One likely mechanism in learning new skills is change in synchronous connections between distributed neural networks, which can be measured by coherence analysis of electroencephalographic patterns. This study examined coherence changes during the learning of two tasks, a word association task and a figure association task. Although learning curves were similar for both tasks, distinct patterns of coherence change were observed. Coherence tended to increase as learning progressed in the figure association task. In contrast, coherence tended to decrease in the word association task, especially within hemisphere. Word learning was coupled with negative intrahemispheric and positive interhemispheric performance-coherence relations in the gamma frequency. Unique to the figure learning task was an increase in the number of positive coherence-performance relations in both delta and theta frequencies across blocks. Results are discussed in light of ongoing efforts to identify the mechanisms that coordinate distributed brain activities during the process of learning. Further research is needed to define patterns of coherence change for different tasks, goals, and brain regions.