Spotify data shows how music preferences change with latitude

In Anchorage, Alaska, in the bleakest midwinter, the Sun rises just after 10am, hovers close to the horizon for a few hours, and dips back down again shortly before 4pm—all in all, just five hours and 28 minutes of daylight. Around that winter solstice (just in time for Christmas), Spotify listeners in the most northerly latitudes of the world dial down the intensity of their music choices, choosing calmer and more relaxing music.

No, it’s not because of chilled-out Christmas playlists: in the Southern Hemisphere, the exact opposite happens, with a peak in intensity just after the summer solstice in December. A paper in Nature Human Behaviour this week drew on the listening data of nearly a million Spotify listeners from around the world, describing the daily and seasonal variations in how people listen. The researchers suggest that the results point to a universal human habit that probably sounds familiar: choosing your music to both match and change your mood.

Morning music vs. midday music

Cornell PhD student Minsu Park and her colleagues were interested in how mass trends in music choices could illustrate the rhythms of music’s role in people’s emotional lives. So, when she ended up interning at Spotify, Park used the opportunity to work with Spotify colleagues and her advisor to dig into the treasure trove of information that’s now available on people’s music choices.

Although Spotify does recommend music to listeners, it reported that in 2016—the year used in the study—more than 80 percent of tracks were users’ personal choices. The researchers also limited their focus to tracks that had played all the way through, reasoning that users would skip tracks they didn’t like or weren’t in the mood for (presumably they decided that Spotify’s limitations on skips for free accounts weren’t a big enough problem to throw their results out substantially).

The researchers took data from listeners in 51 countries, making sure that their samples matched the demographics of each country but otherwise selecting users randomly. Using Spotify-provided data on the music, they tracked a variable they called musical intensity, “ranging from highly relaxing (acoustic, instrumental, ambient, and flat or low tempo) to highly energetic (strong beat, danceable, loud, and bouncy).”

Those intensity preferences tracked daily rhythms more or less exactly as you might expect: lower-intensity songs in the morning, rising until normal work hours, then staying steady before dropping off in the evening, with weekends looking a little different. These results matched up neatly with a previous study tracking emotions in Twitter users’ speech, but it differed on one point: language showed an afternoon slump, but there was no such slump in the music choices. It’s possible, the authors suggest, that people might be choosing music that gives them a boost.

The data also showed some cultural differences—more energetic music, on average, in Latin America, more relaxing music in Asia—and a gender difference that depended on hemisphere: women listen to less intense music in the Northern Hemisphere and more intense in the Southern Hemisphere.

Seasonal intensity

But the annual variation is where things really get intriguing, suggesting that music choices track day length. Peaks in intensity matched the summer solstice in each hemisphere, and these swings were more extreme at more extreme latitudes. Near the equator, changes in intensity were much flatter across the whole year, while more northerly and southerly places (which have greater variation in day lengths) had larger changes in music preferences. Day length accounted for musical intensity better than a range of other options.

This might sound like a glaring example of how humans respond to factors like temperature or sunlight, but it’s possible that there’s a hefty dose of culture bound up in there, too. In a culture that has a long and distinct fall season, there will be reams of associations with what fall means—flavors, clothes, and feelings—which could influence people’s music choices. More danceable music during summer holidays also makes sense. Those cultural factors are obviously also influenced by biological and environmental factors, but it’s all bound in a knot that’s difficult to untangle.

The work comes with caveats, of course. For one thing, the researchers relied on IP addresses and user profiles to classify users by their demographics and geographical regions. Users can lie or use VPNs, so there’s likely to be some fuzziness in the data. And all of the results are purely descriptive—they can’t tell us anything certain about the reasons for the patterns, like whether people are really using music to intentionally affect mood. Intuition can try to fill in the blanks, but different kinds of research—like lab studies—would be needed to confirm those intuitions.

Large datasets like this also contain a lot of noise, which can be mistaken for patterns, especially if weird or confusing results are discarded and not reported. And of course, Spotify users are a distinct (and relatively wealthy) demographic—not necessarily an accurate model of all human experience.

Nature Human Behaviour, 2018. DOI: 10.1038/s41562-018-0508-z (About DOIs).

http://platform.twitter.com/widgets.js