Also in the Article

PPI analyses
This protocol is extracted from research article:
The neural circuitry of affect-induced distortions of trust

Procedure

PPI analyses were conducted using the generalized form of context-dependent PPIs toolbox (gPPI toolbox (35), using the same statistical model as outlined above. All voxels that survived SV FWE correction for the interaction contrast in the left TPJ (−60, −54, 19; k = 95) were used as a seed region (shown in blue color in Fig. 2C). To obtain an estimate of neural activity within the seed region, the BOLD signal from the seed region was extracted, corrected by removing effects of noise covariates, and deconvolved. Psychological interaction regressors for each of the task type and stimulation intensity combinations [control decisions during (i) weak and (ii) strong stimulation and trust decisions during (iii) weak and (iv) strong stimulation] were created by multiplying the estimated neural activity during the relevant decisions with condition-specific onset and offset times convolved with the canonical HRF. A new GLM (general linear model mentioned above) was then estimated for each participant that consisted of the original design matrix with the addition of the four psychological interaction regressors and the time course from the seed region.

To investigate the impact of aversive affect on trust-specific functional connectivity of the left TPJ, we probed the functional connectivity data for an interaction between threat and game type. To investigate the interaction between threat and game type, we entered the contrast estimates obtained from first-level PPI models into a flexible factorial model with the factors game type (trust and NS control), threat (absent and present), and separate covariates reflecting mean transfer in each condition. A subject factor was also included in the model. Given that we were particularly interested in trust-specific changes in functional connectivity, we first contrasted the covariates reflecting mean transfers in the trust game and mean transfers in the NS control task in the absence of threat (Trustno threat > NS Controlno threat). This comparison identifies regions for which connectivity with the TPJ correlates more strongly with mean transfers in the trust game than with mean transfers in the NS control game. As the next step, we then examined how threat of shock changed the relationship between TPJ connectivity and mean transfers in the trust game by examining the interaction between game type and threat estimated over the covariates. We illustrated these results in Fig. 3 by regression plots generated with coefficients reflecting functional connectivity strength for each of the conditions and extracted from 6-mm spheres around the peak voxel of the interaction contrast. The displayed regression plots were generated by the following regression model implemented in R (using the Regression Modeling Strategies package, RMS)$yik=β0+β1Transferik+β2GameTypek+β3Threatk+β4(Transferik×GameTypek)+β5(Transferik×Threatk)+β6(GameTypek×Threatk)+β7(Transferik×GameTypek×Threatk)+εik$

The dependent variable yik is the functional connectivity strength between a given brain region and the TPJ for individual i in GameTypek. Transferik is the mean amount sent by individual i in GameTypek. GameType is a dummy variable encoding whether decisions were made in the trust or the control task (1 indicates trust, and 0 indicates NS control task). Threat is a dummy variable encoding whether decisions were made in the presence or absence of threat (1 indicates presence, and 0 indicates absence of threat). In this regression, the coefficient for Transferik1) measures the slope of the relationship between TPJ connectivity and mean transfers in the absence of threat in the control task (see blue lines in Fig. 3, A, B, and D), and the sum of the coefficients for Transferik1) and the interaction term between Transferik × GameTypek4) measures the trust-related slope increase in the relationship between TPJ connectivity and mean transfers in the absence of threat (see orange lines in Fig. 3, A, B, and D). Equivalent analyses were performed to probe for significant differences between the threat and the no-threat condition in the trust game in the relationship between functional TPJ connectivity and mean transfer levels. Here, the sum of the coefficients for Transferik1), the interaction term between Transferik × GameTypek4), the interaction term between Transferik × Threatk5), and the interaction term between Transferik × GameTypek × Threatk7) measures the slope of the relationship between TPJ connectivity and mean trust in the presence of threat (see red line in Fig. 3C).

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

Q&A