Amplitude and Spike Timing Dependent Plasticity

Many spike timing dependent plasticity (STDP) rules generate a bimodal distribution of synaptic weights because there is no stable equilibrium state. Our approach augments STDP with amplitude dependence providing negative feedback of synaptic weight to plasticity resulting in weights being driven toward stable values and unimodal distributions. The affects of input correlation on synaptic weight are shown using simulated cortical neurons. It was found that pre-and post-synaptic spike trains effect the mean, variance, and skew of the synaptic weight distributions using amplitude and spike-timing dependent plasticity. In addition, multiplicative synaptic modification noise was found to increase the variance of the weight distribution and induce positive skew.