The Brain Keeps Its Two Greatest Tricks in Different Rooms. Even in Toddlers.

A three-year-old is watching a cartoon. On screen, a cloud character stuffs a pelican chick with increasingly dangerous animals; the pelican, visibly uncomfortable, tolerates it. The child laughs, then winces, understanding without being told that the cloud means well and the pelican is suffering, that intentions and feelings are two different things. Somewhere in the right side of the child’s brain, a network for reading other minds is humming. In the left hemisphere, another network parses the narrator’s words. What a new study finds is that these two systems, handling perhaps the most distinctly human things we do, have absolutely nothing to say to each other. They never did.

The question of how language and the capacity to understand other people’s mental states came to occupy separate territories in the brain has nagged at neuroscientists for years. One influential theory held that they probably share a common ancestor in neural terms, that the left-hemisphere machinery we use for grammar and meaning evolved from older social-processing regions on the right, and that the two systems might still be entangled in young children who are only beginning to acquire both skills. A new study from Ohio State University, published April 23 in Communications Biology, suggests that theory is wrong.

Zeynep Saygin, an associate professor of psychology at Ohio State who led the study, wanted to know whether the neural separation seen in adults had always been there, or whether it was something the brain assembled gradually through years of practice. Her team scanned 42 children aged roughly 3 to 9, once while they listened to spoken sentences and once while they watched Pixar’s Partly Cloudy without sound. Language activated networks in the left temporal lobe; mentalizing lit up corresponding regions on the right. The two patterns barely overlapped at all, not even in the superior temporal lobe where both networks converge and where, if there was ever going to be shared real estate, you would expect to find it.

“That was our first question: Are these skills distinct in both their function and location? And we see really broadly, yes,” said Kelly Hiersche, the doctoral student who led the study. “We demonstrate this for the first time in kids, extending an adult finding to development. They’re really distinct there, which is pretty cool.”

The researchers then went looking for developmental change, the kind of gradual disentangling that the shared-origin theory would predict. They compared children at different ages in their cross-sectional sample and, crucially, scanned a subset of children at two timepoints about a year apart. The idea was that if language and mentalizing start out overlapping and pull apart over time, you might catch that process in progress. They didn’t catch it. Older children showed no more separation than younger ones, and individual children showed no meaningful change between visits.

Wiring, Not Experience

This is where the study gets most interesting. The researchers also collected resting-state fMRI scans, images of the brain when it’s not doing any task in particular but still sending signals between regions. From these, they built what they call connectivity fingerprints: maps of which regions each part of the superior temporal lobe tends to communicate with at rest. The logic is that a brain region’s function is largely determined by who it talks to. Language regions in the left hemisphere chat constantly with frontal areas involved in syntax and meaning; mentalizing regions on the right stay in contact with the precuneus and temporal parietal junction, involved in thinking about others’ beliefs and intentions.

“If you observe a voxel’s connectivity, or how it talks to the rest of the brain, that’s going to give you an idea about how that voxel is going to function,” Hiersche explained. Using machine-learning models trained on these connectivity patterns, the team could predict with reasonable accuracy which parts of each child’s superior temporal lobe would respond to language versus mentalizing, without any task data at all. Critically, the fingerprints for language and the fingerprints for mentalizing were almost entirely non-overlapping, about 90 percent of the brain connections that predicted one function were significantly different from those predicting the other. And that separation was just as stark in three-year-olds as in adults.

“We knew these regions were localized in different parts of the brain, but also showed that there’s nothing in how they communicate with the rest of the brain that indicates that they were at any point overlapping,” Saygin said. The connectivity fingerprints were also remarkably stable within individual children over time: a child’s resting-state connectivity at age five predicted their language and mentalizing responses a year later just as well as their concurrent connectivity did, suggesting these patterns aren’t something the brain assembles through use but something that was perhaps already in place well before the children entered the scanner.

