Evaluations of Smiles in Relation to Social Anxiety and Psychopathic Traits PDF

Cognition and Emotion ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/pcem20 Evaluations of three different types of smiles in relation to social anxiety and psychopathic traits Anna L. Dapprich, Eva Gilboa-Schechtman, Eni S. Becker & Mike Rinck To cite this artic...

Cognition and Emotion ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/pcem20 Evaluations of three different types of smiles in relation to social anxiety and psychopathic traits Anna L. Dapprich, Eva Gilboa-Schechtman, Eni S. Becker & Mike Rinck To cite this article: Anna L. Dapprich, Eva Gilboa-Schechtman, Eni S. Becker & Mike Rinck (2022) Evaluations of three different types of smiles in relation to social anxiety and psychopathic traits, Cognition and Emotion, 36:3, 535-545, DOI: 10.1080/02699931.2021.2016638 To link to this article: https://doi.org/10.1080/02699931.2021.2016638 © 2021 The Author(s). Published by Informa View supplementary material UK Limited, trading as Taylor & Francis Group Published online: 24 Dec 2021. Submit your article to this journal Article views: 1906 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=pcem20 COGNITION AND EMOTION 2022, VOL. 36, NO. 3, 535–545 https://doi.org/10.1080/02699931.2021.2016638 BRIEF ARTICLE Evaluations of three different types of smiles in relation to social anxiety and psychopathic traits a Anna L. Dapprich , Eva Gilboa-Schechtmanb, Eni S. Beckera and Mike Rincka a Experimental Psychopathology & Treatment, Behavioural Science Institute, Radboud University, Nijmegen, the Netherlands; b Department of Psychology, Gonda Brain Science Center, Bar-Ilan University, Ramat Gan, Israel ABSTRACT ARTICLE HISTORY Research has identified three different types of smiles – the reward, affiliation and Received 25 March 2021 dominance smile – which serve expressions of happiness, connectedness, and Revised 28 November 2021 superiority, respectively. Examining their explicit and implicit evaluations by Accepted 6 December 2021 considering a perceivers’ level of social anxiety and psychopathy may enhance our KEYWORDS understanding of these smiles’ theorised meanings, and their role in problematic Social phobia; callous– social behaviour. Female participants (N=122) filled in questionnaires on social unemotional traits; cognitive anxiety, psychopathic tendencies (i.e. the affective-interpersonal deficit and bias; approach-avoidance antisocial lifestyle) and callous–unemotional (CU) traits. In order to measure explicit behaviour; facial emotional and implicit evaluations of the three smiles, angry and neutral facial expressions, expressions an Explicit Valence Rating Task and an Approach-Avoidance Task were administered. Results indicated that all smiles were explicitly evaluated as positive. No differences in implicit evaluations between the smile types were found. Social anxiety was not associated with either explicit or implicit smile evaluations. In contrast, CU-traits were negatively associated with explicit evaluations of reward and dominance smiles. These findings support the assumptions of non-biased explicit information processing in social anxiety, and flattened emotional sensitivity in CU-traits. The importance of a multimethod approach to enhance the understanding of the effects of smile types on perceivers is discussed. From a social-functional perspective (Keltner & Haidt, dominance smiles have different, namely positive 1999; Niedenthal et al., 2010), at least three different and negative, functions (Martin et al., 2018, 2021; types of smiles with distinct morphologies, as well Orlowska et al., 2018; Rychlowska et al., 2017, 2021), as distinct meanings and functions can be identified. many studies used smiles homogenously to represent First, symmetrical smiles together with raised eye positive social stimuli. brows express happiness and social approval Although these studies did not differentiate (reward smile). Second, symmetrical smiles and a between different smile types, they suggest that lowered upper lip signal the willingness to bond certain perceiver characteristics are related to differ- socially (affiliation smile). Finally, asymmetrical ences between explicit and implicit evaluations of smiles, a wrinkled nose and a raised upper lip both positive and negative facial expressions (Eisen- express dominance and superiority (dominance barth et al., 2008; Heuer et al., 2007; von Borries smile; Martin et al., 2017; Niedenthal et al., 2010; et al., 2012). In particular, this was observed for psy- Rychlowska et al., 2017). Despite the growing body chopathic traits and social anxiety: Aberrant implicit of research showing that the reward, affiliation and evaluations of facial expressions, in the form of CONTACT Anna L. Dapprich [email protected] Experimental Psychopathology & Treatment, Behavioural Science Institute, Radboud University, 6500 HE, Nijmegen, the Netherlands Supplemental data for this article can be accessed https://doi.org/10.1080/02699931.2021.2016638. © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons. org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way. 536 A. L. DAPPRICH ET AL. impulsive approach-avoidance reactions, were reactivity; Martin et al., 2018). However, participants observed. These might in turn explain problematic with higher baseline high-frequency heart rate varia- social behaviour, such as aggression and withdrawal, bility (HF-HRV), i.e. an index of emotion recognition as seen in psychopathic traits and social anxiety, abilities (Lischke et al., 2017; Quintana et al., 2012), respectively (Dapprich et al., 2021; Heuer et al., 2007; showed the lowest stress responses when confronted von Borries et al., 2012). In order to further validate with the affiliation smile (i.e. decreased heart rate) and the distinct positive and negative meanings of the the highest stress responses when confronted with three smiles, the current study therefore investigated the dominance smile (i.e. increased cortisol reactivity), whether their explicit and implicit evaluations differ as respectively (Martin