AI RESEARCH

Interpreting and Enhancing Emotional Circuits in Large Vision-Language Models via Cross-Modal Information Flow

arXiv CS.CV

ArXi:2605.21980v1 Announce Type: new Large Vision-Language Models (LVLMs) represent a significant leap towards empathetic agents, nstrating remarkable capabilities in emotion understanding. However, the internal mechanisms governing how LVLMs translate abstract visual stimuli into coherent emotional narratives remain largely unexplored, primarily due to the scarcity of visual counterfactuals and the diffuse nature of emotional expression. In this paper, we bridge this gap by