Deontic Moral Reasoning Task: Is Moral Reasoning Special?
Keywords:moral reasoning, convention, metacognition, deontic logic
AbstractDomain theory suggests that moral rules and conventions are perceived differently and elicit a different response. A special procedure was designed to test this hypothesis in a laboratory setting using a deontic reasoning task. The goal was to gain insight into the cognitive and metacognitive processes of deontic reasoning from simple deontic premises. In the 3x2x2 within-subjects design, we varied rule-content (moral, conventional, abstract), rule-type (obligation, permission) and the induced dilemma (punishment dilemma, reward dilemma). Participants (N = 78) were presented with 12 laws. After memorizing a law, eight cases were presented to participants so that they make a quick judgment. Participants were tasked with punishing rule-violators, ignoring rule-conformists, and rewarding rule-supererogation. Response times (RT) and accuracy were measured for each judgment, and final confidence was measured after a set of judgments. No differences were expected between rule-types, except for superior performance for moral content and punishment dilemmas. RT correlated negatively with confidence levels, while accuracy correlated positively. Moral reasoning was more accurate than conventional and abstract reasoning, and produced higher confidence levels. Better performance was found for punishment dilemmas than reward dilemmas, likely due to the presence of a cheater-detection module; but the differences were not found in moral reasoning. Moral reasoning was also independent of rule-type, while conventional and abstract reasoning produced superior performance in obligation-type than in permission-type rules. A large drop-off in accuracy was detected for rules that allowed undesirable behaviour, a phenomenon we termed the "deontic blind spot". However, this blind spot was not present in moral reasoning. Three lines of evidence indicate a qualitative difference between the moral and other deontic domains: (1) performance for moral content was independent of rule-type, (2) moral content produced an equal activation of violator- and altruist-detection modules, and (3) moral content produces higher levels of confidence.