Evolution by Aesthetic Design

By Maryam Zaringhalam, @thisisartlab

evolution analogiesBiological evolution is the change in gene frequencies in populations over successive generations through forces like mutation, natural selection, and genetic drift. But at its most basic conceptual level, evolution is simply change over time. Since life is not stagnant, but perpetually moving forward, we can make analogies between evolution and just about anything we experience. But how can we use these analogies to glean something meaningful about our experiences?

In an experiment called DarwinTunes, bioinformatician Robert MacCallum at Imperial College London put the analogy into practice in an attempt to evolve music from noise.  By applying basic evolutionary principles, he hoped to gain some insight into what aural // aesthetic forces underlie audience experience of music.

For musical evolution to proceed, MacCullum and his team first generated a population of noises—the origin for [Darwinian] musicality to come. Because the origin of life was devoid of any human intervention, they used an algorithm to generate a series of computer programs, or “digital genomes,” thereby limiting their influence on the generative process. Just as our DNA genomes hold all the information needed to build us, each program specifies how to build a particular short sound loop by determining where and when particular notes are played and what instruments should play them.

But, for evolution to occur—for these loops to meld and morph into something new and different—this aural system needed some selective pressure acting on it. Instead of evolution by natural selection, MacCullum’s group recruited almost 7,000 volunteers to lend their ears for an experiment in evolution by audience selection. Volunteers listened to and rated sound clips for their musical appeal on a scale of 1 [can’t stand it] to 5 [love it!] and the top 50% were chosen to pair off and “mate” with each other—”survival of the funkiest.”

During these sonic sexual encounters, bits of the computer programs that contain the information needed to make the sounds get exchanged, shuffled around, and even mutated to make the next generation of sound loops. Remarkably, the quality of the loops increases rather rapidly in the first 900 generations, transitioning from nonsensical noise to what could qualify as music. And all without a composer! Of course, the results don’t hold a candle to Beethoven or the Beatles. [In music, intelligent design // divine intervention still prevails!]

Nevertheless, what emerges resembles musical sounds we’re more accustomed to. In fact, these loops amazingly exhibit a rhythmic complexity and share chords commonly used in popular music. Unconsciously, audiences were picking sounds that converged on music that we hear all the time! [1]

Of course, at some point music [and art in general] becomes a subjective experience where enjoyment relies on individual aesthetic tastes. Consequently, after those first 900 generations, the average musical appeal of the sound loops level off at an average rating of about 3 and not the max rating of 5. [2]  MacCullum poses a few detailed explanations for this leveling off, but to me [jargon aside], this aural plateau results from a mass consensus of sorts that defines the baseline is for musical enjoyment – a stabilizing selection by consensus. Because the sounds are generally more pleasant and familiar to the listener, the individual’s taste begins to play more of a role in selection. As a result, the average musical appeal remains the same, while individual musical appeal is all over the place. [3]

That great, mighty current of evolution which is advancing the life of everything in creation is simply invincible—no one can resist it. -Maharishi Mahesh Yogi

By employing patterns found in nature, MacCullum’s group was able to observe the emergence of patterns [however much we may take them for granted] found in music: the emergence of certain common chords, rhythms, musical structures. For me, then, the truly amazing part about MacCullum’s experiment in music synthesis is really not that his group was able to evolve music from randomly generated and rather unpleasant noises. Instead, it’s that they were able, quite remarkably, to apply the most pervasive natural phenomenon—that thing which is responsible for life on Earth as we know it—to study, and even replicate, a musical phenomenon.

This post was originally featured on Maryam’s blog, ArtLab. To read more of her thoughts on the intersection of art and science, check out http://thisisartlab.com.


[1]  Of course, this is not wholly surprising considering that our ears have been selected  to be more receptive to certain kinds of musical sounds; however, for these sounds to be so pervasive they must have some intrinsic appeal to us [sometimes even causing us to release the feel good neurotransmitter dopamine on listening].
[2]  Results from this experiment in music synthesis are detailed in: Robert MacCullum et al. “Evolution of music by public choice.” PNAS: 2012.
[3]  To create your own individual evolutionary music, check out digital artist // programmer // researcher Jeffrey Ventrella’s musical gene pool!

1 thought on “Evolution by Aesthetic Design”

Join the conversation!