Activity-dependent Growth and Self-Organization in a Neural Network Model:

Abstract

The development of neural circuits is determined to a large extent by experience during critical periods of early postnatal life. Several interacting processes have been identified to be involved in the formation of new synaptic connections and the elimination of already existing connections: functional competition between inputs, neuronal activity, structural consolidation at the end of the critical period and regulation of synaptic connections by experience.

In this study we introduce a neuron model for activity-dependent growth and self-organization of a neural network based on a neurobiologically plausible mechanism of competition for neurotrophic growth factors among synaptic connections inside every neuron. We extend previous models by including a detailed topological model of neuronal outgrowth in combination with recurrent lateral excitation among neurons. In a series of simulation studies we investigate the influence of neurotrophic factors and firing rate dynamics on the connectivity of generated neural networks.

We find that providing a network of initially fully connected neurons with two different baseline input patterns leads to differentiation into two corresponding clusters. Furthermore, we show that the total amount of neurotrophic factors available for each neuron plays an important role in the development of the connectivity of the neural network and we hypothesize that this might be involved in the initialization and dynamics of critical periods.