et al., 2018). Thus, perceiver a function of the perceiver’s social anxiety and psy- characteristics seem to play a role in the recognition chopathic traits. of the positive and negative meanings of different The distinctiveness of the reward, affiliation and smiles. dominance smile has been well underpinned by Two social dysfunctions that are related to biased both explicit and implicit assessments. With regard explicit and implicit responses to positive and negative to explicit assessments, such as recognition tasks, it facial expressions are social anxiety and psychopathic has been found that the reward, affiliation and dom- tendencies. While social anxiety is defined by high inance smile were generally recognised as such fear in social situations and avoidance behaviour (Orlowska et al., 2018; Rychlowska et al., 2017). Percei- (Heimberg et al., 1999; Rapee & Heimberg, 1997), psy- vers also recognised the positive meaning of the chopathy is defined by a lack of fear and rule-transgres- reward smile by assuming that expressers intend to sive, aggressive behaviour (Cleckley, 1941; Hare & convey happiness, joy and contentment. The same Neumann, 2008; Lykken, 1957). The definitions seem was true for the negative meaning of the dominance to describe opposing emotional (i.e. high fear versus smile, which was interpreted as conveying superiority, lack of fear) and behavioural (i.e. avoidance versus disapproval and contempt (Martin et al., 2021; aggression) components. Negative correlations Orlowska et al., 2018; Rychlowska et al., 2017). between self-reported social anxiety and psychopathic However, the meaning of the affiliation smile, tendencies have indeed been found (Dapprich et al., namely promoting social connectedness, could not 2021; Hofmann et al., 2009), although self-reported be differentiated from the meaning of the reward avoidance and aggressive behaviour may correlate smile (Orlowska et al., 2018; Rychlowska et al., 2017), positively (Dapprich et al., 2021). Another shared and its expression was labelled as faked (Martin mechanism of social anxiety and psychopathy might et al., 2021). Yet, another study showed that be distortions in the evaluations of facial expressions. opponents who expressed the affiliation smile, as Even though explicit evaluations, i.e. valence compared to the other two smiles, were assumed to ratings, of smiles do not seem to differ as a function have less negative intentions and a greater willing- of social anxiety or psychopathy (Eisenbarth et al., ness to repair the relationship after breaches of trust 2008; Heuer et al., 2007), implicit evaluations, i.e. in economic games (Rychlowska et al., 2021). Thus, impulsive approach-avoidance tendencies, do differ when seen out of context, positive intentions commu- (Dapprich et al., 2021; Heuer et al., 2007; Lange nicated by the affiliation smile might be more difficult et al., 2008; Roelofs et al., 2010; von Borries et al., to understand (Martin et al., 2021). Next to context 2012). The Approach-Avoidance Task (AAT) is a fre- information, perceiver characteristics may be crucial quently used task to assess implicit evaluations, build- to recognise the meaning of the smiles. ing on the innate motivation to approach pleasure Implicit assessments, such as physiological and avoid harm (Phaf et al., 2014). In this task, partici- measurements, further support the positive meanings pants respond to pictures of facial expressions by of both the reward smile and the affiliation smile, and either pulling them closer with a joystick (approach) the negative meaning of the dominance smile. During or pushing them away (avoid). Healthy participants a social speech task, male participants who faced are generally quicker to approach smiling faces and observers who expressed reward or affiliation smiles to avoid angry ones (Phaf et al., 2014). However, showed lower stress responses (i.e. decreased cortisol socially anxious individuals quickly avoid both reactivity), whereas male participants who faced smiling and angry faces, suggesting that they observers who expressed dominance smiles showed implicitly evaluate social interactions, regardless of higher stress responses (i.e. increased cortisol their valence, as aversive (e.g. Heuer et al., 2007). In  COGNITION AND EMOTION 537 contrast, individuals with higher psychopathic traits In order to test these hypotheses, self-report ques- only slowly avoid angry faces, suggesting that they tionnaires on social anxiety and psychopathic traits, implicitly evaluate angry faces as less aversive (Dap- an explicit valence rating task, and an AAT were admi- prich et al., 2021; von Borries et al., 2012). Overgener- nistered in female students. Only female students were alised avoidance of both positive and negative facial tested since they might better recognise subtle facial expressions might underlie social avoidance, expressions (Hoffmann et al., 2010). We operationalised whereas a lack of avoidance of negative facial psychopathic traits by means of two well-validated expressions might underlie aggressive behaviour questionnaires to comprehensively capture its charac- (Heuer et al., 2007; von Borries et al., 2012). Yet, pre- terising attributes. That is, first, the Levenson Self- vious research did not consider that both positive Report Psychopathy Scale was used to measure both and negative meanings can be conveyed through the affective-interpersonal deficit (Factor I) and the smiles. antisocial lifestyle (Factor II; Levenson et al., 1995). Therefore, the current study examined whether Second, the Inventory of Callous–Unemotional Traits perceivers’ levels of social anxiety and psychopathic was used to measure callous–unemotional traits traits are related to differences in the explicit and (Frick, 2004). In all analyses, we also controlled for implicit evaluations of the reward, affiliation and dom- gender of the expresser based on the stereotypes inance smile. Based on the theoretically positive that females are more affiliative and males more domi- meaning of both the reward smile (i.e. expressing nant (Hess et al., 2005), which