We investigate a general and analytically tractable model for the activity-dependent formation of neuronal connectivity patterns. Previous models are contained as limiting cases. As an important example we analyze the formation of ocular dominance patterns in the visual cortex. A linear stability analysis reveals that the model undergoes a Turing-type instability as a function of interaction range and receptive field size. The phase transitions is of second order. After the linear instability the patterns may reorganize which we analyze in terms of a potential for the dynamics. Our analysis demonstrates that the experimentally observed dependency of ocular dominance patterns on interocular correlations of visual experience during development can emerge according to two generic scenarios: either the system is driven through the phase transition during development thereby selecting and stabilizing the first unstable mode or a primary pattern reorganizes towards larger wavelength according their lower energy. Experimentally observing the time course of ocular dominance pattern formation will decide which scenario is realized in the brain.

• Received 22 July 1998

DOI:https://doi.org/10.1103/PhysRevE.59.6977