The chill

There is a moment in certain songs when the hair on your arm stands up. An unexpected chord, a voice leaping an octave, a swell out of near-silence. The sensation arrives uninvited, spreads fast, and has a name in the literature: frisson. It is a strange thing to feel, because the machinery it runs on was built for something else.

Goosebumps are a pilomotor reflex, a leftover of being furry. Cold air or a threat fires tiny muscles at the base of each hair; the hair stands up to trap warmth or to make the animal look bigger. On a bare human arm, neither job gets done. Yet organized sound, which carries no temperature and no teeth, fires the same reflex.

In 2001, Anne Blood and Robert Zatorre put the question in a PET scanner. Volunteers brought their own music, the specific recordings that reliably gave them shivers down the spine, and listened while the machine tracked cerebral blood flow. As the reported chills grew more intense, blood flow shifted in the ventral striatum, midbrain, amygdala, and orbitofrontal cortex, the same map that responds to food, sex, and drugs of abuse. Heart rate and breathing moved with the chills. An abstract pattern of sound was drawing on circuitry the body normally spends on survival.

The chemistry can be interrupted. Back in 1980, Avram Goldstein gave volunteers an opioid blocker and found the thrills reduced: part of the pleasure runs on the brain's own opioids. In 2011, Valorie Salimpoor and colleagues caught the dopamine itself. A tracer that reveals endogenous dopamine showed the striatum releasing it during peak musical emotion, and fMRI on the same listeners split the timing. The caudate was more involved during the anticipation, the seconds of hearing the moment approach. The nucleus accumbens took over at the peak itself. The dopamine release itself is split in time and place: the caudate while the moment approaches, the nucleus accumbens when it lands.

Which explains where chills cluster. When researchers examine the music instead of the listener, the trigger is usually a violated expectation: an unexpected harmony, a sudden leap from soft to loud, a modulation. In John Sloboda's 1991 catalogue of musical moments, new and unexpected harmonies produced shivers, while appoggiaturas, notes that lean and then resolve, produced tears. The shiver is what a prediction error feels like when the error turns out beautiful.

It is not universal. In one set of measurements only about half of reported chills came with gooseflesh at all, and the location wanders; the arms get it more often than the famous spine. Whether it happens at all tracks the wiring: the strength of the connection between a person's auditory areas and their reward circuitry predicts frisson. The songs are the same. The payout schedule is not.

So a reflex calibrated for winter and wolves gets spent on a chord change. Frisson may be the cheapest evidence that the reward system does not much care what a reward is, only that it was predicted, violated, and resolved.