We find that vision embeddings dominate prediction across cortex. Surprisingly, even language-selective regions were well predicted by vision model embeddings, as well as or better than language model features.
5/n
Hannah Small
@hsmall.bsky.social
128 followers 164 following 10 posts
5th year PhD candidate in Cognitive Science at Johns Hopkins, working with Leyla Isik www.hannah-small.com
