A Nyangatom woman from East Africa sings in an up and down lilt, her unadorned voice rising and falling like a hilly landscape. If you heard this song, with no geographical or cultural context, would you know it was a lullaby meant to soothe babies to sleep? It turns out you probably would, according to a new study published in Current Biology.

In the largest experiment of its kind, cognitive scientists asked hundreds of English speakers from around the world to listen to lullabies, dance songs, love songs, and healing songs from a wide swath of non-Western cultures. Although listeners had trouble identifying love songs, many could distinguish a healing song. But what struck the researchers most was the high confidence with which people identified lullabies and dance songs.

“What that suggests is that lullabies, dance songs, and, to a lesser extent, healing songs, share enough features across cultures to be universal features of human behavior,” says Samuel Mehr, a cognitive scientist at Harvard University and the Victoria University of Wellington in New Zealand, and an author on the new study. “You don’t need to know anything about a particular culture to be able to make some really interesting and accurate inferences about their music.”

What that means is that music could indeed be universal—and not just in a broad, metaphorical sense. Every culture has its own music, but some researchers have hypothesized that certain features and patterns hidden among the notes and rhythms are common to all cultures. So far, however, evidence for these hypothesized universal features has been lacking.

Researchers have only recently begun hunting for universal features. Traditionally, anthropologists and ethnomusicologists, who study the diversity of the world’s music, have eschewed comparative studies, says Greg Bryant, a cognitive scientist at the University of California at Los Angeles, who wasn’t part of the new study. “A lot of cultural anthropologists weren’t as interested in comparing cultures because they thought it was comparing apples and oranges,” says Bryant. Instead, they focus on the nuances and complexities of individual cultures.

Perhaps as a result, a survey that the researchers conducted of 940 academics found that only about half of music scholars, and less than 30 percent of ethnomusicologists, thought people would be able to identify a song’s function just by listening to it. Meanwhile, more than 70 percent of cognitive scientists, who seek commonalities stemming from biology, thought otherwise.

But even if we something seems obvious, it’s not always true. “It’s really easy to think something is true across cultures because of our biased perspectives,” Bryant says. (Westerners may think of beans as part of a savory dish like chili, but in Asia, they’re often found in dessert.) You have to look at the data—which is why the new experiments stand out. “They’re really looking at a lot of different songs from different places, and using a pretty big sample of listeners,” he says.

The researchers are the first to use such a large, diverse database of music, part of a project based at Harvard called the Natural History of Song. The collection contains 118 field recordings and 5,000 descriptions of song and dance. To find possible universal features in song, researchers are culling descriptions and recordings of vocal music from around the world, including data such as a song’s tempo and the demographics of singers and the audience.

Previous studies have been few and limited. They generally compared only two types of music, or they used databases that didn’t accurately represent music from around the world, Mehr says. In 2015, researchers led by Patrick Savage, a musicologist at Oxford University, identified several features—specific characteristics of rhythm and pitch, for example—that were statistically common in a collection of 304 recordings. According to Savage and his coauthors, their study provided the first quantitative evidence for universal features in music. But, according to the new paper, the database didn’t represent an even sampling of the world’s music.

In contrast, the new study involved 118 recordings from 86 small, isolated societies scattered evenly across the globe. In the first experiment, 750 English speakers from 60 countries listened to a random sample songs online. Then they ranked how confident they were that a particular clip was used for dancing, to soothe a baby, to heal an illness, to express love, to mourn the dead, or to tell a story. (The latter two options, which didn’t apply to any of the samples, were thrown in to keep the listeners from answering simply by elimination.)

A second experiment explored the general song characteristics that influenced the listeners’ decisions. A different set of 1,000 online listeners from the U.S. and India rated contextual features—such as the singer’s gender, and musical features, such as whether the song had a steady beat. A statistical analysis revealed that the features that explain how people identify lullabies were the complete opposite of those for dance songs. Lullabies sounded sadder, simpler and slower while dance songs were happier, faster and more complex.

Overall, the analysis shows that people recognized the purpose of a song based on both context and, to a greater degree, musical features. But neither feature could explain the full extent of the rankings in the first experiment, the researchers say. Something else inherent to the music was tipping off the listeners.  

As of now, the experiments can only hint at the existence of specific, universal features; it can’t yet tell us what they are. “In itself, it’s not saying much about universals,” says Sandra Trehub, a psychologist at the University of Toronto whose 1993 experiments suggested adults could identify lullabies from other cultures, and who wasn’t involved in the new research. “But I think it’s leading in that direction.”

To see if their results are really universal, the researchers are now repeating their experiments with non-English speakers and with small-scale societies. In the future, they hope to analyze the music itself—maybe even with artificial intelligence techniques—and zero in on the specific features that may make music universal. “I’m really excited to find out what happens next,” Mehr says.

Equally tantalizingly, this kind of study could even help pinpoint the evolutionary origin of music. Some researchers hypothesize that songs are the most primitive form of music, Bryant says. Lullabies, for example, may have evolved from parent-offspring interactions, according to a recent theory proposed by Mehr. In the new study, the data on both lullabies and dance songs are so pronounced that they might point at something deep and fundamental, says Manvir Singh, an author of the paper.

“Both of these may be important for why music evolved,” he says.