Data analysis paradigms for metabolic‐flow data: Combining neural modeling and functional neuroimaging

Two primary data analysis techniques are used with functional neuroimaging data: the subtraction paradigm and the covariance paradigm. The first is based on the assumption that a brain region participating in a specific task should show altered neural activity (relative to a control task). The second, which is not mutually contradictory with the first, assumes that tasks are mediated by networks of interacting regions and that different tasks involve differences in the pattern of functional connections among brain regions. The neurobiological assumptions governing each paradigm and the kinds of questions each seeks to address are discussed. Illustrations of each technique using positron emission tomographic data of a blood flow study of visual processing are given. The use of neural modeling techniques for analyzing functional neuroimaging data is also discussed, with explicit reference to such systems‐level approaches as structural equation modeling.