Moral inferences from androgynous faces are beyond categorical uncertainty : Evidence of a positive bias towards androgynous targets

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Moral inferences from androgynous faces are beyond categorical uncertainty : Evidence of a positive bias towards androgynous targets © 2024 the Authors Published version Ansani, Alessandro; Olivera‐La Rosa, Antonio Ansani, A., & Olivera‐La Rosa, A. (2024). Moral inferences from androgynous faces are beyond categorical uncertainty : Evidence of a positive bias towards androgynous targets. European Journal of Social Psychology, Early View. https://doi.org/10.1002/ejsp.3048 2024 Received: 22 June 2023 Accepted: 29 January 2024 DOI: 10.1002/ejsp.3048 RESEARCH ARTICLE Moral inferences from androgynous faces are beyond categorical uncertainty: Evidence of a positive bias towards androgynous targets Alessandro Ansani1,2Antonio Olivera-La Rosa3,4 1Department of Music, Art and Culture Studies, University of Jyväskylä, Jyväskylä, Finland 2Department of Philosophy, Communication, and Performing Arts, Roma Tre University, Rome, Italy 3Department of Psychology and Social Sciences, Universidad Católica Luis Amigó, Medellín, Colombia 4Human Evolution and Cognition Group, University of the Balearic Islands, Palma, Spain Correspondence Antonio Olivera-La Rosa, Department of Psychology and Social Sciences, Universidad Católica Luis Amigó, Medellín, Colombia. Email: antonio.oliver[email protected] Abstract Postulating a negative bias towards social ambiguity, we conducted cross-cultural online research to assess whether categorical discrepancies in the perception of androgynous faces were associated with the uncanny feeling and inferences of different morality. Across four studies, we found that androgynous faces were harder to classify into a binary sex category than sex-typical faces, but this difficulty did not influence social judgements of androgynous targets in a negative fashion. In Study 1 (Spanish-speaking sample, N=76), we found that androgynous faces were rated as more trustworthy, less creepy, and less morally different than sex-typical faces. Study 2 replicated most of the findings from Study 1 in an Italian sample (N=45). Positive bias towards androgyny was not replicated with a different set of stimuli featuring faces of diverse ethnic backgrounds (Study 3, Spanish-speaking sample, N=140). However, results revealed a main effect of ethnicity in participants’ responses. When controlling for the effect of morphing procedures in stimuli selection, an overall positive bias towards androgynous targets arose, especially when compared to masculine targets (Study 4, Spanish-speaking sample, N=85). These findings suggest that, at least in certain conditions, a positive social bias towards androgynous faces may emerge that does not depend on categorical uncertainty and facial attractiveness. KEYWORDS androgyny, creepiness, morality, social perception, trustworthiness 1INTRODUCTION The study of the cognitive basis of prejudice can contribute to a more comprehensive study and reduction of stigmatizing behaviours. Recently, Olivera-La Rosa, Chica-Franco & Ingram (2023a) suggested that the uncanny valley hypothesis (Mori et al., 1970/2005) constitutes an insightful framework to account for the negative bias towards physically androgynous individuals. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2024 The Authors. European Journal of Social Psychology published by John Wiley & Sons Ltd. According to this hypothesis, human-like stimuli that closely resemble ‘real’ humans but still bear a slightly artificial quality can elicit a negative emotional response in an observer. Various hypotheses have been proposed to elucidate the psychological mechanisms underlying the uncanny response (Kätsyri et al., 2015; Wang et al., 2015). One of the most pertinent is the categorical uncertainty hypothesis, which posits that ambiguity in categorizing extremely realistic artificial characters can trigger uneasy feelings in an observer (Burleigh et al., 2013; Eur J Soc Psychol. 2024;1–16. wileyonlinelibrary.com/journal/ejsp 1 2ANSANI and OLIVERA-LA ROSA Wang & Rochat, 2017). A proclivity towards categorical thinking is consistent with the broader goal of cognitive economy: cognitive processes are oriented to use minimal mental effort to process the most useful information (Allport, 1954; Kahneman, 2011). The categorical uncertainty hypothesis is consistent with the hedonic fluency model: information that is easily processed is often accompanied by a positive affective response, which can impact preference evaluations in a congruent manner; (Reber & Schwarz, 2006; Winkielman et al., 2003). Some studies showed that stimuli that defy easy categorization (beyond the natural–artificial distinction) can generate a negative affective response (Ferrey et al., 2015; Yamada et al., 2013). Notably, the uncanny response is not confined solely to artificial entities: it has been associated with clowns, people deemed unattractive, or even ‘Botox victims’ (Olivera-La Rosa, Villacampa, Corradi, & Ingram 2023b; Pollick, 2010;Smith,2014). Human entities appraised as unnatural and/or emotionally unpredictable may trigger the uncanny feeling, which may function as a heuristic mechanism of moral inferences (‘if it is physically disturbing, it is morally disturbing’; Olivera-La Rosa, 2018). For instance, uncanny faces were rated as less socially desirable and more morally different than aesthetically neutral faces (Olivera-La Rosa et al., 2023b). Do androgynous faces follow a similar pattern? Androgyny can introduce challenges in sexual categorization because it represents an ambiguous state between the sexual categories of man and woman. Indeed, despite the rising prominence of political movements and support groups advocating for sex and gender diversity, state-of-the-art research suggests that sex perception from an early age is still categorical and binary (Campanella et al., 2001; Freeman et al., 2010; Martin & Slepian, 2021; Quinn et al., 2002). Some evidence suggests that gender is a critical attribute of perceiving humanness and that removing gender from human targets causes dehumanization (Martin & Mason, 2022). Therefore, difficulty in categorizing androgynous faces into a binary category may signal ‘unnaturalness’ and social unpredictability. The integration of these assessments with the negative emotional response stemming from processing disfluency can potentially trigger the uncanny feeling and the associated deductions of social exclusion (Olivera-La Rosa et al., 2023a). Built on these insights, in this research, we assess whether categorical uncertainty in the perception of androgynous faces is related to appraisals of uncanniness and perception of moral traits. 