To calculate brain-behavior matchmaking, i utilized a powerful statistical approach based on the strong-correlation arsenal (Pernet ainsi que al
TNT univariate analyses.
The preprocessed time series in each voxel from the main TNT task was concatenated across sessions to facilitate subsequent DCM analyses. Regressors within a GLM for each voxel were created by convolving a boxcar function (modeled as a 3 s short epoch) at stimulus onset for each condition of interest (i.e., Think, Intrusion, and Non-Intrusion for both Negative and Neutral scenes) with a canonical HRF. In addition to the regressors of interest, further regressors of no interest were included, specifically the six realignment parameters, sines and cosines of up to three cycles per run to capture low-frequency drifts, and constant terms to remove the mean of each run. Filler items, along with the few items with no button press or not recalled during Think condition, were also entered into a single regressor of no interest. For ROI analyses (see ROI definition), individual parameter estimates were then extracted and averaged in each ROI. ROIs were analyzed using ANOVAs with Hemisphere, Region, Emotion, and Awareness (Intrusion vs Non-Intrusion) as within-subject factors for MTL, and Emotion and Awareness (Intrusion vs Non-Intrusion) as within-subject factors for the right MFG. Planned comparisons between experimental conditions of interest were performed using paired t test. Voxel-based analyses were also performed by entering first-level activation maps for each condition of interest into flexible analyses of variance (A (RRID:SCR_007037), which used pooled error and correction for nonsphericity to create t statistics. The SPMs were thresholded for voxels whose statistic exceeded a peak threshold corresponding to p < 0.05 family-wise error (FWE) correction across the whole brain or within the appropriate search volumes of interest using random field theory.
Brain-behavior correlation analyses.
, 2013). Basic, we declined brand new null hypothesis in line with the percentile bootstrap CI, a strategy smaller sensitive to heteroscedasticity of your own study compared to the conventional t shot. Second, we fixed those individuals bootstrapped CIs having multiple comparisons across all of our six MTL ROIs (remaining and proper parahippocampal cortex, hippocampus, and you will amygdala), yielding good 99.3% CI. Third, i made use of missed correlations (accounting getting bivariate outliers utilizing the S-estimator departure code) (Rousseeuw and you can Van Drissen, 1999), and that imagine the genuine connection having appropriate incorrect positive manage and you may in place of death of energy.
For every single Return on your investment, i determined the brand new Invasion as opposed to Low-Invasion compare and you may coordinated that it that have one another attack ratio (the number of Zero-Imagine products where an attack was reported split of the level of No-Believe examples around the instruction) and you can apply at suppression score (No-Consider ? Baseline modified rating).
Behavioral PLS relationship analyses.
ROI-behavior correlation analyses do not, however, allow us to ascertain whether the exact same voxels contribute to both mnemonic and affective regulation. Indeed, the hypothesis of a shared regulation mechanism across domains proposes that a shared set of voxels across the control network produces relationships for affect and memory suppression. However, ROI-behavior correlations could, in principle, be supported by different sets of voxels. Behavioral PLS correlation (McIntosh and Lob) is ideally suited to disentangle this issue and examine more closely the relationship between neural markers of inhibition and behavioral scores across the putative memory control network (see below) and also MTL voxels. Behavioral PLS is a multivariate technique that reduces a set of voxels (i.e., variables) into a ranked series of independent latent variables (LVs) that express the largest possible covariance (or correlation) with behavioral scores. Put simply, this technique tries to identify separate sets of voxels that express quantitatively different relationships with behavioral measures. Voxel activity first has to be aligned and stacked across participants into a brain activation matrix X of 22 rows (i.e., participants) and N voxels. Normalized brain images are therefore used for that purpose. In a first series of PLS analyses, the mask used to create brain activity matrix X was extracted from the univariate analysis of No-Think > Think contrast (p < 0.001, uncorrected) and reflected memory control activity (i.e., frontoparietal control network). Within this mask, the Intrusion > Non-Intrusion contrast (for each type of emotional material) was computed for each voxel and the resulting vector of voxels was then stacked across participants. The same procedure was applied to the neural marker of downregulation in MTL (i.e., Non-Intrusion > Intrusion contrast). Given that this procedure requires the voxels to be aligned across participants, the MTL mask was derived by combining left and right parahippocampal cortex, hippocampus, and amygdala, defined anatomically using the AAL atlas (Tzourio-Mazoyer et al., 2002) (RRID:SCR_003550). The AAL atlas only includes a depal gyrus, which was then divided by splitting the mask into anterior (corresponding to perirhinal cortex) and posterior portions to isolate parahippocampal cortex.