Two Kinds of Human

There is a reason this matters beyond its interest to developmental neuroscientists. Language and theory of mind are the two cognitive capacities most often cited as foundations of human social life. They develop together in early childhood and are behaviourally intertwined in ways that have led many researchers to suspect they might share neural origins. Studies of clinical populations, people with aphasia who lose language but retain mentalizing ability, or autistic individuals who show atypical theory of mind alongside sometimes intact language, have long suggested the two can be dissociated. But whether the separation is built in from the beginning or constructed over time has been harder to answer.

The Ohio State study is the first to examine both systems together in children young enough that their skills in both domains are still visibly emerging. Three-year-olds often fail standard tests of false belief understanding, the classic measure of theory of mind; their language is still maturing too. And yet their brains already run the two systems in separate, non-communicating hardware. “It seems that these processors that help us mentalize and that help us speak and understand were dissociated very, very early in the evolutionary process, such that we can’t even see traces of overlap right now in human development,” Saygin said.

When Adults Break the Pattern

There is one intriguing wrinkle. While children showed cleanly separate connectivity fingerprints for language and mentalizing, adults showed a small but statistically significant degree of similarity between the two. The language network and the mentalizing network, in mature brains, have apparently begun talking to some of the same regions. Saygin suspects this reflects something functional: as adults constantly use language to communicate about mental states, gossip, negotiation, storytelling, the two systems may become more coordinated without becoming fused. “In adults, the mentalizing theory of mind network starts to talk to slightly similar regions as the language areas,” she said. The same process may be ongoing in children as their skills develop, though the underlying architecture keeps the systems apart.

The results push back against theories that frame language as a social-cognitive technology bootstrapped from older mind-reading circuitry. Instead they point toward something more like two parallel evolutionary inventions, distinct from each other even before experience could shape them, each solving a different piece of the problem of living in a world full of other people. “It’s a fundamental question humans ask themselves: ‘What is it that makes us human? How does human cognition emerge?’ I think this sheds some light on that,” Saygin said. Whether the same separation holds in infants younger than three, or in children before they can follow scanner instructions, remains genuinely open.

Hiersche, Osher & Saygin, Communications Biology, April 2026

Frequently Asked Questions

What is theory of mind and why is it compared to language?

Theory of mind is the ability to understand that other people have beliefs, desires, and intentions that may differ from your own. It’s sometimes called mentalizing. Both theory of mind and language are considered uniquely human cognitive abilities that emerge in early childhood, develop together, and are essential for social life, which is why scientists have long wondered whether they might share common brain origins. This study suggests they don’t.

How young were the children in this study, and why does that matter?

The youngest children scanned were around 3 years old. That’s significant because three-year-olds are still actively developing both language and theory of mind, often failing standard tests of false belief understanding and still maturing linguistically. Finding that their brains already keep the two systems in entirely separate, non-overlapping regions suggests the separation isn’t something the brain builds up through practice but something that may be present from very early on.

Is it true that language and mind-reading are processed on opposite sides of the brain?

Broadly, yes. Language processing is predominantly left-hemisphere dominant, anchored in the left superior temporal lobe, while theory of mind tends to be right-lateralized, relying on corresponding regions in the right hemisphere as well as areas like the temporal parietal junction and precuneus. The two networks sit in roughly mirror-image positions across the two hemispheres, which had led some researchers to suspect they might be evolutionary relatives. This study found no evidence to support that idea even at the level of how the regions communicate with the rest of the brain.

What does a connectivity fingerprint actually tell scientists?

A connectivity fingerprint is essentially a profile of which other brain regions a given area tends to send and receive signals from during rest. Decades of research have shown that a brain region’s function is largely shaped by its connections: regions that talk to visual processing areas tend to do visual tasks, regions networked with memory areas tend to support memory, and so on. In this study, the fingerprints for language regions and mentalizing regions were almost entirely distinct, meaning the wiring that sets these systems up appears to be different from the very start.

Could this research change how we understand language disorders or autism?

Potentially. If language and mentalizing are truly separate from the earliest stages of brain development, it raises questions about why the two so often develop in tandem behaviourally and why difficulties in one are associated with difficulties in the other. It’s possible the link is mediated by experience and education rather than shared neural substrate. For conditions like autism, where theory of mind is often atypical, or aphasia, where language is damaged, understanding the precise boundaries of each system could help researchers better target diagnosis and intervention.