This series of two articles develops a hypothesis on the modular organization of somatosensory cortex. The hypothesis is built around two functional entities: the segregate, a discrete somatosensory cortical macrocolumn approximately 0.5 mm in diameter, and the minicolumn, a smaller column approximately 0.05 mm in diameter, 40–80 of which make up a segregate. The hypothesis proposes that during perinatal development, minicolumns, acting via their short-range inhibitory and longer-range excitatory lateral connections, play an important role in the selection of thalamic connections to neighboring minicolumns. More specifically, the thalamic connections to each minicolumn are shaped by the interaction of that minicolumn primarily with those neighbors that belong to the same segregate. The outcome of this within-segregate self-organizational process is that (1) the minicolumns in a segregate acquire a complex but orderly pattern of afferent connections; (2) this con-nectional pattern, along with lateral inhibition, gives the minicolumns diverse receptive fields, arranged in a shuffled but orderly manner; and, most importantly, (3) the minicolumns and the segregate as a whole acquire a variety of stimulus feature-extracting properties.
A computer-based model of a segregate is developed to show that under conditions found in the developing cerebral cortex, the thalamocortical connections within a segregate readily form complex patterns as proposed by the hypothesis. Furthermore, the connectional patterns developed by the model segregate, its receptive field organization, and its feature-extracting properties (the latter are described in the following article) reproduce many experimentally observed features of real cortical networks.