Listeners bring genre expectations into your song

One groove, two verdicts

You build a loop in seven, every bar counted 2+2+3 with the long count at the end. A friend who grew up on Balkan brass nods along from the first bar. A friend who listens to pop and house keeps losing the downbeat and asks whether the edit slipped. They heard the same file, and each one heard it against the music they already know.

Why it matters

Surprising for whom?

When you decide that a move is too strange, too safe or just right, you consult your own expectations, and those were built by your listening. If you have spent years inside one genre, your sense of normal is deep there and shallow elsewhere. Your audience may have a different history, and a crossover track may have two audiences with two different histories. Feedback on a risky bar says as much about the listener as about the bar.

The science

Expectations are learned from a style

Pearce (2018) reviews the case that listeners absorb the statistical regularities of the music they hear and use them to predict what comes next. In that work, a computer model trained on one culture's music simulates the expectations of listeners from that culture, and training it on other music plausibly simulates listeners with other backgrounds. The probabilities come from each listener's own history. The lesson on surprise shows how such predictions turn into a measure of surprise.

Rhythm gives a clean test. Hannon and Trehub (2005a) played folk melodies in simple metres and in complex metres common in Balkan music, then tested listeners on altered versions that either kept or broke the original metre. North American adults told the two kinds of change apart in simple metre but not in the complex one. Adults of Bulgarian or Macedonian origin managed both, and so did 6-month-old infants. A follow-up (Hannon & Trehub, 2005b) tested 40 college students, most raised in North America, in two sessions one or two weeks apart. Those assigned to play a CD of Balkan folk music at home in between reported doing so about once a day, and in the second session they were still at chance on the complex metre. Infants about a year old, given similar home listening, learned to tell the changes apart. The authors concluded that the adults fitted the foreign rhythms into a Western even-beat frame.

Expertise inside a genre matters too. Hansen, Vuust and Pearce (2016) played phrases from Charlie Parker solos to 22 jazz musicians, 20 classical musicians and 20 non-musicians. Listeners rated how expected each of nine possible next notes was, and how certain they felt about what would come next. Both musician groups rated expectedness more in line with a model trained on bebop than non-musicians did. Only the jazz musicians' sense of certainty tracked that model. Specialists seem to know consciously what the style predicts. The groups were small, differed in gender balance and main instrument, and the analyses were not corrected for multiple comparisons, so treat the details with care.

You can hear a learned expectation in harmony right now. For most listeners raised on Western pop, a phrase that stops on the V chord sounds like a question, and the I chord sounds like the answer. On the statistical-learning account that Pearce reviews, much of that sense of closure comes from having heard V move to I countless times. Listen for how open the V ending sounds next to the I ending, and keep in mind that this pull is your learned expectation, shared by your audience only if they learned it too.