The Effect of Spike Redistribution in a Reciprocally Connected Pair of Neurons with Spike Timing-Dependent Plasticity

This work considers a system of two spike-response cells, mutually connected by excitatory synapses that modify according to a multiplicative form of spike timing-dependent plasticity (STDP). The units are driven by independent external inputs as well as one common input, representing a global cortical rhythm. Our simulations show that the synaptic weights converge to attractors that encode input and plasticity parameters and are independent of initial conditions. In certain parameter regimes, the temporal correlation induced by coupling and the shared cortical signal lead to redistribution of spikes to favor doublets. We show that when this occurs, the weight changes associated with these doublets dominate convergence of weights to their attractors. Thus, even without invoking molecular mechanisms, spike timing-dependent plasticity can allow subsets of spikes to play a predominant role in the evolution of synaptic weights