I hummed the Star Wars theme on a loop when I came out of anesthesia. Many of them, who have been under anesthesia, end up doing some comic acts that lighten the mood for their kin around them. While they might think it’s you being far from awareness as the anesthesia wanes, a new study at Baylor College of Medicine shows that the human brain can continue to perform advanced language tasks even when a person is fully unconscious under general anesthesia.
The research once again challenges our concept of consciousness and cognition and makes us truly wonder how clearly defined the boundaries between the conscious and the unconscious are. The scientists offer new insights that could mold future research on memory, language, and brain-computer interfaces. They also suggest that the unconscious brain has more in common with large language models (LLMs) than we can fathom.
Recording Brain Activity under Anesthesia
Dr. Sameer Sheth, professor and Cullen Foundation Endowed Chair of neurosurgery and a McNair Scholar at Baylor, and his colleagues recorded the activity of individual neurons in the hippocampus to see what the unconscious brain is capable of. The recordings were made while patients were under general anesthesia during epilepsy surgery, using Neuropixels probes—silicon electrodes that record the electrical activity of individual neurons.
The overall process used in the first-of-its-kind study of the hippocampus allowed the scientists to study this part of the brain directly. They observed how the brain responded to sounds and language, even when the patients were unconscious.
The Brain Kept Processing Language
The first experiment exposed patients to a series of repeating tones, with unexpected sounds added to the mix. The researchers found that neurons in the hippocampus reliably detected the unusual tones. However, the real deal that left the researchers in awe was how the accuracy of detecting out-of-place tones improved over time. Even when unconscious, the brain was learning, and neural plasticity—the brain’s lifelong ability to adapt and reorganize—still occurred during anesthesia.
The researchers shifted gears, upping the complexity by playing short stories while continuing to record brain activity. The hippocampus showed clear signs of real-time language processing, as neural activity patterns revealed that the brain could distinguish among parts of speech, such as nouns, verbs, and adjectives.
What the scientists found next rocked what we believe to be true about consciousness and AI.
The Unconscious Brain Anticipates
The researchers found that the brain could anticipate what comes next in a story, predicting words before they were spoken. This kind of predictive coding has so far been associated with an attentive, awake brain, but it seems to occur quite regularly in an unconscious state.
The findings imply that important cognitive abilities, including language comprehension and prediction, are independent of conscious awareness. Instead of seeing consciousness as a crowded activity within a single area, such as the hippocampus, the trait might arise out of communication across multiple brain regions. The researchers also found matches between the brain’s predictive behavior and artificial intelligence (AI), especially the ability of large language models to generate text by anticipating the next word. The hippocampus apparently makes similar predictions during language processing, and these shared features could help scientists better understand both human and artificial intelligence.
The work may also be pivotal in the design of future communication technologies, such as speech prosthetics for people who have lost the ability to speak. “Can we use these signals to deploy and run a speech prosthetic for some of the parts of the brain that are damaged by stroke or injury? These are questions that we can now consider in relation to this part of the brain,” said Dr. Vigi Katlowitz, first author and a neurosurgery resident with Baylor.
Is an Unconscious Brain “AI-like”?
The scope of the research goes beyond explaining human awareness, prompting us to wonder whether AI is conscious too. At first glance, it seems that AI understands us and has a mind of its own, capable of decoding human emotions. However, it’s just an impeccable assumption of what might come next. In a sense, AI has tamed randomness—the chaos within probability.
Put simply, AI is unconscious because its actions result from pattern matching across vast amounts of data, allowing it to compress reality into the most probable next steps. We need an intrinsic experience to feel, to be conscious of something, while AI takes the best route to satisfy what the user’s consciousness would likely demand.
An unconscious brain is in an “AI-like” state, running predictive mechanisms on autopilot without subjective experience. Further research will examine other unconscious states, such as sleep or coma, and will focus on one particular region of the brain. However, it is still unknown how these processes occur throughout the brain.
Copyright @smorescience. All rights reserved. Do not copy, cite, publish, or distribute this content without permission.
