Brain organoids classify Braille signals at 83%

Researchers used lab-grown human forebrain organoids to process electrical signals based on Braille letters. A machine-learning decoder classified signals from one organoid with 61% accuracy, rising to 83% when it combined responses from three organoids. ^[1][2]

The experiment converted tactile sensor readings into electrical stimulation delivered to organoids on a microelectrode array. Researchers then used a support vector machine to classify the organoids’ neural activity; the study also reported improved performance against artificially added noise. ^[1][2]

The results show signal processing in a limited, open-loop task—not that the organoids learned to read. The researchers said they did not test learning or long-term changes, and the study used a fixed encoding scheme and one tactile task. ^[1][2]