This study presents a reconfigurable electrolyte-gated transistor (EGT) that can perform both physical reservoir and synaptic functions, mimicking the dynamic and static information processing capabilities of the biological nervous system. The device exhibits tunable time scales under optical and electrical stimuli, making it suitable for reservoir computing and multimodal pre-processing. The non-volatile and programmable properties of the device, achieved through ion insertion/extraction via electrolyte gating, are verified. The device's superior performance in simulating human perception of dynamic and static multisensory information is demonstrated, showing an accuracy exceeding 90% in recognizing the Fashion-MNIST dataset. The study also simulates human audio-visual integration, further validating the device's potential for mimicking biological multisensory fusion. The proposed design principle can be applied to a broad range of materials, including InGaZnO4 (IGZO), and offers a promising approach for advanced neuromorphic applications.This study presents a reconfigurable electrolyte-gated transistor (EGT) that can perform both physical reservoir and synaptic functions, mimicking the dynamic and static information processing capabilities of the biological nervous system. The device exhibits tunable time scales under optical and electrical stimuli, making it suitable for reservoir computing and multimodal pre-processing. The non-volatile and programmable properties of the device, achieved through ion insertion/extraction via electrolyte gating, are verified. The device's superior performance in simulating human perception of dynamic and static multisensory information is demonstrated, showing an accuracy exceeding 90% in recognizing the Fashion-MNIST dataset. The study also simulates human audio-visual integration, further validating the device's potential for mimicking biological multisensory fusion. The proposed design principle can be applied to a broad range of materials, including InGaZnO4 (IGZO), and offers a promising approach for advanced neuromorphic applications.