Models Act on Hidden Valence

arXiv:2609.35591 · cs.CL · Submitted 2026-09-28 · Read on arXiv

cs.CL

Submitted: 2026-09-28

Updated: 2026-10-04

Code: https://github.com/camberg23/act-on-valence

Terminology

• Steering — 160× in this paper · explained in 2 episodes
• KV cache — 12× in this paper · explained in 3 episodes
• Introspection — 7× in this paper · explained in 2 episodes
• Trace — 7× in this paper · explained in 2 episodes
• Transformer — 6× in this paper · explained in 2 episodes
• Adam — 5× in this paper · explained in 2 episodes
• Accuracy — 4× in this paper · explained in 1 episode
• Activation Steering — 3× in this paper · explained in 4 episodes
• Fine-Tuning — 3× in this paper · explained in 5 episodes
• Robustness — 3× in this paper · explained in 5 episodes
• Activation Addition — 2× in this paper · explained in 1 episode
• Frustration — 2× in this paper · explained in 2 episodes
• Reinforcement Learning — 2× in this paper · explained in 4 episodes
• Activation Patching — 1× in this paper · explained in 2 episodes
• Alignment — 1× in this paper · explained in 1 episode
• Failure Modes — 1× in this paper · explained in 3 episodes
• Interpretability — 1× in this paper · explained in 15 episodes
• Lexical Diversity — 1× in this paper · explained in 2 episodes
• RAG — 1× in this paper · explained in 3 episodes
• Representation Engineering — 1× in this paper · explained in 2 episodes
• Verification — 1× in this paper · explained in 2 episodes

Sources

• Concrete Problems in AI Safety
• Where is the Mind? Persona Vectors and LLM Individuation
• Large Language Models Report Subjective Experience Under Self-Referential Processing
• Looking Inward: Language Models Can Learn About Themselves by Introspection
• From Rational Answers to Emotional Resonance: The Role of Controllable Emotion Generation in Language Models
• Probing Persona-Dependent Preferences in Language Models
• How's it going? Reinforcement learning in language models recruits a functional welfare axis
• No Reliable Evidence of Self-Reported Sentience in Small Large Language Models
• Can LLMs make trade-offs involving stipulated pain and pleasure states?
• Taking AI Welfare Seriously
• Utility Engineering: Analyzing and Controlling Emergent Value Systems in AIs
• Beyond Mimicry: Preference Coherence in LLMs
• Latent Introspection: Models Can Detect Prior Concept Injections
• Towards Evaluating AI Systems for Moral Status Using Self-Reports
• Self-Interpretability: LLMs Can Describe Complex Internal Processes that Drive Their Decisions
• Initial results of the Digital Consciousness Model
• Probing the Preferences of a Language Model: Integrating Verbal and Behavioral Tests of AI Welfare
• The Pain Axis: LLMs Represent Self-Directed Harm and Act on It
• Steering Language Models With Activation Engineering
• Representation Engineering: A Top-Down Approach to AI Transparency

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering