Perceptual fluency refers to a subjective feeling or experience of the individual's difficulty in processing external information at the level of perception, reflecting the low-level processing of the attributes related to stimulating external forms. Previous studies have found that reducing the perceptual fluency of learning materials appropriately, could increase the subjective difficulty perceived by learners, which would stimulate the learners' analytical thinking, improve their academic performance ultimately. This phenomenon is called “disfluency effect” in educational psychology. While some difficult learning conditions can improve learning, the findings regarding the contribution of disfluent materials have been inconsistent. We seek breakthroughs in terms of more rational theoretical frameworks and boundary conditions of disfluent effects. It is suggested that the qualitative match of the evoked encoding processes of the applied difficulty with respect to the required retrieval processes of the memory test and the time delayed of the test may be two key factors affecting the disfluent effect. We believe that learning is a dynamic and complex process, whether memory is improved and enhanced is a time-dependent dynamic problem. Therefore, immediate performance does not reflect the effect of disfluency on memory intervention. Disfluent effects will appear over time. This study explored the delayed disfluent effects of perception through three experiments, the results discussed when and under which circumstances disfluency can promote learning as a desirable difficulty. The study explored how does the perceptual blurry affect judgments of learning and memory (immediate/delayed) in the recall in Experiment 1 and in the recognition in Experiment 2, by manipulating word clarity and using the continuous identification (CID) task paradigm. CID task is more sensitive in measure perceptual fluency. It can focus on word recognition in a short period of time without too much processing of semantic information, and more importantly, it can minimize the perception of disfluent experimental manipulation. The Experiments used different memory tests to detect disfluent effects in the same experimental manipulation modes, in order to explore whether the effect depends on the similarity of encoding process and retrieval process. The Experiment 3 replaced the experimental material with a picture, changes the picture clarity to explore whether the delayed disfluent effect of perception still exists in non-semantic concepts such as images. The above studies mainly obtained the following results: (1) Perceptual disfluency as a cue affected our judgments of learning. Compared with clear and fluent materials, we will give blurry and disfluent materials to lower JOLs. It is harder to remember for items that experienced difficulty. (2) The perceptual blurry of the materials had no effect on the immediate recall or recognition performance. After a certain test time delayed, the blurry material which with lower fluency had better performance, and the naming fluency did not interfere with judgments of learning and memory performance in the Experiment. (3) The qualitative match of the evoked encoding processes of the applied difficulty with respect to the required retrieval processes of the memory test as well as when there was a certain time delayed between the experimental manipulation and memory test, the perceptual disfluency can promote learning as a desirable difficulty. (4) This perceptual disfluent delayed effect exists not only in semantic concepts such as words, but also in non-semantic concepts such as images, which have certain stability and universality. The results showed that the encoding process triggered by the experimental manipulation mode matches the retrieval process required by the memory test in nature, and when there was a certain time delay between the experimental manipulation and memory test, perceptual disfluency can be used as a reasonable difficulty to facilitate learning.