A Sensitive and Specific Neural Signature Robustly Predicts Graded Computations of Moral Wrongness

Humans universally condemn what they see as moral violations, yet the perceived wrongness of distinct moral transgressions varies across individuals, neurodiverse populations, and cultures. We currently do not know how and to what extent the human brain universally computes and translates moral wrongness into subjective, graded moral judgments. Here, we combined fMRI with pattern recognition techniques to identify and evaluate a neural signature predictive of graded moral wrongness judgments. Drawing on to date’s largest database for studying the neural basis of moral judgment, spanning independent, multi-culture fMRI datasets of moral vignettes for discovery (n = 64), validation (n = 30), replication (n = 27), and generalization (n = 30) analyses (n _total =151), we demonstrate that accurate prediction of graded moral wrongness relies on a distributed neural circuit, with important contributions from cortical and subcortical areas. We further evaluate common and domain-isolated (e.g., care, fairness, purity) brain systems for graded moral wrongness and demonstrate shared and dissociable neural representations with negative affect and subjective disgust. Together, we find that graded moral wrongness judgments are robustly computed via a shared and distributed neural code and provide a sensitive and specific moral wrongness biomarker for future studies.