Abstract: PURPOSE: In addition to facial physical properties, the social context in which a face is presented also influences its attractiveness. When a target face is presented in a group, its attractiveness may be higher (cheerleader effect) or lower (reverse cheerleader effect) than when it is presented alone. The visual system processes the group attractiveness, and then the attractiveness of the target face comes toward (assimilation effect) or away from (contrast effect) the group attractiveness, resulting in the different degrees of the cheerleader effect. However, social groups are diverse in the real world, and the whole group can be split into different subgroups. Different social groups may have different weights in group attractiveness, thus moderating the cheerleader effect. For example, the observers’ own-race face captures attention faster than other-race face, thus it may have greater weight in the cheerleader effect. The current study aimed to explore how social context affected the attractiveness of the target face when there were multiple social groups.
METHODS: We conducted three experiments recruiting Chinese university students to explore this question. The whole face group was split into the ingroup and the outgroup of the target face. We manipulated the social context into four conditions: high attractive ingroup and high attractive outgroup (H_IH_O), high attractive ingroup and low attractive outgroup (H_IL_O), low attractive ingroup and high attractive outgroup (L_IH_O), low attractive ingroup and low attractive outgroup (L_IL_O). The cheerleader effect was measured as the attractiveness increment of the target face in each context compared to the presented alone. Experiment 1 used the psychophysical method to calculate the point of subjective equivalence of attractiveness judgment for the target faces in each context, with race as the social group category. We used Black faces and White faces in Experiment 1A, and Asian faces and White faces in Experiment 1B. Experiment 2 and Experiment 3 asked participants rating the facial attractiveness of the target face. Experiment 2 used race as the social group category, with Black faces and White faces in Experiment 2A, Asian faces and White faces in Experiment 2B, and Asian faces and Black faces in Experiment 2C. Experiment 3 showed nationality labels (China, Japan, Singapore) to Asian faces to create Chinese, Japan, and Singapore faces, and manipulated threat priming (Japan: threat, Singapore: non-threat). We used Japanese faces and Singaporean faces in Experiment 3A, Chinese faces and Singaporean faces in Experiment 3B, and Chinese faces and Japanese faces in Experiment 3C.
RESULTS: In Experiment 1, we found the cheerleader effect in L_IL_O, showing the tendency for the contrast effect. In Experiment 2 and Experiment 3, we constructed the linear mixed model to examine the cheerleader effect in each context. In Experiment 2, the cheerleader effect in L_IL_O was always larger than that in H_IH_O. When Black faces were presented with Asian/White faces, the cheerleader effect in L_IH_O was larger than in H_IL_O for Black target faces, and no such difference was observed for Asian target faces or White target faces. When Asian faces and White faces were presented simultaneously, the cheerleader effect in L_IH_O was larger than that in H_IL_O for Asian target faces, and there was no such difference for White target faces. These results manifested the contrast effect with a greater weight to the ingroup of the target faces and the ingroup of participants (i.e., Chinese faces). Experiment 3 replicated the result of Experiment 2 that the cheerleader effect in L_IL_O was consistently larger than H_IH_O. Furthermore, when Japanese and Singaporean faces were presented simultaneously, there was no difference between H_IL_O and L_IH_O. However, when Japanese/Singaporean faces were presented with Chinese faces, there was a tendency of a larger cheerleader effect in H_IL_O than in L_IH_O for the Japanese and Singaporean target face. These results implied the contrast effect with greater weight to participants’ ingroup (i.e., Chinese faces). In addition, the attractiveness of target faces negatively predicted the cheerleader effect in both Experiment 2 and Experiment 3.
CONCLUSIONS: The current study explored the cheerleader effect in multiple social groups and found that the attractiveness of target faces contrasted with the whole group. There was an ingroup bias of both the target face and the observers if social groups were categorized by the attributes (e.g., race) of the face itself, while only an ingroup bias of the observers if social groups were manipulated by the labels. We proposed an observer-target-context model, and suggested that the triadic relationship among the observer, target face, and context faces affected selective attention, social inference mechanisms, etc., moderating the cheerleader effect.

Key words: facial attractiveness, cheerleader effect, social group, selective attention