Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Pretty crowds are happy crowds: the influence of attractiveness on mood perception

Pretty crowds are happy crowds: the influence of attractiveness on mood perception Empirical findings predominantly support a happiness superiority effect in visual search and emotion categorization para- digms and reveal that social cues, like sex and race, moderate this advantage. A more recent study showed that the facial attribute attractiveness also influences the accuracy and speed of emotion perception. In the current study, we investigated whether the influence of attractiveness on emotion perception translates into a more general evaluation of moods when more than one emotional target is presented. In two experiments, we used the mood-of-the-crowd (MoC) task to investigate whether attractive crowds are perceived more positively compared to less attractive crowds. The task was to decide whether an array of faces included more angry or more happy faces. Furthermore, we recorded gaze movements to test the assumption that fixations on happy expressions occur more often in attractive crowds. Thirty-four participants took part in experiment 1 as well as in experiment 2. In both experiments, crowds presenting attractive faces were judged as being happy more frequently whereas the reverse pattern was found for unattractive crowds of faces. Moreover, participants were faster and more accurate when evaluating attractive crowds containing more happy faces as well as when judging unattractive crowds composed of more angry expressions. Additionally, in experiment 1, there were more fixations on happy compared to angry expressions in attractive crowds. Overall, the present findings support the assumption that attractiveness moderates emotion perception. Introduction postulated for angry faces (Hansen & Hansen, 1988; Öhman, Lundqvist, & Esteves, 2001; Pinkham, Griffin, Baron, Sas- The fast and correct identification of emotional expressions son, & Gur, 2010), but recent studies have found compelling in human faces is essential for social interactions, because evidence for an advantage for happy faces, termed “hap- facial expressions signal a person’s potential intentions and piness superiority effect” (Becker, Anderson, Mortensen, behavior (Calvo & Marrero, 2009). While a smiling face sig- Neufeld, & Neel, 2011; Savage, Lipp, Craig, Becker, & nals affiliation and approachability, an angry face signals a Horstmann, 2013). These studies used a visual search para- lack of approachability or even threat of aggression (Scherer digm (the “face-in-the-crowd paradigm”, FitC), in which & Wallbott, 1994). Due to the centrality of facial emotion participants have to decide as quickly and accurately as perception for adaptive social interaction, many theorists possible whether an array of faces (all displaying the same have argued that humans can automatically detect signs of expression, e.g. all neutral/happy/angry) contains a deviating affiliation or threats of aggression in faces (e.g., Öhman & target face. In this paradigm, the happiness superiority effect Mineka, 2001). Building on this assumption, some have pos- manifested itself in faster reaction times and lower error tulated that specific emotional expressions have a perception rates for targets with happy expressions compared to angry advantage over others. Initially, a perception advantage was but also other negative expressions (Becker et al., 2011; Sav- age et al., 2013). In emotion categorization tasks, a faster and more correct categorization of happy facial expressions * Alica Mertens was also observed (Leppänen & Hietanen, 2004; Leppänen, alica.mertens@psychologie.uni-heidelberg.de Tenhunen, & Hietanen, 2003). Institute of Psychology, Heidelberg University, Hauptstr. Previous empirical studies suggest that there are several 47-51, 69117 Heidelberg, Germany facial features that can influence the size of the happiness Department of Psychosomatic Medicine, Central Institute superiority effect, including sex and race (e.g., Becker, Ken- of Mental Health, Medical Faculty Mannheim/Heidelberg rick, Neuberg, Blackwell, & Smith, 2007; Craig & Lipp, University, Mannheim, Germany Vol.:(0123456789) 1 3 1824 Psychological Research (2021) 85:1823–1836 2017; Bijlstra, Holland, & Wigboldus, 2010). Specifically, this study was too easy and incurred very low error rates that the happiness superiority effect was larger for female tar - might impede the detection of response biases. At the same gets (Becker et al., 2007; Bucher & Voss, 2019; Bucher, time, the absence of a detection advantage for unattractive Voss, Spaniol, Hische, & Sauer, 2019; Craig & Lipp, 2017; angry faces can also be seen as evidence against evaluative Hugenberg & Sczesny, 2006) and for male own-race targets congruence. The present study aimed to replicate findings compared to male other-race targets (Bijlstra et al., 2010; by Lindeberg and colleagues in a novel paradigm that over- Craig, Mallan, & Lipp, 2012). In addition to this research, a comes some of the limitations of single target paradigms recent study also found that facial attractiveness moderates and includes eye-tracking analyses to assess underlying pro- the happiness superiority effect (Lindeberg, Craig, & Lipp, cesses. This way, we hoped to shed light on whether unat- 2019). In four experiments, the authors found evidence for tractive-angry faces indeed do not entail a detection advan- a faster and more accurate identification of happiness (ver - tage (as suggested by the previous evidence) and which sus anger) in attractive faces compared to unattractive faces, would reject the evaluative congruence account, or whether when presenting one target in each trial. this effect would be present in a paradigm that leaves more The evidence of a faster and more accurate identification room for variance in participant performance. of attractive faces displaying happiness can be explained Specifically, the present study relied on a novel paradigm through an interplay of two theoretical accounts: the attrac- that incurs higher cognitive demands and lower accuracy tiveness stereotype and the evaluative congruence account. rates than single-target tasks. The current paradigm might The attractiveness stereotype, also known as the “what is thus leave more room for motives to bias emotion percep- beautiful is good” effect (Dion, Berscheid, & Walster, 1972), tion and enables testing the evaluative congruence account describes the phenomenon that humans believe physical for both attractive and unattractive faces. In the “mood-of- attractiveness to be associated with a diverse range of posi- the-crowd paradigm” (MoC), participants have to judge tive attributes (for a meta-analysis, see Eagly, Ashmore, the overall mood of the crowd by indicating whether more Makhijani, & Longo, 1991a, b). For example, attractiveness angry or happy faces are present in an array of faces (Bucher increases the likelihood of getting a job interview (Watkins & Voss, 2019). Previous studies were able to replicate the & Johnston, 2000), entails higher ratings on favorable per- happiness superiority effect in this setting: happy crowds sonality traits such as Agreeableness (Borkenau & Liebler, were identified faster and more accurately than angry crowds 1992, 1995; Smits & Cherhoniak, 1976), and higher intel- (Bucher & Voss, 2019; Elias, Dyer, & Sweeny, 2017). The ligence ratings (Jackson, Hunter, & Hodge, 1995). Thus, MoC paradigm has a number of advantages over the FitC the attractiveness stereotype suggests that attractiveness and paradigm. First, judging the mood of a crowd represents positive attributes are strongly associated in the minds of a more ecologically valid task than detecting a single face people. To explain, why the pairing of attractive happy faces in a crowd, because crowds with multiple target emotions entails a faster and more accurate processing, one can addi- are highly prevalent in daily life (e.g., in class, at a sport- tionally draw on the evaluative congruence account (Hugen- ing event, etc.). Moreover, the presentation of more than berg, 2005; Hugenberg & Sczesny, 2006). The evaluative one emotional target prevents confusion about whether an congruence account postulates that emotions are processed observed effect results from a target or a crowd effect (see faster when they match the evaluation of a social cue that Bucher & Voss, 2019). As outlined above, the MoC task is they are presented in conjunction with, for instance attrac- also a more complex and cognitively demanding task and tiveness. Based on the attractiveness stereotype account, we therefore does not incur similar floor effects for error rates know that attractiveness is evaluated positively, as is a happy compared to single target paradigms. Furthermore, the MoC affect. Therefore, happiness and attractiveness are both posi- task can be well combined with process tracing methods tive cues and, based on the evaluative congruence account, such as eye-tracking. Eye-tracking is less applicable to the should be processed faster than if the pairing of affect and a FitC paradigm, for example, because the search process ends social cue were incongruent (e.g. angry + attractive). abruptly as soon as the participant detects the target, which However, the attractiveness stereotype also has a flipside: makes it difficult to identify underlying search patterns. Humans also tend to ascribe negative qualities to unattrac- The implementation of eye-tracking analysis is advanta- tive persons (Dion et al., 1972). Based on the evaluative geous for the current investigation of the evaluative congru- congruence account, one would therefore expect that unat- ence account for facial attractiveness because it makes it tractive faces paired with a negative emotion (e.g. anger) possible to pinpoint the locus of the effect of attractiveness are also processed faster and more accurately. However, on emotion perception. If an attractiveness effect was appar - this effect was not observed in the one previous study that ent in eye movements (e.g., an increased probability and investigated the categorization of attractive and unattractive longer duration of fixations on happy-attractive/angry-unat - faces displaying happy and angry affect (Lindeberg et al., tractive faces), this would indicate that congruent stimuli 2019). It is possible that the categorization task applied in (e.g. happy-attractive faces) are preferentially or selectively 1 3 Psychological Research (2021) 85:1823–1836 1825 attended to and thus attentional processes guide the judge- whether the gender-attractiveness imbalance accounts for the ment of whether the crowd is perceived as predominantly reported attractiveness effect on emotion perception. happy or angry. In contrast, if participants showed no differ - We derived our hypotheses based on the evaluative con- ences in their fixation patterns between congruent (happy- gruence account (Hugenberg, 2005; Hugenberg & Sczesny, attractive/ angry-unattractive) and incongruent (happy-unat- 2006) which suggests a processing advantage for emotions tractive/angry-attractive) stimuli, but did show differences at that are evaluated congruently as the social cue they are the choice level, this would suggest effects in the evaluation paired with (herein attractiveness). Therefore, we hypothe- phase (e.g., a biased response towards happiness for attrac- sized that participants would identify attractive crowds more tive crowds). In this case, participants attend to both con- often as happy than unattractive crowds. Reversely, we also gruent and incongruent stimuli in a similar way but the final expected that participants would identify unattractive crowds judgment is biased towards one emotion, indicating that the more often as angry than attractive crowds. Moreover, we evaluation of the selected faces influences the final judgment hypothesized that attractive crowds containing a higher about the general mood of the crowd. Without the inclusion number of happy faces, and unattractive crowds containing of eye-tracking measures, this differentiation is not possible. a higher number of angry expressions, should entail faster and more accurate responses. This effect has previously been supported by Lindeberg et al. (2019), who observed a faster The current study and more accurate categorization of happy attractive com- pared to angry attractive expressions. Further corroborating The aim of the current study was to test the evaluative con- this, there is evidence that participants evaluate happy faces gruence account, thus also aiming for conceptual replication as more attractive than faces displaying negative emotions of the findings by Lindeberg et al. (2019) in a new paradigm. (Mueser, Grau, Sussman, & Rosen, 1984). Moreover, Golle, The present study used the MoC paradigm, in which the par- Mast and Lobmaier (2014) showed that judgment of rela- ticipant’s task is to decide as fast and accurately as possible tively happier compared to neutral faces is facilitated when whether more angry or more happy faces are presented in a those faces were attractive. Based on evaluative congruence, crowd of faces. The merits of this task (compared to single we expected a general tendency to judge attractive crowds target paradigms) include increased cognitive demand (and as being happy and unattractive crowds as being angry more less likelihood of floor effects), increased ecological validity, frequently. With regard to eye-tracking indices, we expected separation of crowd and target effects, and the possibility of higher fixation rates and fixation durations on happy-attrac- including eye-tracking indices in addition to response times tive and angry-unattractive faces than on happy-unattractive and accuracy rates. Note that in other paradigms (e.g., cat- or angry-attractive faces, based on a higher salience of con- egorization task or FitC paradigm) the use of single targets gruently evaluated social cues. prevents investigation of attentional preferences regarding specific emotions or attractiveness-emotion combinations. For example, in the FitC paradigm, the search process stops Experiment 1 abruptly as soon as the target is detected and in single face categorization tasks, there is no possibility to investigate Method search asymmetries. The MoC paradigm, however, allows for analyzing such gaze patterns. An inclusion of eye-track- Participants ing was desirable in this case to test whether perceptual preferences guide the evaluation of the crowd, indicated Prior to recruiting participants, we conducted a power analy- by an increased fixation rate on happy attractive and angry sis using G*Power 3 (Faul, Erdfelder, Lang, & Buchner, unattractive faces, or whether attractiveness influences the 2007) to determine the required sample size. The required speed and accuracy of the judgements itself, indicated by a sample size to detect an effect of medium size with a power lack of an attractiveness effect in gaze movements. Lastly, of 0.80 and an alpha-error of 0.05 in a repeated measures the present study aimed to target a confound that was present ANOVA setting was 34. Thirty-four adults participated in in the studies by Lindeberg et al. (2019), including the fact that attractive female faces were consistently rated as more attractive compared to attractive male faces, whereas there 1 When conducting the power analysis for our studies we used the was no difference for unattractive male and female faces. studies by Lindeberg et  al. (2019) as a basis. They predominantly found medium effect sizes for the interaction effect between emo- Thus, we used the same selection criterion for target faces as tional expression and attractiveness regarding accuracy and catego- Lindeberg et al. (2019) in our first experiment, but matched rization time across their four experiments. Although the task is not target gender and attractiveness in the second experiment. entirely comparable to the MoC task, we perceived this effect size to Using these different selection criteria, we were able to test be the best anchor for our studies. 