Previous studies have demonstrated that there are separate neural mechanisms underlying semantic and response conflicts in the Stroop task. However, the practice effects of these conflicts need to be elucidated and the possible involvements of common neural mechanisms are yet to be established. We employed functional magnetic resonance imaging (fMRI) in a 4-2 mapping practice-related Stroop task to determine the neural substrates under these conflicts. Results showed that different patterns of brain activations are associated with practice in the attentional networks (e.g., dorsolateral prefrontal cortex (DLPFC), anterior cingulate cortex (ACC), and posterior parietal cortex (PPC)) for both conflicts, response control regions (e.g., inferior frontal junction (IFJ), inferior frontal gyrus (IFG)/insula, and pre-supplementary motor areas (pre-SMA)) for semantic conflict, and posterior cortex for response conflict. We also found areas of common activation in the left hemisphere within the attentional networks, for the early practice stage in semantic conflict and the late stage in "pure" response conflict using conjunction analysis. The different practice effects indicate that there are distinct mechanisms underlying these two conflict types: semantic conflict practice effects are attributable to the automation of stimulus processing, conflict and response control; response conflict practice effects are attributable to the proportional increase of conflict-related cognitive resources. In addition, the areas of common activation suggest that the semantic conflict effect may contain a partial response conflict effect, particularly at the beginning of the task. These findings indicate that there are two kinds of response conflicts contained in the key-pressing Stroop task: the vocal-level (mainly in the early stage) and key-pressing (mainly in the late stage) response conflicts; thus, the use of the subtraction method for the exploration of semantic and response conflicts may need to be further examined.