How does the human brain learn to read, to calculate, to think mathematically? What happens when a piece of information becomes conscious?
Professor at the Collège de France, Chair of Experimental Cognitive Psychology
Scientific Director of NeuroSpin (CEA, Saclay)
President of the French Scientific Council for Education
Anthropic researchers have found, inside the language model Claude, a "J-space" with many analogies to the global neuronal workspace of human consciousness. Stanislas Dehaene and Lionel Naccache examine the parallels and their limits, and propose new tests drawn from cognitive neuroscience.
Read the commentary (PDF)The brain network that responds to mathematics is organised from the very start of schooling, then expands progressively over the first school years.
Stanislas Dehaene's latest book, on how Homo sapiens invented geometry, receives the award.
Stanislas Dehaene's new book on the origins of geometry will be published in English by MIT Press in 2027.
A mathematician turned explorer of the human brain.
Trained as a mathematician at the École normale supérieure, Stanislas Dehaene studies the brain mechanisms of the cognitive functions that make our species unique: calculation, reading, language and consciousness. This research combines experimental psychology, brain imaging (fMRI, MEG, EEG), intracranial recordings and modelling, comparing adults, children, infants, non-human primates, and literate and non-literate people.
Professor Dehaene founded the INSERM-CEA Cognitive Neuroimaging Unit in 2002. Since 2018, Stanislas Dehaene has chaired the French Scientific Council for Education, which brings research evidence to bear on education policy.
Six threads of a single programme: understanding how the circuits, neurons and synapses of the human brain encode objects of thought as complex as numbers, letters, words and sentences.
We are born with an intuition of quantity that we share with many animals. Stanislas Dehaene showed that numbers automatically evoke space — a "mental number line" — and that the intraparietal sulcus, in both hemispheres, lies at the heart of this number sense. Exact and approximate calculation rely on partly distinct circuits. The number sense is present in infants and in adults without schooling or number words, such as the Munduruku of Amazonia.
Today Recent work suggests that number concepts are compositional, built by combining a few primitives in a "language of thought" shared with geometry. It also shows how strongly school shapes them: a maths gender gap emerges within the first months of first grade.
In 1993, a mathematical model proposed by Stanislas Dehaene and Jean-Pierre Changeux predicted the existence of "number neurons", each tuned to a preferred quantity, with Gaussian tuning curves on a logarithmic scale. Such neurons were later recorded in macaques; Professor Dehaene then found the same code in humans with fMRI, where a written or spoken numeral alone is enough to activate it.
Today The search for neural codes now continues at the level of single neurons in humans: in the hippocampus, neurons reactivate a concept even when it is referred to by a mere pronoun. And how often we use each number reveals its compositional structure.
Even with the most abstract concepts — topology, algebra, analysis — professional mathematicians engage the parietal and prefrontal circuits of the number sense, distinct from those of language. Stanislas Dehaene's recent work shows that a sense of geometric regularity, present without schooling, sets humans apart from other primates: a clue to the origins of our singularity.
Today Brain imaging now reveals a "geometric regularity effect" in the human brain, in adults and children alike. And the brain's math network is already organised when children start school, then expands over the first school years.
With Laurent Cohen, Stanislas Dehaene identified the "visual word form area", a left occipito-temporal region that becomes specialised for letters as children learn to read. Comparing literate and non-literate adults revealed how deeply reading changes the brain. Hence the theory of "neuronal recycling": cultural inventions such as writing repurpose pre-existing circuits — our letters reuse shapes the visual system already knows how to recognise.
Today Artificial neural networks now help "crack" the neural code for word recognition. And these findings are reaching the classroom: the Kalulu app, designed from this research, improves first graders' reading in Brazil.
Using masking, which renders a word or digit invisible, Stanislas Dehaene showed that the brain processes a great deal of information without awareness, including its meaning. When information becomes conscious, a sudden, global "ignition" occurs, notably involving prefrontal cortex. With Jean-Pierre Changeux, Professor Dehaene formalised the "global neuronal workspace" theory, whose signatures are now used to assess consciousness in non-communicating patients.
