A groundbreaking study by scientists at the University of Göttingen and MPI-DS uses data science to unravel how emotions arise in music, highlighting distinct differences between jazz improvisations and classical compositions such as those by Bach and Mozart.
It is well-known that music can stir deep emotions, but the scientific basis for how this occurs has long remained a mystery. A new study by a team of scientists at the Max Planck Institute for Dynamics and Self-Organization (MPI-DS) and the University of Göttingen has taken significant strides in unveiling this mystery by employing modern data science methods to explore how musical expectations and surprises evoke emotions.
Published in the journal Nature Communications, the study builds on a 70-year-old hypothesis by music philosopher Leonard Meyer, who proposed that emotions in music arise from the interplay between expectations and their fulfillment or disruption.
The researchers, led by Theo Geisel, director emeritus at MPI-DS and a professor of theoretical physics at Göttingen, analyzed the “memory” of musical pitch sequences using time series analysis to measure their autocorrelation functions. This essentially quantifies how similar a sequence of tones is to previously heard sequences, shedding new light on predictability and variability in music.
The team meticulously analyzed over 450 jazz improvisations and 99 classical compositions, including multi-movement symphonies and sonatas. They discovered that predictability in music decreases over time and varies significantly between different genres and composers. Notably, jazz improvisations were found to have shorter transition times, making them less predictable compared to many classical compositions.
A key finding of the study is the identification of transition times where the music shifts from predictable to unpredictable patterns. For instance, in Johann Sebastian Bach’s works, these transitions occur between five and twelve quarter notes, whereas in Wolfgang Amadeus Mozart’s compositions, they range from eight to 22 quarter notes. This suggests that Mozart’s music maintains a longer period of predictability compared to Bach’s, which is more varied and surprising.
The implications of these findings extend beyond mere academic interest. They offer a new lens through which we can appreciate the emotional complexity of music and understand personal preferences. Geisel shared a personal anecdote that connects to these results.
“In my youth, I shocked my music teacher and conductor of our school orchestra by saying that I often couldn’t show much enthusiasm for Mozart’s compositions,” he said in a news release. “With the transition times between highly correlated and uncorrelated behavior, we have now found a quantitative measure for the variability of music pieces, which helps me to understand why I liked Bach more than Mozart.”
The results highlight how data science can contribute to arts and humanities, opening new avenues for cross-disciplinary research and deepening our understanding of how music operates on a cognitive and emotional level. This pioneering approach could also pave the way for future studies exploring other aspects of art and human experience through the lens of data science.