Vision-language encoding models reveal an image-computable food-quality dimension in human occipitotemporal cortex
Giuseppe Marrazzo, Anne Roefs, Leonardo Pimpini
The paper decomposes perceived calorie ratings from food images using CLIP embeddings and shows that a visual-semantic component explains neural representations in ventral visual cortex.
bioRxiv · Sep 28, 2026 · PreprintNeuroscience