Compassion training alters altruism and neural responses to suffering

Abstract

Compassion is a key motivator of altruistic behavior, but little is known about individuals' capacity to cultivate compassion through training. We examined whether compassion may be systematically trained by testing whether (a) short-term compassion training increases altruistic behavior and (b) individual differences in altruism are associated with training-induced changes in neural responses to suffering. In healthy adults, we found that compassion training increased altruistic redistribution of funds to a victim encountered outside of the training context. Furthermore, increased altruistic behavior after compassion training was associated with altered activation in brain regions implicated in social cognition and emotion regulation, including the inferior parietal cortex and dorsolateral prefrontal cortex (DLPFC), and in DLPFC connectivity with the nucleus accumbens. These results suggest that compassion can be cultivated with training and that greater altruistic behavior may emerge from increased engagement of neural systems implicated in understanding the suffering of other people, executive and emotional control, and reward processing.

Keywords: altruism; compassion; decision making; emotion regulation; emotional control; fMRI; individual differences; meditation; neuroimaging; social behavior.

Conflict of interest statement

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Figures

Fig 1

The Redistribution Game. (a) The dictator transfers an unfair amount of money ($1/$10) to the victim, while the participant observes. (b) The participant can choose to spend any X amount of $5 in order to compel redistribution of funds from the dictator to the victim. (c) After two weeks of training, COM trainees (n = 20) redistributed more funds than REP trainees (n = 21; * independent sample t(39) = 2.09, p < .05, d = 0.65; COM mean rank = 24.725 or $1.14, REP mean rank = 17.45 or $0.62) and participants with no training in an independent validation sample (Fig. S2). Redistribution of $2 (rank = 35.5/41) results in an equal distribution between the Dictator and Victim ($5 each). The $1.14 spent by COM was 1.84 times more money than the $0.62 spent by REP and increased the distribution between the dictator and the victim by 57% (inequality of $3.44 instead of $8). In contrast, REP increased the distribution by only 31% (inequality of $5.52 instead of $8). Redistribution responses were rank-transformed based on 41 participants. Error bars denote standard error of the mean rank.

Fig 2

Greater redistribution after COM is predicted by changes in the neural response to suffering. (a) Training-induced (POST-PRE) BOLD changes in right IPC while regulating emotional responses evoked by images of human suffering were differentially associated with post-training altruistic redistribution in COM vs. REP. The cluster was significant after correction (p < .01) using cluster-extent thresholding based on Monte Carlo simulation within the whole brain (Tables S3–4). Images and coordinates are in MNI space. Interaction R2 indicates the proportion of variance in BOLD change accounted for by the Group (COM, REP) × Redistribution interaction. (b) Trainingrelated increases in right IPC activation were associated with greater subsequent altruistic redistribution in COM (n = 20; *** P < 0.001; Table S5), whereas no significant effect was found in REP (n = 21). (c) The right DLPFC was identified by a conjunction test of (1) the correlation with IPC POST-PRE change scores (voxelwise threshold of p < 0.01) and (2) the original Group × Redistribution Interaction (voxelwise threshold of p < 0.01, shown in red-yellow). The cluster was significant after correction (p < 0.01) using cluster-extent thresholding based on Monte Carlo simulation within the whole brain (Tables S3–4). The blue line indicates that the cluster was identified by the conjunction test. (d) Training-related increases in right DLPFC activation were associated with greater redistribution in COM (*** p < 0.001) and less redistribution in REP (* p < 0.05; Table S5).Δ BOLD (% change) in b and d indicates POST-PRE changes in brain response to human suffering (SUFFERING-NEUTRAL), and sr indicates the semipartial correlation of redistribution and neural change in each group. The dashed line indicates redistribution of $2 (rank = 35.5/41) which results in an equal $5 distribution between the dictator and victim.

Fig 3

Greater training-induced connectivity with the DLPFC predicts redistribution in COM vs. REP. (a) DLPFC cluster identified by the conjunction test was used as the seed region in the PPI analysis. (b) Training-related changes (POST-PRE) in right DLPFCNAcc connectivity were differentially associated with post-training altruistic redistribution in COM vs. REP. The cluster was significant after correction (p < .01) using cluster-extent thresholding based on Monte Carlo simulation within a bilateral NAcc ROI (shown in purple; Tables S3–4). Images and coordinates are in MNI space. Interaction R2 indicates the proportion of variance in BOLD change accounted for by the Group (COM, REP) × Redistribution interaction. (c) Training-related increases in right DLPFC-NAcc connectivity were associated with greater redistribution in COM (n = 20;†p = 0.07) and less redistribution in REP (** p < 0.01; Table S5). Δ Connectivity (®) indicates COM-induced changes (POST-PRE) in PPI connectivity betas (SUFFERING vs. NEUTRAL), and sr indicates the semipartial correlation of redistribution and PPI change in each group. The dashed line indicates redistribution of $2 (rank = 35.5/41), which results in an equal $5 distribution between the dictator and victim.