1 3 1826 Psychological Research (2021) 85:1823–1836 the study (M = 23.06 years, SD = 5.22, range 19–47; 50% the criteria of a minimal duration of 100 ms and maximal age female). We recruited participants from a large participant dispersion of 100 pixels. pool at a German university, in which students of diverse After providing written, informed consent, participants subjects are registered. Before participants took part in the were seated approximately 60 cm in front of the laptop com- study, they received information about the study details and puter. First, participants provided demographic information provided written, informed consent. As compensation for the and then received information that the upcoming task was study, participants could choose between course credit or a to judge whether the crowd contained more happy or angry financial compensation of five Euros. faces. We specifically instructed participants to make their decisions as fast and accurately as possible to prevent them Stimulus material from simply counting the presented faces. The experiment comprised a practice block of 16 trials and two experimental Following Lindeberg et al. (2019), we chose the stimulus blocks of 50 trials each. Prior to each block (practice block material for the first experiment by selecting the most and and experimental blocks) we used a 9-point-calibration for least attractive faces from a pool of faces. We selected the calibrating the eye-tracker. Each calibration was followed by stimulus faces from the Chicago Face Database (Ma, Cor- a 4-point validation. After a successful validation (partici- rell, & Wittenbrink, 2015), restricting the stimulus material pants’ gaze was within 1° visual angle), participants started to faces of Caucasian individuals that provided both angry with the experimental trials. In case of a non-successful vali- and happy (closed mouth) expressions. We used the attrac- dation, we repeated calibration and validation. tiveness ratings of the remaining 37 female and 36 male In each trial, participants saw a crowd of 18 faces faces provided by the norming data of the Chicago Face (each 1.4  cm × 1.9  cm). Faces were randomly allo- Database and selected nine faces that were rated the most cated on the screen in front of a light-grey background attractive and nine faces that achieved the lowest attractive- (34.5 cm × 19.5 cm) to ensure naturalistic eye movements, ness ratings for female and male faces, respectively. This which is not possible when presenting faces in a circle or procedure resulted in 18 female (nine attractive and nine matrix (due to systematic scanning paths: e.g. clockwise unattractive) and 18 male (nine attractive and nine unattrac- or row-wise). To avoid overlap between the pictures, we tive) individuals with happy and angry expressions availa- ensured a minimum distance between the edges of the pic- ble. The attractiveness ratings of the ‘attractive’ female mod- tures (minimum distance was 3.4 cm). Crowds contained els (M = 4.60, SD = 0.32; Models 3, 11, 12, 15, 22, 24, 25, a varying number of angry faces (6, 8, 10 and 12 angry 27 and 29) differed from those of the ‘unattractive’ female faces), and crowds were either completely made up of models (M = 2.32, SD = 0.45; Models 2, 8, 10, 23, 26, 28, 30, attractive or unattractive faces. The different compositions 34 and 37). Likewise, the attractiveness of the ‘attractive’ of angry and happy faces within the crowds (six angry/12 male models (M = 3.89, SD = 0.46; Models 3, 4, 6, 9, 14, 15, happy, eight angry/ten happy, ten angry/eight happy, 12 24, 29 and 33) differed from that of the ‘unattractive’ male angry/six happy) occurred equally often in attractive and models (M = 2.36, SD = 0.16; Models 2, 10, 17, 19, 35, 37, unattractive crowds to prevent any systematic confound. 38, 39, 41). Lighting and visual contrast were similar across Moreover, crowds always contained the same number of all faces in the set. female and male faces (e.g., in case of ten angry and eight happy faces the crowd contained five female angry and five Procedure male angry expressions as well as four female happy and four male happy expressions). All presented face pictures The experiment was run on a Dell laptop computer. Stimulus within one trial were from different individuals. displays and response time measurement were controlled by Each block started with two warm-up trials. The eight a C program using SDL libraries (www.libsdl.or g). We used trial combinations (attractiveness of presented faces × num- a SMI RED250MOBILE eye-tracker to record gaze move- ber of angry pictures) were presented six times in random ments at a frequency of 250 Hz. We defined fixations using order in each experimental block. Prior to each trial, a We chose the composition of the number of happy/angry faces The norming data of the Chicago Face Database includes mean based on the last experiment by Bucher and Voss (2019) which also attractiveness ratings of 1087 participants (Ma et  al., 2015) for each included four trial types (7, 9, 11, 13 angry expressions in crowds of face picture. Because only the averaged ratings of each picture are 20 faces). However, because we wanted to balance target gender in available, the selection is based descriptively on these mean attrac- our crowds of faces, we needed to build crowds of even numbers of tiveness ratings. Therefore, we report the mean and standard devia- angry and happy expressions. Therefore, we changed the total number tion averaged for the selected pictures of the respective types (female of pictures from 20 to 18 and changed the number of presented angry attractive, female unattractive, male attractive, male unattractive). faces to 6, 8, 10 and 12. 1 3 Psychological Research (2021) 85:1823–1836 1827 fixation cross was presented at the center of the screen before attractive, happy and female targets were rated as being more the crowd was shown. Only when participants fixated on this attractive compared to the unattractive, angry and male tar- cross, the crowd appeared. If participants did not focus on gets. Moreover, we found a significant interaction between the fixation cross until 5000 ms were exceeded, a recalibra- attractiveness and target gender, F(1, 32) = 97.83, p < 0.001, tion started. The crowd remained visible until participants η = 0.75, 95% CI [0.61, 0.82]. Follow-up pairwise com- gave their response. If participants’ responses were slower parisons revealed that for attractive faces, female targets than 8000 ms, a message (“please try to respond faster”) were perceived as more attractive compared to male targets, appeared. Participants could press the S-key and the K-key t(33) = 9.06, p < 0.001, d = 1.55, whereas the unattractive to indicate whether they perceived the crowd as predomi- male and female targets did not differ significantly, t < 1. nantly happy or angry. The assignment of the emotions to We ran the same analysis for the emotional intensity rat- the keys was balanced across participants. As a reminder, the ings. There was a significant main effect of attractiveness, words “happy” and “angry” appeared on the respective sides F(1, 32) = 32.24, p < 0.001, η = 0.50, 95% CI [0.28, 0.63], on the bottom of the screen. Subsequent to the eye-tracking and gender, F(1, 32) = 40.48, p < 0.001, η = 0.56, 95% CI experiment, participants rated the happy and angry faces on [0.34, 0.68], showing that attractive and female targets were attractiveness, happiness, and anger on a 7-point Likert scale perceived as showing a higher emotional intensity. The main in a randomized sequence. effects were qualified by an attractiveness × target gender interaction, F(1, 32) = 15.89, p < 0.001, η = 0.33, 95% CI Analysis [0.12, 0.50] and an attractiveness × emotional expression interaction, F(1, 32) = 71.26, p < 0.001, η = 0.69, 95% CI A 2 (attractiveness: attractive, unattractive) × 2 (trial type: [0.52, 0.77]. Follow-up pairwise comparisons showed that mainly happy, mainly angry ) repeated measures ANOVA female expressions were rated as more emotionally inten- was conducted to analyze responses, accuracy rates and sive compared to male faces both for attractive, t(33) = 2.37, response times. For the analyses of the number of fixations p = 0.024, d = 0.41, and unattractive targets, t(33) = 7.45, and fixation duration on the respective faces, the emotion of p < 0.001, d = 1.28. However, this difference was much more the focused picture (happy vs. angry) served as an additional pronounced for unattractive faces, t(33) = 3.96, p < 0.001, within-subjects factor. This way, we assessed how many d = 0.68. Additionally, pairwise comparisons revealed that faces of each type received at least one fixation during a trial happiness was expressed more strongly on attractive com- and how long each type of face was fixated. We excluded pared to unattractive faces, t(33) = 8.37, p < 0.001, d = 1.44, responses faster than 500 ms and slower than 8000 ms from whereas there was no significant difference for angry faces, the response time analysis (0.51% of all trials). Addition- t(33) = − 1.90, p = 0.066, d = 0.33. ally, we adjusted for participants’ gender as a covariate in Mean and standard deviations of the attractiveness and all analyses. If not reported in the text, gender did not show emotional intensity ratings are summarized in Table 1. any significant main effects or interactions. Response time Results We computed mean correct response times for each par- ticipant and each factorial combination. There was a sig- Manipulation check nificant interaction between attractiveness and trial type, F(1, 32) = 6.18, p = 0.018, η = 0.16, 95% CI [0.02, 0.34]. Attractiveness ratings were obtained in a 2 (emotional Follow-up pairwise comparisons revealed that participants expression: happy, angry) × 2 (attractiveness: attractive, were faster when more happy faces were presented in attrac- unattractive) × 2 (target gender: female, male) repeated tive crowds compared to unattractive crowds, t(33) = 2.97, measures ANOVA. There was a main effect of attractive- p = 0.006, d = 0.51, whereas there was no difference when ness, F(1, 32) = 472.53, p < 0.001, η = 0.94, 95% CI [0.89, more angry faces were shown, t < 1 (Fig. 1a). 0.95], emotional expression, F(1, 32) = 15.53, p < 0.001, η = 0.33, 95% CI [0.11, 0.49], and gender, F(1, 32) = 46.78, Response p < 0.001, η = 0.59, 95% CI [0.39, 0.70], confirming that Mean responses (proportion of “angry” responses) were cal- culated for each participant and each factorial combination. To ease interpretation of the results, trial types containing more A main effect of attractiveness, F (1, 32) = 32.53, p < 0.001, happy faces (6 and 8 angry out of 18 faces) and those who consisted 2 η = 0.50, 95% CI [0.28, 0.63], and a main effect of trial of more angry faces (10 and 12 angry out of 18 faces) were com- type emerged, F(1, 32) = 315.00, p < 0.001, η = 0.91, 95% bined. When analyzing the data using the original four categories of CI [0.85, 0.93]. Attractive crowds were judged as “happy” trial type, findings were largely identical. 1 3 1828 Psychological Research (2021) 85:1823–1836 Table 1 Attractiveness and emotional intensity ratings for happy and angry female and male faces from both experiments Measures Female Male Happy Angry Happy Angry Experiment 1  Attractiveness   Attractive 5.63 (0.69) 5.06 (0.86) 3.99 (1.23) 3.52 (1.04)   Unattractive 2.76 (1.07) 2.13 (0.84) 2.66 (1.13) 2.11 (1.04)  Emotional intensity   Attractive 5.59 (0.73) 5.41 (0.73) 5.36 (0.55) 5.35 (0.63)   Unattractive 5.20 (0.66) 5.69 (0.64) 4.76 (0.73) 5.23 (0.71) Experiment 2  Attractiveness   Attractive 4.55 (0.76) 3.70 (0.96) 4.17 (0.94) 3.52 (1.10)   Unattractive 3.40 (0.84) 2.51 (0.82) 3.13 (0.84) 2.36 (0.75)  Emotional intensity   Attractive 5.18 (0.69) 5.19 (0.74) 5.11 (0.71) 4.97 (0.88)   Unattractive 4.92 (0.69) 5.42 (0.80) 5.08 (0.71) 5.23 (0.75) Values in parantheses represent 1 SD and unattractive ones as “angry” more frequently (Fig. 2a). Participants evaluated crowds containing more happy faces as being happy more often compared to those with more angry faces. Accuracy Mean accuracies were calculated for each participant and each factorial combination. There was a main effect of trial type, F(1, 32) = 4.23, p = 0.048, η = 0.12, 95% CI [0.00, Fig. 1 Interaction effect of dominant emotion and attractiveness on 0.29], indicating slightly increased accuracy rates in trials response times (in ms) for experiment 1 (a) and experiment 2 (b). with more angry faces. This main effect was qualified by Trials with ten and 12 happy expressions were combined as well as an attractiveness × trial type interaction, F(1, 32) = 32.53, trials with ten and 12 angry expressions. Error bars indicate standard p < 0.001, η = 0.50, 95% CI [0.28, 0.63]. Follow-up t errors tests indicated that when more happy faces appeared in a crowd, participants showed higher accuracy rates for attractive crowds, t(33) = 6.31, p < 0.001, d = 1.08, and the a main effect of emotional expression, F(1, 32) = 8.30, reversed pattern when confronted with unattractive faces, p = 0.007, η = 0.21, 95% CI [0.04, 0.38], and attractive- t(33) = − 3.34, p = 0.002, d = 0.57 (Fig. 3a). ness, F(1, 32) = 10.20, p = 0.003, η = 0.24, 95% CI [0.05, 0.42], indicating that happy faces and unattractive faces were Fixation duration fixated more frequently compared to angry and attractive faces. Moreover, we found a significant interaction between Mean durations were calculated for each participant and attractiveness and emotional expression, F(1, 32) = 4.65, each factorial combination. There was no significant main or p = 0.039, η = 0.13, 95% CI [0.00, 0.30]. Follow-up pair- interaction effect of any of the investigated factors, F < 2.10, wise comparisons revealed that in unattractive crowds happy p > 0.16, η < 0.06. and angry expressions were fixated to the same degree, t < 1, whereas in attractive crowds significantly more happy faces Number of fixations were fixated, t(33) = 3.85, p = 0.001, d = 0.66 (Fig. 4). There was a significant interaction between emotional expression The number of fixations assesses how many faces of each and trial type, F(1, 32) = 301.35, p < 0.001, η = 0.90, 95% type received at least one fixation during a trial. There was CI [0.84, 0.39], with more fixations on angry (happy) faces 1 3 Psychological Research (2021) 85:1823–1836 1829 Fig. 2 Main effect of attractiveness on response tendency (0 = happy, 1 = angry) for experiment 1 (a) and experiment 2 (b). Error bars indi- Fig. 3 Interaction effect of dominant emotion and attractiveness on cate standard errors accuracy (0 = false, 1 = correct) for experiment 1 (a) and experiment 2 (b). Trials with ten and 12 happy expressions were combined as well as trials with ten and 12 angry expressions. Error bars indicate stand- ard errors when more angry (happy) faces were presented. Lastly, an interaction between all factors emerged, F(1, 32) = 4.55, 2 6 p = 0.041, η = 0.13, 95% CI [0.00, 0.30]. Discussion In experiment 1, we assessed whether stimulus attractiveness When conducting the post-hoc analyses to shed light on this inter- has an effect on the happiness superiority effect in visual action between all factors, no meaningful interpretation of this effect search, specifically in the MoC paradigm. We hypothesized could be revealed. Thus, this interaction effect does not contribute to a faster and more accurate perception of attractive crowds the other findings from our analyses. containing more happy faces and of unattractive crowds We re-ran all analyses and included the difference scores of the emotional intensity ratings for happy attractive and happy unattrac- comprising more angry faces. We expected a response bias tive expressions (happy attractive minus happy unattractive) as well towards happiness in attractive crowds and towards anger as for angry attractive and angry unattractive expressions (angry in unattractive crowds. Results revealed that participants attractive minus angry unattractive) to account for this influence on judged crowds containing attractive faces as happy more our reported findings. However, including these difference scores did not alter the effects. There was still a significant effect of attractive- ness on response tendency, F(1, 30) = 32.25, p < 0.001, η = 0.52, Footnote 6 (continued) 95% CI [0.29, 0.65], a significant interaction between attractiveness and dominant emotion on response time, F(1, 30) = 5.95, p = 0.021, p < 0.001, η = 0.52, 95% CI [0.29, 0.65], and a significant interac- η = 17, 95% CI [0.01, 0.35], a significant interaction between tion between attractiveness and target emotion on number of fixa- attractiveness and dominant emotion on accuracy, F(1, 30) = 32.25, tions, F(1, 30) = 4.38, p = 0.045, η = 0.13, 95% CI [0.00, 0.31]. 