Atypical processing of social anticipation and feedback in borderline personality disorder

Abstract

Background-: Borderline personality disorder (BPD) is characterized by maladaptive social functioning, and widespread negativity biases. The neural underpinnings of these impairments remain elusive. We thus tested whether BPD patients show atypical neural activity when processing social (compared to non-social) anticipation, feedback, and particularly, how they relate to each other.

Methods-: We acquired functional MRI data from 21 BPD women and 24 matched healthy controls (HCs) while they performed a task in which cues and feedbacks were either social (neutral faces for cues; happy or angry faces for positive and negative feedbacks, respectively) or non-social (dollar sign; winning or losing money for positive and negative feedbacks, respectively). This task allowed for the analysis of social anticipatory cues, performance-based feedback, and their interaction.

Results-: Compared to HCs, BPD patients expressed increased activation in the superior temporal sulcus during the processing of social cues, consistent with elevated salience associated with an upcoming social event. BPD patients also showed reduced activation in the amygdala while processing evaluative social feedback. Importantly, perigenual anterior cingulate cortex (pgACC) activity during the presentation of the social cue correlated with reduced amygdala activity during the presentation of the negative social feedback in the BPD patients.

Conclusions-: These neuroimaging results clarify how BPD patients express altered responses to different types of social stimuli (i.e. social anticipatory cues and evaluative feedback) and uncover an atypical relationship between frontolimbic regions (pgACC-amygdala) over the time span of a social interaction. These findings may help to explain why BPD patients suffer from pervasive difficulties adapting their behavior in the context of interpersonal relationships and should be considered while designing better-targeted interventions.

Keywords: Borderline personality disorder; FMRI; Reward; Social cognition.

Fig.. 1

Schematic depiction of the social reward task (A) and example stimuli (B).

Fig.. 2

Model 1 whole brain fMRI results for the Social Condition by Group Cue Contrast. A) Whole brain analysis results from Model 1 showing the contrast social > non-social cues. The STS (yellow cluster) was shown to be significantly more activated by the BPD patients (95 consecutive voxels, peak [60 −37 −14]). Images are thresholded at p=.001, k>88, Monte Carlo corrected and overlaid on the averaged normalized T1-weighted anatomical images created from all participants (N = =45). B) Dot plots of the beta estimates extracted from an 8 mm sphere around the peak STS (from A). Average estimates of each category are illustrated by the yellow diamonds. a.u.=arbitrary units; BPD=borderline personality disorder patients; HC=healthy controls; STS=superior temporal sulcus.

Fig.. 3

Model 2 amygdala ROI activation in the interaction Social Condition by Group Feedback Contrast. A) Significant amygdala ROI activation is shown in yellow (right: 26 consecutive voxels, peak [21 −1 −14], t = =3.57, p_FWE =0.007; left: 1 voxel, peak [−18, −4, −14], t = =2.90, p_FWE =0.048) in the 2-way interaction (Social Condition x Group) overlaid on the ROI mask created using the WFU PickAtlas (red). ROI activations are shown with p=.05 FWE corrected, and overlaid on the averaged normalized T1-weighted anatomical images created from all participants. B) Dot plots underlaid by boxplots of the beta estimates extracted from the bilateral amygdala mask (i.e. red voxels in A). Average estimates of each category are illustrated by the yellow diamonds. a.u.=arbitrary units; BPD=borderline personality disorder patients; HC=healthy controls; ROI=region of interest.

Fig.. 4

Model 3 Whole-brain fMRI results from the independent samples t-test amygdala feedback regression analysis. A) The change in signal (i.e. beta estimates) from the bilateral amygdala ROI (Fig. 2) was extracted from the processing of the social loss>non-social loss feedback and entered as a covariate in an independent samples t-test for each group (i.e. HCs and BPDs) testing the difference between the processing of the social>non-social cues. The negative regression in the BPD group is shown in yellow and includes the pgACC, bilateral middle frontal gyrus, and left putamen. The main effect of group (i.e. HC>BPD) is shown in red (and the overlap is shown in orange), and includes solely the pgACC (93 consecutive voxels, peak [9 41 10], t = =4.97). B) Illustration of the above-mentioned modulation. The y-axis shows the extracted amygdala activity (which was entered as a covariate in the fMRI model), and the mean pgACC activity extracted from A (white circle) is shown on the x-axis. There was a significant correlation in the BPD group (rho= −0.74, p<.001) but not the HCs (rho=0.31, p=.146).*p<.05; a.u.=arbitrary units; BPD=borderline personality disorder patients; HC=healthy controls.