Habituation variables were then created separately for Ku-0059436 negative and neutral by subtracting the average β for the first half from the average β from the second half. Partial correlations with the relevant anxiety type (PSWQ or MASQ-AA) were computed, with the variance associated with the other anxiety type and MASQ-AD-LI partialled out. Psychophysiological interaction analyses PPI analyses were performed on the preprocessed functional time series for each participant using FILM and did not differentiate between the first and second half of the time series. A cluster in Broca’s area, identified during the main

analyses, was used Inhibitors,research,lifescience,medical as the seed cluster. For each participant, Inhibitors,research,lifescience,medical the cluster was warped from Montreal Neurological Institute (MNI) space to functional space, and the time series was extracted. Six predictors were entered in the within-participant analyses: (1) the time series of Broca’s area, (2) a predictor that modeled the difference between negative and neutral conditions (coded as 1 during the negative condition, Inhibitors,research,lifescience,medical −1 during the neutral condition, 0 at all other times), (3) the interaction of these two predictors, and

(4–6) three predictors of no interest that modeled the variance associated with the positive condition, the rest condition, and the sum of the negative and neutral conditions. The negative versus neutral predictor and the three predictors of no interest were convolved with a gamma Inhibitors,research,lifescience,medical function to better approximate the temporal course of the BOLD hemodynamic response function (this convolution

was performed on the negative vs. neutral predictor prior to creating the interaction term). Group inferential statistical analyses were carried out using FLAME. The β maps corresponding to the PPI interaction term were regressed on the psychopathology questionnaires. Thresholding and correction for multiple comparisons Inhibitors,research,lifescience,medical were conducted in the manner described above. One-tailed t-tests were used to test the PSWQ β map in the a priori regions of interest. Two-tailed t-tests were used in conjunction with a whole-brain gray-matter mask to examine the β maps for MASQ-AA and MASQ-AD-LI, because no a priori hypotheses about these predictors were made. To almost determine whether the observed effects held separately in the first and second halves of the task, time series data for each cluster were extracted from the negative and neutral blocks, and the Broca’s area time course X PSWQ X Emotion interaction was tested separately for each half of the task in a hierarchical linear model using the Mixed procedure in SPSS v19. Participant was the nesting variable, block and timepoint were repeated factors, and a lag 1 autoregressive function was used.