Working memory capacity in spiking neural network with two types of plasticity

Working memory (WM) is a brain system for short-term storage and manipulation of information and plays an important role in complex cognitive tasks. Most WM models contain pre-formed content-specific structures of neurons for loading specific objects into memory, without considering the flexibility of WM. We develop on previous research and propose a flexible model of WM based on synaptic theory considering two types of plasticity: Tsodyks-Markram plasticity and spike-timing-dependent plasticity (STDP). The synaptic theory of WM hypothesizes that items are stored silently by temporally potentiated synapses in a distinct neuronal population. We show that external stimulation can form such populations due to the mechanisms of STDP. The capacity of the WM model with two types of plasticity was studied in dependence of synaptic time scales and learning rate. It is shown that estimated WM capacity of model with two types of plasticity depends on intrinsic neural and synaptic parameters and could be scaled by STDP parameters.