We perceive facial expressions of emotions in ethnic outgroup members as less intense than those of ingroup members’ expressions, as if they had a more superficial emotional life

Fischer, Agneta, Kai Jonas, and Pum Kommattam. 2019. “Perceived to Feel Less: Intensity Bias in Interethnic Emotion Perception.” Journal of Experimental Social Psychology, Volume 84, September 2019, 103809. https://doi.org/10.1016/j.jesp.2019.04.007

Abstract: The current research focuses on a bias in intensity perception and tests the hypothesis that individuals perceive facial expressions of emotions in ethnic outgroup members as less intense than those of ingroup members’ expressions. In addition to nine previously conducted and reported studies (focussing only on embarrassment, Kommattam, Jonas, & Fischer, 2017, Studies 1 - 9), we conducted a series of three additional studies including white Dutch, U.S., and U.K. participants (N total = 3201) judging the intensity of nine different emotions displayed by different ethnic group members. A random effects model meta-analysis shows that individuals perceive less intense emotions in ethnic outgroup members than in ethnic ingroup members (d = .33 [0.08 – 0.59], (r = .16)). This intensity bias in interethnic emotion perception points to a systematic downplaying of the intensity of outgroup emotions and suggests an empathy gap towards members from other ethnic groups.

Keywords: emotion perception, facial expressions, intensity bias, intergroup empathy, empathy gap

---
General Discussion

Our data provide support for the idea that European descended individuals perceive less intense emotions on the faces of ethnic outgroup members. We used a meta-analytical approach and found an overall effect, showing that white European, European-American, and British perceivers judged emotions expressed by models of Turkish or Moroccan descent as less intense than emotions expressed by Europeans or European-Americans. A random effects model allowed us to estimate the effect size of this intensity bias in the population. The effect was small (d = .33, r = .16) and shows variability (as indicated by the confidence interval [0.08 – 0.59]), but provided clear support for an intensity bias. In Study 5 and 6 reversed effects were found (Kommattam, Jonas, & Fischer, 2017, Study 5 and 6), which may suggest that the intensity bias only occurs when using a specific stimulus set. However, the fact that we again found the bias in a third stimulus set (Study 12) speaks against this explanation. We think that these reverse effects may be due to the unfortunate low power in those two studies (N=59 and N=60 respectively, while N=729 in Study 12).

In the studies included in the meta-analysis, the intercultural or interethnic context was made salient. We would like to stress that we defined this as a precondition for the intensity bias to occur (see also Hogg & Turner, 1987). Indeed, we did not find an intensity bias in earlier studies in which we did not make the intergroup context salient (i.e. without the ranking task)1. These studies were therefore not included in the meta-analysis, as we defined the presence of an interethnic context as one of the inclusion criteria. The salience of the interethnic context aimed to guarantee that the faces of strangers were seen in an interethnic context, and thus categorized as belonging to an ethnic ingroup or outgroup.

We also found evidence for the two boundary conditions. First, information on the emotional context reduces the intensity bias, as reflected in a difference between conditions where emotional (e.g., this woman just won the lottery’) versus neutral or no information was provided. This boundary condition is crucial since clear emotional cues are oftentimes absent in everyday interactions. In professional or educational settings, or when interacting with strangers for example, we often may see subtle signs of emotions, but we do not know what happened to that person, nor do we want to ask. Second, the intensity bias was more pronounced for so-called secondary emotions, namely contempt, pride and embarrassment. We cannot draw conclusions on whether these latter findings support one of the two explanations, however, because the secondary emotions are also least well recognized. In short, the bias may increase when the context is unknown, and when perceivers do not exactly know what happened to a person. Both boundary conditions suggest that more room for interpretation increases the intensity bias in an interethnic context.

The current data speak to various lines of research. The intensity bias may be an indicator of an interethnic empathy gap (Forgiarini, Gallucci, & Maravita, 2011; Gutsell & Inzlicht, 2010; Gutsell & Inzlicht, 2012; Trawalter, Hoffman, & Waytz, 2012), as well as a signal of dehumanization (Harris & Fiske, 2006) or infra humanization (Castano & Giner –Sorolla, 2006; Čehajić, Brown, & Gonzales, 2009; Haslam, Bain, Douge, Lee, & Bastian, 2005; Haslam & Loughnan, 2014; Leyens et al., 2000; Vaes, Leyens, Paladino, & Miranda, 2012; Wohl, Hornsey & Bennett, 2012). We think that the intensity bias especially plays a role in small-scale social interactions, which may lead to misunderstandings, conflicts or exclusion of ethnic outgroup members. The interpretation that members of ethnic outgroups have a more superficial emotional life may lead to ignoring or disregarding their feelings. Such perceptions may in turn be fertile ground for the rise of xenophobia targeting Muslims, migrants, and asylum seekers all over Europe and the United States.

There are also some limitations with regard to our present research. First, it should be emphasized that our data do not justify any conclusions on directionality. It could be that the perception of less intense emotions in ethnic outgroup members may lead to less empathy or more infra-humanization, but the reverse relations are also possible. Further research is needed in order to determine how these two biases are related. In addition, another limitation is that we have only included participants of European descent who perceive emotions displayed by ethnic outgroup members, especially from a Turkish or Moroccan descent. It is important to test a larger variety of perceiver groups in addition to only European descended perceivers in order to examine the generalizability of the effect.

Despite these limitations, we believe that these findings provide new and valuable insights. We demonstrated the presence of an intensity bias across nine different emotions, different ingroup and outgroup models, and with different European descended white samples, with a total 3201 participants. Furthermore, we established the effect using low intensity still images (Kommattam, Jonas, & Fischer, 2017, Studies 1-4 and 7- 9), as well as with standardized dynamic stimuli across three levels of intensity (Study 12). Finally, we performed a meta-analysis, which allowed us to estimate an effect size in the population, unlike null hypothesis significance testing (Field & Gillet, 2010; Rosenthal, 1995). Accordingly, these findings may be seen as a basis for follow up research testing different moderators of the intensity bias.