1.1 First impressions from faces: The primacy of moral inferences in person perception Traditionally, research on how people form first impressions of unfamiliar targets has focused on warmth (i.e., a social dimension that revolves around connection and community) and competence (i.e., a cognitive dimension that revolves around agency and goal pursuit) as the two fundamental dimensions driving person perception (Fiske et al., 2007; Martin & Slepian, 2021). However, some studies suggest that two-dimensional models of impression formation do not fully capture the importance of moral information in overall impressions of social targets. Goodwin and colleagues (2014) showed that, despite warmth and morality being ‘social’ dimensions, both dimensions diverge (e.g., a person can be charming but of dubious moral character). Further, ‘pure’ moral traits (e.g., honesty) were shown to be the strongest predictors of impressions, suggesting that perceptions of moral character may be the most important source of social information in impression formation (Goodwin, 2015). The primacy of moral character in person perception may reflect both social-functionalist (e.g., assessing intentions) and symbolic considerations related to what it means to be human (Goodwin, 2015; Olivera-La Rosa, 2018; Strohminger & Nichols, 2014). Indeed, inferring moral information from unfamiliar targets is a ubiquitous practice and often results from evaluations based on whatever information is available. For instance, moral behaviours (Uhlmann et al., 2015), affective reactions (Olivera-La Rosa et al., 2021; Kastendieck et al., 2021) and uncanny faces (Olivera-La Rosa et al., 2023b) have been revealed as sources of moral inferences. The latest finding is consistent with decades of research on face perception, showing that people spontaneously draw several first impressions from faces, a process that can be characterized as mainly automatic (Bar et al., 2006; Engell et al., 2007). Some authors claim that initial social perception should be understood as a dynamic negotiation between bottom-up visual features inherent to a target (e.g., facial traits) and social cognitive factors that perceivers bring to the perceptual process (e.g., stereotypes and attitudes; Freeman & Ambady, 2011; Freeman et al., 2020). This model suggests that, in those cases when facial information is particularly ambiguous (e.g., androgynous facial features), social-conceptual knowledge may influence representational competition one way or another (e.g., man or woman). For instance, perceiving a face as real (instead of artificial) can impact perceived trustworthiness (Tucciarelli et al., 2022), which seems consistent with the influence of the uncanny feeling in social judgements (Olivera-La Rosa, 2018). From an evolutionary standpoint, a quick and accurate evaluation of another’s intentions can assist people to engage in the proper behavioural response (e.g., approach vs. avoid) (Atkinson & Adolphs, 2005). Negative biases towards ambiguous social interactions have been explained as social perception mechanisms prioritizing caution to reduce false-positive errors at the cost of increasing false-negative errors (Schaller & Park, 2011). Trustworthiness is a moral trait that has been of special interest in previous face perception research, playing a crucial role in automatic first impressions of unfamiliar faces (Oosterhof & Todorov, 2008; Todorov et al., 2009). Indeed, social traits critical to survival, such as trustworthiness, are more likely to be automatically inferred from faces (Engell et al., 2007). Judgements of trustworthiness approximate the global evaluation of faces (Oosterhof & Todorov, 2008), which is consistent with the primacy of moral information in impression formation (Goodwin, 2015). For instance, unfamiliar faces elicit judgements of trustworthiness or untrustworthiness in as little as 33–100 ms (Todorov et al., 2009; Willis & Todorov, 2006). Untrustworthy faces elicited increased amygdala activity (Said et al., 2010),whichisrelated to basic affective evaluations (e.g., fear). Perception of trustworthiness can be influenced by face typicality, with locally typical faces being judged as more trustworthy than faces typical of other places (Sofer 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License SOCIAL PERCEPTION AND ANDROGYNOUS TARGETS 3 et al., 2015, 2017). Based on this research, it can be argued other types of atypical faces (e.g., androgynous faces) may be more likely to be perceived as untrustworthy. Some evidence showed that facial attractiveness impressions precede and influence the perception of trustworthiness (Gutiérrez- Garcia et al., 2019). Indeed, facial attractiveness positively influences social judgements in a plethora of fields (Langlois et al., 2000); however, unattractive faces often provoke sounder effects on social judgements than attractive faces. This negative bias has been explained as due to unattractive faces being perceived as more ambiguous and, as a consequence, harder to categorize (Griffin & Langlois, 2006). Appraisals of unpredictability and ambiguous social threats are associated with the creepiness response (i.e., an unpleasant and confusing psychological reaction), which is mainly triggered by facial cues (McAndrew & Koehnke, 2016; Watt et al., 2017). According to these authors, ambiguity in social information derived from the face is involved in this negative response. Altogether, these findings suggest that face evaluation may be related to heuristic mechanisms for inferring moral information (e.g., harmful intentions) (Olivera-La Rosa, 2018; Oosterhof & Todorov, 2008). 