Hip-hop artist Childish Gambino (aka actor Donald Glover) made a splash in 2018 with the release of his Grammy-winning hit single, “This is America.” With its stark, sudden shifts between choral melodies in major chords and menacing percussive elements drawn from the trap subgenre, the song constantly defies the listener’s expectations throughout.
That’s why “This is America” also tops the list of pop songs rich in so-called “harmonic surprise,” or points when the music deviates from listener expectations. This is according to a recent study published in the journal Frontiers in Human Neuroscience that analyzes Billboard hits from 1958 to 2019. And it’s no coincidence that the tune is among the most recent: the study also found that harmonic surprise in popular music has been increasing over the decades—a phenomenon the authors have dubbed “inflationary surprise.”
“Music is culture. Culture evolves over time, so the content of music needs to evolve as time goes by, just to have the same success as previously released music,” said co-author Scott Miles, a neuroscientist specializing in how music preference forms in the brain. “This study gets to the heart of what is a dynamic effect within popular culture, in a very concrete and measurable way. These findings [also] help further our understanding of how music is processed in the brain.”
Miles is co-founder and president of music startup Secret Chord Laboratories (SCL). His SCL co-founder, CEO David Rosen, was part of a study published last year in which researchers at Drexel University imaged the brains of improvisational jazz musicians as they played. As we reported previously, the team found that, while the right hemisphere is associated with creativity in fairly inexperienced jazz musicians, experts with high mastery of improvisational skills actually rely primarily on the left hemisphere of the brain.
While the right brain is associated with adapting to novel, unfamiliar situations, the left side of the brain is generally more engaged during habitual or routine tasks. Because of their years of experience, experts approached the task of jazz improvisation with the most habitual routines. Rosen attributed this in part to the fact that a novice must go through 16 bars of music on a lead sheet quarter note by quarter note, for instance, whereas an expert will be able to find patterns, like a ii-V-I progression. So creativity is associated with the right hemisphere when we’re dealing with an unfamiliar situation and associated with the left hemisphere when we are highly experienced with the task at hand.
All this is part of a growing body of research at SCL, which develops software to predict the response listeners will have to a given piece of music. The core technology grew out of a 2017 study Rosen did with Miles on music perception. Together, they examined what, if any, patterns of sound produce a pleasure response in the brain. They looked at Hot 100 songs on the Billboard charts, from 1958’s “Johnny B. Goode” through Nirvana’s “Smells Like Teen Spirit” in 1991.
The researchers found that the most popular songs had a high level of harmonic surprise, including the use of relatively rare chords in verses, for example, instead of just sticking with, say, a standard C major chord progression (C, G, F). The best songs follow up that harmonic surprise with a catchy common chorus. The resulting patents—since expanded to include rhythm, melody, timbre, and lyrics, as well as harmonies—led to the formation of SCL.
The notion of a software program telling songwriters how to make their songs better (from a pop hit standpoint) is bound to raise hackles. But Rosen—a composer himself—insists that the algorithm is only meant to enhance human creativity, not replace it. “Our algorithm doesn’t tell you what the melody should be,” he said. “Rather, it can identify if a certain part of your melody needs more or less surprise, empowering artists to take risks to satisfy their own creativity needs and those of their audiences.”
This latest paper expands Miles and Rosen’s research on what makes a hit pop song by adding the dimension of time. According to Miles, one of the reviewers on their 2017 study asked whether the measured effect was static or changed over time. The question intrigued Miles, so in this latest analysis, he and his co-authors combined the dataset from their 2017 study with SCL’s database of Billboard Hot 100 songs from 2000-2019. They divided the songs into four consecutive time bins, each spanning about five years of release dates, and the company’s propriety machine-learning software calculated the harmonic surprise for each chord.
That analysis showed that harmonic surprise increased across the board over time and that the increase was much more pronounced in the most popular hit songs, regardless of musical style—whether it be Elvis, Madonna, Nirvana, Beyoncé, Drake, or Taylor Swift. And “This is America” came out on top. According to the authors, not only is there an element of surprise in the verses, attained by altering the D minor/B flat major chord progression, but the refrain at the 2:56 mark (3:17 in the official music video) departs from the F minor key entirely for a moment. Instead, it briefly introduces E minor and A minor chords before returning to the original key for the duration of the song.
As for why this might be the case, the authors posit that listening to songs high in harmonic surprise triggers our neural reward system and releases dopamine. “You can make the analogy to a drug, where it’s a dopaminergic reward,” Rosen told Ars. “There’s kind of a unique cohort of 14- to 20-year-olds who drive the success of pop music. The expectations for that cohort are different in 2000 from the expectations of teenagers in 2005, because when those 2005 teens were young, they were being exposed to music that had a certain surprise content. When they became the ones driving pop music, they needed a higher level of harmonic surprise to elicit the same dopamine effect than teens in 2000. And we saw that [trend] in two different sets of music.”
But surely composers of hit songs can’t keep upping the levels of harmonic surprise indefinitely without hitting some kind of aesthetic threshold. That’s part of the next stage of Miles and Rosen’s research. This latest study focused on chords, but fortunately, there is more than one way to inject elements of surprise into a song, including dynamics, rhythm, tempo, timbre, and so forth. Based on a few pilot tests last summer that took all those various elements into account, when listeners do peak on a certain level of surprise, “The fastest way to reset or recalibrate how surprise is functioning in music is to dive into a new genre,” said Rosen.
“Nobody listens to just a collection of chords, they listen to the whole song,” said Miles. “When we look at other features, we’ve found some evidence that surprise in a particular feature will get saturated, and it will migrate [to a different feature].”
So the rich harmonics of 1970s disco, progressive rock, and R&B gave way to the synth-heavy shifts in timbre of the 1980s, for example. “Smells Like Teen Spirit,” by grunge rock pioneers Nirvana, “changed the whole landscape of the sonic experience in the ’90s,” Miles told Ars, despite having no harmonic shifts. What the song does have is a lot of dynamic shift, possibly stemming from the influence of The Pixies on lead singer/songwriter Kurt Cobain. (I personally have a soft spot for the many surprising elements in Guns N’ Roses’ “Welcome to the Jungle” from 1987.) And rap music might not have much melodic complexity, but there is a great deal of complexity to the rhymes in the lyrics (internal rhyming, etc).
Miles is hopeful that their work will one day help shed light on how tonality is established in the human brain—very much an open question in neuroscience. For instance, do human beings have a natural, innate affinity for, say, Pythagorean tuning? Miles believes their findings are evidence that tonality is much more plastic and is learnt over time.
“It all comes down to how we orchestrate learning and how we’ve evolved to develop culture,” he said. “Our statistical expectations of harmonic regularities are not hard-coded. These statistical learning centers are very close to the emotional bonding centers of our brains. It’s not only dopamine, it’s also serotonin and oxytocin [involved]. So as technology has advanced, nurture and nature has become a lot more blurred.”