Beware a dishonest face?: Updating face-based implicit impressions using diagnostic behavioral information

People quickly infer trustworthiness from faces and use it to guide judgments and behaviors. Past research has suggested that face information is too powerful to be overridden by behavioral evidence, especially implicitly, suggesting that untrustworthiness in faces might be an insurmountable influence in impressions of others. In 5 studies, however, we found that implicit impressions of untrustworthy faces can be effectively updated by learning new, counterattitudinal behavioral information under certain conditions. We showed that face-based implicit impressions were rapidly updated regardless of the degree of untrustworthiness in the face (Study 1), but only when new, propositional information was extreme and diagnostic (Study 2). Successful updating of untrustworthy faces occurred even under low cognitive capacity (Studies 3A and 3B), suggesting efficient processing of inconsistent information. In addition, expanding the conditions under which updating can emerge, we also found that face-based implicit impressions of trustworthiness can be updated using valenced information (positive, negative) rather than just trustworthy-specific counterevidence (Study 4). We discuss the implications of these findings for whether facial cues of untrustworthiness are surmountable in person perception, how people learn about the trustworthiness of others, and the updating of implicit impressions more generally.