▎ 摘　　要

Recently, neuromorphic devices have attracted great attention in the field of artificial intelligence technology at the hardware level. In particular regarding the developments of flexible electronics, there is an increasing interest in brain-inspired multi-functional perception learning system. In the present work, flexible cognitive memristors based on poly(vinyl alcohol)-graphene oxide (PVA-GO) hybrid nanocomposite are proposed. These devices exhibit excellent electrical performance at ultralow voltage below 0.5 V with high stability against mechanical stress. The threshold voltage of resistance switching is as low as approximate to 0.2 V. Good endurance and stability are observed. Most importantly, the PVA-GO-hybrid-based memristor demonstrates an ultralow power consumption of approximate to 0.5 mu W during the SET process. Moreover, such cognitive memristor-based neuron circuits demonstrate Pavlovian-associative learning behaviors. These memristors have potential applications in ultralow-power multi-functional perception learning systems.