Mechanistic Decoding of Cognitive Constructs in LLMs

arXiv cs.CL / 4/17/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper proposes a Cognitive Reverse-Engineering framework using Representation Engineering to mechanistically analyze complex emotions in LLMs, focusing on social-comparison jealousy.
It combines appraisal theory with subspace orthogonalization, regression-based weighting, and bidirectional causal steering to isolate two psychological antecedents of jealousy.
Experiments across eight LLMs from the Llama, Qwen, and Gemma families find that jealousy is represented as a structured linear combination of “Superiority of Comparison Person” and “Domain Self-Definitional Relevance.”
The study suggests these internal representations align with the human construct, where Superiority acts as a foundational trigger and Relevance functions as an intensity multiplier.
The framework also indicates toxic emotional states can be mechanically detected and surgically suppressed, pointing to representational monitoring and safer interventions in multi-agent settings.

Abstract

While Large Language Models (LLMs) demonstrate increasingly sophisticated affective capabilities, the internal mechanisms by which they process complex emotions remain unclear. Existing interpretability approaches often treat models as black boxes or focus on coarse-grained basic emotions, leaving the cognitive structure of more complex affective states underexplored. To bridge this gap, we propose a Cognitive Reverse-Engineering framework based on Representation Engineering (RepE) to analyze social-comparison jealousy. By combining appraisal theory with subspace orthogonalization, regression-based weighting, and bidirectional causal steering, we isolate and quantify two psychological antecedents of jealousy, Superiority of Comparison Person and Domain Self-Definitional Relevance, and examine their causal effects on model judgments. Experiments on eight LLMs from the Llama, Qwen, and Gemma families suggest that models natively encode jealousy as a structured linear combination of these constituent factors. Their internal representations are broadly consistent with the human psychological construct, treating Superiority as the foundational trigger and Relevance as the ultimate intensity multiplier. Our framework also demonstrates that toxic emotional states can be mechanically detected and surgically suppressed, suggesting a possible route toward representational monitoring and intervention for AI safety in multi-agent environments.