Rate chaos and memory lifetime in spiking neural networks

Rate chaos is a collective state of a neural network characterized by slow irregular fluctuations of firing rates of individual neurons. We study a sparsely connected network of spiking neurons which demonstrates three different scenarios for the emergence of rate chaos, based either on increasing the synaptic strength, increasing the synaptic integration time, or clustering of the excitatory synaptic connections. Although all the scenarios lead to collective dynamics with similar statistical features, it turns out that the implications for the computational capability of the network in performing a simple delay task are strongly dependent on the particular scenario. Namely, only the scenario involving slow dynamics of synapses results in an appreciable extension of the network's dynamic memory. In other cases, the dynamic memory remains short despite the emergence of long timescales in the neuronal spike trains.