Temporal Pattern Coding in Ionic Memristor-Based Spiking Neurons for Adaptive Tactile Perception

Biological neurons encode signals through firing voltage spike trains having unique temporal patterns, enabling efficient information representation and processing. Realization of these rich neuronal firing characteristics in a single electronic device, without circuitry and software assistance, promise compact and functional neuromorphic hardware for advanced artificial intelligence applications. Here, a Pt/Co3O4-x/ITO-based ionic memristor is reported that can faithfully produce voltage spike trains exhibiting diverse temporal patterns of biological neurons, under electric current stimulation. The spiking behaviors stem from the redistribution of ions in the device, governed by the current induced electric field and Joule heating effects. Tonic, phasic, burst, and adaptive firing patterns of neurons are demonstrated. Particularly, the adaptive firing characteristics allow the memristor to reduce the response to invariant current stimulation and to respond to current changes with enhanced sensitivity, implementing neuronal adaptive coding function. Integrating such memristors with pressure sensors yields an artificial tactile sensory system that can adaptively perceive small pressure variations in the presence of strong static pressure backgrounds, enabling accurate identification of touched objects in ever-changing environments. This work opens up an avenue toward advanced neuromorphic hardware for smart neural prosthetics and bionic robotics applications.