Genetic algorithms and neural networks are two tools in a growing
arsenal of computational methods being applied to fundamental issues
of theoretical biology. Much of this work is now being called
``artificial life'' (ALife). I was asked by the editor of IEEE
Expert for a survey article that motivates this area for its
pragmatic audience of AI expert system engineers
[6][24].
I argue here that there are striking
parallels between AI's investigation of intelligence and ALife's
investigation of life. Further, ALife represents an important new
``grounding'' of our notions of intelligence in fundamental abilities
that arise directly from what it means to be alive. I have also
focused on the special role computer science has to a science of
ALife, above and beyond simply toolsmithing for biologists
[8].
A recurring theme in much ALife work is the ``evolution of complexity,'' i.e., the development of progressively more elaborate adaptive responses to increasingly difficult environments. A central research issue, therefore, is a careful characterization of what it means to be a more complex environment. Towards this purpuse, we have developed a simulation environment we call Latent Energy Environments (LEE) [38] and used it to investigate to important interactions between the cognition of individuals and the evolution of populations: evolution of the sensors by which individuals perceive their world [39] and evolution of ``life history'' characteristics such as when an organism is most susceptible to imitative learning and when it becomes reproductively mature [29].