SOURCE-LINKED INTELLIGENCE

Conditional Visual Evidence Utility: State-Dependent Rank Reversals in Frozen Vision-Language Encoders

arXiv · AI, language, vision and robotics · article · Aug 28, 2026 · UTC

Static importance scores compress visual evidence into a single ranking, but the value of remaining evidence can change after one cue has been observed. We study this possibility in controlled compositional visual search, where color, shape, and texture evidence can be independently exposed and their conditional marginal utility measured across acquisition states. In a held-out confirmation on 800 scenes, frozen OpenCLIP and SigLIP exhibit robust state-dependent rank reversals that concentrate in candidate-overlap regimes designed to induce ordering changes. The structure persists across two e

Read original source ↗ Open in workspace

recordType

paper

region

Global