As a part of the Cultural Analytics Workshop Week, the Berkeley Institute for Data Science, the School of Information, and the Department of Scandinavian welcomed Noah Askin from UC Irvine to present his research on creativity as a process, and which mechanisms inspire creativity. His research focuses specifically on the music industry, and the forces and outcomes that influence the creative process.
What is creativity? Askin defined creativity as the “process of generating something novel that is deemed useful or appropriate.” Using datasets from Spotify and Deezer, Askin explored the origins of creativity by evaluating musical output. Earlier research in creativity emphasized that the primary source of creative inspiration was individual differences. However, in recent years, research into the effect of social structures on creativity has gained traction.
Mechanisms
As Askin explained, one perspective of the origin of creativity is that collaboration and the creative material shared via an artist’s network can enhance a person’s creativity. This mechanism is known as recombination, where others’ ideas influence an artist. Another view holds that creativity is enhanced by collaborating directly alongside other creative individuals. This mechanism, called stimulation, implies that others can push an artist to be more creative.
Spheres of Influence
In addition, Askin described how larger cultural context could influence creativity. While the mechanisms of recombination and stimulation both describe direct sources of influence, categorial affiliation, or genre, may influence artists as well.
Photo: Askin uses Bob Dylan as an example of cultural and social influence.
For example, in music, genres are defined by certain norms and audience expectations. The goal is to understand how the boundaries of genre influence the creative output of artists. Askin used the example of Bob Dylan. Did playing with other artists, such as Johnny Cash, influence Dylan? What inspired Dylan to experiment with genre and ultimately create his rock album, Bringing It All Back Home?
Linking Mechanisms and Spheres of Influence
Askin then presented a 2x2 matrix to link the mechanisms of creative influence with the social spheres of influence. Askin’s work further examines how each quadrant of this matrix impacts creative output.
Image: Askin’s 2x2 matrix linking the mechanisms and social spheres of creative influence
Askin and his co-authors compiled a global music dataset of millions of songs released from 1955-2000, including song and artist level data, from The Echo Nest, Spotify, and MusicBrainz. This dataset included objective sonic attributes, such as tempo and musical key, as well as subjective measures such as danceability, acousticness, and energy.
Then, they determined how to measure creativity. For each given song in the dataset, the team developed equations to define similarity and novelty across relevant song-pairs. The measure of novelty is what the team defined as “creativity.” They found that the distribution of creativity skews right, indicating that a lot of music tends to sound and look familiar to other previously released music.
Askin then explained how the definition of both the social and cultural “spheres” were operationalized in their research. The social sphere included active artists who shared track credits with the focal artist. The cultural sphere included active artists sharing genre affiliations with the focal artist. He also explained how they defined recombination and stimulation, using collaborators and genre peers for focal artists to mathematically calculate recombination and stimulation as a part of the 2x2 matrix.
The Origins of Creativity
After running the models, Askin shared the team’s results. They found that cultural stimulation, or a “genre’s creative inspiration and constraints,” has a 13x greater influence than other sources of creative influence. This means that an artist’s “ability to be creative or not” is heavily dependent on the genres they are a part of. Association with a certain genre can influence how experimental an artist can be, and what is permitted musically within that genre - both factors that influence creativity. Other significant effects the team noticed are more novelty with female artists, and more novelty in the classical and jazz genres. While the team found evidence of both recombination and stimulation, cultural stimulation was the most “powerful predictor of song novelty.”
Photo: Askin explains song clustering from the dataset.
The goal of providing this framework is to help people take a more holistic perspective to understand how creativity works. In addition, using algorithms decreases some bias in analyzing results. Askin also explained some of the limitations of his research. This framework doesn’t provide a singular theory to explain all results. In addition, they have only focused on acoustic features, and a perspective from the production side. The next steps of his research include trying to unpack other specific mechanisms within the matrix. Another extension includes incorporating other metrics, such as audience evaluations and markets.
Creativity does not occur in a vacuum - rather, social, cultural, and personal influences all have varying impacts on an individual’s output. By exploring these mechanisms more fully, we can understand how production is impacted by these different forces.
Watch the recording of Askin’s talk on the I School website here. To stay in touch and join these far-ranging conversations of critical cultural importance, please join our Cultural Analytics mailing list by visiting this page or emailing bids-cultural-analytics+subscribe@lists.berkeley.edu