Relation to other work

The relation to the evolution of biological species is clear, however it is highly simplified. There is no interaction between creatures and goals are static. No actual learning occurs during a lifetime: only changes to the input of neural networks, no changes to the structure. However, knowledge is accumulated across generations.

There are many examples of research into Artificial Life3.5. Thomas Ray's Tierra shows an ecology of programs competing for resources. Parasite-host, parasite-hyperparasite relationships arise, etc. Ray's Tierra is an example of 'programs' that evolve native to the digital medium. The former example illustrates, by analogy, the evolution of insect-like3.6 creatures. While this example makes use of both ANNs and EAs, all behavior is purely instinctive. While some individuals may appear to act intelligently, this is certainly not consciously. Of any individual it would depend on chance if it would cope well under novel circumstances.

Erik de Bruijn 2007-10-19