LogicGaze: Benchmarking Causal Consistency in Visual Narratives via Counterfactual Verification

Rory Driscoll, Alexandros Christoforos, Chadbourne Davis

While sequential reasoning enhances the capability of Vision-Language Models (VLMs) to execute complex multimodal tasks, their reliability in grounding these reasoning chains withi…