Re-evaluating the relationship between pathogen avoidance and preferences for facial symmetry and sexual dimorphism: A registered report. Joshua M. Tybur et al. Evolution and Human Behavior, Volume 43, Issue 3, May 2022, Pages 212-223. https://doi.org/10.1016/j.evolhumbehav.2022.01.003
Abstract: Over the past decade, a small literature has tested how trait-level pathogen-avoidance motives (e.g., disgust sensitivity) and exposure to pathogen cues relate to preferences for facial symmetry and sexual dimorphism. Results have largely been interpreted as suggesting that the behavioral immune system influences preferences for these features in potential mates. However, findings are limited by small sample sizes among studies reporting supportive evidence, the use of small stimulus sets to assess preferences for symmetry and dimorphism, and design features that render implications for theory ambiguous (namely, largely only investigating women's preferences for male faces). Using a sample of 954 White young adult UK participants and a pool of 100 White young adult stimuli, the current registered report applied a standard two-alternative forced-choice approach to evaluate both men's and women's preferences for both facial symmetry and dimorphism in both same- and opposite-sex targets. Participants were randomly assigned to either a pathogen prime or a control prime, and they completed instruments assessing individual differences in pathogen avoidance (disgust sensitivity and contamination sensitivity). Results revealed overall preferences for both facial symmetry and dimorphism. However, they did not reveal a relation between these preferences and disgust sensitivity or contamination sensitivity, nor did they reveal differences in these preferences across control and pathogen prime conditions. Null results of pathogen-avoidance variables were consistent across participant sex, target sex, and interactions between participant sex and target sex. Overall, findings cast doubt on the hypothesis that pathogen-avoidance motives influence preferences for facial symmetry or dimorphism.
Keywords: DisgustHealthMasculinitySymmetryMate preferencesBehavioral immune system
4. Discussion
The current registered report evaluated the relation between pathogen-avoidance motives and preferences for facial symmetry and dimorphism. It sought to test whether any such relation applied to preferences for both same- and opposite-sex targets – a phenomenon that might result from these features being interpreted as cues to infectiousness – or only in opposite sex targets – a phenomenon that might result from these features being treated as information regarding indirect benefits (i.e., genes that increase offspring fitness). Using a set of 100 target faces and a sample of 954 participants, we did not detect evidence consistent with either perspective. That is, we did not detect a relation between individual differences measures (pathogen disgust sensitivity and germ aversion) and general preferences for facial symmetry or dimorphism, nor did we detect a difference in this relation across same- and opposite-sex faces. Similarly, we did not detect an effect of a pathogen prime (relative to a control prime) on preferences for symmetry or dimorphism, nor did we detect differences in such preferences across same- versus opposite-sex targets. We discuss the implications of these findings for both the behavioral immune system literature and the face preferences literature below.
4.1. Implications for the behavioral immune system and face preferences
The null results observed here have some implications for how we view the functional specificity of the behavioral immune system. Current thinking conceptualizes the behavioral immune system as a suite of psychological mechanisms that monitor the environment for features that correlate with pathogen presence (i.e., cues to pathogens) and, when those features are detected, motivates behaviors that reduce the likelihood of infection (Ackerman et al., 2018; Schaller & Duncan, 2007; Lieberman & Patrick, 2014; Tybur & Lieberman, 2016). Byproducts of infection in conspecifics are some of the best candidates for such cues. And, indeed, people can distinguish between individuals experiencing an immune response from those who are not (Arshamian et al., 2021), and they avoid (and are sometimes disgusted by) individuals with rashes, ulcers, and pustules on their faces – some of the key symptoms of communicable diseases (Curtis et al., 2004; Kurzban & Leary, 2001; Oaten et al., 2011).
