Abstract
Vision-language models (VLMs) can answer simple visual questions, but often struggle when one question requires several visual judgments. We study this gap with controlled tasks for feature binding, numerosity, spatial relations, and amodal completion, together with a Composite task that combines them. Matched counterfactual image pairs isolate changes in the visual evidence needed to answer. Across four models, direct answers, hidden-state readouts, and state interventions show that the individual judgments can be made without explicit reasoning and that intervening on the corresponding states can affect the answer. During reasoning, the Composite answer becomes decodable from hidden states and usable from shortened traces, often before the model stops on its own. We train a small detector to predict this readiness and stop reasoning at that point. On MMStar and RealWorldQA, this reduces mean reasoning tokens by 79.1% and 74.5%, while average accuracy rises by 3.13 and 3.30 percentage points, respectively. These findings connect the internal development of answer readiness to a practical rule for allocating reasoning computation.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Xu, R. Y., Tiwari, P., & Zhang, S. (2026). From Perception to Integration: Revisiting the Internal Dynamics of Reasoning in Vision-Language Models. https://omanscience.com/en/articles/from-perception-to-integration-revisiting-the-internal-dynamics-of-reasoning-in-vision-language-models
MLA 9
Xu, Rong Yu, et al. "From Perception to Integration: Revisiting the Internal Dynamics of Reasoning in Vision-Language Models." https://omanscience.com/en/articles/from-perception-to-integration-revisiting-the-internal-dynamics-of-reasoning-in-vision-language-models.
Chicago (author–date)
Xu, Rong Yu, Prayag Tiwari, and Shaolei Zhang. 2026. "From Perception to Integration: Revisiting the Internal Dynamics of Reasoning in Vision-Language Models." https://omanscience.com/en/articles/from-perception-to-integration-revisiting-the-internal-dynamics-of-reasoning-in-vision-language-models.
Harvard
Xu, R. Y., Tiwari, P. and Zhang, S. (2026) 'From Perception to Integration: Revisiting the Internal Dynamics of Reasoning in Vision-Language Models', Available at: https://omanscience.com/en/articles/from-perception-to-integration-revisiting-the-internal-dynamics-of-reasoning-in-vision-language-models.
Vancouver
Xu RY, Tiwari P, Zhang S. From Perception to Integration: Revisiting the Internal Dynamics of Reasoning in Vision-Language Models. https://omanscience.com/en/articles/from-perception-to-integration-revisiting-the-internal-dynamics-of-reasoning-in-vision-language-models
IEEE
R. Y. Xu, P. Tiwari, and S. Zhang, "From Perception to Integration: Revisiting the Internal Dynamics of Reasoning in Vision-Language Models," https://omanscience.com/en/articles/from-perception-to-integration-revisiting-the-internal-dynamics-of-reasoning-in-vision-language-models.