%0 Journal Article
%T Modelling Biological Systems with Competitive Coherence
%A Vic Norris
%A Maurice Engel
%A Maurice Demarty
%J Advances in Artificial Neural Systems
%D 2012
%I Hindawi Publishing Corporation
%R 10.1155/2012/703878
%X Many living systems, from cells to brains to governments, are controlled by the activity of a small subset of their constituents. It has been argued that coherence is of evolutionary advantage and that this active subset of constituents results from competition between two processes, a Next process that brings about coherence over time, and a Now process that brings about coherence between the interior and the exterior of the system at a particular time. This competition has been termed competitive coherence and has been implemented in a toy-learning program in order to clarify the concept and to generate！and ultimately test！new hypotheses covering subjects as diverse as complexity, emergence, DNA replication, global mutations, dreaming, bioputing (computing using either the parts of biological system or the entire biological system), and equilibrium and nonequilibrium structures. Here, we show that a program using competitive coherence, Coco, can learn to respond to a simple input sequence 1, 2, 3, 2, 3, with responses to inputs that differ according to the position of the input in the sequence and hence require competition between both Next and Now processes. 1. Introduction The quest for universal laws in biology and other sciences has led to the development！and sometimes the acceptance！of concepts such as tensegrity [1], edge of chaos [2, 3], small worlds [4], and self-organised criticality [5]. This quest has also led to the pioneering ( ) model developed by Kauffman in which is the number of nodes in an arbitrarily defined Boolean network and is the fixed degree of connectivity between them [6]. The actual use of the ( ) model to the microbiologist, for example, is that it might help explain how a bacterium negotiates the enormity of phenotype space so as to generate a limited number of reproducible phenotypes on which natural selection can act. Although the ( ) model successfully generates a small number of short state cycles from an inexplorable vast number of combinations！which might be equated to generating a few phenotypes from the vast number apparently available to the cell！the model has its limitations for the microbiologist as, for example, it has a fixed connectivity, it does not evolve, and it does not actually do anything. In a different attempt to find a universal law in biology, one of us began working on the idea of network coherence in the seventies. This idea is related to neural networks (though the idea was developed with no knowledge of them) which have indeed been proposed as important in generating phenotypes [7]. The network
%U http://www.hindawi.com/journals/aans/2012/703878/