Science
A New Paper Decodes What Someone Sees From Their Brain Activity — Without Training on That Person
Fifteen researchers from CMU, HKU, Columbia, and Harvard report a meta-learning method that reads visual content out of fMRI signals on a completely new subject with no fine-tuning at all. Until now, every brain-decoding model had to be retrained per person; this paper frames the result as 'a critical step towards a generalizable foundation model for non-invasive brain decoding.'