Anesthetized Brains Process Language and Predict Words — SkimNews
Get the Health newsletter
Daily health & science — research, biotech, public health, the studies worth knowing. Free.
- Baylor College of Medicine researchers found that hippocampal neurons in fully anesthetized patients distinguished nouns, verbs, and adjectives in recorded stories and carried signals that anticipated upcoming words, according to a study published in Nature.
- Neuropixels probes recorded activity from hundreds of individual hippocampal neurons during epilepsy surgery — the first time this region was examined this way in unconscious patients, the researchers said.
- Hippocampal neurons showed signs of neural plasticity under anesthesia, improving their response to unexpected tones amid a series of repeating sounds as the experiment continued.
- Dr. Sameer Sheth, the study's principal investigator and a neurosurgeon at Baylor St. Luke's Medical Center, said "the brain is far more active and capable during unconsciousness than previously thought."
- Dr. Vigi Katlowitz, first author and a neurosurgery resident, said the findings open the door to using these signals for speech prosthetics in patients who lost the ability to speak from stroke or injury.
- The researchers drew a parallel to large language models, noting both biological brains and AI systems use patterns in language to predict what comes next, though they operate in fundamentally different ways.
Why it matters: For neuroscience, the finding reframes what consciousness actually contributes — sophisticated language processing and next-word prediction appear to run without subjective awareness. Practically, the Baylor team flagged brain-computer interfaces as a downstream application, making hippocampal language signals a potential target for prosthetics that decode speech from neural activity in patients impaired by neurological damage.
Ask SkimNews
