Evolution Produces Tasteful Tunes

Crowdsourcing music composes pleasant sounds.
Image
Music notes
Media credits

 photosteve101 via flickr

Media rights

Image rights: http://bit.ly/cGotEb

Joel Shurkin, Contributor

(ISNS) -- In music, as in evolutionary biology, it could be the survival of the fittest.

Scientists, using loops of computer sounds as building blocks, found that by obtaining small contributions from many people, or crowdsourcing, they can create pleasant-sounding music, just the way biology produces adapted organisms.

In a paper published in this week's Proceedings of the National Academy of Sciences, Bob MacCallum of Imperial College, London, and his colleagues demonstrate how evolutionary pressure from audiences can shape music.

"What interests us is not that we can do a composer's job," he said. "What interests us is that we can quantify the consumer or listener's creative role."

Music research is not MacCallum's day job. He actually is an expert on the genetics of mosquitoes. He also loves music.

His paper is full of biological references, describing how music is changed by the same mechanisms organisms and languages use to evolve.

MacCallum wrote a program called DarwinTunes and posted it on the web.

DarwinTunes began with a series of sound loops, eight seconds long, consisting of sounds produced by computer from sine waves -- not very natural sounding. Any melodies are purely coincidental.

The researchers did set limits. MacCallum said. They designed the loops to have four beats to the bar, to continue for four bars, and used the conventional Western 12-tone scale so it would not sound alien to their western subjects.

The sounds are entirely artificial, MacCallum said, because if he used real instruments the personal preferences of the listeners might alter their decision. There are people for instance who hate accordions. 

Participants in the study vote on a scale of one to five on whether they liked the loops or not. Every time 20 loops have been rated, the top 10 are paired with each other randomly and the bottom 10 dropped from the program.

"If it's good music it survives, makes babies, and if it's bad music according to the audience, then they just die off," MacCallum said.

The loops periodically replicate (MacCallum calls it "having sex") and produce new loops ("babies"). The second generation is not identical to the first because they are formed by the combination of two loops and there is some random mutation, just like in life.

At any time, there are 100 loops in the program.

So far, he reported, 6,931 listeners made 85,533 ratings over the course of 2,513 generations. In all, 50,480 loops were created. A few short melodies accidentally evolved. The results sound something like the kind of minimalist music that gained attention from critics toward the end of the last century.

But, as each generation evolved, the changes diminished, which became noticeable about a third of the way in, and the result now appears to have stopped evolving.

MacCallum said this is mirrored in nature. There are some species of reptiles, for instance, which essentially haven't evolved in 100 million years, probably because they didn't need to for survival. Their environment stabilized. Also, there were a lot more of them than tones in his program and with large populations; you need only a small fitness advantage.

MacCallum said they were changing the program to take it off its plateau and resume evolution.

The process also happens in life. A thousand teenaged bands practice in garages, he said, and sometimes a few are so good they are encouraged to play publicly. And the really good ones become popular around the country and the world. Their success breeds imitators.

Is what the program produced real music?

Seems like a stretch, says Sharon Levy, who teaches music literature at the Juilliard School in New York and the Peabody Institute at Johns Hopkins University. 

"Certainly, as the authors note, this kind of program can only evolve so far. That's partly due to the limited choice within the program itself, as it's about very short segments of musical material, whereas some of what's considered the most important musical contributions, at least in the West, has to do with how such materials are arranged, manipulated, and transformed across a much larger unit of musical time," Levy said.

"Another issue," Levy said, "is that everyone -- brilliant musician or not -- plays an equally limited role. If the authors wanted to continue the evolutionary model, then perhaps folks like Bach or Beethoven might be seen as kinds of musical meteors who dramatically altered the evolutionary process, which, if nothing else, is a nice metaphor."

MacCallum thinks it's music within a limited definition.

"It's a fact we've made music that is more pleasing to western people," said MacCallum. "But more pleasing is not the same as a Mahler symphony."

Author Bio & Story Archive

Joel Shurkin is a freelance writer in Baltimore who has also taught journalism and science writing.