Two groups of adults were taught the same set of invented characters, one by copying them out by hand and one by typing them. When they were later shown those characters and asked only to recognise them, the handwriting group's brains fired in the regions used for producing speech and movement — as if reading a letter meant silently drawing it again.

The study we are talking about was published in 2008 in the Journal of Cognitive Neuroscience by Marieke Longcamp and colleagues at the CNRS in Marseille. The design is straightforward. Adults were shown a set of unfamiliar graphic characters they had never seen before. Half of them spent three weeks learning to produce those characters by hand, copying each one repeatedly with a pen on paper. The other half spent the same time learning to produce them on a keyboard. At the end of the training, both groups could recognise the characters equally well on the first test. But when the researchers tested them again three and six weeks later, the handwriting group was still discriminating those characters correctly from their mirror images at a much higher rate than the typing group, whose recognition had gradually decayed. Something about the act of forming the characters by hand had stayed in the recognition system.

What the fMRI showed

To understand what that something was, Longcamp’s team put a subset of the participants into a functional MRI scanner and showed them the characters again. The instruction this time was passive. Look at the characters, do not respond, do not move, do not attempt to produce them. What the scanner picked up was that the participants who had learned to write those characters by hand were activating, during purely visual recognition, brain regions that have nothing to do with vision. Specifically, the handwriting-trained participants showed elevated activation in the left Broca’s area, which is one of the classical language-production regions, and in the bilateral inferior parietal lobules, which are associated with the execution, imagery and observation of hand actions. The typing-trained group showed no such activation for the same stimuli. Their brains were treating the characters as pictures. The hand-trained brains were treating them as actions.

If you want a friendly, practical companion to all this, below is a short video on the six things a pen-and-paper journalling habit quietly does for you, with a few concrete prompts to try.

The action-perception loop the study exposed

What Longcamp’s group had documented, on the strongest current reading of that result, was a version of what cognitive neuroscience calls an action-perception coupling. When a person learns to produce a specific movement associated with a specific visual stimulus, the visual stimulus subsequently activates the motor programme for producing it, even when there is no requirement to act. The same phenomenon has been documented for tool use, musical instruments, and sign language. What Longcamp’s 2008 paper added was that the effect operates at the level of individual written characters, and that it appears within a few weeks of training. Reading, in this view, is not a pure visual process. It is a partially motor one, in which recognising a shape includes silently rehearsing the movement that would produce it.

This was already implied by an earlier fMRI study by the same group in 2003, which had shown that adult right-handers passively viewing alphabetical letters activated a region of left premotor cortex that overlapped with the region they activated when actually writing those letters. The 2008 study extended the finding by demonstrating that the coupling can be built experimentally in adults, in a matter of weeks, and that it does not develop in the same way when the training uses a keyboard.

What later work has added

The Longcamp finding has been replicated and extended by groups working with different techniques and different tasks. In 2021, a research team at the National Institute of Information and Communications Technology in Japan, led by Aya Ihara, ran a conceptually similar experiment on adult native Japanese speakers who were learning novel Indonesian words. According to the resulting paper in Frontiers in Human Neuroscience, participants who learned the words by writing them out by hand, whether with an ink pen on paper or with a digital pen on a tablet, subsequently showed a larger N400 event-related potential response to those words than participants who had learned the same words by typing. The N400 is one of the most consistently measured electrophysiological indices of semantic processing during word recognition, and a larger repetition-priming effect on N400 is generally taken to indicate that the learner has formed a more accessible semantic representation of the stimulus. The Ihara team’s finding, in other words, is what Longcamp’s finding predicts: written-by-hand stimuli come with a richer downstream response when later read.

The oscillation pattern picked up separately

A different NTNU study in 2020, published by Ose Askvik, Van der Weel and Van der Meer in Frontiers in Psychology, used high-density 256-channel EEG in twelve adults and twelve 12-year-olds to look at what their brains were doing during the act of writing itself. When adults wrote cursively with a digital pen on a touchscreen, their parietal and central brain regions showed event-related synchronised activity in the theta band, between 4 and 8 Hz, which is the frequency range that a large existing literature associates with the encoding of new information and the formation of working memory. When the same adults typed the same words on a keyboard, those regions showed the opposite pattern, event-related desynchronisation in the theta and alpha bands, which does not have that association with encoding. The 12-year-olds showed the same pattern less distinctly, consistent with a system that is still developing.

What the accumulated evidence points to

Taken together, on the accumulated evidence from these three lines of work, the underlying claim is fairly specific. When a person learns a written character by producing it with their hand, the visual representation of that character in their brain becomes coupled to the motor programme that produced it, and subsequent recognition of the character will partially reactivate that motor programme. When the character is learned by pressing a key, no such coupling is built, because the motor programme for pressing a key does not encode the character’s shape. This does not mean typing is useless, or that reading is broken in people who never learned to write by hand. It does mean that reading, in the fully wired version that literate cultures have historically produced, is a coupled visual-motor skill rather than a purely visual one. The recognition of a letter is, at the neural level, partially a silent enactment of drawing it.