Abstract
This paper presents how morphologically more realistic artificial neural networks have been obtained by using vectorial- stochastic grammars and used as subsidies for modeling biological neural systems and developing novel artificial neural structures. The paper includes the description of the vectorial-stochatic grammars, a review of the primate striate cortex, a mathematical analysis of the principles under- lying orientation encoding by centric domains, and the development and application of morphologically realistic neural centric models of orientation encoding.