Search papers, labs, and topics across Lattice.
The paper introduces ValueGround, a benchmark to evaluate culture-conditioned visual value grounding in MLLMs using minimally contrastive image pairs representing opposing response options from the World Values Survey (WVS). They find that MLLM accuracy in identifying a country's value tendency drops from 72.8% in text-only settings to 65.8% when options are visualized, despite high accuracy in aligning images with text options. This highlights a challenge in cross-modal transfer of cultural value judgments, even for stronger models.
MLLMs struggle to ground cultural values in visual scenes, losing ~7% accuracy compared to text-only inputs, even when they understand the visual content.
Cultural values are expressed not only through language but also through visual scenes and everyday social practices. Yet existing evaluations of cultural values in language models are almost entirely text-only, making it unclear whether models can ground culture-conditioned judgments when response options are visualized. We introduce ValueGround, a benchmark for evaluating culture-conditioned visual value grounding in multimodal large language models (MLLMs). Built from World Values Survey (WVS) questions, ValueGround uses minimally contrastive image pairs to represent opposing response options while controlling irrelevant variation. Given a country, a question, and an image pair, a model must choose the image that best matches the country's value tendency without access to the original response-option texts. Across six MLLMs and 13 countries, average accuracy drops from 72.8% in the text-only setting to 65.8% when options are visualized, despite 92.8% accuracy on option-image alignment. Stronger models are more robust, but all remain prone to prediction reversals. Our benchmark provides a controlled testbed for studying cross-modal transfer of culture-conditioned value judgments.
