Several empirical studies have documented an asymmetry in people’s assessments of intentional action, so-called ‘Knobe effect’. Accordingly, foreseen (yet undesired) outcomes that are harmful are judged intentional, whereas foreseen (yet undesired) outcomes that are helpful are judged unintentional. The Knobe-effect has been standardly conceived of in bivalent terms: The presence or absence of perceived intentionality contingent on a negative or positive outcome valence. Unsurprisingly, explanat…
Read moreSeveral empirical studies have documented an asymmetry in people’s assessments of intentional action, so-called ‘Knobe effect’. Accordingly, foreseen (yet undesired) outcomes that are harmful are judged intentional, whereas foreseen (yet undesired) outcomes that are helpful are judged unintentional. The Knobe-effect has been standardly conceived of in bivalent terms: The presence or absence of perceived intentionality contingent on a negative or positive outcome valence. Unsurprisingly, explanations thereof have a similar bivalent structure: Intentionality ascriptions in Knobe-effect cases are viewed as contingent on the presence or absence of a binary feature—a blameworthy agent, a norm violation, a morally bad outcome, and so on. In this paper, we report the results of two experiments exploring attributions of intentionality (and knowledge) across a range of graded outcomes: very bad, somewhat bad, neutral, somewhat good and very good. The findings suggest that the Knobe-effect data points are but two data points of a broader, more fine-grained phenomenon, and that Knobe effect explanations that have conceived of it in bivalent terms are at best incomplete.