Design and Application of Biomimetic Memory Circuit Based on Hippocampus Mechanism

Abstract

<p>Hippocampus, a special neuroanatomical structures, has been a realistic research model for the storage and retrieval of short-term and long-term memory. This article proposes a mathematic hippocampus model and bionic memory circuit which not only emulates the memory generation but also realizes the transformation from low to high-level memory. Based on the interior connections of the hippocampus, the proposed circuit reconstructs an episodic memory processing model and achieves the functions of multilevel memory generation. Memristor plays a vital role in imitating the plasticity of synapses in hippocampus recurrence, and its characteristics of switching dynamics are applied for controlling multilevel memory generation. Leveraging the proposed circuit, we propose a multilevel memorial generation system which has the capacities of perception quantification, memorial generation, and comprised the following: 1) receiver module; 2) quantitative module; 3) three-layer hippocampus memory circuit; and 4) memory generation module. The simulation results in PSpice indicate that the application of the model can quantize the episodic memory, afterward processing it by a three-layer hippocampus memory circuit to generate the multilevel memory. Moreover, this work paves the way for the memorial architecture in robotics by emulating the hippocampus memory principle.<br></p>