Pleasure from pop music

Enjoyment of pop songs is down to the uncertainty and surprise created by certain chords...
10 November 2019


this is a cartoon of a human brain, made from musical notation


Enjoyment of pop songs is linked to uncertainty and surprise created by certain chords...

We all have a favourite pop song that brings us pleasure - in some cases even guilty pleasure - but it may not be for the reasons we believe.

A research team in Europe have found that two aspects of expectancy - "uncertainty" and "surprise" - account for musical pleasure. Uncertainty is the lack of expectation when anticipating an event, while surprise is what happens when what is actually heard varies from one's expectations.

The research, published in the journal Current Biology, used machine learning and brain scanning to establish the link between expectancy in music and impact on our emotions.

The machine learning model was used to predict how uncertain or surprising a musical chord is. The model uses information theory - the mathematical study of transmission and processing of information - to efficiently and precisely measure uncertainty and surprise through analysis of the audio signals. The model was taught to measure the uncertainty and surprise in 80,000 chords extracted from 745 classic pop songs, ranging from 1960's Beatles numbers to ABBA and hits from the 1990s.

A group of participants were asked to rate how much they liked a subset of these chords, which did not include lyrics or melody. A second machine learning model was then used to look at the relationship between the participants' ratings and the theoretical ratings produced by the first machine learning model.

Researchers found that a chord was “more pleasantly perceived when it struck a good balance between the uncertainty and the surprise,” says PhD student Vincent Cheung from the Max Planck Institute for Human Cognitive and Brain Sciences. More specifically, participants tended to prefer chords with a high uncertainty and low surprise, and vice versa.

A second group of participants were then asked to listen to another subset of chords whilst undergoing magnetic resonance imaging (MRI) of their brains. The resulting scans were consistent with the initial results - the effects of uncertainty and surprise were seen in the auditory cortex, an area of the brain used to process sound, and also in the amygdala and hippocampus, areas of the brain used to process emotions and memory.

The researchers claim that these new findings have wider implications. According to Cheung, “now that we know how the effects of statistical regularity can modulate our preferences, then this could be used to enhance artificial music generation, or even help song writers to write their own songs, or even predict upcoming musical trends!”

So who knows, it's possible that, within the next decade, we won’t be able to tell whether we are listening to a song created by a human, or an algorithm...


Add a comment