Chinese researchers have developed a new method for decoding visible images from the human visual cortex. With its help, you can find out by brain activity which letter or number is shown to the participant in the experiment. The method is based on deep learning of neural networks using functional magnetic resonance imaging (fMRI) data. According to the authors, the new approach allows for much more accurate image reproduction than other known methods. A preprint of the study was published on the arXiv.org server, and MIT Technology Review briefly reported on it.