Inspired by the multimodal information processing mechanisms of biological visual systems, the optoelectronic synaptic transistors integrate light sensing, memory, and computation functions, offering an innovative pathway for constructing neuromorphic systems with sensing-memory-computing integration. In this work, indium oxide (In2O3) thin-film transistors (TFTs) were fabricated via a low-temperature solution process. Leveraging the persistent photoconductivity of oxide semiconductors, a dual-wavelength optoelectronic synapse capable of both light perception and information retention was designed. Under optical stimulation with 405- and 532-nm light pulses, the device successfully emulated key synaptic plasticity behaviors, including excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), and spike number-dependent plasticity (SNDP). Furthermore, the device was utilized to simulate cognitive functions, such as learning-forgetting-relearning processes, Pavlovian classical conditioning, and a$3\times 3$array-based optical memory. Benefiting from the multilevel photoelectronic memory characteristics of the device, handwritten digit recognition based on the MNIST dataset was simulated, achieving a high recognition accuracy of 95%.