Evolution Of ALife

evolveArtificial Life (ALife) is an area of research, which has its origins in the 1980's. The initial remit of much of this research was to simply model and understand the fundamental mechanisms that seems to underpin life.

“...the principle assumption made in Artificial Life is that the ‘logical form’ of an organism can be separated from its material basis of construction” C.G. Langton

Within these initial confines, Alife is an 'organism' that is essentially imaginary in that it only 'existed' as a software construct within a computer-generated world that defined an equally artificial ecology in which evolution of the organism might take place. Therefore, within the confines of these models, the adaptive behaviour of an Alife organisms was often based on the operation of neural networks, the electronic type not biological, while the evolution was based on software algorithms that mimicked some subset of known genetics, e.g.

  1. Create a initial population
  2. Evaluate the fitness of each organism
  3. Link fitness and ability to produce offspring
  4. Evaluate species based on overall population growth

Evolutionary Theory

Within this initial research context, evolution was modelled on the trial-and-error process of natural selection. This not only made reference to Darwinian theory, but also tried to parallel some of the basic mechanisms of biological evolution. Therefore, at this point, we need to clarify any inference that might be associated with 'natural selection' as opposed to 'artificial selection'. Initially, we might characterise artificial selection, not in terms of some abstract computer algorithm, but rather as a more real example of a farmer breeding cows for greater milk production. Within this context, natural selection is only 'natural' in the sense that there is no obvious artificial 'cognitive system' making the selection. As such, natural selection is a by-product of the system itself, which appears to operate without plan or design, but still seems to lead to an evolutionary process that results in ever-greater complexity, although we should be cautious about this last assumption. As such, much of the early work within Alife was focused on trying to understand the basic mechanisms that led to the evolution of the organism, at least, in terms of diversity and population. However, over time, there was a gradual shift towards modelling the evolution of cognition abilities and, today, many Alife evolutionary models are actively researching the notion of learning, via neural networks, and exploring the idea of adaptive behaviour. In this respect, there was a convergence of research in the Alife field and AI in terms of the computer sciences and the cognitive sciences.

While this may well be the focus of mainstream Alife research, it does not preclude the notion of artificial life evolving via other means. If our definition of Alife is predicated on whether evolution is driven by natural or artificial selection, then any major change to the homo sapien blueprint, which is carried out by design, is a form of Alife. The following subsection will introduce just such an idea.