1.1.1 Social perception of androgyny Androgynous individuals possess similar levels of feminine and masculine traits (Kark, 2020); that is, they exhibit aesthetic and/or behavioural characteristics of an ambiguous nature. It is worth mentioning that the current research focuses on the perception of androgynous faces (i.e., physical androgyny via morphing of sex-typical female and male faces) and not on the sexuality of androgynous individuals (for a similar approach; see Atwood & Axt, 2021). Previous research suggests the existence of a negative social bias towards androgyny. Stern and Rule (2018) demonstrated that negative attitudes towards physically androgynous transgender individuals can be attributed to participants’ difficulty in determining their sex. The authors concluded that the struggle in sorting androgynous-looking faces into a binary category represents a metacognitive variable capable of influencing negative social judgements towards them (see also Owen et al., 2016). This finding is consistent with research by Lick and colleagues, showing that metacognitive processes underlying categorization influence social judgements and interpersonal prejudice (Lick & Johnson, 2013; Lick et al., 2015). For instance, gender-atypical faces were associated with bisexual categorizations and were rated more negatively, partially by disfluent processing in the early stages of social perception (Lick et al., 2015). Such a negative bias against androgynous people has also been shown by implicit procedures (Atwood & Axt, 2021; Axt et al., 2021). 2THE PRESENT RESEARCH Although these findings are certainly insightful, the impact of categorical difficulty in moral inferences (besides attitudes) of androgynous faces requires further research. Our research aims to fill this gap by assessing the moral perception of androgynous targets from the standpoint of the uncanny valley framework, which connects metacognitive processes underlying categorization with moral inferences (Olivera-La Rosa, 2018). We believe that the centrality of moral information in person perception justifies the dependent variables chosen in this research. For instance, the perception of moral character may be more informative of moral judgements than that of the permissibility of acts (Uhlmann et al., 2015). Moreover, previous studies on negative biases towards androgynous faces were largely based on United States citizens (Atwood & Axt, 2021; Axt et al., 2021; Stern & Rule, 2018), which restricts the generality of these findings. Therefore, we assessed whether categorical difficulty in the perception of androgynous faces is related to judgements of untrustworthiness, appraisals of creepiness (uncanniness) and judgements of a different moral stance (i.e., ‘the target person does not share my moral values’) in two samples that were underrepresented in previous research: Spanishspeaking and Italian samples. Based on these findings, we predicted that H1: Androgynous faces should be rated as creepier, more untrustworthy, and more morally different than sex-typical faces. H2: A negative bias towards androgynous faces could be caused by the fact that the participants would have a harder time resolving androgynous faces’ biological sex, thus leading to an increase in the reaction times (RTs). We tested these hypotheses in four online experiments conducted with Spanish-speaking and Italian participants. In Study 1, we assessed the role of categorization difficulty in moral perception of androgynous targets in a Spanish-speaking population. Study 2 was conducted to extend the generalizability of Study 1 results by applying the same experimental design in a different cultural context (Italian sample). Further, Study 2 explores whether potential differences in stimulus familiarity resulting from the particularities of the experimental procedure play a role in the obtained results. Study 3 incorporated a different set of face stimuli (previously used by Atwood & Axt, 2021) varying in ethnicity to assess the role of stimuli selection in the results of Studies 1 and 2. Finally, Study 4 used morphed faces both for androgynous and sex-typical faces to control the potential influence of features of the face stimuli (e.g., realness/attractiveness related to morphing procedures) in our findings. 3STUDY 1 3.1 Introduction The main aim of this online study was to investigate whether categorization difficulty played an important role in influencing social judgements of androgynous targets in a Spanish-speaking sample. We adopted three social judgement variables – trustworthiness, creepiness and shared moral values – to assess the social perception 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 4ANSANI and OLIVERA-LA ROSA of androgyny. Please note that the concept of ‘uncanniness’ has encountered translation challenges. The initial term, Bukimi No Tani, was initially translated as ‘strangeness’ (Mori, 1970/2005)andsubsequently substituted with expressions like ‘creepiness’ or ‘eeriness’ (Wang et al., 2015). 3.2 Methods 3.2.1 Participants We recruited 116 participants via internal email and social networking. Most participants were from Colombia (73.7%), Peru (21.1%), or Spain (2.6%). Before the experiment, all participants affirmed their consent to participate in the study and provided written consent in accordance with ethical procedures approved by the Bioethics Committee of Universidad Católica Luis Amigó, Medellín, Colombia. They then indicated their age, biological sex, sexual preferences, and country of residence. 