GoKawiilWeizmann researchers unveil improved fMRI-to-image brain decoder
Researchers led by Michal Irani at the Weizmann Institute built an AI system that reconstructs images a person is viewing from their fMRI brain scans, aiming for closer matches in structure and content than prior decoders. The system uses two coupled models—a decoder that splits predictions into image structure and content, and an encoder that predicts brain activity from images—trained together to refine outputs via a diffusion model. About 70% of training images were never actually shown to subjects in a scanner, since the encoder generated synthetic brain-activity pairings for them.
GoKawiil's interpretation of the reporting above, not reported fact.
By using an encoder to simulate brain responses to unseen images, the team sidesteps the scarcity of paired fMRI-image datasets, which could let future decoders scale well beyond the small pool of people who have undergone extensive scanning. This approach suggests brain-decoding accuracy may improve faster than new scanning data alone would allow, though it remains unclear how well it generalizes beyond the eight subjects studied.
- New fMRI decoder splits predictions into image structure and content for closer reconstructions.
- A paired encoder-decoder training loop lets researchers use images never actually scanned in a person.
- About 70% of training data came from such unpaired, synthetically modeled brain-image examples.
Source: technologyreview.com — Jessica Hamzelou, 2026-10-01
Published there as: “An AI “mind-reading” tool can reconstruct what you’re looking at based on a brain scan”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.