Neurocomputational mechanisms of personal benefits and justifications in moral decision-making

Confronted with the temptation to cheat, people do not only care about their personal benefits, but also consider whether they can justify their dishonest behavior. Previous research has shown how reward and justifications alone influence dishonest decisions. However, little is known about how they jointly guide dishonest behaviors and which neurocomputational mechanisms are involved. Here, we combined Bayesian computational modeling with neuroimaging techniques to investigate the neurocomputational processes underlying the weighting of justifications and personal benefits during moral decision-making. Using a perceptual decision-making task, we found that participants were more likely to be lured into dishonest choices by rewards (greater personal benefits) when perceptual ambiguity was higher (lying was more easily justifiable), and that individual, model-based weighting of perceptual ambiguity more strongly predicted participants' dishonesty. Model-based choice values correlated with neural activity in the medial prefrontal cortex (PFC) and bilateral anterior insula (AI), with left AI predicting honest choices. On the contrary, more dishonest behavior was associated with stronger functional connectivity between left AI and ventrolateral PFC. These results demonstrate how individuals integrate different aspects of moral behaviors, unifying various predictions of previous accounts on human morality.