Recently, transistor-type optoelectronic synaptic device has garnered widespread attention due to its high potential for realizing artificial visual systems and neuromorphic computing. However, some basic mechanisms, such as the contribution of shallow and deep traps on the synaptic behavior, are still not fully understood. In this work, a channel-only transistor-type optoelectronic synaptic device has been employed as platform to clarify the contribution of shallow and deep traps to the synaptic response. We firstly demonstrated that the channel-only transistor-type synaptic device can successfully mimic almost all synaptic behavior, such as excitatory postsynaptic current spike (ΔEPSC), paired-pulse facilitation (PPF), etc. And then, two individual double-exponential models have been employed to fit the generation and decay part of ΔEPSC response to distinguish the role of shallow and deep traps on the synaptic behavior. The results suggest that the shallow trap gives rise to the fast response in both generation and decay component, and the deep trap contributes to the slow component in both generation and decay part. In addition, the number of deep traps is critical to determine the metastable current in the decay part because of the photogating effect. This explanation has been further confirmed by increasing the number of electrons that can be captured by increasing the light pulse intensity, and tuning the number of trap sites by storing the synaptic device in ambient environment or functionalizing the SiO2 surface with SAM agent containing strong electron withdrawing end group. Thus, this work not only clarify the contribution of shallow and deep traps, but also provide several strategies to tune the synaptic behavior.