Parallel emulation of visual adaptation and memory towards biochemically mediated motion recognition in a real scene

Aqueous neuromorphic devices using essential biological mechanisms have recently appeared as promising candidates for high-level neurosynaptic emulation. Towards the important functionality of motion recognition, while conventional solid-state neuromorphic vision sensors have made significant progress, aqueous motion recognition based on biochemical transmission remains challenging. Taking inspiration from biology, here we report an organic photoelectrochemical transistor biosensor capable of parallel emulation of visual adaptation and memory towards biochemically mediated motion recognition in a real scene. Based on the rational design and implementation of photoelectrochemical events, two artificial Magno and Parvo pathways are emulated to produce visual adaptation and memory in aqueous conditions, respectively. Dynamic and static visual information could be correspondingly processed and applied for integrated image filtering in a real scene and recognition of moving objects by artificial neural networks.