Between kindergarten and CE1, how do children’s brains reorganize themselves to structure mathematics learning? To find out, researchers from the CEA, the CNRS and the Collège de France carried out a large brain imaging study by following children from the end of kindergarten to CE1. They showed that the architecture dedicated to mathematics was already in place before entry into first grade, and that it then continued to be refined and enriched over time. Communiqué.
The mathematics network already in place at 5 years old
To carry out this longitudinal study, researchers from NeuroSpin at the CEA followed around fifty children for three years, from the main kindergarten section to CE1. Each year, the children underwent a 3 Tesla functional MRI session – a brain imaging technique which indirectly measures brain activity in real time – while they were made to listen to mathematical sentences (arithmetic and geometry), general knowledge or social.
The observation is striking: from the age of 5-6, before a more formal introduction to learning, children already activate the same brain regions as adults when faced with mathematical statements. Better yet, within this network, the activity already differs depending on whether it concerns arithmetic or geometry sentences.
Between 5 and 9 years old, a network that is being shaped
So what happens in the brain between kindergarten and CE1? The team identified three complementary mechanisms at work.
First, areas that are already active become more responsive : the measured signal is increasingly stronger with mathematical stimuli.
Second, new voxels (volume of neurons), within these same regions, are progressively recruited, slightly extending the cortical space of the network.
Third, when a child masters a concept well, brain activity decreases for this specific concept: the brain automates and mobilizes fewer neural resources for what it already knows how to do effortlessly.
Finally, over the three years, the “mental map” of mathematical concepts in the brain becomes richer and more varied: each notion occupies an increasingly distinct place in neuronal space, the brain exploiting directions of this space that were previously little or not used. This mechanism explains how an increasing number of concepts can be encoded within a relatively stable cortical space.
« Ce que cette étude apporte, c'est qu'au-delà d'une simple observation de corrélats cérébraux, on comprend mieux les mécanismes qui sous-tendent l'apprentissage des mathématiques, et on peut désormais les modéliser.On a ainsi pu montrer que plusieurs mécanismes sont à l'œuvre : durant les premières années d'école ce réseau s'amplifie là où c'est nécessaire, s'automatise quand c'est possible, et s'enrichit à mesure que l'enfant apprend et comprend de nouveaux concepts », précise Théo Morfoisse, premier auteur de cette étude.
The brain circuit for mathematics exists from kindergarten and is refined at school, July 2026, CNRS
The study was carried out at NeuroSpin, within the Cognitive Neuroimaging Unit (Unicog, CEA/Inserm/Université Paris-Saclay) in collaboration with the Collège de France, under the direction of Ghislaine Dehaene-Lambertz and Stanislas Dehaene. It is published in Proceedings of the National Academy of Sciences (PNAS).
