Uncovering religious and occupational stereotypes using implicit face perception measures

Stereotypes refer to the generalizations individuals make about members of social categories, which can affect their thoughts and behaviours towards those who are stereotyped. Stereotypes can be difficult to assess directly. Implicit measures, which tap into attitudes without an individual’s conscious awareness, are therefore useful in this area of research. In the three studies that make up this dissertation, we used implicit face perception methods to uncover stereotypes about religious and occupational groups. In the first study, we used the reverse correlation procedure to visualize and compare the mental representations Christian and Muslim individuals have for their religious ingroup and the outgroups. Our aim was to uncover their religious stereotypes and determine whether they favoured their ingroup, or instead favoured the majority group. First, a set of Christian and Muslim participants selected faces in a two-image forced choice task that resembled Christians and Muslims to them. We averaged the faces they selected to form classification images (CIs) and had a naive set of participants rate them on several demographic and valenced characteristics to reveal their stereotypes and intergroup preferences. We found that the CIs for Christian faces were consistently rated more positively on valenced characteristics than Muslim CIs were, regardless of whether the CI was made of images selected by Christian or Muslim participants. This suggests that a preference for the majority religious group exists among both Christian and Muslim adults in Canada, and this preference is not overridden by ingroup favouritism. In the next study, we tested which cues of religious identity would be effective at signalling religious group membership, leading individuals to categorize faces as members of separate groups. We used a category-contingent aftereffects paradigm, where participants viewed faces belonging Christian and Muslim individuals which were artificially contracted and expanded respectively. The identity of the faces was cued through audio that either explicitly stated their religious affiliation, or stated a food preference or country of origin that was associated with Christianity or Islam. If the cues led to the perception of discrete groups, we would observe opposing changes in preference for Christian and Muslim faces (e.g., a preference for contracted Christian faces and expanded Muslim faces), known as a category- contingent aftereffect. We observed significant category-contingent aftereffects in the audio conditions with explicit religious labels and food preferences, but not country of origin. This suggests that the first two cues are effective at signalling group membership, enough that they act as a top-down influence on the unconscious process of face perception, and may be leading to rapid categorization and stereotyping in social interactions. In the final study, we used the reverse correlation procedure once again to study stereotypes towards scientists, rather than religious groups, and compare them to stereotypes of heroes, geniuses, and the superordinate “person” category. First we presented our participants with a two-image forced choice task where they selected images that looked like a scientist, hero, genius, and person in separate blocks. We averaged the images they selected to create CIs for each category, and then had a naive set of participants rate them on demographic and valenced traits. We found that the Scientist CI was rated as more White and male than the Person CI, which suggests that scientists are stereotyped as the most historically represented group in the sciences. The Scientist CI was also rated lower than the other CIs on some valenced traits suggesting that scientists are stereotyped as being unsociable, incompetent, and poor communicators.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.