Frontal-Brainstem Pathways Mediating Placebo Effects on Social Rejection

Abstract

Placebo treatments can strongly affect clinical outcomes, but research on how they shape other life experiences and emotional well-being is in its infancy. We used fMRI in humans to examine placebo effects on a particularly impactful life experience, social pain elicited by a recent romantic rejection. We compared these effects with placebo effects on physical (heat) pain, which are thought to depend on pathways connecting prefrontal cortex and periaqueductal gray (PAG). Placebo treatment, compared with control, reduced both social and physical pain, and increased activity in the dorsolateral prefrontal cortex (dlPFC) in both modalities. Placebo further altered the relationship between affect and both dlPFC and PAG activity during social pain, and effects on behavior were mediated by a pathway connecting dlPFC to the PAG, building on recent work implicating opioidergic PAG activity in the regulation of social pain. These findings suggest that placebo treatments reduce emotional distress by altering affective representations in frontal-brainstem systems.SIGNIFICANCE STATEMENT Placebo effects are improvements due to expectations and the socio-medical context in which treatment takes place. Whereas they have been extensively studied in the context of somatic conditions such as pain, much less is known of how treatment expectations shape the emotional experience of other important stressors and life events. Here, we use brain imaging to show that placebo treatment reduces the painful feelings associated with a recent romantic rejection by recruiting a prefrontal-brainstem network and by shifting the relationship between brain activity and affect. Our findings suggest that this brain network may be important for nonspecific treatment effects across a wide range of therapeutic approaches and mental health conditions.

Keywords: emotion regulation; nociception; opioid; placebo; resilience; social.

Figure 1.

Study overview. A, Hypothetical and simplified brain network mediating placebo effects on affective states, adapted from current models of placebo effects on somatic pain. Placebo analgesia in somatic pain is mediated at least partially by the opioidergic descending pain modulatory system. dlPFC may represent treatment context and expectations, thereby modulating value and appraisal processes in vmPFC, which in turn recruits brainstem regions, such as the PAG, to regulate bodily and brain responses to emotional events. Bottom row represents the extent of corresponding ROIs used for the main fMRI analysis. B, Experimental task design. Participants performed the social rejection and heat pain tasks in separate runs (counterbalanced order) before (Pre) and after (Post) a Placebo or Control intervention. Each trial started with the presentation of a fixation cross, followed by 15 s presentation of a picture of the Ex-Partner or a Friend, together with a short prompt that reminded participants to relive the emotions associated with the respective event. They then had to rate how they felt on a 1–5 Likert scale (1 = very bad; 5 = very good). The visuospatial control task between trials served to reduce washover effects to the next trial. The heat pain task had a parallel trial structure, but instead of the pictures, included a 15 s painful (Heat) or nonpainful (Warm) thermal stimulation. C, Behavioral results demonstrate a strong placebo effect on “social pain” and a moderate placebo effect on heat pain. Participants in the Placebo group experienced a greater improvement in affect than participants in the Control group. D, To investigate the brain mechanisms underlying these behavioral placebo effects, we used a multilevel mediation approach, with Postintervention versus Preintervention as a predictor, single-trial brain activity as mediator, and affect ratings as an outcome. Group (Placebo vs Control) was included as a second-level moderator, to investigate placebo-induced changes specifically.

Figure 2.

Mediation of placebo treatment effects on social rejection induced negative affect in key regions. A, Right dlPFC. B, vmPFC/OFC/rACC. C, PAG and brainstem. Gray outlines in brain plots indicate ROIs (see also Fig. 1A). Red represents group differences in changes in brain activity to Ex-Partner Post versus Preplacebo treatment (placebo effect on Path a). Boxes with red borders represent the direction of the intervention effects. Error bars indicate SEM. Plots with green borders represent group differences in brain activity predictive for affect (green clusters, placebo effect on Path b). Individual regression lines indicate the relationship between single-trial activity and affect ratings in the Placebo (left) and the Control (right) group. Yellow represents brain activity formally mediating the relationship between placebo treatment and changes in affect (placebo effect on Path ab) (right dlPFC).

Figure 3.

Exploratory whole-brain analysis of placebo treatment effects on social pain. A, Moderated Path a effect, illustrating activity increases Post-treatment > Pretreatment in the Placebo > Control group. B, Moderated Path b effect, indicating changes in the relationship between activity and affect in the Placebo compared with the Control group. C, Moderated Path ab effect, showing formal mediation effects in the Placebo > Control group. All results thresholded at p < 0.05, FDR corrected.

Figure 4.

Three-path mediation model of placebo effects on social rejection-related pain. A, Based on current models of placebo effects, we searched the whole brain for regions that would mediate the effect of placebo treatment on PAG, which in turn would directly influence social rejection-related pain (affect ratings). B, The results of this analysis revealed clusters in the right dlPFC, alongside with left vlPFC, and right postcentral sulcus, which showed significant three-path mediation effects (thresholded for display at p < 0.01, peak voxels of each cluster were significant at p < 0.05, FDR corrected).

Figure 5.

Mediation of placebo treatment effects on heat pain in key regions. A, Right dlPFC. B, vmPFC/OFC/rACC. C, PAG and brainstem. Gray outlines in brain plots indicate ROIs (see also Fig. 1A). Red represents group differences in changes in brain activity to Heat pain Post versus Pre placebo treatment (placebo effect on Path a). Box with red borders represents the direction of these intervention effects. Error bars indicate SEM. Plots with green borders represent placebo effects on Path b effects (green), depicting individual regression lines for the relationship between single-trial brain activity and affect ratings in the Placebo (left) and the Control (right) group. Whereas activity in the right dlPFC was predictive of more positive affect in the Placebo group, it correlated negatively with affect ratings in the Control group, similar to the findings regarding social rejection. Yellow represents brain activity formally mediating the relationship between placebo treatment and changes in affect (placebo effect on Path ab) (right dlPFC and vmPFC/OFC/rACC).

Figure 6.

Exploratory whole-brain analysis of placebo treatment effects on physical heat pain. A, Moderated Path a effect, illustrating activity increases Post-treatment > Pretreatment in the Placebo > Control group. B, Moderated Path b effect, indicating changes in the relationship between activity and affect in the Placebo compared with the Control group. C, Moderated Path ab effect, showing formal mediation effects in the Placebo > Control group. All results thresholded at p < 0.05, FDR corrected.

Figure 7.

Separable mediation of placebo effects on physical and social pain in the dlPFC. A, The multivariate classifier of dlPFC activity was able to accurately classify mediation effects of social versus physical pain placebo effects in the Placebo (80% cross-validated accuracy), but not in the Control group (40%). Error bars indicate SEM. B, Voxel weights of the classifier in the Placebo group. Pale green represents weights toward social pain placebo mediation. Pink represets weights toward physical pain placebo effects.