Using different acquisition techniques then EPI and looking for deep brain responses requires novel methods to detect activation. The dynamic MR response function to activation not only depends on the brain structure and location, but also on the used MR method. In the past twenty years, task-based fMRI studies have primarily focused on signal amplitude changes, or on connectivity related to few selected nodes. We will follow an alternative view on fMRI data by shifting the focus away from signal amplitudes towards large-scale, task-induced synchronization networks. We developed a new data analysis algorithm called ``SyncMap'' that is designed to identify sets of brain locations that collectively synchronize in response to a task while being spatially connected in an underlying network. SyncMap does not require a presegmentation of the data for dimensionality reduction as it can handle large networks consisting of thousands of nodes. Because its conceptual basis is task-induced synchronization it does not depend on a hemodynamic response model. SyncMap identified several task-specific, large-scale patterns of synchronization. The strongest network hubs coincided with the sites of highest BOLD amplitude changes. However, additional hubs emerged where there were no significant amplitude changes, and several hubs even exhibited antagonistic behavior. This provides an entirely new window into the immense complexity of human brain function by taking large-scale synchronization patterns into account.