Following the logic presented in previous work investigating the relation between pathogen avoidance and preferences for facial symmetry and/or dimorphism, the hypotheses tested here were based on the idea that facial symmetry and dimorphism provide information regarding health, and that the behavioral immune system should motivate preferences for healthy targets (and, perhaps especially, healthy mates). However, features perceived as “healthy” need not be treated as information regarding infection threat. Health can refer to absence of infectious disease, but it can also refer to a number of other aspects of condition, including the absence of non-contagious parasites, the absence of non-contagious metabolic diseases, the absence of injury, the absence of psychopathology, etc. Just as the behavioral immune system should not be expected to influence fear of tigers or heights, both of which can be thought of as preserving some aspect of “health”, it should not be expected to influence preferences for facial symmetry or dimorphism unless those features act as cues to infectiousness. Given that the structural features that give rise to variation in facial symmetry and dimorphism are fairly stable across the lifespan – and given recent findings suggesting that dimorphism and symmetry (along with multiple other aspects of facial appearance) have poor validity as cues to multiple dimensions of health that might relate to infection proneness (Foo, Simmons, & Rhodes, 2017, Cai et al., 2019; see Jones, Holzleitner, & Shiramizu, 2021) – they are unlikely candidates as infection cues. These considerations (and, naturally, the results of the current study) raise questions regarding interpretations of earlier findings that pathogen avoidance relates to preferences for facial symmetry and dimorphism.
4.2. Implications regarding preferences for facial symmetry and dimorphism
Although this investigation was designed to evaluate the relation between pathogen avoidance and preferences for facial symmetry and dimorphism, its sample size and other design features (e.g., assessment of both same- and opposite-sex preferences for both facial symmetry and dimorphism) can contribute to the field's understanding of preferences for symmetry and dimorphism, at least in the population sampled from here. Consider, for example, comparing the current results with those reported by Little, Jones, DeBruine, and Feinberg (2008), who inferred that symmetry and dimorphism provide common information based on the observation that preferences for facial dimorphism correlate with preferences for facial symmetry. The current study similarly detected a positive relation between preferences for facial symmetry and preferences for facial dimorphism (see Table S1). It also replicates other findings reported by Little et al.: that men prefer dimorphism in female faces more than women do, and that women prefer dimorphism in male faces more than men do. However, it did not replicate a third finding from the same paper: that symmetry preferences are contingent on the sex of the rater and the target. Instead, we found that symmetric faces were preferred equally in same-sex and opposite-sex targets, for both men and women. The current data might prove useful for evaluating the robustness of other findings in the face preferences literature.
4.3. Limitations and future directions
4.3.1. Statistical power and potential false negatives
Non-significant results can emerge for multiple reasons, including experimenter error or participant inattention. Multiple aspects of our findings suggest that neither of these factors explains the critical null findings observed here. The fact that we detected global preferences for facial symmetry and facial dimorphism – with the latter preference moderated by participant sex and target sex – suggests that participants were (1) able to detect these features and (2) preferred them in a manner consistent with past studies sampling from the same population. Other incidental findings discount the null results reflecting systematic errors in data collection. For example, the sex difference in pathogen disgust sensitivity observed here (d = .41) was virtually identical to the meta-analyzed sex difference observed in a study of 11,501 participants across 30 nations (d = .41) (Tybur et al., 2016).
Even without experimenter error or participant inattention, null results can still reflect Type II errors. In random effects designs such as the one employed here, the probability of making such errors is influenced by myriad factors, including (1) the magnitude of the fixed effect(s), (2) the number of participants, (3) the number of stimuli, (4) variance accounted for by participants, (5) variance accounted for by stimuli, (6) variance in the relation between participant-level individual differences (e.g., pathogen disgust sensitivity) and preferences across different stimuli, etc. We aimed to minimize the probability of making such Type II errors, even if effect sizes were small, by (1) having a large sample size (N = 954), (2) having a large pool of stimuli (N = 100), and (3) manipulating multiple factors within-participants. However, because we were unable to model all random effect components in our power analyses, results from these power analyses might be imprecise, and we cannot state with confidence the effect sizes that we had adequate power (>80%) to detect. Nevertheless, inspection of the 95% confidence intervals around effect size estimates can provide an idea of the uncertainty in our parameter estimates and the plausible upper bounds of population-level effect sizes (see Table 2). These confidence intervals are narrow and largely centered around zero. Inspection of the confidence intervals collapsing across stimuli can also be informative (see Tables S1–S3), since most prior studies in this literature have not used random effects analyses. Using this approach, the upper limit of the 95% confidence interval for the main effect of pathogen disgust sensitivity on facial dimorphism preferences was r = .12, and the upper limit of the 95% confidence interval for the relation between pathogen disgust sensitivity and facial symmetry preferences was r = .08. Given the nature of the indirect benefits hypothesis, confidence intervals around simple effects within participant sex by target sex interactions (for both symmetry and dimorphism preferences, and for both pathogen disgust and germ aversion as predictors) can also be informative, especially concerning cross-sex preferences. For men, none of the upper limits of these confidence intervals exceeded r = .15; for women, none exceeded r = .07. In total, these results suggest that any relations we failed to detect are likely to be small in magnitude. Future studies on this topic should be designed to detect effect sizes no larger than the upper limits of these confidence intervals.
