The babies looked at the screen for about the same amount of time, no matter what. Eyes visible, eyes shaded, eyes hidden behind dark glasses: attention, measured as looking time, barely moved.
Learning did. Infants aged eight to 10 months in the United Kingdom and Singapore were exposed to three invented languages, each paired with the same speaker wearing different glasses. They learned the one language taught by a speaker whose eyes they could fully see.
"Even when the infants seemed to be staring at a screen attentively, unless there was eye contact, nothing was going in," said Kaili Clackson of the University of Cambridge's Department of Psychology. The information, she said, washed over them.
The study was published in Nature Communications on July 22, with first author W. Zhang and senior author Victoria Leong of Nanyang Technological University in Singapore and the University of Cambridge.
Three Nonsense Languages and One Speaker
The design was built to isolate a single variable. Each of the three artificial languages was made of syllable triplets repeated over and over, and the same speaker delivered all three on video. Only her glasses changed: clear, partially shaded, or fully dark.
While infants were familiarized with the languages, researchers recorded their brain activity using electroencephalography with electrodes on a skull cap. Forty-two infants from the two countries took part in that portion. The team had separately recorded the speaker's own brainwaves at the time the videos were made.
To test whether the babies had learned anything, the speaker then presented individual phrases from the languages alongside newly invented words. Infants typically look longer at unfamiliar material, which is the standard behavioral indicator that something has been learned.
"They're not just passive sponges," Leong said. "They heard all three languages, but only learned the one, which was when they could see the speaker's eyes."
The artificial-language setup is standard in infant language research precisely because it removes prior exposure as a confound. No baby in either country had heard any of the three languages before, so differences in learning could not be explained by familiarity with particular sounds.
The Adult's Brainwaves Predicted the Baby's Learning
The neural result is stranger than the behavioral one, and that is why the paper exists.
When a baby was learning a language, that baby's brainwaves were more likely to be in sync with the speaker's. The level of synchrony predicted learning better than the infants' own brain activity did, and the paper's title frames the coupling as running one way, from adult to infant.
That synchrony only appeared when the baby could fully see the speaker's eyes. The researchers describe eye contact as a switch rather than an attention magnet, since looking time did not change much across conditions.
The pattern held in both Cambridge and Singapore, which the team reads as evidence that the mechanism is not culture-specific. Infants in Singapore did spend more time overall looking at the speaker, possibly because her ethnicity was less familiar, but that did not affect learning.
Leong's group had previously shown that infant and adult brainwaves synchronize during eye contact. What was missing was evidence that the synchrony did anything at all. This study connects the two by showing that the coupling predicts a measurable learning outcome.
Eye Contact Is One Signal Among Several
The researchers are careful not to elevate eye contact into a universal requirement, and the caution has cross-cultural weight.
Other cues that signal intention do similar work, including smiling, pointing, and using an infant's name. Those alternatives may matter especially in cultures where sustained eye contact between a child and an adult is discouraged.
"These cues are really powerful because when you preface new information with them, they tell the baby: this is important, you need to listen," Clackson said.
Leong drew a blunter implication for parents, arguing that social cues do not merely focus a child's attention but activate the learning process itself. Her example was a distracted adult trying to teach something while looking at a phone.
The research was supported by A*STAR and the Ministry of Education, Singapore, and the team spans Cambridge's Department of Pediatrics and Department of Psychology and NTU's School of Social Sciences, according to Cambridge Neuroscience.
What This Study Cannot Tell Parents
Forty-two infants is a modest sample, and the setup involved a speaker on a screen rather than live interaction.
The finding concerns a specific laboratory learning task involving invented languages, not real vocabulary acquisition or long-term developmental outcomes. It measures learning within a session, not months later.
It also should not be read as a claim that screen time or phone use damages children's brains, or that any single interaction determines development. The study did not test those questions.
What it does supply is a mechanism for something developmental psychologists have long assumed: that adults help infants decide what is worth learning, and that this filtering shows up in brain activity before it shows up in language.
Reduced eye contact in an infant can also be an early sign warranting evaluation for developmental differences, and parents who notice this should raise it with a pediatrician rather than interpreting it themselves.
Key Questions Answered
What did the study test?
Whether an adult speaker's eye contact affects whether infants learn a language, and what happens in the brain when they do.
How many infants took part?
Forty-two infants aged 8 to 10 months from the United Kingdom and Singapore participated in the brainwave portion of the experiment.
What was the key neural finding?
Synchrony between the speaker's brainwaves and the infant's predicted learning was greater than that of the infants' own brain activity, and it appeared only with full eye contact.
Did babies pay less attention without eye contact?
No. Looking time was similar across conditions. Learning was what differed.
Is eye contact the only cue that works?
No. Researchers note that smiling, pointing, and using a child's name serve similar signaling functions, which matters in cultures where eye contact is discouraged.
Did the result hold in both countries?
Yes. The researchers found the same pattern in Cambridge and Singapore, which they read as evidence that the mechanism is not culture-specific.
Does this prove screens harm babies?
No. The study did not examine the effects of screen time or long-term developmental outcomes.
