Memory accuracy involves two major processes: pattern separation and pattern completion. Pattern separation refers to the ability to reduce overlap among similar inputs to avoid interference, and pattern completion refers to the ability to retrieve the whole information from partial or degraded cues. Impairments in pattern separation/pattern completion contribute to cognitive deficits in several diseases of the nervous system. Therefore, it is better to evaluate both pattern separation and pattern completion in one apparatus. However, few tools are available to assess pattern separation and pattern completion within the same apparatus for rodents. In this study, we designed a series of images with varying degrees of similarity to the correct image to evaluate pattern separation and pattern completion. First, mice were trained to discriminate between two totally different images, and once the correct percentage reached above 77% for two consecutive days, the images with different degrees of similarity were used to measure pattern separation and pattern completion. The results showed the mice performed progressively worse from S0 to S4 (increasing similarity) when discriminating similar images in pattern separation, and the mice performed progressively worse from C0 to C4 (decreasing cues information) when recalling the correct image according to partial cues in pattern completion, implying a good image similarity-dependent manner for memory accuracy evaluation. In sum, we designed a convenient, effective paradigm to evaluate pattern separation and pattern completion based on a touchscreen pairwise discrimination task, which may provide a new method for the studies of the effects and mechanisms of memory accuracy enhancing drugs.