Artificial Life (AL) algorithms have been extensively used in the last two decades for Machine Learning tasks [9], [10], both in the symbolic and in the subsymbolic paradigm. As a matter of fact, genetic algorithms have proved to provide good solutions to a wide class of Knowledge Acquisition (KA) problems, often independently on the representation used.
The most common ``justification'' for this serendipitous coincidence of techniques between conceptually different disciplines is that - when there is no strong knowledge about the domain in which the learning has to be performed - genetic exploration of the hypothesis space happens to be the best choice, precisely because it does not require previous assumptions about the domain (while in particular cases where more information is available about the problem, domain-specific heuristics can be used more reliably).
This is not an explanation, as evident, but a simple observation. The epistemological question to be answered is: Why does the simulation of a biological process like natural evolution lead to the emulation of a cognitive process like learning? Is this just a coincidence, or does a deep conceptual explanation for it exist?
A possible answer to this philosophical question is provided by an extension of Evolutionary Epistemology (EE), which will be shortly outlined in the next section. This is a general model of Knowledge Acquisition processes, proposed by the late Karl Popper, and happens to account for a number of different phenomena in many distinct fields, where the acquisition of knowledge is the central aspect. It applies to cases where no previous knowledge is available, and a central point of Popper philosophy is that it is impossible to justify any of our prior assumptions.
Genetic or Evolutionary Learning Systems will be shown to be straightforward implementations of this simple model, which has the advantage of being conceptually and philosophically well grounded, and indeed can be considered the only reasonable one from a certain point of view [16].