1 3 1830 Psychological Research (2021) 85:1823–1836 experiment and chose different selection criteria for the stimulus material. Experiment 2 In experiment 1, the focus when choosing the stimulus material was to achieve the greatest possible attractiveness difference between unattractive and attractive targets. This approach leads to a larger difference in perceived attractive- ness and maximizes the chances of finding an attractiveness effect. Lindeberg and colleagues (2019) critically discussed this selection procedure and demonstrated that it leads to a greater difference in rated attractiveness for attractive female compared to attractive male targets. As target gender also influences emotion perception to a large degree, it might be Fig. 4 Interaction effect of target emotion and attractiveness on num- the case that results from experiments using this selection ber of fixations. Error bars indicate standard errors method are somewhat confounded. Therefore, in experi- ment 2, we used different selection criteria for the mate- often than crowds containing unattractive faces. Moreover, rial to ensure that attractive female and male targets as well participants were faster and more accurate when crowds as unattractive male and female targets match with regard were both attractive and containing many happy faces. In to their perceived attractiveness. Furthermore, we used the line with the evaluative congruence account, we also found emotion intensity ratings for the neutral faces provided in the that participants judged unattractive crowds as angry more norming data of the Chicago Face Database to better match frequently, and that judgments were more accurate in trials the attractive and unattractive face pictures. Despite these with more angry targets presented. This is a novel finding, as modifications, the experimental setting and hypotheses in the previous study testing this model (Lindeberg et al., 2019) experiment 2 were identical to experiment 1. only found support for part of the evaluative congruence account, showing an advantage for attractive-happy faces but not for unattractive-angry targets. Further extending Methods previous findings, we included eye-tracking analyses that revealed that fixations occurred more frequently on happy Participants compared to angry facial expressions when attractive faces were presented. Based on the same power analysis as in experiment 1, we The results supported our hypotheses and showed recruited 34 adults (M = 23.26  years, SD = 5.66, range age medium to large effect sizes, but were limited by an imbal- 18–46; 53% female) via a large participant pool of a Ger- ance between target gender and attractiveness. Thus, we man university. Before starting with the experiment, par- cannot completely rule out the possibility that this gender- ticipants received information about the upcoming task and attractiveness disparity affected our results. Mirroring the provided written, informed consent. Again, participants findings by Lindeberg et al. (2019), the difference between could receive course credit or five Euros as compensation attractiveness ratings was also larger for attractive female for their participation. compared to attractive male faces. Furthermore, we found that participants rated happy attractive faces as more emo- Stimulus material tionally intense than happy unattractive faces. Although we believe that attractiveness influenced the emotion rat- We again selected the stimulus material from the Chicago ings to the same degree as accuracy, response tendency and Face Database (Ma et al., 2015), considering only pictures response times in the MoC task, other explanations might be possible as well. It is imaginable that the completion Although the emotion ratings for the neutral expressions do not of the MoC task influenced the following ratings, because necessarily match the ratings of the actual emotions, it nevertheless we observed the same pattern of results for the experiment provides a tendency that might influence the intensity of the emo- and the ratings. Moreover, it is possible that unattractive tional expressions. targets showed less intense happy expressions than attrac- We used the same power analysis criteria for the second experi- tive targets. To address these issues, we conducted a second ment, to allow comparability between the two experiments. 1 3 Psychological Research (2021) 85:1823–1836 1831 of Caucasian men and women that provided both happy and Analysis angry expressions. In addition to the attractiveness ratings, happiness and anger ratings were taken into account when The analytic strategy was identical to that in experiment 1. choosing the stimulus material. Happiness and anger ratings Again, we excluded responses faster than 500 ms and slower of the neutral face pictures of the unattractive and attractive than 8000 ms from the response time analysis (0.43% of all male and female pictures were matched to control for pos- trials). sible emotional intensity differences between the pictures. Furthermore, we matched target gender and perceived attrac- tiveness, so that attractive males and females as well as unat- Results tractive males and females had similar attractiveness rating. We again selected 18 female (nine attractive and nine unat- Manipulation check tractive) and 18 male (nine attractive and nine unattractive) individuals. When comparing the attractiveness ratings, the Ratings from the 45 participants of the pretest were used attractive female (M = 3.95, SD = 0.38; M = 2.47, to analyze perceived attractiveness and emotional intensity. att att happy SD = 0.49; M = 2.60, SD = 0.65; Models 6, 11, There were main effects of attractiveness, F (1, 43) = 169.28, happy angry angry 13, 15, 16, 18, 21, 25 and 31) and unattractive female mod- p < 0.001, η = 0.80, 95% CI [0.70, 0.85], target emotion, els (M = 2.79, SD = 0.17; M = 2.41, SD = 0.54; F(1, 43) = 40.79, p < 0.001, η = 0.49, 95% CI [0.30, 0.61], att att happy happy p M = 2.46, SD = 0.76; Models 5, 7, 8, 19, 23, 28, and target gender, F(1, 43) = 14.15, p = 0.001, η = 0.25, angry angry 30, 36 and 37) as well as the attractive male (M = 3.89, 95% CI [0.08, 0.40], indicating that attractive, happy and att SD = 0.46; M = 2.63, SD = 0.32; M = 2.21, female faces were rated as more attractive compared to att happy happy angry SD = 0.39; Models 3, 4, 6, 9, 14, 15, 24, 29 and 33) unattractive, angry and male faces. This time, there was angry and unattractive male models (M = 2.71, SD = 0.13; no significant interaction between attractiveness and target att att M = 2.48, SD = 0.67; M = 2.39, SD = 0.50; gender, F(1, 43) = 0.63, p = 0.432, η = 0.01, 95% CI [0.00, happy happy angry angry p Models 12, 13, 20, 21, 23, 25, 32, 34 and 37) differed in 0.12]. Lastly, we found a significant interaction between tar - rated attractiveness, respectively, but not with regard to hap- get emotion and target gender, F(1, 43) = 9.56, p = 0.003, piness or anger. Lighting and visual contrast were similar η = 0.18, 95% CI [0.04, 0.34]. Follow-up pairwise com- across all faces in the set. parisons revealed that happy faces were rated more attrac- As the attractiveness, happiness and anger ratings were tively compared to angry faces both for female, t(44) = 8.25, only available for the neutral face expressions, the happy p < 0.001, d = 1.23, and male faces, t(44) = 6.74, p < 0.001, and angry expressions of the attractive and unattractive male d = 1.00, however, this difference was significantly larger for as well as female individuals were additionally rated by 45 female faces, t(44) = 2.76, p = 0.008, d = 0.41. participants (M = 34.98 years, SD = 13.60, range 19–59; With regard to emotional intensity, there was a signifi- age 73% female). We present the results for the attractiveness cant main effect of target gender, F (1, 43) = 13.48, p = 0.001, and emotional intensity ratings in the results section. η = 0.24, 95% CI [0.07, 0.39], and participants’ gender, F(1, 43) = 9.64, p = 0.003, η = 0.18, 95% CI [0.02, 0.37], Procedure indicating that female faces were perceived as more emo- tionally intense and that female raters gave higher inten- The experimental setup and procedure were identical to sity ratings. The main effect of target gender was qualified those in experiment 1. The only difference was with regard by an interaction between target gender and target emo- to the stimulus material and that the emotional face expres- tion, F(1, 43) = 8.13, p = 0.007, η = 0.16, 95% CI [0.03, sions were rated prior to the study regarding attractiveness 0.32]. Whereas happy and angry faces were perceived as and emotional intensity (happiness, anger) by an independ- equally intense in male faces, t < 1, angry female faces ent sample to prevent transfer effects. were rated more intensely compared to happy female faces, t(44) = − 2.65, p = 0.011, d = 0.40. Furthermore, there was a signic fi ant interaction between attractiveness and target emo - tion, F(1, 43) = 34.89, p < 0.001, η = 0.45, 95% CI [0.26, There were several reasons why we decided to recruit a separate 0.58]. Happy attractive faces were rated more intensely than sample to rate the attractive and unattractive emotional face expres- happy unattractive faces, t(44) = 3.76, p < 0.001, d = 0.56, sions. In the first experiment, ratings were collected after the MoC experiment, so it might be possible that the completion of the experi- whereas the reverse pattern was found for angry expres- mental task influenced the attractiveness and emotion ratings after - sions, t(44) = − 4.94, p < 0.001, d = 0.74. Lastly, a signifi- wards. Furthermore, when measuring the ratings prior to the experi- cant interaction between participants’ gender and target gen- ment, it might happen that the ratings influence the completion of the der emerged, F(1, 43) = 11.34, p = 0.001, η = 0.21, 95% experimental task afterwards which is also not ideal. Collecting the p ratings in a “pretest” therefore seemed appropriate. CI [0.05, 0.37]. Male participants rated female faces more 1 3 1832 Psychological Research (2021) 85:1823–1836 emotionally intense compared to male face, t(44) = 4.56, Fixation duration p = 0.001, d = 0.68, whereas there was no such difference for female participants, t < 1. Mean durations were calculated for each participant and Mean and standard deviations of the attractiveness and each factorial combination. Male participants fixated sig- emotional intensity ratings are summarized in Table 1. nificantly longer on the presented faces compared to female participants, F(1, 32) = 4.92, p = 0.034, η = 0.13, 95% CI Response time [0.01, 0.31]. The only other significant effect was a three- way interaction between emotional expression, trial type and We computed mean correct response times for each par- participants’ gender, F(1, 32) = 4.47, p = 0.042, η = 0.12, ticipant for each factorial combination. Again, there was a 95% CI [0.00, 0.30]. Whereas there was no significant inter - significant interaction between attractiveness and trial type, action between emotional expression and trial type for male 2 2 F(1, 32) = 10.39, p = 0.003, η = 0.25, 95% CI [0.06, 0.42]. participants, F(1, 15) = 1.14, p = 0.303, η = 0.07, 95% CI Follow-up pairwise comparisons showed that in attractive [0.00, 0.30], there was a marginal significant interaction for crowds participants were faster when crowds contained female participants, F(1, 17) = 3.82, p = 0.067, η = 0.18, more happy compared to angry faces, t(33) = 2.57, p = 0.015, 95% CI [0.00, 0.41], indicating that women fixated longer d = 0.44, whereas in unattractive crowds the pattern pointed on angry (happy) faces when the crowds consisted of mainly towards the opposite direction, t(33) = − 1.91, p = 0.065, angry (happy) faces. d = 0.33 (Fig. 1b). Lastly, there was a significant effect of participants’ gender, F(1, 32) = 4.55, p = 0.041, η = 0.12, Number of fixations 95% CI [0.00, 0.34], indicating that women were faster in judging the mood of the crowd. This measure indicates how many faces of each type received at least one fixation during a trial. A significant Response interaction between emotional expression and trial type was found, F(1, 32) = 201.13, p < 0.001, η = 0.86, 95% CI [0.77, We calculated mean responses (proportion of “angry” 0.90], with more fixations on angry (happy) faces when responses) for each participant and each factorial combina- more angry (happy) faces were presented. Lastly, there was a tion. Again, there was a main effect of attractiveness, F (1, significant interaction between attractiveness and trial type, 2 2 32) = 68.62, p < 0.001, η = 0.68, 95% CI [0.50, 0.77], and F(1, 32) = 9.31, p = 0.005, η = 0.23, 95% CI [0.05, 0.40]. A of trial type, F(1, 32) = 316.51, p < 0.001, η = 0.91, 95% CI higher number of faces was fixated in attractive compared to [0.85, 0.93]. Participants tended to judge attractive crowds unattractive crowds when more angry faces were presented, more often as happy, whereas they judged unattractive t(33) = 2.38, p = 0.024, d = 0.41, whereas there was no sig- crowds more often as angry (Fig. 2b). Participants evalu- nificant difference in crowds containing more happy faces, ated crowds dominated by happy faces as being happy more t(33) = − 1.69, p = 0.101, d = 0.29. often compared to those containing more angry expressions. Accuracy Discussion Mean accuracies were calculated for each participant and In experiment 2, we replicated the main findings of experi- each factorial combination. A main effect of trial type ment 1 in a different sample and using new stimulus material reached significance, F (1, 32) = 5.37, p = 0.027, η = 0.14, that was better matched for attractiveness between female 95% CI [0.01, 0.32], showing higher accuracy rates in trials and male stimuli. Again, participants evaluated attractive with more angry expressions. This main effect was qualified crowds as happy more frequently than unattractive crowds. by an attractiveness × trial type interaction, F(1, 32) = 68.62, Reversely, participants evaluated unattractive (compared p < 0.001, η = 0.68, 95% CI [0.50, 0.77]. Follow-up t to attractive) crowds more often as angry. Moreover, par- tests revealed that accuracy rates were increased when ticipants were faster and more accurate in judging attractive more happy faces were presented in attractive compared crowds dominated by happy faces and unattractive crowds to unattractive crowds, t(33) = 6.50, p < 0.001, d = 1.11, dominated by angry expressions. These findings are in line and the reversed pattern was found for angry expressions, with the evaluative congruence account, which suggests a t(33) = − 5.41, p < 0.001, d = 0.93 (Fig. 3b). facilitated perception when the presented emotion matches the evaluation of the respective social cue. Extending previ- ous findings by Lindeberg et al. (2019), we found evidence 10 supporting the evaluative congruence account not only for The findings with regard to participants’ gender need to be inter - preted with caution as ratings from only 12 men were available. attractive, but also for unattractive faces. 1 3 Psychological Research (2021) 85:1823–1836 1833 In contrast to the findings for reaction times and choices, faces in attractive crowds and higher fixation rates on angry which were well in line with the findings from experiment faces in unattractive crowds. 1, eye-tracking results did not replicate as closely. In experi- Across both experiments, we found evidence for an inu fl - ment 1, we observed an increased number of fixations on ence of attractiveness on emotion perception. Participants in happy attractive compared to angry attractive faces but this both samples evaluated attractive crowds of faces as happy was not the case in experiment 2. It is possible that the dif- more frequently. Moreover, their evaluations were faster ference in material explains this discrepancy. As described and more accurate in attractive crowds dominated by happy above, we performed a new matching of the target material faces. These findings are in line with the evaluative con - to reduce attractiveness die ff rences between female and male gruence account (Hugenberg, 2005; Hugenber & Sczesny, faces that were present in experiment 1. While successful in 2006). The results of the two studies also corroborate recent this regard, the new matching also resulted in smaller overall evidence by Lindeberg et al. (2019), who found that par- differences between attractive and unattractive targets. It is ticipants categorized attractive happy faces faster and more possible that the attention capturing power of the attractive accurately than unattractive happy faces. smiling faces was thereby diminished, leading to a similar In contrast to previous studies, we also found evidence allocation of attention to attractive and unattractive emo- for evaluative congruence with regard to unattractive-angry tional face expressions. faces. Participants rated unattractive crowds as angry more Even though we were able to match the perceived attrac- frequently and evaluated these crowds faster and more accu- tiveness of female and male targets in experiment 2, we rately, when they were dominated by angry faces. Therefore, again found that participants rated happy attractive faces as we were not only able to find support for the evaluative con - more emotionally intense compared to happy unattractive gruence account with regard to attractive crowds of faces, faces and the opposite for angry attractive and unattractive but also for unattractive facial expressions. The reason for face pictures. This time, we collected ratings in a separate this might be that the MoC paradigm exerts higher cognitive sample, ruling out the possibility that the completion of the demands and thus prevents floor effects (i.e. very low error MoC task influenced the consecutive rating. Furthermore, rates with little variance) from occurring. we consider it unlikely that, again, the emotional intensity of A further limitation that was present in previous studies the happy (angry) unattractive faces was truly lower (higher) we also observed in the present experiment 1. Participants compared to the happy (angry) attractive faces as we used rated attractive female faces as more attractive than attrac- two different stimulus sets from the Chicago Face Database. tive male faces, whereas there was no such difference with Therefore, the most plausible explanation might be that respect to unattractive male and female faces (and the same attractiveness also influenced/biased the emotion intensity limitation afflicted studies by Lindeberg et al., 2019). To ratings. Using computer-generated emotional expressions rule out the possibility that the happiness superiority effect matched for attractiveness, gender, and emotional intensity for attractive crowds might be due to this dissimilarity, we would be necessary in future investigations to ensure equal matched female and male faces on attractiveness for the intensities of the emotional faces. second experiment. Even though this led to smaller attrac- tiveness differences between the unattractive and attractive faces, we were able to replicate our findings. General discussion To additionally address potential mechanisms underlying the attractiveness effect on emotion perception, we combined The present study set out to test the evaluative congruence the MoC paradigm with eye-tracking analyses. We hypoth- account, which assumes a faster perception of an emotion esized that evaluative congruence would manifest in higher when it matches the evaluation of a social cue, in two experi- fixation rates and longer fixation durations on happy-attrac- ments. The social cue tested in the present study was attrac- tive and angry-unattractive faces than on happy-unattractive tiveness and we employed a visual search paradigm, the or angry-attractive faces. We partly found this only in the MoC task, in which participants’ task is to judge the over- first experiment, where participants showed a higher number all mood of a crowd (instead of detecting single emotional of fixations on happy compared to angry attractive faces. targets as in previous studies, see Lindeberg et al., 2019). In the second experiment, we found no such evidence. One Based on the evaluative congruence account, we expected explanation might be the reduced attractiveness difference to see a faster and more accurate identification of attractive between the attractive and unattractive crowds as a result crowds dominated by happy faces and of unattractive crowds of the matching procedure in study 2. Moreover, it is nec- dominated by angry faces. Further extending previous stud- essary to differentiate that fixation durations on the happy ies, we also incorporated eye-tracking to assess potential and angry attractive faces were similar in study 1. In con- mechanisms underlying evaluative congruence. For the eye- trast, happy attractive faces were fixated more numerously tracking indices, we expected higher fixation rates on happy compared to angry attractive faces in the first experiment. 