Neural Traveling Waves Identified as Key Mechanism for Visual Processing

Salk Institute neuroscientists have proposed that neural traveling waves may serve as a computational engine in the visual cortex, based on recent research. These researchers combined physiological and computational studies to advance the understanding of how these brain waves operate, suggesting they assist in creating internal representations of the external world, thereby aiding in perception, memory reconstruction, and anticipation of future events. Their findings were published in the journal Neuron on July 21, 2026.

The term “traveling brain waves” was first introduced by Salk neuroscientist John Reynolds, PhD, in 2020, when he identified these waves in the visual systems of awake animals. His subsequent research demonstrated a direct link between these waves and an animal’s capability to notice objects, shedding light on why individuals may overlook visible items, such as a set of keys, at certain moments.

Reynolds highlighted the significance of their latest findings: “This paper lays out, for the first time in a single integrated framework, what the brain can actually compute by virtue of having this recurrent wave-generating circuitry.”

The research focused on four potential functions of neural traveling waves. These include moment-to-moment adjustments in perception, the transformation of sensory information into internal representations, the generation of short-term predictions, and the preservation and replay of memories associated with past events. The results indicate that these traveling waves play a crucial role in how the brain interprets incoming sensory data rather than merely representing random background electrical activity.

Under this new framework, neural waves are viewed as more than mere electrical noise within the nervous system. The neural connections generating these waves possess the ability to modify their physiology in reaction to sensory experiences, encapsulating learned information. According to Reynolds, these processes resemble what large language models like ChatGPT do: they learn statistical structures from data and generate coherent outputs based on that knowledge.

The brain’s challenge when processing sensory information lies in identifying the most likely stimuli present in a complex environment. This research suggests that the brain learns to recognize recurring patterns and organizes them within synaptic networks. These networks subsequently produce traveling waves that enable the brain to discern likely sources of sensory information, thereby crafting an internal model of the external world.

The study’s co-authors include Lyle Muller of UT Dallas and Fields Institute, Alexandra Busch of Fields Institute and Western University, and Zachary Davis of the University of Utah. The research was funded by the National Institutes of Health and several Canadian organizations and institutes.