Long-Term Evolutionary Dynamics in Heterogeneous Cellular Automata

It is a truism that the complexity of life increases with time, even if the means or measures of it are controversial. Recreating this effectively endless self-generation of new mechanisms and capabilities would be fascinating for its insight into our own origins, and enticing as a harnessable creative force. In this paper, we introduce a new artificial life model based on a discrete dynamical systems framework. We have extended classical 2D cellular automata (CA), our chief modification being the allowance for heterogeneous transition functions. These changes convert a CA system into a new kind of “ecosystemic” model, where different genomes compete for existence. The value of such a model resides precisely in its simplicity: we aim to observe that long-term dynamics can be achieved quite naturally, given appropriate and plausible hypotheses. To demonstrate this open-endedness, we will show that our device is capable of supporting long-term dynamical behaviour, more so than control systems such as homogeneous CA and the Game of Life. In particular, we will exhibit examples of the strategies discovered by our system, which are characterized by the emergence of competitive behaviour.