Pairing frequency experiments in visual cortex reproduced in a neuromorphic STDP circuit

Abstract

Previous studies show that the conventional pair-based form of STDP (PSTDP), is not able to account for many biological experiments including frequency-dependent pairing experiments performed in the visual cortex region of the brain. However, new improved synaptic plasticity rules, such as Triplet-based Spike Timing Dependent Plasticity (TSTDP), are capable of replicating many biological experiments outcomes including the results of the experiments carried out in the visual cortex. This paper proposes a programmable analog neuromorphic circuit, which is capable of reproducing pairing frequency experiments in the visual cortex. The circuit utilizes transistors working in their subthreshold region of operation. In addition, it implements a minimal model TSTDP learning rule, which needs a low number of transistors compared to its PSTDP circuit counterparts. These features result in low-power compact circuits that are suitable for large-scale VLSI implementations of Spiking Neural Networks (SNNs) with improved synaptic plasticity and learning capabilities.