might lead to more posi- happiness) and the affiliation smile (i.e. promoting tive evaluations and faster approach tendencies for social connectedness), we expected both to be evalu- expressions that conform to these stereotypes. ated positively, both explicitly and implicitly (i.e. approached quickly). In contrast, based on the theor- Methods etically negative meaning of the dominance smile (i.e. expressing dominance, which is reciprocally related to The current study combined different research inter- anger; Cabral et al., 2016), we expected the domi- ests. Both the whole study set-up (#13747) and the nance smile to be evaluated negatively, both explicitly current research questions including the statistical and implicitly (i.e. avoided quickly). approach (#30627) were pre-registered on aspredicte- Based on the seemingly opposing correlates of d.org. In the following, only the measurements which social anxiety and psychopathy, we first tested are relevant for the current research question will be whether self-reported social anxiety and psychopathic explained in detail.1 tendencies did indeed correlate negatively with each other. Explicit evaluations were not expected to differ Participants as a function of social anxiety or psychopathic traits (Eisenbarth et al., 2008; Heuer et al., 2007), but implicit We recruited 123 female psychology students from evaluations, operationalised as approach-avoidance the participant pool of Radboud University, Nijmegen, tendencies, were expected to differ. Participants with the Netherlands.2 Participants were between 17 and higher social anxiety were expected to avoid all three 48 years old (M[SD] = 20.37 [4.02]). One participant smiles more quickly than individuals with lower was excluded due to technical problems. The result- social anxiety, in line with their overgeneralised avoid- ing final sample consisted of 122 females. Partici- ance of smiling and angry facial expressions (e.g. Heuer pation was rewarded with course credit. et al., 2007). Differences in the speed of these avoid- ance tendencies might indicate which smile they find Procedure most aversive. Participants with higher psychopathic traits were expected to avoid the dominance smile All participants first completed the Approach-Avoid- more slowly than individuals with lower psychopathic ance Task and a Motivated Viewing Task (the latter traits, in line with their reduced avoidance of angry is unrelated to the current research questions since facial expressions (Dapprich et al., 2021; von Borries it did not include pictures of the three smiles). Then et al., 2012). Weaker avoidance tendencies in response participants completed the Explicit Valence Rating to dominance smiles might indicate in how far more Task and filled in the Inventory of Callous–Unemo- subtle negative expressions, as compared to angry tional Traits, the Levenson Self-Report Psychopathy expressions, are perceived as provocative. Scale, the Liebowitz Social Anxiety Scale and the 538 A. L. DAPPRICH ET AL. State Self-Esteem Scale. Only the Approach-Avoid- computer screen with a joystick mounted on the ance Task and the Explicit Valence Rating Task con- table. Single pictures of the facial expressions were pre- tained pictures of the three different smiles, sented on the screen in medium size. Upon pulling therefore only these two tasks are relevant for the versus pushing of the joystick, the picture grew or current paper. Validated pictures of white actors shrank in size, respectively. The picture disappeared expressing the three smiles were derived from when a full movement in the correct direction was Rychlowska et al. (2017) and Martin et al. (2021). made, and the latency of this full movement was auto- Only full-intensity expressions of the three smiles matically recorded. Participants had to respond to the were used. The whole experiment took one hour, colour of each picture (sepia or grey). Half of the partici- after which participants were thanked and debriefed. pants had to pull all grey pictures and to push all sepia pictures. This was reversed for the other half. The 40 Explicit Valence Rating Task (EVRT) pictures that were also used in the EVRT were mixed with 80 filler pictures of other actors showing only In the EVRT, participants rated the valence of five neutral expressions. Each picture was pushed and different facial expressions: the affiliation smile, the pulled once, yielding a total of 240 trials. In the ana- reward smile, the dominance smile, an angry and a lyses, only the 80 trials of the 8 actors expressing the neutral expression. Eight different white actors (4 three smiles, angry and neutral expressions were females and 4 males) displayed each of these used. The internal consistency of the reaction time expressions (Martin et al., 2021; Rychlowska et al., (RT) differences in the whole AAT was poor (RT of 2017), yielding a total of 40 trials. The instructions pre- pushing minus the corresponding RT of pulling; α sented with each picture read: “Imagine someone =.25), and it was even lower for the different combi- would look at you like this. Do you think this expression nations of facial expressions and gender (see sup- is rather positive or rather negative?” Pictures were pre- plementary material for the descriptive statistics). sented with 72 × 72 dpi resolution, and they remained on the screen until a response was given, using a rating scale ranging from −100 (negative) to +100 (positive) Liebowitz Social Anxiety Scale (LSAS; with 0 being neutral. Internal consistencies of the Liebowitz, 1987) valence ratings ranged from poor (α =.36 for female The LSAS subscale measuring fear of social situations affiliation smiles) to excellent (α =.91 for angry faces). was used. For 24 social situations, participants rated See Table 1 for the descriptive statistics. their perceived level of anxiety (e.g. “talking to people in authority”). Answers were given on a Approach-Avoidance Task (AAT; Heuer et al., Likert-scale ranging from 0 (none) to 3 (severe) and 2007) were summed up to the anxiety subscale score. An adapted version of the AAT was used including