Judgments of musical quality (or value) are the product of the complex ways in which our brains interpret sound, heavily modulated by social and cultural pressures that bombard us throughout our lives. The success of individual musical ideas over time is largely determined by the ways in which ideas (or memes) are introduced and molded into the greater fabric of culture. In some ways this process is analogous to Darwinian natural selection; only the best mutations survive.
It’s worth noting that in order for us to observe how biological traits are introduced and adapted over time, it’s not necessary to dissect every offspring that ever existed on earth. We only require a sample of data from a small number of representative offspring at various stages of the evolutionary process to know how fish gills adapted into air bladders which became lungs etc. By studying these changes over time, we can connect the dots and determine which genetic mutations offered the biggest advantages.
From Darwin to Beethoven: Only the BEST Ideas Survive
A related process can be traced through artistic evolution. For example, musical innovations developed by Haydn are found in the work of Dussek, who directly influenced Beethoven, who laid the foundations for the harmonic and formal innovations of Schubert and Brahms, and so on.
An interesting side observation is that while artistic agency is needed to generate mutations, cultural evolution doesn’t require the agent to be at all aware of the larger process at work. Given the right conditions, it’s completely possible for an artist who is unaware of what has come before to have a significant influence on future generations. However, if the artist is interested in the possibility of creating work that offers long-term cultural value, it’s clearly a benefit to be well-educated in the artistic forces that have preceded them.
Musical value is a form of musical information that exists on a spectrum. This means that there are choices available to musicians that will produce better or worse musical experiences for their audiences. Another way of saying this is that we can avoid creating the worst musical experience possible by leveraging accumulated value data acquired through the analysis of cultural artifacts. As a collective it seems we do have standards after all.
The Search for Universal Musical Value
Given the accumulation and analysis of enough musical data, we can determine key aspects of influence throughout a given corpus. The influence of an individual work can be measured by analyzing its repeatability over time (how frequently and widely was/is it performed?) and the degree to which its unique features are found (or abandoned) in later works.
Additionally, 10,000 years of cultural evolution can point us to which musical elements most frequently drive musical value. Scholars like Trehub, Demany, Cohen, and Thorpe are investigating this and have begun to tease out guidelines that may point to possible universals.
| Primary | Secondary | Tertiary |
| 7 or fewer scale degrees descending or arched melodic shape non-equidistant scale divisions 2 or 3 beat subdivisions | motivic patterns (w/fewer than 5 durational values) limited (few) durational values phrase repetition metrical hierarchy | small intervals isochronous beat short phrases (<9s) chest voice (tesatura) |
So what can we take from these early observations? Given how frequently and consistently they show up in recorded music history, these limitations may well represent foundational principles that underpin our general musical experience. More specifically…
Perhaps culturally evolved music favors certain limits on the number of pitches and their duration, the existence of identifiable motivic patterns, limits on phrase length and register, and the need for easily identifiable repetition and hierarchy.
If our goal is to develop intelligent music AI systems that have a strong chance of exhibiting high levels of musical value, we must look to identify the successful aspects of musical information (memes) that have been preserved and adapted (repeatedly altered and expressed) by our evolved musical culture, and welcome their influence in the design of our AI algorithms.