Birdlike multisensory integrated oxide transistors for motion perception

Abstract

Multisensory inputs have been observed to augment behavioral and neural responses in avian species, underscoring the importance of converging perceptual enhancement, cognitive function, and motion perception into a unified system. To address the need for neuromorphic devices that replicate this process, we present a bio-inspired neuromorphic system that integrates cognitive-vestibular sensory motion perception. The system leverages a multi-input device fabricated from an ion-gel-gated In₂O₃ transistor, which exhibits adaptive plasticity, stable modulation, and spatiotemporal integration. The system embodies efficacious multisensory integration by processing dual-modal motion signals as spike trains, assigning salience weights via mean-firing rates and postsynaptic currents. Validation experiments demonstrate the system's ability to recognize human activities and classify aircraft flight motion modes, aligning with biologically plausible principles of multisensory integration. This neuromorphic system has far-reaching implications for developing advanced sensory artificial devices, smart electronics, and interactive intelligent electronics.