This study aimed to investigate whether stimulus modality influenced the acquisition and use of the rule-based strategy and the similarity-based strategy in category learning and whether the use of the two strategies was supported by shared or separate neural substrates. To address these issues, we combined behavioral and fNIRS methods in a modified prototype distortion task in which each category member has one rule feature and ten similarity features, and each type of feature can be presented in either the visual modality or the auditory modality. The results in Experiment 1 revealed that the learning effect in the "auditory rule-visual similarity" condition was the highest among all four conditions; further analysis revealed that in the "auditory rule-visual similarity" condition, the number of participants who used the rule-based strategy was more than the number of participants who used the similarity-based strategy, and the learning effect was always much higher for the rule-based strategy than for the similarity-based strategy. The behavioral results in Experiment 2 replicated the main findings in Experiment 1, and the fNIRS results showed that the use of the visual rule-based strategy was mediated by the dorsolateral prefrontal cortex, whereas the use of the auditory similarity-based strategy mainly engaged in the superior temporal gyrus, and the use of the visual similarity-based strategy mainly engaged in the inferior temporal gyrus. The results in Experiment 3 revealed that when the stimuli had only one type of feature, the visual rule rather than the auditory rule was learned more easily. The results provide new evidence that the stimulus modality can influence the acquisition and use of the rule-based strategy and the similarity-based strategy in category learning and that the use of the two types of strategies is supported by separate neural substrates both in the auditory modality and the visual modality.