[2603.03590] Social Norm Reasoning in Multimodal Language Models: An Evaluation

Computer Science > Multiagent Systems arXiv:2603.03590 (cs) [Submitted on 3 Mar 2026] Title:Social Norm Reasoning in Multimodal Language Models: An Evaluation Authors:Oishik Chowdhury, Anushka Debnath, Bastin Tony Roy Savarimuthu View a PDF of the paper titled Social Norm Reasoning in Multimodal Language Models: An Evaluation, by Oishik Chowdhury and 2 other authors View PDF HTML (experimental) Abstract:In Multi-Agent Systems (MAS), agents are designed with social capabilities, allowing them to understand and reason about social concepts such as norms when interacting with others (e.g., inter-robot interactions). In Normative MAS (NorMAS), researchers study how norms develop, and how violations are detected and sanctioned. However, existing research in NorMAS use symbolic approaches (e.g., formal logic) for norm representation and reasoning whose application is limited to simplified environments. In contrast, Multimodal Large Language Models (MLLMs) present promising possibilities to develop software used by robots to identify and reason about norms in a wide variety of complex social situations embodied in text and images. However, prior work on norm reasoning have been limited to text-based scenarios. This paper investigates the norm reasoning competence of five MLLMs by evaluating their ability to answer norm-related questions based on thirty text-based and thirty image-based stories, and comparing their responses against humans. Our results show that MLLMs demonstrate ...