Games of Proto-Life in Masked Cellular Automata (MCA) Syed Mustafa All and R. M. Zimmer Brunel University Uxbridge UB8 3PH England (UK) Email: Syed .Must af a .Ali@brunel . ac . uk Abstract. We investigate a relativistic variant of Conway's Life rule in the context of two- dimensional cellular automata with neighbourhood templates defined by lattice geometry. The Proto-Life rule is based on an abstraction of crystal growth processes and we discuss how it provides a model for the emergence of a crystalline precursor to life from an initial random prebiotic soup. 1. Introduction Standard cellular automata (CA) models of naturalistic processes largely ignore the geometric relationships between the components in the physical system under study; this results in a 'coarse grain' model of a natural phenomenon. For certain classes of natural phenomena such as crystal growth processes, this granularity problem may be significant. In order to demonstrate the significance of geometrical considerations to the mod- elling and interpretation of natural phenomena with CA, a relativistic variant of the Game of Life [1] based on an abstraction of crystal growth processes — Proto-Life'— is described. A new CA model is proposed — Masked Cellular Automata (MCA) — in which geometrical properties of the CA lattice are taken into consideration in the definition of neighbourhood template. Three binary classification criteria — (i) homogeneity of state-transition rule, (ii) homogeneity of neighbourhood template, (iii) referentiality - are presented to distinguish between the two types of MCA used in a series of simulation experiments designed to determine whether the Proto-Life rule, as applied to MCA, can support universal computation. Universal computation provides a useful benchmark for studying potential "artificial life" and is the basis of the computational definition of life [2]. The results of experiments conducted under initial conditions of a randomised lattice of specified density are described, and the behaviours of the structures gener- ated are classified according to the Wolfram classification scheme. It is observed that `phase' transitions between behavioural classes are constrained by the granularity of the simulation substrate — the computer hardware on which the CA executes. Parallels between the Life rule, the Proto-Life rule, and crystal growth processes are presented with a view to determining the relevance of Proto-Life to CA studies of the origin and evolution of life via emergent self-organisation. The Game of Life is a metaphor for biological life; Proto-Life is a metaphor for the processes involved in the construetion of a crystalline precursor to life from an initially random prebiotic 'soup'. 77 Ali, S.M., Zimmer, R.M. (1994) Games of Proto-Life in Masked Cellular Automata (MCA). In Complex Systems: Mechanism of Adaptation. Edited by Stonier, R.J., Yu, X.H. Amsterdam, IOS Press, pp. 77-84.