NormViz: A Benchmark and Framework for Grounding Multimodal Reasoning in Global Cultures
On the map: LLM reasoning → Multimodal reasoning
Abstract (arXiv)
AI systems are used worldwide, but they struggle to serve the needs of culturally diverse populations. Prior work on cultural understanding evaluates AI systems on text-only settings or on visual artifact recognition (e.g. foods, clothing). The ability to reason about visually observable behaviors through local social norms, which we call visual norm understanding, remains unexamined. We introduce NormViz-Bench, a high quality, human-validated benchmark of 3,268 contrastive image pairs (6,536 images) spanning 16 countries. Each pair varies only in the culturally relevant behavior (e.g., objects, attributes, spatial relations, and actions) that alters how each image is interpreted. Each image is labeled as conforming to, violating, or irrelevant to local social norms, and pair-level evaluation requires both images to be correctly classified, thereby preventing reliance on superficial visual shortcuts. Even the strongest VLMs, Gemini 3.0 Flash and Qwen2.5 VL 7B, succeed on only 26.6% and 21.6% of pairs, struggling most with identifying violating and culturally benign visual behaviors. Towards bridging this, we introduce NormViz-Train, a training dataset of 64k images paired with explanations. Though absolute performance remains low (<30%), finetuning on NormViz-Train improves pair accuracy relatively by up to 125% and 36% Qwen3-VL 4B and 8B respectively, showing a path forward to teach models to connect visual perception to cultural significance. Together, NormViz-Bench and NormViz-Train establish visual norm understanding as a challenging and consequential frontier for multimodal AI.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 13 of its 71 references that are in the map. First posted on arXiv 6 Sep 2026.
Builds on
13 of 71 references.
References · 13
- 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
- 2025Qwen3-VL Technical Reportcited by 2,793
- 2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,188
- 2023Towards Measuring the Representation of Subjective Global Opinions in Language Modelscited by 437
- 2024Towards Measuring and Modeling "Culture" in LLMs: A Surveycited by 167
- 2024CVQA: Culturally-diverse Multilingual Visual Question Answering Benchmarkcited by 120
- 2025CultureVLM: Characterizing and Improving Cultural Understanding of Vision-Language Models for over 100 Countriescited by 42
- 2024Beyond Aesthetics: Cultural Competence in Text-to-Image Modelscited by 30
- 2024CulturalBench: A Robust, Diverse, and Challenging Cultural Benchmark by Human-AI CulturalTeamingcited by 30
- 2024Cultural Palette: Pluralising Culture Alignment via Multi-agent Palettecited by 16
- 2025Culture in Action: Evaluating Text-to-Image Models through Social Activitiescited by 10
- 2025Multimodal Cultural Safety: Evaluation Framework and Alignment Strategiescited by 8
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025Seeing Culture: A Benchmark for Visual Reasoning and Grounding
- 2025RCI: A Score for Evaluating Global and Local Reasoning in Multimodal Benchmarks
- 2025Toward Socially Aware Vision-Language Models: Evaluating Cultural Competence Through Multimodal Story Generation
- 2026Social Norm Reasoning in Multimodal Language Models: An Evaluation
- 2025VERIFY: A Benchmark of Visual Explanation and Reasoning for Investigating Multimodal Reasoning Fidelity
- 2025BLEnD-Vis: Benchmarking Multimodal Cultural Understanding in Vision Language Models
- 2025All-in-one: Understanding and Generation in Multimodal Reasoning with the MAIA Benchmark
- 2025HumaniBench: A Human-Centric Framework for Large Multimodal Models Evaluation