The time series data were deconvolved with a reference vector that coded the hypothesized BOLD signal for the alternating task conditions across the time series of the task while covarying for linear trends and the degree of motion correction previously applied (58). The reference vector was convolved with a vector that modeled the typical hemodynamic response (59). All data were transformed into standardized space (60). The functional data were resampled into 3 mm cubic voxels, and a spatial smoothing Gaussian filter (FWHM = 5 mm) was applied. These steps resulted in a fit coefficient for each voxel, representing BOLD response to SWM relative to the vigilance baseline condition. A three-step process was used to identify relevant activations for analysis (61). First, a stereotaxic brain atlas (60) was used to define the a priori regions of interest (ROIs). Second, significant clusters of activation (α = .025; volume > 1,323 μL) were identified for each group using AFNI 3dttest within the ROIs. Third, the peak activation within each significant cluster was extracted for each participant, and screened for multivariate outliers and non-normal distribution. The final values represented each subject's maximal contrast between the SWM and baseline vigilance conditions.
