| Abstract: | Cognitive neuroimaging studies of individual differences seek to reveal brain mechanisms of cognition by associating intersubject variability in brain activation with other variables of interest, such as sex, personality trait, or mood state. The choice of a priori regions of interest (ROIs) raises problems, because the selection criterion is typically consistent activation across prior studies, suggesting little intersubject variability. Here, we introduce a novel approach for selecting regions that are defined on the basis of their variance characteristics, rather than on the basis of their location or because of theoretical expectations. These regions of variance (ROVs) constitute the search space with which to assess how much of the observed variance can be ascribed to specific variables of interest. We compare the ROI and ROV approaches by applying each to the same data set and suggest that the conjunction of both methods may yield the greatest likelihood of capturing the rich relation between brain and behavior, while limiting the search space for statistical analyses and minimizing false positive errors. |
