On the wrong side of the trolley track: neural correlates of relative social valuation

Abstract

Using moral dilemmas, we (i) investigate whether stereotypes motivate people to value ingroup lives over outgroup lives and (ii) examine the neurobiological correlates of relative social valuation using fMRI. Saving ingroup members, who seem warm and competent (e.g. Americans), was most morally acceptable in the context of a dilemma where one person was killed to save five people. Extreme outgroup members, who seem neither warm nor competent (e.g. homeless), were the worst off; it was most morally acceptable to sacrifice them and least acceptable to save them. Sacrificing these low-warmth, low-competence targets to save ingroup targets, specifically, activated a neural network associated with resolving complex tradeoffs: medial PFC (BA 9, extending caudally to include ACC), left lateral OFC (BA 47) and left dorsolateral PFC (BA 10). These brain regions were recruited for dilemmas that participants ultimately rated as relatively more acceptable. We propose that participants, though ambivalent, overrode general aversion to these tradeoffs when the cost of sacrificing a low-warmth, low-competence target was pitted against the benefit of saving ingroup targets. Moral decisions are not made in a vacuum; intergroup biases and stereotypes weigh heavily on neural systems implicated in moral decision making.

Fig. 1

The stereotype content model warmth by competence space, stereotyped group exemplars and associated emotions. Source data from Fiske et al. (2002).

Fig. 2

An example of a stimulus presentation block. Each dilemma comprised the following sequence: a picture of the 1 person sacrificed (2 s), a collaged picture of the five people saved (2 s), a response prompt regarding the acceptability of Joe’s actions (4 s), followed by an interstimulus interval of 12 s, during which participants passively viewed a fixation cross in the center of the screen, allowing the hemodynamic response to return to baseline after each trial. The targets in each dilemma were randomly selected (see Methods section).

Fig. 3

Warmth by competence interaction predicting moral acceptability of saving targets from each of the four SCM quadrants. Bars represent standard error.

Fig. 4

Selected brain regions (see Table 3) exhibiting significantly increased activity for sacrificing low warmth, low competence to save high warmth, high competence as compared to the other 15 conditions: mPFC (BA 9, extending caudally to include ACC), left lateral OFC (BA 47), left DLPFC (BA 10), bilateral precuneus (BA 7) and left posterior cingulate (BA 30). Statistical maps of voxelwise t-scores were thresholded for significance (P < 0.001) and cluster size (≥34 voxels). (A) Sagittal slice plane is x = −6; (B) axial slice plane is z = 16; (C) coronal slice plane is y = 33 (Talairach and Tournoux, 1988). Images (B) and (C) are reversed right to left according to radiologic convention.