Category learning is an important cognitive ability of human beings. Previous studies inferred that memory representation precision affects information-integration category learning performance. Whether the precision of external visual representations of learning materials improve the performance of information-integration category learning? The study aimed to investigate the role of visual item representation precision on information-integration category learning under the influence of feedback delay. For the purpose, visual item representation precision (precise vs. imprecise) and feedback delay (immediate feedback vs. delayed feedback) are two independent variables in this experiment with between-subject design. Each stimulus is a Gabor patch which has specific frequency and orientation, as shown in Figure 1. In this way, experiment manipulated visual item representation precision from more precise frequency and orientation value in every category. The experiment adopted the 500 ms or 5000 ms feedback delay paradigm. 122 college students participated in this experiment. Each volunteer learned to classify a Gabor patch in one from four categories in the whole experiment. The whole category learning process takes about an hour including four 80-trial blocks. In each trial, the stimulus came first, then volunteers press the keyboard to select the category. After the selection, the mask was presented for either 500ms or 5000ms, and then the visual feedback (right or wrong) was presented. The mean accuracy is the dependent variable. There are 2 (visual item representation precision) × 2 (feedback delay) × 4 (block) mixed design ANOVA with significance level of 0.05. Main effect of the visual item representation precision is significant, indicating precise visual item representation improves the performance of the information-integration category learning. Interaction between visual item representation precision and feedback delay is significant. The simple effect analysis suggests the mean accuracy of category learning under the precise visual item representation are higher than that in imprecise condition no matter how long the delay of feedback is. Although the difference in dependent variable data is significant, effect size is not large. Whether the small effect size is caused by a potential variable that is not controlled by the experiment. The experiment needs to compensate for the small effect size in significant difference and to prove the stability of the impact of visual representation precision. For these reasons, state-trace analysis take place in this experiment which reveals that visual item representation precision has a credibly impact on information-integration category learning. The state-trace plot of data is shown in Figure 4 in the main text part. Based on the results, we were able to reject the one-dimensional model，fit=8.21，p=.03, which eliminates an irrelevant latent variable. The results indicate that effects of visual item representation precision are stable. In conclusion, first, visual item representation precision adjusts the weakening effect of delayed feedback on information-integration category learning, indicating that precise visual representation is one of the factors to improve the performance of information-integration category learning. Second, state-trace analysis provides a new idea of statistics to eliminate the interference of a single potential variable for functional separation. State-trace analysis enhances the persuasion of the results.