Preprocessing and statistical analyses were performed using SPM8 (Wellcome Department of Imaging Neuroscience, London, UK). To correct for head motion, all functional volumes were realigned to the first volume using septic b-spline interpolation and subsequently unwarped to remove residual movement-related variance due to susceptibility-by-movement interactions. Slice timing correction was performed after realignment/unwarping. To improve coregistration, bias-corrected anatomical and mean EPI images were created and subsequently coregistered using the new segment toolbox in SPM. Images were normalized to the Montreal Neurological Institute T1 template using the parameters (forward deformation fields) derived from the nonlinear normalization of individual gray matter tissue probability maps. Last, functional data underwent spatial smoothing using an isotropic 6-mm full-width at half maximum Gaussian kernel.

Statistical analyses were carried out using the general linear model (GLM). Regressors of interest were modeled using a canonical hemodynamic response function (HRF) with time and dispersion derivatives to account for subject-to-subject and voxel-to-voxel variation in response peak and dispersion. Since our main interest was the impact of aversive affect on trust-taking, we modeled the decision period for the full RT on each trial, that is, from the onset of the decision screen until participants pressed the confirm button. This approach implicitly controls for differences in decision latencies (59). This was performed in the following four conditions: (i) trust game during relatively high-intensity stimulation expectancy (threat condition), (ii) trust game during relatively low-intensity stimulation expectancy (no-threat condition), (iii) control game during relatively high-intensity stimulation expectancy (threat condition), and (iv) control game during relatively low-intensity stimulation expectancy (no-threat condition). Last, the following regressors of no interest were included in our model: the actually realized weak and strong tactile stimulations during a block (modeled as two separate regressors, one including all the predictable and unpredictable weak and one including all the predictable and unpredictable strong shocks), block cues indicating game type (trust, control), and stimulation intensity of the reminder shock (weak, strong) at the beginning of each block, as well as omissions of behavioral responses during a trial.

The main goal of the current investigation was to identify the impact of aversive affect on the neural correlates of trust decisions. Trust-specific neural effects of aversive affect can be identified via an interaction between threat and game type in which threat significantly alters the neural correlates of decision-making in trust relative to NS control trials. To investigate the interaction between threat and game type, an ANOVA was computed by entering contrast estimates obtained from first-level models into a flexible factorial model with the factors game type (trust and control), threat (absent and present), and participant. We were particularly interested in trust-specific affect-induced suppression of activity and connectivity, which we tested via the interaction contrast (trustno threat > trustthreat) > (NS controlno threat > NS controlthreat) in the context of the flexible factorial design. A covariate reflective of each participant’s mean transfer in each condition was also included to probe for brain-behavior correlations. All analyses were also conducted without the behavioral covariate, and results did not change.

Our main analyses rely on a priori ROIs as we expected regions commonly implicated in the major cognitive and affective component processes of trust to be affected by aversive affect. Specifically, we hypothesized that participants needed to assess the trustworthiness of the trustee to make predictions about payout probability in the trust game, which involves regions commonly implicated in theory of mind and social cognition (20). To identify regions implicated in theory of mind, we consulted neurosynth.org (31), which offers a means to obtain automated meta-analyses over a large number of previous fMRI investigations and thereby provides an independent method to obtain masks for ROI analyses. To guide and constrain our ROI selection, we computed the conjunction of the neurosynth meta-analyses for the terms emotion (forward inference to identify regions that consistently show modified activity) and theory of mind (reverse inference to identify regions that are specifically involved in theory of mind). This approach identified overlap between these networks in the left TPJ and dmPFC, which agrees particularly well with results from recent a meta-analysis identifying the TPJ and dmPFC as core social cognition regions (19). Furthermore, because of its prominent role in signaling emotional salience (11), we included the amygdala as an additional ROI (see also neurosynth search: emotion).

ROI analyses in relevant cortical regions were conducted using small volume correction with masks created via relevant search terms on neurosynth.org, while anatomically well-defined subcortical ROI masks were created using the AAL (automated anatomical labeling) atlas implemented in WFU Pickatlas. The following independent ROI masks were created via automated meta-analyses from neurosynth.org: (i) bilateral TPJ (neurosynth term: theory of mind), with peak voxels in the left (−60,−56,14) and right TPJ (56,−58,20) and sizes of 1031 and 1416 voxels, respectively. Note that the ventral part of this mask also contains voxels from pSTS. For simplicity, we refered to this mask nevertheless as “TPJ”. (ii) dmPFC (neurosynth term: theory of mind), with a peak voxel in the medial dmPFC (−2, 28, 62) and a size of 3175 voxels. The ROI mask for the amygdala, which is an anatomically well-defined region, was created via the AAL atlas using an anatomical mask for bilateral amygdala with sizes of 439 (left) and 492 (right) voxels. Additional exploratory analyses were conducted in regions involved in evaluating the anticipated outcomes of choice options, such as the ventral striatum and vmPFC (60) (neurosynth term: reward), as well as a region implicated recently in one-shot trust decisions by a recent meta-analysis (28) using the following masks: bilateral ventral striatum (combined mask of AAL putamen and caudate up to z = 8), with sizes of 3239 (left) and 3429 (right) voxels, respectively; vmPFC (neurosynth search term: ventromedial) with a peak voxel in medial vmPFC (8, 24, −12), with a size of 1327 voxels; left anterior insula with peak voxels at −42, 18, 2 and a size of 13,844 voxels; and right anterior insula with peak voxels at 42, 18, 2 and a size of 13,871 voxels.

Furthermore, to identify whether extended networks outside our ROIs show effects of interest, we conducted exploratory whole-brain analyses at an FWE-corrected extent threshold of P < 0.05 (k > 226, initial cluster-forming height threshold P < 0.001). Last, to characterize activation patterns of interest, such as time courses and activation differences due to aversive affect, regression coefficients (beta weights) for the canonical HRF regressors were extracted with rfxplot (61) from 6-mm spheres around individual participants’ peak voxel that showed significant effects of interest on BOLD (Blood-oxygen-level dependent) responses and functional connectivity. Follow-up tests that characterize the single components of significant interaction effects were conducted in neuroimaging space via tests of simple effects of interest.

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.