Implications of activity-dependent neurite outgrowth for developing neural networks
Van Ooyen, A., and Van Pelt, J. (1993). In: Gielen, S., and Kappen, B., eds. Artificial Neural Networks - ICANN 1993, 3rd International Conference on Artificial Neural Networks, Amsterdam, The Netherlands, pp. 177-182. [Full text: PDF]
The presence of only two basic neuronal properties—a firing threshold and activity-dependent outgrowth—is sufficient to cause a transient overproduction of connections in developing neural networks. This overproduction is enhanced by inhibition. Solely as the result of activity-dependent outgrowth and local cell interactions, the neuritic field of an inhibitory cell becomes smaller than that of an excitatory cell. Furthermore, a specific distribution of cell sizes is generated in the area surrounding an inhibitory cell.