Abstract:Most memristor-based associative memory neural networks are focused on classical conditioning and ignored operant conditioning. In this paper, a memristor-based neural network of operant conditioning accorded with biological feature is designed. The designed circuit includes a voltage control module, an operant module and synapse modules. It realizes learning, forgetting, long-term memory, reinforcement and punishment functions based on variable synapse structure and double self-protection measure. Meanwhile, the four factors that affect operant conditioning such as contingency, immediacy, magnitude and deprivation are discussed and implemented. The simulation results in PSPICE show that the circuit can be used to simulate actual conditioned reflex and complicated applications. The memristor-based neural network circuit of operant conditioning provides more references for further development of neural networks.

Memristor-Based Neural Network Circuit of Operant Conditioning Accorded With Biological Feature