An Equivalent Single Spiking Neuron Model of the Working Memory

In this paper, we propose a biologically plausible computational working memory (WM) model implemented using a spiking neuron model representing a predictable WM mechanism in a single neuron. Empirical evidence from single neuron animal brain recordings has shown that WM is processed in a neuron model by encoding associations and exhibiting persistent activity. The model implemented using an adaptive exponential integrate and fire neuron model, was able to replicate the dynamics observed in WM tasks, such as the Delayed Match to Sample (DMS) paradigm. The input patterns were encoded as numbers, representing the spike train patterns in the neurons, and the frequencies of transient discharges of corresponding neurons were the outputs. By simulating this task, the model demonstrated how cognitive processes such as encoding, maintaining, and retrieving information during the delay period could be performed by single neurons. The model was examined by modifying parameters including the duration of delay, number of inputs, and retrieval probe count attributed to cognitive load. Through this soft computing-based approach, our simulations allow us to elaborate equivalents in emergent dynamics, including persistent neuronal activity during the delay period.