Mimicking Classical Conditioning of Fear Using a Dynamic Synaptic Memristor

Abstract

The growing demand for energy-efficient computing has prompted investigations into the diverse functionalities of resistive switching memristors, which show promise for neuromorphic computing. These memristors can emulate artificial synapses, nociceptors, and computational capabilities like reservoir computing. However, the integration of emotions, a critical aspect of brain function, remains unexplored in memristors. This study explores the emulation of fear, a crucial emotion that enables individuals to avoid potential danger through learned behavior, using a two-terminal Al/NbOx/Pt memristor structure. Leveraging the volatile behavior and non-filamentary switching mechanism of the memristor, synaptic functions and synaptic plasticity changes based on incoming spikes are mimicked. Furthermore, classical fear conditioning is employed to demonstrate fear simulation within the memristor, including the crucial aspects of extinction, generalization, and avoidance. The results showcase the potential of the Al/NbOx/Pt memristor for efficient synapse emulation and neuromorphic applications, as well as its ability to provide enhanced insights into brain function through emotion emulation, enabling versatile future applications of the memristive device. This study explores the emulation of fear conditioning using a two-terminal artificial Al/NbOx/Pt memristor, highlighting its potential in neuromorphic computing. Through the demonstration of fear conditioning using extinction, generalization, and avoidance properties of fear, the memristor advances the integration of emotional functionalities into memristors, paving the way for more sophisticated artificial intelligence systems. image