The many faces of compensation: The similarities and differences between social and facial models of perception.

Previous research shows that stereotypes can distort the visual representation of groups in a top-down fashion. In the present endeavor, we tested if the compensation effect-the negative relationship that emerges between the social dimensions of warmth and competence when judging two social targets-would bias the visual representations of these targets in a compensatory way. We captured participants' near spontaneous facial prototypes of social targets by means of an unconstrained technique, namely the reverse correlation. We relied on a large multi-phase study (N = 869) and found that the expectations of the facial content of two novel groups that differed on one of the two social dimensions are biased in a compensatory manner on the facial dimensions of trustworthiness, warmth, and dominance but not competence. The present research opens new avenues by showing that compensation not only manifests itself on abstract ratings but that it also orients the visual representations of social targets.

Influence of membership in outgroups varying in competence and warmth on observers' Level-2 visual perspective taking.

Visual perspective taking (VPT), the ability to adopt another person's viewpoint, entails two distinct processes, Level-1 (L1)-VPT and Level-2 (L2-VPT), referring to the ability to perceive whether and how a target sees an object, respectively. Whereas previous efforts investigated the impact of targets' social characteristics on L1-VPT, the present work is the first to do so regarding L2-VPT. Specifically, we investigate the impact of targets' membership in outgroups varying in perceived competence and warmth, the two fundamental dimensions of social perception. Participants in four experiments engaged in a L2-VPT task. Avatars belonged to a low competence low warmth group (LCLW; e.g. the homeless) or to a high competence low warmth group (HCLW; e.g. bankers) in Experiments 1-3, and to a LCLW or high competence high warmth group (HCHW; e.g. female students) in Experiment 4. Participants answered as quickly as possible whether a cued number matched a number present in a scene from either their own or the avatar's perspective. We consistently found support for the presence of both egocentric and altercentric interference, but this was not modulated by group competence and warmth, suggesting that membership in outgroups varying in competence and warmth does not influence L2-VPT. We discuss the findings' implications in the light of recent views on VPT.