Salk Institute neuroscientists announced on July 21 that traveling brain waves act as a computational engine in the visual cortex, according to a review article published in Neuron. The research synthesizes physiological and computational information about neural traveling waves, concluding that these waves enable the visual cortex—and likely other brain areas—to build representations of the external world. This process supports prediction, reconstruction, and perception.
John Reynolds, PhD, was the first to identify traveling brain waves in the visual systems of awake animals in 2020. His lab found that these waves directly correlated with whether animals could perceive an object directly in front of them. This finding offers insight into everyday experiences such as searching for visible objects like keys.
After establishing that traveling brain waves exist in awake animals and influence perception at specific moments, Reynolds and his team focused on understanding their function. They propose that neural traveling waves allow the visual cortex to modulate perception moment by moment, transform recent sensory inputs into internal representations, generate short-term predictions about the external world, and store or replay patterns representing memories over time.
The researchers suggest these neural brain waves are more than just electrical noise. According to their findings, "the neural connections that generate these waves aren't simply relaying signals; rather, they change their physiology ('synaptic weights') to reflect the outside world." Each sensory experience alters these connections and builds circuitry used by the brain for constructing internal models of reality.
Reynolds said, "This is, in a meaningful sense, analogous to what large language models like ChatGPT do. They learn statistical structure from language and use that knowledge to generate meaningful and appropriately structured text that reflects the patterns of language. The brain may be doing something functionally similar—a biological generative model built from the ground up by experience." He added that each time sensory input is received by the brain, it must decide what it is most likely sensing at any given moment.
The paper proposes that regularities from physical laws and environmental context are encoded within networks of synapses through wave generation. These mechanisms help infer causes behind sensory input and construct an internal model of reality.