Today In 2025, a large "adversarial" collaboration pitted the global workspace theory against its main rival, and a cortex-wide model explained conscious access as a dynamical bifurcation. Professor Dehaene is now examining whether artificial intelligence systems possess an analogous workspace.
With Ghislaine Dehaene-Lambertz, Stanislas Dehaene showed that by two months of age, the infant brain already has language circuits organised and lateralised much as in adults. In adults, Professor Dehaene's work describes how the brain encodes the tree structure of sentences. These findings feed a theory of learning — attention, active engagement, error feedback, consolidation — put to work for schools.
Today Comparing infants and macaque monkeys recently revealed brain areas for "reversible symbolic reference", a possible singularity of the human brain. And following children through their first school years shows how education and music training shape abstract thinking.
Full publication list: Google Scholar, UNICOG
A selection of papers published in 2025–2026.
Full list: Google Scholar, UNICOG
Books for every reader, translated into more than fifteen languages.
How We Invented Geometry
In the Lascaux cave, beneath the painting of a great stag, someone traced a humble rectangle. Starting from this discreet figure, Stanislas Dehaene shows that geometry is as old as humanity itself, and that even the simplest shape reveals what is unique about the human mind: the ability to think in symbols and to build ever more abstract concepts upon them. The English edition of the book that won the 2026 Prix Bourrienne will be published by MIT Press in 2027.
The Lascaux Rectangle: How We Invented GeometryOn the origins of geometry, from prehistoric markings to the brain signatures of human singularity.
Prix Bourrienne 2026
Une idée dans la têteForty short, joyful nuggets about the brain and learning.
The 40 episodes of the radio series (in French) are still available on France Inter's website.
Seeing the MindOne hundred of the most spectacular images of neuroscience, and what they reveal about our neuronal selves.
How We LearnThe four pillars of learning, and what a child's brain still does better than artificial intelligence.
Book of the year, French Society of Neurology
Other translations: Turkish, Portuguese…
Consciousness and the BrainWhat distinguishes, in the brain, a conscious thought from unconscious processing.
Prix LIRE, best science book
Other translations: Chinese…
Reading in the BrainHow a primate brain learns to read, and what that means for teaching.
Grand Prix RTL-Lire
Other translations: Chinese, Croatian, Russian, Korean…
Steps toward a science of mental lifeThe inaugural lecture at the Collège de France: in search of the universal laws of mental life, from psychophysics to brain imaging.
See the inaugural lecture on the Collège de France website (in French)
The Number SenseThe founding book on our intuition of numbers, in animals, infants and adults. Revised and expanded edition 2010–2011.
Jean Rostand Prize
Other translations: Chinese, Korean
Since 2005, a public lecture course every year. All sessions are available as video and podcast on the Collège de France website (mostly in French): click on a course to watch it.
The whole chair: college-de-france.fr
Bringing research from the laboratory to the classroom, and sharing science with everyone.
Since 2018, Stanislas Dehaene has chaired the French Scientific Council for Education, which advises the Ministry on scientific approaches to learning and teaching.
With the non-profit Excello Education: Kalulu for learning to read, Pionniers des maths for primary-school mathematics, and evidence-based tools.
Each year, a free public course, available as video and podcast (in French): see the complete list of courses.
All Collège de France lectures: college-de-france.fr
Academies, prizes and national honours.
Recipient of the 2014 Brain Prize with Giacomo Rizzolatti and Trevor Robbins, Stanislas Dehaene is a member of eleven academies, including the French Academy of Sciences, the US National Academy of Sciences and the Royal Society.
University of Lisbon (2011), HEC Paris (2013), Université catholique de Louvain (2014), Autonomous University of Santo Domingo (2018), Tel Aviv University (2020), Ricardo Palma University, Lima (2022). Honorary Professor, East China Normal University, Shanghai (2010).
The Cognitive Neuroimaging Unit, at NeuroSpin.
Stanislas Dehaene directs the INSERM-CEA Cognitive Neuroimaging Unit (UNICOG) at NeuroSpin in Saclay, France's ultra-high-field brain imaging centre. Current research, supported in particular by a European Research Council Advanced Grant ("MathBrain"), focuses on the languages of the brain: mathematics, geometry and natural language, from neural populations to behaviour.