An artificial withdrawal reflex arc that can realize neuromorphic tactile perception, neural coding, information processing, and real-time responses was fabricated at the device level without dependence on algorithms. As an extended application, the artificial reflex arc was used to perform an object-lifting task based on tactile commands, and it can easily lift a 200-g weight. A fiber-exploiting electro-optical synaptic transistor (FEST) was fabricated to emulate synaptic plasticity modulated by electrical or optical spikes. Due to an ultrahigh spike duration-dependent plasticity index (~ 12,651%), the FEST was applied in electro-optical encrypted communication tasks and effectively increased signal recognition accuracy. In addition, the FEST has excellent bending resistance (bending radii = 0.6–1.4 cm, bending cycles > 2000) and stable illumination responses for a wide range of incident angles (0°–360°), demonstrating its potential applicability in wearable electronics. This work presents new design strategies for complete artificial reflex arcs and wearable neuromorphic devices, which may have applications in bioinspired artificial intelligence, human–machine interaction, and neuroprosthetics.