3.2.2 Exclusion criteria Two different criteria were adopted to clean the data before the analyses. First, we excluded the participants who did not complete the whole experiment (N=35). Second, we applied the Mahalanobis’ criterion for identifying multivariate outliers (Masnan et al., 2015)toexclude the participants whose response times for masculine, feminine and androgynous faces exhibited a considerable distance from the centroid (N=5). After such exclusions, the final sample was 76 participants (49 females, Mage =33.12, SD =13.42). The task was completed in a median time of 17.28 min. 3.2.3 Face evaluation task and stimuli creation We assessed the relationship between categorization difficulty and social perception of sex-typical versus androgynous faces using a modified version of the face evaluation task. Following Stern and Rule’s (2018) experimental design, the task consisted of two parts. First, participants categorized target faces in a random order as either male or female faces by pressing the A and L keys, respectively. Consistent with previous research (Owen et al., 2016; Stern & Rule, 2018), response latency was included as a measure for ease (or difficulty) of the categorization difficulty of the target. Concerning the stimuli, we selected 15 faces (five androgynous, five feminine, and five masculine faces) after two pilot studies directed to control for sex typicality and attractiveness (see Supporting Information). Androgynous faces were created by morphing male and female faces (i.e., 50% Female–50% Male faces) from DeBruine and Jones (2017). Masculine and feminine faces were selected from the same database. 3.2.4 Social judgements We measured social judgements towards androgynous faces through three independent scales. Thus, participants then viewed the same target faces again and rated each face on the dimensions of trustworthiness, creepiness/uncanniness and morality. We randomized the order of the faces for this task, along with the order of presentation of the three social judgement scales. We measured trust by asking participants to indicate the perceived trustworthiness of the target on a 7-point scale ranging from ‘not at all trustworthy’ to ‘completely trustworthy’. The question was framed as follows: ‘Based on your intuition, how trustworthy does this person seem to you?’. Higher ratings correspond to higher trustworthiness judgements. Participants rated the perceived creepiness by indicating how creepy they perceived the target person to be on a 7-point scale from ‘not creepy at all’ to ‘extremely creepy’. We used 7-point Likert scales to measure trustworthiness and creepiness because we aimed at building a catchy task that was feasible in a relatively short amount of time. All the studies are <18 min and this can be considered satisfactory; especially considering the works on the scarce reliability of long (>20 min) online studies (Revilla & Höhne, 2020; Revilla & Ochoa, 2017). Moreover, trustworthiness and creepiness were measured by using a 7-point scale in previous online research on face perception (Olivera La Rosa et al., 2020; Olivera-La Rosa et al., 2021). Participants responded to the shared moral values scale (Szczurek et al., 2012), by indicating how much they felt the target face shared their moral values (‘I feel the target individual shares most of the same moral values as me’), using a 6-point scale ranging from 1 (strongly disagree)to6(strongly agree). The shared moral values scale was used as a 6-point scale as in previous research (Olivera-La Rosa et al., 2021; Szczurek et al., 2012). 3.3 Results 3.3.1 Statistical notes In all the analyses, a mixed approach was adopted. Jamovi (The Jamovi project, 2022) was used as the analytic software. In greater detail, the Generalized Linear Mixed-effect Models (GLMMs; Stroup, 2013; Faraway, 2006) were run on GAMLj (Gallucci, 2019). This procedure has already been proven to be effective and satisfactory in analysing reaction times data in psychological studies for several reasons: firstly, GLMMs do not average across individual responses; secondly, they can handle data that do not meet the normality assumption. Lastly, such a procedure allows differences between individuals to be measured appropriately, especially with skewed data (Lo & Andrews, 2015). 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License SOCIAL PERCEPTION AND ANDROGYNOUS TARGETS 5 TABLE 1 Descriptive statistics of all studies. EXP Faces’ appearance Androgynous Feminine Masculine d MSEMSEMSE RT EXP1 2.40 0.21 1.60 0.12 1.43 0.12 0.62 EXP2 1.75 0.19 1.24 0.12 0.98 0.09 0.70 EXP3 3.07 0.19 1.87 0.14 1.87 0.15 0.48 EXP4 2.34 0.19 1.55 0.11 1.23 0.08 0.42 Trustworthiness EXP1 3.83 0.32 3.20 0.23 2.94 0.22 0.16 EXP2 4.24 0.35 3.59 0.27 2.82 0.23 0.29 EXP3 3.35 0.18 3.00 0.20 3.15 0.20 0.07 EXP4 3.38 0.25 3.42 0.22 2.83 0.19 0.13 Creepiness EXP1 1.96 0.14 2.61 0.20 2.87 0.23 0.28 EXP2 1.73 0.15 2.43 0.23 2.74 0.28 0.42 EXP3 2.26 0.13 2.20 0.16 2.29 0.16 0.02 EXP4 2.55 0.18 2.93 0.20 3.25 0.23 0.16 Shared Moral Values EXP1 3.01 0.24 2.41 0.19 2.37 0.18 0.21 EXP2 3.20 0.26 2.88 0.23 2.61 0.19 0.16 EXP3 3.00 0.16 2.74 0.18 2.84 0.18 0.07 EXP4 3.03 0.19 2.87 0.18 2.52 0.17 0.12 To control for the participants’ and stimuli’ variability, within the GLMMs, the participants and stimuli were modelled as random intercepts (Judd et al., 2012). Moreover, the effect of the face’s appearance (i.e., masculine, feminine, or androgynous) was modelled as a random slope within each participant; thus implying that the effect of the face’s appearance could differ from participant to participant. Therefore, the following formula was used: DV (Trustworthiness, Creepiness or Shared Moral values) =Faces’ appearance +Participant sex +(1 | Stimulus) +(1 +Faces’ appearance | Participant) In all models, gamma distribution was used with an inverse link function, except for the response times model, where we used a gamma distribution with a log link function (for a rationale about distributions and link functions; see Ng & Cribbie, 2017; Yousefi et al., 2015). Moreover, we excluded the sex of the participants from the RT models as we did not expect any difference between male and female participants in RTs. Generalized Logistic Mixed Models with the same formula were used with the sex chosen by the participants for each face as the model-dependent variable. This was done to assess the probability of androgynous faces being categorized as females. In all models, we estimated the parameters of masculine and feminine faces contrasting them with the androgynous faces. The significance of the only missing comparison (i.e., masculine vs. feminine) was investigated through a planned pairwise contrast. In the results of the GLMMs, the reader will find the description of the main effects in terms of χ2, degrees of freedom (df), p-value and effect size in terms of Cohen’s d.1 For each estimate, a description is provided of the estimated coefficient (β), standard error (SE), level of statistical significance (p-value), and 95% bias-corrected bootstrap confidence interval (95% CI). All the descriptive statistics can be found in Table 1. Furthermore, to enhance the reader’s comprehension of the findings, we have included a Bayesian perspective on the results in the Supporting Information. Specifically, we present the Bayes Factors (BF10) derived from Bayesian paired samples t-tests with all pairwise comparisons. For all tests conducted, we employed JASP (JASP Team, 2023) with weakly informative Cauchy priors (location parameter was centred to µ =0 and the scale parameter was set to σ=0.70). This Bayesian approach offers an alternative lens through which to interpret the data, providing additional insights into the strength of evidence in support of the alternative hypothesis (H1) relative to the null hypothesis (H0). 