@article{howvisionbecomeslanguagealayerwiseinform, title = {How Vision Becomes Language: A Layer-wise Information-Theoretic Analysis of Multimodal Reasoning}, author = {Hongxuan Wu and Yukun Zhang and Xueqing Zhou}, year = {2026}, eprint = {2602.15580}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2602.15580}, }