Dynamics of pulse packet in coincidence detector networks

Propagation of pulse packets, called synfire chain, is one of the most attractive topics in recent neuroscience. In this study, we consider a network that consists of coincidence detectors. A coincidence detector fires if and only if it receives more spikes than its threshold. Neither noise or background activity are necessary. It means that the neuron model represents the temporal information more essentially than the I+F model. We derive the dynamics of the two macroscopic parameters of a pulse packet, as in the literature with the I+F model, when it propagates through a layered coincidence detector network. By considering the distribution of the pulses the neuron receives under the assumption that the threshold of the coincidence detectors is low enough, we obtain the statistical variance of the spiking time when the neuron fires. The results show the existence of stable propagation of pulse packets in the coincidence detector network without any noise or background activity. The theoretical results agree well with the experimental ones given by computer simulations