Brain2VLM: Hierarchical Alignment Between Cortical Representations and Vision-Language Latent Spaces
N. A. Adarsh Pritam, O. Jeba Shiney, Sanyam Jain
A residual MLP beats ridge by Δ≈0.47 correlation decoding CLIP embeddings from higher visual areas, but only Δ≈0.05 to 0.06 decoding Stable Diffusion latents from early visual cortex.
bioRxiv · Aug 11, 2026 · PreprintNeuroscience✦ In-depth digest