Networks of Polarized Evolutionary Picture Processors

We define a new bio-inspired computational model for deciding 2-dimensional languages similar to those presented in [2] and [3]. The novel factor being the communication protocol, which is based on the polarity associated with the symbols on the picture frame. This model can be viewed as a 2-dimensional extension of the one presented in [1]. We compare the class of languages accepted by these networks to other classes of 2-dimensional languages, namely the classes of Local Languages and Tiling Recognizable Languages [12]. We show that these networks can accept the complement of any local picture language and some picture languages that are not recognizable by tiling systems. We also show that these networks can recognize any input that contains a given sub-picture, provided that the sub-picture is made up of at most 3 rows or 3 columns.