Cognitive functions, including attention and working memory, have been recognized to be associated with tone processing. However, it remains unclear how these cognitive functions might modulate the process of tone merging. In Macau Cantonese, three pairs of tones are currently undergoing or have completed merging, exhibiting different rates and ranges of merging within the speech community, providing a valuable opportunity to investigate this issue. Through a perception task, a production task, and a series of cognitive function tasks, we examined differences in cognitive performances among individuals distinguishing different tone pairs and explored the correlations between various indices of tone processing and cognitive functions. The results revealed that the completed merging of the T2–T5 pair showed no significant correlation with cognitive functions in perception and production. The T3–T6 pair, with an intermediate merging speed, exhibited a significant positive correlation between cognitive functions and both their perception and production. In contrast, the T4–T6 pair, characterized by a slower merging speed, showed a correlation only in the perception modality. Our findings suggested that cognitive functions affect tones at different stages of the merging process, influencing their perception and production in distinct ways. The results provide a novel perspective on the origin of tonal variation and the relationship between perception and production.