1 3 1834 Psychological Research (2021) 85:1823–1836 This indicates that attentional processes manifest in a biased perceive another person’s attractiveness is strongly influ- selective attention towards happy attractive faces but not in a enced by their own attractiveness (Sim, Saperia, Brown, longer fixation duration for the happy attractive faces. How - & Berinieri, 2015). Therefore, future investigations could ever, compared to the large effects with regard to accuracy, address participants’ attractiveness as an additional mod- response tendencies, and response times, only small effects erator. Especially individuals who perceive themselves as on the eye-tracking variables were found in experiment 1. highly attractive might judge attractive crowds more favora- Therefore, we argue that differences in mood judgements in bly compared to unattractive crowds whereas this might not attractive compared to unattractive crowds are likely only be the case for persons who rate themselves as being less slightly influenced by attentional processes but manifest attractive. more strongly in the evaluation phase. Lastly, it would be interesting to further investigate A further, broader implication of the present studies is whether the strength of the effect of attractiveness on emo- that it could be necessary to control for attractiveness when tion perception is influenced by attractiveness-related ste - investigating the influence of other social cues (e.g. gen- reotypes. It might be possible that the attractiveness effect is der or race) on emotion perception, or even in visual tasks potentiated for participants who hold the attractiveness ste- with emotional targets in general (Lindeberg et al., 2019). reotype to a stronger degree (e.g. participants who evaluate It is possible that past studies have confounded constructs attractive individuals more favorably compared to unattrac- such as target gender and attractiveness, and therefore the tive ones and associate attractiveness with more beneficial selection of the stimulus material may have exaggerated or outcomes). Moreover, it would also be interesting to add underestimated effects of social cues on emotion perception an additional social cue to the current experimental design (Lindeberg et al., 2019). Hence, controlling for attractive- and systematically investigate the interplay between attrac- ness when selecting targets appears to be highly important. tiveness and this social cue. For example, further varying— While knowledge about the far-reaching effects of other instead of experimentally controlling for— target gender traits such as target gender or race are relatively well known in the experiment (by showing only female and only male and well addressed in the field, the same does currently not crowds) would enable to test the interaction between target apply to attractiveness. Our findings thus also imply a need attractiveness and target gender on emotion perception in for existing and developing stimulus databases to obtain the mood-of-the-crowd paradigm. The attractiveness effect attractiveness ratings to allow researchers to account for this could be even stronger in female compared to male crowds factor in their studies. as females are perceived more favorably compared to males (Eagly, Mladinic, & Otto, 1991a, b). However, this would Limitations require face databases with a larger number of male and female faces and with a greater variance of attractiveness to The present experiments entailed several limitations. First, build the respective crowds. we found that attractive happy and unattractive angry faces were rated as more emotionally intense compared to attrac- tive angry and unattractive happy faces in both experiments. Conclusion As we used different target faces in the two studies, we believe it to be unlikely that happiness was expressed more In two experiments, we demonstrated that attractiveness strongly in attractive faces and anger more strongly in unat- affects emotion perception in a visual search paradigm tractive face expressions. To us, it seemed more plausible with multiple emotional targets, the MoC paradigm. Spe- that influences of attractiveness that were observed in the cifically, participants evaluated attractive crowds containing mood judgements in the MoC task also transfer to the emo- more happy expressions faster and more accurately, and the tional intensity ratings of the faces pictures. Future studies same was true for unattractive crowds dominated by angry should aim to incorporate material in which attractive and expressions, which corroborates the evaluative congruence unattractive faces express respective emotions to the same account. Moreover, attractive crowds were judged as being degree. One possibility would be to use computer-gener- happy more often whereas unattractive crowds were per- ated faces that are closely matched in terms of emotionality, ceived as being angry more frequently. Additionally, eye- gender and attractiveness. Another option would be to use tracking analyses revealed that there is also a small effect machine learning approaches to select attractive and unat- of attractiveness on gaze movements, though this was pre- tractive faces that show the same emotional intensity. sent only in experiment 1. Specifically, we observed higher A second limitation is that we did not investigate partici- fixation rates on happy compared to angry attractive targets, pants’ own attractiveness, which may also play a role when implying that attractiveness plays a role even in the early judging the mood of attractive and unattractive crowds. Pre- stages of perception. These results imply that face attractive- vious studies suggest that the degree to which individuals ness should be carefully considered when selecting material 1 3 Psychological Research (2021) 85:1823–1836 1835 Bucher, A., Voss, A., Spaniol, J., Hische, A., & Sauer, N. (2019). Age for future studies that examine emotion perception, and stim- differences in emotion perception in a multiple target setting: an ulus attractiveness should be experimentally controlled or eye-tracking study. Emotion. https ://doi.or g/10.1037/emo00 00645 . statistically adjusted for. Calvo, M. G., & Marrero, H. (2009). Visual search of emotional faces: the role of affective content and featural distinctiveness. Cogni- tion and Emotion, 23(4), 782–806. https://doi.or g/10.1080/02699 93080 21516 54. Funding Open access funding enabled and organized by Projekt Craig, B. M., & Lipp, O. V. (2017). The influence of facial sex cues on DEAL. This research received no specific grant from any funding emotional expression categorization is not fixed. Emotion, 17(1), agency in the public, commercial or not-for-profit sectors. 28–39. https ://doi.org/10.1037/emo00 00208 . Craig, B. M., Mallan, K. M., & Lipp, O. V. (2012). The effect of poser Availability of data and material Data from the experiments can be race on the happy categorization advantage depends on stimulus downloaded from https://heida t a.uni-heidelber g.de/dat ase t.xhtml?persi type, set size, and presentation duration. Emotion, 12(6), 1303– stent Id=doi:10.11588 /data/LYWYG N 1314. https ://doi.org/10.1037/a0028 622. Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful Compliance with ethical standards is good. Journal of Personality and Social Psychology, 24(3), 285–290. https ://doi.org/10.1037/h0033 731. Eagly, A. H., Ashmore, R. D., Makhijani, M. G., & Longo, L. C. Conflict of interest The authors declare no conflict of interest. (1991a). What is beautiful is good, but…: a meta-analytic review of research on the physical attractiveness stereo- Open Access This article is licensed under a Creative Commons Attri- type. Psychological Bulletin, 110(1), 109–128. https ://doi. bution 4.0 International License, which permits use, sharing, adapta- org/10.1037/0033-2909.110.1.109. tion, distribution and reproduction in any medium or format, as long Eagly, A. H., Mladinic, A., & Otto, S. (1991b). Are women evalu- as you give appropriate credit to the original author(s) and the source, ated more favorably than men? An analysis of attitudes, beliefs, provide a link to the Creative Commons licence, and indicate if changes and emotions. Psychology of Women Quarterly, 15(2), 203–216. were made. The images or other third party material in this article are https ://doi.org/10.1111/j.1471-6402.1991.tb007 92.x. included in the article’s Creative Commons licence, unless indicated Ebner, N. C., & Johnson, M. K. (2010). Age-group differences in otherwise in a credit line to the material. If material is not included in interference from young and older emotional faces. Cognition the article’s Creative Commons licence and your intended use is not and Emotion, 24(7), 1095–1116. https://doi.or g/10.1080/02699 permitted by statutory regulation or exceeds the permitted use, you will 93090 31283 95. need to obtain permission directly from the copyright holder. To view a Elias, E., Dyer, M., & Sweeny, T. D. (2017). Ensemble perception copy of this licence, visit http://creativ ecommons .or g/licenses/b y/4.0/. of dynamic emotional groups. Psychological Science, 28(2), 193–203. https ://doi.org/10.1177/09567 97616 67818 8. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). GPower 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Meth- References ods, 39(2), 175–191. https ://doi.org/10.3758/BF031 93146 . Golle, J., Mast, F. W., & Lobmaier, J. S. (2014). Something to smile Becker, D. V., Anderson, U. S., Mortensen, C. R., Neufeld, S. L., & about: the interrelationship between attractiveness and emo- Neel, R. (2011). The face in the crowd effect unconfounded: tional expression. Cognition and Emotion, 28(2), 298–310. https Happy faces, not angry faces, are more efficiently detected in sin- ://doi.org/10.1080/02699 931.2013.81738 3. gle- and multiple-target visual search tasks. Journal of Experi- Hansen, C. H., & Hansen, R. D. (1988). Finding the face in the mental Psychology: General, 140(4), 637–659. https ://doi. crowd: an anger superiority effect. Journal of Personal- org/10.1037/a0024 060.supp. (Supplemental). ity and Social Psychology, 54(6), 917–924. https ://doi. Becker, D. V., Kenrick, D. T., Neuberg, S. L., Blackwell, K. C., & org/10.1037/0022-3514.54.6.917. Smith, D. M. (2007). The confounded nature of angry men and Hugenberg, K. (2005). Social categorization and the perception of happy women. Journal of Personality and Social Psychology, facial affect: target race moderates the response latency advan- 92(2), 179–190. https ://doi.org/10.1037/0022-3514.92.2.179. tage for happy faces. Emotion, 5(3), 267–276. https ://doi. Becker, D. V., Neel, R., Srinivasan, N., Neufeld, S., Kumar, D., & org/10.1037/1528-3542.5.3.267. Fouse, S. (2012). The vividness of happiness in dynamic facial Hugenberg, K., & Sczesny, S. (2006). On wonderful women and displays of emotion. PLoS ONE. https ://doi.org/10.1371/journ seeing smiles: social categorization moderates the happy face al.pone.00265 51. response latency advantage. Social Cognition, 24(5), 516–539. Bijlstra, G., Holland, R. W., & Wigboldus, D. H. J. (2010). The social https ://doi.org/10.1521/soco.2006.24.5.516. face of emotion recognition: evaluations versus stereotypes. Jour- Jackson, L. A., Hunter, J. E., & Hodge, C. N. (1995). Physical attrac- nal of Experimental Social Psychology, 46(4), 657–663. https :// tiveness and intellectual competence: a meta-analytic review. doi.org/10.1016/j.jesp.2010.03.006. Social Psychology Quarterly, 58(2), 108–122. https ://doi. Borkenau, P., & Liebler, A. (1992). Trait inferences: Sources of validity org/10.2307/27871 49. at zero acquaintance. Journal of Personality and Social Psychol- Juth, P., Lundqvist, D., Karlsson, A., & Öhman, A. (2005). Looking ogy, 62(4), 645–657. https://doi.or g/10.1037/0022-3514.62.4.645. for foes and friends: perceptual and emotional factors when Borkenau, P., & Liebler, A. (1995). Observable attributes as manifes- finding a face in the crowd. Emotion, 5(4), 379–395. https://doi. tations and cues of personality and intelligence. Journal of Per- org/10.1037/1528-3542.5.4.379. sonality, 63(1), 1–25. https ://doi.org/10.1111/j.1467-6494.1995. Langlois, J. H., Kalakanis, L., Rubenstein, A. J., Larson, A., Hal- tb007 99.x. lam, M., & Smoot, M. (2000). Maxims or myths of beauty? A Bucher, A., & Voss, A. (2019). Judging the mood of the crowd: atten- meta-analytic and theoretical review. Psychological Bulletin, tion is focused on happy faces. Emotion, 19(6), 1044–1059. https 126(3), 390–423. https://doi.or g/10.1037/0033-2909.126.3.390. ://doi.org/10.1037/emo00 00507 . 1 3 1836 Psychological Research (2021) 85:1823–1836 Leppänen, J. M., & Hietanen, J. K. (2004). Positive facial expres- learning. Psychological Review, 108(3), 483–522. https ://doi. sions are recognized faster than negative facial expressions, but org/10.1037/0033-295X.108.3.483. why? Psychological Research Psychologische Forschung, 69, Pinkham, A. E., Griffin, M., Baron, R., Sasson, N. J., & Gur, R. C. 22–29. https ://doi.org/10.1007/s0042 6-003-0157-2. (2010). The face in the crowd effect: Anger superiority when Leppänen, J. M., Tenhunen, M., & Hietanen, J. K. (2003). Faster using real faces and multiple identities. Emotion, 10(1), 141–146. choice-reaction times to positive than to negative facial Savage, R. A., Lipp, O. V., Craig, B. M., Becker, S. I., & Horstmann, expressions: the role of cognitive and motor processes. G. (2013). In search of the emotional face: anger versus happiness Journal of Psychophysiology, 17(3), 113–123. https ://doi. superiority in visual search. Emotion, 13(4), 758–768. https://doi. org/10.1027//0269-8803.17.3.113.org/10.1037/a0031 970. Lindeberg, S., Craig, B. M., & Lipp, O. V. (2019). You look pretty Scherer, K. R., & Wallbott, H. G. (1994). Evidence for universality happy: attractiveness moderates emotion perception. Emotion, and cultural variation of differential emotion response patterning. 19(6), 1070–1080. https ://doi.org/10.1037/emo00 00513 .supp. Journal of Personality and Social Psychology, 66(2), 310–328. (Supplemental).https ://doi.org/10.1037/0022-3514.66.2.310. Ma, D. S., Correll, J., & Wittenbrink, B. (2015). The Chicago Face Sim, S. Y.-L., Saperia, J., Brown, J. A., & Bernieri, F. J. (2015). Judg- Database: a free stimulus set of faces and norming data. Behavior ing attractiveness: biases due to raters’ own attractiveness and Research Methods, 47(4), 1122–1135. intelligence. Cogent Psychology, 2(1). Mueser, K. T., Grau, B. W., Sussman, S., & Rosen, A. J. (1984). You’re Smits, G. J., & Cherhoniak, I. M. (1976). Physical attractiveness and only as pretty as you feel: facial expression as a determinant of friendliness in interpersonal attraction. Psychological Reports, physical attractiveness. Journal of Personality and Social Psychol- 39(1), 171–174. https ://doi.org/10.2466/pr0.1976.39.1.171. ogy, 46(2), 469–478. https://doi.or g/10.1037/0022-3514.46.2.469. Watkins, L. M., & Johnston, L. (2000). Screening job applicants: the Öhman, A., Juth, P., & Lundqvist, D. (2010). Finding the face in a impact of physical attractiveness and application quality. Interna- crowd: relationships between distractor redundancy, target emo- tional Journal of Selection and Assessment, 8(2), 76–84. https :// tion, and target gender. Cognition and Emotion, 24(7), 1216– doi.org/10.1111/1468-2389.00135 . 1228. https ://doi.org/10.1080/02699 93090 31668 82. Öhman, A., Lundqvist, D., & Esteves, F. (2001). The face in the crowd Publisher’s Note Springer Nature remains neutral with regard to revisited: a threat advantage with schematic stimuli. Journal of jurisdictional claims in published maps and institutional affiliations. Personality and Social Psychology, 80(3), 381–396. https:/ /doi. org/10.1037/0022-3514.80.3.381. Öhman, A., & Mineka, S. (2001). Fears, phobias, and pre- paredness: toward an evolved module of fear and fear 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Psychological Research Springer Journals

Pretty crowds are happy crowds: the influence of attractiveness on mood perception

Download PDF
14 pages

Loading next page...