3.3.2 Response times and face evaluation The model showed that the RTs were significantly affected by the manipulated face’s appearance (χ2=27.8, df =2, p<.001, d=0.62). Considering the androgynous faces as the reference category, the 1Cohen’s dwas derived from the proportion of the variance explained by the predictor (i.e., R2) as per the following formula (Ruscio, 2008; Marini et al., 2024): d=√−4R (R2−1) 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 6ANSANI and OLIVERA-LA ROSA FIGURE 1 Response times as a function of faces’ appearance and experiment (violin plot). The form of the violin indicates the distribution curve. The boxplots within each violin represent interquartile ranges (IQRs). Black vertical lines within the boxplots indicate median values. Black circles within the boxplots indicate mean values. The dashed line indicates the median. In the computation of the median, Study 3 was excluded. model indicated that both the masculine (β=−.50, 95% CI [−0.64, −0.36], SE =0.07, p<.001) and feminine (β=−.39, 95% CI [−0.53, 0.25], SE =0.07, p<.001) faces fostered a significant decrease in RTs, thus showing greater difficulty in categorizing androgynous faces. No difference was found between the RTs of feminine and masculine faces (p=.140) (Figure 1). The probability that an androgynous face was characterized as female was 60.8%, 95% CI [47.6, 72.5].2 3.3.3 Social judgements Trustworthiness had a significant effect (χ2=8.8, df =2, p=.012, d =0.16). The androgynous faces were perceived as more trustworthy than feminine (β=.05, 95% CI [0.002, 0.10], SE =0.02, p=.043) and masculine faces (β=.08, 95% CI [0.02, 0.13], SE =0.02, p=.004). A planned pairwise contrast of the estimated marginal means revealed that the difference between feminine and masculine faces was not significant (p=.319) (Figure 2). The creepiness GLMM led to very similar results (χ2=19.6, df =2, p<.001, d=0.28), but in the opposite direction. Namely, the androgynous faces were rated as significantly less creepy than masculine (β= −.17, 95% CI [−0.25, −0.09], SE =0.04, p<.001) and feminine faces (β 2A Bayesian model comparison approach was taken to make sure that the effect of the manipulated faces’ appearance was not merely due to the sex chosen by the participants. In other words, to make sure that the effects found were not merely due to a gender bias. The results of the comparison of the predictors are in theSupporting Information. =−.12, 95% CI [−0.19, −0.06], SE =0.03, p<.001). The same pairwise contrast found no difference between masculine and feminine faces (p =.177). Female participants reported higher levels of creepiness overall as opposed to male participants (β=.08, 95% CI [0.001, 0.17], SE = 0.04, p=.046) (Figure 2). The shared moral values model reached significance (χ2=12.0, df =2, p=.002, d=0.21). The androgynous faces had higher ratings in Shared Moral Values (SMV) than masculine (β=.09, 95% CI [0.03, 0.14], SE =0.03, p=.003) and feminine faces (β=.08, 95% CI [0.02, 0.14], SE =0.03, p=.005). Feminine and masculine faces did not differ (p=.784) (Figure 3). 3.4 Discussion Our findings indicate that categorizing androgynous faces into a binary sex category is more challenging compared to sex-typical faces. However, this difficulty does not result in unfavourable social judgements towards androgynous targets. Contrary to our expectations, androgynous faces were perceived as less creepy, more trustworthy, and less morally different than sex-typical faces, suggesting a positive social bias towards androgynous faces independent of processing disfluency. Interestingly, our results diverge from prior research on the social perception of androgynous faces (Atwood & Axt, 2021; Axt et al., 2021) and the perception of facial ambiguity (Griffin & Langlois, 2006; Olivera-La Rosa et al., 2023), and also, inconsistent with the hedonic 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License SOCIAL PERCEPTION AND ANDROGYNOUS TARGETS 7 FIGURE 2 Trustworthiness as a function of faces’ appearance and experiment (violin plot). The form of the violin indicates the distribution curve of the predicted values. The boxplots within each violin represent IQRs. Black horizontal lines within the boxplots indicate median values. Black circles within the boxplots indicate mean values. The black dashed line indicates the grand mean. In the computation of the grand mean, Study 3 was excluded. fluency model (Reber & Schwarz, 2006; Winkielman et al., 2003), may reflect another side of social perception of androgynous targets that occurs under certain specific circumstances. Therefore, we have initiated a second study to validate and expand upon our current findings. 4STUDY 2 4.1 Introduction Negative biases towards androgynous faces were found in samples that were mainly composed of United States citizens (Atwood & Axt, 2021; Axt et al., 2021; Stern & Rule, 2018), which opens the possibility that the divergence of those findings with Study 1 results reflects cultural differences in social perception. Therefore, Study 2 tested a different sample (Italian participants) to ensure that any observed results generalize beyond the sample used in the first study (Spanish-speaking participants). In addition, the fact that participants took longer to categorize androgynous faces than sex-typical faces implies that they were exposed to androgynous faces for a longer period of time before the social judgement task. As a result, it cannot be ruled out that androgynous faces have become more ‘familiar’ than sex-typical faces and are rated more positively (see Stern & Rule, 2018). Therefore, by controlling the order of the experimental tasks (face evaluation task vs. social judgements), Study 2 explores whether our results were explained by the particularities of the experimental procedure. 