A recent study published in the journal Cerebral Cortex suggests that adults who are better at math tend to rely less on the brain areas associated with physical movement when processing numbers. These findings provide evidence that as people develop advanced math skills, their brains shift toward more automatic and abstract ways of thinking about numbers.
Number processing relies on multiple mental formats. Scientists describe a verbal format for number words, a visual format for written digits, and a semantic format for the actual meaning or quantity. In recent years, scientists have proposed that an embodied format also exists, where physical experiences like counting on fingers help shape how the brain understands quantities.
To explore how these mental formats interact at different life stages, the authors aimed to understand how physical representations of numbers relate to formal math competence in both children and adults. Xueying Ren, a postdoctoral scholar in psychology and human development at Vanderbilt University’s Peabody College, explained the motivation behind the research.
“While we know that number processing is foundational for mathematical competence, the underlying brain mechanisms have remained heavily debated,” Ren said. “Theories of embodied cognition suggest that our abstract understanding of math is initially rooted in physical, sensory, and motor experiences, like counting on our fingers in early childhood. We wanted to look closely at both children and adults using fMRI to see how the brain’s sensorimotor regions are recruited during number processing, and how that neural engagement actually tracks with real-world math abilities across different stages of development.”
Functional magnetic resonance imaging, or fMRI, is a type of brain scan that measures blood flow to detect active brain areas. To conduct the study, the researchers collected imaging data from 104 adults with an average age of about 23 years. They also tested 88 fourth-grade children with an average age of nearly 10 years.
While inside the scanner, participants completed a number comparison task and a sound-based task. During the tasks, participants looked at two types of images on a screen. One type was symbolic Arabic numerals, like the visual number four. The other type was embodied representations, which consisted of color photographs of human hands holding up different numbers of fingers.
In the number task, participants had to decide if the number shown on the screen was larger or smaller than a specific target number. The participants pushed buttons to answer as quickly as possible. In the sound-based phonological task, participants had to judge if the starting sound of the number matched the starting sound of a cartoon object, like a fan or a sun.
The researchers also measured the participants’ overall math abilities outside the scanner using a standardized assessment called the Woodcock-Johnson Third Edition Tests of Achievement. This assessment included three specific math tests. The Calculation subtest measured basic computation skills across various types of math. The Math Fluency subtest measured how many simple arithmetic problems the participant could solve in three minutes.
Finally, the Applied Problems subtest measured the ability to analyze and solve spoken word problems. To ensure the brain activity was specifically linked to math, the scientists also tested basic reading skills. They used two reading subtests to measure letter identification and the ability to sound out unfamiliar words. By comparing the math scores and reading scores against the brain scans, the researchers could isolate the specific neural networks responsible for numerical cognition.
When looking at the brain scans, the scientists observed that adults engaged a widespread network of brain regions when processing numbers compared to processing sounds. These areas included the occipital, temporal, parietal, and insular regions of the brain. Children activated a smaller, more localized set of brain areas during the same tasks.
“What surprised us most was the dramatic shift in how the brain is recruited for number processing as we grow up,” Ren told PsyPost. “When looking at the overall brain maps, adults engage a much wider, more expansive network of regions across the brain compared to children.”
“Yet, within that broader adult network, individuals with higher math proficiency actually showed reduced activation across sensorimotor and attentional areas, a pattern completely absent in children. This reveals a fascinating paradox: as the brain gains years of experience, actual math proficiency becomes marked not by working the brain harder, but by a transition toward incredible neural efficiency and automaticity.”
In adults, lower activity in the somatosensory and motor cortices during the number task was associated with higher math skills. These cortices are the parts of the brain responsible for processing physical touch sensations and voluntary body movements. The authors also found that adults with better math skills showed reduced activation in the right insular cortex.
The insular cortex is a brain region that detects highly demanding cognitive tasks and signals the brain to apply more effort. Lower activation in this area suggests that mathematically proficient adults perceive basic number tasks as less mentally taxing. These adults operate on a sort of cognitive autopilot, requiring less conscious effort to process quantities.
“The core takeaway is that proficient math performance in adulthood is characterized by a fundamental neural shift toward efficiency and automaticity,” Ren said. “While children rely heavily on basic quantity processing and sensory grounding to make sense of numbers, adults with higher math skills actually show reduced activation in sensorimotor and attentional brain areas. This suggests that as we gain experience, higher math proficiency isn’t about working the brain harder, but rather about transitioning away from a physical ‘scaffold’ to more abstract, automated mental representations.”
The scientists also examined the left intraparietal sulcus, a brain region known for handling numerical quantities. For adults, less activity in this region correlated with better math performance, supporting the neural efficiency hypothesis. For children, the exact opposite was true. Higher activity in the left intraparietal sulcus predicted better math scores in the fourth graders, indicating that young learners still rely heavily on basic quantity processing to succeed in math.
None of these brain activity patterns correlated with the participants’ reading scores. This lack of correlation provides evidence that the reduced reliance on motor and quantity-processing regions is highly specific to mathematical skills. It does not simply reflect general intelligence or advanced reading comprehension.
A potential misinterpretation of these findings is that physical methods like finger counting are unhelpful for learning math. The authors note that physical representations often serve as a necessary scaffold for young learners as they grasp basic number concepts.
“An important caveat is that our findings do not imply that sensorimotor strategies, like a child using their fingers to count, are bad or should be abandoned early,” Ren said. “Sensorimotor experiences serve as an essential, adaptive scaffold when we first learn mathematical concepts. The key is that this relationship changes over time; while physical grounding is vital for early learning, our long-term math proficiency relies on the brain eventually learning to offload that effortful physical processing to achieve automaticity.”
A limitation of the study is that the data for adults and children were collected using two different brain scanners. This was partially due to scheduling constraints caused by the global pandemic. While scanner differences usually affect overall signal strength rather than specific behavioral correlations, future studies should use consistent equipment to rule out any potential interference.
“Because this study looked at separate groups of adults and fourth graders, one important next step is to utilize longitudinal designs to trace these neural transitions within the same individuals over time,” Ren said. “It would be interesting and critical to pinpoint exactly when and how the brain shifts away from its reliance on sensorimotor scaffolding. Ultimately, understanding this developmental trajectory can help us design better, more tailored educational strategies and interventions for individuals who face persistent challenges in learning math.”
These findings highlight a broader trend in brain development and cognition. “Overall, I think this study beautifully illustrates a broader principle in cognitive neuroscience: learning and high expertise are often marked by the brain doing less work, adaptively reducing activity as effortful control gives way to smooth automaticity,” Ren said.
The study, “Reduced dependence on sensorimotor processing in the brain is associated with higher math skills in adults,” was authored by Xueying Ren, Marc N. Coutanche, Julie A. Fiez, and Melissa E. Libertus.
