Memristive Circuit Design for Personalized Emotion Generation With Memory and Retrieval Functions

Abstract

The generation of human emotion is often influenced by memory and personality. However, few reported circuits of emotion generation have considered this view. This work proposes a memristive emotion generation circuit with the functions of emotional memory and retrieval, and personality traits deployment. It mimics the emotional circuit of the brain and consists of six modules, which are input module, sensory cortex module, amygdala module, orbitofrontal module, hippocampus module and personality module. When inputting multimodal signals, the proposed design can not only generate two-dimensional continuous emotions, but also perform emotional memory and retrieval. Meanwhile, it provides various personality traits to mimic individual differences amid emotion generation based on the Big Five model. All in all, the designed memristive circuit provides a brain-like hierarchical and parallel emotion-generating method. It offers advantages in lower power consumption, area, and in-memory processing. In addition, the non-ideality analysis shows the circuit’s robustness. This work could provide a reference for the implementation of human-like emotional robots.