Neuroscientists understand much about how the human brain is organized into systems specialized for recognizing faces or scenes or for other specific cognitive functions. The questions that remain relate to how such capabilities arise. Are these networks—and the regions comprising them—already specialized at birth? Or do they develop these sensitivities over time? And how might structure influence the development of function? “This is an age-old philosophical question of how knowledge is organized,” says psychologist Daniel Dilks of Emory University. “And where does it come from? What are we born with, and what requires experience?”

Dilks and his colleagues addressed these questions in an investigation of neural connectivity in the youngest humans studied in this context to date: 30 infants ranging from six to 57 days old (with an average age of 27 days). Their findings suggest that circuit wiring precedes, and thus may guide, regional specialization, shedding light on how knowledge systems emerge in the brain. Further work along these lines may provide insight into neurodevelopmental disorders such as autism.

In the study, published Monday in Proceedings of the National Academy of Sciences USA, the researchers looked at two of the best-studied brain networks dedicated to a particular visual function—one that underlies face recognition and another that processes scenes. The occipital face area and fusiform face area selectively respond to faces and are highly connected in adults, suggesting they constitute a face-recognition network. The same description applies to the parahippocampal place area and retrosplenial complex but for scenes. All four of these areas are in the inferior temporal cortex, which is behind the ear in humans.

The team used a technique called resting-state functional magnetic resonance imaging (rsfMRI), which measures the level of synchronization of activity in different brain regions to assess how connected they are. The infants were scanned while sleeping and tightly swaddled. “Getting fMRI data from newborns is a new frontier in neuroimaging,” says neuroscientist and lead study author Frederik Kamps, now at the Massachusetts Institute of Technology. “You need participants’ head to be still, and a sleeping baby is one that’s willing to lie still.”

The researchers found that the face regions were highly connected to one another but not to the scene regions, and vice versa, at this young age. It would be months before they became selective for faces or scenes, suggesting connectivity precedes the development of function.

The team also assessed connections between these regions and the part of the brain where visual input first arrives from the retina: the primary visual cortex, or V1. This region is structured so that such inputs from the center of the retina arrive at a different area than those from the periphery of the field of vision, forming a map of the visual world. The face network was strongly connected to V1’s central area, while the scene network was more tightly linked to its peripheral area. This arrangement likely relates to the fact that we usually fixate on faces, whereas scenes extend across our entire visual field. These networks, present in an infant’s earliest days, are therefore connected so as to receive the most appropriate input for the function they will eventually perform.

Does that mean face recognition and scene processing are innate? Researchers disagree on this point. In 2017 neurobiologist Margaret Livingstone of Harvard Medical School published a study of newborn macaques that found connectivity precedes function—but only as far as visual maps. Livingstone, who was not an author of the new paper, thinks sensitivities to specific categories of things, such as faces, arise from accumulating experiences of seeing them. “You’re born with these maps, and that’s what drives the final organization of the brain,” she says. “That’s the scaffolding on which experience acts.” In another study, she found that monkeys raised without seeing faces did not develop face selectivity.

Others, however, have shown that congenitally blind people have face- and scene-selective regions (using tactile or auditory stimuli, for example), suggesting these functions may be innate—or at least, that they may depend on more than just visual input. Dilks notes that faces are not the only things we fixate on, and other researchers have proposed that “top-down” connections from high-level cortical regions involved in social interaction (between mother and baby, for instance) may also shape the development of face selectivity. This debate shows no sign of being settled soon. “It all boils down to this philosophical question: Are humans special? Do they have parts of their brain predestined to become these special things?” Livingstone says. “Or can we explain it using low-level principles we’ve inherited from lower animals?”

Beyond this theoretical wrangling, Dilks has an eye on possible clinical applications. He is particularly interested in two neurodevelopmental disorders that are thought to involve differences in brain wiring: People with autism have social impairments that may relate to face processing. And a condition called Williams syndrome causes problems with navigation.

Siblings of children with autism could be studied to ask whether connectivity in face regions might predict the onset of the condition, which is usually not diagnosed until at least two years of age. Dilks also hopes to study babies with Williams syndrome to ask whether connectivity between scene-processing regions is a problem. “That’s important to know,” he says, “because maybe we can harness the incredible malleability of the infant brain to intervene earlier.”