Experimental setup, trial sequence (highlighting analyzed epochs) and behavioral data: (A) The Receiver (a confederate) sits in an adjoining testing laboratory to the scanning facility where the Decider (true subject) is undergoing fMRI. The Decider is told that any money left at the end of the task will be randomly multiplied up to 10 times, giving Deciders as much as £200 to take home. The Decider is also required to view, via prerecorded video feed, the administration of any painful stimulation to the Receiver, who is hooked up to an electric stimulation generator. (B) All three tasks (Real PvG, Imagine PvG and Non-Moral task) follow the same event-related design, with the same structure and timing parameters. Our analytical focus was on the Decide event (>11 s). The Video event (4 s), which was spaced a fixed 11 s after the Decide event, was also used in the analysis. (C) Still images of each task illustrating the video the Decider saw while in the scanner: Real PvG video, Imagine PvG video, and Non-Moral video, respectively. VAS scale Deciders used to indicate amount of money to give up/stimulation to deliver per trial. (D) Significantly more Money Kept in the Real PvG Task as compared to the Imagine PvG Task (P = 0.025; error bars = 1 S.E.M). (E) No significant differences between distress levels in response to the Video event across moral tasks.

Real and Imagine Moral networks: (A) Imagine Moral Network: Comparing the Imagine PvG Decide event > Real PvG Decide event reveals significant activation in the PCC, mPFC, posterior parietal cortex, superior frontal sulcus and hippocampus. A priori ROIs (indicated by circles and corrected at P < 0.05 FWE) and parameter estimates reveal that hypothetical moral decisions map closely onto the brain’s construction system. (B). Real Moral Network: Contrasting the Decide event of the Real PvG > Imagine PvG activates bilateral TPJ and amygdala. A priori ROIs and parameter estimates for these regions were found to be more significant during the Real decision than during the Imagine decision. (C). Shared Moral Network: A conjunction analysis of Real and Imagine moral decisions reveals robust activation in the empathy for pain matrix, and parameter estimates of the middle cingulate and bilateral insula illustrate comparable activations for both conditions. All coordinates in MNI space and results portrayed on sections of the mean structural scan at P < 0.005 uncorrected. Both whole brain analysis (P < 0.001 uncorrected) and a priori regions of interest (FWE P < 0.05) were used for all contrasts. A complete list of activated areas and ROIs can be found in Tables 2–4.

Dissociable networks for real selfish and pro-social moral decisions (A) Parametric regression analysis (trial-by-trial) of the Decide event of the Real PvG for increasingly selfish behaviors (greater Money Kept) activates the dACC, bilateral OFC and bilateral dlPFC. A priori ROIs (indicated by circles and corrected at P < 0.05 FWE) were found to be significantly activated for these regions. (B) Parametric regression analysis of the Real PvG Decide event for increasingly pro-social decisions (greater money given up) reveals significant activation in the rostral ACC/mPFC, right temporal pole and right anterior insula. An a priori ROI for the rACC corrected at P < 0.05 FWE was found to be significantly activated. All results portrayed on both axial sections and rendered images at P < 0.005 uncorrected. Both whole brain analysis (P < 0.001 uncorrected) and a priori regions of interest (FWE P < 0.05) were used for all contrasts. All coordinates in MNI space: a complete list of activated areas and ROIs can be found in Tables 6 and 7.