Filtering Capability of Neural Networks from the Developing Mammalian Hippocampus

Synchronous population activity is present in neuronal networks in both normal conditions1–3 and pathological states such as epilepsy4–8. During early development of the hippocampus, similar bursts are also recorded2,3,9. These spontaneous bursts are generated by the synchronized action of interneurons acting as excitatory cells9. Nevertheless, the mechanism leading to synchronization still remains unclear. Here, we investigate the conditions in which synchronization arises in developing hippocampal networks. Using simultaneous recordings, we demonstrate that bursts result from a local cooperation of active cells within an integration period prior to their onset. During this interval, an increase in the number of excitatory postsynaptic potentials (EPSPs) takes place. By comparing EPSP frequency with burst ocutience we show that bursting takes place in a non-linear, allor-none fashion. This non-linear dependency with EPSP frequency is characterized by a threshold from which synchronized bursting arises (17 Hz) which can be reproduced by extracellular stimulation. We propose that this threshold is a property of the network, and determines the critical frequency at which cellular populations become phase locked. This frequency-threshold mechanism endows hippocampal networks with high-pass filtering behavior. We discuss its functional implication in neuronal computation10, stimulus encoding11–13 and in pathological conditions such as epilepsy5,14.