Seeing a face for the first time elicits spontaneous inferences about the personality traits of that person. The notion that facial appearance is related to inner character is common and has a long tradition which was formally conceptualized in a book on physiognomy by Lavater (1775–1778; Hassin & Trope, 2000). More recently, research has shown not only that trait inferences are pervasive but also that there is a surprising consistency in adults' trait impressions based on facial appearance (e.g., Carré, McCormick, & Mondloch, 2009; Zebrowitz & Rhodes, 2002, Oosterhof & Todorov, 2008). While the ability to make quick decisions given limited information is an adaptive behavioral skill, making decisions about who a person is, given how they look, runs the risk of amplifying implicit biases and racial stereotypes to which all humans are susceptible (Marini & Banaji, 2020).
Implicit biases can be elicited by different types of social cues such as socioeconomic status, or cultural, national, racial, ethnic, and gender identity (Kaufmann et al., 2017; Oh, Shafir, Todorov, 2020; Olivola & Todorov, 2010; Zebrowitz & Montepare, 2010). Racially based stereotypes, feelings, beliefs, and notions can lead to negative judgements of individuals of other races and can have a profound impact on interaction between individuals of different races (Pennington et al., 2016). Dovidio and Gaertner (1986), in their theory of Aversive Racism, posited that stereotypical qualities of Black people and other racial minorities affected how White Americans reacted around them even when maintaining that discrimination is wrong. Despite the strong subjective impression that one might be racially unbiased, it has been shown that subliminal biases appear not in self-report but, for instance, in measures of association strength between skin color (which informs, but does not fully explain, perceived race and ethnicity) (Uhlmann et al., 2002) and affectively valenced words (Greenwald, Ghee & Schwartz, 1998). Implicit racial attitudes have subtle but measurable influences on the judgement of trustworthiness in strangers’ faces (Stanley, Sokol-Hessner, Banaji, & Phelps, 2011). Such implicit tests allow us to assess biases that people are either unaware of or might not want to disclose.
One established measurement device for these biases, the Implicit Association Test (Greenwald, Ghee & Schwartz, 1998; Lane, Kang & Banaji, 2007), has been used extensively in social psychology research. Several meta-analyses show that this measure is reliably replicable, although there are debates in the field about the extent to which it can predict biased behavior (Greenwald, Poehlman, Uhlmann & Banaji, 2009; Blanton, Jaccard, Klick, Mellers, Mitchell, & Tetlock, 2009). A meta-analysis of the race IAT has found that it explains on average 5% of variance in racially biased behavior (Greenwald, Poehlman, Uhlmann, & Banaji, 2009). The race IAT is suitable to reveal a bias that exists within a population, but it is not well suited to predicting behavior on an individual basis (Blanton, Jaccard, Klick, Mellers, Mitchell, & Tetlock, 2009; Hehman, Calanchini, Flake & Leitner, 2019). Therefore, it should not be interpreted as a diagnostic tool for measuring racial bias on the participant level. However, there is some evidence that the IAT could be diagnostic of biases at the population level (Hehman, Calanchini, Flake & Leitner, 2019). This metric, as the name clearly suggests, is simply an approximate measure of implicit bias. Because of its extensive use in research in social psychology, we have adopted the IAT in the present study to measure a racial bias toward White and Black faces (Gawronski, 2002; Vianello & Bar-Anan, 2020).
The aim of the present study was to investigate whether there is a reliable relationship between implicit racial biases and the way we make snap judgments about new faces. In contrast to previous studies, we focused on judgments of attractiveness as well as inferences about several different personality traits based on facial appearance. We tested trait inferences for Black faces, White faces and faces from a third race: East Asian faces. We found that implicit racial biases predict biases in ratings of attractiveness and inferred personality traits, with implicit preference for Black faces predicting higher attractiveness ratings for Black faces compared to White faces (and vice versa). We were also able to predict how a participant would rate attractiveness and some personality traits of East Asian faces relative to White faces based on skin tone preference as measured with the IAT. Stronger implicit preferences for White faces predicted more positive ratings of White faces relative to East Asian faces. The race IAT did not predict how a participant would judge East Asian faces relative to Black faces.