🇷🇺 Russian, 6 November 2026, Visual Cortical Stimulation
Neural Network Improves Accuracy of Visual Cortical Stimulation
Researchers use deep learning to improve the accuracy of neural responses to visual cortical stimulation in a blind patient.
A team of researchers from the University of California, Santa Barbara, ETH Zurich, and the University of Miguel Hernandez used deep learning to improve the accuracy of neural responses to visual cortical stimulation in a blind patient. The study demonstrates that artificial intelligence can increase the accuracy of neural responses, requiring less electrical current compared to traditional methods. The researchers trained a neural network to predict the brain’s responses to different stimulation patterns, taking into account the patient’s resting brain activity.
The study involved a 27-year-old man who had lost his vision due to a head injury. A temporary implant containing 96 electrodes was inserted into his brain, stimulating the visual cortex and causing phosphenes - perceptions of light in the form of flashes or figures. The neural network was able to predict the patient’s visual perceptions better than the parameters of the electrical signals themselves.
The study’s key finding is that recorded brain activity predicted visual perceptions better than the electrical signal parameters. This suggests that the brain is not a pixelated screen, but rather a complex system that interacts with electrical signals in a non-linear way.