One of the many luxuries of my superannuated life is the bountiful gift of time, specifically the four or five hours a day (yes!) I can happily burn through doing nothing but listening to music.
Lately, however, these listening sessions have also been driven by a bit of scientific research. Curated by Remi de Fleurian, a data scientist at Spotify, and Marcus Pearce, who teaches cognitive science at Queen Mary University of London, the playlist was compiled as part of a study on frisson; the goosebumps, or literal shivers down the spine, that certain music produces.
Part of it is pure unadulterated passion, part is the joyful occupational “hazard” of writing this column. I’ve been diving into a 715-song playlist known to researchers as the “chill corpus”.
Instead of the usual approach — a small sample of people in a lab, wired up and subjected to tracks — Fleurian and Pearce took a computational route. The research itself concedes this. A song one has never heard before, however scientifically vetted, might never carry the weight of a song picked by one’s own life. It can state, with some confidence, that a slow tempo and an unresolved chord are more likely to produce a shiver than a wall of noise. The corpus, in the end, can point to the neighbourhood. It’s up to you to find the right house.
The study unfolded in two phases, over five years: a 2020-21 phase that built the corpus and conducted the acoustic comparison (this is what produced the viral 715-song playlist), and a 2024-25 phase that used machine learning to try to predict exactly when a chill would arrive. Richard Wagner’s Die Walküre (1856) may leave plenty of listeners cold. So could Wilco’s Jesus, Etc (2002). Which is really the only honest conclusion to draw from a 715-song list built by data scientists: It’s a very good map, but not necessarily the territory. It cannot say why a Shostakovich symphony lands exactly right on a particular afternoon, or why a concerto ignored as a child turns out, 60-odd years on, to have been waiting quietly the whole time.
All in an attempt to isolate what made the “chills” songs different. What both phases converged on was fairly consistent: chill songs, compared to the same artist’s average output, tend to be slower, sadder, more acoustic and less “busy”. About half the people polled report no musical chills at all, and this correlates less with taste than with a personality trait the researchers call “openness to experience” (as well as a fair amount of personal history, which no algorithm can account for).
They gathered songs that earlier studies had identified as reliable chill-triggers; ran a computer program comparing them with other songs by the same artists; and matched for length and popularity.
11 (1957) turned up. I later learned that it commemorates Bloody Sunday, 1905, when Tsarist troops fired on unarmed civil-rights demonstrators in Saint Petersburg. I didn’t expect electronic dance music to do anything for me, but Echoes (2007) by the German duo Digitalism, a track often cited as a decent entry point into EDM, does. Prince’s Purple Rain (1984), unsurprisingly, delivers, the guitar solo arriving exactly when the song has stretched one’s patience. Elvis Costello’s Alison (1977) works in the opposite way, quietly. Boston’s Peace of Mind (1976) sent me back to the early ’80s, which is when I last listened to this band. The Smiths’ How Soon Is Now? (1984) did it with the tremolo guitar riff running underneath Morrissey’s vocals the whole way through. Pixies’ Gigantic (1988) works off the quiet-loud dynamic that the band all but patented. Ibrahim Maalouf’s Beirut (2011) earns its place the moment it breaks into a passage where jazz improvisation and Arabic modal music sit on top of each other. Phish’s You Enjoy Myself, off Junta (1989), took me back to the early days of encountering that group. The Grateful Dead’s Ripple (1970) belongs on the list for plenty of reasons, including the way its slow, sparse, fingerpicked sound leaves one’s expectations a bit unresolved. Nearly 50 years of being a Deadhead means the chills I get from it, though, have less to do with chord changes than with all the memories now associated with the song. Oasis’s Wonderwall (1995) got to me, almost against my will. So did Adele’s Hello (2015), and Amy Winehouse’s Back to Black (2006). Also on the list: Charles Mingus’s Fables of Faubus (1959), which isn’t really built for chills in the conventional sense. It’s a scathing, explicit protest against Arkansas Governor Orval Faubus, who in 1957 called out the National Guard to stop a group of Black children nicknamed the Little Rock Nine from integrating a high school.
The strongest predictor of a chill, it turns out, is not volume or a beat drop, but expectation: The brain quietly anticipating the next note, and then being surprised. TURN UP THE SHIVERS My own experience with the corpus has been less scientific and more excursionary. I’ve been moving up and down the list at random, waiting to see what does it for me. I put it on shuffle while writing this, and Soviet composer Dmitri Shostakovich’s Symphony No. It turned out to be a grand, cinematic piece that had just the effect the researchers were looking for. Some of the chills, as the music continued, weren’t about the music at all. They were about memory. Beethoven’s Piano Concerto No. It is only decades later that music like that catches up with you. The list traverses genres in a way that surprised me. It’s a long, meandering, half-improvised piece that ultimately rewards one’s patience. Pop, of all things, held up its end. The chill here is less about harmony and more about knowing what the music is angry about. None of this, of course, works the same way for everyone.
5 (1809), the “Emperor Concerto”, did this to me instantly: Some 60 years ago, Baba playing his records, never too loud, for himself and Ma, while we kids sat around mostly unmoved.
