The role of implicit social bias on holistic processing of out-group faces

Abstract

Although faces of in-group members are generally thought to be processed holistically, there are mixed findings on whether holistic processing remains robust for faces of out-group members and what factors contribute to holistic processing of out-group faces. This study examined how implicit social bias, experience with out-group members, and ability to process in-group faces holistically might predict the magnitude of holistic processing for faces of two out-groups: other-race and other-age groups. In Experiment 1, Caucasian participants viewed Caucasian (own-race) and East Asian (other-race) faces. In Experiment 2, young adult participants viewed young adult (own-age) and baby (other-age) faces. Each participant completed a composite task with in-group and out-group faces, an implicit association test, and questionnaires about their experience with in-group and out-group members. We found that while the participants had relatively extensive experience with the other-race group, they had limited experience with the other-age group. Nonetheless, implicit social bias was found to positively predict the magnitude of holistic processing for both other-race and other-age faces. Exploratory analyses on the interactions among the predictors suggest that the effect of implicit social bias was primarily observed in participants with strong holistic processing ability of in-group faces but with low level of experience with members of the out-groups. These findings suggest that observers utilize different kinds of information when processing out-group faces, and that social features, such as race or age, are incorporated to influence how out-group faces are processed efficiently.

Keywords: Composite task; Face perception; Implicit association test; Individual differences.

Fig. 1

Task design and sample trial sequence of the face composite task (not actual stimuli used in the experiment due to copyright issues; the sample faces were taken from the Radboud Faces Database, Langner et al., 2010). In this task, the face composites were always aligned. The relationship between the top and bottom halves of the study and test faces could be either congruent or incongruent. To improve reliability of the measure by increasing variability in task difficulty (Wang et al., 2016), the faces were shown for 200 ms in the first half of the study and for 150 ms in the second half of the study

Fig. 2

Schematic procedure of the implicit association test (IAT). Participants categorized East Asian and White faces in Block 1, and positive and negative words in Block 2. In Block 3, participants categorized the faces and words presented in a random order. In Block 4, participants again categorized the faces with the response key mapping switched. In Block 5, participants again categorized the randomly presented faces and words

Fig. 3

Mean congruency effect for East Asian faces (∆accuracy), as a function of 1SD above and below the means of the measures of the magnitude of holistic processing of Caucasian faces (strong vs. weak), experience with East Asian individuals (high vs. low), and implicit bias toward East Asians (positive vs. negative)

Fig. 4

Sample baby stimuli in the face composite task (not included in the actual experiment). The face composites were always aligned. The relationship between the top and bottom halves of the study and test faces could be either congruent or incongruent

Fig. 5

Mean congruency effect for baby faces (∆RT), as a function of 1SD above and below the means of the magnitude of holistic processing of adult faces (strong vs. weak) and implicit bias toward babies (positive vs. negative)