EPSRC Network on Evolvability in Biology & Software Systems
Evolvability, Genetics & Development in Natural and Constructed Systems: Abstracts of the EPSRC Evolvability Network Symposium
Tewin Bury Farm Hotel, Hertfordshire, England, UK
C. L. Nehaniv, P. J. Bentley & S. Kumar (Editors)
CTRNN WORLD: A Model of Open-ended Evolution in a Dissipative System
We present a model of open-ended evolution in a thermodynamically constrained ecosystem consisting of agents that are continuous time recurrent neural networks (able to map any function). Agents interact with each other by subtraction of wave forms, so effectively the entire ecosystem is a large CTRNN. Agents pass energy between each other and exist in a non-spatial environment allowing complex non-transitive interactions. We explore whether the system will evolve to maximise the rate of dissipation of energy (See Swenson, R. and M. T. Turvey (1991). ``Thermodynamic Reasons for Perception-Action Cycles." Ecological Psychology 3(4): 317-348 & Kauffman, S. (2003). ``Molecular Autonomous Agents." Phil. Trans. R. Soc. Lond. A 361: 1089-1099.) as observed in Benard cells. The tendency of physical systems form ordered structures to maximise the rate of energy dissipation been hypothesised to be a constraint favouring the production of order in open-ended evolutionary systems. Our Preliminary results show that ``Cannibalistic Altruism" (See Kerszberg, M. and M. Wolpert (1998). ``The Origin of Metazoa and the Egg: a Role for Cell Death." Journal of Theoretical Biology 193: 535-537.) evolved. This is a primitive form of multicellularity. The ultimate aim of the model is to allow higher levels of organisation to evolve indefinitely, something that has not been possible in Tierra, Avida, ECHO, Polyworld, or COSMOS.