A landmark study published by scientists at the University of California San Diego is redefining science's understanding of the way learning takes place. The findings, published in the journal Nature and supported by the National Institutes of Health and U.S. National Science Foundation, provide novel insights on how brain wiring changes during learning periods, offering a path to new therapies and technologies that aid neurological disorders.
For many years, neuroscientists have isolated the brain's primary motor cortex (M1), an area in the frontal lobe region, as a hub for sending out signals related to complex movements during episodes of learning. More recently, the motor thalamus, located in the center of the brain, has been implicated as an area that influences M1 during motor learning functions.
But even with such advancements, evidence was lacking on how this learning process unfolds, mainly due to the complex nature of monitoring the interactions of cells across brain areas.
A research team led by Professor Takaki Komiyama's laboratory used powerful neurobiological research techniques to describe these mechanisms in mice for the first time. Using high-tech imaging and a novel data analysis method, the researchers identified the thalamocortical pathway, a communication bridge between the thalamus and the cortex, as the key area that is modified during learning.
Beyond identifying the main pathway, the researchers found that links between regions physically change during learning. Motor learning does much more than adjust activity levels, it sculpts the circuit's wiring, refining the conversation between the thalamus and cortex at a cellular level.
"Our findings show that learning goes beyond local changes — it reshapes the communication between brain regions, making it faster, stronger and more precise," said Assaf Ramot, the study's lead author and a postdoctoral scholar in the Komiyama Lab. "Learning doesn't just change what the brain does — it changes how the brain is wired to do it."
The study, during which mice learned specific movements, revealed that learning causes a focused reorganization of the thalamus and cortex interaction. During times of learning, the thalamus was found to activate M1 neurons to encode the learned movement and to halt the activation of neurons not involved with the movement being learned.
"During learning, these parallel and precise changes are generated by the thalamus activating a specific subset of M1 neurons, which then activate other M1 neurons to generate a learned activity pattern," said Komiyama, a professor in the Departments of Neurobiology (School of Biological Sciences) and Neurosciences (School of Medicine), with appointments in the Halıcıoğlu Data Science Institute (School of Computing, Information and Data Sciences) and Kavli Institute for Brain and Mind.
To bring the activity of specific neurons into focus — a key insight of the study — the researchers developed a novel analytical method called ShaReD (Shared Representation Discovery) with Neurobiology Assistant Professor Marcus Benna and graduate student Felix Taschbach, study coauthors.
According to Taschbach, who spearheaded development of the data analysis procedure, identifying behaviors that are commonly encoded across different subjects presents a significant challenge because behaviors and their neural representations can vary substantially between animals. To address this issue, the researchers developed ShaReD, which identifies a single shared behavioral representation that correlates with neural activity across different subjects, allowing them to map subtle behavioral features to the activity of different neurons in each animal.
Existing methods typically enforce artificial alignment to reduce individual variability — similar to requiring everyone to follow exactly the same route to a destination. In contrast, ShaReD functions more like identifying which landmarks consistently help travelers navigate, regardless of their specific route choices. The ShaReD method was critical to the study's findings.
"This new method allows us to combine data from multiple experiments to make detailed discoveries that would not have been possible using only the limited number of relevant neurons recorded in an individual brain," said Benna, a computational neuroscientist and co-corresponding author of this study.
The new study is the second recently led by the Komiyama lab that illuminates how our brains learn. In April, William Wright, Nathan Hedrick and Komiyama published a study in Science that describes the multiple rules that neurons follow during episodes of learning, with synapses in different regions following different rules.
With the Nature study's findings, the researchers further science's understanding of the learning process with a new comprehensive model of how the neural circuits underlying learned movements emerge during learning. The new information also offers hope for those who suffer from neurological disorders.
"The study shows that learning isn't just repetition," said Ramot. "It's about your brain literally rewiring itself in a targeted way. Whether you're learning a new skill, recovering from a stroke or using a neuroprosthetic, understanding how brain regions reorganize their communication helps us design better therapies and technologies that work with the brain's natural learning mechanisms."
The paper is dedicated to the memory of An Wu, an assistant project scientist in Komiyama's lab who tragically died in a 2023 Montreal building fire. She is remembered as a brilliant neuroscientist who elevated the many lives she touched.