pic- Internal consistency was high (α =.90). tures of the three smiles, next to pictures of angry and neutral faces. Participants were seated in front of a Inventory of Callous–Unemotional Traits (ICU; Frick, 2004) Table 1. Descriptive statistics of explicit valence ratings (N = 122). Emotion Min M (SD) Max Cronbach’s α The ICU measures callous–unemotional traits. It con- Reward smile 13 71 (20) 100.85 sists of 24 items (e.g. “I do not show my emotions to male 6.5 68 (23) 100.72 others”), which were rated on a Likert-scale ranging female 5.5 75 (20) 100.79 Affiliation smile −7 38 (17) 75.67 from 0 (not at all true) to 3 (definitely true). Internal con- male −13.5 37 (19) 78.64 sistency for the total score was acceptable (α =.70). female −19 40 (19.5) 82.36 Dominance smile −16 28 (19) 85.74 male −25.5 32 (22) 90.56 Levenson Self-Report Psychopathy Scale female −33 24 (20) 80.55 Neutral −85 −21 (23) 20.5.72 (LSRPS; Levenson et al., 1995) male −81 −20 (15.5) 8.55 female −90 −21 (18) 34.59 The LSRPS measures psychopathic tendencies in non- Angry −100 −64 (21) −2.91 clinical samples. It consists of 26 items measuring the male −100 −67 (21) −6.83 affective-interpersonal deficit (i.e. Factor I; e.g. female −100 −61 (23) 3.84 “Looking out for myself is my top priority”) and  COGNITION AND EMOTION 539 antisocial lifestyle (i.e. Factor II; e.g. “I find myself in the materials. We found significant positive correlations of same kind of trouble time after time”). They are rated the anxiety subscale of the Liebowitz Social Anxiety on a Likert-scale ranging from 1 (strongly disagree) to 4 Scale with the Levenson Self-Report Psychopathy (strongly agree). Internal consistency of Factor II (α Scale total score (r =.30, p =.010) and with the Leven- =.63) was questionable, but acceptable for Factor I son Self-Report Psychopathy Scale Factor II scale (r (α =.72) and the total score (α =.77). =.34, p =.002): Participants with higher social anxiety had higher levels of psychopathic tendencies. Statistical approach Data were analysed using the statistical software s R Explicit valence ratings (version 3.5.0; R Core Team, 2019) and R studio Explicit valence ratings were predicted as a function (version 1.2.5033; RStudio Team, 2019). Pearson’s cor- of the categorical predictors facial expression relations between questionnaires and valence ratings (reward smile/ affiliation smile/ dominance smile/ were calculated with the function corr.test of the angry/ neutral) and actors’ gender (male/ female), package “psych” using Bonferroni adjustment to which were sum-to-zero coded, as well as the continu- control for multiple testing (version 1.8.12; Revelle, ous predictors social anxiety (i.e. Liebowitz Social 2018). Bayesian linear mixed models were conducted Anxiety Scale anxiety score) and psychopathic ten- to analyse our main research questions by using the dencies (i.e. Inventory of Callous–Unemotional Traits function brm of the package “brms” (version 2.11.1; total score and Levenson Self-Report Psychopathy Bürkner, 2017). The models are described in more Scale total score), which were standardised. The detail in the supplementary materials. Model conver- four-way interactions between facial expression, gence was assessed in terms of Rhat (values >.9 gender, social anxiety and the measurements for psy- and < 1.1 were considered as good) and diagnostic chopathic traits, respectively, as well as the underlying plots (Bürkner, 2017). Importantly, we considered an three- and two-way interactions were included. effect to be statistically significant, if the 95% pos- Random intercepts for participant and actor were terior credible interval (CI) did not include zero. Sig- included. Facial expression, gender, and their inter- nificant interactions were followed up using the action were entered as random slopes varying package “emmeans” (Lenth, 2020). across participants. Facial expression was included as random slope varying across actor. A more detailed Transparency statement description of the statistical model can be found in the supplementary materials. Please note that the preregistration of the current study As mentioned above, we expected a significant contains a slightly different terminology, less nuanced main effect of facial expression on valence ratings, hypotheses, and a different analysis for the explicit indicating more positive evaluations of both the valence ratings. First, in the pre-registration, we refer reward smile and the affiliation smile, and more nega- to the three smiles by using rather plain terms. Here, tive evaluations of the dominance smile. Furthermore, we decided to use the established terms in order to we expected a significant interaction of facial enhance the comparability of our and previous expression and gender on valence ratings, indicating results. Second, we were able to specify our hypotheses more positive evaluations of affiliation smiles regarding gender and explicit evaluations after reading expressed by females, and dominance smiles more literature on social stereotypes. Finally, we expressed by males. We did not expect that social realised that the explicit valence ratings can be ana- anxiety or psychopathic tendencies would be lysed on the trial level, thereby controlling for partici- related to any differences in the valence ratings. pant and actor. Thus, we used mixed models for the The model converged without warnings and the analysis of the explicit valence ratings, too. diagnostic plots indicated that sampling succeeded. Most important for the current questions, the Results reward smile (estimated regression coefficient [B] = 60.62), lower and upper bounds of the 95% posterior Correlations between questionnaires credible interval (95% CI) [50.47, 70.57], the affiliation Descriptive statistics of our measures and a complete smile (B = 27.51, 95% CI [10.57, 44.79]) and the domi- correlation matrix are depicted in the supplementary nance smile (B = 17.13, 95% CI [2.16, 32.01]) were all 540 A. L. DAPPRICH ET AL. rated as more positive than the overall mean, whereas were specified, namely for facial expression, move- angry