A memristor-based associative memory neural network circuit with emotion effect.

Generally, people tend to learn or recall pleasant experiences during positive feelings. Similarly, people tend to learn or recall unpleasant things during negative feelings. The research in psychological field has demonstrated that human memory is closely related to emotion. On one hand, emotion helps store the memory that possesses the same emotion valence, which is known as the mood congruency memory (MCM). On the other hand, memory stored in a certain emotional state will be associated easily when the same emotion occurs, which is called mood-dependent memory (MDM). Inspired by the mechanisms of MCM and MDM, a memristor-based circuit of emotion-affected associative memory neural network is proposed in this work. The designed circuit mainly contains MCM module and MDM module. The functions, such as learning, forgetting, variable learning rate, MCM effect, MDM effect, and time interval, are implemented by the circuit. The simulation results in PSPICE show that the proposed memristive circuit can learn and associate the memory based on emotional effects like humans. [ABSTRACT FROM AUTHOR]