The University of Sussex

Enact: an artificial-life world in a family of cellular automata

Pedro Paulo Balbi de Oliveira

Enact is a family of two-dimensional, non-deterministic cellular automata, whose temporal evolution on a periodic background can be described in terms of the metaphor of an artificial-life world where a population of worm-like organisms of arbitrary length undergo a coevolutionary process. During their lifetime, the organisms roam around, sexually reproducing, interacting with the environment, and being subjected to a developmental process which includes ageing and death. An organism is formed by a sequence of contiguous cells, so that the cells at each end can be intuitively thought of as its head and tail, whereas the cells in between constitute its body. One single cell of the body is the organism's genotype; the others are phenotype-like cells, since they are subjected to change through environmental interaction. The gene separates the phenotype-like cells into two parts: a single cell between the head and the gene, which we consider the actual organism's phenotype, and the others along the body (towards the tail), which is supposed to be the organism's memetype. Such a distinction comes from the fact that the initial state (during the neonate development) of the phenotypic cell depends on the organism's gene, whereas the state of the memetype does not, being determined through direct parental inheritance. As a consequence, the coevolutionary process supported in Enact is in general both genotypic and memetic. The former refers to the evolution of a coordinated movement of the population, while the latter allows to explore the effects of the genotypic evolution on the state of the organism's memetype. From a computational point of view, Enact can be regarded as a programmable, virtual, parallel machine defined by the artificial-life processes it supports, and relying upon six categories of states which represent environment, the organism's terminals, genotype, phenotype, memetype, and an additional category to allow the organisms to move. Its overall qualitative dynamics depends primarily on the ageing rate of the organisms, its tuning being very straightforward so as to prevent extinction of the organisms or deadlocks due to over-population, and guaranteeing the existence of very long transients. Here we show the artificial-life processes and overall dynamics involved in Enact; discuss computational issues that come out of its definition; exemplify its use in the context of a particular world set-up; analyse a case of parasite-like behaviour involved in a pair of organisms evolved from this set-up; and discuss conceptual issues underlying its use and design, such as its links with the notions of ``enaction'' and ``exaptation''.

This paper is not available online