For the first time, researchers have obtained direct evidence that the human brain can perform complex speech processing even during deep general anesthesia. In the study, neuroscientists recorded the activity of individual neurons in the hippocampus of seven patients undergoing surgery for severe epilepsy. While the participants were under propofol anesthesia, they were played sounds and podcasts, and brain activity was recorded with Neuropixels microelectrodes.
Patients were played sequences of identical sounds with rare unusual signals. As the experiment progressed, the neural response became more precise, and computer modeling showed that the brain continued to learn even during anesthesia. This indicates that neuroplasticity was preserved — the ability of the nervous system to reorganize its connections in response to new information.
In the second part of the experiment, the participants listened to recordings of ordinary conversational speech. Neural activity depended on how unexpected the next word was in context. Machine-learning algorithms were able to determine from these signals whether the word heard referred to objects, places or emotions, and also to distinguish nouns from other parts of speech. Moreover, hippocampal activity made it possible to statistically predict the semantic category of the word the speaker would pronounce about a second later. In other words, the brain continued to build expectations about how the speech would develop.
For a long time, it was believed that such information processing was impossible without consciousness. Previous studies had shown only the preservation of primary sensory areas that respond to sound or light. The new work is the first to demonstrate that much more complex processes can continue in the deep structures of the brain, which are responsible for forming memory, meaning and context. This significantly expands our understanding of how human thought is organized.
