During mammalian neural development, mitotic cells sense local cues and morphogenic gradients to form complex and coherent processing systems, such as the cerebral cortex, that contains billions of neurons classified into different cell types. We use agent-based simulations of cellular growth and differentiation to explore and understand the regulatory principles governing cortical development in mouse and macaque. We use the java platform Cx3D, which provides methods for detailed simulations of genetic networks and also physical cellular processes in a 3-D space. First, we derive an artificial gene regulatory network (GRN) consisting of only 36 genes from experimental lineage and phenotypic data of mouse corticogenesis. Then, in Cx3D, a small group of precursor cells loaded with this GRN develop through mitosis, into a portion of the six-layered mouse neocortex with the correct quantitative and dynamical measurements observed experimentally for the primary somatosensory area (S1). One line of investigation explores the control of cortical lamination. In this regard, we find that modulation of one gene in the model GRN distinguishes the specification of the primary motor area (M1), suggesting that key genetic control points are able to shift developmental programs. This observation raises the question whether small changes in the GRN could also account for species specific differences of primate neocortex. We find that adding a state node to the mouse GRN resulted in a new precursor pool that following differentiation resembled the stereotypical macaque primary visual cortex (V1). Our models show how autonomous mitotic cells, which have a restricted repertoire of only local and simple actions, develop into the six-layered murine neocortex. Simple changes to the murine GRN give rise to a very different macaque neocortex, which follows nearly identical self-assembling principles.