Minimal and Mechanistic Conditions for Behavioral Self-Awareness in LLMs

2025/11/06 by Bozoukov, Matthew, Nguyen, Matthew, Singh, Shubkarman +2
Computer Science · #Adversarial Robustness in Machine Learning #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Reinforcement Learning in Robotics

paper · doi:10.48550/arxiv.2511.04875

openalex publication_date 2025/11/06 · openalex created_date 2025/11/11 · openalex updated_date 2026/07/28

Abstract

Recent studies have revealed that LLMs can exhibit behavioral self-awareness: the ability to accurately describe or predict their own learned behaviors without explicit supervision. This capability raises safety concerns as it may, for example, allow models to better conceal their true abilities during evaluation. We attempt to characterize the minimal conditions under which such self-awareness emerges, and the mechanistic processes through which it manifests. Through controlled finetuning experiments on instruction-tuned LLMs with low-rank adapters (LoRA), we find: (1) that self-awareness can be reliably induced using a single rank-1 LoRA adapter; (2) that the learned self-aware behavior can be largely captured by a single steering vector in activation space, recovering nearly all of the fine-tune's behavioral effect; and (3) that self-awareness is non-universal and domain-localized, with independent representations across tasks. Together, these findings suggest that behavioral self-awareness emerges as a domain-specific, linear feature that can be easily induced and modulated.

Citations

• Simple Mechanistic Explanations for Out-Of-Context Reasoning
• Does It Make Sense to Speak of Introspection in Large Language Models?
• One-shot Optimized Steering Vectors Mediate Safety-relevant Behaviors in LLMs
• Tell me about yourself: LLMs are aware of their learned behaviors
• Looking Inward: Language Models Can Learn About Themselves by Introspection
• Self-Cognition in Large Language Models: An Exploratory Study
• External Invariants: A Cryptographic Trust Architecture for Institutional AI Inference
• The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets
• LoRA: Low-Rank Adaptation of Large Language Models

Related

• Tell me about yourself: LLMs are aware of their learned behaviors
• Evidence for Limited Metacognition in LLMs
• Revealing Hidden Model Behaviors with Task-Specific Self-Reports
• Mechanisms of Introspective Awareness
• LLMs Position Themselves as More Rational Than Humans: Emergence of AI Self-Awareness Measured Through Game Theory