expressions (B = −74.30, 95% CI [−84.93, ment, gender, social anxiety (i.e. Liebowitz Social −63.60]) were rated as more negative than the Anxiety Scale) and psychopathic tendencies (i.e. overall mean. Post-hoc analyses showed that reward Inventory of Callous–Unemotional Traits and Leven- smiles were rated more positively than affiliation son Self-Report Psychopathy Scale, respectively). In smiles (B = −33.1, higher and lower posterior density addition to that, the underlying four-, three- and interval (HPD) [−53.2, −12.7]), and also more positively two-way interactions were included. Random inter- than dominance smiles (B = −43.7, HPD [−62.3, cepts for participant and actor were entered. Facial −24.5]), whereas the difference between affiliation expression, movement, gender and their interactions smiles and dominance smiles was not significant (B were specified as random slopes varying across par- = 10.6, HPD [−14.5, 35.6]). ticipants. Facial expression and movement were Furthermore, there were significant interactions included as random slopes varying across actor. A between the Inventory of Callous–Unemotional complete description of the statistical model can be traits total score and the ratings of the reward smile, found in the supplementary material. the dominance smile and angry expressions. Thus, As mentioned above, we expected a significant with increasing levels of callous–unemotional traits, two-way interaction of facial expression and move- the reward smile (B = −5.01, 95% CI [−8.25, −1.73]) ment on reaction times, indicating shorter reaction and the dominance smile (B = −2.80, 95% CI [−5.43, times for the approach of both the reward and affilia- −0.19]) were rated as less positive, whereas angry tion smile, as well as shorter reaction times for the expressions were rated as less negative (B = 6.28, avoidance of dominance smiles (and angry 95% CI [2.08, 10.35]). For an illustration, see Figure 1. expressions). Furthermore, we expected significant Finally, there was a significant interaction between three-way interactions between facial expression, ratings of angry female faces and Levenson Self- movement and social anxiety, indicating shorter reac- Report Scale scores (B = 2.01, 95% CI [0.42, 3.63]), tion times for the avoidance of all smiles, as well as suggesting that higher psychopathic tendencies significant three-way interactions between facial were related to less negative ratings of females’ expression, movement and psychopathic tendencies, angry expressions. However, neither the effect of indicating shorter reaction times for the avoidance gender nor the effect of social anxiety was significant, of dominance smiles (and angry expressions). indicating that the explicit valence ratings did not The model converged without warnings and the differ between male and female actors, nor did they diagnostic plot indicated that sampling succeeded. differ for participants with varying levels of social There was a significant main effect of movement on anxiety. reaction time, indicating that participants were gener- ally faster to push than to pull the joystick (B = 0.03, 95% CI [0.02, 0.05]). Neither a significant interaction Approach-avoidance response times between facial expressions and movement, nor Approach-avoidance tendencies (operationalised as between facial expressions, movement and any reaction time differences in the Approach-Avoidance other predictor of interest was found. Thus, social Task) were predicted as a function of the same anxiety, CU-traits, psychopathic tendencies, and model as for explicit valence ratings, though the cat- expresser gender were unrelated to the responses egorical predictor movement (push/ pull) was to facial expressions. added. That is, reaction times of correct trials that were performed with the pictures of the same actors Discussion as in the explicit valence rating were entered as the dependent variable. The categorical predictors facial The aim of the current study was to examine the expli- expression (reward smile/ affiliation smile/ dominance cit and implicit evaluations of three different types of smile/ angry/ neutral), movement (push/ pull) and smiles by also taking perceivers’ social anxiety and actors’ gender (male/ female) were sum-to-zero psychopathic traits into account. At an explicit level, coded, whereas the continuous predictors Liebowitz all smile types were evaluated as more positive than Social Anxiety Scale, Inventory of Callous–Unemo- neutral and angry expressions. Reward smiles were tional Traits and Levenson Self-Report Psychopathy evaluated most positively. Affiliation smiles and dom- Scale were standardised. Two five-way interactions inance smiles were evaluated positively, too, but did  COGNITION AND EMOTION 541 Figure 1. Explicit valence ratings per facial expression as a function of social anxiety and callous-unemotional traitsNote: Differences in explicit valence ratings as a function of social anxiety (i.e. LSAS anxiety score, upper panel) and callous-unemotional traits (i.e. ICU scores, lower panel). Social anxiety had no siginificant effect on valence ratings. Lines that are marked with an asterisk differ significantly from the overall mean across emotions with increasing ICU total scores. Importantly, for displaying purposes, valence ratings were averaged within emotions, whereas in the analyses, data of single trials were analyzed. not differ from each other in perceived positivity. With theoretical meaning and previous findings. That is, increasing levels of callous–unemotional (CU) traits, its distinct morphological features, namely symmetric the reward smile and the dominance smile were eval- activations of the zygomaticus major and raised eye uated less positively, and angry faces less negatively. brows, have consistently been interpreted as spon- Neither perceivers’ level of social anxiety nor expres- taneous expressions of happiness, joy or contentment sers’ gender were related to the evaluations of the (Martin et al., 2021; Orlowska et al., 2018; Rychlowska smiles. In implicit evaluations, none of the expected et al., 2017). Moreover, they were