The Strong Platonic Representation Hypothesis suggests independently learned embeddings may be translated using geometry alone, without shared data.
We ask whether this also holds across human brains: do visual-cortex representations form compatible versions of a shared neural geometry?
