Ordered Asynchronous Processes In Natural And Artificial Systems
David J. Cornforth, David Geoffrey Green, David Newth, Michael Kirley · 2001
Models of multi-agent systems with fixed network structure usually update the states of all agents in synchronous fashion. Examples include Cellular Automata, Random Boolean Networks and Artificial Neural Networks. Some recent studies have shown that the behaviour of such models can change dramatically if random asynchronous updating is used. Here we show that many real systems, both natural and artificial, undergo updating that is asynchronous, but ordered in some way. We use examples to demonstrate some of the properties of ordered asynchronous updating in Lsystems and cellular automata. In many cases, models of such processes effectively hide both their asynchronous nature, and the ordering, by embedding them in the model's details. This practice has prevented earlier recognition of ordered asynchronicity as well as some important implications. Among these implications are its role in the rise of modularity within complex systems. As an example, we introduce the “spotlight model ” of gene regulation, a random Boolean network in which controller nodes create modules by unfreezing different sets of nodes in turn. We argue that such models are not only more realistic representations of nature, but have potential advantages for solving complex problems.