Most neuroimaging studies average across heterogeneous individuals while they perform a small number of tasks, leading to imprecise and incomplete maps of neural architecture. This talk describes an alternative strategy, which is to densely image individual participants while they complete a large battery of tasks, in order to obtain a more accurate and complete understanding of person-specific neural architecture. Topics discussed will include assumptions and problems with group averaging, techniques available to parcellate or cluster the brain into functionally similar regions, and methods of assessing within- and between-subject parcellation reliability.