Human iEEG recordings during audiovisual speech violations support and extend predictive routing models

The human brain does not merely record the world like a camera capturing a still image; it actively constructs our experience of reality. Instead of waiting for every detail to arrive through our senses, the brain constantly generates predictions about what is coming next, using these internal guesses to guide how it processes new information. When the world matches these expectations, the brain saves energy by letting the prediction do the work. But when reality clashes with what was expected, the brain must generate a signal to update its model of the world. This process, known as predictive processing, is thought to rely on specific electrical rhythms within the brain. One leading theory suggests that the brain uses a system of "routing" to manage this: it uses slower, rhythmic waves to suppress sensory input that matches predictions, while allowing faster, high-speed bursts of activity to carry the signal of any unexpected errors.

To understand how this works in the human mind, researchers turned to a unique group of volunteers: seven patients being treated for epilepsy. Because these patients had electrodes already implanted in their brains to locate the source of their seizures, scientists could record electrical activity directly from the surface of the brain with a precision that non-invasive methods cannot achieve. The team focused on two key areas: the superior temporal gyrus, a region near the ear that processes sound, and the prefrontal cortex, a higher-level area involved in planning and expectation. By observing these regions while the patients watched and listened to speech, the researchers could see exactly how the brain handles the moment when what we see and what we hear do not match.

The experiment used a classic trick of perception called the McGurk effect. In this scenario, a person on a screen mouths a syllable while a different audio track plays a conflicting sound. This creates an audiovisual incongruence that can elicit the McGurk illusion, where the brain struggles to reconcile the visual prediction of the mouth movement with the actual sound it hears. The researchers used this mismatch to test their theory. They recorded the brain's electrical power across different speed ranges, looking for the specific signatures of prediction and error. They found that when the brain received a sound that matched a visual cue, it behaved one way, but when the visual cue lied and the sound was unexpected, the brain reacted with a distinct, measurable shift in its electrical rhythm.

The results revealed a sophisticated division of labor between different parts of the brain. In the auditory region near the ear, the brain responded to the mismatch with a sharp spike in high-frequency activity, known as high gamma power. This surge appeared only when the visual prediction was wrong, acting as a local alarm bell signaling that the incoming sound did not fit the expected pattern. Interestingly, this same high-frequency spike did not appear in the higher-level frontal areas of the brain. Instead, the frontal regions showed a different kind of response. While they did not show the same high-frequency error signal, they did exhibit a strong increase in a slower, rhythmic wave called theta activity. This theta rhythm seemed to act as a broader signal, spreading across both the auditory and frontal areas to coordinate a response to the complex confusion of the mismatch.

These findings support the idea that the brain uses different electrical tools for different jobs. The study suggests that simple errors in sound are handled locally by a fast, high-frequency burst in the hearing center. However, when the error involves a complex conflict between sight and sound, the brain recruits a slower, more widespread rhythm to organize the correction across the entire network. The researchers also observed that before the sound even began, the presence of a visual cue changed the brain's state, increasing slower alpha and beta waves to prepare for the expected sound. When that expectation was violated, these preparatory waves dropped, and the error signals took over. This confirms that the brain is not a passive receiver but a dynamic system that constantly primes itself for the future, only shifting gears when reality proves the prediction wrong.

The study does not claim to have solved the entire mystery of human perception, and the small number of patients means the findings are specific to this group. However, the direct measurements provide a clear window into the timing and location of these processes. The authors suggest that the brain's predictive routing system is more flexible than previously thought, utilizing a combination of fast local signals and slower, coordinating waves to manage the constant stream of sensory information. By showing how the brain switches between these rhythms when faced with a lie from the eyes, the research offers a concrete glimpse into the neural machinery that allows us to navigate a world that is often different from what we expect.