From: adrienne wang (axwang@cs.washington.edu)
Date: Wed Dec 01 2004 - 11:05:23 PST
The evolutionary origin of complex features
The evolutionary theory is hard to prove due to the fact that we are not
able to investigate the origin and keep a complete record along the
development of many organisms. In this article, the authors present an
artificial model, called digital organism, on which they apply the
evolutionary principles so that the programs can self-replicate, mutate,
compete and evolve. They claim that these digital organisms can be used
to examine the evolutionary origin of complex features.
The digital organisms are modeled to replicate by copying commands from
an ancestor, and mutations occur when errors occur during replication.
They develop by performing logic functions and gain energy depending on
the complex level of logic functions they perform. Their experiments
demonstrate that digital organisms follow exactly what evolutionary
theory requires and can serve as experimental evidence to the validity
of the hypothesis. In detail, they start the digital organisms in simple
forms which can’t perform logic functions, and show that they develop
through simple modifications to existing structures and functions and
finally perform the most complex logic function. They also show that one
complex function could be produced through many different paths.
However, the applicability of the experiments is somewhat restricted.
The digital organism can never replicate the exact developing paths of
biological organisms. The logical functions and rewards associated to
them are not comparable to the complexity of natural selection. In
nature, organisms could face more complicated environment and the same
mutation could have different results under different circumstances.
Their experiments also make us wonder how the digital organisms would
develop if the rewards to those simple logic functions are modified.
There are several questions raised for future research. They mention in
the article that the consequences of asexual and sexual reproduction for
the evolution of complex features deserve some attentions. Their model
seems to be an existing model for genetic algorithm. This looks like a
parallel search model with rewards being the heuristic and performing
EQU being the goal state and all mutations being the search space.
This archive was generated by hypermail 2.1.6 : Wed Dec 01 2004 - 11:05:23 PST