 
/lp/springer-journals/pretty-crowds-are-happy-crowds-the-influence-of-attractiveness-on-mood-tnbJ1Wihmg

References (42)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2020
ISSN
0340-0727
eISSN
1430-2772
DOI
10.1007/s00426-020-01360-x
Publisher site
See Article on Publisher Site

Abstract

Empirical findings predominantly support a happiness superiority effect in visual search and emotion categorization para- digms and reveal that social cues, like sex and race, moderate this advantage. A more recent study showed that the facial attribute attractiveness also influences the accuracy and speed of emotion perception. In the current study, we investigated whether the influence of attractiveness on emotion perception translates into a more general evaluation of moods when more than one emotional target is presented. In two experiments, we used the mood-of-the-crowd (MoC) task to investigate whether attractive crowds are perceived more positively compared to less attractive crowds. The task was to decide whether an array of faces included more angry or more happy faces. Furthermore, we recorded gaze movements to test the assumption that fixations on happy expressions occur more often in attractive crowds. Thirty-four participants took part in experiment 1 as well as in experiment 2. In both experiments, crowds presenting attractive faces were judged as being happy more frequently whereas the reverse pattern was found for unattractive crowds of faces. Moreover, participants were faster and more accurate when evaluating attractive crowds containing more happy faces as well as when judging unattractive crowds composed of more angry expressions. Additionally, in experiment 1, there were more fixations on happy compared to angry expressions in attractive crowds. Overall, the present findings support the assumption that attractiveness moderates emotion perception. Introduction postulated for angry faces (Hansen & Hansen, 1988; Öhman, Lundqvist, & Esteves, 2001; Pinkham, Griffin, Baron, Sas- The fast and correct identification of emotional expressions son, & Gur, 2010), but recent studies have found compelling in human faces is essential for social interactions, because evidence for an advantage for happy faces, termed “hap- facial expressions signal a person’s potential intentions and piness superiority effect” (Becker, Anderson, Mortensen, behavior (Calvo & Marrero, 2009). While a smiling face sig- Neufeld, & Neel, 2011; Savage, Lipp, Craig, Becker, & nals affiliation and approachability, an angry face signals a Horstmann, 2013). These studies used a visual search para- lack of approachability or even threat of aggression (Scherer digm (the “face-in-the-crowd paradigm”, FitC), in which & Wallbott, 1994). Due to the centrality of facial emotion participants have to decide as quickly and accurately as perception for adaptive social interaction, many theorists possible whether an array of faces (all displaying the same have argued that humans can automatically detect signs of expression, e.g. all neutral/happy/angry) contains a deviating affiliation or threats of aggression in faces (e.g., Öhman & target face. In this paradigm, the happiness superiority effect Mineka, 2001). Building on this assumption, some have pos- manifested itself in faster reaction times and lower error tulated that specific emotional expressions have a perception rates for targets with happy expressions compared to angry advantage over others. Initially, a perception advantage was but also other negative expressions (Becker et al., 2011; Sav- age et al., 2013). In emotion categorization tasks, a faster and more correct categorization of happy facial expressions * Alica Mertens was also observed (Leppänen & Hietanen, 2004; Leppänen, alica.mertens@psychologie.uni-heidelberg.de Tenhunen, & Hietanen, 2003). Institute of Psychology, Heidelberg University, Hauptstr. Previous empirical studies suggest that there are several 47-51, 69117 Heidelberg, Germany facial features that can influence the size of the happiness Department of Psychosomatic Medicine, Central Institute superiority effect, including sex and race (e.g., Becker, Ken- of Mental Health, Medical Faculty Mannheim/Heidelberg rick, Neuberg, Blackwell, & Smith, 2007; Craig & Lipp, University, Mannheim, Germany Vol.:(0123456789) 1 3 1824 Psychological Research (2021) 85:1823–1836 2017; Bijlstra, Holland, & Wigboldus, 2010). Specifically, this study was too easy and incurred very low error rates that the happiness superiority effect was larger for female tar - might impede the detection of response biases. At the same gets (Becker et al., 2007; Bucher & Voss, 2019; Bucher, time, the absence of a detection advantage for unattractive Voss, Spaniol, Hische, & Sauer, 2019; Craig & Lipp, 2017; angry faces can also be seen as evidence against evaluative Hugenberg & Sczesny, 2006) and for male own-race targets congruence. The present study aimed to replicate findings compared to male other-race targets (Bijlstra et al., 2010; by Lindeberg and colleagues in a novel paradigm that over- Craig, Mallan, & Lipp, 2012). In addition to this research, a comes some of the limitations of single target paradigms recent study also found that facial attractiveness moderates and includes eye-tracking analyses to assess underlying pro- the happiness superiority effect (Lindeberg, Craig, & Lipp, cesses. This way, we hoped to shed light on whether unat- 2019). In four experiments, the authors found evidence for tractive-angry faces indeed do not entail a detection advan- a faster and more accurate identification of happiness (ver - tage (as suggested by the previous evidence) and which sus anger) in attractive faces compared to unattractive faces, would reject the evaluative congruence account, or whether when presenting one target in each trial. this effect would be present in a paradigm that leaves more The evidence of a faster and more accurate identification room for variance in participant performance. of attractive faces displaying happiness can be explained Specifically, the present study relied on a novel paradigm through an interplay of two theoretical accounts: the attrac- that incurs higher cognitive demands and lower accuracy tiveness stereotype and the evaluative congruence account. rates than single-target tasks. The current paradigm might The attractiveness stereotype, also known as the “what is thus leave more room for motives to bias emotion percep- beautiful is good” effect (Dion, Berscheid, & Walster, 1972), tion and enables testing the evaluative congruence account describes the phenomenon that humans believe physical for both attractive and unattractive faces. In the “mood-of- attractiveness to be associated with a diverse range of posi- the-crowd paradigm” (MoC), participants have to judge tive attributes (for a meta-analysis, see Eagly, Ashmore, the overall mood of the crowd by indicating whether more Makhijani, & Longo, 1991a, b). For example, attractiveness angry or happy faces are present in an array of faces (Bucher increases the likelihood of getting a job interview (Watkins & Voss, 2019). Previous studies were able to replicate the & Johnston, 2000), entails higher ratings on favorable per- happiness superiority effect in this setting: happy crowds sonality traits such as Agreeableness (Borkenau & Liebler, were identified faster and more accurately than angry crowds 1992, 1995; Smits & Cherhoniak, 1976), and higher intel- (Bucher & Voss, 2019; Elias, Dyer, & Sweeny, 2017). The ligence ratings (Jackson, Hunter, & Hodge, 1995). Thus, MoC paradigm has a number of advantages over the FitC the attractiveness stereotype suggests that attractiveness and paradigm. First, judging the mood of a crowd represents positive attributes are strongly associated in the minds of a more ecologically valid task than detecting a single face people. To explain, why the pairing of attractive happy faces in a crowd, because crowds with multiple target emotions entails a faster and more accurate processing, one can addi- are highly prevalent in daily life (e.g., in class, at a sport- tionally draw on the evaluative congruence account (Hugen- ing event, etc.). Moreover, the presentation of more than berg, 2005; Hugenberg & Sczesny, 2006). The evaluative one emotional target prevents confusion about whether an congruence account postulates that emotions are processed observed effect results from a target or a crowd effect (see faster when they match the evaluation of a social cue that Bucher & Voss, 2019). As outlined above, the MoC task is they are presented in conjunction with, for instance attrac- also a more complex and cognitively demanding task and tiveness. Based on the attractiveness stereotype account, we therefore does not incur similar floor effects for error rates know that attractiveness is evaluated positively, as is a happy compared to single target paradigms. Furthermore, the MoC affect. Therefore, happiness and attractiveness are both posi- task can be well combined with process tracing methods tive cues and, based on the evaluative congruence account, such as eye-tracking. Eye-tracking is less applicable to the should be processed faster than if the pairing of affect and a FitC paradigm, for example, because the search process ends social cue were incongruent (e.g. angry + attractive). abruptly as soon as the participant detects the target, which However, the attractiveness stereotype also has a flipside: makes it difficult to identify underlying search patterns. Humans also tend to ascribe negative qualities to unattrac- The implementation of eye-tracking analysis is advanta- tive persons (Dion et al., 1972). Based on the evaluative geous for the current investigation of the evaluative congru- congruence account, one would therefore expect that unat- ence account for facial attractiveness because it makes it tractive faces paired with a negative emotion (e.g. anger) possible to pinpoint the locus of the effect of attractiveness are also processed faster and more accurately. However, on emotion perception. If an attractiveness effect was appar - this effect was not observed in the one previous study that ent in eye movements (e.g., an increased probability and investigated the categorization of attractive and unattractive longer duration of fixations on happy-attractive/angry-unat - faces displaying happy and angry affect (Lindeberg et al., tractive faces), this would indicate that congruent stimuli 2019). It is possible that the categorization task applied in (e.g. happy-attractive faces) are preferentially or selectively 1 3 Psychological Research (2021) 85:1823–1836 1825 attended to and thus attentional processes guide the judge- whether the gender-attractiveness imbalance accounts for the ment of whether the crowd is perceived as predominantly reported attractiveness effect on emotion perception. happy or angry. In contrast, if participants showed no differ - We derived our hypotheses based on the evaluative con- ences in their fixation patterns between congruent (happy- gruence account (Hugenberg, 2005; Hugenberg & Sczesny, attractive/ angry-unattractive) and incongruent (happy-unat- 2006) which suggests a processing advantage for emotions tractive/angry-attractive) stimuli, but did show differences at that are evaluated congruently as the social cue they are the choice level, this would suggest effects in the evaluation paired with (herein attractiveness). Therefore, we hypothe- phase (e.g., a biased response towards happiness for attrac- sized that participants would identify attractive crowds more tive crowds). In this case, participants attend to both con- often as happy than unattractive crowds. Reversely, we also gruent and incongruent stimuli in a similar way but the final expected that participants would identify unattractive crowds judgment is biased towards one emotion, indicating that the more often as angry than attractive crowds. Moreover, we evaluation of the selected faces influences the final judgment hypothesized that attractive crowds containing a higher about the general mood of the crowd. Without the inclusion number of happy faces, and unattractive crowds containing of eye-tracking measures, this differentiation is not possible. a higher number of angry expressions, should entail faster and more accurate responses. This effect has previously been supported by Lindeberg et al. (2019), who observed a faster The current study and more accurate categorization of happy attractive com- pared to angry attractive expressions. Further corroborating The aim of the current study was to test the evaluative con- this, there is evidence that participants evaluate happy faces gruence account, thus also aiming for conceptual replication as more attractive than faces displaying negative emotions of the findings by Lindeberg et al. (2019) in a new paradigm. (Mueser, Grau, Sussman, & Rosen, 1984). Moreover, Golle, The present study used the MoC paradigm, in which the par- Mast and Lobmaier (2014) showed that judgment of rela- ticipant’s task is to decide as fast and accurately as possible tively happier compared to neutral faces is facilitated when whether more angry or more happy faces are presented in a those faces were attractive. Based on evaluative congruence, crowd of faces. The merits of this task (compared to single we expected a general tendency to judge attractive crowds target paradigms) include increased cognitive demand (and as being happy and unattractive crowds as being angry more less likelihood of floor effects), increased ecological validity, frequently. With regard to eye-tracking indices, we expected separation of crowd and target effects, and the possibility of higher fixation rates and fixation durations on happy-attrac- including eye-tracking indices in addition to response times tive and angry-unattractive faces than on happy-unattractive and accuracy rates. Note that in other paradigms (e.g., cat- or angry-attractive faces, based on a higher salience of con- egorization task or FitC paradigm) the use of single targets gruently evaluated social cues. prevents investigation of attentional preferences regarding specific emotions or attractiveness-emotion combinations. For example, in the FitC paradigm, the search process stops Experiment 1 abruptly as soon as the target is detected and in single face categorization tasks, there is no possibility to investigate Method search asymmetries. The MoC paradigm, however, allows for analyzing such gaze patterns. An inclusion of eye-track- Participants ing was desirable in this case to test whether perceptual preferences guide the evaluation of the crowd, indicated Prior to recruiting participants, we conducted a power analy- by an increased fixation rate on happy attractive and angry sis using G*Power 3 (Faul, Erdfelder, Lang, & Buchner, unattractive faces, or whether attractiveness influences the 2007) to determine the required sample size. The required speed and accuracy of the judgements itself, indicated by a sample size to detect an effect of medium size with a power lack of an attractiveness effect in gaze movements. Lastly, of 0.80 and an alpha-error of 0.05 in a repeated measures the present study aimed to target a confound that was present ANOVA setting was 34. Thirty-four adults participated in in the studies by Lindeberg et al. (2019), including the fact that attractive female faces were consistently rated as more attractive compared to attractive male faces, whereas there 1 When conducting the power analysis for our studies we used the was no difference for unattractive male and female faces. studies by Lindeberg et  al. (2019) as a basis. They predominantly found medium effect sizes for the interaction effect between emo- Thus, we used the same selection criterion for target faces as tional expression and attractiveness regarding accuracy and catego- Lindeberg et al. (2019) in our first experiment, but matched rization time across their four experiments. Although the task is not target gender and attractiveness in the second experiment. entirely comparable to the MoC task, we perceived this effect size to Using these different selection criteria, we were able to test be the best anchor for our studies. 