Perceptual category learning is a fundamental cognitive mechanism which human beings classify perceptual stimulus and obtain category knowledge. In category learning, feedback is an indispensable part. There are some controversies in whether category learning is a single learning system or multiple systems and each one has its different views. According to COVIS model, which is an assumption of the multiple systems in category learning, feedback that delays a few seconds will affect the implicit system of category learning but not the explicit system of category learning. However, it is not clear that how the mask type influent on the perceptual category learning. Two experiments are designed for exploring the effect of mask type on perceptual category learning. The experiment 1, which is under the condition of the Gabor mask, is a 2(category structures: rule-based vs. information-integration)×2(delay feedback: 500ms vs.3000ms) between-subject design. The results showed that the performance of information-integration category structure decline due to the delay feedback, and rule-based category structure is not affected. The separation effect arises between implicit category structure and explicit category structure. The experiment 2, which is the same design as Exp 1 but replaces the Gabor mask with black mask, showing that the separation effect is disappeared according to the results. That is, delayed feedback does not affect the information-integration category structure. Our experiment shows that the COVIS model has limitations. Maybe the impairment of II learning is caused by the levels of perceptual and criterial noise. Both perceptual and criterial noise were impaired category learning by increasing uncertainty about the location of the stimulus in relation to the regions of perceptual space. The role of the mask in category learning is discussed in this paper. Furthermore, we can also explore the effect of physical properties of masking stimulus itself (such as the contrast, line) on category learning. In addition, because little known about the location of the mask in category learning, it is worth doing the research about whether the mask impact on the delayed feedback between the response and feedback or between the stimulus and the response. Two experiments are designed for exploring the effect of mask type on perceptual category learning. The experiment 1, which is under the condition of the Gabor mask, is a 2(category structures: rule-based vs. information-integration)×2(delay feedback: 500ms vs.3000ms) between-subject design. The results showed that the performance of information-integration category structure decline due to the delay feedback, and rule-based category structure is not affected. The separation effect arises between implicit category structure and explicit category structure. The experiment 2, which is the same design as Exp 1 but replaces the Gabor mask with black mask, showing that the separation effect is disappeared according to the results. That is, delayed feedback does not affect the information-integration category structure. Our experiment shows that the COVIS model has limitations. Maybe the impairment of II learning is caused by the levels of perceptual and criterial noise. Both perceptual and criterial noise were impaired category learning by increasing uncertainty about the location of the stimulus in relation to the regions of perceptual space. The role of the mask in category learning is discussed in this paper. Furthermore, we can also explore the effect of physical properties of masking stimulus itself (such as the contrast, line) on category learning. In addition, because little known about the location of the mask in category learning, it is worth doing the research about whether the mask impact on the delayed feedback between the response and feedback or between the stimulus and the response.