Nov 30

Reported by Brandon Keim in November 29, 2010 in Wired Science

One hundred and fifty-one years after the publication of On the Origin of Species, digital creatures have evolved to communicate like fireflies in a computer program that blurs the boundaries of life.

Recorded in line-by-line detail, their development in a software platform called Avida may provide insight into biological behavior and inspiration for the design of distributed computer networks.

“Evolutionary programs have been around for a while, but we haven’t seen them applied to distributed computing,” said computer scientist Philip McKinley of Michigan State University. Synchronized communication can be “seen in the natural world. But in Avida, we can go back to how and why it evolved. We can see the key points that allowed this relatively complex behavior to emerge.”

The new synchronization findings, made by McKinley and fellow MSU computer scientist David Knoester, were published November 18 in Artificial Life.

Inside the program, developed in the early 1990s at the California Institute of Technology and refined at MSU’s Digital Evolution Laboratory, digital organisms called Avidians take the form of self-replicating code. Their genomes are written in assembly language and stored in separate regions of memory, executed again and again at electronic speeds. Programmers set the parameters of mutation and natural selection, and evolutionary principles manifest themselves in silico.

“We like to say ‘it’s not a simulation of evolution, it’s evolution.’ The difference is that these are computer programs,” McKinley said.

In a previous and well-known study, researchers supported a key tenet of evolutionary theory by demonstrating how easily complexity could emerge in Avidians through incremental changes in simple, existing functions.

McKinley and Knoester specialize in organismal interactions: How complexity emerges not only in individuals, but also in groups.

Their earlier work examined the evolution of collective perception, cooperation and decision making. In the new study, however, they emphasized communication and selected for groups of Avidians that best synchronized their flashing with others.

Fireflies, which coordinate their blinking across distances spanning miles, are the best-known synchronized communicators of the biological world. How they do it isn’t fully understood, but Knoester said “it was literally a three- or four-line change” in Avida.

Crucial to Avidian synchronization was the handling of the computational version of “junk DNA,” or genetic code that seems to have no apparent purpose. In biology, junk DNA is now appreciated as having crucial regulatory functions. In the Avidians, individuals evolved to change their flash timing by adjusting the speed at which “junk” instructions were executed.

McKinley and Knoester don’t think that fireflies necessary synchronize the same way, as Avida provided a computational and likely different route to the same outcome. More importantly, it gave the researchers algorithms they would not have otherwise imagined.

The algorithms could inspire functional code beyond Avida’s confines.

“Avidians build network topologies. What sort of topologies do they come up with that are robust to damage, if the routing nodes fail?” Knoester said. “We’re also collaborating with a professor in the electrical engineering department who works on robotic fish. We’re not really interested in schooling; we want robots to track oil slicks, to monitor water quality. To do those things, you need to stay connected.”

As for the upper limit on Avidian complexity, “I’m not sure we know yet,” Knoester said.

Video: Organisms in Avida, a software platform for artificial life, running their genomic instructions. Eventually they evolve to flash in synchrony, like fireflies./Philip McKinley and David Knoester.

Comments are closed.

preload preload preload