Incoprea Theory
Volume XI · nine fields

Theory 05/Music & Sound Culture/Music cognition

Music Moves by Prediction Error

Why a song stops working after a hundred listens is usually answered with taste. It may be arithmetic.

The claim

The emotional effect of music is set by prediction error in the rhythmic and harmonic hierarchy rather than by mode or tempo, so a piece loses its effect with familiarity and recovers it with subtle variation.

01The evidence

Musical pleasure is associated with anticipation and surprise, and familiar pieces show reduced response in reward regions.

02The tension

Explanations invoke mode, tempo and cultural association, all of which are static properties of the piece and cannot explain why the same piece becomes less moving over time.

03The mechanism

The auditory system predicts the next event at several timescales. Pleasure peaks when the prediction is almost right, which is an intermediate error; too predictable is boring and too unpredictable is noise.

Familiarity reduces error, which moves the listener down the curve. Variation, performance nuance and reharmonisation restore it without changing the piece's nominal mode or tempo.

04What this is not

Static-feature explanations

Those attribute effect to mode and tempo. This attributes it to error, which explains the change over repeated listening.

05Predictions

  • P1Pleasure should track prediction error across repetitions.
  • P2Subtle variation should restore response to a familiar piece.
  • P3Mode and tempo should predict poorly once error is controlled.

06Refutation

Pleasure constant across repetition at fixed mode and tempo would favour static features.

07What it changes

It would change how music is composed for repeated listening and how playlists are sequenced.