4.2 Methods All relevant methodological variables were held constant with Study 1, except for one critical modification. We balanced the appearance of the face evaluation task between subjects: some participants completed the face evaluation task before the social judgement scales (i.e., after the individual differences section), while others completed it after the social judgement scales (i.e., at the end of the experiment). 4.2.1 Participants Initially, we recruited 55 participants. Subsequently, we executed some exclusions with the same criteria as Study 1. The valid sample was 45 participants (26 females, Mage =30.80 years, SD =7.72). All participants were Italian. The task was completed in a median time of 12.21 min. As in Study 1, all participants confirmed that they agreed to undergo the study and provided written consent in accordance with ethical procedures approved by the Bioethics Committee of Universidad Católica Luis Amigó, Medellín, Colombia, before the experiment. Demographic information was collected as in Study 1. 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 8ANSANI and OLIVERA-LA ROSA FIGURE 3 Creepiness as a function of faces’ appearance and experiment (violin plot). The form of the violin indicates the distribution curve of the predicted values. The boxplots within each violin represent IQRs. Black horizontal lines within the boxplots indicate median values. Black circles within the boxplots indicate mean values. The black dashed line indicates the grand mean. In the computation of the grand mean, Study 3 was excluded. 4.3 Results 4.3.1 Response times and face evaluation A GLMM identical to that of Study 1 was run. The androgynous faces increased the RTs both with respect to the masculine (β=.54 95% CI [0.42, 0.68] SE =0.06 p<.001) and feminine faces (β=.31, 95% CI [0.17, 0.44], SE =0.06, p<.001) (Figure 1). Feminine faces required more time than masculine to be categorized (p<.001). The main effect of the faces’ appearance was significant (χ2=31.6, df =2, p<.001, d= 0.70). The probability that an androgynous face was characterized as female was 77.0%, 95% CI [62.9, 86.9]. 4.3.2 Social judgements The same GLMMs were run, and the presentation order was added to the model to take into account its potential impact. The results were very similar to those of Study 1. The trustworthiness omnibus effect reached significance (χ2=11.7, df =2, p=.003, d=.29); proving that androgynous faces were rated as more trustworthy than masculine (β=.11, 95% CI [0.04, 0.17], SE =0.03, p<.001) and feminine faces, although this result just approached significance (β=.04, 95% CI [−0.004, 0.09], SE =0.02, p =.077). The result of the pairwise comparison indicated that the difference between masculine and feminine faces was also significant (p =.028), with feminine faces being perceived as more trustworthy than masculine ones (Figure 2). The effect of the faces’ appearance on creepiness was significant (χ2 =20.8, df =2, p<.001, d=0.42). The androgynous faces were perceived as less creepy than both masculine (β=−.20, 95% CI [−0.28, −0.11], SE =0.04, p<.001) and feminine (β=−.15, 95% CI [−0.23, −0.07], SE =0.04, p<.001) faces. The difference between masculine and feminine faces was not significant (p=.205). Comparably with Study 1, female participants’ creepiness scores were higher as opposed to those of male participants (β=.12, 95% CI [0.001, 0.25], SE =0.06, p =.047) (Figure 3). As for the shared moral values, the effect of the faces’ appearance failed to reach significance (χ2=3.2, df =2, p=.197, d=0.16) (Figure 4). In all models, no effect of presentation order was found (ps>.29) and the presentation order did not interact with the faces’ appearance (ps>.42). Furthermore, we inspected Akaike and Bayesian information criteria (i.e., AIC and BIC) values of the same models with and without the presentation order. In all cases, the model without the presentation order fits better with the data (i.e., lower values of AIC and BIC). For this reason, although we always manipulated presentation order, we will not include it in the models of Studies 3 and 4. 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License SOCIAL PERCEPTION AND ANDROGYNOUS TARGETS 15 Gutiérrez-García, A., Beltrán, D., & Calvo, M. G. (2019). Facial attractiveness impressions precede trustworthiness inferences: Lower detection thresholds and faster decision latencies. Cognition and Emotion,33(2), 378–385. https://doi.org/10.1080/02699931.2018. 1444583 Hies, O., & Lewis, M. B. (2022). Beyond the beauty of occlusion: medical masks increase facial attractiveness more than other face coverings. Cognitive Research: Principles and Implications,7, 1–6. Jackson, L. A. (1983). The perception of androgyny and physical attractiveness: Two is better than one. Personality and Social Psychology Bulletin, 9(3), 405–413. JASP Team. (2023). JASP (Version 0.17.3)[Computer software]. Jentsch, E. (1997). On the psychology of the uncanny (1906). Angelaki: Journal of the Theoretical Humanities,2(1), 7–16.