1 3 1826 Psychological Research (2021) 85:1823–1836 the study (M = 23.06 years, SD = 5.22, range 19–47; 50% the criteria of a minimal duration of 100 ms and maximal age female). We recruited participants from a large participant dispersion of 100 pixels. pool at a German university, in which students of diverse After providing written, informed consent, participants subjects are registered. Before participants took part in the were seated approximately 60 cm in front of the laptop com- study, they received information about the study details and puter. First, participants provided demographic information provided written, informed consent. As compensation for the and then received information that the upcoming task was study, participants could choose between course credit or a to judge whether the crowd contained more happy or angry financial compensation of five Euros. faces. We specifically instructed participants to make their decisions as fast and accurately as possible to prevent them Stimulus material from simply counting the presented faces. The experiment comprised a practice block of 16 trials and two experimental Following Lindeberg et al. (2019), we chose the stimulus blocks of 50 trials each. Prior to each block (practice block material for the first experiment by selecting the most and and experimental blocks) we used a 9-point-calibration for least attractive faces from a pool of faces. We selected the calibrating the eye-tracker. Each calibration was followed by stimulus faces from the Chicago Face Database (Ma, Cor- a 4-point validation. After a successful validation (partici- rell, & Wittenbrink, 2015), restricting the stimulus material pants’ gaze was within 1° visual angle), participants started to faces of Caucasian individuals that provided both angry with the experimental trials. In case of a non-successful vali- and happy (closed mouth) expressions. We used the attrac- dation, we repeated calibration and validation. tiveness ratings of the remaining 37 female and 36 male In each trial, participants saw a crowd of 18 faces faces provided by the norming data of the Chicago Face (each 1.4  cm × 1.9  cm). Faces were randomly allo- Database and selected nine faces that were rated the most cated on the screen in front of a light-grey background attractive and nine faces that achieved the lowest attractive- (34.5 cm × 19.5 cm) to ensure naturalistic eye movements, ness ratings for female and male faces, respectively. This which is not possible when presenting faces in a circle or procedure resulted in 18 female (nine attractive and nine matrix (due to systematic scanning paths: e.g. clockwise unattractive) and 18 male (nine attractive and nine unattrac- or row-wise). To avoid overlap between the pictures, we tive) individuals with happy and angry expressions availa- ensured a minimum distance between the edges of the pic- ble. The attractiveness ratings of the ‘attractive’ female mod- tures (minimum distance was 3.4 cm). Crowds contained els (M = 4.60, SD = 0.32; Models 3, 11, 12, 15, 22, 24, 25, a varying number of angry faces (6, 8, 10 and 12 angry 27 and 29) differed from those of the ‘unattractive’ female faces), and crowds were either completely made up of models (M = 2.32, SD = 0.45; Models 2, 8, 10, 23, 26, 28, 30, attractive or unattractive faces. The different compositions 34 and 37). Likewise, the attractiveness of the ‘attractive’ of angry and happy faces within the crowds (six angry/12 male models (M = 3.89, SD = 0.46; Models 3, 4, 6, 9, 14, 15, happy, eight angry/ten happy, ten angry/eight happy, 12 24, 29 and 33) differed from that of the ‘unattractive’ male angry/six happy) occurred equally often in attractive and models (M = 2.36, SD = 0.16; Models 2, 10, 17, 19, 35, 37, unattractive crowds to prevent any systematic confound. 38, 39, 41). Lighting and visual contrast were similar across Moreover, crowds always contained the same number of all faces in the set. female and male faces (e.g., in case of ten angry and eight happy faces the crowd contained five female angry and five Procedure male angry expressions as well as four female happy and four male happy expressions). All presented face pictures The experiment was run on a Dell laptop computer. Stimulus within one trial were from different individuals. displays and response time measurement were controlled by Each block started with two warm-up trials. The eight a C program using SDL libraries (www.libsdl.or g). We used trial combinations (attractiveness of presented faces × num- a SMI RED250MOBILE eye-tracker to record gaze move- ber of angry pictures) were presented six times in random ments at a frequency of 250 Hz. We defined fixations using order in each experimental block. Prior to each trial, a We chose the composition of the number of happy/angry faces The norming data of the Chicago Face Database includes mean based on the last experiment by Bucher and Voss (2019) which also attractiveness ratings of 1087 participants (Ma et  al., 2015) for each included four trial types (7, 9, 11, 13 angry expressions in crowds of face picture. Because only the averaged ratings of each picture are 20 faces). However, because we wanted to balance target gender in available, the selection is based descriptively on these mean attrac- our crowds of faces, we needed to build crowds of even numbers of tiveness ratings. Therefore, we report the mean and standard devia- angry and happy expressions. Therefore, we changed the total number tion averaged for the selected pictures of the respective types (female of pictures from 20 to 18 and changed the number of presented angry attractive, female unattractive, male attractive, male unattractive). faces to 6, 8, 10 and 12. 1 3 Psychological Research (2021) 85:1823–1836 1827 fixation cross was presented at the center of the screen before attractive, happy and female targets were rated as being more the crowd was shown. Only when participants fixated on this attractive compared to the unattractive, angry and male tar- cross, the crowd appeared. If participants did not focus on gets. Moreover, we found a significant interaction between the fixation cross until 5000 ms were exceeded, a recalibra- attractiveness and target gender, F(1, 32) = 97.83, p < 0.001, tion started. The crowd remained visible until participants η = 0.75, 95% CI [0.61, 0.82]. Follow-up pairwise com- gave their response. If participants’ responses were slower parisons revealed that for attractive faces, female targets than 8000 ms, a message (“please try to respond faster”) were perceived as more attractive compared to male targets, appeared. Participants could press the S-key and the K-key t(33) = 9.06, p < 0.001, d = 1.55, whereas the unattractive to indicate whether they perceived the crowd as predomi- male and female targets did not differ significantly, t < 1. nantly happy or angry. The assignment of the emotions to We ran the same analysis for the emotional intensity rat- the keys was balanced across participants. As a reminder, the ings. There was a significant main effect of attractiveness, words “happy” and “angry” appeared on the respective sides F(1, 32) = 32.24, p < 0.001, η = 0.50, 95% CI [0.28, 0.63], on the bottom of the screen. Subsequent to the eye-tracking and gender, F(1, 32) = 40.48, p < 0.001, η = 0.56, 95% CI experiment, participants rated the happy and angry faces on [0.34, 0.68], showing that attractive and female targets were attractiveness, happiness, and anger on a 7-point Likert scale perceived as showing a higher emotional intensity. The main in a randomized sequence. effects were qualified by an attractiveness × target gender interaction, F(1, 32) = 15.89, p < 0.001, η = 0.33, 95% CI Analysis [0.12, 0.50] and an attractiveness × emotional expression interaction, F(1, 32) = 71.26, p < 0.001, η = 0.69, 95% CI A 2 (attractiveness: attractive, unattractive) × 2 (trial type: [0.52, 0.77]. Follow-up pairwise comparisons showed that mainly happy, mainly angry ) repeated measures ANOVA female expressions were rated as more emotionally inten- was conducted to analyze responses, accuracy rates and sive compared to male faces both for attractive, t(33) = 2.37, response times. For the analyses of the number of fixations p = 0.024, d = 0.41, and unattractive targets, t(33) = 7.45, and fixation duration on the respective faces, the emotion of p < 0.001, d = 1.28. However, this difference was much more the focused picture (happy vs. angry) served as an additional pronounced for unattractive faces, t(33) = 3.96, p < 0.001, within-subjects factor. This way, we assessed how many d = 0.68. Additionally, pairwise comparisons revealed that faces of each type received at least one fixation during a trial happiness was expressed more strongly on attractive com- and how long each type of face was fixated. We excluded pared to unattractive faces, t(33) = 8.37, p < 0.001, d = 1.44, responses faster than 500 ms and slower than 8000 ms from whereas there was no significant difference for angry faces, the response time analysis (0.51% of all trials). Addition- t(33) = − 1.90, p = 0.066, d = 0.33. ally, we adjusted for participants’ gender as a covariate in Mean and standard deviations of the attractiveness and all analyses. If not reported in the text, gender did not show emotional intensity ratings are summarized in Table 1. any significant main effects or interactions. Response time Results We computed mean correct response times for each par- ticipant and each factorial combination. There was a sig- Manipulation check nificant interaction between attractiveness and trial type, F(1, 32) = 6.18, p = 0.018, η = 0.16, 95% CI [0.02, 0.34]. Attractiveness ratings were obtained in a 2 (emotional Follow-up pairwise comparisons revealed that participants expression: happy, angry) × 2 (attractiveness: attractive, were faster when more happy faces were presented in attrac- unattractive) × 2 (target gender: female, male) repeated tive crowds compared to unattractive crowds, t(33) = 2.97, measures ANOVA. There was a main effect of attractive- p = 0.006, d = 0.51, whereas there was no difference when ness, F(1, 32) = 472.53, p < 0.001, η = 0.94, 95% CI [0.89, more angry faces were shown, t < 1 (Fig. 1a). 0.95], emotional expression, F(1, 32) = 15.53, p < 0.001, η = 0.33, 95% CI [0.11, 0.49], and gender, F(1, 32) = 46.78, Response p < 0.001, η = 0.59, 95% CI [0.39, 0.70], confirming that Mean responses (proportion of “angry” responses) were cal- culated for each participant and each factorial combination. To ease interpretation of the results, trial types containing more A main effect of attractiveness, F (1, 32) = 32.53, p < 0.001, happy faces (6 and 8 angry out of 18 faces) and those who consisted 2 η = 0.50, 95% CI [0.28, 0.63], and a main effect of trial of more angry faces (10 and 12 angry out of 18 faces) were com- type emerged, F(1, 32) = 315.00, p < 0.001, η = 0.91, 95% bined. When analyzing the data using the original four categories of CI [0.85, 0.93]. Attractive crowds were judged as “happy” trial type, findings were largely identical. 1 3 1828 Psychological Research (2021) 85:1823–1836 Table 1 Attractiveness and emotional intensity ratings for happy and angry female and male faces from both experiments Measures Female Male Happy Angry Happy Angry Experiment 1  Attractiveness   Attractive 5.63 (0.69) 5.06 (0.86) 3.99 (1.23) 3.52 (1.04)   Unattractive 2.76 (1.07) 2.13 (0.84) 2.66 (1.13) 2.11 (1.04)  Emotional intensity   Attractive 5.59 (0.73) 5.41 (0.73) 5.36 (0.55) 5.35 (0.63)   Unattractive 5.20 (0.66) 5.69 (0.64) 4.76 (0.73) 5.23 (0.71) Experiment 2  Attractiveness   Attractive 4.55 (0.76) 3.70 (0.96) 4.17 (0.94) 3.52 (1.10)   Unattractive 3.40 (0.84) 2.51 (0.82) 3.13 (0.84) 2.36 (0.75)  Emotional intensity   Attractive 5.18 (0.69) 5.19 (0.74) 5.11 (0.71) 4.97 (0.88)   Unattractive 4.92 (0.69) 5.42 (0.80) 5.08 (0.71) 5.23 (0.75) Values in parantheses represent 1 SD and unattractive ones as “angry” more frequently (Fig. 2a). Participants evaluated crowds containing more happy faces as being happy more often compared to those with more angry faces. Accuracy Mean accuracies were calculated for each participant and each factorial combination. There was a main effect of trial type, F(1, 32) = 4.23, p = 0.048, η = 0.12, 95% CI [0.00, Fig. 1 Interaction effect of dominant emotion and attractiveness on 0.29], indicating slightly increased accuracy rates in trials response times (in ms) for experiment 1 (a) and experiment 2 (b). with more angry faces. This main effect was qualified by Trials with ten and 12 happy expressions were combined as well as an attractiveness × trial type interaction, F(1, 32) = 32.53, trials with ten and 12 angry expressions. Error bars indicate standard p < 0.001, η = 0.50, 95% CI [0.28, 0.63]. Follow-up t errors tests indicated that when more happy faces appeared in a crowd, participants showed higher accuracy rates for attractive crowds, t(33) = 6.31, p < 0.001, d = 1.08, and the a main effect of emotional expression, F(1, 32) = 8.30, reversed pattern when confronted with unattractive faces, p = 0.007, η = 0.21, 95% CI [0.04, 0.38], and attractive- t(33) = − 3.34, p = 0.002, d = 0.57 (Fig. 3a). ness, F(1, 32) = 10.20, p = 0.003, η = 0.24, 95% CI [0.05, 0.42], indicating that happy faces and unattractive faces were Fixation duration fixated more frequently compared to angry and attractive faces. Moreover, we found a significant interaction between Mean durations were calculated for each participant and attractiveness and emotional expression, F(1, 32) = 4.65, each factorial combination. There was no significant main or p = 0.039, η = 0.13, 95% CI [0.00, 0.30]. Follow-up pair- interaction effect of any of the investigated factors, F < 2.10, wise comparisons revealed that in unattractive crowds happy p > 0.16, η < 0.06. and angry expressions were fixated to the same degree, t < 1, whereas in attractive crowds significantly more happy faces Number of fixations were fixated, t(33) = 3.85, p = 0.001, d = 0.66 (Fig. 4). There was a significant interaction between emotional expression The number of fixations assesses how many faces of each and trial type, F(1, 32) = 301.35, p < 0.001, η = 0.90, 95% type received at least one fixation during a trial. There was CI [0.84, 0.39], with more fixations on angry (happy) faces 1 3 Psychological Research (2021) 85:1823–1836 1829 Fig. 2 Main effect of attractiveness on response tendency (0 = happy, 1 = angry) for experiment 1 (a) and experiment 2 (b). Error bars indi- Fig. 3 Interaction effect of dominant emotion and attractiveness on cate standard errors accuracy (0 = false, 1 = correct) for experiment 1 (a) and experiment 2 (b). Trials with ten and 12 happy expressions were combined as well as trials with ten and 12 angry expressions. Error bars indicate stand- ard errors when more angry (happy) faces were presented. Lastly, an interaction between all factors emerged, F(1, 32) = 4.55, 2 6 p = 0.041, η = 0.13, 95% CI [0.00, 0.30]. Discussion In experiment 1, we assessed whether stimulus attractiveness When conducting the post-hoc analyses to shed light on this inter- has an effect on the happiness superiority effect in visual action between all factors, no meaningful interpretation of this effect search, specifically in the MoC paradigm. We hypothesized could be revealed. Thus, this interaction effect does not contribute to a faster and more accurate perception of attractive crowds the other findings from our analyses. containing more happy faces and of unattractive crowds We re-ran all analyses and included the difference scores of the emotional intensity ratings for happy attractive and happy unattrac- comprising more angry faces. We expected a response bias tive expressions (happy attractive minus happy unattractive) as well towards happiness in attractive crowds and towards anger as for angry attractive and angry unattractive expressions (angry in unattractive crowds. Results revealed that participants attractive minus angry unattractive) to account for this influence on judged crowds containing attractive faces as happy more our reported findings. However, including these difference scores did not alter the effects. There was still a significant effect of attractive- ness on response tendency, F(1, 30) = 32.25, p < 0.001, η = 0.52, Footnote 6 (continued) 95% CI [0.29, 0.65], a significant interaction between attractiveness and dominant emotion on response time, F(1, 30) = 5.95, p = 0.021, p < 0.001, η = 0.52, 95% CI [0.29, 0.65], and a significant interac- η = 17, 95% CI [0.01, 0.35], a significant interaction between tion between attractiveness and target emotion on number of fixa- attractiveness and dominant emotion on accuracy, F(1, 30) = 32.25, tions, F(1, 30) = 4.38, p = 0.045, η = 0.13, 95% CI [0.00, 0.31]. 