ReportThe social face of emotion recognition: Evaluations versus stereotypes
Introduction
Fast and correct recognition of emotional expressions is a prerequisite for fluent social interaction. A face carries a wealth of informative cues and can say more than a thousand words. Apart from unchangeable features in the face that carry information regarding identity or social category (e.g. sex or ethnicity), changeable features reveal how someone is feeling at that specific moment (e.g. happy, angry or sad). Previous research has shown that unchangeable features can influence the interpretation of changeable facial features in emotional recognition. More specifically, this research focused on the interplay between social categorization and recognition of emotional expressions (e.g. Becker et al., 2007, Elfenbein and Ambady, 2002, Hugenberg, 2005, Hugenberg and Bodenhausen, 2003, Hutchings and Haddock, 2008).
How do social categorization processes influence emotion recognition? One route that has been recently explored is that evaluative category associations facilitate or inhibit the recognition of evaluative congruent emotions. For example, Hugenberg (2005) demonstrated, in line with an evaluative processes account, that white American participants show a recognition speed advantage when judging whether a white target’s face displays a positive (happy) facial expression compared to negative (angry or sad) facial expressions. This phenomenon is called the happy face advantage (Leppänen & Hietanen, 2003). This effect was found by Hugenberg to be reversed when the expresser of the emotion was black, with white participants showing a recognition advantage for both negative emotions (angry and sad) compared to the positive emotion. In a similar vein, Hugenberg and Sczesny (2006) found evidence for differences in emotion recognition depending on gender: for female target faces a greater happy face advantage was found than for male target faces. The authors conclude that these effects are best explained by the spreading of evaluative associations: the valence of the target face, as triggered by the category, serves as a prime for recognizing emotional expressions independently of the discrete emotion being expressed.
However, a vast amount of research revealed evidence that social categories not only trigger evaluative associations, but also stereotype associations (e.g. Devine, 1989). The latter contain specific traits and cognitions that are linked to the social category. An increasing body of evidence emphasizes the importance of differentiating between evaluative and stereotype associations. For example, Amodio and Devine (2006) recently demonstrated that evaluative and cognitive processes differentially influence behavioral responses. Moreover, they argue that different neurological mechanisms underlie these effects. However, in emotion recognition literature such a distinction between processes has yet to be made.
We argue that also stereotype associations influence the ease of emotion recognition. Specifically, discrete emotions such as anger and sadness differ from each other despite both having a negative valence. Similarly to traits, some discrete emotional expressions are more stereotypical for some social categories than for others, independent of their general valence. For instance, anger is more strongly associated with men than women, while sadness is more strongly associated with women than men (Fischer et al., 2004, Plant et al., 2000). Although both emotions have a negative valence, they are associated with different social categories. In this sense, emotions do not differ from other stimuli or characteristics that may have a better normative fit with one social category than the other (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Importantly, these stereotype associations may facilitate or inhibit the recognition of emotional expressions (e.g. sadness is recognized faster on female faces than on male faces). In this way, stereotypical views may confirm themselves in biased emotion recognition.
Interestingly, the findings of Hugenberg, 2005, Hugenberg and Sczesny, 2006 did not reveal any evidence of stereotype effects on emotion recognition. No categorical differences in the processing of discrete negative emotions between social categories were found. Therefore, their findings suggest that social category information influences emotion recognition by means of general evaluative connotations with the category, rather than by specific associations between the category (e.g. black targets) and discrete emotional expressions (e.g. anger). In our view, however, these null findings for stereotype associations could be due to the research method that was employed in these earlier studies. In prior work, participants categorized both positive and negative emotional expressions on faces of two distinct social categories within one task (e.g. Hugenberg, 2005, Hugenberg and Sczesny, 2006). As a result, differences in valence were highly salient within the task and participant performance was influenced by general affective reactions towards the social categories under investigation. Consequently, recognition effects concerning discrete emotional expressions may have been overshadowed by general evaluative effects.
The goal of the present two studies was to demonstrate the influence of not only valence but also of stereotype associations on emotion recognition. In order to do this, comparative context was taken into account. On the one hand, the experimental design employed ‘dual-valence comparative context conditions’ which contained a positive versus a discrete negative emotional expression. On the other hand, we used a ‘single-valence comparative context condition’ in which two distinct negative emotional expressions were compared. In the dual-valence comparative context, participants categorized happy versus angry, or happy versus sad expressions, whereas in the single-valence comparative context, anger versus sadness were categorized. In the remainder of this article we will refer to these comparative context conditions as single- or dual-valence conditions.
When people are asked to classify two distinctive negative emotional expressions, the influence of valence on emotion recognition advantages is thought to be decreased. If category information influences emotion recognition only by means of valence, both negative emotions should be recognized faster when expressed by a member of a relatively negatively associated social category compared to members of a more positively evaluated social category. However, if specific stereotype associations (or cognitive instead of affective processes) influence emotion recognition speed, especially the discrete negative emotional expression most strongly associated with the social category should be recognized faster compared to less strong or unassociated negative emotional expressions. We therefore expected associated category-valence to influence emotion recognition speed in the dual-valence conditions, while stereotype associations were expected to facilitate emotion recognition processes in the single-valence condition. The social categories employed were White Dutch and Moroccan Dutch men (Study 1) and men versus women (Study 2).
