HU, H. Right Answers, Wrong Reasons: Perception Errors in Multimodal Chain-of-Thought and the Blind Spot of Outcome-Based Rewards. International Journal of Advanced AI Applications, [S. l.], v. 2, n. 10, p. 77–101, 2026. Disponível em: https://www.dawnclarity.press/index.php/ijaaa/article/view/193. Acesso em: 13 oct. 2026.