On January 3, a paper by Ms. Shirin Vafaei, a fifth-year PhD student in our lab, was published in Communications Biology.

Title: “Brain-aligning of semantic vectors improves neural decoding of visual stimuli“

In brain decoding, which aims to predict stimuli from neural activity, conventional image- and text-based feature vectors are misaligned with neural representations, leading to reduced accuracy.
In this study, we propose “brain-aligned vectors,” which fine-tune these feature vectors to better match the structure of neural representations.
Moreover, this method, trained on fMRI data, improved decoding accuracy across fMRI, MEG, and ECoG datasets.