1 3 1830 Psychological Research (2021) 85:1823–1836 experiment and chose different selection criteria for the stimulus material. Experiment 2 In experiment 1, the focus when choosing the stimulus material was to achieve the greatest possible attractiveness difference between unattractive and attractive targets. This approach leads to a larger difference in perceived attractive- ness and maximizes the chances of finding an attractiveness effect. Lindeberg and colleagues (2019) critically discussed this selection procedure and demonstrated that it leads to a greater difference in rated attractiveness for attractive female compared to attractive male targets. As target gender also influences emotion perception to a large degree, it might be Fig. 4 Interaction effect of target emotion and attractiveness on num- the case that results from experiments using this selection ber of fixations. Error bars indicate standard errors method are somewhat confounded. Therefore, in experi- ment 2, we used different selection criteria for the mate- often than crowds containing unattractive faces. Moreover, rial to ensure that attractive female and male targets as well participants were faster and more accurate when crowds as unattractive male and female targets match with regard were both attractive and containing many happy faces. In to their perceived attractiveness. Furthermore, we used the line with the evaluative congruence account, we also found emotion intensity ratings for the neutral faces provided in the that participants judged unattractive crowds as angry more norming data of the Chicago Face Database to better match frequently, and that judgments were more accurate in trials the attractive and unattractive face pictures. Despite these with more angry targets presented. This is a novel finding, as modifications, the experimental setting and hypotheses in the previous study testing this model (Lindeberg et al., 2019) experiment 2 were identical to experiment 1. only found support for part of the evaluative congruence account, showing an advantage for attractive-happy faces but not for unattractive-angry targets. Further extending Methods previous findings, we included eye-tracking analyses that revealed that fixations occurred more frequently on happy Participants compared to angry facial expressions when attractive faces were presented. Based on the same power analysis as in experiment 1, we The results supported our hypotheses and showed recruited 34 adults (M = 23.26  years, SD = 5.66, range age medium to large effect sizes, but were limited by an imbal- 18–46; 53% female) via a large participant pool of a Ger- ance between target gender and attractiveness. Thus, we man university. Before starting with the experiment, par- cannot completely rule out the possibility that this gender- ticipants received information about the upcoming task and attractiveness disparity affected our results. Mirroring the provided written, informed consent. Again, participants findings by Lindeberg et al. (2019), the difference between could receive course credit or five Euros as compensation attractiveness ratings was also larger for attractive female for their participation. compared to attractive male faces. Furthermore, we found that participants rated happy attractive faces as more emo- Stimulus material tionally intense than happy unattractive faces. Although we believe that attractiveness influenced the emotion rat- We again selected the stimulus material from the Chicago ings to the same degree as accuracy, response tendency and Face Database (Ma et al., 2015), considering only pictures response times in the MoC task, other explanations might be possible as well. It is imaginable that the completion Although the emotion ratings for the neutral expressions do not of the MoC task influenced the following ratings, because necessarily match the ratings of the actual emotions, it nevertheless we observed the same pattern of results for the experiment provides a tendency that might influence the intensity of the emo- and the ratings. Moreover, it is possible that unattractive tional expressions. targets showed less intense happy expressions than attrac- We used the same power analysis criteria for the second experi- tive targets. To address these issues, we conducted a second ment, to allow comparability between the two experiments. 1 3 Psychological Research (2021) 85:1823–1836 1831 of Caucasian men and women that provided both happy and Analysis angry expressions. In addition to the attractiveness ratings, happiness and anger ratings were taken into account when The analytic strategy was identical to that in experiment 1. choosing the stimulus material. Happiness and anger ratings Again, we excluded responses faster than 500 ms and slower of the neutral face pictures of the unattractive and attractive than 8000 ms from the response time analysis (0.43% of all male and female pictures were matched to control for pos- trials). sible emotional intensity differences between the pictures. Furthermore, we matched target gender and perceived attrac- tiveness, so that attractive males and females as well as unat- Results tractive males and females had similar attractiveness rating. We again selected 18 female (nine attractive and nine unat- Manipulation check tractive) and 18 male (nine attractive and nine unattractive) individuals. When comparing the attractiveness ratings, the Ratings from the 45 participants of the pretest were used attractive female (M = 3.95, SD = 0.38; M = 2.47, to analyze perceived attractiveness and emotional intensity. att att happy SD = 0.49; M = 2.60, SD = 0.65; Models 6, 11, There were main effects of attractiveness, F (1, 43) = 169.28, happy angry angry 13, 15, 16, 18, 21, 25 and 31) and unattractive female mod- p < 0.001, η = 0.80, 95% CI [0.70, 0.85], target emotion, els (M = 2.79, SD = 0.17; M = 2.41, SD = 0.54; F(1, 43) = 40.79, p < 0.001, η = 0.49, 95% CI [0.30, 0.61], att att happy happy p M = 2.46, SD = 0.76; Models 5, 7, 8, 19, 23, 28, and target gender, F(1, 43) = 14.15, p = 0.001, η = 0.25, angry angry 30, 36 and 37) as well as the attractive male (M = 3.89, 95% CI [0.08, 0.40], indicating that attractive, happy and att SD = 0.46; M = 2.63, SD = 0.32; M = 2.21, female faces were rated as more attractive compared to att happy happy angry SD = 0.39; Models 3, 4, 6, 9, 14, 15, 24, 29 and 33) unattractive, angry and male faces. This time, there was angry and unattractive male models (M = 2.71, SD = 0.13; no significant interaction between attractiveness and target att att M = 2.48, SD = 0.67; M = 2.39, SD = 0.50; gender, F(1, 43) = 0.63, p = 0.432, η = 0.01, 95% CI [0.00, happy happy angry angry p Models 12, 13, 20, 21, 23, 25, 32, 34 and 37) differed in 0.12]. Lastly, we found a significant interaction between tar - rated attractiveness, respectively, but not with regard to hap- get emotion and target gender, F(1, 43) = 9.56, p = 0.003, piness or anger. Lighting and visual contrast were similar η = 0.18, 95% CI [0.04, 0.34]. Follow-up pairwise com- across all faces in the set. parisons revealed that happy faces were rated more attrac- As the attractiveness, happiness and anger ratings were tively compared to angry faces both for female, t(44) = 8.25, only available for the neutral face expressions, the happy p < 0.001, d = 1.23, and male faces, t(44) = 6.74, p < 0.001, and angry expressions of the attractive and unattractive male d = 1.00, however, this difference was significantly larger for as well as female individuals were additionally rated by 45 female faces, t(44) = 2.76, p = 0.008, d = 0.41. participants (M = 34.98 years, SD = 13.60, range 19–59; With regard to emotional intensity, there was a signifi- age 73% female). We present the results for the attractiveness cant main effect of target gender, F (1, 43) = 13.48, p = 0.001, and emotional intensity ratings in the results section. η = 0.24, 95% CI [0.07, 0.39], and participants’ gender, F(1, 43) = 9.64, p = 0.003, η = 0.18, 95% CI [0.02, 0.37], Procedure indicating that female faces were perceived as more emo- tionally intense and that female raters gave higher inten- The experimental setup and procedure were identical to sity ratings. The main effect of target gender was qualified those in experiment 1. The only difference was with regard by an interaction between target gender and target emo- to the stimulus material and that the emotional face expres- tion, F(1, 43) = 8.13, p = 0.007, η = 0.16, 95% CI [0.03, sions were rated prior to the study regarding attractiveness 0.32]. Whereas happy and angry faces were perceived as and emotional intensity (happiness, anger) by an independ- equally intense in male faces, t < 1, angry female faces ent sample to prevent transfer effects. were rated more intensely compared to happy female faces, t(44) = − 2.65, p = 0.011, d = 0.40. Furthermore, there was a signic fi ant interaction between attractiveness and target emo - tion, F(1, 43) = 34.89, p < 0.001, η = 0.45, 95% CI [0.26, There were several reasons why we decided to recruit a separate 0.58]. Happy attractive faces were rated more intensely than sample to rate the attractive and unattractive emotional face expres- happy unattractive faces, t(44) = 3.76, p < 0.001, d = 0.56, sions. In the first experiment, ratings were collected after the MoC experiment, so it might be possible that the completion of the experi- whereas the reverse pattern was found for angry expres- mental task influenced the attractiveness and emotion ratings after - sions, t(44) = − 4.94, p < 0.001, d = 0.74. Lastly, a signifi- wards. Furthermore, when measuring the ratings prior to the experi- cant interaction between participants’ gender and target gen- ment, it might happen that the ratings influence the completion of the der emerged, F(1, 43) = 11.34, p = 0.001, η = 0.21, 95% experimental task afterwards which is also not ideal. Collecting the p ratings in a “pretest” therefore seemed appropriate. CI [0.05, 0.37]. Male participants rated female faces more 1 3 1832 Psychological Research (2021) 85:1823–1836 emotionally intense compared to male face, t(44) = 4.56, Fixation duration p = 0.001, d = 0.68, whereas there was no such difference for female participants, t < 1. Mean durations were calculated for each participant and Mean and standard deviations of the attractiveness and each factorial combination. Male participants fixated sig- emotional intensity ratings are summarized in Table 1. nificantly longer on the presented faces compared to female participants, F(1, 32) = 4.92, p = 0.034, η = 0.13, 95% CI Response time [0.01, 0.31]. The only other significant effect was a three- way interaction between emotional expression, trial type and We computed mean correct response times for each par- participants’ gender, F(1, 32) = 4.47, p = 0.042, η = 0.12, ticipant for each factorial combination. Again, there was a 95% CI [0.00, 0.30]. Whereas there was no significant inter - significant interaction between attractiveness and trial type, action between emotional expression and trial type for male 2 2 F(1, 32) = 10.39, p = 0.003, η = 0.25, 95% CI [0.06, 0.42]. participants, F(1, 15) = 1.14, p = 0.303, η = 0.07, 95% CI Follow-up pairwise comparisons showed that in attractive [0.00, 0.30], there was a marginal significant interaction for crowds participants were faster when crowds contained female participants, F(1, 17) = 3.82, p = 0.067, η = 0.18, more happy compared to angry faces, t(33) = 2.57, p = 0.015, 95% CI [0.00, 0.41], indicating that women fixated longer d = 0.44, whereas in unattractive crowds the pattern pointed on angry (happy) faces when the crowds consisted of mainly towards the opposite direction, t(33) = − 1.91, p = 0.065, angry (happy) faces. d = 0.33 (Fig. 1b). Lastly, there was a significant effect of participants’ gender, F(1, 32) = 4.55, p = 0.041, η = 0.12, Number of fixations 95% CI [0.00, 0.34], indicating that women were faster in judging the mood of the crowd. This measure indicates how many faces of each type received at least one fixation during a trial. A significant Response interaction between emotional expression and trial type was found, F(1, 32) = 201.13, p < 0.001, η = 0.86, 95% CI [0.77, We calculated mean responses (proportion of “angry” 0.90], with more fixations on angry (happy) faces when responses) for each participant and each factorial combina- more angry (happy) faces were presented. Lastly, there was a tion. Again, there was a main effect of attractiveness, F (1, significant interaction between attractiveness and trial type, 2 2 32) = 68.62, p < 0.001, η = 0.68, 95% CI [0.50, 0.77], and F(1, 32) = 9.31, p = 0.005, η = 0.23, 95% CI [0.05, 0.40]. A of trial type, F(1, 32) = 316.51, p < 0.001, η = 0.91, 95% CI higher number of faces was fixated in attractive compared to [0.85, 0.93]. Participants tended to judge attractive crowds unattractive crowds when more angry faces were presented, more often as happy, whereas they judged unattractive t(33) = 2.38, p = 0.024, d = 0.41, whereas there was no sig- crowds more often as angry (Fig. 2b). Participants evalu- nificant difference in crowds containing more happy faces, ated crowds dominated by happy faces as being happy more t(33) = − 1.69, p = 0.101, d = 0.29. often compared to those containing more angry expressions. Accuracy Discussion Mean accuracies were calculated for each participant and In experiment 2, we replicated the main findings of experi- each factorial combination. A main effect of trial type ment 1 in a different sample and using new stimulus material reached significance, F (1, 32) = 5.37, p = 0.027, η = 0.14, that was better matched for attractiveness between female 95% CI [0.01, 0.32], showing higher accuracy rates in trials and male stimuli. Again, participants evaluated attractive with more angry expressions. This main effect was qualified crowds as happy more frequently than unattractive crowds. by an attractiveness × trial type interaction, F(1, 32) = 68.62, Reversely, participants evaluated unattractive (compared p < 0.001, η = 0.68, 95% CI [0.50, 0.77]. Follow-up t to attractive) crowds more often as angry. Moreover, par- tests revealed that accuracy rates were increased when ticipants were faster and more accurate in judging attractive more happy faces were presented in attractive compared crowds dominated by happy faces and unattractive crowds to unattractive crowds, t(33) = 6.50, p < 0.001, d = 1.11, dominated by angry expressions. These findings are in line and the reversed pattern was found for angry expressions, with the evaluative congruence account, which suggests a t(33) = − 5.41, p < 0.001, d = 0.93 (Fig. 3b). facilitated perception when the presented emotion matches the evaluation of the respective social cue. Extending previ- ous findings by Lindeberg et al. (2019), we found evidence 10 supporting the evaluative congruence account not only for The findings with regard to participants’ gender need to be inter - preted with caution as ratings from only 12 men were available. attractive, but also for unattractive faces. 1 3 Psychological Research (2021) 85:1823–1836 1833 In contrast to the findings for reaction times and choices, faces in attractive crowds and higher fixation rates on angry which were well in line with the findings from experiment faces in unattractive crowds. 1, eye-tracking results did not replicate as closely. In experi- Across both experiments, we found evidence for an inu fl - ment 1, we observed an increased number of fixations on ence of attractiveness on emotion perception. Participants in happy attractive compared to angry attractive faces but this both samples evaluated attractive crowds of faces as happy was not the case in experiment 2. It is possible that the dif- more frequently. Moreover, their evaluations were faster ference in material explains this discrepancy. As described and more accurate in attractive crowds dominated by happy above, we performed a new matching of the target material faces. These findings are in line with the evaluative con - to reduce attractiveness die ff rences between female and male gruence account (Hugenberg, 2005; Hugenber & Sczesny, faces that were present in experiment 1. While successful in 2006). The results of the two studies also corroborate recent this regard, the new matching also resulted in smaller overall evidence by Lindeberg et al. (2019), who found that par- differences between attractive and unattractive targets. It is ticipants categorized attractive happy faces faster and more possible that the attention capturing power of the attractive accurately than unattractive happy faces. smiling faces was thereby diminished, leading to a similar In contrast to previous studies, we also found evidence allocation of attention to attractive and unattractive emo- for evaluative congruence with regard to unattractive-angry tional face expressions. faces. Participants rated unattractive crowds as angry more Even though we were able to match the perceived attrac- frequently and evaluated these crowds faster and more accu- tiveness of female and male targets in experiment 2, we rately, when they were dominated by angry faces. Therefore, again found that participants rated happy attractive faces as we were not only able to find support for the evaluative con - more emotionally intense compared to happy unattractive gruence account with regard to attractive crowds of faces, faces and the opposite for angry attractive and unattractive but also for unattractive facial expressions. The reason for face pictures. This time, we collected ratings in a separate this might be that the MoC paradigm exerts higher cognitive sample, ruling out the possibility that the completion of the demands and thus prevents floor effects (i.e. very low error MoC task influenced the consecutive rating. Furthermore, rates with little variance) from occurring. we consider it unlikely that, again, the emotional intensity of A further limitation that was present in previous studies the happy (angry) unattractive faces was truly lower (higher) we also observed in the present experiment 1. Participants compared to the happy (angry) attractive faces as we used rated attractive female faces as more attractive than attrac- two different stimulus sets from the Chicago Face Database. tive male faces, whereas there was no such difference with Therefore, the most plausible explanation might be that respect to unattractive male and female faces (and the same attractiveness also influenced/biased the emotion intensity limitation afflicted studies by Lindeberg et al., 2019). To ratings. Using computer-generated emotional expressions rule out the possibility that the happiness superiority effect matched for attractiveness, gender, and emotional intensity for attractive crowds might be due to this dissimilarity, we would be necessary in future investigations to ensure equal matched female and male faces on attractiveness for the intensities of the emotional faces. second experiment. Even though this led to smaller attrac- tiveness differences between the unattractive and attractive faces, we were able to replicate our findings. General discussion To additionally address potential mechanisms underlying the attractiveness effect on emotion perception, we combined The present study set out to test the evaluative congruence the MoC paradigm with eye-tracking analyses. We hypoth- account, which assumes a faster perception of an emotion esized that evaluative congruence would manifest in higher when it matches the evaluation of a social cue, in two experi- fixation rates and longer fixation durations on happy-attrac- ments. The social cue tested in the present study was attrac- tive and angry-unattractive faces than on happy-unattractive tiveness and we employed a visual search paradigm, the or angry-attractive faces. We partly found this only in the MoC task, in which participants’ task is to judge the over- first experiment, where participants showed a higher number all mood of a crowd (instead of detecting single emotional of fixations on happy compared to angry attractive faces. targets as in previous studies, see Lindeberg et al., 2019). In the second experiment, we found no such evidence. One Based on the evaluative congruence account, we expected explanation might be the reduced attractiveness difference to see a faster and more accurate identification of attractive between the attractive and unattractive crowds as a result crowds dominated by happy faces and of unattractive crowds of the matching procedure in study 2. Moreover, it is nec- dominated by angry faces. Further extending previous stud- essary to differentiate that fixation durations on the happy ies, we also incorporated eye-tracking to assess potential and angry attractive faces were similar in study 1. In con- mechanisms underlying evaluative congruence. For the eye- trast, happy attractive faces were fixated more numerously tracking indices, we expected higher fixation rates on happy compared to angry attractive faces in the first experiment. 