(Original work published 1906). Judd, C. M., Westfall, J., & Kenny, D. A. (2012). Treating stimuli as a random factor in social psychology: A new and comprehensive solution to a pervasive but largely ignored problem. Journal of Personality and Social Psychology,103(1), 54–69. https://doi.org/10.1037/ a0028347 Kahneman, D. (2011). Thinking, fast and slow. Macmillan. Kark, R. (2020). Androgyny. Encyclopedia of personality and individual differences (pp. 160–166). Springer. https://doi.org/10.1007/978-3-319- 24612-3_1041 Kastendieck, T., Mauersberger, H., Blaison, C., Ghalib, J., & Hess, U. (2021). Laughing at funerals and frowning at weddings: Top-down influences of context-driven social judgments on emotional mimicry. Acta Psychologica,212, 103195. Kätsyri, J., Förger, K., Mäkäräinen, M., & Takala, T. (2015). A review of empirical evidence on different uncanny valley hypotheses: support for perceptual mismatch as one road to the valley of eeriness. Frontiers in psychology,6, 390. https://doi.org/10.3389/fpsyg.2015.00390 Langlois, J. H., Kalakanis, L., Rubenstein, A. J., Larson, A., Hallam, M., & Smoot, M. (2000). Maxims or myths of beauty? A meta-analytic and theoretical review. Psychological Bulletin,126(3), 390–423. Lick, D. J., & Johnson, K. L. (2013). Fluency of visual processing explains prejudiced evaluations following categorization of concealable identities. Journal of Experimental Social Psychology,49(3), 419–425. Lick, D. J., & Johnson, K. L. (2015). The interpersonal consequences of processing ease: Fluency as a metacognitive foundation for prejudice. Current Directions in Psychological Science,24(2), 143–148. Lick, D. J., Johnson, K. L., & Rule, N. O. (2015). Disfluent processing of nonverbal cues helps to explain anti-bisexual prejudice. Journal of Nonverbal Behavior,39, 275–288. Lo, S., & Andrews, S. (2015). To transform or not to transform: Using generalized linear mixed models to analyse reaction time data. Frontiers in Psychology,6, 1171. https://doi.org/10.3389/fpsyg.2015.01171 Ma, D. S., Correll, J., & Wittenbrink, B. (2015). The Chicago face database: A free stimulus set of faces and norming data. Behavior Research Methods, 47(4), 1122–1135. https://doi.org/10.3758/s13428-014-0532-5 Marini, M., Ansani, A., Demichelis, A., Mancini, G., Paglieri, F., & Viola, M. (2024). Real is the new sexy: The influence of perceived realness on selfreported arousal to sexual visual stimuli. Cognition and Emotion, Advance online publication. https://doi.org/10.1080/02699931.2023.2296581 Martin, A. E., & Mason, M. F. (2022). What does it mean to be (seen as) human? The importance of gender in humanization. Journal of Personality and Social Psychology,123(2), 292–315. https://doi.org/10.1037/ pspa0000293 Martin, A. E., & Slepian, M. L. (2021). The primacy of gender: Gendered cognition underlies the big two dimensions of social cognition. Perspectives on Psychological Science,16(6), 1143–1158. Masnan, M. J., Mahat, N. I., Shakaff, A. Y. M., Abdullah, A. H., Zakaria, N. Z. I., Yusuf, N., Subari, N., Zakaria, A., & Aziz, A. H. A. (2015). Understanding Mahalanobis distance criterion for feature selection. AIP Conference Proceedings,1660(1), 050075. https://doi.org/10.1063/1.4915708 Major, B., Carnevale, P. J., & Deaux, K. (1981). A different perspective on androgyny: Evaluations of masculine and feminine personality characteristics. Journal of Personality and Social Psychology,41(5), 988. McAndrew, F. T., & Koehnke, S. S. (2016). On the nature of creepiness. New Ideas in Psychology,43, 10–15. https://doi.org/10.1016/j.newideapsych. 2016.03.003 Mendes, W. B., & Koslov, K. (2013). Brittle smiles: Positive biases toward stigmatized and outgroup targets. Journal of Experimental Psychology: General,142, 923–933. Mori, M. (2005). The uncanny valley. (K. F. MacDorman, & T. Minato, Trans.). Energy,7, 33–35. (Original work published 1970). Ng, V. K. Y., & Cribbie, R. A. (2017). Using the gamma generalized linear model for modeling continuous, skewed and heteroscedastic outcomes in psychology. Current Psychology,36(2), 225–235. https://doi.org/10. 1007/s12144-015-9404-0 Norton, M. I., Frost, J. H., & Ariely, D. (2007). Less is more: The lure of ambiguity, or why familiarity breeds contempt. Journal of Personality and Social Psychology,92(1), 97. Norton, M. I., Frost, J. H., & Ariely, D. (2013). Less is often more, but not always: Additional evidence that familiarity breeds contempt and a call for future research. Journal of Personality and Social Psychology,105(6), 921–923. https://doi.org/10.1037/a0034379 Olivera-La Rosa, A., Chica-Franco, T., & Ingram, G. (2023a, December 18). A cognitive approach to the genesis of negative attitudes towards androgyny: Insights from the uncanny valley. https://doi.org/10.31234/osf.io/ fhwr3 Olivera-La Rosa, A., Villacampa, J., & Ingram, G. P (2023b, December 18). The creepy, the bad and the ugly: Exploring perceptions of moral character and social desirability in uncanny faces. Current Psychology,42, 1146–1156. https://doi.org/10.1007/s12144-021-01452-w Olivera-La Rosa, A. (2018). Wrong outside, wrong inside: A social functionalist approach to the uncanny feeling. New Ideas in Psychology,50, 38–47. https://doi.org/10.1016/j.newideapsych.2018.03.004 Olivera-La Rosa, A., Chuquichambi, E. G., & Ingram, G. P. (2020). Keep your (social) distance: Pathogen concerns and social perception in the time of COVID-19. Personality and Individual Differences,166, 110200. Olivera-La Rosa, A., Chuquichambi, E. G., Carmona, C. A., Arango-Tobón, O. E., Amador, O., Rosselló, J., & Ingram, G. P. (2021). The “Joker” laugh: Social judgments towards affective deviants in a sample of young offenders with callous unemotional traits. Personality and Individual Differences, 183, 111148. Olivera-La Rosa, A., Villacampa, J., Corradi, G., & Ingram, G. P. (2023). The creepy, the bad and the ugly: Exploring perceptions of moral character and social desirability in uncanny faces. Current Psychology,42, 1146–1156. https://doi.org/10.1007/s12144-021-01452-w Oosterhof, N. N., & Todorov, A. (2008). The functional basis of face evaluation. Proceedings of the National Academy of Sciences,105(32), 11087–11092. Orghian, D., & Hidalgo, C. A. (2020). Humans judge faces in incomplete photographs as physically more attractive. Scientific Reports,10(1), 110. Owen, H. E., Halberstadt, J., Carr, E. W., & Winkielman, P. (2016). Johnny Depp, reconsidered: How category-relative processing fluency determines the appeal of gender ambiguity. PloS ONE,11(2), e0146328. https://doi.org/10.1371/journal.pone.0146328 Perrett, D. I., Lee, K. J., Penton-Voak, I., Rowland, D., Yoshikawa, S., Burt, D. M., Henzi, S. P., Castles, D. L., & Akamatsu, S. (1998). Effects of sexual dimorphism on facial attractiveness. Nature,394(6696), 884–887. https://doi.org/10.1038/29772 Pollick, F. E. (2010). In search of the uncanny valley. Lect. Note Inst. Comput. Sci. Telecomm,40, 69–78. doi: 10.1007/978-3-642-12630-7_8 Quinn, P. C., Yahr, J., Kuhn, A., Slater, A. M., & Pascalis, O. (2002). Representation of the gender of human faces by infants: A preference for female. Perception,31(9), 1109–1121. https://doi.org/10.1068/p3331 Reber, R., & Schwarz, N. (2006). Perceptual fluency, preference, and evolution. Polish Psychological Bulletin,37(1), 16. 