and

Form generation or morphogenesis has a major role in both artificial and natural development. This paper provides results from experiments in which a genetic algorithm (GA) was used to evolve cellular automata (CA) in order to produce predefined bidimensional shapes. The shapes considered were a square, a diamond, a triangle and a circle. CA were chosen as models for shape formation and they were implemented as NetLogo models. The CA's rule was defined as a lookup table where the input was determined by the interaction neighbourhood’s cell state values, and the output established whether or not a cell was to be reproduced at the empty objective cell. Four different interaction neighbourhoods were considered: von Neumann, Moore, 2-Radial, and Margolus. The GA worked by evolving the CA rule table's output bits and the number of iterations that the model was to run. After the final chromosomes were obtained for all shapes, the CA model was allowed to run starting with a single cell in the middle of the lattice until the allowed number of iterations was reached and a shape was formed. In some cases, mean fitness values of evolved chromosomes were 100%