1 3 1834 Psychological Research (2021) 85:1823–1836 This indicates that attentional processes manifest in a biased perceive another person’s attractiveness is strongly influ- selective attention towards happy attractive faces but not in a enced by their own attractiveness (Sim, Saperia, Brown, longer fixation duration for the happy attractive faces. How - & Berinieri, 2015). Therefore, future investigations could ever, compared to the large effects with regard to accuracy, address participants’ attractiveness as an additional mod- response tendencies, and response times, only small effects erator. Especially individuals who perceive themselves as on the eye-tracking variables were found in experiment 1. highly attractive might judge attractive crowds more favora- Therefore, we argue that differences in mood judgements in bly compared to unattractive crowds whereas this might not attractive compared to unattractive crowds are likely only be the case for persons who rate themselves as being less slightly influenced by attentional processes but manifest attractive. more strongly in the evaluation phase. Lastly, it would be interesting to further investigate A further, broader implication of the present studies is whether the strength of the effect of attractiveness on emo- that it could be necessary to control for attractiveness when tion perception is influenced by attractiveness-related ste - investigating the influence of other social cues (e.g. gen- reotypes. It might be possible that the attractiveness effect is der or race) on emotion perception, or even in visual tasks potentiated for participants who hold the attractiveness ste- with emotional targets in general (Lindeberg et al., 2019). reotype to a stronger degree (e.g. participants who evaluate It is possible that past studies have confounded constructs attractive individuals more favorably compared to unattrac- such as target gender and attractiveness, and therefore the tive ones and associate attractiveness with more beneficial selection of the stimulus material may have exaggerated or outcomes). Moreover, it would also be interesting to add underestimated effects of social cues on emotion perception an additional social cue to the current experimental design (Lindeberg et al., 2019). Hence, controlling for attractive- and systematically investigate the interplay between attrac- ness when selecting targets appears to be highly important. tiveness and this social cue. For example, further varying— While knowledge about the far-reaching effects of other instead of experimentally controlling for— target gender traits such as target gender or race are relatively well known in the experiment (by showing only female and only male and well addressed in the field, the same does currently not crowds) would enable to test the interaction between target apply to attractiveness. Our findings thus also imply a need attractiveness and target gender on emotion perception in for existing and developing stimulus databases to obtain the mood-of-the-crowd paradigm. The attractiveness effect attractiveness ratings to allow researchers to account for this could be even stronger in female compared to male crowds factor in their studies. as females are perceived more favorably compared to males (Eagly, Mladinic, & Otto, 1991a, b). However, this would Limitations require face databases with a larger number of male and female faces and with a greater variance of attractiveness to The present experiments entailed several limitations. First, build the respective crowds. we found that attractive happy and unattractive angry faces were rated as more emotionally intense compared to attrac- tive angry and unattractive happy faces in both experiments. Conclusion As we used different target faces in the two studies, we believe it to be unlikely that happiness was expressed more In two experiments, we demonstrated that attractiveness strongly in attractive faces and anger more strongly in unat- affects emotion perception in a visual search paradigm tractive face expressions. To us, it seemed more plausible with multiple emotional targets, the MoC paradigm. Spe- that influences of attractiveness that were observed in the cifically, participants evaluated attractive crowds containing mood judgements in the MoC task also transfer to the emo- more happy expressions faster and more accurately, and the tional intensity ratings of the faces pictures. Future studies same was true for unattractive crowds dominated by angry should aim to incorporate material in which attractive and expressions, which corroborates the evaluative congruence unattractive faces express respective emotions to the same account. Moreover, attractive crowds were judged as being degree. One possibility would be to use computer-gener- happy more often whereas unattractive crowds were per- ated faces that are closely matched in terms of emotionality, ceived as being angry more frequently. Additionally, eye- gender and attractiveness. Another option would be to use tracking analyses revealed that there is also a small effect machine learning approaches to select attractive and unat- of attractiveness on gaze movements, though this was pre- tractive faces that show the same emotional intensity. sent only in experiment 1. Specifically, we observed higher A second limitation is that we did not investigate partici- fixation rates on happy compared to angry attractive targets, pants’ own attractiveness, which may also play a role when implying that attractiveness plays a role even in the early judging the mood of attractive and unattractive crowds. Pre- stages of perception. These results imply that face attractive- vious studies suggest that the degree to which individuals ness should be carefully considered when selecting material 1 3 Psychological Research (2021) 85:1823–1836 1835 Bucher, A., Voss, A., Spaniol, J., Hische, A., & Sauer, N. (2019). Age for future studies that examine emotion perception, and stim- differences in emotion perception in a multiple target setting: an ulus attractiveness should be experimentally controlled or eye-tracking study. Emotion. https ://doi.or g/10.1037/emo00 00645 . statistically adjusted for. Calvo, M. G., & Marrero, H. (2009). Visual search of emotional faces: the role of affective content and featural distinctiveness. Cogni- tion and Emotion, 23(4), 782–806. https://doi.or g/10.1080/02699 93080 21516 54. Funding Open access funding enabled and organized by Projekt Craig, B. M., & Lipp, O. V. (2017). The influence of facial sex cues on DEAL. This research received no specific grant from any funding emotional expression categorization is not fixed. Emotion, 17(1), agency in the public, commercial or not-for-profit sectors. 28–39. https ://doi.org/10.1037/emo00 00208 . Craig, B. M., Mallan, K. M., & Lipp, O. V. (2012). The effect of poser Availability of data and material Data from the experiments can be race on the happy categorization advantage depends on stimulus downloaded from https://heida t a.uni-heidelber g.de/dat ase t.xhtml?persi type, set size, and presentation duration. Emotion, 12(6), 1303– stent Id=doi:10.11588 /data/LYWYG N 1314. https ://doi.org/10.1037/a0028 622. Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful Compliance with ethical standards is good. Journal of Personality and Social Psychology, 24(3), 285–290. https ://doi.org/10.1037/h0033 731. Eagly, A. H., Ashmore, R. D., Makhijani, M. G., & Longo, L. C. Conflict of interest The authors declare no conflict of interest. (1991a). What is beautiful is good, but…: a meta-analytic review of research on the physical attractiveness stereo- Open Access This article is licensed under a Creative Commons Attri- type. Psychological Bulletin, 110(1), 109–128. https ://doi. bution 4.0 International License, which permits use, sharing, adapta- org/10.1037/0033-2909.110.1.109. tion, distribution and reproduction in any medium or format, as long Eagly, A. H., Mladinic, A., & Otto, S. (1991b). Are women evalu- as you give appropriate credit to the original author(s) and the source, ated more favorably than men? An analysis of attitudes, beliefs, provide a link to the Creative Commons licence, and indicate if changes and emotions. Psychology of Women Quarterly, 15(2), 203–216. were made. The images or other third party material in this article are https ://doi.org/10.1111/j.1471-6402.1991.tb007 92.x. included in the article’s Creative Commons licence, unless indicated Ebner, N. C., & Johnson, M. K. (2010). Age-group differences in otherwise in a credit line to the material. If material is not included in interference from young and older emotional faces. Cognition the article’s Creative Commons licence and your intended use is not and Emotion, 24(7), 1095–1116. https://doi.or g/10.1080/02699 permitted by statutory regulation or exceeds the permitted use, you will 93090 31283 95. need to obtain permission directly from the copyright holder. To view a Elias, E., Dyer, M., & Sweeny, T. D. (2017). Ensemble perception copy of this licence, visit http://creativ ecommons .or g/licenses/b y/4.0/. of dynamic emotional groups. Psychological Science, 28(2), 193–203. https ://doi.org/10.1177/09567 97616 67818 8. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). GPower 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Meth- References ods, 39(2), 175–191. https ://doi.org/10.3758/BF031 93146 . Golle, J., Mast, F. W., & Lobmaier, J. S. (2014). Something to smile Becker, D. V., Anderson, U. S., Mortensen, C. R., Neufeld, S. L., & about: the interrelationship between attractiveness and emo- Neel, R. (2011). The face in the crowd effect unconfounded: tional expression. Cognition and Emotion, 28(2), 298–310. https Happy faces, not angry faces, are more efficiently detected in sin- ://doi.org/10.1080/02699 931.2013.81738 3. gle- and multiple-target visual search tasks. Journal of Experi- Hansen, C. H., & Hansen, R. D. (1988). Finding the face in the mental Psychology: General, 140(4), 637–659. https ://doi. crowd: an anger superiority effect. Journal of Personal- org/10.1037/a0024 060.supp. (Supplemental). ity and Social Psychology, 54(6), 917–924. https ://doi. Becker, D. V., Kenrick, D. T., Neuberg, S. L., Blackwell, K. C., & org/10.1037/0022-3514.54.6.917. Smith, D. M. (2007). The confounded nature of angry men and Hugenberg, K. (2005). Social categorization and the perception of happy women. Journal of Personality and Social Psychology, facial affect: target race moderates the response latency advan- 92(2), 179–190. https ://doi.org/10.1037/0022-3514.92.2.179. tage for happy faces. Emotion, 5(3), 267–276. https ://doi. Becker, D. V., Neel, R., Srinivasan, N., Neufeld, S., Kumar, D., & org/10.1037/1528-3542.5.3.267. Fouse, S. (2012). The vividness of happiness in dynamic facial Hugenberg, K., & Sczesny, S. (2006). On wonderful women and displays of emotion. PLoS ONE. https ://doi.org/10.1371/journ seeing smiles: social categorization moderates the happy face al.pone.00265 51. response latency advantage. Social Cognition, 24(5), 516–539. Bijlstra, G., Holland, R. W., & Wigboldus, D. H. J. (2010). The social https ://doi.org/10.1521/soco.2006.24.5.516. face of emotion recognition: evaluations versus stereotypes. Jour- Jackson, L. A., Hunter, J. E., & Hodge, C. N. (1995). Physical attrac- nal of Experimental Social Psychology, 46(4), 657–663. https :// tiveness and intellectual competence: a meta-analytic review. doi.org/10.1016/j.jesp.2010.03.006. Social Psychology Quarterly, 58(2), 108–122. https ://doi. Borkenau, P., & Liebler, A. (1992). Trait inferences: Sources of validity org/10.2307/27871 49. at zero acquaintance. Journal of Personality and Social Psychol- Juth, P., Lundqvist, D., Karlsson, A., & Öhman, A. (2005). Looking ogy, 62(4), 645–657. https://doi.or g/10.1037/0022-3514.62.4.645. for foes and friends: perceptual and emotional factors when Borkenau, P., & Liebler, A. (1995). Observable attributes as manifes- finding a face in the crowd. Emotion, 5(4), 379–395. https://doi. tations and cues of personality and intelligence. Journal of Per- org/10.1037/1528-3542.5.4.379. sonality, 63(1), 1–25. https ://doi.org/10.1111/j.1467-6494.1995. Langlois, J. H., Kalakanis, L., Rubenstein, A. J., Larson, A., Hal- tb007 99.x. lam, M., & Smoot, M. (2000). Maxims or myths of beauty? A Bucher, A., & Voss, A. (2019). Judging the mood of the crowd: atten- meta-analytic and theoretical review. Psychological Bulletin, tion is focused on happy faces. Emotion, 19(6), 1044–1059. https 126(3), 390–423. https://doi.or g/10.1037/0033-2909.126.3.390. ://doi.org/10.1037/emo00 00507 . 1 3 1836 Psychological Research (2021) 85:1823–1836 Leppänen, J. M., & Hietanen, J. K. (2004). Positive facial expres- learning. Psychological Review, 108(3), 483–522. https ://doi. sions are recognized faster than negative facial expressions, but org/10.1037/0033-295X.108.3.483. why? Psychological Research Psychologische Forschung, 69, Pinkham, A. E., Griffin, M., Baron, R., Sasson, N. J., & Gur, R. C. 22–29. https ://doi.org/10.1007/s0042 6-003-0157-2. (2010). The face in the crowd effect: Anger superiority when Leppänen, J. M., Tenhunen, M., & Hietanen, J. K. (2003). Faster using real faces and multiple identities. Emotion, 10(1), 141–146. choice-reaction times to positive than to negative facial Savage, R. A., Lipp, O. V., Craig, B. M., Becker, S. I., & Horstmann, expressions: the role of cognitive and motor processes. G. (2013). In search of the emotional face: anger versus happiness Journal of Psychophysiology, 17(3), 113–123. https ://doi. superiority in visual search. Emotion, 13(4), 758–768. https://doi. org/10.1027//0269-8803.17.3.113.org/10.1037/a0031 970. Lindeberg, S., Craig, B. M., & Lipp, O. V. (2019). You look pretty Scherer, K. R., & Wallbott, H. G. (1994). Evidence for universality happy: attractiveness moderates emotion perception. Emotion, and cultural variation of differential emotion response patterning. 19(6), 1070–1080. https ://doi.org/10.1037/emo00 00513 .supp. Journal of Personality and Social Psychology, 66(2), 310–328. (Supplemental).https ://doi.org/10.1037/0022-3514.66.2.310. Ma, D. S., Correll, J., & Wittenbrink, B. (2015). The Chicago Face Sim, S. Y.-L., Saperia, J., Brown, J. A., & Bernieri, F. J. (2015). Judg- Database: a free stimulus set of faces and norming data. Behavior ing attractiveness: biases due to raters’ own attractiveness and Research Methods, 47(4), 1122–1135. intelligence. Cogent Psychology, 2(1). Mueser, K. T., Grau, B. W., Sussman, S., & Rosen, A. J. (1984). You’re Smits, G. J., & Cherhoniak, I. M. (1976). Physical attractiveness and only as pretty as you feel: facial expression as a determinant of friendliness in interpersonal attraction. Psychological Reports, physical attractiveness. Journal of Personality and Social Psychol- 39(1), 171–174. https ://doi.org/10.2466/pr0.1976.39.1.171. ogy, 46(2), 469–478. https://doi.or g/10.1037/0022-3514.46.2.469. Watkins, L. M., & Johnston, L. (2000). Screening job applicants: the Öhman, A., Juth, P., & Lundqvist, D. (2010). Finding the face in a impact of physical attractiveness and application quality. Interna- crowd: relationships between distractor redundancy, target emo- tional Journal of Selection and Assessment, 8(2), 76–84. https :// tion, and target gender. Cognition and Emotion, 24(7), 1216– doi.org/10.1111/1468-2389.00135 . 1228. https ://doi.org/10.1080/02699 93090 31668 82. Öhman, A., Lundqvist, D., & Esteves, F. (2001). The face in the crowd Publisher’s Note Springer Nature remains neutral with regard to revisited: a threat advantage with schematic stimuli. Journal of jurisdictional claims in published maps and institutional affiliations. Personality and Social Psychology, 80(3), 381–396. https:/ /doi. org/10.1037/0022-3514.80.3.381. Öhman, A., & Mineka, S. (2001). Fears, phobias, and pre- paredness: toward an evolved module of fear and fear 1 3

Journal

Psychological Research – Springer Journals

Published: May 25, 2020

There are no references for this article.