Can Darwinian Mechanisms Make Novel Discoveries?: Learning from discoveries made by evolving neural networks
Author: Pennock, R.T.
Source: Foundations of Science, Volume 5, Number 2, 2000 , pp. 225-238(14)
Abstract:Some philosophers suggest that the development of scientific knowledge is a kind of Darwinian process. The process of discovery, however, is one problematic element of this analogy. I compare Herbert Simon's attempt to simulate scientific discovery in a computer program to recent connectionist models that were not designed for that purpose, but which provide useful cases to help evaluate this aspect of the analogy. In contrast to the classic A.I. approach Simon used, ``neural networks'' contain no explicit protocols, but are generic learning systems built on the model of the interconnections of neurons in the brain. I describe two cases that take the connectionist approach a step further by using genetic algorithms, a form of evolutionary computation that explicitly models Darwinian mechanisms. These cases show that Darwinian mechanisms can make novel discoveries of complex, previously unknown patterns. With some caveats, they lend support to evolutionary epistemology.
Document Type: Regular Paper
Affiliations: Lyman Briggs School and Department of Philosophy, Michigan State University
Publication date: 2000