4.3.2. Validity of the dependent measure and stimuli
In line with previous studies in this literature, we investigated the degree to which pathogen-avoidance motives relate to attraction to facial symmetry and sexual dimorphism. Perceptions of attractiveness need not fully regulate the physical proximity, direct contact, or indirect contact that influences pathogen transmission, though. Recent studies in the pathogen-avoidance literature have asked participants how comfortable they would be with physical contact with a target (e.g., Van Leeuwen & Petersen, 2018), and one of these studies found only a modest relationship between target facial attractiveness and contact comfort (Tybur, Lieberman, Fan, Kupfer, & de Vries, 2020). Although the current study did not detect a relation between pathogen avoidance and attraction to facial symmetry or dimorphism, future research could better test whether people are more averse to infection-risky acts with individuals with low dimorphism or low symmetry faces (cf. Kupfer, 2018; Ryan et al., 2012).
As is standard in this literature, we used a two-alternative forced-choice response format. Recent work has suggested that this method partially assess face matching ability rather than variation in preferences (Lewis, 2020), and that it can produce results that differ from those obtained with paradigms in which individual faces are rated for attractiveness (Jones & Jaeger, 2019; Lee, De La Mare, Moore, & Umeh, 2021). Also following standard procedures in this literature, we manipulated base faces to be 50% more similar to male or female prototypes (for the dimorphism manipulation) or 50% more or less similar to a perfectly symmetric version of the base face. We cannot rule out the possibility that pathogen avoidance would relate to preferences for facial dimorphism or symmetry if transformations were more or less extreme.
4.3.3. Effects of the COVID-19 pandemic
We collected data in May 2021, after approximately 4,500,000 COVID-19 cases and 125,000 deaths had been confirmed in the UK in the 14 months since the pandemic began (Roser, 2021). Some recent work has argued that the SARS-CoV-2 outbreak has increased pathogen disgust sensitivity (Boggs, Ruisch, & Fazio, 2022; Stevenson, Saluja, & Case, 2021). Such increases, if sufficiently strong, could attenuate the relation between pathogen disgust sensitivity and preferences for facial symmetry or dimorphism. Our data give no reason to suspect that pathogen disgust sensitivity was unusually high in the population we sampled from, though. The mean observed here was virtually indistinct (and, if anything, slightly lower) from that in the sample of U.S. college students (N = 507) used to validate the Three-Domain Disgust Scale (Tybur et al., 2009) and that in a large (N = 7166) online English-speaking sample recruited shortly before the pandemic (O'Shea, DeBruine, & Jones, 2019) (see the online supplement for more details). There are also reasons to question whether, how, and why the presence of SARS-CoV-2 would affect how the behavioral immune system detects or processes cues to pathogens. Like many other respiratory pathogens, SARS-CoV-2 is largely spread via invisible respiratory droplets and aerosols expelled when (often asymptomatic or pre-symptomatic) individuals breath, talk, or sing (Greenhalgh et al., 2021). Those infected with SARS-CoV-2 typically exhibit symptoms similar to those caused by the myriad endemic respiratory pathogens that circulated widely before the COVID-19 pandemic (e.g., coughing, sneezing, headache, fatigue, fever) (Tostmann et al., 2020). And, while SARS-CoV-2 causes serious illness in some people, its appearance coincided with the virtual elimination of many other respiratory viruses from circulation (Yeoh et al., 2021). These reasons raise doubts that the pandemic conditions that began in early 2020 would affect the behavioral immune system, at least via increases in the presence of detectable transmission risks, changes in observable illness symptoms in others, or increases in encounters with pathogens oneself (Ackerman, Tybur, & Blackwell, 2021). Future work can clarify whether, how, and why the pandemic affects the behavioral immune system in other manners.