found to buffer per- differences emerged. ceivers’ stress responses (Martin et al., 2018). The posi- The finding that all three smiles were explicitly tive explicit valence ratings found in the current study evaluated as positive is partly in line with theory further support its rewarding function to perceivers. and research. First, reward smiles received the most Second, the generally positive evaluations of dom- positive evaluations, which is in line with both its inance smiles were unexpected and in line neither 542 A. L. DAPPRICH ET AL. with theory nor with previous findings. Previous were associated with decreased intensity ratings of research found that pictures and video clips of reward smiles, dominance smiles and angry actors showing an asymmetric activation of the zygo- expressions (decreased positivity in the case of maticus major, a wrinkled nose and a raised upper lip smiles, and decreased negativity in the case of were clearly recognised as signals of social domi- anger). Previous research showed that these nance, disapproval or contempt (Martin et al., 2021; expressions were more easily recognised than other Orlowska et al., 2018; Rychlowska et al., 2017). More- expressions (e.g. Montagne et al., 2007; Rychlowska over, they increased stress responses in perceivers et al., 2017). Perhaps an easier recognition also (Martin et al., 2018), underlining a negative meaning enabled participants with higher CU-traits to be less for perceivers. In the current study, dominance affected by the valence of the respective expression. smiles were less positively evaluated than reward Since CU-traits are characterised by unemotionality, smiles and received the least positive ratings of the carelessness and uncaringness (Frick, 2004), the three smile types, suggesting that participants per- diminished ability to experience affect might be ceived the dominance smile as the most negative related to diminished ratings of another person’s smile – but still generally positive. If no context infor- affect. Similarly, male participants with higher CU- mation is available, people might fall back on the pro- traits evaluated positive video clips as less positive totypical assumption that smiles are rather positive. and negative clips as less negative (Fanti et al., Yet, the circumstances under which the dominance 2016). This flattened sensitivity could also explain smile has a negative or positive meaning require why CU-traits are associated with aggressive, rule- further investigation. breaking behaviour. Consistent with our predictions, Finally, positive evaluations of the affiliation smile social anxiety was not found to be associated with are in line with its theoretical meaning, though only explicit evaluations of either smiles or angry partly in line with previous research. Expressions of expressions. This finding has been explained by the the affiliation smile, i.e. symmetric activations of the notion that biased implicit processes rather than zygomaticus major and a lowered upper lip, were biased explicit evaluations maintain the strong behav- less easily interpreted as intending to improve social ioural avoidance tendencies observed in individuals relationships (Orlowska et al., 2018; Rychlowska with social anxiety (Heuer et al., 2007). However, the et al., 2017), and rather labelled as faked (Martin current attempt to study implicit evaluations of the et al., 2021). A positive, prosocial function of affiliation three smiles was unfortunately inconsistent with our smiles could only be detected when taking context predictions. and perceiver characteristics into account (Martin Unexpectedly, implicit evaluations of the smile et al., 2018; Rychlowska et al., 2021). The current types did not differ from each other, and neither results also suggest that participants perceived social anxiety nor psychopathy played a role in this stand-alone affiliation smiles neither as clearly posi- hypothesised link. The Approach-Avoidance Task tive, nor as clearly negative: Their explicit valence (AAT) has often been used to measure implicit evalu- ratings were lower than those of reward smiles, but ations (for a meta-analysis, see Phaf et al., 2014) and could not be distinguished from those of dominance biased automatic action tendencies in response to smiles. Similarly, the range of explicit evaluations facial expressions have been replicated for both ascribed to the affiliation smiles (range = −7 – 75), as social anxiety (Heuer et al., 2007; Lange et al., 2008; well as the low Cronbach’s alpha values (α =.67) Roelofs et al., 2010) and psychopathic traits (Dapprich suggest that the meaning of the affiliation smile et al., 2021; von Borries et al., 2012). However, in the may be ambiguous, varying across specific stimuli current study, AAT response tendencies had very and perceivers. However, different to previous low reliabilities for the whole task and for the research (Martin et al., 2018), we did not find that per- different facial expressions. It might be that we had ceiver characteristics related to different responses too few actors with too much variability for the towards affiliation smiles, but to different responses same facial expressions, which did not consistently towards both reward and dominance smiles. evoke the same tendency to approach or avoid. As Our findings with respect to individual differences the responses to our stimuli were not internally con- in the explicit evaluations of the three smiles were sistent, a larger variety of actors and more validation only partially consistent with our predictions. In con- studies of the three smiles are needed. Not only auto- trast to our predictions, increasing levels of CU-traits matic action tendencies showed an unexpected  COGNITION AND EMOTION 543 pattern, but also the link between self-reported social psychopathy. The present results suggest that all anxiety and psychopathic tendencies disconfirmed smiles are evaluated as positive. However, further vali- our hypotheses. dation of the three smiles using multimethod Against our expectations and some previous approaches seems desirable. A more