Studying the Emergence of Multicellular Organisms in Artificial Life

ABSTRACT
Traditional systems building often consists of using a top-down approach. That is, conceptualizing the overall organization for how a system should operate and then breaking it down so that it fits into logical pieces. Each of these pieces is then divided again and again until manageable units are developed in detail. There is some belief that in order to build very complex systems, such as applications that are able to learn and repair themselves, a bottom-up approach must be taken to systems design. System can be built with small simple components that together organize to function as a complex system. This is similar to how nature creates complex organisms out of cells. Each cell is a relatively simple unit with a minimal instruction set, which is capable of organizing with other cells to create a large complex organism.

This paper studies the emergent properties of single celled organisms and how in an evolutionary simulation, multicelled organisms might originate. This study may give insights into emergent systems, and will hopefully aid in understanding how these systems can be harnessed and created. Experiments are performed in an artificial life simulation on a computer using Cartesian Genetic Programming. These experiments produced populations of cells, which exhibited novel behavior. Sometimes, interesting group behaviors emerged which showed properties of multicellularity.

DOWNLOAD
Report: Download in pdf format (600 KB)

Published as a peer reviewed paper at UKCI 2002, Birmingham UK. Also presented as a late breaking paper at GECCO 2002, NYC. UKCI'02 Paper: pdf (100 KB) UKCI'02 Presentation: pdf (300 KB)