Self-oscillatory dynamics in recurrent excitatory networks

We present results concerning the capacity of recurrent excitatory neural networks to reach self-sustained oscillatory firing. We simulated networks made up of leaky integrate-and-fire based neurons, with excitatory connections. We studied the conditions of emergence of sustained self-oscillations, in which synchronic and asynchronic activities alternate, leading to a network bursting pattern. Finally, we analyzed the encoding by the network of external inputs depending on its firing activity. The results support the idea that the network encoding may change along time. (C) 2002 Elsevier Science B.V. All rights reserved.