comprehensive research (Dapprich et al., 2021; Hofmann et al., understanding of the influence of perceivers’ charac- 2009), social anxiety and psychopathic tendencies teristics on the interpretation of the meaning of correlated significantly positively with each other. facial expressions may inform us about the possible We found that participants who reported more fear behavioural consequences of such expressions. in social situations also reported more psychopathic tendencies, such as shallow affect, lacking prosocial Notes emotions and impulsive behaviour. However, to be more precise, social anxiety was only significantly 1. In addition to the measures mentioned here, a Motivated positively associated with the total score and with Viewing Task which did not include pictures of the three smiles, and the State Self-Esteem Questionnaire Factor II of the Levenson Self-Report Psychopathy (Heatherton & Polivy, 1991) were administered. Scale, but not with Factor I or CU-traits. These 2. Based on an a-priori power analysis, we aimed to test 120 findings might indicate that different subtypes of participants. We powered for a small-to-medium corre- (social) anxiety and psychopathy are differentially lation between an interindividual difference variable (e.g. social anxiety) and AAT difference scores (r=.25, linked to each other (for a review, see Derefinko, two-tailed), with p=.05 and 1−ß=.80. 2015). Indeed, Dapprich et al. (2021) as well as Hofmann et al. (2009) used other questionnaires than we did here. Thus, it will be necessary to test Acknowledgements whether different operationalisation of social anxiety We would like to thank our students for helping to conduct the and psychopathy relate differently to each other and experiment, as well as our participants for participating in it. to other concepts. Finally, the helpful suggestions by the editor and two anon- Several limitations should be considered when ymous reviewers are gratefully acknowledged. interpreting the present results. First, the current study used static pictures of facial expressions, Disclosure statement which lack context information and have lower eco- logical validity. Previous research already stressed No potential conflict of interest was reported by the author(s). the importance of using a multimethod approach in order to study the three smiles, as well as their Funding social motives and their effects on perceivers (Martin This study was supported by the Behavioural Science Institute et al., 2018, 2021; Orlowska et al., 2018; Rychlowska (BSI) of Radboud University, as well as the Department of Psy- et al., 2021). For instance, a variety of different chology and the Gonda Brain Science Center of Bar-Ilan measures has been used, including physiological University. measures, economic decisions, facial mimicry and explicit rating tasks. These studies revealed that realis- ORCID tic presentations of the smiles and context infor- mation are crucial. Thus, examining valence ratings Anna L. Dapprich http://orcid.org/0000-0002-2208-698X with paradigms of higher ecological validity would be desirable. Furthermore, given the partly low References internal consistencies of our measures, future research needs to include additional stimuli. Finally, Bürkner, P. (2017). Brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 80(1), 1– as we only tested healthy female participants, the 28. https://doi.org/10.18637/jss.v080.i01 effects cannot be generalised to males or clinical Cabral, J. C. C., Tavares, P. d. S., & de Almeida, R. M. M. (2016). populations. Future research should examine Reciprocal effects between dominance and anger: A systema- whether the same effects occur in other samples. tic review. Neuroscience & Biobehavioral Reviews, 71, 761–771. The current study was one of the first to examine https://doi.org/10.1016/j.neubiorev.2016.10.021 Cleckley, H. M. (1941). The mask of sanity: An attempt to reinter- the evaluation of three theoretically central types of pret the so-called psychopathic personality. Mosby. smiles in relation to two psychopathologies that Dapprich, A. L., Lange, W.-G., von Borries, A. K. L., Volman, I., interfere with social behaviour – social anxiety and Figner, B., & Roelofs, K. (2021). The role of psychopathic 544 A. L. DAPPRICH ET AL. traits, social anxiety and cortisol in social approach avoidance Levenson, M. R., Kiehl, K. A., & Fitzpatrick, C. M. (1995). tendencies. Psychoneuroendocrinology, 128, 105207. https:// Assessing psychopathic attributes in a noninstitutiona- doi.org/10.1016/j.psyneuen.2021.105207 lized population. Journal of Personality and Social Derefinko, K. J. (2015). Psychopathy and low anxiety: Meta-ana- Psychology, 68(1), 151–158. https://doi.org/10.1037/0022- lytic evidence for the absence of inhibition, not affect. Journal 3514.68.1.151 of Personality, 83(6), 693–709. https://doi.org/10.1111/jopy. Liebowitz, M. R. (1987). Social phobia. Modern Problems in 12124 Pharmacopsychiatry, 22, 141–173. doi:10.1159/000414022 Eisenbarth, H., Alpers, G. W., Segrè, D., Calogero, A., & Angrilli, A. Lischke, A., Lemke, D., Neubert, J., Hamm, A. O., & Lotze, M. (2008). Categorization and evaluation of emotional faces in (2017). Inter-individual differences in heart rate variability psychopathic women. Psychiatry Research, 159(1–2), 189– are associated with inter-individual differences in mind- 195. https://doi.org/10.1016/j.psychres.2007.09.001 reading. Scientific Reports, 7(1), 11557. https://doi.org/10. Fanti, K. A., Panayiotou, G., Lombardo, M. V., & Kyranides, M. N. 1038/s41598-017-11290-1 (2016). Unemotional on all counts: Evidence of reduced Lykken, D. T. (1957). A study of anxiety in the sociopathic person- affective responses in individuals with high callous-unemo- ality. The Journal of Abnormal and Social Psychology, 55(1), 6– tional traits across emotion systems and valences. Social 10. https://doi.org/10.1037/h0047232 Neuroscience, 11(1), 72–87. https://doi.org/10.1080/174709 Martin, J. D., Abercrombie, H. C., Gilboa-Schechtman, E., & 19.2015.1034378 Niedenthal, P. M. (2018). Functionally distinct smiles elicit Frick, P. J. (2004). Inventory