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 16 ANSANI and OLIVERA-LA ROSA Revilla, M., & Höhne, J. K. (2020). How long do respondents think online surveys should be? New evidence from two online panels in Germany. International Journal of Market Research,62(5), 538–545. https://doi.org/ 10.1177/1470785320943049 Revilla, M., & Ochoa, C. (2017). Ideal and maximum length for a web survey. International Journal of Market Research,59(5), 557–565. Rhodes, G., Harwood, K., Yoshikawa, S., Nishitani, M., & McLean, I. (2002). The attractiveness of average faces: Cross-cultural evidence and possible biological basis. In G. Rhodes & L. A. Zebrowitz (Eds.), Facial attractiveness: Evolutionary, cognitive, and social perspectives (pp. 35–58). Ablex. Rieger, G., Gygax, L., Linsenmeier, J. A. W., Siler-Knogl, A., Moskowitz, D. A., & Bailey, J. M. (2011). Sex typicality and attractiveness in childhood and adulthood: Assessing their relationships from videos. Archives of Sexual Behavior,40(1), 143–154. https://doi.org/10.1007/s10508-009- 9512-8 Ruscio, J. (2008). A probability-based measure of effect size: Robustness to base rates and other factors. Psychological Methods,13(1), 19–30. https:// doi.org/10.1037/1082-989x.13.1.19 Sadr, J., & Krowicki, L. (2019). Face perception loves a challenge: Less information sparks more attraction. Vision Research,157, 61–83. Said, C. P., Dotsch, R., & Todorov, A. (2010). The amygdala and FFA track both social and non-social face dimensions. Neuropsychologia,48, 3596–3605. https://doi.org/10.1016/j.neuropsychologia.2010.08.009 Schaller, M., & Park, J. H. (2011). The behavioral immune system (and why it matters). Current Directions in Psychological Science,20(2), 99–103. Sedikides, C., & Gregg, A. P. (2008). Self-enhancement: Food for thought. Perspectives on Psychological Science,3(2), 102–116. Smith, P. (2014). Of ‘near pollution’and non-linear cultural effects: Reflections on Masahiro Mori and the Uncanny Valley. American Journal of Cultural Sociology,2(3), 329–347. Sofer, C., Dotsch, R., Wigboldus, D. H., & Todorov, A. (2015). What is typical is good: The influence of face typicality on perceived trustworthiness. Psychological Science,26(1), 39–47. Sofer, C., Dotsch, R., Oikawa, M., Oikawa, H., Wigboldus, D. H., & Todorov, A. (2017). For your local eyes only: Culture-specific face typicality influences perceptions of trustworthiness. Perception,46(8), 914–928. Stern, C., & Rule, N. O. (2018). Physical androgyny and categorization difficulty shape political conservatives’ attitudes toward transgender people. Social Psychological and Personality Science,9(1), 24–31. https://doi.org/ 10.1177/1948550617703172 Strohminger, N., & Nichols, S. (2014). The essential moral self. Cognition, 131(1), 159–171. Stroup, W. W. (2013). Generalized linear mixed models: Modern concepts, methods and applications. CRC Press. Szczurek, L., Monin, B., & Gross, J. J. (2012). The stranger effect: The rejection of affective deviants. Psychological science,23(10), 1105– 1111. The jamovi project. (2022). jamovi. (Version 2.3) [Computer Software]. https://www.jamovi.org Todorov, A., Said, C. P., Engell, A. D., & Oosterhof, N. N. (2008). Understanding evaluation of faces on social dimensions. Trends in Cognitive Sciences, 12(12), 455–460. http://doi.org/10.1016/j.tics.2008.10.00 Todorov, A., Pakrashi, M., & Oosterhof, N. N. (2009). Evaluating faces on trustworthiness after minimal time exposure. Social Cognition,27(6), 813–833. Tucciarelli, R., Vehar, N., Chandaria, S., & Tsakiris, M. (2022). On the realness of people who do not exist: The social processing of artificial faces. IScience,25(12), 105441. Uhlmann, E. L., Pizarro, D. A., & Diermeier, D. (2015). A person-centered approach to moral judgment. Perspectives on Psychological Science,10(1), 72–81. Wang, S., Lilienfeld, S. O., & Rochat, P. (2015). The uncanny valley: Existence and explanations. Review of General Psychology,19(4), 393–407. Wang, S., & Rochat, P. (2017). Human perception of animacy in light of the uncanny valley phenomenon. Perception,46(12), 1386–1411. Ward, C., Flowe, H., & Humphries, J. (2012). The effects of masculinity and suspect gender on perceptions of guilt. Applied Cognitive Psychology, 26(3), 482–488. https://doi.org/10.1002/acp.2823 Watt, M. C., Maitland, R. A., & Gallagher, C. E. (2017). A case of the “heeby jeebies”: An examination of intuitive judgements of “creepiness”. Canadian Journal of Behavioural Science/Revue canadienne des sciences du comportement,49(1), 58–69. Willis, J., & Todorov, A. (2006). First impressions: Making up your mind after a 100-ms exposure to a face. Psychological Science,17, 592–598. https:// doi.org/10.1111/j.1467-9280.2006.01750.x Winkielman, P., Schwarz, N., Fazendeiro, T., & Reber, R. (2003). The hedonic marking of processing fluency: Implications for evaluative judgment. InEnJ.Musch&K.C.Klauer(Eds.),The psychology of evaluation: Affective processes in cognition and emotion (pp. 189–217). Lawrence Erlbaum Associates. Yamada, Y., Kawabe, T., & Ihaya, K. (2013). Categorization difficulty is associated with negative evaluation in the “uncanny valley” phenomenon. Japanese Psychological Research,55(1), 20–32. https://doi.org/10.1111/j. 1468-5884.2012.00538.x Yousefi, A., Paulk, A. C., Deckersbach, T., Dougherty, D. D., Eskandar, E. N., Widge, A. S., & Eden, U. T. (2015). Cognitive state prediction using an EM algorithm applied to Gamma distributed data. In 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (pp. 7819–7824). IEEE. https://doi.org/10.1109/EMBC. 2015.7320205 SUPPORTING INFORMATION Additional supporting information can be found online in the Supporting Information section at the end of this article. How to cite this article: Ansani, A., & Olivera-La Rosa, A. (2024). Moral inferences from androgynous faces are beyond categorical uncertainty: Evidence of a positive bias towards androgynous targets. European Journal of Social Psychology, 1–16. https://doi.org/10.1002/ejsp.3048 10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3048 by University Of Jyväskylä Library, Wiley Online Library on [08/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License