Turning brain prediction models into testable explanations

Researchers introduce generative causal testing, which translates black box models into clear hypotheses and verifies them in the scanner, revealing what specific brain regions respond to in language.

Raccontata damicrosoft.com