of Callous-Unemotional Traits. different physiological responses in an evaluative context. Hare, R. D., & Neumann, C. S. (2008). Psychopathy as a clinical Scientific Reports, 8(3558), 1–8. https://doi.org/10.1038/ and empirical construct. Annual Review of Clinical s41598-018-21536-1 Psychology, 4(1), 217–246. https://doi.org/10.1146/annurev. Martin, J. D., Rychlowska, M., Wood, A., & Niedenthal, P. M. clinpsy.3.022806.091452 (2017). Smiles as multipurpose social signals. Trends in Heatherton, T. F., & Polivy, J.. (1991). Development and vali- Cognitive Sciences, 21(11), 864–877. https://doi.org/10.1016/ dation of a scale for measuring state self-esteem. Journal of j.tics.2017.08.007 Personality and Social Psychology, 60(6), 895–910. http://dx. Martin, J. D., Wood, A., Cox, W. T. L., Sievert, S., Nowak, R., Gilboa- doi.org/10.1037/0022-3514.60.6.895 Schechtman, E., Zhao, F., Witkower, Z., Langbehn, A. T., & Heimberg, R. G., Horner, K. J., Juster, H. R., Safren, S. A., Brown, E. Niedenthal, P. M. (2021). Evidence for distinct facial signals J., Schneier, F. R., & Liebowitz, M. R. (1999). Psychometric of reward, affiliation, and dominance from both perception properties of the Liebowitz Social Anxiety Scale. and production tasks. Affective Science, 2(1), 14–30. https:// Psychological Medicine, 29(1), 199–212. https://doi.org/10. doi.org/10.1007/s42761-020-00024-8 1017/S0033291798007879 Montagne, B., Kessels, R. P. C., De Haan, E. H. F., & Perrett, D. I. Hess, U., Adams, R. B. J., & Kleck, R. E. (2005). Who may frown and (2007). The emotion recognition task: A paradigm to who should smile? Dominance, affiliation, and the display of measure the perception of facial emotional expressions at happiness and anger. Cognition & Emotion, 19(4), 515–536. different intensities. Perceptual and Motor Skills, 104(2), 589– https://doi.org/10.1080/02699930441000364 598. https://doi.org/10.2466/pms.104.2.589-598 Heuer, K., Rinck, M., & Becker, E. S. (2007). Avoidance of Niedenthal, P. M., Mermillod, M., Maringer, M., & Hess, U. (2010). emotional facial expressions in social anxiety: The The simulation of smiles (SIMS) model: Embodied simulation approach-avoidance task. Behaviour Research and and the meaning of facial expression. Behavioral and Therapy, 45(12), 2990–3001. https://doi.org/10.1016/j.brat. Brain Sciences, 33(6), 417–433. https://doi.org/10.1017/ 2007.08.010 S0140525X10000865 Hoffmann, H., Kessler, H., Eppel, T., Rukavina, S., & Traue, H. C. Orlowska, A. B., Krumhuber, E. G., Rychlowska, M., & Szarota, P. (2010). Expression intensity, gender and facial emotion rec- (2018). Dynamics matter: Recognition of reward, affiliative, ognition: Women recognize only subtle facial emotions and dominance smiles from dynamic vs. static displays. better than men. Acta Psychologica, 135(3), 278–283. Frontiers in Psychology, 9(938), 1–8. https://doi.org/10.3389/ https://doi.org/10.1016/j.actpsy.2010.07.012 fpsyg.2018.00938 Hofmann, S. G., Korte, K. J., & Suvak, M. K. (2009). The upside of Phaf, R. H., Mohr, S. E., Rotteveel, M., & Wicherts, J. M. (2014). being socially anxious: Psychopathic attributes and social Approach, avoidance, and affect: A meta-analysis of anxiety are negatively associated. Journal of Social and approach-avoidance tendencies in manual reaction time Clinical Psychology, 28(6), 714–727. https://doi.org/10.1521/ tasks. Frontiers in Psychology, 5(Article 378), 1–16. https:// jscp.2009.28.6.714 doi.org/10.3389/fpsyg.2014.00378 Keltner, D., & Haidt, J. (1999). Social functions of emotions at four Quintana, D. S., Guastella, A. J., Outhred, T., Hickie, I. B., & Kemp, levels of analysis. Cognition & Emotion, 13(5), 505–521. A. H. (2012). Heart rate variability is associated with emotion https://doi.org/10.1080/026999399379168 recognition: Direct evidence for a relationship between the Lange, W.-G., Keijsers, G., Becker, E. S., & Rinck, M. (2008). Social autonomic nervous system and social cognition. anxiety and evaluation of social crowds: Explicit and implicit International Journal of Psychophysiology, 86(2), 168–172. measures. Behaviour Research and Therapy, 46(8), 932–943. https://doi.org/10.1016/j.ijpsycho.2012.08.012 https://doi.org/10.1016/j.brat.2008.04.008 Rapee, R. M., & Heimberg, R. G. (1997). A cognitive-behavioral Lenth, R. (2020). emmeans: Estimated Marginal Means, aka Least- model of anxiety in social phobia. Behaviour Research and Squares Means (R package version 1.4.6). https://cran.r- Therapy, 35(8), 741–756. https://doi.org/10.1016/S0005-7967 project.org/package=emmeans (97)00022-3  COGNITION AND EMOTION 545 R Core Team. (2019). R: A language and environment for statistical Rychlowska, M., Jack, R. E., Garrod, O. G. B., Schyns, P. G., Martin, computing. R Organisation for Statistical Computing. https:// J. D., & Niedenthal, P. M. (2017). Functional smiles: Tools for www.r-project.org/ love, sympathy, and war. Psychological Science, 28(9), 1259– Revelle, W. (2018). psych: Procedures for personality and 1270. https://doi.org/10.1177/0956797617706082 psychological research (R package version 1.8.12). Rychlowska, M., van der Schalk, J., Niedenthal, P., Martin, J., Northwestern University. https://cran.r-project.org/ Carpenter, S. M., & Manstead, A. S. R. (2021). Dominance, package=psych reward, and affiliation smiles modulate the meaning of unco- Roelofs, K., Putman, P., Schouten, S., Lange, W. G., Volman, operative or untrustworthy behaviour. Cognition and I., & Rinck, M. (2010). Gaze direction differentially affects Emotion, 35(7), 1281–1301. https://doi.org/10.1080/ avoidance tendencies to happy and angry faces in 02699931.2021.1948391 socially anxious individuals. Behaviour Research and von Borries, A. K. L., Volman, I., de Bruijn, E. R. A., Bulten, B. H., Therapy, 48(4), 290–294. https://doi.org/10.1016/j.brat. Verkes, R. J., & Roelofs, K. (2012). Psychopaths lack the auto- 2009.11.008 matic avoidance of social threat: Relation to instrumental RStudio Team. (2019). RStudio: Integrated development environ- aggression. Psychiatry Research, 200(2–3), 761–766. https:// ment for R. doi.org/10.1016/j.psychres.2012.06.026