Implicit evaluations of moral agents reflect intent and outcome

Explicit (directly measured) evaluations of moral agents reflect both the externally observable consequences of actions and inferences about the agents hidden mental states: Negative outcomes without negative intent (e.g., someone getting killed accidentally) and negative intent without a negative outcome (e.g., a failed attempt to kill someone) are each sufficient for negative explicit evaluations of a moral agent to emerge. Across two studies (final N = 826; Study 2 preregistered), we newly investigated implicit (indirectly measured) evaluations of moral agents, as assessed by an Implicit Association Test (IAT). Study 1 included 3 between-participant conditions: accident (negative outcome + positive intent), attempt (positive outcome + negative intent), and harm (negative outcome + negative intent), each compared to a harmless (positive outcome + positive intent) control. Study 2 had a 2-by-2 design, in which outcome (positive vs. negative) and intent (positive vs. negative) were manipulated orthogonally, with targets in each condition compared to a neutral control whose actions did not carry moral implications. Mirroring prior findings obtained using explicit measures, implicit evaluations of moral agents tracked both manifest outcomes (e.g., someone falling from a bridge) and inferences about latent mental states (e.g., the intent to let someone fall off a bridge) in both paradigms. These results are difficult to reconcile with dual-process theories positing that implicit evaluations arise from low-level associative learning but they are readily accounted for by propositional theories according to which implicit evaluations are sensitive to high-level inferential reasoning.