LLM Assertiveness can be Mechanistically Decomposed into Emotional and Logical Components

2025/08/24 by Hikaru Tsujimura, Arush Tagade, Tsujimura, Hikaru +1
Computer Science · Social Sciences · #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #Computational and Text Analysis Methods #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Topic Modeling

paper · pdf · doi:10.48550/arxiv.2508.17182

openalex publication_date 2025/08/24 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Large Language Models (LLMs) often display overconfidence, presenting information with unwarranted certainty in high-stakes contexts. We investigate the internal basis of this behavior via mechanistic interpretability. Using open-sourced Llama 3.2 models fine-tuned on human annotated assertiveness datasets, we extract residual activations across all layers, and compute similarity metrics to localize assertive representations. Our analysis identifies layers most sensitive to assertiveness contrasts and reveals that high-assertive representations decompose into two orthogonal sub-components of emotional and logical clusters-paralleling the dual-route Elaboration Likelihood Model in Psychology. Steering vectors derived from these sub-components show distinct causal effects: emotional vectors broadly influence prediction accuracy, while logical vectors exert more localized effects. These findings provide mechanistic evidence for the multi-component structure of LLM assertiveness and highlight avenues for mitigating overconfident behavior.

Citations

• Epistemic Integrity in Large Language Models
• Taming Overconfidence in LLMs: Reward Calibration in RLHF
• External Invariants: A Cryptographic Trust Architecture for Institutional AI Inference
• Large Language Models in Law: A Survey
• Cognitive Reframing of Negative Thoughts through Human-Language Model Interaction
• Governance Architecture for Neural Network Superposition: A Structural Solution to Hallucination via Routing and Interference Filtering
• Measuring Sentence-Level and Aspect-Level (Un)certainty in Science Communications
• LoRA: Low-Rank Adaptation of Large Language Models

Related

• Wired for Overconfidence: A Mechanistic Perspective on Inflated Verbalized Confidence in LLMs
• Do Language Models Mirror Human Confidence? Exploring Psychological Insights to Address Overconfidence in LLMs
• From Prompts to Constructs: A Dual-Validity Framework for LLM Research in Psychology
• Do Emotions Really Affect Argument Convincingness? A Dynamic Approach with LLM-based Manipulation Checks
• Gauging Overprecision in LLMs: An Empirical Study