Abstract
Over a century of research has focused on the consistency and inconsistency of human moral decision preferences. We proposed and found that moral decision preferences are flexible and shift towards newly learned moral rules when their application leads to utilitarian choices. Hence, decision-makers' psychological concept of morality is continually under construction (on the fly); based on learning, informed by changes in moral rules and specific moral contexts. Accordingly, in two experiments we developed and employed a two-stage supervised learning task, where participants learned novel moral rules based on corrective feedback of their moral decisions. Our empirical findings revealed that participants learn new moral rules, transfer these rules to tasks where no feedback is provided. However, participants make decisions based on the principle of maximizing utility rather than a learned rule when the rule conflicts with this principle, demonstrating further the flexibility of moral decision-making. In light of our proposal and findings that decision-makers' psychological concept of morality is continually under construction, moral decision-making researchers should integrate learning into their respective models and predictions.