Representation of Information in Spatial Maps which are Produced by Self-Organization

Abstract

There are processes of very different kinds which have been discussed under the heading of self-organisation; adaptive phenomena in neural-like networks constitute a subset of them. In the original sense, self-organization obviously meant autonomous generation of stable complex forms that appear especially in biological organisms. On the other hand, it is frequently maintained that structures of information, or rather its representations, may somehow evolve in similar autonomous processes. Based on results recently acquired in the research of adaptive processes, a different view is held by this author: it is demonstrated in this paper that at least the first step in the formation of representations of information may be generation of simplified images or abstractions of the observable world. The original model of such a process is in the biological brain; if the conditions described in this work are valid, such abstractions may be formed in artificial systems, too.