The emulation of neuromorphic computations like the human brain exhibits an intriguing potential in the development of next-generation computing architectures and artificial intelligence. In this work, a transparent Ta₂O_5–3x/Ta₂O_5−x homo-structured optoelectronic memristor was proposed and developed using the atomic layer deposition technique, capable of replicating the intricate processes of associative learning in the brain. Basic synaptic plasticity was successfully emulated in the presence of electric stimuli, and an artificial neural network was constructed in line with the weight-updating rules of long-term potentiation/depression properties, achieving a handwriting recognition accuracy of 84.61%. Leveraging the photoelectrical effect and carrier dynamics at the Ta₂O_5–3x/Ta₂O_5−x interface, light-induced synaptic plasticity and optoelectronic synergistic response were achieved. With light as the conditioned stimuli (CS) and electrical stimuli as the neutral unconditioned stimuli (US), the associative learning behavior in the brain was effectively mimicked. One aspect involved the validation of classical conditioning behavior through the simultaneous application of the US and CS to the device. Moreover, sophisticated operant conditioning was achieved by pre-applying the enhanced or extinct US to control subsequent CS-triggered behaviors. This work provides an optoelectronic operating platform for sophisticated neuromorphic computing, demonstrating significant prospects in terms of adaptive intelligent sensing systems and autonomous robotics.