Vision Language Models Cannot Reason About Physical Transformation
University of Michigan - Ann Arbor · University of California, San Diego · University of California, Berkeley · Johns Hopkins University · Emory University · Westlake University · Columbia University · Auburn University · University of Michigan
PDF 由论文原始站点提供，PaperCompass 不保存论文文件。
摘要
Understanding physical transformations is fundamental for reasoning in dynamic environments. While Vision Language Models (VLMs) show promise in embodied applications, whether they genuinely understand physical transformations remains unclear. We introduce ***ConservationBench*** evaluating ***conservation***—whether physical quantities remain invariant under transformations. Spanning four properties with paired conserving/non-conserving scenarios, we generate 23,040 questions across 112 VLMs. Results reveal systematic failure: performance remains near chance with improvements on conservation tasks accompanied by drops on controls. Control experiments show strong textual priors favoring invariance, yet models perform worse with visual content. Neither temporal resolution, prompting, nor curated sampling helps. These findings show that current VLMs fail to maintain transformation-invariant representations of physical properties across dynamic scenes.