4.3.4. Generalizability to other populations
The current study sampled from a population of young adult (<35) heterosexual White individuals from the UK, and it assessed attraction toward young adult White targets. Some findings indicate that preferences for facial dimorphism – perhaps especially in male targets – varies across ecologies (DeBruine, Jones, Crawford, Welling, & Little, 2010; Marcinkowska et al., 2019; Scott et al., 2014), as do preferences for at least some other dimensions of facial appearance (e.g., coloration; Han et al., 2018). Hence, our findings might not generalize to other populations. However, most studies that have reported relations between pathogen avoidance and preferences for facial symmetry or dimorphism have sampled from similar populations and assessed attraction toward similar targets (though see Saribay et al., 2021 and Zheng et al., 2016). Future work could certainly test whether pathogen avoidance relates to such preferences in other populations, even if such a relationship does not exist in the population sampled from here.
4.3.5. Validity of priming method and concluding thoughts
Most studies in the behavioral immune system literature assess individual differences in pathogen-avoidance motives using either the Perceived Vulnerability to Disease Scale or the Three-Domain Disgust Scale (Oosterhoff, Shook, & Iyer, 2018; Tybur et al., 2014). Multiple studies have clarified the validity of these instruments (e.g., Duncan et al., 2009; Tybur et al., 2009). There is less consistency in approaches used to experimentally manipulate pathogen-avoidance motives and, relatedly, less evidence supporting the validity of these procedures. For example, studies have reported that each of the following experimental manipulations produces effects consistent with behavioral immune system hypotheses: (1) asking participants to consciously reflect upon past experiences with infection (e.g., Moran et al., 2021; Murray, Kerry, & Gervais, 2019); (2) exposing participants to olfactory cues to pathogens (e.g., Tybur, Bryan, Magnan, & Hooper, 2011); (3) having participants read essays describing pathogen-risky situations (e.g., White, Kenrick, & Neuberg, 2013); (4) having participants complete a disgust sensitivity instrument immediately before the dependent measure (e.g., Lee & Zietsch, 2011; Navarrete & Fessler, 2006; Watkins et al., 2012); and (5) exposing participants to disgust-eliciting images or slideshows showcasing pathogen risks (e.g., Faulkner, Schaller, Park, & Duncan, 2004; Hill, Prokosch, & DelPriore, 2015; Mortensen, Becker, Ackerman, Neuberg, & Kenrick, 2010; Park, Schaller, & Crandall, 2007). Using a combination of those last two approaches – methods used in studies that have reported effects of pathogen primes on preferences for facial symmetry or dimorphism (Ainsworth & Maner, 2019; Little et al., 2010; Watkins et al., 2012; Young et al., 2011) – we did not detect an effect of the priming manipulation. Other recent studies have similarly reported not detecting effects of pathogen primes on, among other things, conformity (Van Leeuwen & Petersen, 2021), political attitudes (Shook & Oosterhoff, 2020), moral sentiments (Makhanova, Plant, Monroe, & Maner, 2019), and attitudes toward immigrants (Ji, Tybur, & van Vugt, 2019). Following these null findings, the behavioral immune system literature would benefit from large-scale, registered, collaborative work using multiple priming approaches to test the same hypothesis. Such an endeavor would be valuable for multiple reasons. Like the current study, it could be used to replicate studies that used methods that, in retrospect, might not be as robust as originally assumed. It could also give an unbiased assessment of the effect sizes that researchers should expect from priming methods; such an assessment would prove valuable for future study designs. And it could indicate which of the multiple manipulations used in the literature – from images to essays to odors – give rise to the largest of such effect sizes. In sum, taking a look at the methods and results used in past behavioral immune system work can improve future developments in this area.
No comments:
Post a Comment