What transforms noise from album filler to dancefloor killer? Why do some tracks turn us on while others make us tune out?
DarwinTunes, a computer program that employs the principles of natural selection and sexual reproduction (see below), aims to show how music evolves from noise to aesthetically pleasing music.
The authors of the related paper, Evolution of Music by Public Choice, are to be congratulated on developing a model that convincingly “shed(s) light on the evolution of real musical cultures”.
Such an approach adds to the growing stock of material that addresses music as an evolved phenomenon: both in terms of its origination as a human characteristic; and as a memetic system – that is, a cultural practice of social behaviour passed on through generations, such as fashion.
But there are problems. The design of DarwinTunes mimics genetic evolution. But music is quintessentially memetic, and apt to fluctuate, where biological evolution cannot as easily put itself, as it were, into reverse; and do so at accelerated rates.
Know your meme
Within the meme “hairstyle”, a population can switch from mullet to Mohican and back at runaway/runway speed. Hairstyling suggests a key property of the interface of biological and cultural evolution that has not been captured by DarwinTunes.
The model the authors have devised adopts a largely sexual selection premise, based on the information theory approach of broadcast-receive, modelled as produce-consume.
A significant role in the evolution of human music, at least where its origins as instinctive vocalisation with ritual connotations are concerned, is argued by authors such as Bjorn Merker, Bruce Richman, Robin Dunbar and Iain Morley. This relates to the capacity for simultaneous chorusing, whereby arousal and emotional empathy are recruited through performance in which pitch and rhythm are identically entrained.
This is seen as universal in human societies, and forms a significant means for the learning, teaching and sharing of song, affecting both the preservation of the known and the capacity for innovation. Such vocalisation was – and remains so in hunter-gatherer societies – rarely separated from dance.
Darwin dancing
To illustrate this point, let’s imagine a computer model that generates, in a manner similar to DarwinTunes, the transmission of choreography: let’s call it DarwinSteps. We would anticipate that some step patterns would be learnt through observation (produce-consume). A significant mover both in the means by which steps were perfected according to a model and as importantly, motivation applied to expend energy on the activity, would be simultaneous group performance.
As with our hairstyle example, the object is looking identical to other members of the community: what Merker has termed “conformal motive”.
So, sounding identical, in unison performance, also has an observable function in the evolution of music in social settings.
Could this be captured in a model such as DarwinTunes? How would this be done? And what might it tell us in comparison with the current version?
Play it again, Sam
A checking procedure that mimicked the condition “repeat that identically” would double as a kind of approval, both in music and in dance. Of course, it could also propel what is planned as an evolutionary procedure into a closed loop in which the same pattern is reiterated ad infinitum – a bit like some modern dance music!
Perhaps this is what produces “classics”? But even classics give rise to arrangements and cover versions.
It would nevertheless tell us more about the naturalistic transmission of musical material were a social selection element to be factored in: an algorithm for “performativity” and repetition by acclaim.
Two other factors come to mind: the receiver/consumer within the current model biases selection towards Western Popular Music; this is evident both in the rating by a web community and in the adoption of the feature detection programme Chordino, assumed to represent the biases of such listeners.
What would happen if the population responding to the material drew predominantly on other cultures? It would be a valuable test of DarwinTunes to expose it to this condition.
Musical mutations
A further thought comes to mind. Namely, the “progressivist” tendency whereby what counts as music is what is available in our current generation: equally tempered, with all instruments and voices able to carry material at all available pitch levels to which mutation might transpose them. But this is a decidedly post-modern condition for music to find itself in.
Brass instruments in the Classical period were limited principally to the notes of the [Harmonic Series](http://en.wikipedia.org/wiki/Harmonic_series_(music) so random mutation of music they “played” would have to take this into account lest listeners recognise “impossible” examples.
In turn, the quality of the sounds humans encounter relates strongly to the environment and technology available to them.
Some cultures have resonant metal instruments, others do not. Some have access to gourds and bamboo from which rudimentary wind instruments of great expressivity can be made, as explored by Pedro Espi-Sanchis.
The Inuit are limited to duet singing (see video above), which they achieve with great ingenuity but in a manner that DarwinTunes as currently configured would be unable to compute.
All these ideas arise in response to this fascinating project. What they illustrate is that speculation on the origins of the human capacity for music – perhaps the holy grail of evolutionary musicology – needs to take into account social and environmental factors responsible for what can be performed. Perhaps this is beyond the grasp of a model that begins with the synthesis of heard results.
Further reading:
Artificial selection on music and on rappers – Rob Brooks