Figure 2.

Population activity remains asynchronous in the absence of local excitation. (A) Global network activity, as measured by the average number of neurons firing within a time bin of 1 ms, increases over time due to ongoing synaptic plasticity. However, the network activity remains relatively low and noisy, which indicates that the neurons in the network are firing asynchronously. While there is a tendency for the peak of the population activity to increase slightly, reflecting more synchronous activity, there is no evidence for global synchrony in the network. (B) Time course of 500 randomly chosen synaptic weights. Most of the synaptic weights fluctuate between the bounding values, which are g = 0 mV and g = 0.04 mV, respectively. Since STDP eliminates synaptic loops in neural networks, after a sufficiently long simulation time, about 50% of the synaptic weights reach the threshold values, but there is no specific direction for the elimination of synaptic connections. The reason is that the heterogeneity between the firing rates are large enough and the initial synaptic weights are not strong enough to overcome the inhomogeneity (disorder) in the network and synaptic connections are eliminated in a random directions in the network.