The team then played these songs to 750 volunteers recruited through the internet—a third from the United States, a third from India, and another third from a mix of 58 other countries. Every participant listened to 36 recordings, and rated how likely each one was to be, say, a dance song or healing song.
They were surprisingly accurate, for every category except love songs. Dance songs and lullabies, in particular, share enough features around the world that naïve listeners can identify them with no experience of the cultures from which they arise. The three groups of volunteers were also surprisingly consistent. “Some random person from Texas who’s doing our survey is expected to have a similar conception of what a healing song should be to someone at her computer in India,” says Singh.