Section snippets
Study 1: White Dutch versus Moroccan Dutch
The Moroccan Dutch community is currently one of the most negatively stereotyped groups in the Netherlands (Verkuyten & Zarembe, 2005) and is associated with criminality (Dotsch, Wigboldus, Langner, & van Knippenberg, 2008), a trait more strongly related to anger than to sadness. Furthermore, Otten and Stapel (2007) demonstrated that Dutch students perceive the Moroccan Dutch social category as more aggressive and hostile than the White Dutch social category. We expected that in dual-valence
Participants and design
A total of 64 white European participants (56 women, mean age = 20.75) at the Radboud University Nijmegen were randomly assigned to one of three comparative context conditions (dual-valence: ‘happy versus angry’ and ‘happy versus sad’; single-valence: angry versus sad). In each condition, participants saw two emotional expressions displayed by both White Dutch and Moroccan Dutch targets.
Materials and procedure
Eight models, four White Dutch and four Moroccan Dutch, were selected from the Radboud Faces Database on the
Results
The primary dependent variable in this study is the mean response time needed to categorize emotional expressions. Before analyzing the data, incorrect trials (8.20%) and response latencies below 200 ms or above 3000 ms (<1%) were excluded. Due to the skewed distribution of the response latencies, all analyzes were performed on log-transformed response latencies. To facilitate the interpretation of our findings, we report mean response latencies in untransformed milliseconds. In order to test
Discussion
In line with our hypotheses, a two-way interaction between target ethnicity and expressions valence but no three-way interaction was found in the dual-valence conditions. This suggests that the evaluative connotation with the social category influences emotion recognition in situations where people have to classify positive and negative emotional expressions within one task. Furthermore, we found evidence for stereotype-congruency when judgments could not be made on a more general affective
Study 2: male versus female
In our second study we aimed to replicate the findings of Study 1 using gender as social category, which allowed us to test a full design in which male and female participants judged male and female target faces. Research has shown that both men and women are evaluated positively, however, women are evaluated more positively than men (e.g., Eagly and Mladinic, 1989). In the dual-valence conditions this should lead to a pattern of results similar to the effects found in Study 1. That is, in line
Participants and design
A total of 117 participants (81 women, mean age 22.48) at the Radboud University Nijmegen were randomly assigned to one of three comparative context conditions (dual-valence: ‘happy versus angry’ and ‘happy versus sad’; single-valence: angry versus sad) in which they saw male and female targets displaying two emotional expressions.
Materials and procedure
Twelve models, six White Dutch men and six White Dutch women, were selected from the RaFD, again, on the basis of recognition data from a validation study (Langner et
Results
As with Study 1, the primary dependent variable was the mean response time needed to categorize the displayed emotional expressions. Incorrect responses (10.50%) and response latencies falling outside the range of 200–3000 ms (<1%) were excluded from subsequent analyzes. An analysis with participants’ gender in both, the single- and dual-valence, comparative context condition as between subject factor revealed no main or interaction effects and therefore is not included in the analyzes. Similar
General discussion
The results from these two studies strongly suggest that comparative context moderates the influence of social categorization on emotion recognition. The goal of the present research was to show that depending on the comparative context, emotion recognition could be influenced by both spreading activation of evaluative associations and by stereotypically based associations. Specifically, we expected and found that when positive and negative emotional expressions (dual-valence conditions) were
Acknowledgment
We thank Daniel Fitzgerald for his comments on this manuscript.
References (20)
- et al.
Look black in anger: The role of implicit prejudice in the categorization and perceived emotional intensity of racially ambiguous faces
Journal of Experimental Social Psychology
(2008) - et al.
Stereotyping and evaluation in implicit race bias: Evidence for independent constructs and unique effects on behavior
Journal of Personality and Social Psychology
(2006) - et al.
The confounded nature of angry men and happy women
Journal of Personality and Social Psychology
(2007) - et al.
Gender stereotypes and attitudes toward women and men
Personality and Social Psychology Bulletin
(1989) Are there basic emotions?
Psychological Review
(1992)- et al.
On the universality and cultural specificity of emotion recognition: A meta-analysis
Psychological Bulletin
(2002) Stereotypes and prejudice. Their automatic and controlled components
Journal of Personality and Social Psychology
(1989)- et al.
Ethnic out-group faces are biased in the prejudiced mind
Psychological Science
(2008) - et al.
Gender and culture differences in emotion
Emotion
(2004) Social categorization and the perception of facial threat: Target race moderates the response latency advantage for happy faces
Emotion
(2005)
Cited by (82)
Facial first impressions following a prison sentence: Negative shift in trait ratings but the same underlying structure
2024, Journal of Experimental Social PsychologyThe effect of face masks on the stereotype effect in emotion perception
2022, Journal of Experimental Social PsychologyCitation Excerpt :Following Bayesian models, the perception of the stimulus is influenced by both the prior and the visual input, which means that both sources of information interact: The weaker the visual input (i.e., 50% emotion intensity) and the stronger the prior (i.e., stronger emotion – social category associations), the stronger the tendency to respond in a stereotype-congruent manner regardless of the emotion – social category combination actually being presented. In the research of Bijlstra et al. (2010), the likelihood can be described as P(face | emotion), the posterior as P(emotion | face), and the prior, which is influenced by participants stereotypical associations between Moroccan-Dutch and anger, and White-Dutch and sadness, respectively, as P(emotion). The experimental design facilitated reliance on the prior, i.e., the stereotypes, because the emotional faces were presented only briefly.
Gender Biases in Estimation of Others’ Pain
2021, Journal of PainWhat factors predict anti-Black bias in pain perception? An internal meta-analysis across 40 experimental studies
2024, Social and Personality Psychology CompassFacial Expression Stereotypes of Male and Female Adults and Children: Do Facial Expression Stereotypes of Adults Apply to Children?
2024, Archives of Sexual Behavior