21 Sept 2012

Evolution - Not Just for Carbon-Based Organisms

by L E Carmichael

For about twenty years now, scientists have been studying evolution using a new kind of life form - a digital one.

Digital organisms are programs designed to replicate their own code, and were originally inspired by computer viruses.  Their computer-based habitats are restricted in space (RAM) and restricted in energy (the CPU time required to run the program).   As the organisms reproduce, errors are introduced into their codes at a fixed frequency - the mutation rate.  This results in a genetically variable population which is competing for resources.  And because some variants in the population can reproduce faster, while randomly-chosen organisms are periodically deleted from the system (simulating death), the composition of the population shifts over time.  In other words, the digital species evolves.

There are major advantages to using digital organisms in the study of evolution.  For one thing, complex carbon-based life forms tend to have long generation times, making it difficult to observe evolution in action.  With some exceptions, such as Darwin's finches, scientists are usually restricted to studies of the fossil record, or relationships between living species.  This research is therefore largely observational.  Experiments on evolution can normally be done only on carbon-based bacteria and viruses, and even then, it takes years to propagate the population through enough generations for evolution to occur.  In contrast, digital organisms can go through thousands of generations in just a few hours.

More importantly, because digital organisms share no history with life on earth, they are a completely independent example of evolution.  This allows scientists to distinguish general principles of evolution from quirks that might be specific to DNA-based organisms.

Some of the most interesting results to come out of this research relate to complexity.  In most cases, digital organisms develop faster replication rates by losing code - stripping themselves down to the essential instructions required for reproduction.  In other words, they get simpler over time.  Under computational scenarios, however, simple organisms can obtain numbers from their environment (equivalent to necessary chemicals or food resources).  If they have the right code (genes), they can perform calculations using these numbers (just as carbon-based life forms use enzymes to transform nutrients).  They then release the products of the computation back into the environment.  If those products are deemed beneficial, the organism is rewarded with an increase in fitness - the ability to replicate more quickly.    As mutations which allow new computations are rewarded, what results is a spontaneous increase in complexity over time.  In fact, over the course of several hundred generations, digital organisms evolve that can perform up to 50 distinct calculations.

Another mind-blowing result of this research is the spontaneous evolution of parasites - digital organisms with stripped-down genomes that piggy-back on more complex organisms, using the code of their hosts to copy themselves.  This is exactly what the average cold virus does when it infects a human nose.  What's even more incredible is that the host organisms begin to evolve parasite resistance, launching a digital arms race analogous to that between influenza and annual flu shots!

Digital organisms are a powerful model system, but there is also scientific interest in studying them as life forms in their own right.  I suspect Isaac Asimov would approve.

No comments: